Yamada Winnewiser Interstellar Molecules

Springer Tracts in Modern Physics
Volume 241
Managing Editor
G. Höhler, Karlsruhe, Germany
Editors
A. Fujimori, Tokyo, Japan
J. Kühn, Karlsruhe, Germany
Th. Müller, Karlsruhe, Germany
F. Steiner, Ulm, Germany
W. C. Stwalley, Storrs, CT, USA
J. Trümper, Garching, Germany
P. Wölfle, Karlsruhe, Germany
For further volumes:

http://www.springer.com/series/426
Springer Tracts in Modern Physics
Springer Tracts in Modern Physics provides comprehensive and critical reviews of topics of
current interest in physics. The following fields are emphasized: elementary particle
physics, solid-state physics, complex systems, and fundamental astrophysics.
Suitable reviews of other fields can also be accepted. The editors encourage prospective
authors to correspond with them in advance of submitting an article. For reviews of topics
belonging to the above mentioned fields, they should address the responsible editor,
otherwise the managing editor.
See also springer.com
Managing Editor Solid-State Physics, Editors

Gerhard Höhler Atsushi Fujimori
Institut für Theoretische Teilchenphysik Editor for The Pacific Rim
Karlsruhe Institut für Technologie KIT, Department of Physics
Postfach 69 80 University of Tokyo
76128 Karlsruhe, Germany 7-3-1 Hongo, Bunkyo-ku
Phone: +49 (7 21) 6 08 33 75 Tokyo 113-0033, Japan
Fax: +49 (7 21) 37 07 26 Email: fujimori@phys.s.u-tokyo.ac.jp
Email: gerhard.hoehler@KIT.edu http://wyvern.phys.s.u-tokyo.ac.jp/
www-ttp.physik.uni-karlsruhe.de/ welcome_en.html
Peter Wölfle
Elementary Particle Physics, Editors
Institut für Theorie der Kondensierten Materie
Johann H. Kühn Karlsruhe Institut für Technologie KIT,
Institut für Theoretische Teilchenphysik Postfach 69 80
Karlsruhe Institut für Technologie KIT, 76049 Karlsruhe, Germany
Postfach 69 80 Phone: +49 (7 21) 6 08 35 90
76049 Karlsruhe, Germany Fax: +49 (7 21) 6 08 77 79
Phone: +49 (7 21) 6 08 33 72 Email: peter.woelfle@KIT.edu
Fax: +49 (7 21) 37 07 26 www-tkm.physik.uni-karlsruhe.de
Email: johann.kuehn@KIT.edu
www-ttp.physik.uni-karlsruhe.de/~jk Complex Systems, Editor
Frank Steiner
Thomas Müller
Institut für Theoretische Physik
Institut für Experimentelle Kernphysik Universität Ulm
Karlsruhe Institut für Technologie KIT, Albert-Einstein-Allee 11
Postfach 69 80 89069 Ulm, Germany
76049 Karlsruhe, Germany Phone: +49 (7 31) 5 02 29 10
Phone: +49 (7 21) 6 08 35 24 Fax: +49 (7 31) 5 02 29 24
Fax: +49 (7 21) 6 07 26 21 Email: frank.steiner@uni-ulm.de
Email: thomas.muller@KIT.edu www.physik.uni-ulm.de/theo/qc/group.html
www-ekp.physik.uni-karlsruhe.de
Atomic, Molecular and Optical Physics
Fundamental Astrophysics, Editor William C. Stwalley
Joachim Trümper University of Connecticut
Max-Planck-Institut für Extraterrestrische Physik Department of Physics
Postfach 13 12 2152 Hillside Road, U-3046
85741 Garching, Germany Storrs, CT 06269-3046, USA
Phone: +49 (89) 30 00 35 59 Phone: +1 (860) 486 4924
Fax: +49 (89) 30 00 33 15 Fax: +1 (860) 486 3346
Email: jtrumper@mpe.mpg.de Email: w.stwalley@uconn.edu
www.mpe-garching.mpg.de/index.html www.phys.uconn.edu/faculty/stwalley.html
Koichi M. T. Yamada Gisbert Winnewisser
Editors
Interstellar Molecules
Their Laboratory and Interstellar Habitat
123
Editors
Dr. Koichi M. T. Yamada Prof. Dr. Gisbert Winnewisser
National Institute of Advanced Industrial Alte Str. 8a
Science & Technology (AIST) 79249 Merzhausen
EMTech Germany
Onogawa 16-1
305-8569 Tsukuba
Ibaraki, Japan
e-mail: kmt.yamada@aist.go.jp
ISSN 0081-3869 e-ISSN 1615-0430
ISBN 978-3-642-16267-1 e-ISBN 978-3-642-16268-8
DOI 10.1007/978-3-642-16268-8
Springer Heidelberg Dordrecht London New York
Ó Springer-Verlag Berlin Heidelberg 2011
This work is subject to copyright. All rights are reserved, whether the whole or part of the material is
concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation,
broadcasting, reproduction on microfilm or in any other way, and storage in data banks. Duplication
of this publication or parts thereof is permitted only under the provisions of the German Copyright Law
of September 9, 1965, in its current version, and permission for use must always be obtained from
Springer. Violations are liable to prosecution under the German Copyright Law.
The use of general descriptive names, registered names, trademarks, etc. in this publication does not
imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
Cover design: eStudio Calamar, Berlin/Figueres
Printed on acid-free paper
Springer is part of Springer Science+Business Media (www.springer.com)

Preface
About 30 years ago, the Cologne laboratories were founded with a design such that
laboratory spectroscopy and appropriate interstellar observations would support
each other by supplying new results. From the very early stage, the Terahertz
spectroscopy has been one of the most important subjects of the laboratory, but it
has remained to be one of the most poorly explored areas of the entire electro-
magnetic spectrum. It was our dream to have a space observatory, with which we
can observe the spectra in Terahertz region without the strong attenuation of the
radiation by the Earth’s atmosphere.
It is a great pleasure of us to publish this textbook for the students who are
learning molecular astrophysics at the occasion when the tremendous amount of
new results for the interstellar molecules in the Terahertz region are reported
successively from the Herschel space observatory.
In the first chapter of this book, the history from ancient days to now is briefly
presented by Gisbert Winnewisser with the editorial help of Koichi M. T. Yamada.
In the second chapter, the molecules observed in the interstellar medium and
circumstellar envelops are reviewed by Karl M. Menten and Friedrich Wyrowski.
Quite large number, more than 150, of molecular species have been detected up to
today, and thus it is impossible to describe all of those molecules. We have limited
ourselves to present only the selected molecules: molecules detected in the early
stage, prototypical molecules, and those detected most recently.
The third chapter accounts for the molecular spectroscopy, which is the irre-
placeable tool for the astrophysics. The basic theory for the spectroscopy of atoms,
diatomic molecules, and polyatomic molecules are briefly presented in Sects. 3.1–
3.3; Sects. 3.1 and 3.2 are contributions of K. M. T. Yamada, and 3.3 of Per Jensen
with the appendix by K. M. T. Yamada. Since many text books have been pub-
lished concerning to the molecular spectroscopy, the descriptions are limited to the
very fundamental level.
The number of molecules with internal rotation observed in the interstellar
space is increasing steadily. The spectra of such molecules are rather complicated
and are of interest not only for astrophysics but also for molecular spectroscopy.
Section 3.4 is a contribution of Nobukimi Ohashi, which accounts for the internal
v
vi Preface
rotation problems with one- and two-methyl tops. Various methods for the energy
calculation for such molecules are described in detail. The symmetry labeling of
the energy levels are also discussed on the basis of permutation-inversion group.
We wish to acknowledge Prof. Kaori Kobayashi’s assistance to produce many of
the figures in Sects. 3.4.
In the fourth chapter, the most recent progress in the laboratory spectroscopy in
Cologne is presented by Thomas F. Giesen, and the development of heterodyne
mixers, the essential device for the radio observation, is reviewed by Karl Jacobs.
In the last chapter, the molecules observed in the interstellar space are listed on
the basis of the Cologne Database for Molecular Spectroscopy (CDMS).
We would like to thank the Cologne students, post doctoral fellows, and all
collaborators, whose enthusiastic and untiring efforts are essential for many of the
beautiful results presented in this book. We would like to express special thanks to
Helga Winnewisser, who has supported Gisbert mentally and physically during the
preparation of this book.
On behalf of the contributors
August, 2010 Koichi M. T. Yamada

Contents
1 History . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Prelude . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.1 The Ancient Mechanical Universe . . . . . . . . . . . . . . . 1
1.1.2 The Dark Ages . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.1.3 The Renaissance of Science . . . . . . . . . . . . . . . . . . . 4
1.1.4 Discovery of Dark Nebulae. . . . . . . . . . . . . . . . . . . . 5
1.1.5 Dawn of Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . 6
1.2 Interstellar Medium (ISM) . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.2.1 Atomic and Molecular Hydrogen . . . . . . . . . . . . . . . . 8
1.2.2 The 21-cm Line of Neutral Hydrogen . . . . . . . . . . . . 9
1.2.3 Recombination Lines . . . . . . . . . . . . . . . . . . . . . . . . 10
1.2.4 Atomic Fine- and Hyperfine-Structure Lines . . . . . . . . 10
1.3 Interstellar Molecular Lines. . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.3.1 Small Molecules . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.3.2 Large Molecules . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.3.3 Non-polar Molecules . . . . . . . . . . . . . . . . . . . . . . . . 14
1.3.4 Atomic Carbon and CO Molecule . . . . . . . . . . . . . . . 15
1.3.5 Molecular Cloud and Star Formation . . . . . . . . . . . . . 16
1.4 Molecular Spectroscopy: An Irreplaceable Tool. . . . . . . . . . . . 17
1.4.1 Scope of Interstellar Atomic
and Molecular Spectroscopy . . . . . . . . . . . . . . . . . . . 18
1.4.2 Terahertz Region: Far–InfraRed Domain . . . . . . . . . . 19
1.5 Episodical Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
1.5.1 Life Story of Fraunhofer . . . . . . . . . . . . . . . . . . . . . . 21
1.5.2 Life Story of Herschel . . . . . . . . . . . . . . . . . . . . . . . 22
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2 Molecules Detected in Interstellar Space . . . . . . . . . . . . . . . . . . . . 27

2.1 Detection of the First Interstellar Molecules . . . . . . . . . . . . . . 27
2.2 The habitat of Interstellar Molecules . . . . . . . . . . . . . . . . . . . 30
vii
viii Contents
2.3 Metal-bearing Molecules in the Interstellar Medium. . . . . . . . . 30

2.4 Light Hydrides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.5 Ions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
2.5.1 Cation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
2.5.2 Anions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
2.6 Chain Molecules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.7 Symmetric and Asymmetric Top Molecules . . . . . . . . . . . . . . 35
2.8 Cyclic Molecules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
2.9 Complex Molecules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
2.10 Isotopologues and Deuterated Molecules. . . . . . . . . . . . . . . . . 37
2.11 Prospects for Future Detections . . . . . . . . . . . . . . . . . . . . . . . 39
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3 Fundamentals of Spectroscopy for Astrophysics . . . . . . . . . . . . . . 43

3.1 Atoms. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.1.1 Hamiltonian for the Hydrogen Atom . . . . . . . . . . . . . 43
3.1.2 Schrödinger Equation . . . . . . . . . . . . . . . . . . . . . . . . 44
3.1.3 Angular Wavefunction . . . . . . . . . . . . . . . . . . . . . . . 45
3.1.4 Radial Wavefunction . . . . . . . . . . . . . . . . . . . . . . . . 46
3.1.5 Atomic Orbit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
3.1.6 Electron Spin and Fine Structure . . . . . . . . . . . . . . . . 48
3.1.7 Nuclear Spin and Hyperfine Structure . . . . . . . . . . . . 49
3.1.8 Atoms with More Than One Electron. . . . . . . . . . . . . 49
3.2 Diatomic Molecules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
3.2.1 Hamiltonian for a Diatomic Molecule without Spin . . . 57
3.2.2 Hamiltonian for a Diatomic Molecule without
Electronic Angular Momentum . . . . . . . . . . . . . . . .. 60
3.2.3 Hamiltonian for a Diatomic Molecule with
Electronic Angular Momentum . . . . . . . . . . . . . . . . . 65
3.2.4 Hyperfine Structure . . . . . . . . . . . . . . . . . . . . . . . . . 70
3.3 Polyatomic Molecules. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
3.3.1 The Born–Oppenheimer Approximation . . . . . . . . . . . 71
3.3.2 The Choice of Rotation–Vibration Coordinates . . . . . . 73
3.3.3 The Eckart Equations . . . . . . . . . . . . . . . . . . . . . . . . 78
3.3.4 Normal Coordinates . . . . . . . . . . . . . . . . . . . . . . . . . 81
3.3.5 Molecular Rotation–Vibration Energies . . . . . . . . . . . 84
3.3.6 The Rigidly Rotating Molecule . . . . . . . . . . . . . . . . . 87
3.3.7 The Harmonically Vibrating Molecule . . . . . . . . . . . . 93
3.3.8 Effective Hamiltonians . . . . . . . . . . . . . . . . . . . . . . . 97
3.3.9 Molecular Symmetry . . . . . . . . . . . . . . . . . . . . . . . . 101
3.3.10 TAppendix to 3.3. . . . . . . . . . . . . . . . . . . . . . . . . . . 108
3.4 Molecules with Internal Rotation . . . . . . . . . . . . . . . . . . . . . . 119
3.4.1 Molecule with One Methyl Internal Rotor. . . . . . . . . . 120
3.4.2 Two-Top Internal-Rotation Problem . . . . . . . . . . . . . . 150
Contents ix
3.4.3 Case of a Molecule with Two Equivalent Methyl

Internal Rotors. . . . . . . . . . . . . . . . . . . . . . . . . . ... 160
3.4.4 Tunneling Matrix Formulation for Molecules with
Two Methyl Internal Rotors . . . . . . . . . . . . . . . . ... 166
3.4.5 Appendix to 3.4. . . . . . . . . . . . . . . . . . . . . . . . . ... 172
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 177
4 Laboratory Experimental Methods . . . . . . . . . . . . . . . . . . . . . . . . 179

4.1 Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
4.1.1 Molecules Relevant to Laboratory Astrophysics. . . . . . 179
4.1.2 Spectral Line Widths and Intensities . . . . . . . . . . . . . 181
4.2 Monochromatic Radiation Sources for High Resolution
Absorption Spectroscopy. . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
4.2.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
4.2.2 Terahertz Backward-Wave Oscillators . . . . . . . . . . . . 183
4.2.3 BWO Terahertz Spectrometers . . . . . . . . . . . . . . . . . 186
4.2.4 Sub-Doppler Spectroscopy . . . . . . . . . . . . . . . . . . . . 187
4.2.5 Multiplier-Spectrometer . . . . . . . . . . . . . . . . . . . . . . 189
4.2.6 Planar Schottky Multipliers . . . . . . . . . . . . . . . . . . . . 191
4.2.7 Monolithic Membrane Diodes . . . . . . . . . . . . . . . . . . 192
4.2.8 COSSTA: Cologne Sideband Spectrometer for THz
Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
4.3 Spectroscopy in Supersonic Jets. . . . . . . . . . . . . . . . . . . . . . . 195
4.3.1 General Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
4.3.2 SuJeSTA. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
4.3.3 OROTRON . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198
4.4 Heterodyne Mixers for High Resolution Spectroscopy
of Interstellar Molecules . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
4.4.1 Heterodyne Detection . . . . . . . . . . . . . . . . . . . . . . . . 203
4.4.2 SIS Mixers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
4.4.3 SIS Mixer Device Design and Fabrication . . . . . . . . . 207
4.4.4 Superconducting Hot-Electron Bolometers . . . . . . . . . 213
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
5 List of Molecules Observed in Interstellar Space. . . . . . . . . . . . . . 219
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
Chapter 1
History
Koichi M. T. Yamada and Gisbert Winnewisser
1.1 Prelude
1.1.1 The Ancient Mechanical Universe
Claudius Ptolemaeus who lived in the second century (83–161 AD) was an ancient
mathematician, geographer, astronomer, and astrologer in Roman Egypt.1 Ptol-
emy, as he is known in English, was the last great and dominating figure before the
dawn of the renaissance of science during the years 1500–1700. During the life-
span of Ptolemy, arithmetical techniques were developed for calculating astro-
nomical phenomena. Greek astronomers, e.g., Hipparchus, had produced
geometric models for calculating celestial motions. Ptolemy himself contributed to
the fitting of the astronomical data. He had developed elaborate arithmetical
techniques for calculating celestial motions. Ptolemy, however, claimed to have
derived his geometrical models from selected astronomical observations by his
predecessors spanning more than 800 years, though astronomers have for centuries
suspected that his models’ parameters were adopted independently of
K. M. T. Yamada (&)
National Institute of Advanced, Industrial Science and Technology (AIST), EMTech,
Onogawa 16-1, Tsukuba,
Ibaraki 305-8569, Japan
G. Winnewisser
Alte Str. 8a, 79249 Merzhausen, Germany
1
See, e.g., Encyclopedia Britannica: http://www.britannica.com/.
K. M. T. Yamada and G. Winnewisser (eds.), Interstellar Molecules, 1

Springer Tracts in Modern Physics 241, DOI: 10.1007/978-3-642-16268-8_1,
2 K. M. T. Yamada and G. Winnewisser
observations. Ptolemy presented his astronomical observations in a convenient

form of tables, which could be used directly to compute the past or future position
of the planets. The Almagest (‘‘The Great Treatise,’’ originally written by Ptol-
emy) also contains a star catalog, which is an appropriate version of the catalog
created by Hipparchus.
Since ancient times, astronomers of the heavens had noted that from night to
night the stars moved across the sky in a regular procession, trailing from east to
west, following the motions of the sun and the moon. In this night time procession,
the stars maintained their heavenly order. If one inspects the movement of the Big
Dipper, one notices in a short time that the five planets move across. Today, we
find that, with the exception of five stars, the rest would hold their relative posi-
tions. The five stars looped forward and backward from these, west to east, and
then from east to west. The Greeks, observing this remarkable fact repeatedly,
called each of these five celestial bodies asteres planetai, the Greek word for
wandering stars. But today, we call them the planets, i.e., Mercury, Venus, Mars,
Jupiter, and Saturn. The looping motion of each planet could be duplicated if the
planet were anchored to the outer rim of a wheel that was located on a track
moving across the sky. When the observed positions of the planets were plotted,
they agreed fairly well with this description of the twelve wheels rolling across the
sky. But if the ancient astronomers tried to make the calculated motions of the
planets fit the observations accurately, then the system became very complicated
and cumbersome. It had to be assumed that some of the wheels themselves rolled
on the rims of other wheels. The best job of fitting was done by Ptolemy. He and
his followers found that 80 wheels within wheels were needed for just the five
planets known at that time. Some planets required wheels within wheels within
wheels.
To account for departures of heavenly bodies from uniform circular motion,
early astronomers devised an elaborate system, based on mechanical or mathe-
matical models. In its simplest form, the system employs an earth-centered circle
called a deferent. The second circle, called an epicycle, has a center Q which lies
on the deferent and revolves uniformly counter-clockwise around the earth from
west to east. The point Q was said to be ‘‘immaterial,’’ since it did not contain
matter. The celestial body, meanwhile, revolves uniformly around Q on the epi-
cycle circle. The second motion can be either clockwise or counter-clockwise. The
apparent motion of the body relative to the earth is the combined motion in the two
circles.
Ptolemy was the author of 13 scientific treatises in astronomy: three of which
were of continuing importance to later Islamic and European science. The first
book he published is an astronomical treatise now known as the Almagest
(in Greek or ‘‘The Great Treatise’’). Discussion of the geographic knowledge is the
content of a second treatise the Geography, which boils down to a thorough
discussion of the geographic knowledge of the Greco-Roman world. The third
one is the astrological treatise known as the tetrabiblos (‘‘Four books’’) in which
he attempted to adapt horoscopic astrology to the Aristotelian natural philosophy:
the first actual star catalog in the second century.
1 History 3
This catalog appeared in the book known later as Almagest (which is a Latin
corruption of the name of the Arabic translation, Al-mijisti). It had 1,025 entries;
the positions of these bright stars had been measured by Hipparchus 250 years
earlier. Ptolemy’s catalog was the only widely used one before to the seventeenth
century, i.e., the renaissance of science. His collected works, amounted to 13
books of the mathematical composition, were written in Alexandria, and these
books comprise the best guide to Ptolemy’s ideas and to the contributions of those
who preceded him. Out of careful observation, logical deduction, and the appli-
cation of basic geometrical principles, without telescopes and with only primitive
measuring devices at hand, the Greeks compiled a tremendous amount of obser-
vational data concerning the motions of the planets with respect to the stars.
Although beclouded here and there with vague philosophical ideas and super-
stitions, the Greek concept of the universe was the prod. The geocentric world of
Ptolemy had as the basic assumption that the Earth is situated at the center of an
un-movable universe. The movement of the moon and the sun on the other hand,
were still described with relative ease as a fairly simple circular motion problem.
1.1.2 The Dark Ages
Peace and prosperity continued to reign until about the end of the second century
AD. Then, a series of catastrophes dissolved the Roman Empire and all but
destroyed civilization itself. Decadent political and social systems collapsed. The
exhaustion of gold and silver coins, as the Spanish mine ran out of the precious
minerals, paralyzed commerce. A horrible pestilence brought back from Asia by
the military, struck down the population at the rate of 2,000–3,000 persons per day,
rendering wide areas of being completely depopulated. Barbarian tribes pressed in,
conquering and devastating southern Europe. The centuries that followed were sad
times for science. Deprived of comfort in this world, people looked forward to
solace in a better world thereafter, and turned away from the science of the past.
Christianity, the major force during the Middle Ages, brought with it a growing
desire to reconcile science in general and concepts of the universe in particular
with the literal statements of the Scriptures. Lactantius ridiculed the concept of a
spherical earth. Most people came to believe that the earth was flat, square, and
supported by four pillars, one at each corner. Under such circumstance, contri-
butions to science were few. The ‘‘meteors’’ became signs in the sky prophetic of
events, evil or good, about to happen on the earth. Superstition and fear were the
motivating forces in astronomical thought for many centuries.
For some time, the torch of learning passed into hands of the Arabs. They made
few advances, but at least they preserved in the Almagest the writings of Ptolemy
and they continued to make a few observations of the motions of the heavenly
bodies. The Koran, fortunately, was silent on questions that related to concepts of
the earth and the universe. By the end of the twelfth century, interest was rekindled
in learning, in agriculture, craft, commerce, and in science. But an aura of
superstition, mysticism, and magic continued to obstruct the development of the

sciences, especially astronomy and chemistry.
The pseudoscience of alchemy burgeoned, bearing a relationship to chemistry
similar to that of astrology to astronomy. There was a widespread belief that with
the use of magic, potions, and incantations, one could turn base metal into gold or
silver. Failure did not discourage the alchemists and magicians. From time to time,
they even made, by accident, some useful chemical discovery. In such an atmo-
sphere, it is not surprising that astrology flourished. The idea that God, through the
stars, could influence mankind was not unacceptable to the church fathers. And so,
with the full approval of the church, astrologers continued to cast horoscopes as
they did in the ancient world.
1.1.3 The Renaissance of Science
Without any obvious or logical reason the fifteenth century undergoes decisive
changes in the lifestyle particularly the way science was handled. This trend
started in Italy, and soon it spread out to the north. As an example of changing, the
mood and the way it can be traced was demonstrated by looking at the setup and
behavior of the imperial court of the Emperor Karl IV (Charles IV) who laid down
in the year 1348, the foundation of Charles University at Prague. Charles IV did
recognize the importance of liberal arts and science at large and in particular for
his fellow country men. By about 1600, the development of modern science
started, which has rendered such a large number of results that we are proud to use
still today. At his residence in Prague, Emperor Karl IV employed two outstanding
experts in the field of astronomy, or as you would nowadays call them, scientists:
Tycho de Brahe (1546–1601), who provided superb quality of the astronomical
measurement, and his successor Johannes Kepler (1571–1630).
Galileo Galilei (1563–1642) and Johannes Kepler, who lived in the same period
but geometrically separated by a large distance, reached to a common conclusion
of the heliocentric movement of the planets, which had been proposed much
earlier by the great pioneer, Nicolaus Copernicus (1473–1543).
The harvest out of their high quality measurements was reaped by Isaac Newton
(1643–1727). Today, the complicated movements of planets, against which
Ptolemy and his successors fought desperately, can be explained by simple prin-
ciples: Newton’s equations of motion and gravitation theory. Newton’s equations
of motion form the beginning of classical mechanics. The observed continuous
nature of macroscopic motion in space and time can be explained by applying
integral and differential calculus to the solution of dynamic problems. The prin-
cipal ideas embodied in the deterministic formulation of classical mechanics.
It consists of the following:
1. The system in question is observationally accessible at all times, and obser-
vations can be executed without disturbing the system under investigation.
1 History 5
2. By defining the state of a system in time, all the mechanical variables of the
system at that particular time can be determined.
3. Combining the specification of the state in statement 2 with the laws of
mechanics allows for a complete description of the system as a function of time.
1.1.4 Discovery of Dark Nebulae
Dark nebulae, or dark dust clouds, as we designate them nowadays, have been
noticed early on as huge voids in an otherwise star-filled heaven. Clearly, it must
have been an utter nuisance for star counters, like Sir F. W. Herschel, who were
hampered in getting reliable numbers for their star-counting business. If Herschel
would have been able to use his scrutinizing eyes, then he probably would have got
great satisfaction out of the fact that the cold dark clouds also take part in star
formation, more so individually, i.e., as single stars. In contrast to the cold clouds,
with temperatures of around 10 K, the warm dark clouds having temperatures well
over 15 K are heated by external sources or by embedded stars. By choosing the
proper molecules or class of molecules, like symmetric top molecules, the kinetic
temperature can be estimated rather quickly from the level-populations.
Therefore, Herschel investigated all astrophysically interesting phenomena,
stars, nebulae, and star clusters. He discovered something like 2,000–3,000
nebulae and clusters. In his article on the ‘‘Construction of the Heavens,’’ he
presented speculation on the evolution of matter by successive steps from the
nebulous material to the final planetary state.
As it has turned out over the years and very recently with the advent of
molecular spectroscopic investigation of the dark clouds, one is not surprised to
find out that one of the earliest references in the literature addresses the Taurus
Cloud Complex. It is mentioned explicitly in an important publication by Herschel
(1784), entitled: ‘‘Account of some observations tending to investigate the con-
struction of the Heavens.’’ In his description of the nebulae, he noted that ‘‘the
spaces preceding them were generally quite deprived of their stars, so as often to
afford many fields without a single star in it.’’ Here, he clearly refers to the task of
star counting. He found himself, as figuratively expressed it ‘‘. . . on nebulous
ground.’’ He refers in this early publication to those regions in the sky, where no
stars are observable, or where the number and the brightness of background stars is
drastically reduced relative to the surrounding areas. A particularly beautiful
example of these conspicuous dust clouds are the dark dust clouds in Taurus
region, with the truly remarkable sources, like TMC 1, TMC 2, . . .; very cold
ðTkin 810 KÞ; very quiet sources with almost no turbulences within the sources
themselves, and therefore with no indication of any start of star-formation.
In Fig. 1.1, which is an excellent ‘‘historic’’ reproduction of Barnard’s plate No. 5
of the nebulous region in Taurus [1]. In his description of this region, Barnard
comments that ‘‘very few regions of the sky are so remarkable as this one . . .’’ and
Fig. 1.1 This is the Taurus

region with the famous dark
clouds which are for the
Terahertz frequencies such an
important dark region, see,
e.g., Winnewisser and
Kramer [46]. The photograph
image is reproduced with the
permission of copyright
holder, Carnegie Institution
he continues ‘‘. . . and bears strongest proof of the existence of obscuring matter
in space.’’
1.1.5 Dawn of Spectroscopy
The very gifted young Joseph von Fraunhofer succeeded in manufacturing high-
quality lenses and constructed one of the finest refractory telescopes ever built. He
invented a heliometer and a diffraction grating for measuring the wavelength of
light precisely. In addition, he acquired great skill in the manufacture of homog-
enous glass prisms, which reached unparalleled quality. These technical achieve-
ments, in particular, his uniquely homogeneous glass prisms, gave him the edge
over all his contemporaries working on similar project in dispersing light. Pointing
his refracting telescope together with his ‘‘self-built spectrograph’’ toward the
sun and some of the close-by stars, he discovered that the sun’s spectrum, when
sufficiently dispersed, was punctuated by a large number of fine dark absorption
lines, about 700 of which he studied in great detail. He collected an atlas of these
lines, naming the strongest features consecutively as A, B, C, D, . . .
Currently, a few of these cataloged lines are still known to spectroscopists by
the famous Fraunhofer notation, such as the two sodium D-lines, which he named
D1 and D2 : We now know that the 3p level of sodium (Na) displays an observable
fine structure, so closely spaced doublet. Thus, the two Fraunhofer lines corre-
spond to the following transitions: D1 ¼ 2P1=2 2S1=2 ¼ 589:5940 nm, and
D2 ¼ 2P3=2 2S1=2 ¼ 588:9973 nm:
About the time when Joseph von Fraunhofer found a large number of fine dark
absorption lines in the solar spectrum, he published his thorough and extensive
research study on the solar spectrum in form of a catalog. This monumental work
1 History 7
on the solar spectrum was followed in 1823 by a second catalog, summarizing

stellar lines discovered in different stars close to the sun. It may be justified to state
that the beginning of spectroscopy was set by the very professional level of
J. v. Fraunhofer in presenting his results.
The optical window was traditionally always used to carry through the obser-
vational program and check out new observational part of the electromagnetic
spectrum to be used in the quest for recording weaker and ever weaker signals from
distances well beyond the sun. The construction of a telescope is a considerably
demanding task during the last century to record absorption lines in the spectrum of
the sun and other stars. Around the year 1860, Kirchhoff and Bunsen marked the
beginnings of what we now call modern astrophysics. Their contribution was the
recognition that every chemical element exhibits a characteristic spectrum which
can be used to detect it in the laboratory and in far distant cosmic objects.
The first molecules in space were discovered in cometary spectra, as early as
1864, and a spirit of science. They were studied for the first time visually by
Donati in comet Temple [2]. The spectral features he saw were presumably the
474.0, 516.5, and 563.0 nm band groups of C2 : But at the time, they were not in a
position to identify the lines as such, because the knowledge of comparing spectra
with each other was just not yet known. This step would have provided the
important final check. In this context, we note that Huggins [3] in 1882 first
observed the 405.0-nm group of lines, which was identified only in 1951 by
Douglas [4] as due to the C3 molecule. In the meantime, C3 has been studied
extensively in the laboratory employing optimized spectrometers, covering the
microwave region almost continuously to the ultraviolet. The C3 molecule has
recently been found in interstellar space.
1.2 Interstellar Medium (ISM)
The enormously vast spaces between the stars in our Galaxy and in between the
stars of the extragalactic domain are by no means void. On the contrary, they
contain an interesting and highly surprising, but totally unexpected composition of
a rarefied gas and dust, which is very in homogeneously distributed.
The study of ISM started about the turn of the century (about 1904), when
Hartmann noticed that matter must be dispersed among the stars. At that time, he
was checking in great detail the Doppler velocities of stellar absorption lines in
the gas associated with the close binary system d-Orionis. He noticed that some of
the CaII resonance absorption lines did not participate in the system’s periodic
Doppler shift, whereas others did. Subsequently, ‘‘stationary’’ Na-D lines
(Fraunhofer notation) were also discovered in d-Orionis and in other stars.
The main components/constituents of interstellar medium, ISM, consist of
molecular particles and interstellar dust, with a % ratio of gas to dust of about
99–1. The interstellar particles are predominantly in the gas phase in neutral form
and consist of atoms, molecules, free radicals, whereas interstellar dust consists
of lm-particles of rather uncertain chemical composition. A smaller fraction of the

gas, ca. 3% is found to be ionized.
Hydrogen, H, is with about 70% by mass, the most abundant element in the
universe and it is followed in abundance by helium (28% by mass). Hydrogen,
helium, and deuterium are considered to be primordial in nature. The available
evidence indicates that these three elements were created during the earliest
evolutionary phases of the universe, i.e., the Big Bang. All elements, heavier than
helium may have been synthesized more slowly by thermonuclear reactions in the
various stellar burning stages involving intensive mass loss from stars in their late
burning stages which include supernovae, novae, and planetary nebula. However,
the details of the processes which led to the present chemical composition of the
Galaxy and the Universe are far from being understood. The composition of the
interstellar dust on the other hand is unknown.
Atoms, both neutral and ionized, exhibit spectral features throughout most of the
electromagnetic spectrum. These spectral lines—in emission and/or absorption—
carry detailed and rather specific information on the physical and chemical condi-
tions of the sources from which they emanate. In fact, the heights of the energy
levels of the observed transitions together with the degree of their population furnish
a reliable measure of the excitation conditions and thus the temperature of the source
itself. The relative intensities of various transitions of a molecule, in conjunction
with the principle of statistical equilibrium, can be used to derive the gaseous
number-density of the source. The widely different line profiles observed can be
analyzed to provide information on the sources’ dynamical states, i.e., whether the
line widths are thermal, broadened by turbulence, or by pressure.
Atomic spectral features have played, and still play a crucial role in the study of
stellar atmospheres. The classification of stars by their atomic optical transitions
has been an important field of astronomy since the turn of the century. In addition
to standard dipole-allowed spectral transitions, atomic lines are found in the radio,
infrared, and optical regions in the form of recombination lines, fine-structure lines
and hyperfine-structure lines.
1.2.1 Atomic and Molecular Hydrogen
H2 molecule forms on the surfaces of dust grains, in the ISM. The dust exerts,
because of its large opacity, a strong influence on the diffuse interstellar radiation
field and undoubtedly it has a fundamental effect on the processes whereby
interstellar clouds collapse to form stars. Energy generated in the stars is absorbed
and re-emitted by the ISM. Molecular hydrogen, H2, stands out to be the most
abundant interstellar molecule. As shown by Spitzer and Zweibel [5, 6], it will be
synthesized rather efficiently on the surface of interstellar dust grains. The dust
contains a significant fraction of the interstellar heavy elements and thereby ties up
important atomic species that can cool the interstellar gas throughout and causes
increasingly excited fine-structure transitions.
1 History 9
Progress in characterizing the nature of interstellar dust will therefore not only
provide important information to secure knowledge about the chemical compo-
sition of significant constituent of the universe forming stars, which in turn
undoubtedly has a fundamental effect to introduce very significant uncertainties in
the energy distributions of astronomical sources located behind dust clouds.
The influence of high mass stars on the ISM is tremendous: during their for-
mation process, they assume various stages of luminosity. Important new results
were obtained within the last 40–50 years. In fact, the early construction of a radio
telescope in the 1930s by K. G. Jansky made him discover with his home-built
telescope in his backyard a completely new era of physics by investigating the
Milky Way in different wavelength regions. The free–free radiation was discov-
ered from a plasma, HII regions were seen for the first time, and most important,
the discovery of the 21-cm line of atomic hydrogen, predicted on account of
theoretical calculations by H. C. van de Hulst in 1944.
1.2.2 The 21-cm Line of Neutral Hydrogen
Approximately 5–10% of the mass of the Milky Way is in the form of interstellar
atomic hydrogen. The study of the physical
properties of this matter is possible
only because the atomic ground state 2 S1=2 can undergo a hyperfine transition
giving rise to the radio emission at the wavelength of 21.1 cm. The transition
occurs when the electron reverses its spin relative to that of the proton. The higher
energy state is the one in which the two spins are parallel to each other. The lower
state is the one where the two spins are anti-parallel. The prediction that this
should be an observable transition was made by van de Hulst in 1944 just before
the end of the Second World War; neutral hydrogen line should emit at a wave-
length about 21 cm, although it is a highly forbidden transition with a spontaneous
transition probability with every twelve million years, because of extremely low
changes in energy when the relative spins of the proton and electron in a hydrogen
atom change. The Doppler effect of the neutral hydrogen line was readily obser-
vable, displayed, and recorded, by producing a frequency shift of the neutral
21 cm atomic hydrogen line.
The use of the 21-cm line of atomic hydrogen for mapping the distribution of
low temperature regions opened new vistas on the structure and dynamics of
galaxies. Since then, additional lines were discovered in the cm- and dm-
spectroscopic areas and assigned to high quantum numbers of hydrogen and
helium lines. Furthermore, a fair number of diatomic and polyatomic molecular
lines have been assigned. They came as a certain surprise, and they are a signature
of the low density interstellar clouds.
This famous and fundamental transition (2 S1=2 ; F ¼ 1 0; 1420.405 751
786(1) MHz) was originally detected in space by Ewen and Purcell [7] and by
Muller and Oort [8]. It has since been used to map out neutral diffuse interstellar
gas in large regions of our

galaxy and others. Despite
the low transition rate for
spontaneous emission A21 cm ¼ 2:86 1015 s1 ; this line is very easily
detectable in space because of the large column densities of atomic hydrogen. The
frequency of transition was accurately determined in the laboratory mmn ¼
1420:405751767 MHz [9].
1.2.3 Recombination Lines
Recombination lines are emitted during the ðfree ! boundÞ capture of an electron
by ions such as Hþ ; Heþþ ; Cþ ; . . .; or during the ensuing bound–bound cascade
processes. These processes typically occur with Dn ¼ 1; 2; 3; where n is the
principal quantum number, and are referred to as alpha, beta, gamma, . . ., tran-
sitions, respectively. According to the Rydberg formula, the position of these
transitions are (in cm1 )
" #
1 1
~mji ¼ Ra 2 2 ð1:2:1Þ
ni nj
where the quantum numbers are given by ni [ nj ; and Ra ¼ 1:097 105 ð1 me =ma Þ
is the atomic Rydberg constant in cm1 , me is the electron mass
ð5:586 104 amuÞ, and ma is the atomic mass. While low-n transitions appear in
the UV (Lyman lines, lower level n ¼ 1; La ; Lb ; . . .; with the suffix a; b; . . .
denoting Dn ¼ 1; 2; . . .), in the optical (Balmer series, n ¼ 2; Ha ; Hb ; . . .), and the
adjoining near-infrared region, high-n (main quantum number *50 and larger)
recombination lines are found throughout the radio region. Transitions with
n [ 110 appear at frequencies lower than 5 GHz. Hydrogen recombination lines
have been used in the last 30 years; see, e.g., Wilson and Filges [10] and Vers-
chuur and Kellermann [11] for detailed studies of ionized regions called Hþ or HII
regions. These ionized regions are always associated with young O- and B-stars
which have recently formed from cool interstellar matter. Thus, recombination
lines, in conjunction with atomic fine–hyperfine-structure lines, as well as
molecular transitions (see below), are now among the tools of astrophysicists
used to study star formation and the interactions of ionized plasma with dense
molecular clouds.
1.2.4 Atomic Fine- and Hyperfine-Structure Lines
Atoms and molecular ions with unpaired outer electrons can possess a non-zero
electronic orbital angular momentum L and a spin angular momentum S: The
total angular momentum J ¼ L þ S in Russel–Saunders coupling scheme.
1 History 11
The configuration of the atoms outer electron shell determines J and thus the
multiplicity of the fine-structure levels, which arise from the magnetic spin–orbit
coupling. The magnitude of the fine-structure splitting is approximately of the
order of the square of the fine-structure constant a * (1/137)2 times the electronic
energy (10–100 eV) and is largest for low n levels. Fine-structure transitions are
found predominantly in the far-infrared region.
Since the first detection of the forbidden far-infrared lines of neutral oxygen
(O, OI) at 63:2 lm [12, 13], and ionized carbon (Cþ ; CII) at 157:7 lm [14], fine-
structure lines have become important tools in probing the atomic and ionized
portions of interstellar matter. Neutral carbon C (CI), neutral O (OI), and ionized
carbon, Cþ (CII), are used as the main tracers to probe interface regions separating
cool neutral atomic and molecular material from ionized material. In star-forming
regions, the surface layers of the cloud or the individual clumps it may consist of,
facing embedded or nearby young OB-stars, are ionized by the far-UV radiation
emitted by these stars. Photo-dissociation regions (PDRs) are zones where the far-
UV radiation beyond the Lyman continuum (13.6 eV) ionizes atoms with lower
ionization potentials than hydrogen, such as C (11.3 eV), the alkali metals
ð 5 eVÞ; S (10.3 eV), and others. In this way, a portion of the stellar luminosity is
absorbed, leading to local heating and subsequent re-radiation, predominantly in
the far-IR region of the spectrum. In fact, since the discovery of the 2P3=2 ! 2P1=2
CII fine-structure line of singly ionized carbon [14], it has become clear [15] that
the CII emission arises from moderately dense ð 1; 000 cm3 Þ and warm
ðT [ 200 KÞ gas associated with the surfaces of molecular clouds (see, e.g.,
Genzel et al. [16]). For lower density regions where hydrogen is atomic, the CII
line is also predicted to be the main coolant [17]. CII emission has also been
observed toward gas-rich nuclei of luminous spiral galaxies [15, 18, 19] with the
result that this line contributes about 1% of the total far–infrared luminosity. This
radiation has to be longward of the 91.2 nm Lyman limit, because hydrogen atoms
blanket stellar radiation below 91.2 nm. Thus, the 157:7409-lm CII line (labo-
ratory measurements by Cooksy et al. [20]) has turned out to be one of the
brightest fine-structure transitions. It plays an important role in cooling denser
regions by collisions with electrons, translational and rotationally excited hydro-
gen molecules in which the energy is first transferred to the CII species and then
radiated away. The excitation of the hydrogen molecules stems from far-UV
pumping into vibrationally excited states.
Actually, the most important cooling lines arise from neutral oxygen O (the two
fine-structure transitions: 3P0 ! 3P1 at 146 lm and 3P1 ! 3P2 at 63 lm). Other
atomic cooling lines arise from neutral carbon, C, and Siþ : Neutral atomic carbon
has a 1s2 2s2 2p2 electronic configuration, and its two outer electrons give rise to a
singlet and triplet system. According to Hund’s rule for terms arising from
equivalent orbitals, those with largest multiplicity lie the lowest. Thus, the triplet
system is energetically the lowest and therefore of particular astrophysical interest
in low-temperature regions. The 3P electronic ground state is split into three fine-
structure levels of different total angular momentum J, i.e., 3P0 ; 3P1 , and 3P2 in
increasing energy, respectively. This ‘‘normal’’ term order in atomic carbon—the

level with lowest J lies the lowest, according to Hund’s rules for less than
half-filled shells—is ‘‘inverted’’ for atomic oxygen. With a 1s2 2s2 2p4 electronic
configuration, oxygen possesses the same ground state ð3PÞ as neutral carbon, but
since the p-subshell is more than half filled, now the level with highest J lies the
lowest, i.e., 3P2 ; 3P1 , and 3P0 in ascending energy.
1.3 Interstellar Molecular Lines
In this section, we present a brief general overview of interstellar molecular lines;

more details are given in Chap. 2. Tables in Chap. 5 list all the presently known
gas phase interstellar molecules. The molecular species range in size from 2 to 13
atoms. They are typically found in the so-called dense(!) interstellar clouds with
tremendous sizes of 1–100 light years, average gas densities of 102 103 cm3 ;
and temperatures in the range 10–60 K, although both higher temperatures and gas
densities are found in localized regions where star formation is occurring. In
addition, differing physical conditions exist not only on the edges of clouds but
also in areas of highly clumped molecular material, where external UV radiation of
close-by sources is important. There are some well-studied cases where UV
radiation penetrates through the dilute inter-cloud medium deeply into the clumpy
interior of the molecular cloud, ionizing the exposed surface layers of the indi-
vidual clumps. The Orion Molecular Cloud and the regions M17 SW and S140 are
typical examples where ionized regions—consisting mainly of hydrogen in the
form of HII ðHþ Þ—border molecular clouds. Some of the organic molecules
shown in the tables in Chap. 5 have been detected in extended circumstellar
envelopes of cool, old stars that are carbon rich, i.e., possess more elemental
carbon than oxygen. The more normal pattern in stars is for the elemental abun-
dance of oxygen to exceed that of carbon: the cool envelopes of these stars do not
contain large abundances of most organic molecules [21].
The most abundant interstellar molecule is H2 ; this is due to both to the high
elemental abundance of hydrogen compared with all heavy elements found in
molecules (see, e.g., Watson [22]), and to an efficient chemical pathway for its
formation from hydrogen atoms.
The second most abundant molecule—carbon monoxide—has a ‘‘fractional
abundance’’ with respect to molecular hydrogen of 10-4. Most of the remaining
molecules are organic or carbon containing in nature. In interstellar clouds, both
normal and highly unusual molecules exist compared to those that we know on
earth; we have discussed their nature in previous articles [23, 24]. But it carries
importance due to the fact that CO is very stable and carries a fairly low dipole-
moment. All organic molecules have significantly smaller fractional abundances
than CO ð\108 Þ and can be regarded as trace constituents, although the amount
of interstellar organic molecules is quite large compared with their terrestrial
counterparts.
1 History 13
In addition to the gas phase molecules, there is additional ambiguous low-

resolution spectroscopic evidence for condensed phase species.
1.3.1 Small Molecules
The first molecule to be discovered in the ISM were the diatomic radicals, CH and
CN, and the positive ion CHþ : Dunham and Adams [25] found in the high reso-
lution optical spectra toward hot background O-stars in addition to a number of
interstellar atomic lines (Na, Ti, . . .), a few fairly sharp absorption features which
were clearly interstellar in character. Shortly thereafter, these lines were identified
and assigned by Swings and Rosenfeld [26], and by Douglas and Herzberg [27].
For details of these fundamental discoveries, we refer the reader to a personal
recollection and several articles by Herzberg [28]. Optical absorption lines of these
three diatomic molecules as well as optical and ultraviolet spectra of other dia-
tomic molecules detected subsequently are observed in the so-called diffuse
interstellar clouds which have gas densities of about 102 cm3 :
Relatively soon after the discovery of the first three poly-atomic molecules,
NH3 ; H2 O; and H2 CO, in space, it became clear that, by employing high spectral
and spatial resolution programs, the interstellar molecules reside in molecular
clouds where they are synthesized and where the sites for star formation are
located. Although there is nothing like a standard molecular cloud, they are highly
diversified and reach rather different sizes, spanning a size range from close to few
solar masses to several 106 M : Obviously, molecular clouds are not only pro-
ducing a rich and highly diversified chemistry, but they are also exposed to vastly
different physical conditions, which reflect themselves in displaying anomalous
line radiation patterns, as will be discussed later on.
The concept of the diffuse interstellar clouds that they were formed out of gas
and dust took about another 30 years before the medium started getting appreci-
ated slowly.
One of the well-known spectacular sources is in the heart of the Taurus
molecular cloud complex. This Taurus region is one of the many dark dust cloud
complexes containing a sizeable number of small dense and cool low mass frag-
ments. Among them, the most well-studied cloudlet is TMC 1, the Taurus
Molecular Cloud 1, which harbors the longest carbon chain molecules detected in
space. It contains unusually high abundance of cyanopolyynes and other linear
carbon chain molecules, forming a ridge of molecular emission.
The OH radicals can be observed by microwave techniques. Some of the light
hydrides are still followed later. This time, the detection was made at a frequency
of 1.6 GHz using a radio telescope which represents the first radio-line to be
discovered. However, the field of interstellar molecules was opened by the dis-
covery of three polyatomic molecules NH3 ; water H2 O, and formaldehyde H2 CO.
These three discoveries were instrumental to open the flood gates to the more than
150 molecules now known to roam the interstellar space. However, the list of
interstellar molecules is not complete. Missing may be others due to a variety of

problems mainly technical, as well as by exhibiting transitions in the Terahertz
region only, because of which many of the possible transitions will be completely
blocked out by the atmospheric extinction.
1.3.2 Large Molecules
The discovery of more than 150 molecular species in interstellar space via their
rotational transitions observed in the microwave, millimeter-wave, and adjoining
Terahertz region has demonstrated investigation of interstellar matter. The dis-
tribution of ISM is far from being homogeneous; on the contrary, the ISM displays
an inhomogeneous distribution, as can be seen and derived from inspection of
results obtained from large scale mappings of molecular clouds. In addition,
mapping of molecular clouds will supply detailed information on local inhomo-
geneity in molecular clouds—densities can reach up to 109 cm3 : It has become
clear that, within the last 20 years of molecular astrophysics, the hydrogen, with
about 50% of interstellar matter being mainly in the form of hydrogen gas, is
molecular and concentrated in extensive molecular–cloud complexes, often called
Giant Molecular Clouds (GMCs): gas densities of 104 cm3 and temperatures
below about 100 K.
The majority of the interstellar molecules are carbon-containing organic sub-
stances (they are summarized in tables in Chap. 5). From those tables, you can see
that the largest group of the interstellar species are diatomic molecules and
radicals. It amounts to 34 individual species, most of which have been detected in
the millimeter-wave and adjoining Terahertz regions. The molecules which con-
tain three atoms are linear or bent and produce accordingly more difficult spectra.
The molecules with four atoms are 23 in number while the five- and six-atomic
count up to 18 and 16, respectively. The larger molecules with 7–12 atoms are
rapidly decreasing, and the HC11 N remains up to the present day the largest
molecule by mass. The spectra of the various molecules will be discussed after we
have explained the theoretical background of the various molecules. The close
agreement with high resolution laboratory spectra furnishes the ultimate exacting
test for their definite identifications. Molecules have been detected throughout the
electromagnetic spectrum, predominantly, however, within the last 20 years in the
microwave, sub-millimeter wave, and adjacent far–infrared regions.
1.3.3 Non-polar Molecules
The list of interstellar molecules (tables in Chap. 5) is not complete, in the sense
that only polar molecules can be observed by microwave only. Molecules which
do not possess an electric dipole moment are still missing because of a variety of
1 History 15
problems, mainly technical and by exhibiting transitions in the Terahertz region

only. The lower rotational lines of light hydrides lie in the far–infrared regions and
this cannot be observed from the ground, since radiation of these wavelengths—
with a few exceptions—is completely blocked by the atmosphere. The molecules
observed in the gas phase and surfaces of dust particles are most likely produced
from atoms, ions, and smaller molecules by local chemical processes, see, e.g.,
Winnewisser and Herbst [29] and Herbst [30]. The results from the Infrared Space
Observatory obtained with the Short Wavelength Spectrometer operated in the
2–20 lm region, are beautiful spectroscopic evidence for a variety of ice mantle
and refractory grain core features, such as solid-state CO, CO2 , and silicates [31].
The chemically important or ‘‘biogenic’’ elements, carbon, nitrogen, and oxy-
gen, have abundances which are fractions between 10-3 and 10-4 of the hydrogen
abundance and are expected to be picked up by their ro-vibrational lines. Carbon,
and, in consequence, the carbon-chain molecules, form the main architectural
theme of the discovered interstellar molecules. Besides the cyanopolyynes and
their various derivatives, pure carbon cluster molecules have been observed in
space as well: CCC [32] and CCCCC [33] have been detected in the circumstellar
shell of the late type giant carbon star IRC 10216. In addition, a tentative
assignment of CCC toward the galactic center source Sagittarius B2, a star-
forming region, has been reported [34]. Very recently the laboratory gas-phase
electronic spectrum of CCCCCCC has been measured [35], and five lines show
close coincidence with visible diffuse interstellar bands (DIBs), probably paving
the way toward future identifications of DIBs.
In molecular clouds, two types of processes dominate by which molecules can
be synthesized from precursor atoms and ions: gas phase chemical reactions and
reactions that occur on the surfaces of dust particles. The latter mechanism is
invoked for the formation of molecular hydrogen, H2 ; as well as for the synthesis
of larger interstellar molecules.
Within the denser parts of the molecular cloud, the conditions for star formation
are met in many locations and, thus, one observes quite often the formation of
multiple star formation. In many real examples, up to 5,000 new stars are born.
1.3.4 Atomic Carbon and CO Molecule
The fine-structure energy-level separation in carbon is smaller than that for oxygen
and lies in the sub-millimeter region, and the two fine-structure transitions have
therefore been observed with radio-astronomical techniques. The forbidden mag-
netic dipole transitions of neutral carbon 3P2 3 P1 and 3P1 3 P0 have been
measured in the laboratory in carbon discharges [20, 36] to lie at 809 343.5(9) and
492 160.651(55) MHz, respectively. Neutral atomic carbon has been found to exist
throughout molecular cloud cores with an abundance relative to CO of 10–20%
[16, 37, 38]. Very recent observations in the S140 cloud [39] revealed that the CI
emission near the interface between the HII and neutral material region consists of
narrow and broad velocity components. In the broad velocity material, the relative
abundances of CI and CO are comparable.
The large extent and spatial coexistence of emission from CI, CII, and warm
CO are in accordance with models of PDRs [40, 41, 42]. These models show that
the chemical structure of a cloud illuminated by UV radiation 105 times the local
(near the sun) interstellar value and with a gas density of 2 105 cm3 changes
sharply as a function of the visual extinction Av : Av is a parameter closely
resembling the optical depth and it provides a measure of the penetration depth
into a cloud (see, e.g., [43, 44]). At Av ¼ 1 2 penetration atomic hydrogen
transforms into its molecular form, i.e., H2 with carbon and oxygen remaining in
the form ionic CII and neutral OI, respectively. Deeper into the cloud ðAv 5Þ,
ionized carbon decreases sharply and neutral carbon, CI, peaks giving way further
inward to material in molecular form, mainly CO. The study of these interface
regions is of strong interest, particularly in view of the observed highly clumped
form of the molecular material and its probable connection to star formation in
these areas.
1.3.5 Molecular Cloud and Star Formation
During the past 50–60 years, powerful methods and highly sensitive instrumen-
tation were developed and finally assigned to specific tasks for purposes of analytic
research for studying atoms and molecules. Among these methods are atomic and
molecular spectroscopy. They can reach linear dimensions of several hundreds of
light years which contain total masses of 106 M : The closest GMC to the Solar
System is the Orion molecular cloud complex with a distance of about 450 pc or
1,500 light years.
Molecular clouds are the most massive and in extent the largest individual
objects, both galactic and extragalactic. These clouds can be studied using
molecular transitions as general probes to reveal to us the deep interior of active
sites of star formation. Star formation is an on-going process, and it can be
monitored ‘‘live’’ by radio telescopes. If one accepts for the moment the ‘‘Big
Bang’’ as the beginning of the universe, then all the elements, with the exception
of hydrogen (H), deuterium (D), and helium (He), have been synthesized slowly by
the thermonuclear reactions in the hot interior of stars. During their final thermo-
nuclear stages, stars are known to return part of their processed material to the
ISM, enriching it with elements heavier than the light elements, hydrogen and
helium, which they have synthesized during the various nuclear burning stages. On
the other hand, there is little doubt that the ISM furnishes the raw material for new
stars to form.
However, star formation takes place in the denser parts of molecular clouds,
where hydrogen has been converted to molecular hydrogen, H2 ; and where the
spectra of molecules such as CO, CS, etc. make an important contribution to the heat
balance. Interstellar molecular radiation from more than 150 molecules—together
1 History 17
with some atomic fine structure lines—continues to excite the astronomers around
the world and arouses their fancy. So far, the interstellar molecules serve as the most
powerful tool for probing deep into the interior of molecular clouds. The field of
interstellar spectroscopy has matured to the point that optical absorption spectros-
copy has only rather limited value, since the high dust content of the dark clouds
makes them opaque to visible and ultraviolet light.
1.4 Molecular Spectroscopy: An Irreplaceable Tool
Historically, the interdependence of laboratory spectra and those received from the
heavens was made obvious by Fraunhofer in the wake of his first, great spectro-
scopic discovery in 1718. Forty years after Fraunhofer’s original discovery of the
solar spectrum, the power of spectroscopy was clearly demonstrated by comparing
spectra which were obtained in the laboratory with those recorded from stars. This
correlation marked the time (1769) and place (Heidelberg) of the establishment of
spectroscopy as an analytic tool. Kirchhoff and Bunsen, professors of physics and
chemistry at the University of Heidelberg recognized the analytic power of
spectroscopy and the unparalleled capability to study both physical and chemical
processes in the laboratory and in cosmic dimensions. The new tool ‘‘spectros-
copy’’ and its future applications were stated in an article of Kirchhoff and Bunsen
that it would open ‘‘an entirely untrodden field, stretching far beyond the limits of
the earth or even our solar system.’’ They praised the almost inconceivable sen-
sitivity and applicability of spectroscopy that would lead to many new discoveries.
The early spectroscopic techniques of Kirchhoff and Bunsen are basically still in
use, but they have been extended duly refined, and it has become a computerized
assignment process.
The basic spectroscopic technique of matching specific wavelengths or fre-
quencies from distant objects with those spectral features obtained in the labora-
tory has remained in use till date and has supplied us with an abundance of detailed
interstellar vistas of a fascinating information. The enormous technological and
scientific advances which have occurred in the last 30 years have allowed for
bringing forth two developments. First, they have made spectroscopic observations
possible throughout virtually the entire electromagnetic spectrum—though with
different resolution and sensitivity—providing access to an overwhelming amount
of spectral features. Second, they have pushed the accuracy of spectral line
identification in some wavelength regions, e.g., at microwave and millimeter
wavelengths, to about one part in 109, thus virtually excluding erroneous identi-
fications. In the optical region, a spectral resolution of about one part in 105 is
achieved and this suffices since the density of a spectrum is not as high as in the
millimeter wave region. Most of the hundred or more interstellar molecules
detected to date have been identified in the microwave, millimeter-, and sub-
millimeter-wave part of the electromagnetic spectrum via their characteristic
rotational transition frequencies. An increased number of new identifications will
come forward in the Terahertz region. The interstellar carbon complex, CCC or the
linear molecule C3 ; has recently been identified. Other gas phase molecules have
been detected in the infrared via characteristic vibrational frequencies.
1.4.1 Scope of Interstellar Atomic and Molecular Spectroscopy
Atoms and molecules, both neutral and ionized, exhibit spectral features
throughout most of the electromagnetic spectrum. These spectral lines—in emis-
sion and/or absorption—carry detailed and rather specific information on the
physical and chemical conditions of the sources from which they emanate. In fact,
the heights of the energy levels of the observed transitions together with the degree
of their population furnishes a reliable measure of the excitation conditions and
thus the temperature of the source itself. The relative intensities of various tran-
sitions of a molecule, in conjunction with the principle of statistical equilibrium,
can be used to derive the gaseous number-density of the source.
Since the first discovery of polyatomic molecules in space, the field of inter-
stellar spectroscopy has branched out into many different areas of physics,
astronomy, chemistry, biology, and technology. Despite this astounding diversity,
two major scientific trends dominate the technical developments:
1. The quest for studies of the atomic and molecular distribution on a wide range
of spatial scales from hundreds of degrees down to about 1 arcsec and less. The
aim is to obtain a broad overview of the large-scale galactic distribution of
molecular clouds and a high spatial-resolution picture of the internal cloud
structure with the hope of ultimately resolving the details of star-forming
regions. This wide range of spatial resolution can be achieved by the use of
small and large millimeter- and sub-millimeter-wave single-dish telescopes and
telescope arrays.
2. The extension of the presently available frequency region into the Terahertz
band of the electromagnetic spectrum accessible for interstellar spectroscopic
study from the ‘‘standard’’ 10 cm to 1 mm wavelengths into the sub-millimeter
and far-infrared regions. Multiline studies, probing high excitation conditions
of hot and dense star-forming regions, are behind the strong push to open the
sub-millimeter and far-infrared regions to astronomical observation, from
earth-based, airborne, and satellite missions. The two disciplines, laboratory
and interstellar research, of microwave, millimeter-wave and Terahertz spectral
lines have always been intimately connected.
However, quite independently of the wavelength region, the spectroscopic
history tells us that the fascinating interplay of the research started from the very
beginning of spectroscopy, at the time of Fraunhofer, Bunsen and Kirchhoff (the
wavelength region—OPTICAL LINES). It has been exciting to witness the
expansion toward the IR-region and since 1968, with the knowledge of the exis-
tence of the first polyatomic interstellar molecule, NH3 at 23 GHz, a fascinating
1 History 19
relationship between laboratory development and interstellar progress, determined

in many cases the pace of the research. It was, and still is, equally rewarding to
monitor the technological advances and at the same time follow the conduct of a
complicated observation. These type of measurements were carried through quite
extensively.
One extremely important constituent of molecular clouds are atoms and mol-
ecules in the form of gas. At present, there are more than 150 interstellar molecules
known to exist in various interstellar clouds, i.e., dark and translucent clouds. The
existence of such a diverse chemistry indicates a chemical evolution of cosmic
dimensions taking place. It is fascinating to observe that a large majority (about 70
molecules) of interstellar molecules contains at least one or more carbon atoms per
molecule. This distribution of carbon indicates strongly that the chemical bonding
in interstellar space resembles the covalent bond character known from laboratory
experiment. Interstellar space does support the formation of carbon and its wide
use in the interstellar chemistry. It is the property of the covalent carbon bond to
withstand the very hostile surroundings associated with interstellar clouds.
1.4.2 Terahertz Region: Far–InfraRed Domain
The entire infrared domain covers the wavelengths region from k 1 mm all the
way down to k 1lm (or 300 THz) in four wavelengths intervals. From the long
wavelength’s end to that of the shortest wavelength, they are named as follows:
k 1--0:3 mm sub-millimeter wave (Sub-MM) region,
k 0:3--50 lm far-infrared (FIR) region,
k 50--3 lm mid-infrared (MIR) region,
k 3--1 lm near-infrared (NIR) region.
The two long wavelength intervals, the sub-millimeter, or sometimes called the
radio region, and the FIR regions, are of high interest, for they turn out to be the
Terahertz (THz) band.
The THz window of the electromagnetic spectrum is sandwiched between the
two more established areas—the millimeter-wave band and the infrared region. At
the long wavelength end, the THz window joins the microwave region and
accordingly, at the short end of the THz domain, one finds eventually closing up to
the optical region. Thus, the THz-domain covers in total approximately the same
region as traditionally has been covered for the Far–Infrared, FIR. Although there
is no official agreement known to us, which would define the area exactly, the
frequency gap between 1011 Hz and 5 1012 Hz represents the THz region. It is
advantageous to note that 1 THz (=1,000 GHz) corresponds to an energy of
4 meV or 50 K. The THz region, if considered from the infrared part of the
spectrum, has been referred to traditionally as the far–infrared region, FIR, which
bridges the gap between microwave and optical region.
Being situated between the radio and infrared domains, the THz region bor-
rowed to a large extent its techniques from the established areas. Predominantly, it
relied to a certain extent, but by no means entirely, on the transfer of technology
from the millimeter-wave region. Despite this enormous transfer, the THz region
has become the hunting ground for a rich, scientific prey. But still it remains to be
one of the most poorly explored areas of the entire electromagnetic spectrum. The
access to the field is still not an easy one, and two physically very different reasons
are dominating the main causes for these persistent access difficulties:
1. The first one is directly related to the problems encountered in fabricating the
mm-wave equipment and optimizing its sub-mm and THz usage.
2. The second problem is of rather different origin, and is the frequency-dependent
attenuating nature of the Earth’s atmosphere.
The combined effects of these two causes have been in the past the main
obstacles toward a quick and speedy recovery of the THz region. The general
difficulties, encountered in scaling down in a linear fashion the intrinsic sub-
millimeter wave hardware, remain a notoriously difficult task. But, in principle, the
problems of producing accurate pieces of millimeter-wave hardware are slowly
mastered by the quality outcome of an in-depth extensive technological program.
A variety of laboratories, including the Cologne Laboratories, have taken great
benefit from these measures.
Other important prerequisites are beyond any doubt, the availability of high
gain sub-millimeter-wave telescopes on appropriate mountain sites, such as the
KOSMA 3-m-submillimeter-wave telescope on Gornergrat, in the Swiss central
Alps, or the reliable supply of sub-millimeter-wave radiation sources, together
with the proper combination of receivers, detectors, low-noise amplifiers, and
spectrometers. These fairly expensive custom-made pieces of equipment often do
stretch the resources available to the different groups. It is to be noted here that
commercially only very few millimeter and/or sub-millimeter items can be pur-
chased. Some of the most important sub-millimeter-wave components have to be
produced by the research teams themselves. These preconditions are often too
prohibitively high for one group so as to overstretch the resources available.
In this book, we will consider molecular lines detected by laboratory tech-
niques, whereby THz spectroscopy can employ broad band scanning, i.e., about
50 GHz, and the ultimate frequency resolution achieved with our spectrometers is
around 40 kHz at about 1 THz. These laboratory-recorded line-width can be
compared with the same lines emitted from interstellar sources. Interstellar lines
compare in width quite favorably, provided they emanate from quiescent dark
clouds, where no star formation takes place.
Another important group of lines are in the region beyond 2 THz: the vibra-
tion–rotation lines of low bending vibrational bands have recently been discovered
in the case of CCC in the m2 bending mode at about 2 THz.
Finally, the increasing attenuation of the Earth atmosphere, as one progresses
into the THz region, i.e., [1 THz, has to be added to the general difficulties.
1 History 21
However, judged by some of the recently secured results, the Terahertz region
promises to be the band of plenty. In fact, it seems to hold the key for some of the
very important aspects of physics, chemistry, astrophysics, and of laboratory and
interstellar spectroscopy. The reason for this behavior rests in the basic property of
the Terahertz radiation to match the energy level schemes of a large variety of
applications. The Terahertz spectrum is dominated by light molecules containing
hydrogen. They are known as the hydrides.
1.5 Episodical Appendix
1.5.1 Life Story of Fraunhofer
Joseph von Fraunhofer (1787–1826), German optician and physicist, was born at
Straubing, Bavaria, March 6, 1787.2 The son of a poor glazier, he was apprenticed
at the age of 12 to a lens grinder in Munich. He later went to work at the
Utzschneider Optical Institute near Munich, and rose to become the institute’s
general manager, in 1818. Fraunhofer acquired great skill in the manufacture of
lenses and constructed some of the finest refracting telescopes ever built. He
invented a heliometer and a diffraction grating for measuring the wavelengths of
light. He studied and carefully measured 576 of the dark lines in the solar spec-
trum, previously noticed by W. H. Wollaston. These became to be known as
Fraunhofer lines. Fraunhofer’s studies were cut short by his early death from
tuberculosis at Munich, June 7, 1826.
In all respects, the discovery that the sun and the stars are made up of the same
chemical elements as the Earth is one of the earliest and greatest achievements of
spectroscopy and its first application to astronomical objects. The identification
of the Fraunhofer lines was begun in the years 1859–1861 by the new methods of
spectral analysis introduced by R. W. Bunsen and G. R. Kirchhoff. This correlation
marked the time and place of the establishment of spectroscopy as an analytic tool
with an unparalleled power to study both physical and chemical processes in the
laboratory and in cosmic dimensions. The power of this new tool ‘‘spectroscopy’’
and its potential for future applications were recognized by Kirchhoff and Bunsen,
who stated that it would open ‘‘an entirely untrodden field, stretching far beyond
the limits of the Earth, or even of our solar system’’ and who noted the ‘‘almost
inconceivable’’ sensitivity of spectroscopy that would lead to the detection of
hitherto unknown elements. Indeed many new elements, such Cs, Rb, and He,
were subsequently detected by their atomic spectra, in the laboratory, and in stars.
In 1914, Martin Schwarzschild tried to explain the dark Fraunhofer lines in terms
of widths and shapes with the idea to derive from the stellar lines the structure and
2
Fraunhofer Gesellschaft, München, Germany, provides a biographic document for Fraunhofer:
http://www.fraunhofer.de/en/about-fraunhofer/joseph-von-fraunhofer/.
evolution of the stars. But a more detailed understanding of the spectra and their
associated energy level diagrams had to await the advent of quantum mechanics in
the 1920s and its application to molecular spectra. These efforts in obtaining
extremely precise data on stars and derived from the observed spectra of 4,000
stars were classified into four spectral types. Finally, all this study culminated in
the famous Harvard classification of stars catalog. Over a time-span of several tens
of years the Henry–Draper Catalog which contains the spectral type of about one
quarter of a million stars. The basic result of this enormous experimental data is
that the Harvard sequence of stars is essentially arranged according to their
temperatures.
1.5.2 Life Story of Herschel
Friedrich Wilhelm Herschel (1738–1822), generally known as Sir William

Herschel, one of the most illustrious of astronomers, was born at Hannover,
November 15, 1738.3 His father was a musician employed as hautboy player in the
Hanoverian guards. The family had immigrated from Bohemia in the early part of
the seventeenth century, on account of religious troubles, they themselves being
Protestants. Herschel’s earlier education was necessarily of a very limited char-
acter, chiefly owing to the troubles in which his country at that time was involved;
but, being at all times an indomitable student, he, by his own exertions, more than
repaired this deficiency of his youth. He became a very skilful musician, both
theoretical and practical; while his attainments as a self-taught mathematician
were fully adequate to the prosecution of those branches of astronomy which, by
his labors and his genius, he so eminently advanced and adorned. Whatever he did,
he did methodically and thoroughly; and in this methodical thoroughness, lay the
secret of what Dominique Arago very properly terms his ‘‘astonishing scientific
success.’’
In 1755, at the age of 17, he joined the band of the Hanoverian guards, and with
his detachment visited England accompanied by his father and eldest brother. In
the following year, he returned to his native country, but two years later, impelled
by the troubles that surrounded him, he finally quit Hannover to seek his fortunes
in England. As might have been expected, the earlier part of his career in his
adopted country was attended with formidable difficulties and much privation. We
find him engaged in several towns in the north of England as an organist and a
teacher of music, but these were occupations not attended with any lucrative
results. In 1766, the tide of his fortunes began to flow, inasmuch as he obtained the
appointment of organist to the Octagon Chapel in Bath, at that time the resort of
the wealth and fashion of the city, and of its numerous distinguished visitors.
3
A few biographic reports have been published for Herschell, e.g., Millman [45].
1 History 23
In the year 1781, Herschel made a discovery which, as we shall see, soon
completely altered the character of his professional life. In the course of his
systematic examination of the heavens with a view to the discovery of the plan of
their construction, he lighted on an object which at first he supposed to be a comet,
but which, by its subsequent motions and appearance, turned out to be a new
planet, moving outside of the orbit of Saturn. To this planet, he in due time
assigned the name of Georgium Sidus; but this name has by general consent been
laid aside in favor of Uranus.
In the year 1782, in consequence of his fame, Herschel was invited to Windsor
by George III, and then accepted the offer made by the king to become his private
astronomer, and thenceforth devoted himself wholly to a scientific career. The
salary offered and accepted was £ 200 per annum, to which an addition of £ 50 per
annum was subsequently made for the astronomical assistance of his faithful sister.
Dr. Watson, to whom alone the amount of this salary was mentioned, made the
natural remark, ‘‘Never before was honor purchased by a monarch at so cheap a
rate.’’ In this way, the great philosophical astronomer was shifted from Bath first to
Datchet and soon afterward permanently to Slough, within easy access of his royal
patron at Windsor.
It was at Slough in 1783 that he wrote his first memorable article on the
‘‘Motion of the Solar System in Space,’’ a sublime speculation, yet through his
genius realized by consideration of the utmost simplicity. He returned to the
same subject with full details in 1805. It was also after his shift to Slough that he
published his first memoir containing his speculative ideas on the Construction
of the Heavens, which from the beginning had been the chief aim of his tools,
both of mind and body. In a long series of remarkable articles, addressed as
usual to the Royal Society, and extending from the year 1784 to 1818, when he
was 80 years age, he demonstrated the fact that our sun is a star situated not far
from the bifurcation of the Milky Way, and that all the stars visible to us lie
more or less in clusters scattered throughout a comparatively thin stratum, but in
the other two dimensions stretching immeasurably further into space. At one
time, he imagined that his powerful instruments had pierced through this stellar
stratum of the Milky Way, and that he had approximately determined the form
of some of its boundaries. In the last of his memoirs, he had convinced himself
of his error, and he admitted that to his telescopes our stratum of stars is
‘‘fathomless.’’ Over this stratum of stars and their planetary attendants, the whole
being in ceaseless motion round some common center of gravity as the resultant
point of the combined gravitation, Herschel discovered on either side a canopy
of discrete nebulous masses, from the condensation of at least a part of which
the whole stellar universe had been formed, a magnificent conception, pursued
with a force of genius and put to the practical test of observation with an
industry almost incredible. It was the work of a single mind, carried to its
termination with no assistance beyond that of a loyal sister, almost as remarkable
a personage as himself.
Hitherto, we have said nothing about the grand reflecting telescope, of 40 foot
focal length and 4 foot aperture, which is too often regarded as the chief effort of
his genius and his perseverance. The full description of this celebrated instru-
ment will be found in the 85th volume of the Transactions of the Royal Society.
Gigantic as it really was, we are disposed to regard it as among the least of his
great works. On the day it was finished (August 28, 1789), Herschel saw at the
first view, in a grandeur not witnessed before, the Saturnian system with all its
six satellites, (Pan, Atlas, Pandora, Prometeus, Epimetheus, Janus, and Mimas)
five of which had been discovered long before by Huygens or Cassini, while the
sixth, subsequently named Enceladus, which he had, two years before, sighted by
glimpses in his exquisite little telescope of 6:5 inches aperture, but now saw
in unmistakable brightness with the towering giant he had just completed.
On September 17, he discovered a seventh, which proved to be the nearest of all
the satellites of Saturn. It has since that time received the name of Mimas. It is
somewhat remarkable that, notwithstanding his long and repeated scrutinies
of this planet, the eighth satellite Hyperion, and the crape ring should have
escaped him.
Herschel married the widow of Mr. John Pitt, a wealthy London merchant, on
May 8, 1788, by whom he had only one son, John Frederick William. The prince
regent conferred a Hanoverian knighthood upon him in 1816. But a far more value
and less tardy distinction was the Copley medal assigned to him by his associates
in the Royal Society in 1781. He died at Slough on August 25, 1822, in the eighty-
fourth year of his age, and was buried under the tower of St. Laurence Church,
Upton, within a few hundred yards of the old site of the 40-foot telescope. A mural
tablet on the wall of the church bears a Latin inscription from the pen of the late
Dr. Goodall, provost of Eton College. A collection edition of his astronomical
memoirs would speak of his genius in unmistakable language; but this has not been
published.
Herschel was an indefatigable worker. For many years, he made nightly sys-
tematic surveys of the sky, examining planets, stars, nebulae, and star clusters. By
day he worked over his results or directed the construction of his telescopes. His
sister Caroline was his assistant and became an independent worker of some
standing. Herschel observed sun-spots, confirmed the gaseous nature of the sun,
rediscovered the Martian polar caps, and believed in the possibility of life on Mars.
He announced the trade-wind theory of Jupiter’s belts and discovered two of
Jupiter’s and two of Uranus’ Moons. His main purpose, he said, was a to gain
‘‘knowledge of the Construction of the Heavens.’’ His principal study was on stars,
and it was in this field that he made two discoveries of primary importance: the
movement of the solar system through space, and the evidence that binary stars
move around a common center of gravity. This evidence substantiated the
universality of Newton’s theory of gravitation. Herschel discovered nearly 1,000
double stars. In 1785, he brought out the disk theory of the stellar system, holding
that all nebulae are clusters of stars, which he called island universes. He aban-
doned the idea of the Milky Way as a collection of stars uniformly distributed and
came to believe that it was composed of local groups and clusters. The nebulae, he
thought, were all clusters not yet resolved into stars. Herschel also discovered
nebulae and clusters numbering from 2,000 to 3,000 on which he published catalog
1 History 25
reports. After years of profound research, he brought out an important article on

the construction of the heavens, theorizing on the evolution of matter by succes-
sive steps from nebulous material to the final planetary state, by means of gravi-
tation. Herschel died at Slough, England, August 25, 1822.
References
1. E.E. Barnard, E.B. Frost, A Photographic Atlas of Selected Regions of the Milky Way
(Carnegie Institution of Washington, Washington, 1927)
2. G.B. Donati, Mon. Not. R. Astron. Soc. 25, 31 (1864)
3. W. Huggins, Proc. R. Soc. Lond. 34, 148 (1882–1883)
4. A.E. Douglas, Astrophys. J. 114, 446 (1951)
5. L. Spitzer Jr., E.G. Zweibel, Astrophys. J. 191, L127 (1974)
6. L. Spitzer, Physical Process in the Interstellar Medium (Wiley-Interscience, New York,
1978)
7. H.I. Ewen, E.M. Purcell, Nature 168, 356 (1951)
8. C.A. Muller, J.H. Oort, Nature 168, 357 (1951)
9. L. Essen, R.W. Donaldson, M.J. Bangham, E.G. Hope, Nature 229, 110 (1971)
10. T.L. Wilson, L. Filges, in Proceedings of IAU Colloq, vol. 125, ed by M.A. Godon,
R.L. Sorochenko (Kluwer, Dordrecht, 1990)
11. G.L. Verschuur, K.I. Kellermann, Galactic and Extra-Galactic Radio Astronomy (Springer,
Berlin, 1988)
12. G. Melnick, G.E. Gull, M. Harwit, Astrophys. J. 227, L29 (1979)
13. J.W.V. Storey, D.M. Watson, C.H. Townes, Astrophys. J. 233, 109 (1979)
14. R.W. Russel, G. Melnick, G.E. Gull, M. Harwit, Astrophys. J. 240, L99 (1980)
15. M.K. Crawford, R. Genzel, C.H. Townes, D.M. Watson, Astrophys. J. 291, 755 (1985)
16. R. Genzel, A.I. Harris, J. Stutzki, D.T. Jaffe, Astrophys. J. 332, 1049 (1988)
17. A. Dalgarno, R.A. McCray, Annu. Rev. Astron. Astrophys. 10, 375 (1972)
18. G.J. Stacy, N. Geis, R. Genzel, J.B. Lugten, A. Poglitsch, A. Sternberg, Astrophys. J. 373,
L99 (1980)
19. S.C. Madden, R. Genzel, F. Herrmann, A. Poglitsch, N. Geis, C.H. Townes, G.J. Stacy, Bull.
Am. Astron. Soc. 24, 1268 (1992)
20. A.L. Cooksy, R.J. Saykally, J.M. Brown, K.M. Evenson, Astrophys. J. 309, 828 (1986)
21. A. Omont, Rev. Mod. Astron. 1, 102 (1988)
22. W.D. Watson, Rev. Mod. Phys. 48, 513 (1976)
23. G. Winnewisser, E. Herbst, Top. Curr. Chem. 139, 119 (1987)
24. G. Winnewisser, E. Herbst, H. Ungerechts, in Spectroscopy of the Earth’s atmosphere and
interstellar medium, ed. by K.N. Rao, A. Weber (Academic Press, Boston, 1992)
25. T. Dunhum Jr., W.S. Adams, Publ. Astron. Soc. Pac. 49, 26 (1937)
26. P. Swings, L. Rosenfeld, Astrophys. J. 86, 483 (1937)
27. A.E. Douglas, G. Herzberg, Astrophys. J. 94, 483 (1941)
28. G. Winnewisser, J.T. Armstrong (eds.), in The Physcis and Chemistry of Interstellar
Molecular Clouds. Lecture Note in Physics 331. (Springer, Berlin, 1989)
29. G. Winnewisser, E. Herbst, Rep. Prog. Phys. 56, 1209 (1993)
30. E. Herbst, Annu. Rev. Phys. Chem. 46, 27 (1995)
31. E.F. van Dishoeck, Faraday Discuss. 109, 31 (1998)
32. K.W. Hinkle, J.J. Keady, P.F. Bernath, Science 241, 1319 (1988)
33. P.F. Bernath, K.W. Hinkle, J.J. Keady, Science 244, 562 (1989)
34. A. van Orden, J.D. Cruzan, R.A. Provencal, T.F. Giesen, R.J. Saykally, in Airborne Astronomy
Symposium on the Galactic Ecosystem, ASP Conference Series, vol. 73 (1995), p. 67
35. M. Tuley, D.A. Kirkwood, M. Pachkov, J.P. Maier, Astrophys. J. 506, L69 (1998)
36. S. Yamamoto, S. Saito, Astro. Phys. J. 370, L103 (1991)
37. T.G. Phillips, P.J. Huggins, Astrophys. J. 251, 533 (1981)
38. J. Keene, G.A. Blake, T.G. Phillips, P.J. Huggins, C.A. Beichman, Astrophys. J. 299, 967
(1985)
39. J. Hernichel, D. Krause, R. Roehrig, J. Stutzki, G. Winnewisser, Astron. Astrophys. 259, L77
(1992)
40. A.G.G.M. Tielens, D. Hollenbach, Astrophys. J. 291, 722, 747 (1985)
41. A. Sternberg, A. Dalgarno, Astrophys. J. 338, 197 (1989)
42. D.J. Hollenbach, T. Takahashi, A.G.G.M. Tielens, Astrophys. J. 377, 192 (1991)
43. C.W. Allen, Astrophysical Quantities (Athlone Press, London, 1973)
44. B.D. Savage, J.S. Mathis, Annu. Rev. Astron. Astrophys. 17, 73 (1979)
45. P.M. Millman, J. R. Astron. Soc. Can. 74, 134 (1980)
46. G. Winnewisser, C. Kramer, Space Sci. Rev. 90, 181 (1999)
Chapter 2
Molecules Detected in Interstellar Space
Karl M. Menten and Friedrich Wyrowski
In this section, an overview is given of molecules detected in the interstellar

medium and circumstellar envelopes. Given that so far more than 150 species have
been observed in interstellar space, this overview can by no means be complete.
Therefore, we focus on the detections of the first molecules and then give
examples of prototypical molecules and recent new detections. This section is
based on the Menten (2003) article presented in the proceedings of the 2003
Zermatt conference [1].
2.1 Detection of the First Interstellar Molecules
The first optical detections of the diatomic radicals CH and CN and the CHþ ion
were already described in Sect. 1.3.1. Here, the first radio detections are discussed.
The late 1960s and early 1970s were an era of great excitement in all branches
of radio astronomy. In spectral line research, 12 years after the detection of the
first line, the 21 cm HI hyperfine structure (hfs) transition [2], finally, the first
molecule was found: the hydroxyl (OH) radical [3].
The OH K-doublet transitions, as well as many other lines at m-, cm- and mm-
wavelengths from a variety of molecules, had been identified as possible astro-
nomical targets in a visionary article by Charles Townes [4].1
K. M. Menten and F. Wyrowski (&)

Max-Planck-Institut f. Radioastronomie, Auf dem Hügel 69, 53121 Bonn, Germany
e-mail: wyrowski@mpifr-bonn.mpg.de
1
see also Shklovsky [5].

Springer-Verlag Berlin Heidelberg 2011
28 K. M. Menten and F. Wyrowski
Shortly after the detection of OH, Hoglund and Mezger [6] discovered the first
radio recombination line, H109a at 5009 MHz. Molecular radio astronomy started
in earnest in 1968/1969, with the discoveries of ammonia (NH3 ; Cheung et al. [7])
and water vapor (H2 O; Cheung et al. [8]) by a group around Townes,2 as well as
the first complex organic species, formaldehyde (H2 CO; Snyder et al. [9]),
breaking loose a real gold rush that has yet to stop. Even today, approximately 2–4
‘‘new’’ interstellar or circumstellar3 molecules are found each year, totaling to
more than 150 species (not counting isotopomers).4
The early history of ‘‘molecule-hunting’’ is well documented in the review by
Snyder (1972; ‘‘Molecules in Space’’ [10]), which gives a very useful introduction
to molecular radio astronomy, that is still very much readable today. In the same
volume, the review, following Snyder’s by Winnewisser et al. and presents an
equally excellent overview of the laboratory side of ‘‘Millimetre Wave
Spectroscopy’’ [11]
As shown in Table 2.1,5 many of the early molecular detections were made at
low radio frequencies (and toward Sagittarius B2). For a full list of interstellar
molecules detected in space, see Chap. 5.
Almost concurrently with cm-wavelength molecular radio astronomy, shorter
wavelength exploration started. Driven by new high-frequency receiver technol-
ogy and, generally, higher line intensities and a much higher ‘‘line density’’
(i.e., lines per GHz) the ‘‘millimeter explosion’’ started with the discoveries of CO
by Wilson et al. [35] and HCOþ (‘‘X-ogen’’) by Buhl and Snyder [36] both in
1970 (the latter published in 1971), and continues through today, having reached
the submillimeter range in the 1980s.
After the millimeter window was opened, (most) people spontaneously lost
interest in the cm-range because
• Most molecules have the bulk or the entirety of their rotational spectrum at
(sub)millimeter wavelengths.
• Almost all ‘‘simple’’ (di- or triatomic), light molecules have (sub)mm or far-IR
(FIR) rotational spectra.
• Many of the complex molecules were (for a long time) only found in Sgr B2.
Interest remained high in special cm-wavelength lines, including the NH3
inversion transitions near 24 GHz, a number of prominent maser lines from OH
(hfs, 1.7 GHz and higher frequencies), H2O (22.2 GHz), CH3 OH (25, 12.2, and
2
Both lines were on Townes’ (1957) list [4]!
3
Circumstellar molecules exist in the envelopes of evolved giant stars. Many species are
detected there which are also found in interstellar clouds, whereas some others, have so far only
been detected in a circumstellar environment.
4
See http://www.ph1.uni-koeln.de/vorhersagen/molecules/main_molecules.html.
5
The frequencies are taken from the JPL catalog [33]; see http://spec.jpl.nasa.gov/ or the
Cologne Database for Molecular Spectroscopy (CDMS) [34]; see http://www.ph1.
uni-koeln.de/vorhersagen/molecules/main_molecules.html. Numbers in parentheses are uncer-
tainties in the last significant digit(s).
2 Molecules Detected in Interstellar Space 29
Table 2.1 The first interstellar ‘‘Radio’’ molecules

Name Species Transition Frequency (MHz) Reference
Hydroxyl OH 2
P32 ; J ¼ 32 1665:4018ð2Þa [3]
Ammonia NH3 (1, 1) b [7]
23694:4955ð1Þ
Formaldehyde H2 CO 111 110 4829:660ð1Þb [9]
Water H2 O 616 523 22235:08ð2Þb [8]
Methyl alcohol CH3 OH 1þ 1 834.267(2) [12]
Formic acid HCOOH 111 110 1638.804(1) [13]
211 212 4916.319(3) [14]
Formamide NH2 CHO 111 111 1539:58ð2Þc [15]
211 212 4618:55ð2Þc [16]
Cyanoacetylene HC3 N 1-0 9098:6537ð3Þc [17]
H13 C3 N 1-0 9060d [18]
H12þ13 C3 N 1-0 9000d [19]
Acetaldehyde CH3 CHO 110 111 1065.075(5) [20]
Methylidine CH 2
P12 ; J ¼ 12 3263:788ð10Þe [21]
Thioformaldehyde H2 CS 211 212 3139.38(3) [22]
Methanimine CH2 NH 110 111 5290.31(4) [23]
Methylamine CH2 NH2 202 110 8776:2ð1Þf [24]
Vinyl cyanide H2 CCHCN 211 212 1371.8262(1) [25]
Methyl formate HCOOCH3 -A 110 111 1610.2445(7) [26]
HCOOCH3 -E 110 111 1610.91(10) [27]
Dimethyl ether (CH3 Þ2 O 202 111 9119:668ð15Þc [28]
Cyanodiacetylene HC5 N 4-3 10650.650(4) [29]
2-1 5325.328(2) [30]
Only those molecules first detected at cm-wavelengths are listed. For early detections in the mm-
range, see Snyder [10]. All molecules were first detected in the transitions listed, except for
dimethyl either which was discovered in its 606 515 line at 90938.10(3) MHz by Snyder et al.
[31]. No detections for isotopomers are listed, except for HC3 N
a
Apart from the F = 1 - 1 line listed, also the 2 - 2 line at 1667.3590(2) MHz was reported in
the detection paper
b
Line with hfs, intensity-weighted frequency is given
c
Line with hfs, average of the lowest and the highest hfs line frequency, is given
d
Various lines with differently placed 13 C between 8815 and 9100 MHz
e
Apart from the F = 0 - 1 line listed, also the F = 1 - 1 and F = 1 - 0 lines at 3335.475(10)
and 3349.185(10) MHz, resp., were reported in the detection paper (frequencies from Rydbeck
et al. [32])
f
Frequency from Fourikis et al. [24]
6.7 GHz), as well as the CH hfs lines (3.3 GHz). All of these, except for CH3 OH,
have all or the bulk of their rotational spectra at submillimeter and far-infrared
wavelengths.
Considering the mm-wavelength data coming in, soon after the radio obser-
vations, it became clear that the emission from many low-frequency, low-J lines of
complex molecules toward Sgr B2 must be (weakly) inverted. Morris et al. [37]
pointed out that this can be understood if the dipole collision is not the dominant
factor of the collisional process; in such cases, population inversion for the J = 0
and 1 levels (at least for linear or near-linear species) may occur easily over a wide
range of physical conditions (e.g., [38]). This is borne out by their statistical
equilibrium calculations.
After the first complex molecules were found in Sgr B2, this source turned into
a Bonanza: The described inversion mechanism together with a relatively strong,
extended continuum background ‘‘boosted’’ many otherwise undetectable lines
into ‘‘observability’’. As a consequence, the vast majority of interstellar molecules
was and still is found in Sgr B2, many of the most complex ones only there. All of
the molecules listed in Table 2.1 are found in this source.
2.2 The Habitat of Interstellar Molecules
The by-far most abundant interstellar molecule is molecular hydrogen, H2 . Since

H2 , because of its symmetry, does not have an electric dipole moment, only weak
electric quadrupole transitions between rotational or ro-vibrational levels can be
observed. Therefore, low rotational transitions of the CO molecule, although about
by 10-4 less abundant than H2 , have become the most important tool to study the
large scale distribution of the molecular gas in the interstellar medium. Most of the
molecular gas can be found in giant molecular clouds such as the region shown in
Fig. 2.1. These clouds have diameters of 10–100 pc, masses of 1056:5 M , and
mean densities of several 100 cm3 , but they are strongly clumped (e.g., [39]). The
higher density clumps are probed in Fig. 2.1 by submillimeter dust continuum
emission and are most closely associated with star formation. Giant molecular
clouds are distributed along the Galactic plane with a small scale height of only
100 pc. Although they occupy only a relatively small volume of the interstellar
medium, they contain a large fraction of its mass.
In addition, molecules can also be found in the so-called diffuse clouds, which
are more tenuous and fully exposed to UV radiation from stars (see the review by
Snow and McCall [41]). In fact, the first detections of molecules in space have
been done by the observations of sharp optical absorption lines in diffuse clouds.
Molecules can even be found in circumstellar envelopes which are due to strong
mass loss of stars in their late stages of evolution. The central stars in this stage
have a low temperature so that molecules and dust can be formed and fueled into
the interstellar medium.
2.3 Metal-Bearing Molecules in the Interstellar Medium
Observations of species containing metals, as well as other refractory species such

as SiO, deliver important information on depletion into dust grains. Moreover,
when released either from sputtering of the grains or their mantles by violent
Fig. 2.1 The giant molecular cloud G29.9 seen in 13 CO (1–0) emission (grayscale) from the
Galactic ring survey by Jackson et al. [40] overlaid with archival dust continuum observations
from SCUBA at the JCMT
dynamics, such as protostellar outflows, they can be used as probes of these

phenomena (see, e.g., Bachiller [42]).
Todate, four diatomic, non-H-bearing metal compounds (NaCl, KCl, AlCl, and
AlF and three triatomic ones (MgNC, MgCN, NaCN, and AlNC) have been
detected in the dense envelope of the carbon star IRC+10216 [43–49]. MgNC,
NaCN, AlF, and NaCl were also found, recently, toward the proto-planetary
nebula (PN) CRL 2688 ([49–51]; see Petrie [52] for a discussion of other candidate
species for detection). Still, only one metal-containing molecule, FeO, has been
detected in the interstellar medium, toward Sgr B2 [53].
As regards metal hydrides, radio and (sub)millimeter wavelength searches for
interstellar NaH [54] and LiH6 have so far been unsuccessful, as was a search for
MgH in IRC+10216’s envelope [56]. Optical wavelength absorption from CaH,
AlH, FeH, and MgH has been found in cool stars and the Sun: MgH even in the
integrated light of external galaxies (see, e.g., [57]).
6
The ground state J = 1 - 0 transition of LiH is near 444 GHz, an atmospherically very
unfavorable frequency. A tentative detection of this line was made in the redshifted (z = 0.68)
dense absorbing cloud toward the B0218+357 gravitational lens system [55].
2.4 Light Hydrides
The history of interstellar hydrides begins with the detection by Dunhum [58] of
three optical absorption lines at 3957.7, 4232.6, and 4300.3 Å from diffuse inter-
stellar clouds [58], which subsequently were found to arise from CH (4300.3 Å;
McKellar [59]; suggested by Swings and Rosenfeld [60]) and CHþ (the former
two; [61]).
It has been thought for a long time that hydrides might play an important rôle in
cooling the highest density regions of interstellar clouds. Hydrides have, first, large
level spacings, allowing them to emit relatively a large amount of energy per
quantum. Second, their large dipole moments require high critical densities for
their excitation. Thus, their rotational levels are only excited in high-density
regions with n [ 106 cm3 , making them effective coolants in this regime, where
even rare CO isotopes are so optically thick that CO cooling becomes negligible
compared to that of (the hydride) H2 O.
This regime was excluded in the classical study by Goldsmith and Langer [62]
on ‘‘Molecular Cooling and Thermal Balance in Dense Molecular Clouds’’, but
was re-addressed by Neufeld et al. [63], who found that for elevated temperatures,
T ¼ 100K; H2 O cooling dominates, and the hydrides’ contribution is comparable
to that of all other molecules combined, i.e., &30% of the former.
Recent detections of hydride rotational transitions were made by the Caltech
Submillimeter Observatory (CSO), namely H3 Oþ (JK ¼ 1 þ þ þ
1 21 ; 32 22 ; 30
20 =307; 365; 396 GHz, [64, 65]), SiH ( P12 ; J ¼ 2 2/625, 628 GHz; [66]),7 HCl
2 3 1
(J = 1 - 0/626 GHz; [67]), H2 O; H18 2 O (110 101 =557; 548GHz; [68]), HDO (the
101 000 /465 GHz and 111 000 /894 GHz ground-state transitions; [69, 70]),8 and
NH2 (J ¼ 32 32; 12 12; 32 12/462, 459, 461 GHz); [71]). H2 Dþ (110 111 =372
GHz), the deuterated isotope of the key Hþ 3 molecule was discovered with the
9
James–Clerck–Maxwell telescope [72]. Observations using the Long Wavelength
Spectrometer aboard the Infrared Space Observatory led to the discoveries of HF
(J = 2 - 1/2.47 THz; [74]), HD (J = 1 - 0/2.68 THz and 2 - 1/5.36 THz;
[75]), plus various H2 O; OH, and CH rotational and some H3 Oþ inversion transi-
tions (many toward Sgr B2; [76] and in the proceedings [1]).
A preliminary conclusion is that (for the only two sources observed, Orion-KL
and Sgr B2) the measured HCl and HF abundances indicate that only a low
percentage of all chlorine is in these molecules, suggesting significant depletion. If
this should hold true for other hydrides as well, then their cooling contribution
7
Tentative detection.
8
A number of higher excitation cm and mm-wavelength H2 O; H182 O, and HDO transitions have
also been found.
9
Hþ
3 itself was finally detected, after many arduous attempts, in absorption at near-infrared
wavelengths by Geballe and Oka [73], benefitting from a monumental body of laboratory
research, mainly gathered by Oka and collaborators (see the proceedings [1]).
Fig. 2.2 New hydrides,

observations using APEX
toward Sgr B2(M) (the upper
two panels show IRAM 30-m
detections for comparison.
The gray lines show synthetic
spectra based on the velocity
components seen in the C3 H2
spectrum
could be much smaller than that discussed above. More sensitive observations of
more transitions toward more lines of sight are needed and are being observed now
using the APEX telescope and at the Herschel Space Observatory.10
Using the APEX telescope in Chile on the 5000-m high Chajnantor plateau,
several hydrides (SHþ ;13 CHþ ; OHþ , and HCl) have recently been studied from
the ground by Menten et al. [77] and Wyrowski et al. [78] in the absorbing diffuse
clouds towards Sgr B2(M). SHþ and OHþ were new interstellar detections, and the
detected absorption of a continuous velocity range on the line-of-sight showed
these molecules to be an abundant component of diffuse clouds (see Fig. 2.2).
10
http://astro.estec.esa.nl/SA-general/Projects/First/first.html.
With Herschel, also the HF ground-state transition has been detected in

absorption (Neufeld et al., [79]), which shows that HF will be the dominant res-
ervoir of interstellar flourine. From Herschel observations also, many detections of
H2 Oþ are reported, which, compared to previous model predictions, turned out to
have higher abundances in the low-density gas of diffuse clouds. Another new
detection by Herschel is interstellar chloronium (H2 Clþ , Lis et al., [80]), again
with surprisingly high abundances.
2.5 Ions
2.5.1 Cation
In 1970, Buhl & Snyder found an unidentified interstellar line during their search
for the HCN (1–0) line at 86.3 GHz and labeled it ‘‘X-ogen’’ [81]. Even before this
article was published in Nature, Klemperer [82] proposed HCOþ as the carrier of
the X-ogen line, but only several years later, this claim could be confirmed by
laboratory measurements and interstellar observations of H13 COþ . This first
detection of an interstellar ion, triggered studies on ion–molecule chemistry as the
basis for the formation of interstellar molecules in the gas phase (e.g., [83]). In
dense molecular clouds, the degree of ionization is determined by cosmic ray
ionization of molecular hydrogen which further reacts with H2 to form Hþ 3 . The
Hþ3 ion, which plays a key role in driving the chemistry in molecular clouds, was
only detected in 1996 by Geballe and Oka by infrared absorption of three rotation–
vibration transitions [73]. About a dozen cations have so far been detected and,
only recently, several anions as well.
2.5.2 Anions
In 1995, Kawaguchi et al. [84] reported in their centimeter spectral line survey of
the circumstellar envelope of IRC+10216 the detection of a series of seven lines
showing a spectral pattern of a linear molecule with a rotational constant of
1377 MHz. Only 11 years later, McCarthy et al. [85] could attribute with new
laboratory measurements these lines to the first anion detected in interstellar space,
C6 H . In addition, the lines from the ions C4 H ; C8 H , and C3 N (and possibly
C5 N ) have been measured in the meantime as well, and these molecules are now
also observed in dark clouds and star forming regions. Very recently, even CN-
was detected in the carbon star envelope IRC+10216 (Agundez et al., [86]). While
based on ion-chemistry models, negative ions have been predicted many years ago,
the new observations have now triggered a new set of chemical models with new
mechanisms for anion formation. A nice review is given by Herbst [87].
2.6 Chain Molecules
Besides simple linear molecules, such as CO and HCN, much larger linear mol-
ecules, chain molecules, have been detected over the years mostly in cold dark
clouds, but also in the envelopes of late-type stars and proto-planetary nebula. A
considerable amount of carbon can be locked up in these chains, with the prime
examples being chains such as HCn N; Cn H, and Cn S. The recent detections of
negatively charged chains have been discussed in the previous section. These
carbon chain molecules are highly unsaturated. The current longest detected chain
is HC11 N which was detected toward the cold Taurus Molecular Cloud 1 (TMC-1).
Laboratory frequencies of even longer chains are available. The longest chain
detected in circumstellar envelopes is HC9 N. Carbon chain molecules might be
related to the formation and destruction of carbonaceous compounds such as
polyaromatic hydrocarbons (PAHs, see also the section about cyclic molecules).
2.7 Symmetric and Asymmetric Top Molecules
Symmetric rotors are being used as interstellar thermometers. Two of their

moments of inertia are equal; therefore, they can be described by only two rota-
tional quantum numbers, the total angular momentum, J, and its component along
the symmetry axis, K. Radiative transitions are only possible within the so-called
K ladder. For symmetric tops, such as CH3 CN and CH3 CCH, the line frequencies
of transitions with the same J but different K have only small separations, and lines
from the complete K ladder can therefore be easily observed simultaneously. The
level population within a given K ladder is mostly a function of the temperature of
the emitting clouds—therefore, being in local thermodynamic equilibrium—
Boltzmann plots can be used to estimate the temperature of the clouds as shown in
Fig. 2.3.
Fig. 2.3 IRAM 30-m telescope observations of the CH3 CCH J ¼ 5 4K ladder in a massive
star forming clump. The temperature is estimated from the Boltzmann plot in the right panel
As already mentioned in Sect. 2.1, the first symmetric rotor detected in space
was ammonia, NH3 , since ammonia has observable lines in the centimeter
wavelength range. These lines are due to the tunneling of the nitrogen through the
plane of the hydrogen atoms and are called inversion lines. This effect causes a
small splitting in energy of the rotational levels, and the so-called metastable
inversion lines with J = K can be observed in a relatively small frequency range
around 24 GHz. An excellent review of ammonia is given in Ho and Townes [88].
Rotational transitions of ammonia were first observed in its lowest vibrationally
excited state by Schilke et al. [89]. Rotational lines in the vibrational ground state
cannot be observed from the ground. The first detection of the (J, K) (1, 0) –(0, 0)
line at 572 GHz was done with the Kuiper Airborne Observatory by Keene et al.
[90].
Slightly asymmetric rotors offer, in addition to their dependence on the clouds’
temperatures, the opportunity to probe the density of the molecular clouds.
Important examples of this class of molecules that are often used in astronomical
investigations are methanol (CH3 OH) and formaldehyde (H2 CO). To determine
the physical parameters T and n of a molecular cloud, a range of transitions with
different dependence on these parameters is observed.
The first asymmetric rotor detected in interstellar space was water, although it is
very difficult to observe thermal water lines from the ground due to the large
amount of water vapor in our atmosphere. Thermal lines of water have been
studied with space observatories. The Infrared Space Observatory (ISO) made
important detections of water. Its short-wavelength spectrometer (2–45 lm) cov-
ered mostly ro-vibrational lines, while its long-wavelength spectrometer
(43–194 lm) covered pure rotational transitions, both with crude spectral reso-
lution though. Velocity-resolved observations of the water ortho ground state line
at 557 GHz have been conducted with the small space observatories, Odin and
SWAS. A great progress in our understanding of interstellar water is currently
being made with the Herschel Space Observatory, which allows to make velocity-
resolved observations of many rotational lines of water with much higher spatial
resolution than with the previous space observatories.
2.8 Cyclic Molecules
The first cyclic molecule detected in space was SiC2 in the carbon star IRC+10216
observed by Thaddeus et al. [91] followed by the first truly hydrocarbon ring
molecule C3 H2 [92], which subsequently was extensively studied in many inter-
stellar clouds. After these detections, only a few new ones joined the group of
interstellar-detected cyclic molecules, namely, C3 H and C2 H4 O, and in 2001,
Benzene (C6 H6 ; Cernicharo et al. [93]) was detected in the infrared with ISO.
Polycyclic aromatic hydrocarbons (PAHs) can be considered as very big cyclic
molecules, but there relation to their smaller siblings and the formation of PAHs
are still not understood. For a recent review on PAHs, see Tielens [94].
2.9 Complex Molecules
A comprehensive review of complex organic interstellar molecules has recently

been given by Herbst and van Dishoeck [95]. By interstellar standards, molecules
containing six or more molecules are viewed as complex molecules. With this
definition, about one third of the molecules detected in the interstellar medium or
circumstellar shells are complex molecules. In the following, we give examples of
the most complex molecules so far observed in interstellar space.
In 2003, Kuan et al. claimed to have detected the simplest amino acid, glycine
(H2 NCH2 COOH) [96] in three warm, dense, interstellar clouds, including Sgr B2.
If true, this would be the conclusion of a long quest (see, e.g., [97–100]).
Any such ‘‘identification’’ must rely on a large number of lines. Even so (or
maybe because of this), the confusion problem is a formidable one. Inspecting any
figure presented by Kuan et al., one finds comparably many glycine lines as
unidentified (‘‘U’’) lines. For a wider view, see the 218–263 GHz line survey by
Nummelin et al. [101, 102], which, for Sgr B2(N), is at or near the confusion limit
throughout that band, even at their relatively moderate sensitivity (compared to the
interferometric data by Snyder and his collaborators).
In the high-density (n 107 cm3 ) Sgr B2, ‘‘Large Molecule Heimat (LMH)’’
level populations are in Local Thermodynamic Equilibrium (LTE) characterized
by rotation temperatures, Trot , around 200 K [103]. At this temperature, the lower
vibrationally excited states of a large number of complex molecules are populated,
giving rise to a plethora of emission lines, probably with comparable intensities to
the putative glycine lines. The width of the LMH’s lines, typically *6–10 km s-1,
adds additional complication. Given the gigantic partition functions, many of the
lines are weak.
An instructive example of a well-founded detection of a large complex mole-
cule is the amino acetonitrile’s (glycinonitrile) detection by Belloche et al. [104] in
Sgr B2(N). This molecule is important since it is chemically related to the simplest
amino acid, glycine. Many lines have been targeted (see Fig. 2.4). While many
lines are blended, an important criterion is that all expected lines are actually seen.
In addition, Belloche et al. performed interferometer follow ups, establishing that
lines contributing to amino acetonitrile originate from the same positions in the
cloud. In the future, the Atacama Large Millimeter Array (ALMA) will be an
important instrument for the search for even more complex molecules, because the
line-blending problem will be much smaller at high angular resolution.
2.10 Isotopologues and Deuterated Molecules
The observations of isotopologues of a given molecule has been important for

several reasons: First of all, in many cases, the emission of the main isotopologue
is optically thick (12 CO being the most prominent example), which leads to
Fig. 2.4 Spectra of amino

acetonitrile lines at 3 mm
obtained with the IRAM
30-m telescope [104]. A fit
which includes all identified
molecules at 3 mm in Sgr
B2(N) is shown in light gray,
whereas a fit of only amino
acetonitrile lines is shown in
dark gray
difficulties in the determination of column densities and the observations of the

hidden, denser parts of the molecular clouds. Here, the rarer isotopologues come to
help, in the case of CO, 13 CO; C18 O, and C17 O which are observed often to study
the mass distribution in a cloud, while the optically thick main isotopologue is then
used to derive the excitation temperature. Even detections of 13 C18 O have been
made. Another important application is to use isotope abundance ratios as a tool
for the understanding of stellar nucleosynthesis and to look for galactic gradients
of these ratios (see, for example, [105]).
The large number of detected deuterated molecules came as a surprise, since the
cosmic D/H ratio is only of order 10-5, while new observations of deuterated
molecules in cold molecular clouds find values of this ratio even above 0.1. The
first radio detection of a deuterated molecule in the ISM was reported by Jefferts
et al. [106] with the detection of the DCN (3–2) line in the Orion molecular cloud.
In the same year, HD also was detected in the ISM with UV absorption mea-
surements by Spitzer et al. [107]. The reason for the increased D/H ratio at low
temperatures is that main reactions, such as Hþ þ
3 þ HD ! H2 D þ H2 , are exo-
thermic. In addition, in cold environments, CO, the main reactant for H2 Dþ , is
depleted further increasing the relative amount of deuterated molecules. This led in
2002 to record the detection of triple-deuterated molecules (ammonia and
methanol).
2.11 Prospects for Future Detections
Science with the Herschel Space Observatory has just started in 2010 has just started,
and the first results are encouraging (see the Astronomy & Astrophysics Herschel and
Herschel/HIFI special issues of 2010). Unbiased line surveys from about 650 to
70 lm that are being performed toward the most fruitful astronomical objects in
search of new detections of molecules in space have revealed already new species
(e.g., H2 Clþ ). Many lines of these surveys will remain unidentified, and more lab-
oratory spectroscopy and systematic computation and collections of frequencies, as
provided by the JPL and CDMS catalogs, will be crucial. For the detections of more
complex molecules, interferometers will be crucial. Besides ALMA, the Expanded
Very Large Array (EVLA) also will play a major role at cm wavelengths, where,
recently again, new heavy species are detected with the Greenbank telescope (e.g.,
the CH3 C6 H detections by Remijan et al. [108] toward TMC-1).
References
1. S. Pfalzner, C. Kramer, C. Staubmeier, A. Heithausen (eds.), The Dense Interstellar Medium

in Galaxies Springer proceedings in physics, vol. 91 (Proceedings of the 4th Cologne-Bonn-
Zermatt Symposium, Zermatt, Switzerland, 22–26, Sept. 2003), (Springer, Berlin-
Heidelberg, 2004), p. 69
2. H.I. Ewen, E.M. Purcell, Nature 168, 356 (1951)
3. S. Weinreb, A.H. Barrett, M.L. Meeks, J.C. Henry, Nature 200, 829 (1963)
4. C.H. Townes, in IAU Symp. 4: Radio astronomy, ed. by H.C. van de Hulst (Cambridge
University Press, London, 1957), p. 92
5. I.S. Shklovsky, Astr. Zhur. 29, 144 (1952)
6. B. Hoglund, P.G. Mezger, Science 150, 339 (1965)
7. A.C. Cheung, D.M. Rank, C.H. Townes, D.D. Thornton, W.J. Welch, Phys. Rev. Lett. 21,
1701 (1968)
8. A.C. Cheung, D.M. Rank, C.H. Townes, D.D. Thornton, W.J. Welch, Nature 221, 626
(1969)
9. L.E. Snyder, D. Buhl, B. Zuckerman, P. Palmer, Phys. Rev. Lett. 22, 679 (1969)
10. L.E. Snyder, Molecules in Space. in MTP International Review of Science, vol 3,
Spectroscopy. ed. by D.A. Ramsay (Butterworths, London, 1972), pp. 193–240
11. G. Winnewisser, M. Winnewisser, B.P. Winnewisser in MTP International Review of
Science, vol 3, Spectroscopy. ed. by D.A. Ramsay (Butterworths, London, 1972), p. 241
12. J.A. Ball, C.A. Gottlieb, A.E. Lilley, H.E. Radford, Astrophys. J. 162, L203 (1970)
13. B. Zuckerman, J.A. Ball, C.A. Gottlieb, Astrophys. J. 163, L41 (1971)
14. G. Winnewisser, E. Churchwell, Astrophys. J. 200, L33 (1975)
15. P. Palmer, C.A. Gottlieb, L.J. Rickard, B. Zuckerman, Bull. Am. Astron. Soc. 3, 499 (1971)
16. R.H. Rubin, G.W. Swenson, R.C. Benson, H.L. Tigelaar, W.H. Flygare, Astrophys. J. 169,
L39 (1971)
17. B.E. Turner, Astrophys. J. 163, L35 (1971)
18. F.F. Gardner, G. Winnewisser, Astrophys. J. 195, L127 (1975)
19. E. Churchwell, C.M.Walmsley, G. Winnewisser, Astron. Astrophys. 54, 925 (1977)
20. J.A. Ball, C.A. Gottlieb, A.E. Lilley, H.E. Radford, IAU Circ. No. 2350 (1971)
21. O.E.H. Rydbeck, J. Elldér, W.M. Irvine, Nature 246, 466 (1973)
22. M.W. Sinclair, N. Fourikis, J.C. Ribes, B.J. Robinson, R.D. Brown, P.D. Godfrey,
Australian J. Phys. 26, 85 (1973)
23. P.D. Godfrey, R.D. Brown, B.J. Robinson, M.W. Sinclair, Astrophys. Lett. 13, 119 (1973)
24. N. Fourikis, K. Takagi, M. Morimoto, Astrophys. J. 191, L139 (1974)
25. F.F. Gardner, G. Winnewisser, Astrophys. J. 197, L73 (1975)
26. R.D. Brown, J.G. Crofts, P.D. Godfrey, F.F. Gardner, B.J. Robinson, J.B. Whiteoak,
Astrophys. J. 197, L29 (1975)
27. E. Churchwell, G. Winnewisser, Astron. Astrophys. 45, 229 (1975)
28. G. Winnewisser, F.F. Gardner, Astron. Astrophys. 48, 159 (1976)
29. L.W. Avery, N.W. Broten, J.M. MacLeod, T. Oka, H.W. Kroto, Astrophys. J. 205, L173
(1976)
30. F.F. Gardner, G. Winnewisser, Mon. Not. R. Astro. Soc. 185, 57 (1978)
31. L.E. Snyder, D. Buhl, P.R. Schwartz, F.O. Clark, D.R. Johnson, F.J. Lovas, P.T. Giguere,
Astrophys. J. 191, L79 (1974)
32. O.E.H. Rydbeck, J. Elldér, A. Sume, Å. Hjalmarson, W.M. Irvine, Astron. Astrophys. 34,
479 (1974)
33. H.M. Pickett, R.L. Poynter, E.A. Cohen, M.L. Delitsky, J.C. Pearson, H.S.P. Müller, J.
Quant. Spectrosc. Rad. Transfer 60, 883 (1998)
34. H.S.P. Müller, S. Thorwirth, D.A. Roth, G. Winnewisser, Astron. Astrophys. 370, L49
(2001)
35. R.W. Wilson, K.B. Jefferts, A.A. Penzias, Astrophys. J. 161, L43 (1970)
36. D. Buhl, L.E. Snyder, Nature 232, 161 (1971)
37. M. Morris, B.E. Turner, P. Palmer, B. Zuckerman, Astrophys. J. 205, 82 (1976)
38. P.F. Goldsmith, Astrophys. J. 176, 597 (1972)
39. J. Stutzki, R. Guesten, Astrophys. J. 356, 513 (1990)
40. J.M. Jackson, J.M. Rathborne, R.Y. Shah, R. Simon, T.M. Bania, D.P. Clemens, E.T.
Chambers, A.M. Johnson, M. Dormody, R. Lavoie, M.H. Heyer, Astrophys. J. Suppl. 163,
145 (2006)
41. T.P. Snow, B.J. McCall, Ann. Rev. Astron. Astrophys. 44, 367 (2006)
42. R. Bachiller, Ann. Rev. Astron. Astrophys. 34, 111 (1996)
43. J. Cernicharo, M. Guélin, Astron. Astrophys. 183, L10 (1987)
44. M. Guélin, R. Lucas, J. Cernicharo, Astron. Astrophys. 280, L19 (1993)
45. K. Kawaguchi, E. Kagi, T. Hirano, S. Takano, S. Saito, Astrophys. J. 406, L39 (1993)
46. L.M. Ziurys, A.J. Apponi, M. Guélin, J. Cernicharo, Astrophys. J. 445, L47 (1995)
47. B.E. Turner, T.C. Steimle, L. Meerts, Astrophys. J. 426, L97 (1994)
48. M. Guelin, M. Forestini, P. Valiron, L.M. Ziurys, M.A. Anderson, J. Cernicharo, C. Kahane,
Astron. Astrophys. 297, 183 (1995)
49. L.M. Ziurys, C. Savage, J.L. Highberger, A.J. Apponi, M. Guélin, J. Cernicharo, Astrophys.
J. 564, L45 (2002)
50. J.L. Highberger, C. Savage, J.H. Bieging, L.M. Ziurys, Astrophys. J. 562, 790 (2001)
51. J.L. Highberger, K.J. Thomson, P.A. Young, D. Arnett, L.M. Ziurys, Astrophys. J. 593, 393
(2003)
52. S. Petrie, Mon. Not. R. Astron. Soc. 302, 482 (1999)
53. C.W. Walmsley, R. Bachiller, G.P. des Forêts, P. Schilke, Astrophys. J. 566, L109 (2002)
54. R.L. Plambeck, N.R. Erickson, Astrophys. J 262, 606 (1982)
55. F. Combes, T. Wiklind, Astron. Astrophys. 334, L81 (1998)
56. L.W. Avery, M.B. Bell, C.T. Cunningham, P.A. Feldman, R.H. Hayward, J.M. McLeod,
H.E. Matthews, J.D. Wade, Astrophys. J 426, 737 (1994)
57. H. Spinrad, D.B. Wood, Astrophys. J. 141, 109 (1965)
58. T. Dunhum Jr., W.S. Adams, Publ. Astron. Soc. Pacific 49, 26 (1937)
59. A. McKellar, Publ. Astron. Soc. Pacific 52, 187 (1940)
60. P. Swings, L. Rosenfeld, Astrophys. J. 86, 483 (1937)
61. A.E. Douglas, G. Herzberg, Astrophys. J. 94, 483 (1941)
62. P.F. Goldsmith, W.D. Langer, Astrophys. J. 222, 881 (1978)
63. D.A. Neufeld, S. Lepp, G.J. Melnick, Astrophys. J. Suppl. 100, 132 (1995)
64. A. Wootten, B.E. Turner, J.G. Mangum, M. Bogey, F. Boulanger, F. Combes, P.J. Encrenaz,
M. Gerin, Astrophys. J. 380, L79 (1991)
65. T.G. Phillips, E.F. van Dishoeck, J. Keene, Astrophys. J. 399, 533 (1992)
66. P. Schilke, D.J. Benford, T.R. Hunter, D.C. Lis, T.G. Phillips, Astrophys. J. Suppl. 132, 281
(2001)
67. G.A. Blake, J. Keene, T.G. Phillips, Astrophys. J. 295, 501 (1985)
68. G.J. Melnick, M.L.N. Ashby, R. Plume, E.A. Bergin, D.A. Neufeld, G. Chin, N.R. Erickson,
P.F. Goldsmith, M. Harwit, J.E. Howe, S.C. Kleiner, D.G. Koch, B.M. Patten, R. Schieder,
R.L. Snell, J.R. Stauffer, V. Tolls, Z. Wang, G. Winnewisser, Y.F. Zhang, Astrophys. J. 539,
L87 (2000)
69. A. Schulz, R. Güsten, C.M. Walmsley, E. Serabyn, Astron. Astrophys. 246, L55 (1991)
70. J.R. Pardo, J. Cernicharo, F. Herpin, J. Kawamura, J. Kooi, T.G. Phillips, Astrophys. J. 562,
799 (2001)
71. E.F. van Dishoeck, D.J. Jansen, P. Schilke, T.G. Phillips, Astrophys. J. 416, L83 (1993)
72. R. Stark, F.F.S. van der Tak, E.F. van Dishoeck, Astrophys. J. 521, L67 (1999)
73. T.R. Geballe, T. Oka, Nature 384, 334 (1996)
74. D.A. Neufeld, J. Zmuidzinas, P. Schilke, T.G. Phillips, Astrophys. J. 488, L141 (1997)
75. E.T. Polehampton, J.-P. Baluteau, C. Ceccarelli, B.M. Swinyard, E. Caux, Astron.
Astrophys. 388, L44 (2002)
76. J.R. Goicoechea, J. Cernicharo, Astrophys. J. 554, L213 (2001)
77. K.M. Menten, F. Wyrowski, A. Belloche, et al., Astron. Astrophys. 525, A77 (2011)
78. F. Wyrowski, K.M. Menten, R. Gusten, and A. Belloche, Astron. Astrophys. 518, A26
(2010)
79. D.A. Neufeld, P. Sonnentrucker, T.G. Phillips, et al., Astron. Astrophys. 518, L108 (2010)
80. D.C. Lis, J.C. Pearson, D.A. Neufeld, et al., Astron. Astrophys. 512, L9 (2010)
81. D. Buhl, Nature 228, 267 (1970)
82. W. Klemperer, Nature 227, 1230 (1970)
83. E. Herbst, W. Klemperer, Astrophys. J. 185, 505 (1973)
84. K. Kawaguchi, Y. Kasai, S.-I. Ishikawa, N. Kaifu, Publ. Astro. Soc. Japan 47, 853 (1995)
85. M.C. McCarthy, C.A. Gottlieb, H. Gupta, P. Thaddeus, Astrophys. J. 652, L141 (2006)
86. M. Agundez, J. Cernicharo, M. Guélin, et al., Astron. Astrophys. 517, L2 (2010)
87. E. Herbst, Astron. Soc. Pac. Conf. Ser. 417, 153 (2009)
88. P.T.P. Ho, C.H. Townes, Ann. Rev. Astron. Astrophys. 21, 239 (1983)
89. P. Schilke, C.M. Walmsley, T.L. Wilson, R. Mauersberger, Astron. Astrophys. 227, 220
(1990)
90. J. Keene, G.A. Blake, T.G. Phillips, Astrophys. J. 271, L27 (1983)
91. P. Thaddeus, S.E. Cummins, R.A. Linke, Astrophys. J. 283, L45 (1984)
92. P. Thaddeus, J.M. Vrtilek, C.A. Gottlieb, Astrophys. J. 299, L63 (1985)
93. J. Cernicharo, A.M. Heras, A.G.G.M. Tielens, J.R. Pardo, F. Herpin, M. Guélin, L.B.F.M.
Waters, Astrophys. J. 546, L123 (2001)
94. A.G.G.M. Tielens, Ann. Rev. Astron. Astrophys. 46, 289 (2008)
95. E. Herbst, E.F. van Dishoeck, Ann. Rev. Astron. Astrophys. 47, 427 (2009)
96. Y. Kuan, S.B. Charnley, H. Huang, W. Tseng, Z. Kisiel, Astrophys. J. 593, 848 (2003)
97. L.E. Snyder, J.M. Hollis, L.W. Brown, D. Buhl, R.D. Suenram, F.J. Lovas, Astrophys. J.
268, 123 (1983)
98. I.I. Berulis, G. Winnewisser, V.V. Krasnov, R.L. Sorochenko, Sov. Astron. Lett. 11, 251
(1985)
99. F. Combes, N. Q-Rieu, G. Wlodarczak, Astron. Astrophys. 308, 618 (1996)
100. C. Ceccarelli, L. Loinard, A. Castets, A. Faure, B. Lefloch, Astron. Astrophys. 362, 1122
(2000)
101. A. Nummelin, P. Bergman, Å. Hjalmarson, P. Friberg, W.M. Irvine, T.J. Millar, M. Ohishi,
S. Saito, Astrophys. J. Suppl. 117, 427 (1998)
102. A. Nummelin, P. Bergman, Å. Hjalmarson, P. Friberg, W.M. Irvine, T.J. Millar, M. Ohishi,
S. Saito, Astrophys. J. Suppl. 128, 213 (2000)
103. S. Liu, D.M. Mehringer, and L.E. Snyder, Astrophys. J. 552, 654 (2001)
104. A. Belloche, K.M. Menten, C. Comito, H.S.P. Müller, P. Schilke, J. Ott, S. Thorwirth, C.
Hieret, Astron. Astrophys. 482, 179 (2008)
105. T.L. Wilson, R. Rood, Ann. Rev. Astron. Astrophys. 32, 191 (1994)
106. K.B. Jefferts, A.A. Penzias, R.W. Wilson, Astrophys. J. 179, L57 (1973)
107. L. Spitzer, J.F. Drake, E.B. Jenkins, D.C. Morton, J.B. Rogerson, D.G. York, Astrophys. J.
181, L116 (1973)
108. A.J. Remijan, J.M. Hollis, L.E. Snyder, P.R. Jewell, F.J. Lovas, Astrophys. J. 643, L37
(2006)
Chapter 3
Fundamentals of Spectroscopy
for Astrophysics
Koichi M. T. Yamada, Per Jensen and Nobukimi Ohashi
3.1 Atoms
3.1.1 Hamiltonian for the Hydrogen Atom
The hydrogen atom is composed of two particles, a proton and an electron, and the
Hamiltonian describing the state of a hydrogen atom is
^20
^ ¼ p p2
^
H þ 1 þ Vðjr1 r0 jÞ; ð3:1:1Þ
2m0 2m1
where m; r and p^ represent the masses, coordinate vectors, and their conjugates
momentum operators of the two particles; subscripts 0 and 1 of these variables
indicate the proton and electron, respectively. The potential energy V is the
attractive Coulomb interaction between the nucleus and the electron;
Ze2
V¼ ; ð3:1:2Þ
4p0 jr1 r0 j
K. M. T. Yamada (&)
National Institute of Advanced Industrial Science and Technology (AIST), EMTech,
Onogawa 16-1, Tsukuba,
Ibaraki 305-8569, Japan
P. Jensen
Fachgruppe Chemie, FB C Mathematik u. Naturwissenschaften, Bergische Universität
Wuppertal, Gaussstr. 20, 42097 Wuppertal, Germany
N. Ohashi
Faculty of Science, Department of Physics, Kanazawa University, Kakuma,
Kanazawa 920-1192, Japan

Springer-Verlag Berlin Heidelberg 2011
44 K. M. T. Yamada et al.
where e is the proton charge (electron charge is ‘‘e’’) and Z is the atomic number;
Ze represents the charge of nuclei and Z ¼ 1 for the hydrogen atom.
Introducing the relative coordinate r; the coordinate of the center of mass R;
total mass M; and the reduced mass l; i.e.,
r ¼ r1 r0 ; ð3:1:3Þ
ðm0 r0 þ m1 r1 Þ
R¼ ; ð3:1:4Þ
m0 þ m1
M ¼ m1 þ m0 ; ð3:1:5Þ
m1 m0
l¼ ; ð3:1:6Þ
m1 þ m0
Eq. 3.1.1 is expressed as
^2 p2
^ ¼ P þ^
H þ VðjrjÞ; ð3:1:7Þ
2M 2l
where P^ and ^ p are the conjugate momenta of R and r; respectively. Since the
electron mass is very much smaller than the proton mass, the reduced mass l is
very close to the electron mass m1 as
l ð1 0:000544Þm1 : ð3:1:8Þ
The first term of Eq. 3.1.7 represents the translation energy of the hydrogen
atom, which appears as the Doppler effect by spectroscopic observation. To
describe the internal motion of the hydrogen atom, we can ignore this term.
3.1.2 Schrödinger Equation
The Schrödinger equation, i.e., wave equation, for the stationary state of the
internal motion of a hydrogen atom is then

h2 2

r þ VðjrjÞ w ¼ Ew: ð3:1:9Þ
2l
By introducing the polar coordinates r ¼ ðr; h; /Þ; i.e.,

x ¼ r sin h cos / ð3:1:10Þ
y ¼ r sin h sin / ð3:1:11Þ
z ¼ r cos h; ð3:1:12Þ
3 Fundamentals of Spectroscopy for Astrophysics 45
the wave equation (3.1.9) can be rewritten as

h2 1 o 2 o 1 o o 1 o2
r þ sin h þ w ¼ fE VðrÞgw:
2l r 2 or or r 2 sin h oh oh r 2 sin2 h o/2
ð3:1:13Þ
Substituting the wavefunction w by a product of a radial part and angular parts;
w ¼ RðrÞHðhÞUð/Þ; ð3:1:14Þ
the separation of variables of the wave equation (3.1.13) can be performed as
described elsewhere in standard text book for quantum mechanics [1];
d2 U
¼ E/ U; ð3:1:15Þ
d/2

1 d dH E/
sin h ¼ Eh 2 H; ð3:1:16Þ
sin h dh dh sin h

h2 1 d
2 dR h2 Eh
r ¼ E VðrÞ R: ð3:1:17Þ
2l r 2 dr dr 2lr 2
The first two equations govern the angular motion and are independent of either
the reduced mass or the potential energy; they are applied commonly for angular
motion in space expressed by the polar coordinate system. The third equation rules
the radial motion.
3.1.3 Angular Wavefunction
Requiring the periodic boundary conditions, the eigenvalues and eigenfunctions of

those angular wave equations (3.1.15) and (3.1.16) are obtained as
E/ ¼ m; ð3:1:18Þ
pffiffiffiffiffiffi
U ¼ Um ð/Þ ¼ ð 2pÞ1 eim/ ; ð3:1:19Þ
Eh ¼ lðl þ 1Þ; ð3:1:20Þ

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ð2l þ 1Þðl MÞ! M
H ¼ Hm
l ðcos hÞ ¼ Pl ðcos hÞ: ð3:1:21Þ
2ðl þ MÞ!
The value l is a positive integer or 0, which is named azimuthal quantum number

and represents the orbital angular momentum. The value m is an integer, which is
named magnetic quantum number, representing the projection of the orbital

angular momentum on an arbitrary defined Z axis; consequently
m ¼ l; l þ 1; . . .; l 1; l: ð3:1:22Þ
The capital M represents the absolute value of m; i.e., M ¼ jmj: The function PM l in
Eq. 3.1.21 is the associate Legendre polynomial. The product of the functions
Um ð/Þ and Hm l ðcos hÞ is a spherical harmonic function, Ylm ðh; /Þ; which is an
eigenfunction of the orbital angular momentum L ^2 and of its component L ^z ; i.e.,
^2 Ylm ðh; /Þ ¼ lðl þ 1Þ

L h2 Ylm ðh; /Þ; ð3:1:23Þ
^Z Ylm ðh; /Þ ¼ m
L hYlm ðh; /Þ: ð3:1:24Þ
3.1.4 Radial Wavefunction
The last term in the large bracket on the right hand of the radial wave equation
(3.1.17) is rewritten using the quantum number l as
lðl þ 1Þh2
2
; ð3:1:25Þ
2lr
which represents the effective potential due to the centrifugal effect. Thus the
radial wave equation depends on l: The eigenvalues of the radial wave equation
(3.1.17) represents the internal energy of a hydrogen atom, and are independent of
l; they are [1]
Z 2 e2
E ¼ En ¼ : ð3:1:26Þ
2a0 n2
The value n is a positive integer and called as principal quantum number. Here we
introduce the radius of the first Bohr orbit for hydrogen atom,
h2
4p0
a0 ¼ 0:0529 nm: ð3:1:27Þ
le2
To be exact, the definition of the Bohr radius is slightly different from a0 ; the Bohr
radius is obtained by substituting the reduced mass l by the electron mass me in
Eq. 3.1.27.
The Eq. 3.1.26 explains the empirical formula of Rydberg,
RH
En ¼ hc ; ð3:1:28Þ
n2
where RH is the Rydberg constant for hydrogen.1 The word term is historically
used to specify the electronic states expressed by the Rydberg formula, Eq. 3.1.28,
and the term value is the electronic energy expressed in wavenumber, i.e., E=hc:
The eigenfunctions of the radial wave equation (3.1.17) depend on n and l;
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
3
2Z ðn l 1Þ! Zr=na0 2Zr l 2lþ1 2Zr
R ¼ Rnl ¼ e Lnþl : ð3:1:29Þ
na0 2n½ðn þ 1Þ!3 na0 na0
The function L2lþ1

nþl is the associated Laguerre polynomial. Between n and l there is
a restriction,
n [ l; ð3:1:30Þ
i.e., for a given value of n;

l ¼ 0; 1; . . .; n 1: ð3:1:31Þ
3.1.5 Atomic Orbit
The motion of the electron in a hydrogen atom is specified by three quantum

numbers n; l; and m; as described above. The internal energy of a hydrogen atom is
expressed by Eq. 3.1.26 and depends only on the principal quantum number n: For
a given n; there are n different l states, Eq. 3.1.31, which are degenerate in energy.
The degeneracy can be lifted as described later by relativistic effects. There are
2l þ 1 different m states for a given l; Eq. 3.1.22. They are degenerate in energy
unless any external field is implied, and the orbital motion of electron in a
hydrogen atom is usually specified by two quantum numbers, n and l:
The state specified by n ¼ 1 is lowest in energy, i.e., the ground state; substi-
tuting Z ¼ 1 and n ¼ 1 in Eq. 3.1.26 we obtain
e2
E1 ¼ 13:6 eV: ð3:1:32Þ
2a0
For n ¼ 1 the orbital angular momentum quantum number l must be 0 as required
by Eq. 3.1.31. The orbit with l ¼ 0 is denoted as ‘‘s’’ orbit. Together with the n
value, the inner most orbit is denoted as 1s: For n ¼ 2; l can be 0 or 1. The orbit
with l ¼ 1 is denoted as ‘‘p’’ orbit.
Because of the m degeneracy, there are three, i.e., 2l þ 1; p orbits. Altogether
there are one 2s orbit and three 2p orbits for n ¼ 2: For n ¼ 3; l can be 0, 1, or 2.
1
The Rydberg constant is usually expressed in wavenumber in unit of cm1 : The factor hc is
explicitly written here to adjust the dimension.
The orbit with l ¼ 2 is denoted as ‘‘d’’ orbit. Thus there are one 3s; three 3p; and
five 3d orbits for n ¼ 3:
The notation s; p; and d are taken from the words sharp, principal, and diffuse,
which represent the spectral characters of the alkali atoms [2]. The l ¼ 3 is denoted
by f ; from the word fundamental, and the orbits with l larger than three are named
alphabetically as ‘‘g’’, ‘‘h’’, and so on.
3.1.6 Electron Spin and Fine Structure
In addition to the orbital angular momentum, an electron has an intrinsic angular

momentum, which is called electron spin. The electron spin angular momentum
operator ^s and its projection ^sZ of an axis Z satisfies;
^s2 n ¼ sðs þ 1Þ

h2 n; ð3:1:33Þ
^sZ n ¼ ms
hn: ð3:1:34Þ
with s ¼ 12 and ms ¼ 12 : The spin eigen function n is often expressed as a and b

for ms ¼ 12 and 12 : Thus for each orbit described in the previous section is
associated with two electron spin states; i.e., 1sa and 1sb; for example. The state of
an electron in the hydrogen atom is characterized by four quantum numbers,
n; l; m; and ms : The state specified by those four quantum numbers is called orbital
in atomic and molecular physics.
The total angular momentum j of the hydrogen atom is the vector sum of the
orbital and spin angular momentum
j ¼ l þ s; ð3:1:35Þ
i.e., the total angular momentum for an s electron is j ¼ 12 ; and that for an l [ 0
state are j ¼ l þ 12 and l 12 :
The electron spin contribution to the internal energy of a hydrogen atom cannot
be evaluated by the Schrödinger equation, Eq. 3.1.9. Required is the Dirac equation
where the relativity contributions are properly taken care of. The degeneracy for a
given n state in the Bohr model, Eq. 3.1.26, is thus removed by this interaction and
the energy levels exhibit fine structure. The fine structure is indicated by j in
subscript as nlj ; for example, 2p32 means ðn; l; jÞ ¼ ð2; 1; 32Þ: The fine structure
splitting of the 2p32 and 2p12 is approximately 10.9 GHz. By the Dirac equation, the
2s12 and the 2p12 states are still degenerate, which can be removed by applying the
QED theory resulting with about 1 GHz splitting, which is called the Lamb shift.
Classically speaking, the electron spin couples with orbital angular momentum
via magnetic interaction, i.e., the spin–orbit interaction;
hso ¼ Aso ð^l ^sÞ:

^ ð3:1:36Þ
The energy shift due to this interaction can be evaluated, for the first order
correction, with the diagonal element of this operator, applying the vector model,
j 2 ¼ l 2 þ 2l s þ s 2 ð3:1:37Þ
as
1
DEso ¼ Aso ½ jðj þ 1Þ lðl þ 1Þ sðs þ 1Þ : ð3:1:38Þ
2
3.1.7 Nuclear Spin and Hyperfine Structure
The proton, the nucleus of the hydrogen atom, has also a spin angular momentum,
nuclear spin I ¼ 12 : The total angular momentum including the nuclear spin, F; is
the vector sum of I and j;
F ¼ j þ I; ð3:1:39Þ
and the energy terms are further split by the nuclear spin–electron spin interaction,
resulting in the hyperfine structure; the interaction operator is similar to that of the
spin–orbit interaction, Eq. 3.1.36,
hHF ¼ að^j ÎÞ:

^ ð3:1:40Þ
The diagonal contribution of this interaction is then evaluated by the vector model
as
1
DEHF ¼ a½ jðj þ 1Þ lðl þ 1Þ sðs þ 1Þ : ð3:1:41Þ
2
For example, 1s12 level splits into two, F ¼ 0 and 1, the energy difference is about
1.4 GHz. This frequency corresponds to the wavelength of 21 cm; the transition
between these hyperfine structures has been used in radio astronomy to observe the
hydrogen atom.
3.1.8 Atoms with More Than One Electron
3.1.8.1 Single Electron Approximation and Shell Structure
The Hamiltonian for the hydrogen atom, Eq. 3.1.1, can be extended for an atom
with more electrons; i.e., numbering the electrons by i ¼ 1; 2; . . .; N
PN X
N X
^20 p2i
^
^ ¼ p
H þ i¼1
þ Vðjri r0 jÞ þ V 0 ðrj ri Þ; ð3:1:42Þ
2m0 2me i¼1 j[i
where m0 and me are the masses of the nucleus and an electron, and r0 and ri
represent the coordinates of them. The potential energy V is the attractive
Coulomb interaction between the nucleus and the electron as in Eq. 3.1.2;
Ze2
V¼ ; ð3:1:43Þ
4p0 jri r0 j
The potential energy V 0 represents the repulsive Coulomb interaction between the
electrons i and j;
e2
V0 ¼ : ð3:1:44Þ
4p0 rj ri
Contrary to the case of hydrogen atom, the Schrödinger equation for the atoms
with more than one electron, derived from the Hamiltonian, Eq. 3.1.42, cannot be
solved analytically. The motion of each electron in such atoms can be approxi-
mated as a motion in the centrally symmetric potential, where the effect of the
other electrons is handled as an effective potential as an average. In this approx-
imation, the motion of an electron can be described by the quantum numbers n; l;
and m; as in the case of a hydrogen atom.
Although the approximation applied above is relatively poor, the number and
type of orbits can be obtained. As in the case of hydrogen atom, the energy
separation between the states with different n is much larger than that between the
different l of a given n: The energy levels with a given n form a cluster which is
called shell. The shell of n ¼ 1 is named K shell, then L; M; and so on for
n ¼ 2; 3; . . .:
Each shell can be further divide into several sub-shells characterized by the l
quantum number. The shell structure and the number of orbitals for each sub-shell
are listed in Table 3.1.
3.1.8.2 Pauli Principle and Closed Shell
The electrons are Fermions with spin 12 ; and therefore, the wavefunctions should
be antisymmetric to any permutation of electrons. If the two or more electrons
share an orbital, the wavefunction describing those electrons must be symmetric to
the permutation of those electrons. Obviously this is forbidden by the symmetry
Table 3.1 Shell structure of Shell K L M

atomic orbitals
Orbit 1s 2s 2p 3s 3p 3d
n 1 2 2 3 3 3
l 0 0 1 0 1 2
m degeneracy 1 1 3 1 3 5
# of orbitals 2 2 6 2 6 10
requirement for Fermions. Thus, each orbital can be occupied by a single electron.
This is known as Pauli exclusion principle.
Consequently, the orbitals of the K shell, the lowest energy shell, can be
occupied by two electrons and then the shell is closed. The L shell is closed by
eight electrons; two of them occupy the s sub-shell, and the rest occupy the p sub-
shell. For each closed sub-shell, the total orbital angular momentum and total spin
vanish;
X
li ¼ 0 ð3:1:45Þ
subshell
X
si ¼ 0: ð3:1:46Þ
subshell
The shells are closed for the ground state of rare gas atoms. By indicating the
number of electrons which occupy each orbital, the electron configuration of rare
gas atoms are expressed as 1s2 for He, 1s2 2s2 2p6 for Ne, 1s2 2s2 2p6 3s2 3p6 3d 10 for
Ar, and so on.
3.1.8.3 Alkali Atoms
The lightest atom of alkali metals is lithium (Li), which has three electrons. In the
ground electronic state, the K shell is closed by the two of them like in He, and
the rest one electron occupies one of the 2s orbitals, i.e., 1s2 2s1 : The motion of
the outermost electron resembles very much that of the hydrogen atom, because
the field caused by the electrons in the closed shell is spherically symmetric in
average. Thus the energy of this outer electron is well approximated by a formula
similar to Eq. 3.1.28;
RLi
En ¼ hc ; ð3:1:47Þ
ðn DÞ2
where D is a small correction term, named quantum defect, to the principal
quantum number n: It is clear from Eq. 3.1.47 the energy correction due to the
quantum defect is larger for the smaller n state.
The next larger atom of the alkali metals is sodium (Na), which contains 11
electrons. In the ground electronic state, ten of them occupy the K and L shells,
like in Ar, and one electron occupies one of the 3s orbitals, i.e., 1s2 2s2 2p6 3s1 :
3.1.8.4 Term Symbols
Before discussing the electron configuration of atoms other than the rare gasses
and alkali metals, we explain the symbols of the electronic term which express the
electron configurations of an atom. First we introduce an important coupling
scheme for the various angular momenta of electrons in an atom, so-called

Russel–Saunders coupling which is an approximation using vector models as
follows.
In this coupling scheme, the orbital angular momentum of each electron cou-
ples together, and they form a vector sum,
X
N
L¼ li ; ð3:1:48Þ
i
which is called total orbital angular momentum. Similarly the spin angular
momentum of each electron couples together, and they form the total electron spin,
X
N
S¼ si : ð3:1:49Þ
i
Those couplings are so strong that the atoms behaves as if they have a single
orbital angular momentum expressed by L and a single spin expressed by S:
Consequently the fine structures due to the spin–orbit coupling is represented by
the total angular momentum J which is the vector sum of L and S;
J ¼ L þ S; ð3:1:50Þ
instead of the individual coupling expressed by Eq. 3.1.35. Surprisingly, this

coupling scheme can be applied to most of the atoms.
Thus the electron configurations of atoms, or electronic terms, are specified by
the three quantum numbers L; S; and J: We denote the term of L ¼ 0 as S (caution,
capital), in analogy to the orbit notation s (lower case) for the l ¼ 0: Successively,
P; D; F; and so on for L ¼ 1; 2; 3; . . .: As a superscript on the left of the angular
momentum symbol, we indicate the spin multiplicity 2S þ 1; which represents a
number of possible different spin states, and as a subscript on the right the total
angular momentum J; which represents the fine structure component, i.e.,
2Sþ1
XJ ð3:1:51Þ
where X ¼ S; P; D; F; . . .:
As we discussed before the vector sum of the orbital angular momenta and spin
angular momenta vanishes for closed sub-shells, Eqs. 3.1.45 and 3.1.46, the terms
of rare gas atoms for their ground electronic states are all 1 S0 : The ground states of
a hydrogen atom and alkali atoms are expressed as 2 S12 :
3.1.8.5 Building-Up Principle
Now we consider a more complicated atom, the carbon atom for example. The
electronic configuration of the carbon atom is 1s2 2s2 2p2 in its ground state. The K
shell and 2s sub-shell are closed, and thus the vector sum of li vanishes upto 2s
sub-shell and that of si as well, see Eqs. 3.1.45 and 3.1.46. Thus we can completely
ignore those electrons in the closed shell by calculating L and S in Eqs. 3.1.48 and
3.1.49. In case where no restriction of the Pauli exclusion principle exists, the
value L can be 0, 1, or 2, and the value S can be 0 or 1; i.e., there are six possible
terms, 1 S; 1 P; 1 D; 3 S; 3 P; and 3 D:
Because of Pauli principle, if the two electrons are distributed in the 2p orbitals,
the number of possible terms are less than six. For finding these possible terms, there
seems to be no simpler way than that explained by the Table 3.11 in Herzberg’s book
[2]. The possible terms are only 1 S; 1 D; and 3 P: Among the three, the ground
electronic state of carbon is 3 P; which can be predicted by the Hund rules.
The Hund rules, which can be wonderfully applied to find out the ground state
term symbol for most of atoms (with exceptions), are
1. The term of largest multiplicity is lowest in energy.
2. Within the same multiplicity terms, the largest L term is lowest in energy.
3. Multiplets are regular (the term with smallest J value is lowest in energy) for
the term where less than half of the shell is filled, and are inverted (the term
with largest J value is lowest) for the term where more than half of the shell is
filled.
For the carbon case, there is only one triplet state. Thus we can find the 3 P to be
ground state from the rule 1, and no need to use the rule 2. This state is triplet and
there are three fine structure components, J ¼ 0; 1, and 2; 3 P0 ; 3 P1 ; and 3 P2 :
3.1.8.6 Fine Structure
Since the 2p sub-shell is less than half filled (two of six), the Hund rule 3 suggests
for the carbon atom that the multiplets are regular and the ground state should be
3
P0 : The energy separations of the fine structure are caused mainly by the spin–
orbit interaction. The spin–orbit coupling energy is thus explained in the first order
approximation as
^ ^
^ SÞ;
hso ¼ Aso ðL ð3:1:52Þ
which should be compared with Eq. 3.1.36. As in the case of single electron, the
diagonal element of this operator
1
DE ¼ Aso ½JðJ þ 1Þ LðL þ 1Þ SðS þ 1Þ : ð3:1:53Þ
2
gives the energy separations for the fine structure. The fine structure energies of
the carbon atom, 12 C and 13 C, in the ground 3 P state are determined experimen-
tally with high precision by observing the direct transitions between the fine
structure components [3, 4] as illustrated in Fig. 3.1. The energy separation of
3
P2 3 P1 is observed to be 809 GHz, which is fairly smaller than that expected by
Eq. 3.1.53 from the observed 3 P1 3 P0 separation (492 GHz).
Fig. 3.1 Fine structure 3

splitting observed for carbon P2
atom
+ASO
3
P
−ASO
3
P1
−2ASO
3
P0
3.1.8.7 Nuclear Spin and Hyperfine Structure
If the nuclear spin angular momentum I is not zero, the vector sum of the nuclear
spin and the electronic angular momentum J forms the total angular momentum F;
similarly to Eq. 3.1.39 in the case of single electron atoms, we have now
F ¼ J þ I: ð3:1:54Þ
The energy levels split further into several sub-levels denoted by F quantum
number which is the hyperfine structure.
Magnetic Interaction
In the Russel–Saunders coupling case the interaction of the nuclear spin I with the
total angular momentum J is expressed as
^
hmag ¼ aðJ^ ÎÞ; ð3:1:55Þ
which should be compared with Eq. 3.1.40. The diagonal contribution of this
operator can be evaluated by the vector model as
1
DEmag ¼ a½FðF þ 1Þ JðJ þ 1Þ IðI þ 1Þ : ð3:1:56Þ
2
The quantity in the square brackets in the equation above is often expressed by C:
i.e.,
C ¼ FðF þ 1Þ JðJ þ 1Þ IðI þ 1Þ: ð3:1:57Þ
For 12 C atom, there is no hyperfine splitting because the nuclear spin I ¼ 0 for
this isotope atom. On the other hand the nuclear spin I is 12 for 13 C isotope; the
hyperfine levels of 13 C atom in the ground state are illustrated in Fig. 3.2.
Fig. 3.2 Hyperfine levels of 3

P2 F = 5/2
13
C atom
F = 3/2
+ASO
3
P
−ASO
3
P1 F = 3/2
−2ASO F = 1/2
3
P0 F = 1/2
Electric Quadrupole Interaction
So far we have assumed that the nucleus is a point mass. It is, however, not the
case; a nucleus has a finite volume, and consequently it may have non-vanishing
electric multi-pole moments, which interact with the electric field caused by the
electrons surrounding the nucleus. Here we explain the interaction closely fol-
lowing the discussion presented Townes and Schawlow [5].
Because the electric dipole moment vanishes in atoms, if the nuclear state is not
degenerate, in most cases the leading term next to the mono-pole (Coulombic)
interaction in the electric multi-pole interaction energy is attributed to the quad-
rupole moment, i.e.,
Z
1 o2 V 1 o2 V 1 o2 V
WQ ¼ qðx; y; zÞ x2 2 þ y2 2 þ z2 2
2 ox 2 oy 2 oz
2 2 2
o V o V oV
þxy þ yz þ zx dv; ð3:1:58Þ
oxoy oyoz ozox
where qðx; y; zÞ is the nuclear charge density and V is the electric potential pro-
duced by the electrons. By removing the spherically symmetric contribution,
which vanishes in a good approximation, the quadrupole interaction energy is
expressed as
Z
1 o2 V o2 V o2 V
WQ ¼ qðx; y; zÞ ð3x2 r 2 Þ 2 þ ð3y2 r 2 Þ 2 þ ð3z2 r 2 Þ 2
6 ox oy oz
2 2 2
oV o V oV
þ6xy þ 6yz þ 6zx dv: ð3:1:59Þ
oxoy oyoz ozox
Since the volume of the nucleus is very small the field gradient V can be con-
sidered constant in the above equation.
Now we introduce an axis system ðxI ; yI ; zI Þ; where zI axis to be parallel to
the nuclear spin angular momentum vector I: Assuming that the nucleus
spins very fast, the charge density can be considered cylindrically symmetric,
i.e., qðxI ; yI ; zI Þ ¼ qðzI Þ; and the cross terms Eq. 3.1.59 vanish. In addition we
have
Z Z
qðzI Þð3x2I r 2 Þdv ¼ qðzI Þð3y2I r 2 Þdv
Z ð3:1:60Þ
1 2 2
¼ qðzI Þð3zI r Þdv;
2
where the relation

3z2I r 2 ¼ ð3x2I r 2 Þ þ ð3y2I r 2 Þ ð3:1:61Þ
is used.
The quantity
Z
1
Q¼ qðzI Þð3z2I r 2 Þdv ð3:1:62Þ
e
is the nuclear quadrupole moment. The quadrupole energy of Eq. 3.1.59 is then,

eQ o2 V 1 o2 V o2 V
WQ ¼ þ 2 : ð3:1:63Þ
6 oz2I 2 ox2I oyI
If the contribution of the electrons in the nucleus region is negligible, the Laplace
equation holds:
o2 V o2 V o2 V
þ 2 þ 2 ¼ 0: ð3:1:64Þ
ox2I oyI ozI
Equation 3.1.63 is thus further simplified as

eQ o2 V
WQ ¼ : ð3:1:65Þ
4 oz2I
The electrons move so fast that the zI direction (the direction of I) can be
considered to be constant during the period of the orbital motion of electrons. If we
take the average over the electronic orbital motion, the field gradient should
be cylindrically symmetric along the axis of the electronic angular momentum
vector J: Then, in the coordinate system ðxJ ; yJ ; zJ Þ; where the zJ axis is parallel to
J; we obtain
2 2
oV oV 1 o2 V
¼ ¼ ð3:1:66Þ
ox2J av oy2J av 2 oz2J av
where we apply the Laplace equation, Eq. 3.1.64. By using the angle h between I
and J; the quadrupole interaction energy, Eq. 3.1.65, is rewritten classically as
eqJ Q

WQ ¼ 3 cos2 h 1 =2; ð3:1:67Þ
4
where

o2 V
qJ ¼ : ð3:1:68Þ
oz2J av
The quantum mechanical expression corresponding to Eq. 3.1.67 was derived by

Casimir as

1
2 3
^ ^ ^ ^ ^ ^2 ^2
HQ ¼ eqJ Q 3 I J þ I J I J =½Ið2I 1ÞJð2J 1Þ: ð3:1:69Þ
2 2
^ Q is
The diagonal contribution of H
1 3CðC þ 1Þ=4 IðI þ 1ÞJðJ þ 1Þ

WQ ¼ eqJ Q ; ð3:1:70Þ
2 Ið2I 1ÞJð2J 1Þ
where C is defined by Eq. 3.1.57. It should be noted here that the quadrupole
moment vanishes for J ¼ 0; I ¼ 0 or 12 ; because qJ vanishes for J ¼ 0 and Q
vanishes for I ¼ 0 or 12 :
3.2 Diatomic Molecules
3.2.1 Hamiltonian for a Diatomic Molecule without Spin
The Hamiltonian of a diatomic molecule, AB, with n electrons, is expressed as

^ ¼H
H ^ nuc ðra ; rb ; ^
pa ; ^ ^ el ðr1 ; . . .; rn ; ^
pb Þ þ H p1 ; . . .; ^pn ; ra ; rb Þ; ð3:2:1Þ
where the effect of spins of electrons and nuclei is ignored, which will be discussed
later. In the Hamiltonian given above, the coordinates of the nuclei A and B are
represented by ra and rb ; their conjugate momenta are ^pa and ^pb ; the coordinates of
electrons are ðr1 ; r2 ; . . .; rn Þ; and their conjugate momenta are ð^p1 ; ^p2 ; . . .; ^pn Þ:
Those two parts of the Hamiltonian in Eq. 3.2.1 are concretely expressed using
individual coordinate and momentum as

2 e2
^ nuc ¼ 1 p
H p2b =mb þ
â =ma þ ^ Za Zb =jra rb j ð3:2:2Þ
2 4p0
!
X
n
e2 X n Xn Xn
^ el ¼ 1
H p2i =m
^ Za =jri ra j þ Zb =jri rb j 1=jri rj j ;
2 i¼1
4p0 i¼1 i¼1 i\j
ð3:2:3Þ
where ma and mb are the masses of the nucleus A and B, Za and Zb are the atomic
number of them, m is the electron mass, and e is the electron charge. The first part,
^ nuc ; contains the kinetic energies of the nuclei and the repulsive Coulombic
H
potential between the nuclei, and depends only on the nuclear variables. The second
part, H ^ el contains the kinetic energies of the electrons, attractive Coulombic
potential between the nuclei and electrons, and the repulsive one between the
electrons.
3.2.1.1 Separation of Nuclear and Electron Variables
The energy of a diatomic molecule can be obtained by solving the Schrödinger

equation,
nuc
^ þH
H ^ el Wðra ; rb ; r1 ; . . .; rn Þ ¼ EWðra ; rb ; r1 ; . . .; rn Þ: ð3:2:4Þ
For solving this equation, we assume that the eigenfunction Wðra ; rb ; r1 ; . . .; rn Þ can
be expressed as a product
Wðra ; rb ; r1 ; . . .; rn Þ ¼ Wnuc ðra ; rb ÞWel ðr1 ; . . .; rn ; ra ; rb Þ: ð3:2:5Þ
Substituting Wðra ; rb ; r1 ; . . .; rn Þ in Eq. 3.2.4 by Eq. 3.2.5, and neglecting the effect
pa and ^
of ^ pb on the function Wel ðr1 ; . . .; rn ; ra ; rb Þ; we obtain
^ nuc Wnuc ðra ; rb Þ þ Wnuc ðra ; rb ÞH

Wel ðr1 ; . . .; rn ; ra ; rb ÞH ^ el Wel ðr1 ; . . .; rn ; ra ; rb Þ
¼ EWnuc ðra ; rb ÞWel ðr1 ; . . .; rn ; ra ; rb Þ: ð3:2:6Þ
Dividing both sides by Wnuc Wel ; Eq. 3.2.6 is rewritten as

nuc nuc el el
H^ W ðra ; rb Þ ^ W ðr1 ; . . .; rn ; ra ; rb Þ
H
þ ¼ E: ð3:2:7Þ
Wnuc ðra ; rb Þ Wel ðr1 ; . . .; rn ; ra ; rb Þ
The first term on the left side of above equation is a function which depends
only on the nuclear coordinates ðra ; rb Þ and the second term is a function of both
nuclear and electron coordinates. To satisfy the equation for any ðr1 ; . . .; rn Þ; the
value of the second term should not depend on ðr1 ; . . .; rn Þ : i.e.,
el el
H^ W ðr1 ; . . .; rn ; ra ; rb Þ
¼ ðra ; rb Þ; ð3:2:8Þ
Wel ðr1 ; . . .; rn ; ra ; rb Þ
which is not other than the Schrödinger equation for the electrons treating the
nuclear coordinates ðra ; rb Þ as parameters:
^ el Wel ðr1 ; . . .; rn ; ra ; rb Þ ¼ ðra ; rb ÞWel ðr1 ; . . .; rn ; ra ; rb Þ:
H ð3:2:9Þ
The value ðra ; rb Þ is thus considered as the energy of the electrons in the state
expressed by Wel ðr1 ; . . .; rn ; ra ; rb Þ:
Substituting Eq. 3.2.8 into Eq. 3.2.7, we obtain

nuc nuc
H^ W ðra ; rb Þ
þ ðra ; rb Þ ¼ E; ð3:2:10Þ
Wnuc ðra ; rb Þ
which is the Schrödinger equation for the nuclear motion:
nuc
^
H þ ðra ; rb Þ Wnuc ðra ; rb Þ ¼ EWnuc ðra ; rb Þ: ð3:2:11Þ
Using Eq. 3.2.2 for H ^ nuc ; and considering that ðra ; rb Þ is actually expressed as
ðjra rb jÞ for a diatomic molecule, the equation above is expressed as

1
2
p =ma þ ^pb =mb þ Vðjra rb jÞ Wnuc ðra ; rb Þ ¼ EWnuc ðra ; rb Þ;
^ 2
ð3:2:12Þ
2 a
where
Za Zb e2
Vðjra rb jÞ ¼ þ ðjra rb jÞ: ð3:2:13Þ
4p0 jra rb j
The quantity V is the potential energy for the nuclear motion. In the terminology of
the molecular orbital theory, the word ‘‘electronic energy’’ implies the quantity V
defined by Eq. 3.2.13, which includes the repulsion energy between the nuclei,
instead of the defined by Eq. 3.2.9.
3.2.1.2 Born–Oppenheimer Approximation
For deriving Eq. 3.2.6, we introduced an assumption
pk Wel ðr1 ; . . .; rn ; ra ; rb Þ ¼ 0;
^ ð3:2:14Þ
where k ¼ a or b; i.e., the momentum operator for the nuclear motion, ^pa or ^pb ; has
no effect on the electronic wavefunctions. This assumption was introduced first by
Born and Oppenheimer in 1927 to simplify the problem [6], and referred as the
Born–Oppenheimer approximation.
This assumption is considered to be reasonable for most of the molecules
because of the following reason [1]. The electron mass m is 1836 times smaller
than the proton mass mp ; and about twenty thousand times smaller than the mass of
carbon nuclei, so that electrons in molecules move much faster than the nuclei.
When the nuclei change their positions, the electrons can follow almost imme-
diately, and thus it is enough to evaluate the motion of electrons for each
instantaneous positions of nuclei. In other words, the effect of the momenta of
nuclei on the electronic wavefunctions can be neglected; Eq. 3.2.14 is a quantum
mechanical expression for this.
The ratio of electron mass to nuclear mass is often used to evaluate the order of
magnitude of perturbations in molecular systems [7, 8]. The parameter often called
as ‘‘Born–Oppenheimer parameter’’ is defined as the quartic root of this ratio,
m 14 1
j¼ ; ð3:2:15Þ
M 10
where M is a typical nuclear mass. Using this parameter, the molecular levels can be
classified into three categories, i.e., the electronic energy ðEel Þ; vibrational energy
ðEvib Þ; and rotational energy ðErot Þ; with following order of magnitude relation:
Evib ¼ j2 Eel ð3:2:16Þ
Erot ¼ j4 Eel : ð3:2:17Þ
The electronic energy Eel is not other than V defined by Eq. 3.2.13.
3.2.2 Hamiltonian for a Diatomic Molecule without Electronic

Angular Momentum
3.2.2.1 Nuclear Motions
The Hamiltonian operator for the nuclear motions of a diatomic molecule, in the
left side of Eq. 3.2.12, is essentially same as that for the hydrogen atom, Eq. 3.1.1.
As we have done for the case of the hydrogen atom, we introduce the relative
coordinate r; the coordinate of the center of mass R; total mass M; and the reduced
mass l; i.e.,
r ¼ rb ra ; ð3:2:18Þ
ðma ra þ mb rb Þ
R¼ ; ð3:2:19Þ
ma þ ma
M ¼ ma þ mb ; ð3:2:20Þ
ma mb
l¼ : ð3:2:21Þ
ma þ mb
Then the Hamiltonian for the nuclear motion of a diatomic molecule is expressed
as
^2 p2
^ ¼ P þ^
H þ VðjrjÞ: ð3:2:22Þ
2M 2l
The first term represents the translational kinetic energy of the molecule in the
space, and will be ignored in the following.
As in the case of the hydrogen atom, we introduce the polar coordinate

system for the relative coordinate r; then the Schrödinger equation based on the
Hamiltonian, Eq. 3.2.22, is expressed as Eqs. 3.1.15–3.1.17. The angular parts are
exactly identical. The radial part, Eq. 3.1.17, can be rewritten as

h2 1 d
2 dwJ ðrÞ h2 JðJ þ 1Þ

r ¼ E VðrÞ wJ ðrÞ: ð3:2:23Þ
2l r 2 dr dr 2lr 2
where the angular momentum quantum number in Eh ; Eq. 3.1.20, is expressed by

J; because it is not the electronic angular momentum here. Equation 3.2.23 can be
solved for each J value, and thus the wavefunction may be expressed by wJ ðrÞ
indicating J as a parameter. The last term in the bracket on the right side can be
considered as an effective potential due to the centrifugal force induced by the
rotation of the molecule. It should be noted that the term can be rewritten as
2 JðJ þ 1Þ
h
; ð3:2:24Þ
2I
where I ¼ lr 2 is the moment of inertia of the molecule for the nuclear distance r:
3.2.2.2 Vibrational Energy
If the molecule is not rotating, J ¼ 0; we obtain from Eq. 3.2.23,

h2 1 d
2 d
r þ VðrÞ wJ¼0 ðrÞ ¼ EwJ¼0 ðrÞ; ð3:2:25Þ
2l r 2 dr dr
which is the Schrödinger equation for the vibrational motion (bond stretching
vibration) of the diatomic molecule, and E is not other than the vibrational energy
Ev : Equation 3.2.25 can be simplified by introducing
wv ðrÞ ¼ wJ¼0 ðrÞ=r ð3:2:26Þ
as

2 d2
h
þ VðrÞ wv ðrÞ ¼ Ev wv ðrÞ: ð3:2:27Þ
2l dr 2
The typical shape of the potential energy VðrÞ for a diatomic molecule is illus-
trated in Fig. 3.3, which is characterized by the three elemental values: (1) the
potential minimum at re ; (2) the dissociation energy measured from the potential
minimum De ; and (3) the infinitely large potential value at r ¼ 0:
If the amplitude of the vibration is small compared to the re ; it is convenient to
express the instantaneous nuclear distance r as
r ¼ re þ q: ð3:2:28Þ
Fig. 3.3 A typical shape of

the potential energy for a
diatomic molecule
V
De
0 re 10
r
Then the potential VðrÞ can be expressed as a power series of q:

X1
1 dn V
Vðre þ qÞ ¼ Vðre Þ þ n
qn : ð3:2:29Þ
n¼1
n! dr r¼re
Because Vðre Þ ¼ 0 and ðdV=dr Þr¼re ¼ 0; the above equation is rewritten as
1 1
Vðre þ qÞ ¼ kð2Þ q2 þ kð3Þ q3 þ : ð3:2:30Þ
2 6
The coefficient kð2Þ is called the quadratic force constant, or the harmonic force
constant. The coefficient kð3Þ is called the cubic force constant, and so on. The
contribution of the cubic and higher order terms is called as the anharmonicity of
the potential. By ignoring the anharmonic terms in the potential, Eq. 3.2.27 is
identical to the Schrödinger equation of a harmonic oscillator,

h2 d 2
1 2 harm
þ kq wv ðre þ qÞ ¼ Evharm wharm
v ðre þ qÞ; ð3:2:31Þ
2l dq2 2
and the eigenvalues, i.e., the harmonic vibrational energy, are well known as
1
Evharm ¼ ðv þ Þhxharm ; ð3:2:32Þ
2
where v is a vibrational quantum number and xharm is the angular frequency, or
angular velocity, of the classical harmonic oscillator,
sffiffiffiffiffiffiffi
kð2Þ
xharm ¼ : ð3:2:33Þ
l
The harmonic vibrational frequency mharm in the unit of Hz is thus expressed as2
mharm ¼ xharm =ð2pÞ ð3:2:34Þ
and Eq. 3.2.32 is rewritten as

1
Evharm ¼ v þ hmharm : ð3:2:35Þ
2
The harmonic oscillator wavefunction, i.e., the eigenfunction of Eq. 3.2.32, for
the state expressed by v is
1 2 2
wharm
v ðre þ qÞ ¼ Nv Hv ðaqÞe2 a q ; ð3:2:36Þ
where Hv is the vth Hermite polynomial with

ð2Þ 14
lk
a¼ : ð3:2:37Þ
h2

3.2.2.3 Rotational Energy
For the case of J 6¼ 0; the rotational correction term, Eq. 3.2.24, should be eval-
uated. If the nuclear distance of the diatomic molecule is fixed at the point of the
potential minimum (equilibrium position), r ¼ re ; then the moment of inertia I is
independent of the vibrational motion, and the rotational energy correction is
simply given as
Erot ¼ hBe JðJ þ 1Þ; ð3:2:38Þ
where Be is the rotational constant at the equilibrium position in the frequency unit
(Hz), and3
2
h
hBe ¼ : ð3:2:39Þ
2Ie
The nuclear distance changes by the vibrational motion, and therefore, the
evaluation of the rotational energy is not so simple as in Eq. 3.2.28. In many cases,
however, the rotational energy contribution is much smaller than that of the
vibration approximately given in Eq. 3.2.32 as discussed earlier: see Eqs. 3.2.16
and 3.2.17. Thus we can handle the rotational correction term as a perturbation to
the molecular vibration, and the first order correction term given by the diagonal
2
This value in wavenumber unit (cm1 ), ~m; is expressed as ~m ¼ xharm =ð2pcÞ with c in cm/s.
3
If the rotational constant is given in wavenumber unit (cm1 ), then hcBe ¼
h2 =ð2Ie Þ with c in
cm/s.
matrix element of the perturbation operator represents the rotational energy Erot ;
i.e.,
Erot ¼ hBv JðJ þ 1Þ; ð3:2:40Þ
where Bv is the effective rotational constant averaged for the vth vibrational state,
i.e.,
2
hBv ¼ hwv
h =ð2IÞwv i; ð3:2:41Þ
which can be approximated as

1 1 2
Bv ¼ Be a v þ þc vþ ð3:2:42Þ
2 2
3.2.2.4 Dunham Expansion
The rovibrational energy of a diatomic molecule is expressed effectively as the

sum of the harmonic vibrational energy, Eq. 3.2.35, and the rotational energy of
the rigid body, Eq. 3.2.40, with anharmonic and centrifugal distortion corrections:

1 1 2 1 2
Ev;J =h ¼ mharm v þ þx vþ þy v þ þ
2 2 2
þ Bv JðJ þ 1Þ Dv ½JðJ þ 1Þ2 þ Hv ½JðJ þ 1Þ3 þ ; ð3:2:43Þ
where x and y are called as anharmonicity constants,4 and D and H are the quartic
and sextic centrifugal distortion constants.
More sophisticated expression was derived by Dunham [9] as
X
1 i
Ev;J =h ¼ Yij v þ ½JðJ þ 1Þj : ð3:2:44Þ
i;j
2
The relation of the Dunham coefficients with the parameters used in the traditional
expression Eq. 3.2.43 can be found in the textbook by Townes and Schawlow [5].
The expressions of the Yi;j coefficients by the potential parameters, Eq. 3.2.30, are
also listed there; actually Dunham expressed the potential as
V ¼ a0 n2 ð1 þ a1 n þ a2 n2 þ Þ; ð3:2:45Þ
where n ¼ ðr re Þ=re :
4
For diatomic molecules the anharmonicity constants x; y; and so on are often defined as mharm x,
etc.
3.2.3 Hamiltonian for a Diatomic Molecule with Electronic

Angular Momentum
3.2.3.1 Vector Model and Hund’s Coupling Cases
Since the spin–orbit interaction of the electrons in a molecule is usually small, we

assume that the electronic orbital angular momentum L and electron spin angular
momentum. S; defined as
X
N
L¼ li ; ð3:2:46Þ
i
X
N
S¼ si ; ð3:2:47Þ
i
are good quantum numbers also for molecules: similar to the Russel–Saunders
coupling discussed for the atomic term expressions, see Sect. 3.1.8.4. In analogy to
the atomic term expressions, we denote the electronic states of L ¼ 0; 1; 2; . . . as
R; D; P; . . .; the molecular term symbols are expressed in Greek, whereas atomic
term symbols in Roman. The spin multiplicity 2S þ 1 is indicated as a superscript
on the left of the angular momentum symbol as in the case of atoms. Thus, the
electronic state discussed in the previous section is of 1 R.
For the cases other than the 1 R state, various coupling schemes of angular
momenta caused by molecular rotation, electronic orbital motion, and electron
spin have to be considered. The contribution of the nuclear spins is usually small
and can be considered separately.
We follow in this text the IUPAC recommendation in 1994 for various angular
momenta [10]:
L: electron orbital angular momentum
S: electron spin angular momentum
R: rotational angular momentum
N: total orbital angular momentum; N ¼ R þ L
J: total angular momentum without the nuclear spin; J ¼ N þ S
I: nuclear spin angular momentum.
The electronic states of a diatomic molecule resemble those of a atom perturbed
by the external field generated by another atom nearby. The perturbation should be
axially symmetric along the axis connecting the two atoms, so-called molecular
axis. Because of this symmetry of the perturbation, called ‘‘axis interaction’’ in this
text, not only the total angular momentum and its projection on a space fixed axis,
but also its projection on the molecular axis is conserved, i.e., the quantum number
specifying them are good quantum numbers. As described below the electronic
states of a diatomic molecule can be classified into the Hund’s coupling cases by
the approximate quantity of this projection on the molecular axis.
3.2.3.2 Hund’s Coupling Case ðaÞ
If the axis interaction is stronger than the interactions between the various angular
momenta, the electron orbital angular momentum L couples first with the axis. The
vector L undergoes a precessional motion along the axis (in view point of the
molecule-fixed system) and its projection on the axis is conserved, which is
denoted by K; which is a good quantum number; the states with different K are
well separated. The electron spin angular momentum S is then coupled with the
axis by the spin–orbit interaction, and its projection on the axis is represented by a
quantum number R: The K and R are the signed quantities: classically they are
K ¼ L eaxis ð3:2:48Þ
R ¼ S eaxis ; ð3:2:49Þ
where the direction of the molecular axis is represented by an unit vector eaxis :
Thus along with the molecular axis there is an angular momentum components, X;
defined as
X ¼ K þ R; ð3:2:50Þ
which is a good quantum number in this case; the multiplets separation for dif-
ferent jXj; which is caused by the spin–orbit interaction, are much larger than the
typical rotational energy separation.
The vector addition of this axis component and the rotational angular
momentum R forms the total angular momentum J:
J ¼ R þ Xeaxis : ð3:2:51Þ
This requires J X:
Since R is perpendicular to eaxis ; X is the axis components of the total angular
momentum J: The coupling scheme described above is illustrated in Fig. 3.4 by
Fig. 3.4 The vector model eaxis

is illustrated for Hund’s
case ðaÞ J S
Σ
Λ L
Ω B
A
the vector model above, which is called as Hund’s coupling case ðaÞ: In case ðaÞ; X
is a good quantum number, and thus we indicate its absolute value, jXj; in the term
symbol as a subscript on the right; for example the term symbol 2 P32 represents the
state of ðS ¼ 12 ; L ¼ 1; jXj ¼ 32Þ:
3.2.3.3 Rotational Energy in Case ðaÞ
Since the two angular momentum vectors, R and Xeaxis ; are perpendicular with
each other, the rotational energies in case ðaÞ can be expressed classically as the
sum of the quantity proportional to R2 and X2 : in quantum mechanics we obtain in
the first order approximation,
Erot =h ¼ Bv R2 þ AX2
ð3:2:52Þ
¼ Bv JðJ þ 1Þ X2 þ AX2 ;
where we employed the vector model, Eq. 3.2.51, to evaluate R2 : Bv is the rota-
tional constant, see Eq. 3.2.41. Since X is purely electronic, the term with A
represents a part of the electronic energy, and thus usually large. This term can be
rewritten as

AX2 ¼ A K2 þ R2 þ 2AKR: ð3:2:53Þ
The cross term on the right originates from the spin–orbit interaction, Aso LS: cf :
Eq. 3.1.52. The diagonal contribution of the interaction is Aso KR in case ðaÞ: Thus
Eq. 3.1.52 can be replaced by

Erot =h ¼ Bv JðJ þ 1Þ X2 þ Aso KR þ A K2 þ R2 : ð3:2:54Þ
The last term of the equation is a constant for a given electronic term, and thus can
be ignored here because it can be included in the electronic energy.5
The energy level diagram of a 2 P state is schematically presented in Fig. 3.5.
For this state, L ¼ 1 and S ¼ 12 ; i.e., K ¼ 1 and R ¼ 12 : Consequently, we have
X ¼ 1 12 ¼ 32 and X ¼ 1 12 ¼ 12 : The states of jXj ¼ 12 and 32 have dif-
ferent energies due to the spin–orbit interaction; the size of splitting is jAso j; if
Aso [ 0; the jXj ¼ 12 level is lower in energy than the jXj ¼ 32 ; and vice versa: the
multiplets split by the spin–orbit interaction with Aso [ 0 is called regular and
with Aso \0 inverted as in the atomic cases, cf : Sect. 3.1.8.6.
5
It should be noted that the spin–orbit constant Aso is different from A in Eq. 3.1.52 by a factor
of 2.
Fig. 3.5 Energy level 7.5

diagram of a 2 P (Hund’s case 6.5
ðaÞ) state is schematically
presented, assuming 6.5
ASO ¼ 10 and Bv ¼ 1 cm1 5.5
in Eq. 3.2.54
5.5 Λ-type doubling
4.5
4.5 3.5
3.5 2.5
1.5
2
2.5 Π Ω=3/2 J
+ASO/2
1.5
0.5 −ASO/2
J Ω=1/2
3.2.3.4 Hund’s Coupling Case ðbÞ
In this coupling case, the orbital angular momentum L couples strongly with the
axis, i.e., K is a good quantum number, but the electron spin S does not couple to
the axis, because of the weak spin–orbit interaction. The rotational angular
momentum R and K form a total orbital angular momentum N as
N ¼ R þ Keaxis ; ð3:2:55Þ
which requires N K:
Then the spin S is magnetically coupled with N; and the total angular
momentum J is formed by the vector addition,
J ¼ N þ S; ð3:2:56Þ
thus the quantum number J should be
J ¼ N þ S; N þ S 1; . . .; jN Sj: ð3:2:57Þ
The vector diagram of the coupling scheme is illustrated in Fig. 3.6
In this case, the spin multiplet components are split by the spin–rotation
interaction, cSN; and the energy is expressed in the first order approximation as
1
Erot =h ¼ Bv NðN þ 1Þ K2 þ c½JðJ þ 1Þ NðN þ 1Þ SðS þ 1Þ; ð3:2:58Þ
2
where c is called spin–rotation coupling constant (Fig. 3.7 illustrates an example).
3.2.3.5 Other Fine Structure Interactions
Here we explain two additional interactions which are often un-ignorable. The first
one is the spin–spin coupling. In case of more than one unpaired electrons,
Fig. 3.6 The vector diagram eaxis

of the coupling scheme is
illustrated for case ðbÞ
J
S
Λ L
N B
Fig. 3.7 The energy level 8 8.5

diagram of a 2 R state (Hund’s 70 7.5
case ðbÞ) is schematically
illustrated, assuming Bv ¼ 1 60
7.5
and c ¼ 0:1 cm1 in 7 6.5
Eq. 3.2.58 50
6.5
Energy [cm ]
−1
6 5.5
40
30 5 5.5
4.5
4 4.5
20
3.5
3 3.5
10 2.5
2.5
2 1.5
1 1.5
0 0 0.5 0.5
N 2 J
Σ
the interactions between the spins of those electrons causes an perturbation to the
energy expressed by

2
^ SS ¼ k 2 3^
H Sz S^2 : ð3:2:59Þ
3
The second one is the K-type doubling, which lifts the degeneracy of the
K ¼ þ1 and 1 state in the P electronic state; the interaction is effectively
expressed by the following operator:

2
H^K ¼ o ^Sþ þ ^S2 =2

p N ^þ^
Sþ þ N ^ ^S =2

2
þq N ^þ þ N^ 2 =2: ð3:2:60Þ
The q term exists for all P states. The p term vanishes for 1 P states, and the o term
vanishes for 1 P and 2 P states.
The K-type doubling interaction, Eq. 3.2.60, however, is an off-diagonal
interaction. The couplings discussed in the previous sections also have off-diag-
onal contributions. In the descriptions given to the previous sections are limited to
the diagonal contributions of the interaction terms, assuming that the coupling
scheme (a) or (b) is good. In order to reduce the off-diagonal contributions as
much as possible, coupling schemes other than the case (a) and (b) have been
proposed.
In order to evaluate correctly the off-diagonal contributions, however, it
becomes common to express the necessary matrix elements (diagonal as well as
off-diagonal) by the case (b) basis and to diagonalize the energy matrix numeri-
cally; thanks to the developments in the computer technology, it works very well
in general, including the hyperfine interactions described in the next section.
3.2.4 Hyperfine Structure
If the nuclear spin angular momentum I is not zero for a nucleus of the diatomic
molecule, the energy levels split further into the hyperfine structure denoted by F
quantum number defined by Eq. 3.1.54. As discussed in Sect. 3.1.8.7 there are two
kinds of interactions which contribute to the hyperfine structure: the magnetic
interaction and the electric quadrupole interaction. For simplicity we assume here
that only one nucleus has a non-vanishing nuclear spin.
3.2.4.1 Magnetic Interaction
Similar to the magnetic interaction in atoms, Eq. 3.1.55, there are interactions
between the nuclear spin angular momentum and the other angular momenta.
Assuming the z-axis is the molecular axis, the magnetic interactions can be
expressed as
^
hmag ¼ aðÎ LÞ
^
^
þ bF ðÎ SÞ
^
þ cðI^z S^z Î S=3Þ
dðS^þÎþ þ ^ SÎ Þ=2; ð3:2:61Þ
where the first line in the right is the nuclear spin–orbit interaction, the second line
the Fermi contact interaction, the third line the diagonal contribution of the
electron–spin nuclear–spin dipole–dipole interaction, and the last line the off-
diagonal contribution for the P electronic state between the K ¼ 1:
For a diatomic molecule, we have additionally the nuclear spin–rotation

coupling:
^
hIJ ¼ CI ðÎ JÞ
^
CI0 ðÎþ ^
Jþ þ Î ^
J Þ=2; ð3:2:62Þ
3.2.4.2 Electric Quadrupole Interaction
The quadrupole Hamiltonian H ^ Q for an atom, Eq. 3.1.69, can be applied also for
diatomic molecules. In diatomic molecules, the field gradient is axially symmetric
along the molecular axis, and thus the quantity qJ in the equation can be related to
the molecule axis component qz by as described in detail by Townes and
Schawlow [5]
J
qJ ¼ qz : ð3:2:63Þ
2J þ 3
^ Q is obtained as
Substituting this into Eq. 3.1.70, the diagonal contribution of H
WQ ¼ eqz QYðJ; I; FÞ; ð3:2:64Þ
where Y is the Casimir’s function

3CðC þ 1Þ=4 IðI þ 1ÞJðJ þ 1Þ
YðJ; I; FÞ ¼ : ð3:2:65Þ
2Ið2I 1Þð2J 1Þð2J þ 3Þ
3.3 Polyatomic Molecules
3.3.1 The Born–Oppenheimer Approximation
We consider a molecule consisting of N nuclei with masses mg and charges

Cg e; g ¼ 1; 2; . . .; N; and n electrons, each with the mass me and charge e: The
molecule is not influenced by external fields. We initially describe each nucleus by
means of its Cartesian coordinates ðXg ; Yg ; Zg Þ in an Cartesian axis system XYZ
rigidly attached to the laboratory frame (a laboratory-fixed axis system), and each
electron by its coordinates ðxi ; yi ; zi Þ; i ¼ 1; 2; . . .; n; in the same axis system, and
we denote the set of all Cartesian coordinates for the nuclei
ðX1 ; Y1 ; Z1 ; X2 ; Y2 ; Z2 ; . . .; XN ; YN ; ZN Þ as Rn and the set of all Cartesian coordinates
for the electrons, ðx1 ; y1 ; z1 ; x2 ; y2 ; z2 ; . . .xn ; yn ; zn Þ as re : The position vector of
nucleus g; with coordinates ðXg ; Yg ; Zg Þ; is called Rg ; and the position vector of
electron i; with coordinates ðxi ; yi ; zi Þ; is called ri : If we consider only Coulomb
interaction between the particles, the non-relativistic Schrödinger equation for the
molecule can be written as

T^n þ Tê þ VCoulomb ðRn ; re Þ wne ðRn ; re Þ ¼ Ene wne ðRn ; re Þ ð3:3:1Þ
with the molecular wavefunction wne ðRn ; re Þ and its corresponding eigenvalue Ene :
In Eq. 3.3.1, the operator T^n representing the kinetic energy of the nuclei is given
by
" #
2X
N 2 2 2

h 1 o o o
T^n ¼ þ þ ; ð3:3:2Þ
2 g¼1 mg oXg2 oYg2 oZg2
the operator Tê representing the kinetic energy of the electrons is given by
n 2
^ h2 X
o o2 o2
Te ¼ þ þ ð3:3:3Þ
2me i¼1 ox2i oy2i oz2i
and the Coulomb potential energy function VCoulomb ðRn ; re Þ is given by

N X
X n
C g e2
VCoulomb ðRn ; re Þ ¼
g¼1 i¼1
4 p 0 jRg ri j
X e2 X Cg Cg0 e2
þ þ ð3:3:4Þ
i\i0
4 p 0 jri0 ri j g\g0 4 p 0 jRg0 Rg j
At the present time, all practical schemes for solving Eq. 3.3.1 are two-stage
procedures: First an ab initio calculation is carried out, that is, the electronic
Schrödinger equation
h i
Tê þ VCoulomb ðRð0Þ ð0Þ ð0Þ ð0Þ
n ; re Þ we ðRn ; re Þ ¼ VðRn Þ we ðRn ; re Þ ð3:3:5Þ
is solved for fixed nuclear coordinates Rð0Þ

n : By solving Eq. 3.3.5 for many values
ð0Þ
of Rn ; we obtain the effective potential energy function VðRn Þ; and in the second
step of the calculation we can solve the Schrödinger equation for the nuclear
motion

T^n þ VðRn Þ wn ðRn Þ ¼ Ene wn ðRn Þ ð3:3:6Þ
where wn ðRn Þ is the nuclear wavefunction. In the approximation defined by Eqs.

3.3.5 and 3.3.6, the total molecular wavefunction is then taken to be
wne ðRn ; re Þ ¼ we ðRn ; re Þwn ðRn Þ: ð3:3:7Þ
This approximation is usually credited to Born and Oppenheimer [6].

We assume here that the ab initio calculation has been carried out and that
consequently the potential energy function VðRn Þ is known. With the known
VðRn Þ we then develop procedures for solving Eq. 3.3.6 in order to obtain the
molecular rovibronic energies Ene and the corresponding wavefunctions wn ðRn Þ:
3.3.2 The Choice of Rotation–Vibration Coordinates
3.3.2.1 Translation
The first problem that one encounters when developing a method for solving the
nuclear Schrödinger equation (3.3.6), is to choose a suitable set of coordinates for
describing the nuclear motion. This problem does not exist in ab initio theory,
where all existing methods appear to describe the motion of the electrons through
the Cartesian coordinates ðxi ; yi ; zi Þ ½i ¼ 1; 2; . . .; n: These coordinates are well
suited to express the atomic-orbital-type basis functions used in ab initio calcu-
lations, and consequently they constitute a rather natural choice for the coordinates
of the electronic problem. However, the analogous nuclear Cartesian coordinates
ðXg ; Yg ; Zg Þ½g ¼ 1; 2; . . .; N are entirely inconvenient for setting up Eq. 3.3.6 in a
form that allows it to be solved in an efficient manner. Consequently, we must
rewrite the total nuclear Hamiltonian
^ total ¼ T^n þ VðRn Þ
H ð3:3:8Þ
by introducing new, more suitable coordinates. There is no natural, generally

applicable choice for these coordinates, and the currently available methods for
solving Eq. 3.3.6 employ different strategies for choosing them. We shall discuss
here the strategy used in the standard approach.
In general, we obviously need 3N coordinates to describe completely a system
consisting of N nuclei. All methods for solving Eq. 3.3.6 have in common that they
use three of these coordinates to describe the translational motion of the molecule.
During a translation of the molecule, all nuclei move uniformly along straight lines
without changing their relative positions. Usually the three translational coordi-
nates ðX0 ; Y0 ; Z0 Þ are chosen as the laboratory-fixed coordinates of the nuclear
center of mass
8 9 8 9
< X0 = 1 XN < Xg =
Y0 ¼ PN mg Yg ð3:3:9Þ
: ; g¼1 mg g¼1
: ;
Z0 Zg
where mg is the mass of atom g:6 The remaining 3N 3 coordinates used to

describe the nuclear motion are socalled internal coordinates; they are invariant
under uniform translation of all nuclei. It is customary to define initially such
6
In rotation–vibration calculations it is accepted practice to use atomic masses rather than
nuclear masses. As explained by Bunker and Moss [11] this compensates to some extent for the
breakdown of the Born–Oppenheimer approximation; see also Gordy and Cook [12].
coordinates ðt1;X ; t1;Y ; t1;Z ; t2;X ; t2;Y ; t2;Z ; . . .; tN1;X ; tN1;Y ; tN1;Z Þ through N 1
equations (see Sutcliffe and Tennyson [13])
8 9 8 9
< ti;X = X N < Xg =
ti;Y ¼ Vgi Yg ; i ¼ 1; 2; . . .; N 1: ð3:3:10Þ
: ; g¼1 : ;
ti;Z Zg
The condition that the coordinates ti;A [A ¼ X; Y; or Z] be invariant under uniform

translation of the molecule can be expressed through the equation
X
n
Vgi ¼ 0 for all i ¼ 1; 2; . . .; N 1: ð3:3:11Þ
g¼1
Further, it necessary that the Vgi be such that the coordinate transformation defined
through Eqs. 3.3.9 and 3.3.10,
ðX1 ; Y1 ; Z1 ; X1 ; Y1 ; Z1 ; . . .; XN ; YN ; ZN Þ
! ðX0 ; Y0 ; Z0 ; t1;X ; t1;Y ; t1;Z ; t2;X ; t2;Y ; t2;Z ; . . .; tN1;X ; tN1;Y ; tN1;Z Þ ð3:3:10Þ
can be inverted so that it is possible to determine the coordinates ðX1 ; Y1 ; . . .; ZN Þ

from the coordinates ðX0 ; Y0 ; Z0 ; t1;X ; . . .; tN1;Z Þ: Traditionally, the ti;A coordinates
have been chosen so that they define the positions of the nuclei in an axis system
with axes parallel to the laboratory-fixed XYZ system but with origin in the nuclear
center of mass (see Chap. 6 of Bunker [14] or Chap. 11 of Wilson et al. [15]).
This leads to the following values for the Vgi matrix elements
mg
Vgi ¼ dgi PN : ð3:3:13Þ
g¼1 mg
In this case, the Vgi matrix elements depend on the nuclear masses and hence the
ti;A coordinates become mass dependent or isotope-dependent. This means that
if we consider two isotopic molecules whose nuclei are arranged such that the two
molecules have the same values of the laboratory-fixed coordinates
ðX1 ; Y1 ; . . .; ZN Þ; these two molecules will be described by different values of the
coordinates ðt1;X ; t1;Y ; . . .; tN1;Z Þ: It is not necessary that the Vgi matrix elements
depend on the nuclear masses, and in many cases it is advantageous to choose
them to be mass-independent.
It can be shown that we can write the nuclear kinetic energy operator T^n as
2
^ h2
o o2 o2
Tn ¼ PN þ þ
2 g¼1 mg oX02 oY02 oZ02

^ o o o o
þ Trv ih ; ih ; i
h ; . . .; ih ð3:3:14Þ
ot1;X ot1;Y ot1;Z otN1;Z
where, as indicated, the kinetic energy operator T^rv (which describes the rotation
and vibration of the molecule) only depends on the momenta iho=oti;A conjugate
to the ti;A coordinates. Since in the absence of external fields, the effective potential
energy function VðRn Þ cannot depend on the coordinates ðX0 ; Y0 ; Z0 Þ;7 this means
that we can write H ^ total (Eq. 3.3.8) as the sum of two commuting operators,
^ total ¼ T^translation þ H
H ^ rv ð3:3:15Þ
where
2
^ h2 o o2 o2
Ttranslation ¼ PN þ þ ð3:3:16Þ
2 g¼1 mg oX02 oY02 oZ02
and
^ rv ¼ T^rv þ VðRn Þ:
H ð3:3:17Þ
^ total can be written as
Consequently the eigenvalues of H
Etotal ¼ Etranslation þ Erv ð3:3:18Þ
where Etranslation is an eigenvalue of T^translation and Erv is an eigenvalue of H

^ rv : We
are not interested in the energy contribution Etranslation from the translational
motion. The reason is that this contribution is of no importance in the prediction
and interpretation of spectroscopic experiments; such experiments do not involve
transitions between translational states. We thus simply ignore T^translation and
concentrate on the solution of the eigenvalue problem
^ rv wrv ¼ Erv wrv
H ð3:3:19Þ
which involves the ti;A coordinates only.
3.3.2.2 Rotation and Vibration
We have been led to introduce the 3N 3 coordinates ti;A to describe the

translation free nuclear motion. These coordinates, however, are not ideal for
this purpose, and we must carry out further coordinate transformations before we
can attempt to solve Eq. 3.3.19. We can think of the nuclei as carrying out a
superposition of two kinds of motion:
• uniform rotation of the molecule as a whole. When the molecule rotates without
the relative positions of the nuclei changing, the effective potential function
VðRn Þ obviously does not change when there are no external fields. Conse-
quently, the rotation is free, i.e., it is not influenced by VðRn Þ:
7
If we let the molecule undergo a uniform translation of all nuclei corresponding to a change in
the center-of-mass coordinates ðX0 ; Y0 ; Z0 Þ; VðRn Þ cannot change.
• motion of the nuclei relative to each other, which we will call vibration. Such
motion obviously is influenced by the effective potential function VðRn Þ:
When we define the coordinates ti;A as linear combinations of laboratory-fixed
coordinates, Eq. 3.3.10, the displacement of any one of the ti;A coordinates will
normally involve both rotation of the molecule as a whole and some relative
motion of the nuclei. Consequently, VðRn Þ will in general depend on all
3N 3 ti;A ’s. We would like to define a new set of 3N 3 coordinates that reflect
correctly the separation of the nuclear motion into rotation and vibration. The
general strategy for doing this, which is common for all methods of rotation–
vibration calculations, is as follows: A Cartesian axis system (the molecule-fixed
axis system) with origin at the nuclear center of mass is attached to the molecule so
that it follows the rotational motion (in a sense that we shall discuss below). The
orientation of the molecule-fixed axis system relative to the laboratory-fixed XYZ
axis system is defined by three Euler angles ðh; /; vÞ; see Appendix A of Papoušek
and Aliev [16]. We use these angles as rotational coordinates. Further, we define
3N 6 new coordinates ðq1 ; q2 ; q3 ; . . .; q3N6 Þ which are invariant not only under
uniform translation of the molecule [as were the ti;A ’s] but also under uniform
rotation of the molecule.8 Obviously we must select the scheme for attaching the
molecule-fixed coordinate system to the molecule and define the qi coordinates
such that the coordinate transformations
ðt1;X ; t1;Y ; t1;Z ; t2;X ; t2;Y ; t2;Z ; . . .; tN1;X ; tN1;Y ; tN1;Z Þ

$ ðh; /; v; q1 ; q2 ; q3 ; . . .; q3N6 Þ ð3:3:20Þ
can be carried out in both directions, so that if we know a set of values of the
coordinates ti;A ; we can calculate the values of the coordinates ðh; /; v; q1 ; . . .;
q3N6 Þ and vice versa.
The traditional spectroscopic approach to solving Eq. 3.3.19 [8, 15–21] uses
vibrational coordinates chosen in order to simplify the expression for the kinetic
energy operator T^rv : This operator can be written as

T^rv ¼ T^rot h; /; v; P
^h; P ^/; P^v

þ T^vib q1 ; q2 ; . . .; q3N6 ; P
^ q1 ; P
^ q2 ; . . .; P
^ q3N6

þ T^rest h; /; v; P ^h; P ^/; P
^ v ; q1 ; q2 ; . . .; q3N6 ; P^ q1 ; P
^ q2 ; . . .; P
^ q3N6 ; ð3:3:21Þ
where P ^ a denotes the momentum conjugate to a coordinate a; and the rotational

and vibrational coordinates are defined such that the rotation–vibration interaction
term T^rest is minimized. The background for this is the following: It follows from
the discussion given above that the effective potential energy function V can only
8
In the ‘‘standard theory’’ for a molecule whose equilibrium geometry is linear, two rotational
coordinates, h and /; and 3N 5 vibrational coordinates are employed [16]. This is a special
case which we discuss further in Sect. 3.3.6.5.
depend on the 3N 6 coordinates qi : When we write the rotation–vibration

^ rv as
Hamiltonian H

^ rv ¼ T^rot þ T^vib þ V þ T^rest ;
H ð3:3:22Þ
T^rot will commute with T^vib þ V since these two operators depend on different
coordinates. In the event that we can neglect T^rest we thus have the remaining
Hamiltonian written as a sum of two commuting operators whose eigenvalue prob-
lems we can solve separately and then construct the rotation–vibration wavefunctions
as the product
wrv ðh; /; v; q1 ; q2 ; . . .; q3N6 Þ ¼ wrot ðh; /; vÞwvib ðq1 ; q2 ; . . .; q3N6 Þ: ð3:3:23Þ
In practice, we cannot neglect T^rest ; but if we can minimize it through our coor-
dinate choice we might be able to treat it as a perturbation.
The coordinate choice outlined here aims at eliminating rotation–vibration inter-
action terms from the kinetic energy operator. Since these terms depend on the nuclear
masses, the coordinates ðh; /; v; q1 ; q2 ; q3 ; . . .; q3N6 Þ which are chosen to eliminate
them also become mass-dependent. However, the potential energy function is not
naturally expressed in these mass-dependent coordinates. We normally use a param-
eterized function of the nuclear coordinates to express the effective potential function
V; we would typically obtain the values of the parameters by fitting this function
through a set of ab initio energies with corresponding nuclear geometries. As noted by
Hoy et al. [21], the values of the parameters in V are independent of isotopic substi-
tution provided that we express V as a function of 3N 6 mass-independent, geo-
metrically defined coordinates which we might call ðR1 ; R2 ; R3 ; . . .; R3N6 Þ: These
coordinates are typically instantaneous values of internuclear distances (‘‘bond
lengths’’) or angles \(ABC) defined by three nuclei A, B, and C (‘‘bond angles’’). They
are different from the mass-dependent coordinates ðq1 ; q2 ; . . .; q3N6 Þ which we have
introduced to simplify the kinetic energy operator, so before we can solve Eq. 3.3.19
we must express the effective potential function in terms of the coordinates chosen for
the kinetic energy operator through coordinate transformations of the type
Ri ¼ Ri ðq1 ; q2 ; . . .; q3N6 Þ: ð3:3:24Þ
These relations are inserted in the effective potential energy function V and this
function can now be seen as depending on the coordinates qi : It is normally
possible to derive exact relations of the type given in Eq. 3.3.24, i.e., the Ri
coordinates can be expressed exactly in terms of the qj coordinates. However,
inserting these relations in the effective potential energy function leads to a
mathematically intractable form of this function. Consequently, one normally
resorts to an approximate coordinate transformation, in which the Ri coordinates
are expressed as power series in the qj coordinates (see Eq. 3 of Hoy et al. [21]):
X
3N6
ðjÞ
3N6 X
X 3N6 ðjkÞ
Ri Ri0 þ Bi q j þ B i qj qk þ ð3:3:25Þ
j¼1 j¼1 k¼1
In the preceding paragraphs we have outlined the choice of what one might call
‘‘spectroscopic coordinates’’. The form of these coordinates is dictated by the
kinetic energy operator, because they are defined so that they minimize the rota-
tion–vibration interaction term T^rest in Eq. 3.3.21. Thus it is necessary to transform
the potential energy function to make it depend on the spectroscopic coordinates.
In the beginnings of rotation–vibration theory 60–70 years ago, one only had hopes
of calculating approximative eigenvalues for a rotation–vibration Hamiltonian
in which rotation and vibration were separated [i.e., in which T^rest could be
neglected), and consequently this type of coordinates constituted the only
feasible choice. However, the ‘‘standard’’ separation of rotation and vibration is
possible only if the molecule carries out its vibration in one deep ‘‘well’’ of the
potential energy surface. This potential well defines one equilibrium structure, and
the amplitudes of the vibrational motion must be small compared to the linear
dimensions of this structure. In this book we are only concerned with the cus-
tomary, standard approach to molecular rotation–vibration theory. Modern
extensions of this theory, which make use of high-capacity computers, are
described, for example, in Jensen and Bunker [22] and the references given there.
3.3.2.3 The Standard Coordinates
The standard approach to rotation–vibration theory [8, 15–21, 23, 24], describes
molecules whose potential energy surfaces have one minimum only, and the
molecule-fixed coordinate system (see Sect. 3.3.2.2) is defined so that the con-
figuration of the molecule corresponding to the one minimum [the equilibrium
configuration] is rigidly attached to the molecule-fixed coordinate system. Hence
the equilibrium configuration is described in the molecule-fixed axis system
through constant position vectors ag for the nuclei g ¼ 1; 2; . . .; N:
8 9
< agx =
ag ¼ agy : ð3:3:26Þ
: ;
agz
Vibrational motion is then defined as the instantaneous displacements of the nuclei

from the positions given by the vectors ag ; measured along the axes of the
molecule-fixed coordinate system. As already mentioned, the standard approach
assumes that these displacements are small compared to the linear dimensions of
the equilibrium configuration.
3.3.3 The Eckart Equations
We have still not given a precise definition of the molecule-fixed coordinate

system. If we think of a non-linear molecule (i.e., a molecule whose equilibrium
structure is non-linear), we imagine for a moment that we could photograph an
instantaneous configuration of the nuclei. On the photograph we would see the

nuclei as points, and we could in principle determine their Cartesian, laboratory-
fixed coordinates ðXg ; Yg ; Zg Þ; but we would initially have no way of deciding to
which extent the molecule on the picture was displaced from its equilibrium
structure, and we would have no way of determining the Euler angles h; /; and v:
We obviously need a mathematical ‘‘recipe’’ allowing us to position the molecule-
fixed axis system [and the equilibrium geometry attached to it] relative to the
instantaneous positions of the nuclei. In the standard approach, this recipe is given
through the Eckart equations [17].
In order to formulate the Eckart equations, we define the following column
vectors
8 9 8 9 8 9
< xg = < Xg = < X0 =
rg ¼ yg ; rg ¼ Yg ; and r0 ¼ Y0 ð3:3:27Þ
: ; : ; : ;
zg Zg Z0
where ðxg ; yg ; zg Þ are the coordinates of nucleus g in the molecule-fixed coordinate

system. Since the molecule-fixed coordinate system is rotated by the Euler angles
relative to the laboratory-fixed axis system, we have that

rg ¼ Sðh; /; vÞ Rg R0 ð3:3:28Þ
where Sðh; /; vÞ is a 3 3 matrix effecting the transformation between the mol-

ecule-fixed and laboratory-fixed coordinate system; the elements of this matrix are
simple trigonometric functions of the angles h; /; and v; see Appendix A and
Table A.1 of [16].
We now define the instantaneous displacements of the nuclei from the equi-
librium configuration, measured along the axes of the molecule-fixed coordinate
system, as column vectors
8 9
< dgx =
dg ¼ dgy ¼ rg ag ; g ¼ 1; 2; . . .; N: ð3:3:29Þ
: ;
dgz
The vector components dgx ; dgy ; and dgz ; g ¼ 1; 2; . . .; N; are the vibrational
coordinates. However, we have seen in Sect. 3.3.2.2 that there exist only 3N 6
independent vibrational coordinates and consequently the 3N components of the
dg -vectors cannot all be independent. Six of the 3N components are redundant, and
consequently there must exist six equations linking all 3N components. We
already know three of these equations: The molecule-fixed coordinate system is
defined to have its origin at the nuclear center of mass (Sect. 3.3.2.2), both when
the molecule is in its equilibrium configuration and when it is in an arbitrary,
displaced configuration. One can easily show that this leads to the equations
X
N X
N
m g ag ¼ mg dg ¼ O ð3:3:30Þ
g¼1 g¼1
where O is a zero, three-component column vector. Equation 3.3.30 delivers three

equations connecting the components of the vectors dg : We need to choose three
additional equations connecting these components, and in order to understand
the choice that is customarily made we consider the classical kinetic energy T of
the molecular rotation and vibration. This energy is given by Eq. 3 in Sect. 11-1 of
[15]:
X
N

XN X
N
2T ¼ mg x rg x rg þ mg r_ g r_ g þ 2x mg rg r_ g ð3:3:31Þ
g¼1 g¼1 g¼1
where the vector x contains the components of the instantaneous angular velocity
of the molecule and, for an arbitrary quantity x; x_ is the time derivative dx=dt:
The last term on the right-hand side of Eq. 3.3.31 is the Coriolis coupling term
which represents the dominant part of the interaction between rotation and
vibration. We would like to choose the coordinates so that this term disappears
completely, i.e.,
X
N
mg rg r_ g ¼ O ð3:3:32Þ
i¼1
where r_ g is the velocity of nucleus g relative to the molecule-fixed axes.

This would mean that the molecule would have no angular momentum relative to
the molecule-fixed axes. It turns out that we cannot in general choose coordinates
that cause the angular momentum in the molecule-fixed axis system (and hence
the Coriolis coupling term) to disappear. The best we can do is to minimize this
angular momentum through the Eckart equations [17] which we can express as
X
N
mg ag dg ¼ O: ð3:3:33Þ
g¼1
Using Eq. 3.3.33 we can easily show that classically

X
N X
N
mg rg r_ g ¼ mg dg d_ g ð3:3:34Þ
g¼1 g¼1
so that the Coriolis coupling term vanishes in the equilibrium configuration when
all dg ¼ O: Since we assume that the molecule carries out small oscillations
around the equilibrium configuration, the angular momentum in the molecule-
fixed coordinate system is then always small and a high degree of rotation–
vibration separation is achieved.
When we insert Eqs. 3.3.28 and 3.3.29 in Eq. 3.3.33 we obtain
X
N
mg ag Sðh; /; vÞ Rg R0 ¼ O: ð3:3:35Þ
g¼1
In a given instantaneous configuration of the molecule, we can in principle

measure the Cartesian laboratory-fixed coordinates contained in the vectors Rg and
use Eq. 3.3.9 to obtain R0 : If we assume that we know the constant vectors ag ; Eq.
3.3.35 then delivers three equations in the Euler angles h; /; and v; and we can
solve the equations to obtain these angles as functions of the laboratory-fixed
Cartesian coordinates.
The first term on the right-hand side of Eq. 3.3.31 is the rotational kinetic
energy, and we can rewrite it as
X
N

mg x rg x rg ¼ xT Ix ð3:3:36Þ
g¼1
where a superscript T denotes transposition and the 3 3 matrix I contains

the instantaneous moments of inertia and products of inertia [see, for example,
Sect. 4.1 of [16]. When the molecule is in its equilibrium configuration, i.e., when
all rg ¼ ag ; we have that I ¼ Ie ; and the elements of Ie depend on the components
of the constant vectors ag only. These vectors must be chosen such that when all
rg ¼ ag the relative positions of the nuclei correspond to the minimum of the
potential energy function. However in defining the ag we are free to rotate them
relative to the molecule-fixed axis system since this does not change the relative
positions of the nuclei. We can use this freedom of choice to make the molecule-
fixed axis system a principal axis system for the molecule in its equilibrium
configuration, i.e., to make Ie diagonal. Hence for the molecule in its equilibrium
configuration we have
xT Ie x ¼ Ie;xx x2x þ Ie;yy x2y þ Ie;zz x2z : ð3:3:37Þ
3.3.4 Normal Coordinates
We have now defined the rotational Euler angle coordinates through the Eckart
equations (3.3.33). We now need to define the 3N 6 vibrational coordinates,
which we called ðq1 ; q2 ; q3 ; . . .; q3N6 Þ in Sect. 3.3.2.2. Through this definition we
aim at simplifying the vibrational Hamiltonian. The classical vibrational energy is
the sum of the vibrational kinetic energy, which is given through the second term
on the right-hand side of Eq. 3.3.31:
1X N
1X N
Tvib ¼ mg r_ g r_ g ¼ mg d_ g d_ g
2 g¼1 2 g¼1
ð3:3:38Þ
1X N n 2 2 2
o 1
¼ mg d_ gx þ d_ gy þ d_ gz ¼ d_ T Md_
2 g¼1 2
and the potential energy VðRn Þ: In Eq. 3.3.38, d is a column vector with 3N rows
containing the components of the vectors dg : d1x ; d1y ; d1z ; d2x ; . . .; dNz ; and M is a
3N 3N diagonal matrix with M11 ¼ M22 ¼ M33 ¼ m1 ; M44 ¼ M55 ¼ M66 ¼ m2
and so on.
In the standard approach the potential energy VðRn Þ is taken to be a Taylor
expansion in the geometrically defined coordinates defined above:
1X
VðR1 ; R2 ; . . .; R3N6 Þ ¼ fjk Rj Rk
2 j;k
1X 1 X
þ fjk‘ Rj Rk R‘ þ fjk‘ Rj Rk R‘ Rm þ
6 j;k;‘ 24 j;k;‘;m
ð3:3:39Þ
where the coordinates Ri are defined as displacements from equilibrium values so
that the configuration in which all Ri ¼ 0 corresponds to the equilibrium geom-
etry, where all first derivatives are zero and where we choose V ¼ 0:
In order to define the vibrational motion of the molecule, we now define 3N 6
independent, linearized internal coordinates Si through the transformation
N X
X
Si ¼ Bi;ga dga ; i ¼ 1; 2; 3; . . .; 3N 6; ð3:3:40Þ
g¼1 a¼x;y;z
with the elements of the ð3N 6Þ 3N matrix B given as

oRi
Bi;ga ¼ ; ð3:3:41Þ
odga
the elements of this matrix can be derived using purely geometrical arguments. In
these coordinates the kinetic energy of the vibrations can be expressed as (see [15])
1 T
Tvib ¼ S_ G1 S_ ð3:3:42Þ
2
with S being a column vector with 3N 6 rows containing the Si coordinates and
G ¼ BM1 BT : ð3:3:43Þ
Through the definition of the B matrix we have for small values of the dga (i.e.,
configurations close to equilibrium) that Si Ri ; and we can then write
1X 1X 1
fjk Rj Rk fjk Sj Sk ¼ ST FS ð3:3:44Þ
2 j;k 2 j;k 2
where the matrix F has the elements fjk : Using a standard technique of classical
mechanics, the GF-calculation, we can change to so-called normal coordinates Qk
given through the transformation
1 3N6 X
dga ¼ pffiffiffiffiffiffi ‘ga;k Qk : ð3:3:45Þ
mg k¼1
The normal coordinates are defined such that
1X 1 T 1 Xn 2
1 3N6 o
Tvib þ fjk Sj Sk ¼ S_ G1 S_ þ ST FS ¼ Q_ k þ kk Q2k : ð3:3:46Þ
2 j;k 2 2 2 k¼1
In Eqs. 3.3.45 and 3.3.46, the kk are the eigenvalues of the matrix GF; they are
related to the harmonic vibration wavenumbers xk of the molecule through
1 pffiffiffiffiffi
xk ¼ kk : ð3:3:47Þ
2pc
The coefficients ‘ga;k can be obtained from the eigenvectors of the GF matrix, they
are given by Eq. 3.3.10 of [16]):
l ¼ M1=2 BT G1 L: ð3:3:48Þ
In Eq. 3.3.48, l is the 3N 3N 6 matrix with elements ‘ga;k ; and the matrix L
contains the eigenvectors of the matrix GF; that is, it is determined such that
L1 GFL ¼ K ð3:3:49Þ

where K is a diagonal matrix with Kkk ¼ kk : When we approximate the vibrational
energy by the expression in Eq. 3.3.469 we can write this energy as the sum of
3N 6 independent contributions, each one only depending on one Qk coordinate
and we have thus simplified considerably the form of the vibrational energy.
Finally, if we define
cr ¼ 2pcxr =h ð3:3:50Þ
it is easy to verify that the coordinate

pffiffiffiffi
qr ¼ cr Q r ð3:3:51Þ
is dimensionless. We will use these dimensionless normal coordinates to describe

the vibrational motion in the following.
We have now defined the coordinates used in the standard approach, and during
this process we have made three choices:
• We have chosen the Eckart equations (3.3.33) to define the Euler angles h; /;
and v: The direct reason for making this choice was to minimize the Coriolis
coupling term (the third term on the right-hand side of Eq. 3.3.31) which con-
tains the major contribution to the rotation–vibration interaction.
9
That is, when we neglect expansion terms of third and higher order from Eq. 3.3.39.
• We have rotated the constant ag -vectors relative to the molecule-fixed axis

system so that the molecule-fixed axes are principal axes when the molecule is
in its equilibrium configuration. Thus we obtain that the products of inertia
vanish at equilibrium, Eq. 3.3.37.
• We have chosen the vibrational coordinates to be the normal coordinates Qk ;
thus eliminating terms coupling different vibrational modes of the molecule
from the vibrational kinetic energy Tvib and from the harmonic potential energy,
Eq. 3.3.44.
Clearly each choice was made in order to eliminate certain terms in the classical
energy. Since the eliminated terms in the kinetic energy are mass-dependent,
the resulting coordinates are also all mass-dependent in the sense defined in
Sect. 3.3.2.1.
3.3.5 Molecular Rotation–Vibration Energies
3.3.5.1 The Watson Hamiltonian
Watson [20] showed that when the classical Hamiltonian function described in
Sect. 3.3.2 is expressed in terms of the coordinates ðh; /; v; q1 ; q2 ; q3 ; . . .q3N6 Þ
and converted to quantum mechanical form using the Podolsky trick (see, for
example, Sect. 9.2 of [25]) the following quantum mechanical Hamiltonian for a
non-linear molecule is obtained:
X X

1 X
3N6
^ rv ¼ 1
H ^ â lab ^
Ja p Jb p
^b þ hc xr ^p2r
2 a¼x;y;z b¼x;y;z 2 r¼1
2 X
h
l þV ð3:3:52Þ
8 a¼x;y;z aa
^
Ja component of the total angular momentum along the molecule-fixed að¼ x; y; zÞ
axis.
P P3N6 ðaÞ pffiffiffiffiffiffiffiffiffiffiffiffiffi
â ¼
p h 3N6 f xs =xr qr ^ ps ; vibrational angular momentum.
ðaÞ
r¼1
ðaÞ Ps¼1 rsPN
frs ¼ fsr ¼ b;c¼x;y;z g¼1 eabc ‘gb;r ‘gc;s ; where the antisymmetric tensor
elements eabc are defined below and ‘gc;s is defined in Eq. 3.3.48.
Qr normal coordinate (Sect. 3.3.4)
P^ r ¼ i ho=oQr ; momentum conjugate to Qr :
pffiffiffiffi
qr ¼ cr Qr ; dimensionless normal coordinate defined in Eq. 3.3.51.
cr ¼ 2pcxr = h; Eq. 3.3.50.
^ r =ð pffiffiffiffi
^pr ¼ P h cr Þ; dimensionless conjugate momentum.
xr harmonic vibration wavenumber, Eq. 3.3.47.
V the potential energy function.

1
lab an element of the 3 3 matrix l ¼ ðI 0 Þ ; the inverse of the matrix ðI 0 Þ with
elements
X
3N6 X 3N6
1 3N6 X X aad adb
ðI 0 Þab ¼ dab Ie;aa þ aab
r Qr þ
r s
Qr Qs ð3:3:53Þ
r¼1
4 r¼1 s¼1 d¼x;y;z Ie;dd
where Ie;aa is an equilibrium moment of inertia in the principal axis system.

P PN 1=2

aab
r ¼ 2 c;d;¼x;y;z eac ebd g¼1 mg ag c ‘gd;r where mg is the mass of nucleus g
and the vector ag is defined in Eq. 3.3.26.
The elements of the unit antisymmetric tensor are defined as
exyz ¼ eyzx ¼ ezxy ¼ þ1

ð3:3:54Þ
exzy ¼ ezyx ¼ eyxz ¼ 1:
If any pair of indices in abc are equal, eabc ¼ 0:
3.3.5.2 The Zero-Order Hamiltonian
We have discussed in connection with Eq. 3.3.22 that it is advantageous to express

the Hamiltonian for the internal motion of a molecule in field-free space as
^ rv ¼ H ð0Þ
^ rot ^ ð0Þ þ H
^ rest ;
H þH vib ð3:3:55Þ
ð0Þ
^ rot
where H ¼ T^rot depends only on the rotational coordinates ðh; /; vÞ and their
conjugate momenta, H ^ ð0Þ ¼ T^vib þ V depends only on the vibrational coordinates
vib
ðq1 ; q2 ; . . .; q3N6 Þ and their conjugate momenta, and H ^ rest depends on both rota-
tional and vibrational coordinates and momenta.
ð0Þ
^ rot
We obtain H from Eq. 3.3.52 by setting all qr ¼ ^pr ¼ 0 (so that the molecule
is considered to be in its equilibrium configuration) and ignoring the small constant
P
term ð h2 =8Þ a¼x;y;z laa : The result is
ð0Þ
X X ð0Þ
^ rot
H ¼ ^
Ja lab ^Jb ð3:3:56Þ
a¼x;y;z b¼x;y;z
ð0Þ
where the superscript ‘‘(0)’’ on lab indicates that this matrix element is calculated
in the equilibrium configuration. It follows from the definition of the matrix l; and
ð0Þ
from Eq. 3.3.53 that lab ¼ dab =Ie;aa ; and so
^
J2 ^
Jy2 ^J 2
ð0Þ
^ rot
H ¼ x þ þ z : ð3:3:57Þ
2Ie;xx 2Ie;yy 2Ie;zz
^ ð0Þ the vibrational

We include in the zero-order vibrational Hamiltonian H vib
kinetic energy operator
1 X
3N6
T^vib ¼ p2r
hc xr ^ ð3:3:58Þ
2 r¼1
from Eq. 3.3.52 together with the harmonic potential energy
X
1 3N6 X
1 3N6
Vharmonic ¼ kk Q2k ¼ hc xr q2r ð3:3:59Þ
2 k¼1 2 k¼1
from Eq. 3.3.46. That is,
X
3N6
2
^ ð0Þ ¼ 1
H ^r þ q2r :
hc xr p ð3:3:60Þ
vib
2 k¼1
In order to determine H ^ rest in Eq. 3.3.55, the potential energy function V and the
elements lab of the inverse inertial tensor in Eq. 3.3.52 are expanded as Taylor
series in the dimensionless normal coordinates qr : The resulting Hamiltonian is a
polynomial (of infinite order) in the operators ^ Ja ; qr ; and ^pr : The coefficients in this
0
polynomial depend on the equilibrium moments of inertia Iaa ; on the zeta con-
ab
stants, on the constants ar and on the potential constants from V:
Traditionally, we solve the Schrödinger equation for H ^ rv by initially neglecting
T^rest : The resulting, zero-order Hamiltonian
ð0Þ
^ rv ð0Þ
^ rot ^ ð0Þ
H ¼H þH vib ð3:3:61Þ
ð0Þ
is the sum of two commuting terms, and we can obtain its eigenvalues Erv as
ð0Þ ð0Þ
Eð0Þ
rv ¼ Erot þ E vib ð3:3:62Þ
with corresponding eigenfunctions

ð0Þ ð0Þ
wð0Þ
rv ¼ wrot wvib ð3:3:63Þ
ð0Þ ð0Þ ð0Þ
^ rot
where ðErot ; wrot Þ is an eigenpair for H :
ð0Þ ð0Þ
^ rot ð0Þ ð0Þ
H wrot ¼ Erot wrot ð3:3:64Þ
ð0Þ ð0Þ ^ ð0Þ :
while ðEvib ; wvib Þ is an eigenpair for H vib
^ ð0Þ wð0Þ ¼ Eð0Þ wð0Þ :

H ð3:3:65Þ
vib vib vib vib
Once the zero-order Schrödinger equations in Eqs. 3.3.64 and 3.3.65 have been
solved, the eigenpairs ðErv;n ; wrv;n Þ (where n is an index numbering the solutions)
^ rv (which includes the initially neglected term T^rest ) are generated with the
for H
‘‘Ansatz’’
X ð0Þ ð0Þ
wrv;n ¼ cðnÞ
nvib ;nrot wvib;nvib wrot;nrot ð3:3:66Þ
nvib ;nrot
where we have introduced indices nrot and nvib to label the solutions of Eqs. 3.3.64
and 3.3.65, respectively. With this Ansatz, the molecular rotation–vibration
energies Erv;n are computed as the eigenvalues of a matrix with elements
Z
ð0Þ
ð0Þ
^ rv wð0Þ 0 wð0Þ 0 ds;
Hðn00vib ;n00rot Þ;ðn0vib ;n0rot Þ ¼ wvib;n00 wrot;n00 H vib;n rot;n ð3:3:67Þ
vib rot vib rot
ðnÞ
and the expansion coefficients cnvib ;nrot in Eq. 3.3.66 are components of the
eigenvectors of this matrix. In Eq. 3.3.67, the volume element ds is appropriate for
the total set of rotation–vibration coordinates ðh; /; v; q1 ; q2 ; . . .; q3N6 Þ:
To obtain a qualitative understanding of molecular spectra it is often sufficient
to consider only the solutions of the zero-order Schrödinger equations Eqs. 3.3.64
and 3.3.65 and to approximate the molecular rotation–vibration energies and
wavefunctions by Eqs. 3.3.62 and 3.3.63, respectively. So we discuss in the fol-
lowing sections the theory of the rigidly rotating molecule, corresponding to the
approximations in Eq. 3.3.64, and the harmonically vibrating molecule, corre-
sponding to the approximations in Eq. 3.3.65.
3.3.6 The Rigidly Rotating Molecule
The zero-order rotational Hamiltonian (i.e., the Hamiltonian for a rigidly rotating
molecule or, generally, for the rigid rotor) is given by Eq. 3.3.57:
^
J2 ^
Jy2 ^J 2
ð0Þ
^ rot
H ¼ x þ þ z
2Ie;xx 2Ie;yy 2Ie;zz
; ð3:3:68Þ
Jâ2 ^Jb2 ^Jc2
¼ þ þ
2Ie;aa 2Ie;bb 2Ie;cc
where the principal axes abc are a permutation of xyz; ordered such that
Ie;aa Ie;bb Ie;cc :
We now consider various rotor types with different patterns of rotational
energies:
• A linear rotor has Ie;aa ¼ 0 and Ie;bb ¼ Ie;cc : Examples of linear rotors are
hydrogen cyanide HCN, carbon dioxide CO2 ; and acetylene HCCH.
• A spherical top has Ie;aa ¼ Ie;bb ¼ Ie;cc : Examples of spherical tops are methane
CH4 ; sulfur hexafluoride SF6 ; and buckminsterfullerene C60 :
• A prolate symmetric top (Example: Methyl fluoride CH3 F) has Ie;aa \Ie;bb ¼
Ie;cc ; while an oblate symmetric top (Example: Benzene C6 H6 ) has Ie;aa ¼
Ie;bb \Ie;cc :
• An asymmetric top has Ie;aa \Ie;bb \Ie;cc : Examples of asymmetric tops are
water H2 O and formaldehyde H2 CO:
It is customary to introduce three rotational constants Ae ; Be ; and Ce ; expressed
either in frequency (MHz, GHz, . . .) or wavenumber ðcm1 Þ units. In frequency
units, we have
1 h2 h
Ae ¼ ¼ 2
; ð3:3:69Þ
h 2Ie;aa 8 p Ie;aa
1 h2 h
Be ¼ ¼ 2
; ð3:3:70Þ
h 2Ie;bb 8 p Ie;bb
and
1 h2 h
Ce ¼ ¼ : ð3:3:71Þ
h 2Ie;cc 8 p2 Ie;cc
In terms of the rotational constants, the zero-order Hamiltonian becomes

2 2 2
^ ð0Þ Jâ ^
Jb ^Jc
Hrot =h ¼ Ae þBe þCe ð3:3:72Þ

h
h h
where the three operators ^ h; a ¼ a; b; c; are dimensionless.

Ja =
3.3.6.1 Rigid Symmetric Top Eigenfunctions
We have already introduced the three operators ð^Jx ; J^y ; J^z Þ that represent the
components of the molecular angular momentum along the molecule-fixed axes
xyz: The analogous operators ð^JX ; ^
JY ; ^
JZ Þ represent the angular momentum com-
ponents along the laboratory-fixed axes XYZ: We define
J^2 ¼ ^
Jx2 þ ^
Jy2 þ ^
Jz2 ¼ ^
JX2 þ ^JY2 þ ^JZ2 ; ð3:3:73Þ
and it can be shown that the three operators J^2 ; J^z ; and ^JZ all commute, so that it
follows from elementary quantum mechanics that they have simultaneous eigen-
functions. We call these functions jJ; k; mi: They depend on the Euler angles
ðh; /; vÞ and are referred to as rigid symmetric top eigenfunctions or simply
symmetric top eigenfunctions. The jJ; k; mi functions have the properties
J^2 jJ; k; mi ¼ JðJ þ 1Þ h2 jJ; k; mi ð3:3:74Þ

with J ¼ 0; 1; 2; 3; . . .;
^
Jz jJ; k; mi ¼ k
h jJ; k; mi ð3:3:75Þ
with k ¼ J; J þ 1; J þ 2; . . .; J 1; J; and finally

^
JZ jJ; k; mi ¼ m
h jJ; k; mi ð3:3:76Þ
with m ¼ J; J þ 1; J þ 2; . . .; J 1; J:
3.3.6.2 Spherical Tops
For a spherical top molecule, we have Ie;aa ¼ Ie;bb ¼ Ie;cc and, therefore, Ae ¼
Be ¼ Ce : The zero-order rotational Hamiltonian in Eq. 3.3.72 becomes
ð0Þ 1 1
^ rot
H =h ¼ Be 2 Jâ2 þ ^ Jc2 ¼ Be 2 J^2 :
Jb2 þ ^ ð3:3:77Þ
h h
Obviously, it follows from Eq. 3.3.74 that in this case, the jJ; k; mi functions are
ð0Þ
^ rot
eigenfunctions for H =h with the eigenvalues
ð0Þ
Erot =h ¼ Be JðJ þ 1Þ; ð3:3:78Þ
where J ¼ 0; 1; 2; 3; . . .: The energy is independent of k and m; and so each level is
ð2J þ 1Þ2 -fold degenerate, since this is the number of possible values for k and m:
3.3.6.3 Symmetric Tops
For a prolate symmetric top, Ie;aa \Ie;bb ¼ Ie;cc so that Ae [ Be ¼ Ce : If we

identify the xyz axes with the abc axes so that xyz ¼ bca; the ‘‘special’’ axis a
ð0Þ
^ rot
becomes the z axis, and we can rewrite H =h as
ð0Þ
^ rot 1 2 1h i
H =h ¼ Ae 2 ^
Jz þ Be 2 ^Jx2 þ ^Jy2

h
h
1 ^2 1
¼ Ae 2 Jz þ Be 2 J^2 ^Jz2 ð3:3:79Þ

h
h
1 1
¼ Be 2 J^2 þ ðAe Be Þ 2 J^z2

h h
where we have used Eq. 3.3.73. We now have from Eqs. 3.3.74 and 3.3.75 that for
ð0Þ
^ rot
a prolate symmetric top, the jJ; k; mi functions are eigenfunctions for H =h; and
the eigenvalues are
ð0Þ
Erot =h ¼ Be JðJ þ 1Þ þ ðAe Be Þ k2 ð3:3:80Þ
where J ¼ 0; 1; 2; 3; . . .; and k ¼ J; J þ 1; J þ 2; . . .; J 1; J: A level with

k ¼ 0 is ð2J þ 1Þ-fold degenerate, ð2J þ 1Þ being the number of possible values
for m; and a level with jkj [ 0 is 2ð2J þ 1Þ-fold degenerate since the energy is
independent of the sign of k: We note that since Ae [ Be ; the energy contribution
ðAe Be Þ k2 is positive for jkj [ 0:
An oblate symmetric top has Ie;aa ¼ Ie;bb \Ie;cc and Ae ¼ Be [ Ce : Here, we
identify xyz ¼ abc so that now the special axis c becomes the z axis, and by a
derivation completely analogous to that for the prolate symmetric top, we can
show that also for the oblate top, the jJ; k; mi functions are eigenfunctions for
ð0Þ
^ rot
H =h with eigenvalues
ð0Þ
Erot =h ¼ Be JðJ þ 1Þ þ ðCe Be Þ k2 ; ð3:3:81Þ
where the degeneracies are as for the prolate top. Since Ce \Be ; the energy con-
tribution ðCe Be Þ k2 is negative for jkj [ 0:
3.3.6.4 Asymmetric Tops
For an asymmetric top, Ae [ Be [ Ce and we cannot simplify the zero-order

Hamiltonian in Eq. 3.3.72 in the way that we did for spherical and symmetric tops.
ð0Þ
It can be shown, however, that the two operators J^2 and ^JZ commute with H
^ rot and
so the associated quantum numbers J and m are ‘‘good’’ quantum numbers that can
be used to label the energies of the asymmetric top. Knowing this, we can expand
ð0Þ
the rotational wavefunction wrot;J;m of an asymmetric top in terms of rigid sym-
metric top eigenfunctions:
ð0Þ
X
J
ðJ;mÞ
wrot;J;m ¼ ck jJ; k; mi; ð3:3:82Þ
k¼J
with this Ansatz for the wavefunction, the molecular rotation–vibration energies
ð0Þ
Erot are computed as the eigenvalues of a matrix with elements
D E
^ ð0Þ 00
Hk0 ;k00 ¼ J; k0 ; mH rot J; k ; m ; ð3:3:83Þ
since J and m are ‘‘good’’ quantum numbers we only require matrix elements
diagonal in them. It can be shown that for Hk0 ;k00 to be non-vanishing, we must
either have k00 ¼ k0 ¼ k and

h2
1 1 2k2
Hk;k ¼ þ JðJ þ 1Þ k2 þ ; ð3:3:84Þ
4 Ie;xx Ie;yy Ie;zz
or k00 ¼ k0 2 ¼ k 2 and

2 1
h 1 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Hk;k2 ¼ JðJ þ 1Þ kðk 1Þ
8 Ie;xx Ie;yy
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
JðJ þ 1Þ ðk 1Þðk 2Þ: ð3:3:85Þ
In using Eqs. 3.3.84 and 3.3.85 we can identify the axis xyz with the principal axes
abc as we like. However, for a near prolate symmetric top with Ie;aa \Ie;bb Ie;cc
it is most convenient to use xyz ¼ bca since this minimizes the off-diagonal matrix
elements in Eq. 3.3.85. For similar reasons, for a near-oblate rotor with Ie;aa
Ie;bb \Ie;cc ; we use xyz ¼ abc:
ð0Þ
^ rot
The matrix representation of H in the basis set of jJ; k; mi functions is block
diagonal in J and m; and the matrix elements do not depend on m; so initially we
can say that we must diagonalize one block, of dimension 2J þ 1; for each J value.
However, Eqs. 3.3.84 and 3.3.85 show that each such block separates further into
one block of basis functions with even k values, and another block of basis
functions with odd k values. All matrix elements connecting an even k value with
an odd k value obviously vanish. It can be shown that if, instead of the basis
functions jJ; k; mi; we use the so-called Wang basis
1
jJ; K; m; i ¼ pffiffiffi ðjJ; K; mi jJ; K; miÞ ð3:3:86Þ
2
where K [ 0; and
jJ; 0; m; þi ¼ jJ; 0; mi; ð3:3:87Þ
we obtain a further factorization into four blocks that we can label Eþ ; E ; Oþ ;

and O : A block labeled E(O) contains basis functions with even(odd) K values
only, and the þðÞ superscript signifies that the block contains only Wang
functions with þðÞ from Eqs. 3.3.86 and 3.3.87.
As a measure for the ‘‘asymmetry’’ of an asymmetric top we use the asymmetry
parameter
2Be Ae Ce
j¼ : ð3:3:88Þ
Ae Ce
A prolate symmetric top has j ¼ 1; and an oblate symmetric top has j ¼ þ1:
We now imagine a progression of asymmetric tops, all with the same values of
Ae and Ce ; but with Be varying continuously between Ce and Ae (or, equivalently,
with j varying between 1 and þ1). For Be ¼ Ce ; or j ¼ 1; we are at the
prolate limit, where the energy is given by Eq. 3.3.80 with k ¼ ka ; where ka is the
projection, in units of
h; of the angular momentum onto the a axis. For a given J;
the energy increases with increasing jka j as discussed above. On the other hand, for
Be ¼ Ae ; or j ¼ þ1; we have reached the oblate limit, where the energy is given
by Eq. 3.3.81 with k ¼ kc ; where kc h is the projection of the angular momentum
250 J =5
Ka Kc
5 55,0 0
1
200 55,1
2
54,1
Ered [cm-1]
150 4 3
54,2
53,2
100
52,3
3 53,3 4
52,4
50 51,4
2
1 51,5
0 0 5
50,5
1 2 3 4 5 6 7 8 9 10
B=C B [cm-1] B=A
Fig. 3.8 The energy levels of a rigid asymmetric top molecule are plotted for J ¼ 5; with
Ae ¼ 10 cm1 and Ce ¼ 1 cm1 ; but with Be varying continuously between Ce and Ae : The
vertical axis represents the reduced energy which is measured from the lowest energy levels for
the J ¼ 5 state
onto the c axis. For a given J; the energy decreases with increasing jkc j: If we plot
the energies of the progression of asymmetric tops as functions of Be varying from
Ce to Ae ; or j varying from 1 to þ1; we obtain a correlation diagram, Fig. 3.8.
Each energy curve extends from the prolate limit to the oblate limit, and so
obviously any energy level of an asymmetric top correlates, in the prolate limit,
with a prolate symmetric top level with a defined value of Ka ¼ jka j and, in the
oblate limit, with an oblate symmetric top level with a defined value of Kc ¼ jkc j:
In addition, it can be shown that for any value of j; where 1\j\ þ 1; the rigid
asymmetric top with a given J-value has 2J þ 1 different energies. Consequently,
the energy curves in the correlation diagram cannot cross. At a j-value where two
of them crossed, there would only be 2J different energies and this cannot happen.
Because of this non-crossing rule, we can unambiguously correlate the energies in
the prolate and oblate limits for a given J value. For example, the highest energy of
the prolate rotor, with Ka ¼ J; must necessarily correlate with the highest level of
the oblate rotor with Kc ¼ 0: Similarly, the lowest level of the prolate rotor, with
Ka ¼ 0; must correlate with the lowest level of the oblate rotor with Kc ¼ J: In
general, we can label any level of the asymmetric rotor as JKa Kc : When, for a given
J value, we
sort the
energies

EðJKa
Kc Þ in ascending

order,
we
have EðJ0 J Þ\
EðJ1 J Þ\E J1 ðJ1Þ \E J2 ðJ1Þ \E J2 ðJ2Þ \E J3 ðJ2Þ \E J3 ðJ3Þ \ \

E JðJ2Þ3 \E JðJ2Þ2 \E JðJ1Þ2 \E JðJ1Þ1 \EðJJ 1 Þ\EðJJ 0 Þ: For any level,
Ka þ Kc ¼ J or Ka þ Kc ¼ J þ 1:
3.3.6.5 Linear Rotors
The theory outlined in the preceding part of this section is concerned with non-
linear molecules, i.e., molecules with a non-linear equilibrium structure (see Sect.
3.3.2.2). For such a molecule, we need three Euler angles ðh; /; vÞ to describe the
rotation. The Euler angle v; however, describes the rotation of the molecule about
the molecule-fixed z axis. For a rigid linear molecule, all nuclei are located on the z
axis, and the rotation about this axis cannot physically take place. Thus the angle v
becomes redundant and we have only the two angles ðh; /Þ which are the usual
polar/azimuthal angles defining the direction of the molecular axis relative to
the XYZ axes. As mentioned above, a rigid linear molecule has Ie;aa ¼ 0 and
Ie;bb ¼ Ie;cc : Its rotational Hamiltonian is
" 2 2 #
ð0Þ
^
Jb ^Jc
^ rot =h ¼ Be
H þ : ð3:3:89Þ

h h
ð0Þ
^ rot are the well-known spherical harmonics YJm ðh; /Þ
The eigenfunctions of H
which are related to the jJ; k; mi functions:
pffiffiffiffiffiffi
YJm ðh; /Þ ¼ 2p jJ; 0; mi ð3:3:90Þ
where, as usual, where J ¼ 0; 1; 2; 3; . . . and m ¼ J; J þ 1; J þ 2; . . .;
J 1; J: The YJm ðh; /Þ functions are simultaneous eigenfunctions of J^2 and ^JZ
ð0Þ
^ rot
with the eigenvalues JðJ þ 1Þ h2 and m
h; respectively. The eigenvalue of H
corresponding to the eigenfunction YJm ðh; /Þ is
ð0Þ
Erot =h ¼ Be JðJ þ 1Þ: ð3:3:91Þ
The energy is independent of m; and so each level is 2J þ 1-fold degenerate.
3.3.7 The Harmonically Vibrating Molecule
The zero-order vibrational energies for a non-linear molecule are the eigenvalues
of the Hamiltonian from Eq. 3.3.60:
X
3N6
2
^ ð0Þ ¼ 1
H ^r þ q2r
hc xr p ð3:3:92Þ
vib
2 r¼1
where qr is a dimensionless normal coordinate from Eq. 3.3.51 and ^pr ¼ io=oqr
is its dimensionless conjugate momentum. The harmonic vibration wavenumber
xr is defined in Eq. 3.3.47.
3.3.7.1 Asymmetric Tops
Whereas a rigid asymmetric top has a more complicated rotational energy pattern
than a symmetric top, a spherical top or a linear rotor, it has a simpler vibrational
spectrum than any of these molecules. For an asymmetric top, the matrix GF
discussed in Sect. 3.3.4 has 3N 6 different eigenvalues kk ; and the zero-order
vibrational energy is given by
X
3N6 X
3N6
ð0Þ 1
Evib =ðhcÞ ¼ Eð0Þ
vr =ðhcÞ ¼ x r vr þ ð3:3:93Þ
r¼1 r¼1
2
where each of the quantum numbers vr ¼ 0; 1; 2; 3; . . .: Here, we have that Eð0Þ

vr ¼
1
h c xr ðvr þ 2Þ is an eigenvalue obtained by solution of the harmonic oscillator

2
pr þ q2r /ð0Þ
hc xr ^ ð0Þ ð0Þ
vr ðqr Þ ¼ Evr /vr ðqr Þ; ð3:3:94Þ
where /ð0Þ
vr ðqr Þ is a standard one-dimensional harmonic oscillator eigenfunction
(see, for example, [25]. The zero-order vibrational wavefunction for an asym-
metric top is
ð0Þ
wvib ¼ /ð0Þ ð0Þ ð0Þ ð0Þ
v1 ðq1 Þ /v2 ðq2 Þ /v3 ðq3 Þ /v3N6 ðq3N6 Þ: ð3:3:95Þ
3.3.7.2 Linear Rotors
As discussed in Sect. 3.3.6.5, the rotation of a rigid linear molecule is described by

two Euler angles ðh; /Þ: The rotation about the molecular axis cannot take place
when the molecule is linear. When the molecule vibrates out of the linear con-
figuration, however, this rotation becomes possible, and it is customarily described
as a vibration so that for a molecule with N nuclei and a linear equilibrium
configuration, we have 3N 5 normal coordinates ðq1 ; q2 ; q3 ; . . .; q3N6 ; q3N5 Þ:
Each of these normal coordinates qr has an associated harmonic vibrational
wavenumber xr ; but not all of these harmonic vibrational wavenumbers are dif-
ferent. Of the 3N 5 normal coordinates, N 1 describe the stretching of the
N 1 bonds of the molecule and these will be different, just like the xr -values of
an asymmetric top. The remaining 2N 4 xr -values, however, describe the
bending motion at each of the N 2 ‘‘inner’’ nuclei of the molecule (no bending
motion takes place at the two terminal nuclei). These 2N 4 xr -values are pair-
wise equal and form N 2 pairs. This is because at each inner nucleus, one normal
coordinate describes the bending motion taking place in the xz plane of the mol-
ecule-fixed axis system xyz (where the z axis is the molecular axis and the direction
of the x axis is arbitrarily chosen), and another normal coordinate describes the
bending motion taking place in the yz plane. These two normal coordinates
necessarily have the same value of xr : Hence, a linear molecule has ðN 1Þ þ

ðN 2Þ ¼ 2N 3 different xr -values.
The two-dimensional bending vibration at a given inner nucleus is described by
two normal coordinates which, in an obvious notation, we can write as ðqsx ; qsy Þ
where the index s numbers the vibrational mode. We have xsx ¼ xsy ¼ xs : In
principle, we can use product wavefunctions /ð0Þ ð0Þ
vsx ðqsx Þ /vsy ðqsy Þ as zero-order
vibrational wavefunctions; as before /ð0Þ v ðqÞ is a standard one-dimensional har-
monic oscillator eigenfunction. The contribution to the vibrational energy from
vibrational mode s is thus

ð0Þ 1 1
Evsx ;vsy =ðhcÞ ¼ xs vsx þ þ xs vsy þ
2 2

¼ xs vsx þ vsy þ 1 ¼ xs ðvs þ 1Þ ð3:3:96Þ
where vs ¼ vsx þ vsy : The zero-order vibrational state with the energy Eð0Þ vsx ;vsy
is ðvs þ 1Þ-fold degenerate since there are vs þ 1 different values of ðvsx ; vsy Þ
½¼ ðvs ; 0Þ; ðvs 1; 1Þ; ðvs 2; 2Þ; . . .; ð1; vs 1Þ; ð0; vs Þ with a common value of vs :
Instead of the ðvs þ 1Þ product wavefunctions /ð0Þ ð0Þ
vsx ðqsx Þ /vs vsx ðqsy Þ with a
common value of vs ; it is customary to use as zero-order vibrational wavefunc-
tions, obtained as linear combinations of the product functions. These new func-
ð0Þ
tions are called /vs ;‘s ðqs ; vs Þ where ‘s ¼ vs ; vs þ 2; vs þ 4; . . .; vs 2; vs ; and
the coordinates qs and vs are such that
qsx ¼ qs cos vs and qsy ¼ qs sin vs : ð3:3:97Þ
The quantum number ‘s is the projection, in units of h; onto the molecular axis of
the angular momentum generated by the vibrational mode s (which, as mentioned
above, also describes the rotation about the molecular axis). The detailed form of
ð0Þ
the functions /vs ;‘s ðqs ; vs Þ is given, for example, by Bunker and Jensen [25].
With the definitions introduced here, the zero-order vibrational energy for a
linear rotor is given as
X
N 1 2N3
X
ð0Þ 1
Evib =ðhcÞ ¼ xr vr þ þ xs ðvs þ 1Þ ð3:3:98Þ
r¼1
2 s¼N
and the zero-order vibrational wavefunction is
ð0Þ
Y
N 1 Y
2N3
ð0Þ
wvib ¼ /ð0Þ
vr ðqr Þ /vs ;‘s ðqs ; vs Þ: ð3:3:99Þ
r¼1 s¼N
In the harmonic (zero-order) approximation, the vibrational energy depends

only on the v quantum numbers; it is independent of the ‘ quantum numbers.
However, to characterize completely the vibrational state of a linear rotor, we
require the quantum numbers v1 ; v2 ; v3 ; . . .; vN1 ; vN ; . . .; v2N3 and ‘N ; . . .; ‘2N3 ;
whereas for an asymmetric top with the same number of nuclei, we use the
quantum numbers v1 ; v2 ; v3 ; . . .; v3N6 :
3.3.7.3 Symmetric Tops
A symmetric top molecule has non-degenerate normal vibrations, analogous to the

stretching vibrations of the linear rotor, and doubly degenerate normal vibrations,
analogous to the bending vibrations of the linear molecule. A non-degenerate
normal vibration is described by a single normal coordinate qr and the eigen-
function /ð0Þ vr ðqr Þ of the corresponding one-dimensional harmonic oscillator. A
doubly degenerate normal vibration is described by two normal coordinates
ðqsa ; qsb Þ; say, or by the equivalent coordinates ðqs ; vs Þ defined by analogy with Eq.
3.3.97.
If a symmetric top molecule with N nuclei has n1 non-degenerate normal
vibrations and n2 doubly degenerate normal vibrations, then we have n1 þ 2n2 ¼
3N 6; but unlike the linear rotor, there is no simple rule to determine n1 and n2 :
For example, the symmetric top CH3 F has three non-degenerate vibrations and
three doubly degenerate ones.
By analogy with the linear rotor, we obtain for the symmetric top that the zero-
order vibrational energy is expressed as
X
n1 nX 1 þn2
ð0Þ 1
Evib =ðhcÞ ¼ x r vr þ þ x s ð v s þ 1Þ ð3:3:100Þ
r¼1
2 s¼n þ1 1
and the zero-order vibrational wavefunction is
ð0Þ
Y
n1 nY
1 þn2
ð0Þ
wvib ¼ /ð0Þ
vr ðqr Þ /vs ;‘s ðqs ; vs Þ: ð3:3:101Þ
r¼1 s¼n1 þ1
To characterize completely the vibrational state of a symmetric top, we require the

quantum numbers v1 ; v2 ; v3 ; . . .; vn1 ; vn1 þ1 ; . . .; vn1 þn2 and ‘n1 þ1 ; . . .; ‘n1 þn2 but in the
zero-order approximation, the energy depends on the v quantum numbers
exclusively.
3.3.7.4 Spherical Tops
The expression for the zero-order vibrational energy of a spherical top is, in
principle, analogous to Eqs. 3.3.98 and 3.3.100. However, spherical tops have
normal vibrations of higher degeneracy than two. Methane CH4 ; for example, has
triply degenerate vibrations described by three normal coordinates ðqsa ; qsb ; qsc Þ;
all corresponding to the same value of xs : Such a coordinate triple will produce a
contribution of h c xs ðvs þ 3=2Þ to the zero-order vibrational energy.
3.3.8 Effective Hamiltonians
3.3.8.1 The Contact Transformation
In connection with Eqs. 3.3.61–3.3.67, we have outlined how the rotation–vibra-

tion Schrödinger equation
^ rv wrv;n ¼ Erv;n wrv;n
H ð3:3:102Þ
can be solved by expanding the wavefunction wrv;n in terms of known basis

functions [Eq. 3.3.66] and thus converting the Schrödinger equation to a matrix
eigenvalue equation. In Eq. 3.3.102, H ^ rv is the Watson Hamiltonian from Eq.
3.3.52 and Erv;n is the eigenvalue corresponding to the eigenfunction wrv;n with the
running index n numbering the solutions.
The matrix representation of H^ rv (with elements given by Eq. 3.3.67), which we
obtain by inserting Eq. 3.3.66 in Eq. 3.3.102, can be diagonalized numerically in a
so-called variational calculation. In spectroscopic theory, however, it is tradi-
tionally solved by perturbation methods, and we outline one such method here, the
so-called contact transformation (Maes and Amat [26]).
We can write a unitary operator as
^
T^ ¼ ei S ð3:3:103Þ
where ^
S is a hermitian, initially arbitrary operator and the exponential is expressed
as a Taylor expansion
^ 1
^ 2
T^ ¼ ei S ¼ 1 þ i ^
Sþ iS þ ð3:3:104Þ
2!
We use the operator T^ to form a new transformed Hamiltonian
0
^ rv
H ^ rv T^ 1 ¼ ei ^S H
¼ T^ H ^ rv ei ^S : ð3:3:105Þ
^ rv by solving the Schrödinger
We obtain the eigenvalues and eigenfunctions for H
equation in Eq. 3.3.102 which we rewrite as
^ rv T^ 1 T^ wrv;n ¼ Erv;n T^ wrv;n :
T^ H ð3:3:106Þ
This shows that the Schrödinger equation for H ^ rv

0
is
0

^ rv
H T^ wrv;n ¼ Erv;n T^ wrv;n : ð3:3:107Þ
Hence H ^ rv
^ rv and H 0 ^ rv
have identical eigenvalues but the eigenfunctions for H 0
are
given by
w0rv;n ¼ T^ wrv;n ð3:3:108Þ

^ rv :
where wrv;n is an eigenfunction for H
The basic idea of a contact transformation is now to determine the operator ^S
such that
ð0Þ ð0Þ ^
wvib;nvib wrot;nrot ¼ T^ wrv;n ¼ ei S wrv;n ð3:3:109Þ
ð0Þ ð0Þ ^ rv
0
where wvib;nvib wrot;nrot is a basis function from Eq. 3.3.66. Ideally, H is diagonal in
ð0Þ ð0Þ
the basis of product functions wvib;nvib wrot;nrot and the rotation–vibration energies
are then simply given by the diagonal matrix elements
D 0 ð0Þ E
ð0Þ ð0Þ ð0Þ
Erv;n ¼ wvib;nvib wrot;nrot H
^ rv w
vib;nvib wrot;nrot : ð3:3:110Þ
The procedure of a contact transformation is described, for example, in [16]. In

practice, the operator ^
S is expanded as a series of terms
^
S¼^
S0 þ ^
S1 þ ^
S2 þ ^
S3 þ ^S4 þ ; ð3:3:111Þ
where the significance of the term Sk diminishes as k increases. Equation 3.3.110

applies in the limit of k ! 1; it involves an approximation that will become
increasingly better as more terms are added in Eq. 3.3.111.
3.3.8.2 The Watsonian
^ rv
In general, we cannot in practice obtain a contact-transformed Hamiltonian H 0
that is exactly diagonal in a basis of functions

ð0Þ ð0Þ
wvib;nvib wrot;nrot :
Normally, contact transformations are used to reduce the matrix elements, which
are off-diagonal in nvib ; to a degree that these matrix elements can be neglected.
^ rv
0 ð0Þ
Hence, we can average H over one vibrational basis function wvib;nvib to obtain the
effective rotational Hamiltonian
D 0 ð0Þ E
~ rot ¼ wð0Þ H ^ rv w ð3:3:112Þ
H vib;nvib vib;nvib :
Bunker and Jensen [25] have advocated the use of the term Watsonian for the
Hamiltonian H~ rot for any molecule; the development and use of this type of
Hamiltonian have benefited greatly from the work of Watson [27, 28]. The
Watsonian is an expansion in the rotational operators ^Ja (a ¼ x; y and z) and for a
non-planar asymmetric top molecule it has the general form [28] (see also Sarka
et al. [29])
~ rot ¼ B200 J^2 þ B020 ^

H Jz2 þ T400 J^4 þ T220 J^2 ^Jz2 þ T040 ^Jz4
1 h i
þ B002 þ T202 J^2 þ T022 ^ Jþ Þ2 þ ð^J Þ2
Jz2 ; ð^
2 þ
^ 4 ^ 4
þ T004 ðJþ Þ þ ðJ Þ ; ð3:3:113Þ
where we have truncated the expansion in the Jâ after the second order (i.e.,
quartic) terms. The quantities Bijk and Tijk in Eq. 3.3.113 are molecular parameters,
and ^ J ¼ ^ Jx i^
Jy : For two operators A ^ and B;
^ the plus commutator ½A; ^ B
^ ¼
þ
^B
A ^þB ^ The eigenvalues of H
^ A: ~ rot are obtained by diagonalization of a Hamil-
tonian matrix set up in a basis of symmetric top functions jJ; k; mi exactly as
described for the Hamiltonian of the rigid asymmetric rotor in Sect. 3.3.6.4.
If we know (an analytical representation of) the potential energy surface and the
equilibrium geometry of a molecule, we can calculate the values of the quantities
Bijk and Tijk in Eq. 3.3.113. However, Watsonians are mostly used in least squares
fittings to experimentally observed transition frequencies; the fittings aim at
determining experimental values for parameters such as Bijk and Tijk (and of the
coefficients of higher order terms in the ^Ja which we neglect here). Initial problems
with determining simultaneously the values of all six Tijk in Eq. 3.3.113 from
experimental data were explained by Watson [27, 28]. There are infinitely many
Hamiltonians H ~ rot ; all of the general form given in Eq. 3.3.113 but with different
values for the Tijk parameters, which have the same eigenvalues and which are
equally well suited for fitting a given set of experimental data. These Hamiltonians
are all connected by contact transformations. Watson [27, 28]: solved this problem
by introducing constraints on the parameters Tijk : For example, for a general H ~ rot
with non-zero values for all six Tijk it is possible to devise a rotational contact
transformation (see also Sarka and Demaison [30]) that produces a transformed
~ rot with T004 ¼ 0: In a notation more familiar than that in Eq. 3.3.113, the
H
resulting Hamiltonian is given by

1 ^ 1
H rot ¼ ð X þ Y Þ J þ Z ð X þ Y Þ ^
~ 2
Jz2 DJ J^4 DJK J^2 ^Jz2 DK ^Jz4
2 2

1 ^ 2 ^ 2 ^ 2 ^ 2
þ ð X Y Þ dJ J dK Jz ; ðJþ Þ þ ðJ Þ : ð3:3:114Þ
4 þ
This is known as Watson’s A-reduced Hamiltonian. In Eq. 3.3.114, X; Y; and Z are

the effective rotational constants of the molecule for the vibrational state in
question [associated with the molecule-fixed axes x; y; and z; respectively], and
the parameters DJ ; DJK ; DK ; dJ ; and dK are centrifugal distortion constants.
The application of a contact transformation to impose selected constraints on the
parameters of a Hamiltonian is called a reduction of the Hamiltonian. Another way
of reducing the Watsonian leads to Watson’s S-reduced Hamiltonian which has
T022 ¼ 0: Reductions are treated in detail in the original papers by Watson
[27, 28]: and in the review articles [29, 30].
For a prolate symmetric top, in the absence of resonances, the eigenvalues of

the Watsonian for a given vibrational state can be written (in cm1 ) as
TVLJK ¼ GVL þ FVLJK ð3:3:115Þ
where
X X
dr dr dr 0
GVL ¼ x r vr þ þ Xrr0 vr þ vr0 þ
r
2 r r0
2 2
X
þ gtt0 lt lt0 þ ð3:3:116Þ
t t0
and
X
FVLJK ¼ BV JðJ þ 1Þ þ ðAV BV ÞK 2 2 Afzta;tb lt k
V
t
DJ J 2 ðJ þ 1Þ2 DJK JðJ þ 1ÞK 2 DK K 4

X X
þ gJt lt JðJ þ 1Þ k þ gKt lt k3 þ ð3:3:117Þ
t t
with
X
dr
BV ¼ Be aBr vr þ þ ; ð3:3:118Þ
r
2
X
dr
AV ¼ Ae aAr vr þ þ ; ð3:3:119Þ
r
2
and
X
dr
Afzta;tb ¼ Ae fzta;tb aAf
r
t
v r þ þ : ð3:3:120Þ
V
r
2
The expression for an oblate symmetric top is obtained by replacing Ae and AV by

Ce and CV ; respectively. In Eqs. 3.3.115–3.3.120 V and L are shorthand notations
for all vr and lr quantum numbers, respectively, of the symmetric top, see Sect.
3.3.7.3. If mr is a non-degenerate normal mode, dr = 1, whereas if mr is doubly
degenerate, dr ¼ 2: The parameter xr is defined in Eq. 3.3.47 for the normal mode
mr : The rotational constants Ae and Be are defined in Eq. 3.3.69–3.3.71; the fzta;tb are
Coriolis coupling constants. Further parameters are Xrr0 ; gtt0 ; the centrifugal dis-
tortion constants DJ ; DJK ; and DK ; and the constants gJt ; gKt ; aBr ; aAr ; and aAf
r : They
t
all depend on the equilibrium geometry (equilibrium bond lengths and angles) of
the molecule and on the parameters determining the potential energy function; see,
for example [16].
For linear rotor, in the absence of resonances, the eigenvalues of the Watsonian
for a given vibrational state can be written (in cm1 ) as
TVLJ‘ ¼ GVL þ FV LJ‘ ð3:3:121Þ
where the expression for GVL is identical to that for a symmetric top molecule, Eq.
3.3.116, and
FVLJ‘ ¼ BV ½JðJ þ 1Þ ‘2 DV ½JðJ þ 1Þ ‘2 2

þ HV ½JðJ þ 1Þ ‘2 3 þ MV ½JðJ þ 1Þ ‘2 4 þ ð3:3:122Þ
P
where ‘ ¼ t ‘t : The parameters BV ; DV ; HV ; MV ; . . . are defined by equations
analogous to Eqs. 3.3.118 and 3.3.119.
If zero-order vibrational states are close in energy, a resonance occurs. For
example, for a symmetric top or a linear rotor, this would mean that for two
different sets of vr values, V and V 0 ; we have GVL GV 0 L0 : In this case the
Watsonian cannot be applied as outlined here. The theory must be extended so that
the molecular energies are obtained as the eigenvalues of a matrix representation
of a Watsonian common for the two vibrational states VL and V 0 L0 : Such reso-
nances and their theoretical description are discussed in detail by, for example,
[16, 19, 30].
3.3.9 Molecular Symmetry
In the preceding sections we have briefly described the rotation–vibration states of

different types of molecules, and we now continue to consider the way these states
can be connected by transitions accompanied by absorption and emission of
electromagnetic radiation energy. Luckily, it turns out that there are very restrictive
selection rules that determine which state pairs can be connected by a transition of a
given type. This is lucky since if it were not so, a given molecule would absorb or
emit light at any frequency or wavenumber, and rotation–vibration spectra could
not be used to identify molecules for example in interstellar space.
3.3.9.1 Symmetry and Symmetry Groups
The selection rules are derived by consideration of molecular symmetry and so we

give here a brief discourse on this subject. We discuss the following symmetry
types:
1. Permutations of identical nuclei, whereby two or more identical nuclei replace
each other.
2. Spatial inversion, where all nuclei and electrons are inverted through the
molecular center of mass.
3. Rotational symmetry, i.e., the fact that owing to the isotropy of space, we can
rotate an isolated molecule arbitrarily (so that all particles, nuclei and electrons,
in the molecule are rotated in the same manner) without changing the physical
situation.
Each of the symmetry types considered above gives rise to symmetry operations
that commute with the molecular Hamiltonian, see [25, 31, 32].
The mathematical tool that is used to describe the permutation–inversion
symmetry of a molecule is called group theory or, more precisely, representation
theory. An introduction is given by Bunker and Jensen [25, 31]. A group theo-
retical analysis of a molecule, in terms of permutation–inversion symmetry,
involves the following steps:
i. The possible permutations of identical nuclei, the inversion operation and all
operations that can be obtained by combining a possible permutation of
identical nuclei with the inversion operation. These operations constitute a
(symmetry) group for the molecule. Groups are mathematical objects that
satisfy certain group axioms (see, for example, [25], Sect. 1.3). The group
just mentioned is called the CNPI group for the molecule (Complete Nuclear
Permutation Inversion Group, see [25], Chap. 2). In most cases we do not use
the complete CNPI group as symmetry group. Normally it is sufficient to use
only a subgroup of the CNPI group. This smaller group is called the
Molecular Symmetry Group (MS Group, see [25], Chap. 3). There are also
cases where the MS group and the CNPI group are identical.
ii. When the MS group has been determined, its so-called irreducible repre-
sentations can be constructed according to unambiguous, mathematical rules.
For most groups, these irreducible representations are already known and can
be found in textbooks such as [25, 31] in the form of so-called character
tables. An irreducible representation can be viewed as a set of rules that
determine how a wavefunction (that belongs to this irreducible representa-
tion) transforms under the operations of the symmetry group. The irreducible
representation dictates how the wavefunction changes,10 when the operations
in the symmetry group (i.e., permutations of identical nuclei and/or inver-
sion) are applied to it. A given group will normally have a relatively small
number of irreducible representations.
iii. The central statement is now that any wavefunction of a given molecule must
belong to an irreducible representation of the symmetry group. When we
investigate how the various wavefunctions of the molecule change under
symmetry operations (i.e., we investigate how they transform), there are only
a very limited number of possibilities for the possible transformation prop-
erties. These possibilities correspond to the irreducible representations of the
10
For degenerate irreducible representations, the transformation properties of the generating
wavefunctions are not uniquely determined, but the matrices determining the transformation
properties of the wavefunctions are related by similarity transformations.
symmetry group. For a given molecular state, we can use the irreducible
representation generated by its wavefunction to label this wavefunction.
Further, representation theory can be used to simplify the calculation of
molecular energies, that is, to save computer time and memory.
Obviously we are giving here a very rough sketch of molecular symmetry and
its applications. We will describe the application of these ideas to the water
molecule as an example:
The Symmetry of H2 O
The water molecule H2 O has two identical protons. Consequently, for this mol-
ecule the only possible permutation of identical nuclei is the interchange of these
two nuclei. We label the protons as 1 and 2 (and the oxygen nucleus as 3) and
denote the interchange of 1 and 2 as (12) (this operation is read as ‘‘one-two’’ and
never as ‘‘twelve’’). Another possible symmetry operation is the inversion E
which inverts the two protons, the oxygen nucleus and the ten electrons in the
center of mass of the molecule. Also, we have the combined operation ð12Þ
(‘‘one-two-star’’) which is carried out by first applying E

and then (12) (or first
(12) and then E
; the sequence plays no role) Finally, for formal reasons the
symmetry group must contain a so-called neutral element E: This element does
nothing; it leaves the system in the initial condition.
The four symmetry operations listed above form a group, a mathematical object
satisfying a set of group axioms (see, for example, [25]). The group that we
consider here is customarily called C2v (M):
C2v ðMÞ ¼ fE; ð12Þ; E

; ð12Þ
g: ð3:3:123Þ
It can be shown that this group has four irreducible representations called
A1 ; A2 ; B1 und B2 ; respectively. The wavefunctions w of the water molecule are
such that when we apply a group operation R [= E; ð12Þ; E
; or ð12Þ
] to w; we
have
Rw ¼ vC ðRÞw; ð3:3:124Þ
where the constant factor vC ðRÞ depends on the symmetry operation R and on the
irreducible representation C [= A1 ; A2 ; B1 or B2 ]. The four irreducible represen-
tations correspond to four kinds of water wavefunctions:
A1 : The function is unchanged by all four symmetry operations E; ð12Þ; E
; ð12Þ
:
A2 : The function is unchanged by the operation E (as all functions are). It also
unchanged by (12) but changes sign under E
: In consequence, the function
must necessarily change sign under the operation ð12Þ
¼ ð12ÞE
:
B1 : The function is unchanged by the operation E: It changes sign under both of
the operations (12) and E
; and so it stays unchanged under ð12Þ
:
Table 3.2 The character E (12) E* (12)*

table for the molecular E C2 r r0
symmetry group C2v (M)
A1 1 1 1 1
A2 1 1 -1 -1
B1 1 -1 -1 1
B2 1 -1 1 -1
Table 3.3 The character table of the CNPI group G12

E (123) (12) E* (123)* (12)*
(132) (23) (132)* (23)*
(31) (31)*
Aþ
1 1 1 1 1 1 1
A
1 1 1 1 -1 -1 -1
Aþ
2 1 1 -1 1 1 -1
A
2 1 1 -1 -1 -1 1
Eþ 2 -1 0 2 -1 0
E 2 -1 0 -2 1 0
B2 : The function is unchanged by the operation E: It changes sign under (12) but is
unchanged by E
: Thus, it changes sign under ð12Þ
:
These definitions imply that for example vA1 ðEÞ ¼ vA1 ðð12ÞÞ ¼ vA1 ðE
Þ ¼
v ðð12Þ
Þ ¼ 1 and vA2 ðEÞ ¼ vA2 ðð12ÞÞ ¼ 1; but vA2 ðE
Þ ¼ vA2 ðð12Þ
Þ ¼ 1: The
A1
quantities vC ðRÞ are called the characters of the irreducible representations. For a
given group, they are compiled in the character table. Table 3.2 gives the char-
acter table for the group C2v ðMÞ:
Each possible wavefunction of a water molecule belongs to one of the irre-
ducible representations A1 ; A2 ; B1 oder B2 : Thus, the wavefunction is labeled by
‘‘its’’ irreducible representation.
The Symmetry of PH3
The CNPI group of the PH3 molecule is a group of 12 elements that we call G12 :
G12 ¼ fE; ð123Þ; ð132Þ; ð12Þ; ð23Þ; ð31Þ;

E
; ð123Þ
; ð132Þ
; ð12Þ
; ð23Þ
; ð31Þ
g ð3:3:125Þ
The rotation–vibration-electronic energy levels of the PH3 molecule (neglecting

nuclear spin) can be labeled with the irreducible representation labels of the group
G12 : The character table of this group is given in Table 3.3.
Fig. 3.9 PH3 inversion 1 1
P P
3 3
2 2
Inversion vibration
Before we consider the results of this symmetry labeling we should consider the
effect of the inversion motion in PH3 : PH3 has two versions (see Bone et al. [33]
for a discussion of this concept) of the numbered equilibrium structure of the
molecule. An observer sitting on the P nucleus will see the three protons labeled
1–3 so that the rotation from 1 to 2 to 3 is a clockwise rotation for the one version,
and an anticlockwise rotation for the other version. The two versions are mirror
images of each other and are taken into each other by the inversion operation E
:
On the potential energy surface of the molecule, the two minima corresponding to
the two versions are separated by a very high potential barrier (11300 cm1 ; see
Špirko et al. [34]) so that, on the timescale of a typical experiment, there is no
observable inversion tunneling splitting. Thus the energy levels can be calculated
by just considering the motion in one of the two minima, and we do not need to
consider both minima. The two minima have identical sets of rotation–vibration
energy levels (Figs. 3.9 and 3.10).
If we were to calculate the rotation–vibration energy levels using the double
minimum potential energy surface we would find that well below the barrier, every
energy level would be doubly degenerate to within measurement accuracy for PH3 :
Fig. 3.10 A typical double

minimum potential is
illustrated. In the case PH3 ;
the coordinates q represents
the ‘‘inversion’’ motion
Energy [arbitrary unit]
shown in Fig. 3.9
Potential Barrier
(I) (II)
−ρe 0 ρe
ρ [arbitrary unit]
Table 3.4 The character E (123) (12)*

table of the molecular (132) (23)*
symmetry group C3v (M) (31)*
A1 1 1 1
A2 1 1 -1
E 2 -1 0
If we symmetry classified the levels using the group G12 we would find that there
were three types of energy level: Aþ þ þ
1 þ A1 ; A2 þ A2 ; or E þ E : This double
degeneracy would be resolved by inversion tunneling and it is an accidental
degeneracy not forced by the symmetry group G12 : If the inversion tunneling is not
resolved it is not necessary to use the group G12 : There are only three distinct types
of energy levels and yet we have classified them in a group of six irreducible
representations. However, Longuet-Higgins [35] showed how to obtain the
appropriate subgroup of G12 that avoids the double labels. This is achieved by just
using the elements of G12 that are appropriate for a single minimum; we delete
elements such as E
and (12) that interconvert the two versions. Longuet-Higgins
termed the deleted elements ‘‘unfeasible.’’ The group obtained is ‘‘the molecular
symmetry (MS) group.’’ In the case of PH3 ; we obtain the particular MS group
C3v ðMÞ ¼ fE; ð123Þ; ð132Þ; ð12Þ

; ð23Þ
; ð31Þ
g: ð3:3:126Þ
Here, the operation (123) replaces nucleus 1 by nucleus 2, nucleus 2 by nucleus 3,
and nucleus 3 by nucleus 1. The operation (132) replaces nucleus 1 by nucleus 3,
nucleus 3 by nucleus 2, and nucleus 2 by nucleus 1. The operations ð12Þ
; ð23Þ
;
and ð31Þ
are analogous the ð12Þ
operation considered for water. The character
table of C3v ðMÞ is given in Table 3.4. Using this group we achieve a sufficient
symmetry labeling of the levels as being either A1 ; A2 or E: All possible interac-
tions can be understood using this group (apart from the effect of inversion
tunneling).
For PH3 we do not save much work by using the MS group rather than the
CNPI group, but for larger molecules, such as, for example, benzene, a great
simplification is achieved if all unfeasible elements of the CNPI group are elim-
inated from consideration. An unfeasible element of the CNPI group is one that
takes the molecule between versions that are separated by an insuperable energy
barrier in the potential energy function. For benzene, the CNPI group has 6!
6! 2 ¼ 1; 036; 800 elements. The MS group that is used to interpret the spectrum
has 24 elements.
Ammonia NH3 is pyramidal like PH3 ; and in its electronic ground state there
are two versions of the numbered equilibrium structure exactly as for PH3 : The
potential barrier between the two versions, however, is around 2; 000 cm1 for
ammonia (Špirko and Kraemer [36]) and thus much lower than in PH3 : This
barrier is so low that the molecule will tunnel through it on the time scale of a
typical spectroscopic experiment, and the tunneling motion gives rise to energy
level splittings that can be resolved experimentally (see, for example, Fig. 15-3
of [25]). Thus, for NH3 ; all elements of the group G12 are feasible, and the
molecular symmetry group of NH3 in its electronic ground state is G12 : This
group is isomorphic to the point group D3h and in the literature it is customarily
called D3h (M).
3.3.9.2 Selection Rules for Optical Transitions
The intensity of an electric dipole transition (in absorption or emission) is

proportional to the square of the integral
Z
ðwn Þ
lA wm dV; ð3:3:127Þ
where wn is the rovibronic wavefunction for the initial state of the transition (with
energy En ), wm is the rovibronic wavefunction for the final state (with energy Em ),
lA is the A component (A ¼ n; g; f; where the ngf axis system has axes parallel to
the laboratory-fixed axes XYZ but origin in the nuclear center of mass) of the
molecular dipole moment operator, and dV is the volume element associated with
the rovibronic coordinates.
An important result of representation theory is that the integral in Eq. 3.3.127
can only be non-vanishing (i.e., be different from 0) when the integrand
ðwn Þ
lA wm belongs to the totally symmetric representation of the MS group. The
totally symmetric representation has the character þ1 under all group operations,
and any group has such a representation which, by convention, is always written in
the first line of the character table. A wavefunction that belongs to the totally
symmetric representation is unchanged by all group operations. For the group
C2v (M) (Table 3.2), A1 is the totally symmetric representation.
The dipole moment component lA always belongs to an irreducible represen-
tation with the character þ1 under all ‘‘pure’’ permutations ðijk. . .Þ and the
character 1 under all permutation–inversion operations [E
or ðijk. . .Þ
]. For
C2v (M) (Table 3.2), this representation is A2 ; and any other group has an analogous
irreducible representation.
We consider a transition in the water molecule for which the wavefunction wn
belongs to the irreducible representation Cn of the group C2v (M), while wm belongs
to the irreducible representation Cm : In order that the integrand ðwn Þ
lA wm can
belong to the totally symmetric representation A1 ; the product ðwn Þ
wm must
belong to the irreducible representation A2 : Table 3.2 shows that this will be true
for ðCn ; Cm Þ ¼ ðA1 ; A2 Þ or for ðCn ; Cm Þ ¼ ðA2 ; A1 Þ or for ðCn ; Cm Þ ¼ ðB1 ; B2 Þ or
for ðCn ; Cm Þ ¼ ðB2 ; B1 Þ: These selection rules can be summarized as
A1 $ A2
ð3:3:128Þ
B1 $ B2
Analogous selection rules can be constructed for any other molecule.

For PH3 ; the molecular dipole moment component lA has the symmetry A2 in
C3v (M) (Table 3.4). The selection rules on the symmetries of the wavefunctions wn
and wm are
A1 $ A2
ð3:3:129Þ
E$E
The analogous selection rules on the symmetries of the wavefunctions wn and wm

for NH3 are
Aþ
1 $ A1
Aþ
2 $ A2 ð3:3:130Þ
Eþ $ E ;
the appropriate MS group is now G12 or D3h (M) (Table 3.3), in which the
molecular dipole moment component lA has the symmetry A 1:
For any molecule, the selection rules obtained from MS group symmetry are
accompanied by the selection rules obtained from rotational symmetry (i.e., from
the fact mentioned above that owing to the isotropy of space, we can rotate an
isolated molecule arbitrarily without changing the physical situation). For electric
dipole transitions these selection rules are, for any molecule, DJ ¼ J 0 J 00 ¼ 1;
0, or 1, and J 0 þ J 00 1: Here J 00 and J 0 are the values of the rotational quantum
number J for the initial and final states, respectively, of the transition considered.
3.3.10 Appendix to 3.3
3.3.10.1 Rotational Spectra of Linear Molecules
Typical rotational spectra of linear molecules are shown in Fig. 3.11 as a stick-
diagram. The line positions and intensities for HCN, HC3 N; HC5 N; HC7 N;
Fig. 3.11 Simulated 0.06

rotational spectra of linear HCN
molecules: HCN, 0.05
HC3 N; HC5 N; HC7 N; and
Intensity [arbitrary]
HC9 N at 20 K 0.04 HC3N
0.03 HC5N
0.02
0.01
0.00
0 100 200 300 400 500
Frequency [GHz]
Fig. 3.12 Simulated 0.06

rotational spectra of linear
molecules: HCN, 0.05
HC3 N; HC5 N; HC7 N; and
HC3N
HC9 N at 20 K: below 100 0.04
GHz
0.03
0.02 HC5N
HCN
HC7N
0.01 HC9N
0.00
0 20 40 60 80 100
Frequency [GHz]
and HC9 N are calculated by using PGOPHER,11 with the rotational constants
B; centrifugal distortion constants D; and the permanent dipole moment l found in
literature [37, 38, 39, 40]. The intensities are calculated for T ¼ 20 K. The
intensities for HCN lines are reduced by a factor 4, because they are too large
compared to those of other molecules.
For linear molecules (including diatomic molecules) the line positions for the
rotational spectra are given by
mðJ þ 1 JÞ ¼ EðJ þ 1Þ EðJÞ 2BðJ þ 1Þ ð3:3:131Þ
for each J; where we assume that the energy and the rotational constant are in
frequency unit. Consequently, the spectra are composed of almost equally spaced
lines, as shown in the figure. The line intensity increases with the increase of J
from 0, reaches the maximum at a value of J depending on the temperature, and
then decreases. Details of the intensity calculation are given in textbooks such as
[12]. In the figure the spectra of HCn N with n [ 5 are very densely overlapped at
the low frequency region (0–100 GHz), which are shown in an expanded scale in
Fig. 3.12.
3.3.10.2 Asymmetry Parameters
In general, the rotational energies of an asymmetric top cannot be expressed

analytically except for a few low J cases (see Table 7.7 of [12]. Nowadays, they
can be easily calculated by using a PC with a proper program. In the old days,
however, the energy term values of an asymmetric top are calculated (or tabulated)
with the help of asymmetry parameter j (Ray’s asymmetry parameter) [5, 12],
11
A Program for Simulating Rotational Structure, by C.M. Western, University of Bristol. http://
pgopher.chem.bris.ac.uk.
Fig. 3.13 Asymmetry 1.0

parameters, j (solid line), bp
(dotted line), and bo (chain
line), are plotted for the case 0.5
of A ¼ 10; C ¼ 1; and
1 B 10 cm1
κ
0.0
bp bo
–0.5
–1.0
1 2 3 4 5 6 7 8 9 10
−1
B=C B [cm ] B=A
2B A C
j¼ ; ð3:3:132Þ
AC
which is 1 and þ1 at the prolate ðA [ B ¼ CÞ and oblate ðA ¼ B [ CÞ sym-
metric top limit, respectively. The relation of the parameter j with the rotational
constants is graphically displayed in Fig. 3.13, for a molecule with A ¼ 10 and
C ¼ 1 cm1 : by definition A B C:
For the molecules close to the symmetric top limit (near prolate or near oblate
symmetric top), the effect of the inertial asymmetry can be evaluated by the
perturbation treatment. In such a case it is useful to introduce other asymmetry
parameters:
CB
bp ¼ ð3:3:133Þ
2A B C
and
AB
bo ¼ : ð3:3:134Þ
2C B A
As shown in Fig. 3.13, they take values between 0 and 1 : bp ¼ 0 at the prolate
symmetric top limit, and bo ¼ 0 at the oblate symmetric top limit. Consequently
they can be used as the perturbation parameter near the symmetric top limits.
3.3.10.3 Rotational Energy Level Diagram: H2 CO as an Example
Figure 3.14 shows schematically the energy level diagram of the formaldehyde
molecule, H2 CO; which is a near prolate asymmetric top. The energies are cal-
culated as a rigid rotor assuming the rotational constants to be A ¼ 9:4053; B ¼
1:22; and C ¼ 1:20 cm1 [41]. All levels with Ka 6¼ 0 are doubled (K-type dou-
bling) due to the inertial asymmetry as indicated.
The molecule is of C2v symmetry, and the a-axis (the molecule-fixed axis along
the smallest moment of inertia) coincides with the C2 symmetry axis, which
corresponds to the permutation operation (12), i.e., exchange of the two protons of
the molecule (see Sect. 3.3.9).
Since the protons are Fermion of nuclear spin I ¼ 1=2; there are four linearly
independent nuclear spin states for H2 CO; noting that the nuclear spin is zero for
12
C and 16 O. Three of them are symmetric and one is anti-symmetric for the
symmetry operation (12) [25, 14]. For a single permutation of two identical
Fermions, the overall wavefunctions should be anti-symmetric. Thus, for H2 CO;
the rotational states which are symmetric to the C2 operation (even Ka levels of the
ground vibronic state12) can be combined with the one anti-symmetric nuclear spin
state, and the anti-symmetric rotational states (odd Ka levels) can be combined
with the three anti-symmetric nuclear spin states. If the effect of the nuclear spin
on the energy is small (i.e., the hyperfine splitting is small), then the three different
nuclear spin sub-states for a given odd Ka level are energetically degenerate.
Consequently, the rotational spectra of H2 CO behaves as if the odd Ka levels were
three times more populated than the even Ka levels. Such a effect is called as
‘‘nuclear spin statistics’’ or ‘‘nuclear spin weights’’.
The nuclear spin weight is three for the odd Ka and one for the even Ka : States
with larger nuclear spin weight are called ortho states and those with less weight
are para states.
3.3.10.4 Rotational Energy Level Diagram: H2 O as an Example
Figure 3.15 shows schematically the energy level diagram of the water molecule,
H2 O; which is a near prolate asymmetric top. The energies are calculated as a rigid
rotator assuming the rotational constants to be A ¼ 27:877; B ¼ 14:512; and
C ¼ 9:285 cm1 [42].
The water molecule is also of C2v symmetry, but different from the case of
H2 CO; b-axis (the molecule-fixed axis along the intermediate moment of inertia)
coincides with the C2 symmetry axis, which corresponds to the permutation
operation (see footnote 12), i.e., exchange of the two protons.
Thus the ortho and para states are distinguished by the parity of Ka þ Kc (see
footnote 12), as indicated by solid and broken lines in Fig. 3.15. Since the ortho
and para states are distributed rather randomly, as compared with the case of
H2 CO (Fig. 3.14), the ortho and para levels are often illustrated side-by-side as
shown in Fig. 3.16.
12
It should be noted that the parity of the quantum number Ka indicates the symmetry of the
rotational wavefunction for the C2 rotation about the molecule-fixed a-axis. Similarly Kc
indicates that for the C2 rotation about the molecule-fixed c-axis. Consequently, the symmetry
for the C2 rotation about the molecule-fixed b-axis is indicated by the parity of Ka þ Kc :
6
250 945
9 946 5
8
200 7
936
9 937 6
Energy [cm ]
5
−1
8
150 927 4
9 928 7
918 6
909 9 8
9 919 5
100 8 7 4
8 3
6
7
7 5
6 4
50 6 3
5 2
5 4
4 3
3 2
1 2 1
0 0
0 1 2 3 4 5
Ka
Fig. 3.14 The rotational energy levels of H2 CO in the ground vibronic state are schematically
illustrated. The odd Ka levels are ortho with the spin weight of three, which are indicated by solid
lines. The even Ka levels are para with the spin weight of one, which are indicated by broken
lines
3.3.10.5 Pure Rotational Spectra of an Asymmetric Top Molecule:

HNCO as an Example
HNCO is an asymmetric top molecule very near prolate symmetric top, and is of
planar structure with non-vanishing dipole moment along the a and b-axis.13
Therefore, we can observe a-type ðDKa ¼ 0Þ and b-type ðDKa ¼ 1Þ transitions.
The spectra, which are of a; b-hybrid type, simulated by PGOPHER (see footnote
11) for T ¼ 50 K are shown up to 3.5 THz in Fig. 3.17, where the molecular
constants given in [43] are used.14 The lower frame shows the spectra as a stick-
diagram, and the upper frame shows the Fortrat diagram, the vertical axis of which
represents the J quantum number for the lower level of the transition.
Strong a-type transitions are indicated as q RK in the lower frame. The notation
x
Yz is used for indicating the transition, which is defined for symmetric top spectra.
The superscript x is p; q; or r for DK ¼ 1; 0, or þ1; respectively. Y is P; Q; or R
for DJ ¼ 1; 0, or þ1; respectively. The subscript z indicates the quantum number
K of the lower level of the transition. For a near prolate symmetric top, K should
be read as Ka : The details of the q RK transitions are illustrated in Fig. 3.18.
13
The molecular plane is the a b plane.
14
The Hamiltonian was truncated for simplicity up to the sextic centrifugal correction terms in
this simulation. Therefore, it does not reproduce the observed spectrum.
532 440
500 533 441
606 606 523
524
400 514 431
432
Energy [cm−1]
505 515 422

423 330
300
413 331
404 414 321

200 322
312
303 313 220
221
100 211
202 212
110
101 111
0 000
0 1 2 3 4
Ka
Fig. 3.15 The rotational energy levels of H2 O in the ground vibronic state are schematically
illustrated. The ortho levels with the spin weight of three, which are indicated by solid lines, and
para levels with the spin weight of one by broken lines
500 532 533

441 440
523 616 606
524
514
400 432 431
505 515
422
Energy [cm−1]
423
300 330 331
413
414 404
321 322
200 312
221 303 313 220
211
100 212 202
110 111
101
Ortho 000 Para
0
4 3 2 1 0 1 2 3 4
Ka
Fig. 3.16 The ortho levels of H2 O are illustrated on the left side, and the para levels on the right.
Since the conversion between the ortho and para levels is very slow, we can handle the ortho and
para states as if they are of ‘‘different’’ molecules
20
15
10
0
0.004 r
Q0
0.003 r
R0
q r r
RK P0 Q1
0.002
r r
0.001 P1 R1
0
0.0 1.0 2.0 3.0
Frequency [THz]
Fig. 3.17 Simulated structures of the a; b-hybrid band of a very near prolate asymmetric top are
displayed: the stick-diagram in the lower frame and the Fortrat diagram in the upper frame
Fig. 3.18 The effective 9

ðKÞ
rotational constants Beff 8
defined by Eq. 3.3.137 are
plotted for J 00 7
6
5
J"
Ka=5
4
Ka=4
3
Ka=3
2
Ka=2
1 l u
Ka=1 Ka=0 Ka=1
0
10940 10960 10980 11000 11020 11040
Beff[MHz]
Strong b-type transitions are as indicated there, r P0 ;r Q0 ;r R0 ;r P1 ;r Q1 ; and r R1 :

Because of the K-type doubling, the Ka ¼ 2 1 transitions are doubled as shown
in the figure.
3.3.10.6 Effective B Plots
The rotational energy of a symmetric top molecule for can be expressed in the
rigid rotor approximation as
ðKÞ ðKÞ
EðJ; KÞ ¼ E0 þ Beff JðJ þ 1Þ ð3:3:135Þ
for each K; where we assume that the energy and the rotational constant are in
frequency unit. The frequencies for the transitions of q RK ðJÞ is thus given by
ðKÞ
mðq RK ðJÞÞ ¼ EðJ þ 1; KÞ EðJ; KÞ ¼ 2Beff ðJ þ 1Þ; ð3:3:136Þ
ðKÞ
which is similar to Eq. 3.3.131. The effective rotational constant Beff is thus
derived from the observed frequencies as
ðKÞ
Beff ¼ mðq RK ðJÞÞ=½2ðJ þ 1Þ: ð3:3:137Þ
This expression can be applied also for asymmetric top molecules, if they are close
to the symmetric top limits, like HNCO or H2 CO:
From the observed a-type R-branch transitions of HNCO the J dependence of
ðKÞ
the Beff values are displayed in Fig. 3.18, where effective B values are taken from
[43]. HNCO is an asymmetric top molecule, and thus because of the K-type
doubling, two effective B values can be obtained for Ka ¼ 1 as shown there. If the
molecule is a true symmetric top, we should have a single effective B value for
K ¼ 1; the plot of which should appear near the center of the two Ka ¼ 1 plots.
This doubling should be observed also for Ka [ 1: Indeed we can see such
doubling for high-J transitions of Ka ¼ 2 as shown in Fig. 3.19, where the central
part of Fig. 3.18 are presented in an expanded scale. HNCO molecule is very much
close to the prolate symmetric top limit, the doubling cannot be observed for
Ka 3 at least for J\10:
3.3.10.7 Rovibrational Spectra of an Asymmetric Top Molecule:

An a-Type Band of a Near Prolate Top
The structure of the m2 band of H2 CO looks like15 the one given in Fig. 3.20 which
is simulated by PGOPHER (see footnote 11) for T ¼ 150 K with the rigid rotor
approximation: the upper frame shows the Fortrat diagram, and the lower the stick-
diagram of the spectra.
The m2 band of H2 CO corresponds to the vibrational mode of CO stretching,
which induce a change of dipole moment along the molecule-fixed a-axis. Thus
the band is of a-type; DKa ¼ 0 transitions, q PK ;q QK ; and q RK ; appear strongly as
shown there. The spectrum is very similar to that of a parallel band of a prolate
symmetric top molecule, except for the K-type doubling effect: an impressive Q-
branch transitions at the center, and P and R transitions on both side with almost
equal spacing about 2B:
The spectra can be displayed in a different manner as the ‘‘Loomis–Wood’’
diagram [44] as shown in Fig. 3.21. In this diagram, the spectrum given in Fig. 3.20
15
It does not reproduces the observed spectrum.
Fig. 3.19 The center part of 9

Fig. 3.18 is plotted in an
expanded scale. The K-type 8
doubling effect for the Ka ¼ 2 7
is visible in this figure
6
5
J"
Ka=5
4
Ka=4
3
Ka=3
2
Ka=2
1
Ka=0
0
10970 10980 10990
Beff [MHz]
15
10
J"
0
q q q
PK QK RK
[arbitrary unit]
Intensity
1700 1720 1740 1760 1780 1800

Wavenumber [cm ] −1
Fig. 3.20 Typical structures of an a-type rovibrational band of a slightly asymmetric top (near
prolate) are illustrated. The spectrum is displayed as a stick-diagram in the lower frame, and the
Fortrat diagram is displayed in the upper frame. The spectrum is calculated by applying the rigid
rotor approximation with the rotational constants and band origin similar to those of H2 CO m2
band
ðobsÞ
is displayed by plotting the line positions mi (i for all observed lines) measured
from the reference frequencies mm ; which are usually calculated by
m m ¼ a0 þ a1 m þ a2 m 2 ; ð3:3:138Þ
where a0 ¼ 1746:1; a0 ¼ 2:41706; a2 ¼ 0:00377 cm1 in the case of Fig. 3.21.

ðobsÞ
With this diagram (m in the vertical axis, and the mi mm in the horizontal axis)
we can identify series of lines which appear with a regular frequency spacing of
Fig. 3.21 The spectrum 30

given in Fig. 3.20 is displayed
as a Loomis Wood diagram.
The circles represent the line 20
positions. The line intensities
are represented by the size of
the circle 10
m
0
-10
-20
-30
-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5
−1
ν−νref [cm ]
about a1 which is about 2:4 cm1 in this case. Details of the diagram are described
in [45].
3.3.10.8 Rovibrational Spectra of an Asymmetric Top Molecule:

A b-Type Band of a Near Prolate Top
The structure of the m5 band of H2 CO looks like16 in Fig. 3.22, which is simulated
by PGOPHER (see footnote 11) for T ¼ 150 K with the rigid rotor approximation:
the upper frames of each part of the spectrum shows the Fortrat diagram, and the
lower the stick-diagram of the spectra.
The m5 band of H2 CO corresponds to the vibrational mode of anti-symmetric
CH stretching, which induce a change of dipole moment along the molecule-fixed
b-axis. Thus the band is of b-type; DKa ¼ 1 transitions, r PK ; r QK ; r RK ; p PK ; p QK ;
and p RK ; appear strongly as shown there. The spectrum is very similar to that of a
perpendicular band of a prolate symmetric top molecule, except for the K-type
doubling effect: impressive r Q and p Q branch transitions appear with almost equal
spacing of about 2A:
The band is widely spread out compared with the a-type band, and it is in
general difficult to display with a single Loomis–Wood diagram.
16
It does not reproduces the observed spectrum.
25
20
15
J"
10
5
0
p p p
0.003 Q5 Q4 Q3
0.002
0.001
0.000
2750 2760 2770 2780 2790 2800 2810 2820
Wavenumber [cm-1]
25
20
15
J"
10
5
0
p p p r r
Q3 Q2 Q1 Q0 Q1
0.003
0.002
0.001
0.000
2800 2810 2820 2830 2840 2850 2860 2870
Wavenumber [cm-1]
25
20
15
J"
10
5
0
r r
r r r
Q0 Q1 Q2 Q3 Q4
0.003
0.002
0.001
0.000
2850 2860 2870 2880 2890 2900 2910 2920
Wavenumber [cm-1]
Fig. 3.22 Typical structures of a b-type rovibrational band of a slightly asymmetric top (near
prolate) are illustrated together with the Fortrat diagram
3.4 Molecules with Internal Rotation
In this section, one- and two-methyl top internal-rotation problems in asymmetric

top molecules will be treated. Internal rotation itself, which is a kind of large-
amplitude vibrations, is a fascinating theme also for spectroscopists.
Let us consider qualitatively at first what effect the internal rotation produces in
the rotational energy level structure and the rotational spectrum. The internal
rotation occurring in a free molecule is hindered generally and it is subject to an
appropriate potential function. The internal-rotation potential function is periodic,
and in the simplest case it is represented by a sinusoidal function, whose amplitude
is the barrier height. Two limiting cases of the internal rotation are a free rotor
limit, corresponding to an extremely low barrier case, and an extremely high
barrier limit.
For understanding aspects of the effect of the internal rotation on the rotational
energy level structure, let us start from a hypothetically high barrier limit case,
where a small-amplitude torsional motion occurs around a potential minimum. In
the extremely high barrier limit, the rotational energy level structure will be the
same as that of an ordinary semi-rigid molecule. As the barrier height of the
internal-rotation potential is lowered, tunneling motions occurs among equilibrium
configurations at the potential minima. A rotational energy level is then split into
two levels, A-symmetry and E-symmetry levels, in a molecule with one methyl
internal rotor.
As the barrier height is further lower, the tunneling motions occur more fre-
quently and accordingly the splitting is larger. If the effect of the internal rotation
is to induce the splitting to be the same for all the rotational levels in a certain
vibrational state, the rotational energy level structure will not be so complicated,
and then we will observe only ordinary semi-rigid rotor transitions overlapped for
two symmetry species since selection rules are A $ A and E $ E:
In reality, since the Coriolis interaction between the overall rotation and the
internal rotation exists necessarily and it makes a rotational energy level split into
two levels whose interval depends on angular momentum quantum numbers, we
observe somewhat complicated rotational spectra such as a rotational spectral line
is split into two lines whose interval depends on the rotational transition.
For group theoretical treatments of flexible molecules such as those with
internal rotors, it is appropriate to use the permutation–inversion group proposed
by Longuet-Higgins [35]. Throughout this section the group theoretical description
will be made using the permutation–inversion group. In this section, molecules
only with methyl internal rotors which can be described by a single group of the
permutation–inversion group will be treated, and internal rotations in molecules
such as CH3 CH3 ; H2 O2 and H2 S2 ; which should be described in a specially
designed method including a double group idea of the permutation inversion
group, are not treated.
We describe below mainly the formulation for understanding high-resolution
torsional–rotational spectra and helping to analyze them.
3.4.1 Molecule with One Methyl Internal Rotor
3.4.1.1 Torsion–Rotation Hamiltonian Operator
At first we note that torsion is used to represent internal rotation sometimes in

Sect. 3.4. The purpose of this section is to show a procedure for deriving a
Hamiltonian operator for molecules with one methyl internal rotor. In literatures,
we find that three approaches to the problem, which are called PAM (principal axis
method), RAM (rho axis method) and IAM (internal axis method), have been
developed. The difference among the three methods lies in the choice of the
reference configuration, in other words, which coordinate system is taken to
describe the torsional and rotational motion. Referring to the paper of Hougen
et al. [46], where the three approaches are summarized systematically, we review
the three methods below.
Principal Axis Methods
It is conceptually convenient to describe the principal axis method (PAM) first for
defining a coordinate system and deriving a Hamiltonian operator. The laboratory-
fixed coordinate Ri ; ðXi ; Yi ; Zi Þ; of the ith atom in the molecule with a methyl top
rotor is expressed in a PAM system as follows:

Ri ¼ R þ S1 ðh; /; vÞS1 ðhP ; /P ; vP Þ S1 ð0; 0; sÞ½ai þ di A
ð3:4:1Þ
¼ R þ S1 ðh; /; vÞrPAM i ;
for methyl protons (i.e., i ¼ 1; 2 and 3), and

Rj ¼ R þ S1 ðh; /; vÞS1 ðhP ; /P ; vP Þ aj þ dj A
ð3:4:2Þ
¼ R þ S1 ðh; /; vÞrPAM
j ;
for other nuclei.

Before denoting various symbols appearing in Eqs. 3.4.1 and 3.4.2, we give the
definition of three axis systems ðX; Y; ZÞ; ðx0 ; y0 ; z0 Þ and ðx; y; zÞ used implicitly:
X; Y; Z: the laboratory-fixed axis system,
x0 ; y0 ; z0 : the molecule-fixed axis system, where the z0 -axis is along the C 3 axis
of the methyl top,
x; y; z: the molecule-fixed principal inertial axis system (usually, a; b; c $ z; x; y).
The vectors Ri ; R; ai ; di and A in Eqs. 3.4.1 and 3.4.2 are defined as follows:
Ri : the position vector of the ith atom expressed in the laboratory-fixed system.
R: the laboratory-fixed coordinates of the center of mass of the molecule.
ai : the equilibrium position of the ith atom expressed in the molecule-fixed x0 y0 z0 -
system.
di : the small-amplitude displacement vector of the ith atom expressed in the

molecule-fixed x0 y0 z0 -system.
A: the center of mass expressed in the x0 y0 z0 -system.
The rotation matrix S1 ð0; 0; sÞ rotates the equilibrium positions and dis-
placement vectors of methyl protons through an angle s about the z0 -axis. The
constant rotation matrix S1 ðhP ; /P ; vP Þ is introduced so that the reference posi-
tions S1 ð0; 0; sÞ½ai þ di for the methyl protons and aj þ dj for the other nuclei
are located in the principal axis system x; y; z: When we assume the methyl top is
of C3v symmetry, A and S1 ðhP ; /P ; vP Þ are constant, and we have
S1 ð0; 0; 2p=3Þa1 ¼ a2 ; S1 ð0; 0; 2p=3Þa2 ¼ a3 ; S1 ð0; 0; 2p=3Þa3 ¼ a1 : ð3:4:3Þ
The rotation matrix S1 used in the equations above is the usual 3 3 direction
cosine matrix defined as
0 10 10 1
cos / sin / 0 cos h 0 sin h cos v sin v 0
S1 ðh; /; vÞ ¼ @ sin / cos / 0 A@ 0 1 0 A@ sin v cos v 0 A:
0 0 1 sin h 0 cos h 0 0 1
ð3:4:4Þ
Classical Kinetic Energy Expression. The first step for deriving a Hamiltonian
operator is to express a classical nuclear kinetic energy with internal variables. The
classical nuclear kinetic energy is a sum over the kinetic energies of individual
atoms within a molecule:
" #
X1 dRi 2 dRi 2 dRi 2
T¼ mi þ þ : ð3:4:5Þ
i
2 dt x dt y dt z
In the present treatment, we ignore the translational energy and the small-ampli-
tude vibrational energy, which are related to R and di in Eqs. 3.4.1 and 3.4.2,
respectively. Then we have the following expression of the velocity vector for the
methyl protons after an algebraic manipulation,
2
dR 3 2
3
i dRi
dt x dt X
6
dRi 7 6
dR 7
6 dt 7 ¼ Sðh; /; vÞ6 i 7 ¼ Sðh; /; vÞ d S1 ðh; /; vÞ rPAM
4 y5 4 dt Y 5 dt i

dR
dR
i i
dt z dt Z

d 1
þ S ðhP ; /P ; vP ÞS1 ð0; 0; sÞ ai
dt ð3:4:6Þ
2 3
2
PAM
3 s_ eint r0i PAM
x ri 6 7
6
x 7 6 x 7
6 PAM 7
¼6 PAM 7
4
x ri y 5 þ 6 s_ eint ri0
7
6 y7
5
PAM
x ri 4
z s_ eint r0i PAM
z
where r0i PAM is the position vector of the ith methyl-proton in the xyz-axis system
from the center of mass of the methyl group
PAM
r0i ¼ rPAM
i þ S1 ðhP ; /P ; vP ÞA: ð3:4:7Þ
For the other nuclei, we have

2 3 2 3
dRj
x rPAM
j
6 dt x7 6 x 7
6 dRj 7 6 7
6 dt 7 ¼ 6 x rPAM 7; ð3:4:8Þ
6 7 6 j
y 7
4 dR 5 4
y
5
dt
j
x rPAM
j
z z
where x, angular velocity vector representing the overall rotation of the molecule;
and eint , unit vector along the methyl internal rotation, which is the C3 axis because
we assume the methyl top to be of C3v symmetry.
Substituting Eqs. 3.4.6 and 3.4.8 into Eq. 3.4.5 we have
1 X 1 X
T¼ Ia x2a þ Is s_ 2 þ Is s_ ka x a ; ð3:4:9Þ
2 a¼x;y;z 2 a¼x;y;z
where Ia ða ¼ x; y; zÞ are the principal moments of inertia of the molecule, Is is the

moment of inertia of the methyl top about the C 3 axis, and ka ða ¼ x; y; zÞ are the
direction cosines of the methyl internal-rotation axis. For obtaining the quantum
mechanical Hamiltonian operator, it is necessary to express the classical kinetic
energy in the momenta conjugate to the variables representing the overall rotation
and the methyl internal rotation. The momenta are
oT
Ja ¼ ¼ Ia xa þ Is ka s_ ða ¼ x; y; zÞ ð3:4:10Þ
oxa
oT X
ps ¼ ¼ Is s_ þ Is ka x a : ð3:4:11Þ
o_s a¼x;y;z
By using the definitions

ka I s X
qa ; r 1 q a ka ð3:4:12Þ
Ia a¼x;y;z
and Eqs. 3.4.10 and 3.4.11, we have

!
X X
ps qa Ja ¼ Is 1 qa ka s_ ¼ rIs s_ : ð3:4:13Þ
a¼x;y;z a¼x;y;z
From Eqs. 3.4.9, 3.4.10, 3.4.11 and 3.4.13, we have

!2
X 1 1 X
T¼ Ja2 þ ps qa Ja ð3:4:14Þ
2I
a¼x;y;z a
2rIs a¼x;y;z
Quantum Mechanical Hamiltonian Operator. The quantum mechanical PAM

Hamiltonian operator is obtained by adding the potential function VðsÞ to T; that
is,
!2
X 1 1 X
2
HPAM ¼ J þ ps qa Ja þVðsÞ; ð3:4:15Þ
2I a 2rIs
a¼x;y;z a a¼x;y;z
where Ja ða ¼ x; y; zÞ are the total angular momentum operators and ps the inter-
o o
nal-rotation angular momentum operator (Jz ¼ ih ov , etc. and ps ¼ ih os ).
Rho Axis Method
The laboratory-fixed coordinate in the RAM coordinate system, Rk ; is expressed as
Rk ¼ R þ S1 ðh; /; vÞS1 ðhRAM ; /RAM ; vRAM ÞrPAM

k ; ð3:4:16Þ
where

rPAM
k ¼ S1 ðhP ; /P ; vP Þ S1 ð0; 0; sÞ½ak þ dk A ð3:4:17Þ
for the methyl protons, and
rPAM
k ¼ S1 ðhP ; /P ; vP Þf½ak þ dk Ag ð3:4:18Þ
for the other nuclei.

The rotation matrix S1 ðhRAM ; /RAM ; vRAM Þ introduced in Eq. 3.4.16
transforms angular momentum component operators Jx ; Jy ; Jz ; in such a way that
2 3 2 3
Jx Jx
4 Jy 5 ¼ S1 ðhRAM ; /RAM ; vRAM Þ4 Jy 5 ð3:4:19Þ
Jz RAM Jz PAM
or
2 3 2 3
Jx Jx
4 Jy 5 ¼ Sþ1 ðhRAM ; /RAM ; vRAM Þ4 Jy 5 : ð3:4:20Þ
Jz PAM Jz RAM
The Eulerian angles ðhRAM ; /RAM ; vRAM Þ are taken so that the transformation
shown in Eq. 3.4.19 or 3.4.20 leads to
X qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
qa Ja ! qJz ; q ¼ q2x þ q2y þ q2z : ð3:4:21Þ
a¼x;y;z
As a result, the RAM Hamiltonian operator is written by

X 1 1
HRAM ¼ Ja Jb þ Jb Ja þ Fðps qJz Þ2 þ VðsÞ; ð3:4:22Þ
2I 2
a¼x;y;z ab
Using the transformation relation of the angular momentum

2 3
Jx
h ihRAM Jy =h ivRAM Jz =h 6
i/RAM Jz = 7
e e e 4 Jy 5eivRAM Jz =h eihRAM Jy =h ei/RAM Jz =h
Jz
3 2 ð3:4:23Þ
Jx
6 7
¼ Sþ1 ðhRAM ; /RAM ; vRAM Þ4 Jy 5;
Jz
and the relations
ei/RAM Jz =h eihRAM Jy =h eivRAM Jz =h ps eivRAM Jz =h eihRAM Jy =h ei/RAM Jz =h ¼ ps ð3:4:24Þ
ei/RAM Jz =h eihRAM Jy =h eivRAM Jz =h VðsÞeivRAM Jz =h eihRAM Jy =h ei/RAM Jz =h ¼ VðsÞ ð3:4:25Þ
we have a relation between the PAM and the RAM Hamiltonian operators
HRAM ¼ ei/RAM Jz =h eihRAM Jy =h eivRAM Jz =h HPAM eivRAM Jz =h eihRAM Jy =h ei/RAM Jz =h :
ð3:4:26Þ
The relation (3.4.23) is derived from the commutation relations

eivRAM Jz =h Jx iJy eivRAM Jz =h ¼ eivRAM Jx iJy
eihRAM Jy =h ½Jz iJx eihRAM Jy =h ¼ eihRAM ½Jz iJx ð3:4:27Þ

ei/RAM Jz =h Jx iJy ei/RAM Jz =h ¼ ei/RAM Jx iJy :
See Appendix (Sect. 3.4.5) for the derivation of Eq. 3.4.27.

Consider a molecule having a plane of symmetry such as methanol ðCH3 OHÞ:
Assuming the symmetry plane to be the xz plane, we have vRAM ¼ 0; /RAM ¼ 0;
qy ¼ 0; and
ei/

h ihRAM Jy =h ivRAM Jz =h
RAM Jz = e e
qx ps Jx þ qz ps Jz eivRAM Jz =h eihRAM Jy =h ei/RAM Jz =h

¼ eihRAM Jy =h qx ps Jx þ qz ps Jz eihRAM Jy =h
¼ ðcos hRAM Þqx ps Jx þ ðsin hRAM Þqx ps Jz þ ðcos hRAM Þqz ps Jz ðsin hRAM Þqz ps Jx :
ð3:4:28Þ
For arriving at

eihRAM Jy =h qx ps Jx þ qz ps Jz eihRAM Jy =h ¼ qps Jz
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
q ¼ q2x þ q2z
the Eulerian angle hRAM satisfies

qx
tan hRAM ¼ : ð3:4:29Þ
qz
Then the RAM Hamiltonian operator is written as

HRAM ¼ Fðps qJz Þ2 þ VðsÞ
þ Aðcos hRAM Jz sin hRAM Jx Þ2 þBðsin hRAM Jz þ cos hRAM Jx Þ2 þCJy2 ;
ð3:4:30Þ
Internal Axis Method
The IAM Hamiltonian operator is obtained by a unitary transformation of the

RAM Hamiltonian
HIAM ¼ eisqJz =h HRAM eisqJz =h ð3:4:31Þ
so as to remove the term 2Fqps Jz in the RAM Hamiltonian operator. The relations

eisqJz =h Jx iJy eisqJz =h ¼ eisq Jx iJy ; ð3:4:32Þ
eisqJz =h Jz eisqJz =h ¼ Jz ; ð3:4:33Þ
eisqJz =h ps eisqJz =h ¼ ps þ qJz ; ð3:4:34Þ
eisqJz =h VðsÞeisqJz =h ¼ VðsÞ; ð3:4:35Þ

lead to the removal of the term 2Fqps Jz :
It is shown from Eq. 3.4.30 and Eqs. 3.4.31–3.4.35 that in the case of a
molecule having a plane of symmetry, the IAM Hamiltonian operator is given by
HIAM ¼ Fp2s þ VðsÞ

2
þ A cos hRAM Jz sin hRAM ðcos qsÞJx ðsin qsÞJy
2
þ B sin hRAM Jz þ cos hRAM ðcos qsÞJx ðsin qsÞJy
2
þ C ðsin qsÞJx þ ðcos qsÞJy : ð3:4:36Þ
The IAM coordinates, corresponding to the IAM Hamiltonian operator given by

Eq. 3.4.31, are expressed as
Rk ¼ R þ eisqJz =h S1 ðh; /; vÞeisqJz =h rRAM
k
ð3:4:37Þ
¼ R þ S1 ðh; /; vÞS1 ð0; 0; qsÞrRAM
k ;
where

rRAM
k ¼ S1 ðhRAM ; /RAM ; vRAM ÞS1 ðhP ; /P ; vP Þ S1 ð0; 0; sÞ½ak þ dk A
ð3:4:38Þ
for the methyl protons, and
rRAM
k ¼ S1 ðhRAM ; /RAM ; vRAM ÞS1 ðhP ; /P ; vP Þf½ak þ dk Ag ð3:4:39Þ
for the other nuclei.

It can be understood that Eqs. 3.4.32 and 3.4.33, together with the equation
2 3 2 3
Jx JX
4 Jy 5 ¼ Sþ1 ðh; /; vÞ4 JY 5; ð3:4:40Þ
Jz JZ
and the commutation relation

½Ja ; Ji ¼ 0 ða ¼ x; y; z; i ¼ X; Y; ZÞ ð3:4:41Þ
lead mathematically to the equations for the IAM coordinates, (3.4.37), (3.4.38)
and (3.4.39) (see Appendix, Sect. 3.4.5).
The physical meaning of the IAM system is seen from Eqs. 3.4.36 and 3.4.37 to
be that, by rotating the whole molecule backward about the q direction (the z
direction in the RAM system), the angular momentum generated by the methyl
internal rotation is canceled and as a result the Coriolis-like 2Fqps Jz term is
eliminated from the Hamiltonian operator.
3.4.1.2 Permutation–Inversion (PI) Group and Transformation Properties

of Variables Under Permutation–Inversion Operations
We assume that in a molecule under consideration a methyl internal rotation is the

only large-amplitude vibration mode. We will treat the following three cases.
1. The molecule has no geometric symmetry at its equilibrium, and the PI group
which belongs to is G3 : The symmetry of such a molecule is denoted as C1 ðG3 Þ:
2. The molecule has a plane of symmetry at its equilibrium configuration, and the
PI group which belongs to is G6 : The symmetry of such a molecule is denoted
as Cs ðG6 Þ:
3. The molecule is of CH3 BF2 type. It is of Cs symmetry at its equilibrium, and
it belongs to the PI group G12 : The symmetry of such a molecule is denoted as
Cs ðG12 Þ:
C1 ðG3 Þ Case
Alanine CH3 CH(COOH)NH2 shown in Fig. 3.23, for an example, corresponds to

this case. In Table 3.5, the character table of the PI group G3 for a C1 ðG3 Þ
molecule is given. In Table 3.5, 1–3 represent protons in the methyl group. (123)
and (132) [¼ ð123Þ2 ] are symmetry operations which become feasible by the
Fig. 3.23 Alanine

CH3 CH(COOH)NH2 ; an
example of a C1 molecule
with one methyl top
Table 3.5 Character table of E (123) (132)

G3 for a molecule of C1
symmetry A 1 1 1
Esep
E+ 1 x x*
E- 1 x* x
x ¼ expði2p=3Þ
Table 3.6 Transformation properties of variables appearing in Eqs. 3.4.1, 3.4.16, 3.4.17, 3.4.37,
and 3.4.38 for a generating operation (123) of G3
R1 ; R2 ; R3 R (translation) h; /; v (rotation) s (internal rotation)
(123)
PAM R2 ; R3 ; R1 R h; /; v s þ 2p=3
RAM R2 ; R3 ; R1 R h; /; v s þ 2p=3
IAM R2 ; R3 ; R1 R h; /; v þ 2pq=3 s þ 2p=3
methyl internal rotation. A brief explanation of the permutation–inversion group is

given in Appendix (Sect. 3.4.5). Although Eþ and E species in Table 3.5 are
different symmetry species mathematically, whose characters are complex con-
jugate with each other, they are degenerate energetically. We will use a notation
Esep for combined Eþ and E species. Transformation properties of variables in
the PAM, RAM and IAM system for the generating operation (123) of G3 are
shown in Table 3.6. We will not treat the small-amplitude displacements di
hereinafter in this Sect. 3.4 since they are out of consideration in the present
problem.
Cs ðG6 Þ Case
Methanol CH3 OH; shown in Fig. 3.24, is well known as an example of this
category. The character table and the transformation properties of variables are
given in Tables 3.7 and 3.8, respectively.
Fig. 3.24 Methanol CH3 OH;

an example of a Cs molecule
with one methyl top
Table 3.7 Character table of E ð123Þ; ð132Þ ð23Þ

; ð13Þ
; ð12Þ
G6 for a methanol-like Cs
molecule A1 1 1 1
A2 1 1 1
E 2 1 0
and 3.4.38 under generating operations of G6 for a methanol-like molecule
R1 ; R2 ; R3 R (translation) h; /; v (rotation) s (internal rotation)
ð123Þ
PAM R2 ; R3 ; R1 R h; /; v s þ 2p=3
RAM R2 ; R3 ; R1 R h; /; v s þ 2p=3
IAM R2 ; R3 ; R1 R h; /; v þ 2pq=3 s þ 2p=3
ð23Þ
PAM R1 ; R3 ; R2 R p h; p þ /; p v s

RAM R1 ; R3 ; R2 R p h; p þ /; p v s
IAM R1 ; R3 ; R2 R p h; p þ /; p v s
Seen in Tables 3.6 and 3.8, there is no difference of transformations between the
PAM and the RAM system while the transformation properties of the IAM vari-
ables is different with those for both the PAM and RAM systems. A further large
difference exists between the IAM system and the PAM or RAM system: the
operation ð123Þ3 is equivalent to the identity operation in the fRi g space for all the
PAM, RAM and IAM systems, but it is not equivalent to the identity operation for
the ðh; /; v; sÞ space in the IAM system while it is the identity operation also for
the ðh; /; v; sÞ in the PAM or the RAM systems. Accordingly, the boundary
conditions are somewhat complicated when we treat the internal-rotation problem
in the IAM system. One method to avoid the complication is to introduce the
extended PI group idea, but we mention it no more in this article.
A permutation–inversion operation ð23Þ
; which exists in the permutation–
inversion group for such a molecule as methanol and corresponds to the reflection
Fig. 3.25 CH3 BF2
Table 3.9 Character table of G12 for a CH3 BF2 -like molecule
E (123), (23)*, (45) (123)(45), (23)(45)*,
(132) (13)*, (132) (45) (13)(45)*,
(12)* (12)(45)*
A01 1 1 1 1 1 1
A02 1 1 -1 1 1 -1
E0 2 -1 0 2 -1 0
A001 1 1 1 -1 -1 -1
A002 1 1 -1 -1 -1 1
E00 2 -1 0 -2 1 0
1, 2, 3 represent protons in the methyl group, and 4, 5 represent F atoms
in the plane of symmetry at an equilibrium configuration, transforms variables

appearing in Eqs. 3.4.1, 3.4.16, 3.4.17, 3.4.37, and 3.4.38 as shown in Table 3.8.
The transformations of the variables for ð23Þ
are the same in all the PAM, RAM
and IAM systems.
Cs ðG12 Þ Case
An example for this type of molecule is CH3 BF2 ; whose chemical structure is
shown in Fig. 3.25. In Fig. 3.25 the PAM and RAM axes are also indicated (note
that CH3 BF2 is nearly an oblate symmetric top molecule). Although the difference
between the PAM and RAM systems is not essential for the case of CH3 BF2 since
its internal rotation axis coincides with its b-principal axis, we define the PAM and
RAM axes as shown in Fig. 3.25. The character table and the transformation
properties of variables are given in Tables 3.9 and 3.10, respectively.
Symmetry operations which become feasible by internal rotation are (123),
(132), (45), (123)(45), (132)(45), being different with the case of a methanol-like
molecule. Symmetry operation ð23Þð45Þ
corresponds to the reflection in the plane
of symmetry at an equilibrium configuration. Note that the operation ½ð123Þð45Þ6
is equivalent to the identity operation in the fRi g space for all the PAM, RAM and
IAM systems, but it is not equivalent to the identity operation for the ðh; /; v; sÞ
space in the IAM system while it is the identity operation also for the ðh; /; v; sÞ in
the PAM or the RAM system.
and 3.4.38 for generating operations of G12 for a CH3 BF2 -like molecule
R1 ; R2 ; R3 R (translation) h; /; v(rotation) s (internal rotation)
ð123Þð45Þ
PAM R2 ; R3 ; R1 R p h; p þ /; v s 2p=6
RAM R2 ; R3 ; R1 R h; /; p þ v s 2p=6
IAM R2 ; R3 ; R1 R h; /; p þ v 2pq=6 s 2p=6
ð23Þð45Þ
PAM R1 ; R3 ; R2 R p h; p þ /; p v s

RAM R1 ; R3 ; R2 R p h; p þ /; p v s
IAM R1 ; R3 ; R2 R p h; p þ /; p v s
3.4.1.3 Qualitative Understanding of Pure Internal-Rotation Energy

Levels for the C1 ðG3 Þ and Cs ðG6 Þ Cases
We consider here a pure internal-rotation problem, which is represented by a

Hw ¼ Ew;
ð3:4:42Þ
H ¼ Fp2s þ VðsÞ;
where, for simplicity, we consider the cases C1 ðG3 Þ and Cs ðG6 Þ; and we adopt the
potential function VðsÞ of the simplest form
1
VðsÞ ¼ V3 ð1 cos 3sÞ: ð3:4:43Þ
2
Barrier Height Parameter and Mathieu Equation
The energy eigenvalues of Eq. 3.4.42 with Eq. 3.4.43 are expressed by a parameter
s defined as
9
V3 ¼ Fs; ð3:4:44Þ
4
since the equation

2 1
Fps þ V3 ð1 cos 3sÞ wðsÞ ¼ EwðsÞ ð3:4:45Þ
2
is transformed to the Mathieu equation, being a differential equation well known in

physical mathematics
d2 w
þ ðb s cos2 xÞw ¼ 0; ð3:4:46Þ
dx2
where
9 9
2x ¼ 3s þ p; V3 ¼ Fs; E ¼ Fb: ð3:4:47Þ
4 4
Here we do not devote ourselves to describe the procedure of solving the Mathieu
equation Eq. 3.4.46. Instead we discuss the two limiting cases in the next.
Energy Level Patterns in Extreme Cases
A Hamiltonian operator of the pure internal-rotation problem for an extreme case

V3 ! 0 (it is called free-internal-rotor case) is given by
d2
H ¼ Fp2s ¼ F ; ð3:4:48Þ
ds2
and for another extreme case V3 ! 1; the internal-rotation motion is restricted to
oscillation of small amplitude around s ¼ 0 (or s ¼ 2p=3 or s ¼ 4p=3). In this
case the Hamiltonian operator is given as
9
H ¼ Fp2s þ V3 s2 ; ð3:4:49Þ
4
by truncating after the second term of the expansion of VðsÞ around s ¼ 0

1 9 2 81 4
VðsÞ V3 1 1 s þ s þ
2 2 24
ð3:4:50Þ
9 81
¼ V 3 s 2 V3 s 4 þ
4 48
As known well, for the Hamiltonian given in Eq. 3.4.48 we have
Hjmi ¼ Em jmi;
1 ð3:4:51Þ
jmi ¼ pffiffiffiffiffiffi eims ; Em ¼ Fm2 ; ðm ¼ 0; 1; 2; . . .Þ
2p
and that for Eq. 3.4.49 we have
Hjvi ¼ Ev jvi;
pffiffiffiffiffiffiffiffi 1 ð3:4:52Þ
Ev ¼ 3 FV3 ðv þ Þ; ðv ¼ 0; 1; 2; . . .Þ
2
where jvi denotes the well-known harmonic oscillator functions. Symmetry
properties of the free-internal-rotor eigenfunctions under the PI group G3 and G6
are as follows:
jmi ðm ¼ 0 mod 3Þ : A symmetry species;
ð3:4:53Þ
j mi ðm ¼ 1 or m ¼ 2 mod 3Þ : Esep symmetry species;
for G3 ; and
pffiffiffi jmi ¼ j0i : A1 symmetry species;

ð1=p2ffiffiffiÞðjmi þ j miÞ ðm ¼ 0 mod 3 [ 0Þ : A1 symmetry species,
ð1= 2Þðjmi j miÞ ðm ¼ 0 mod 3 [ 0Þ : A2 symmetry species,
j mi ðm ¼ 1 or 2 mod 3Þ : E symmetry species,
ð3:4:54Þ
for G6 :
Symmetry properties of jvi under the PI group G3 and G6 is described next.
Qualitative Interpretation of Internal-Rotation Energy Level Patterns

for a High Barrier Case Based on Tunneling Matrix Idea
Using a tunneling matrix approach, we try to understand aspect of energy levels

produced by the Schrödinger equation (3.4.42) with Eq. 3.4.43 of the pure internal-
rotation problem. In the high barrier limit, the internal-rotation motion is con-
sidered to be such that an oscillation of small amplitude around either one of the
three minima tunnels to an oscillation around another minimum. Based on this
idea, we construct approximate eigenfunctions from wavefunctions localized
at three potential minima; that is s ¼ 0; s ¼ 2p=3 and s ¼ 4p=3; we write
those three localized functions as j1i; j2i and j3i; respectively (see Fig. 3.26). We
call these functions ‘‘framework functions’’ at framework 1–3, respectively. The
framework function j1i can be approximated by the solution of the Schrödinger
equation based on the Hamiltonian operator given by Eq. 3.4.49,

9
H0 wv ¼ Fp2s þ V3 s2 wv ¼ Ev wv ; ð3:4:55Þ
4
Fig. 3.26 Potential minima

and framework functions
i.e., the framework function at framework 1 is approximately equal to the solution

wv ;
j1; vi ¼ wv : ð3:4:56Þ
In Eq. 3.4.56, v is used for representing the vth vibrational state: v ¼ 0 corresponds
to the ground vibrational state, v ¼ 1 to the first excited torsional state and so on.
The three framework functions are expressed as
j1; vi Uv ðsÞ ð3:4:57Þ
j2; vi ¼ ð123Þj1; vi ¼ Uv ðs þ 2p=3Þ ð3:4:58Þ
j3; vi ¼ ð132Þj1; vi ¼ Uv ðs þ 4p=3Þ ð3:4:59Þ
(We assume Uv ðsÞ ¼ Uv ðs 2pÞ). By making suitable linear combinations of

framework functions, we have symmetrized wavefunctions belonging to symmetry
species of the PI group G3 or G6 as follows, with S ¼ h1; vj2; vi;
1
jA; vi ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ðj1; vi þ j2; vi þ j3; viÞ;
3ð1 þ 2SÞ
1

jEa ; vi ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi j1; vi þ x2 j2; vi þ xj3; vi ; ð3:4:60Þ
3ð1 SÞ
1

jEb ; vi ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi j1; vi þ xj2; vi þ x2 j3; vi ;
3ð1 SÞ
where Ea and Eb correspond to Eþ and E for the case of a C1 molecule ðG3 Þ; and
they represent two components of E species for the Cs molecule ðG6 Þ: For a Cs
molecule, jA; vi is replaced by jA1 ; vi for even v or jA2 ; vi for odd v: The three
functions shown in Eq. 3.4.60 are (approximate) eigenfunctions of the Schrödinger
Eq. 3.4.42, and eigenvalues are expressed as
1
hA; vjHjA; vi ¼ ðh1; vjHj1; vi þ 2h1; vjHj2; viÞ
1 þ 2S
ð3:4:61Þ
2
h1; vjH0 j1; vi þ h1; vjH 0 j2; vi;
1 þ 2S
hEa ; vjHjEa ; vi ¼ hEb ; vjHjEb ; vi

1
¼ ðh1; vjHj1; vi h1; vjHj2; viÞ ð3:4:62Þ
1S
1
h1; vjH0 j1; vi h1; vjH 0 j2; vi;
1S
where we used
H ¼ H0 þ H 0 ð3:4:63Þ
Fig. 3.27 Schematic diagram of one-top internal-rotation energy levels for cases C1 ðG3 Þ and
Cs ðG6 Þ: (Symmetry species in parentheses are those for G6 )
h1; vjHj1; vi h1; vjH0 j1; vi ¼ Ev ð3:4:64Þ
h1; vjHj2; vi ¼ h1; vjH0 j2; vi þ h1; vjH 0 j2; vi

ð3:4:65Þ
¼ Ev S þ h1; vjH 0 j2; vi
Since H 0 takes negative values in the intermediate region between framework 1

and 2, we have
h1; v ¼ 0jH 0 j2; v ¼ 0i\0; h1; v ¼ 1jH 0 j2; v ¼ 1i [ 0; . . . ð3:4:66Þ
The tunneling matrix approach for the case involving torsional–rotational

interaction is written later. Schematic pure internal-rotation energy level patterns
are shown in Fig. 3.27, where both the extreme cases are depicted.
3.4.1.4 Qualitative Understanding of Pure Internal-Rotation

Energy Levels for Cs ðG12 Þ Case
We describe briefly the tunneling matrix approach for the case of a CH3 BF2 -like
molecule. Although the molecule has a low barrier height to internal rotation, we
consider here a hypothetical case of a CH3 BF2 -like molecule with high barrier. For
this case we have six framework functions expressed as
j1; vi UðsÞ; ð3:4:67Þ
jn; vi ¼ an1 j1; vi ¼ Uðs 2p½n 1=6Þ; ðn ¼ 1; 2; 3; 4; 5; 6Þ ð3:4:68Þ
where a ¼ ð123Þð45Þ: Symmetrized wavefunctions constructed from these six

framework functions are given as
j1; vi þ j2; vi þ j3; vi þ j4; vi þ j5; vi þ j6; vi

jA01 ; vi ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ;
6ð1 þ 2S1 þ 2S2 þ S3 Þ
j1; vi j2; vi þ j3; vi j4; vi þ j5; vi j6; vi
jA002 ; vi ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ;
6ð1 2S1 þ 2S2 S3 Þ
j1; vi þ 2 j2; vi þ 4 j3; vi þ j4; vi þ 4 j5; vi þ 2 j6; vi
jEa0 ; vi ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ;
6ð1 S1 S2 þ S3 Þ
ð3:4:69Þ
j1; vi þ 2 j2; vi þ 4 j3; vi þ j4; vi þ 4 j5; vi þ 2 j6; vi
jEb0 ; vi ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ;
6ð1 S1 S2 þ S3 Þ
j1; vi þ j2; vi þ 2 j3; vi j4; vi þ 2 j5; vi þ 1 j6; vi
jEa00 ; vi ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ;
6ð1 þ S1 S2 S3 Þ
j1; vi þ 1 j2; vi þ 2 j3; vi j4; vi þ 2 j5; vi þ j6; vi
jEb00 ; vi ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ;
6ð1 þ S1 S2 S3 Þ
where ¼ expði2p=6Þ; S1 ¼ h1; vj2; vi; S2 ¼ h1; vj3; vi; and S3 ¼ h1; vj4; vi:
Equations 3.4.69 are expressions for the case of ð23Þð45Þ
j1; vi ¼ þj1; vi: For the
case of ð23Þð45Þ
j1; vi ¼ j1; vi; jA01 ; vi and jA002 ; vi should be changed to jA02 ; vi
and jA001 ; vi; respectively. The energy eigenvalues for the pure internal-rotation
problem are given by
hA01 ; vjHjA01 ; vi ¼ ðh1; vjHj1; vi þ 2h1; vjHj2; vi þ 2h1; vjHj3; vi þ h1; vjHj4; viÞ=NA01 ;
hA002 ; vjHjA002 ; vi ¼ ðh1; vjHj1; vi 2h1; vjHj2; vi þ 2h1; vjHj3; vi h1; vjHj4; viÞ=NA002 ;
hE0 ; vjHjE0 ; vi ¼ ðh1; vjHj1; vi h1; vjHj2; vi h1; vjHj3; vi þ h1; vjHj4; viÞ=NE0 ;
hE00 ; vjHjE00 ; vi ¼ ðh1; vjHj1; vi þ h1; vjHj2; vi h1; vjHj3; vi h1; vjHj4; viÞ=NE00 ;
ð3:4:70Þ
where NA01 ¼ 1 þ 2S1 þ 2S2 þ S3 ; NA002 ¼ 1 2S1 þ 2S2 S3 ; NE0 ¼ 1 S1 S2 þ

S3 ; and NE00 ¼ 1 þ S1 S2 S3 :
Equations 3.4.70 are expressions for the case of
ð23Þð45Þ
j1; vi ¼ þj1; vi: ð3:4:71Þ
For the case of

ð23Þð45Þ
j1; vi ¼ j1; vi; ð3:4:72Þ
jA01 ; vi and jA002 ; vi should be changed to jA02 ; vi and jA001 ; vi; respectively.
Combining Eq. 3.4.70 with the description for cases C1 ðG3 Þ and Cs ðG6 Þ in
the previous section, schematic energy level diagrams depicted in Fig. 3.28 can be
understood easily. V6 used in Fig. 3.28 is the barrier height parameter of the
potential function
1
VðsÞ ¼ V6 ð1 cos 6sÞ: ð3:4:73Þ
2
Fig. 3.28 Schematic diagram of one-top internal-rotation energy levels for case Cs ðG12 Þ
3.4.1.5 Construction of Hamiltonian Matrix with the Use of Free Rotor

Wavefunctions
We will give expressions for Hamiltonian matrix elements in the traditional

treatment, where free rotor torsional functions multiplied by symmetric top rota-
tional functions are used as basis set wavefunctions. We adopt the RAM system
since only DK ¼ 0 matrix elements occur in the torsional–rotational interaction
and it makes the construction of the Hamiltonian matrix somewhat convenient.
C1 ðG3 Þ Case
We consider at first the case C1 ðG3 Þ; where the Hamiltonian operator is given by
H ¼ AJz2 þ BJx2 þ CJy2
þ Dzx ðJz Jx þ Jx Jz Þ þ Dxy ðJx Jy þ Jy Jx Þ þ Dyz ðJy Jz þ Jz Jy Þ
1
þ F ðps qJz Þ2 þ V3 ð1 cos 3sÞ: ð3:4:74Þ
2
As seen in Eq. 3.4.74, we adopt the simplest potential function for simplicity. As a
basis set function, a product form of a free rotor torsional function jmi and a
symmetric top rotational function jJ; K; Mi is used, that is,
jm; J; K; Mi ¼ jmijJ; K; Mi; ð3:4:75Þ
pffiffiffiffiffiffi
where jmi ¼ ð1= 2pÞ expðimsÞ:
Hereinafter we drop the laboratory-fixed angular momentum quantum number

M in the rotational function jJ; K; Mi since we treat an isolated molecule in free
space and its energy does not depend on the quantum number M:
Since jm; J; Ki has a symmetry property

2pm
ð123Þjm; J; Ki ¼ exp i jm; JKi; ð3:4:76Þ
3
symmetry species of jm; J; Ki are given as

jm; J; Ki for m ¼ 0; 3; 6; . . . : A symmetry species,
jm; J; Ki for m ¼ . . .; 5; 2; 1; 4; 7; . . . : Eþ symmetry species, ð3:4:77Þ
jm; J; Ki for m ¼ . . .; 4; 1; 2; 5; 8; . . . : E symmetry species,
Thus, the Hamiltonian matrix is divided into two blocks with A and
Esep ðEþ or E Þ species. Non-zero matrix elements of hm; J; KjHjm0 ; J; K 0 i are
expressed as

1 1 1
hm; J; KjHjm; J; Ki ¼ ðB þ C ÞJ þ A ðB þ C Þ K 2 þ F ðm qK Þ2 þ V3
2
2 2 2
ð3:4:78Þ

1 2 2

hm; J; KjHjm; J; K 2i ¼ J; Kj ðB CÞ Jx Jy þ Dxy Jx Jy þ Jy Jx jJ; K 2
2
ð3:4:79Þ

hm; J; KjHjm; J; K 1i ¼ hJ; KjDzx ðJz Jx þ Jx Jz Þ þ Dyz Jy Jz þ Jz Jy jJ; K 1i
ð3:4:80Þ
1
hm; J; KjHjm 3; J; Ki ¼ V3 ; ð3:4:81Þ
4
In a usual spectral analysis, a suitable type of centrifugal distortion terms is added,
and the numerical diagonalization of the Hamiltonian matrix and the iteration
process of the least squares fit are employed.
The torsional–rotational energy levels of methanol ðCH3 OHÞ in the ground
vibrational state, shown schematically in Fig. 3.29, will help us to understand the
behavior of torsional–rotational energy levels, which are connected with Hamil-
tonian matrix elements described above.
Cs ðG6 Þ Case
In this case, expressions shown above are modified somewhat. Since the
Hamiltonian operator should satisfy ð23Þ
H ¼ H; Dyz and Dxy terms disappear in
Eq. 3.4.74 and a potential function VðsÞ should satisfy VðsÞ ¼ VðsÞ necessarily
Fig. 3.29 Schematic torsional–rotational energy level diagram of methanol in the ground
vibrational state. In order to designate rotational levels, J; Ka ; Kc is used for levels belonging to A-
symmetry species as for those of an ordinary asymmetric top molecule, and J; K is used for E-
symmetry levels
(we assume a molecule has an a; b-plane of symmetry and a; b; c $ z; x; y). Basis

set wavefunctions belonging to each of symmetry species are given as follows: for
A1 species,
1
jA1 ; m; J; Ki ¼ pffiffiffi jmijJ; Ki þ ð1ÞJK j mijJ; Ki
2
ðm ¼ 3n with n ¼ 0; 1; 2; . . .; and K ¼ J; J þ 1; . . .; 0; . . .; þJÞ;
ð3:4:82Þ
for A2 species,
1
jA2 ; m; J; Ki ¼ pffiffiffi jmijJ; Ki ð1ÞJK j mijJ; Ki
2
ðm ¼ 3n with n ¼ 0; 1; 2; . . .; and K ¼ J; J þ 1; . . .; 0; . . .; þJÞ;
ð3:4:83Þ
and for E species,
jE; m; J; Kia ¼ jmijJ; Ki; ð3:4:84Þ
jE; m; J; Kib ¼ ð23Þ

jE; m; J; Kia ¼ ð1ÞJK j mijJ; Ki
ðm ¼ 3n þ 1 or n ¼ 3n þ 2 with n ¼ 0; 1; 2; . . .; and ð3:4:85Þ
K ¼ J; J þ 1; . . .; 0; . . .; þJÞ:
pffiffiffi
In the equations above 1= 2 is replaced by 1=2 when m ¼ K ¼ 0:
In Eqs. 3.4.84 and 3.4.85, jE; m; J; Kia and jE; m; J; Kib represent pair states for
the degenerate E symmetry species.
The matrix element expressions hm; J; KjHjm0 ; J; K 0 i shown in Eqs. 3.4.78–
3.4.81 are available also for this case after setting Dxy ¼ Dyz ¼ 0:
When we treat Hamiltonian matrices for A1 and A2 species separately in writing
a computing program, we have to be very careful since we use Wang-type-like
functions as shown in Eqs. 3.4.82 and 3.4.83, involving a subtle phase factor. The
author recommends to treat the A1 and A2 species simultaneously by using A
species functions such as shown in Eq. 3.4.77 and without using Wang-type-like
functions although the matrix size is doubled.
Cs ðG12 Þ Case
Since the C3 axis of the methyl group is along the inertial b-axis in the CH3 BF2
molecule, the RAM system, in which the axis correspondence is such as b; c; a $
z; x; y; may not be necessarily convenient for such a case. It is thus difficult to tell
which system is preferable for a spectral analysis of the CH3 BF2 molecule, PAM
or RAM. We give anyhow a formulation for a CH3 BF2 -like molecule with the use
of the RAM system. Transformation properties of variables given in Table 3.10
lead to the Hamiltonian operator
1
H ¼ BJz2 þ CJx2 þ AJy2 þ F ðps qJz Þ2 þ V6 ð1 cos 6sÞ; ð3:4:86Þ
2
where we adopt the simplest potential function.
Basis set functions belonging A01 and A02 symmetry species are given as follows.
1
jA01 ; m; J; Ki ¼ pffiffiffi jmijJ; Ki þ ð1ÞJK jmijJ; Ki ;
2 ð3:4:87Þ
with ðm ¼ 0 mod 6; K ¼ even 0Þ or ðm ¼ 3 mod 6; K ¼ odd [ 0Þ;
1
jA02 ; m; J; Ki ¼ pffiffiffi jmijJ; Ki ð1ÞJK jmijJ; Ki ;
2 ð3:4:88Þ
with ðm ¼ 0 mod 6; K ¼ even 0Þ or ðm ¼ 3 mod 6; K ¼ odd [ 0Þ;
1
jA001 ; m; J; Ki ¼ pffiffiffi jmijJ; Ki ð1ÞJK jmijJ; Ki ;
2 ð3:4:89Þ
with ðm ¼ 0 mod 6; K ¼ odd [ 0Þ or ðm ¼ 3 mod 6; K ¼ even 0Þ;
1
jA002 ; m; J; Ki ¼ pffiffiffi jmijJ; Ki þ ð1ÞJK jmijJ; Ki ;
2 ð3:4:90Þ
with ðm ¼ 0 mod 6; K ¼ odd [ 0Þ or ðm ¼ 3 mod 6; K ¼ even 0Þ;
pffiffiffi
where 1= 2 is replaced by 1/2 when m ¼ K ¼ 0; and m takes values of plus,
minus and 0 within the restriction shown in parentheses.
Basis set functions E0 and E00 symmetry species are given by
jEa0 ; m; J; Ki ¼ jmijJ; Ki; ð3:4:91Þ
jEb0 ; m; J; Ki ¼ ð23Þð45Þ
jEa0 ; m; J; Ki
¼ ð1ÞJK j mijJ; Ki;

with m ¼ 6n þ 2; or 6n þ 4; n ¼ 0; 1; 2; . . .; K ¼ even 0;
or m ¼ 6n þ 1; or 6n þ 5; n ¼ 0; 1; 2; . . .; K ¼ odd [ 0; ð3:4:92Þ
jEa00 ; m; J; Ki ¼ jmijJ; Ki; ð3:4:93Þ
jEb00 ; m; J; Ki ¼ ð23Þð45Þ
jEa00 ; m; J; Ki
¼ ð1ÞJK j mijJ; Ki;

with m ¼ 6n þ 1; or 6n þ 5; n ¼ 0; 1; 2; . . .; K ¼ even 0;
or m ¼ 6n þ 2; or 6n þ 4; n ¼ 0; 1; 2; . . .; K ¼ odd [ 0: ð3:4:94Þ
Hamiltonian matrix elements for each of symmetry species are evaluated by
using the basis set functions written above, and the matrix elements
hm; J; KjHjm0 ; J; K 0 i;
which are obtained by making the replacement V3 ! V6 in Eq. 3.4.78 and by

putting
hm; J; KjHjm 6; J; Ki ¼ V6 =4 ð3:4:95Þ
in place of Eq. 3.4.81. We note in passing that a Hamiltonian matrix element

hm; J; KjHjm 3; J; K 1i appear possibly by adding a term such as
aðcos 3sÞJy þ bðsin 3sÞJx ; ða; b: constantsÞ ð3:4:96Þ
to the Hamiltonian operator.
3.4.1.6 An Approximate Expression of Torsional–Rotational Energy
We will derive an approximate expression of torsional–rotational energy for the

C1 ðG3 Þ or Cs ðG6 Þ case of a nearly prolate top molecule. From Eqs. 3.4.78–3.4.81,
we have

1 1
hm; J; KjHjm; J; Ki ¼ ðB þ CÞJ 2 þ A ðB þ CÞ K 2
2 2
1
þ F ðm qK Þ2 þ V3 ; ð3:4:97Þ
2
hm; J; KjHjm; J; K 2i 0; ð3:4:98Þ
hm; J; KjHjm; J; K 1i 0; ð3:4:99Þ
1
hm; J; KjHjm 3; J; Ki ¼ V3 : ð3:4:100Þ
4
The above equations lead to

1 1 1
h3n; J; KjHj3n; J; Ki ¼ ðB þ CÞJ þ A ðB þ CÞ K 2 þ F ð3n qK Þ2 þ V3 ;
2
2 2 2
ð3:4:101Þ

1 1 1
¼ ðB þ CÞJ 2 þ A ðB þ CÞ K 2 þ F ð3n þ qK Þ2 þ V3 ; ð3:4:102Þ
2 2 2

1 1 1
¼ ðB þ CÞJ þ A ðB þ CÞ K 2 þ F ½3ðn þ 1Þ ðqK þ 3Þ2 þ V3 ;
2
2 2 2
ð3:4:103Þ
1
h3n; J; KjHj3ðn 1Þ; J; Ki ¼ V3 : ð3:4:104Þ
4
with n ¼ 0; 1; 2; . . .; for A species, and

1 1
h3n þ 1; J; KjHj3n þ 1; J; Ki ¼ ðB þ CÞJ þ A ðB þ CÞ K 2
2
2 2
2 1
þ F ½3n þ ð1 qKÞ þ V3 ; ð3:4:105Þ
2

1 1 1
¼ ðB þ CÞJ þ A ðB þ CÞ K 2 þ F ½3n ð1 qKÞ2 þ V3 ; ð3:4:106Þ
2
2 2 2

1 1 1
¼ ðB þ CÞJ 2 þ A ðB þ CÞ K 2 þ F ½3ðn þ 1Þ þ ð1 qK 3Þ2 þ V3 ;
2 2 2
ð3:4:107Þ
1
h3n þ 1; J; KjHj3ðn 1Þ þ 1; J; Ki ¼ V3 : ð3:4:108Þ
4
with n ¼ 0; 1; 2; . . .; for E species.
By noting that (i) since the Hamiltonian matrix elements are defined by indices
n’s for n ¼ 0; 1; 2; . . .; 1 as seen in Eqs. 3.4.101, 3.4.104, 3.4.105 and
3.4.108, the eigenvalues are unchanged for n ! n and/or n ! n 1; and (ii)
n ! n and n ! n þ 1 are equivalent to c ! c and c ! c þ 3 (see Eq. 3.4.110
below for the definition of c), respectively, as shown in Eq. 3.4.103 and 3.4.107,
we have
WC;J;K:l ðcÞ ¼ WC;J;K:l ðcÞ ¼ WC;J;K:l ðc þ 3Þ; ð3:4:109Þ
where WC;J;K:l ¼ hC; J:K; ljHjC; J:K; li; ðC ¼ A; E; l ¼ 1; 2; 3; . . .Þ; and
c ¼ qK ðfor C ¼ AÞ;
ð3:4:110Þ
c ¼ qK 1 ðfor C ¼ EÞ;
Accordingly, we have

1 1
WA;J;K;l ¼ ðB þ CÞJ þ A ðB þ CÞ K 2
2
2 2
X1 ð3:4:111Þ
1 2pk
þ V3 þ alk cos qK ;
2 k¼0
3

1 1
WE;J;K;l ¼ ðB þ CÞJ 2 þ A ðB þ CÞ K 2
2 2
X1 ð3:4:112Þ
1 l 2pk
þ V3 þ ak cos ðqK 1Þ ;
2 k¼0
3
where alk ’s are coefficients in the Fourier expansions.
3.4.1.7 Perturbation Treatment and Effective Rotational Hamiltonian

Operator
A rotational Hamiltonian operator effective for a specific vth vibrational state (see
Fig. 3.27 and 3.28) was derived by Herschbach [47] using a second order per-
turbation theory. Such an effective rotational Hamiltonian will be useful for
understanding of the rotational energy level structure in a specific vibrational state.
We put the PAM Hamiltonian operator as
H ¼ H0 þ H0; ð3:4:113Þ
H 0 ¼ AJz2 þ BJx2 þ CJy2 þ Fp2 þ Fp2s þ VðsÞ; ð3:4:114Þ
H 0 ¼ 2Fps p; ð3:4:115Þ
P
where p ¼ a¼x;y;z qa Ja : In terms of the second order perturbation treatment we
have
X jhv; CjH 0 jv0 ; C0 ij2

Heff ¼ hv; CjH 0 jv; Ci þ hv; CjH 0 jv; Ci þ
v0 6¼v
Ev;C Ev0 :C0
¼ AJz2 þ BJx2 þ CJy2 þ Fp2 þ hv; CjFp2s þ VðsÞjv; Ci
X X jhv; Cjps jv0 ; C0 ij2
2F qa Ja hv; Cjps jv; Ci þ 4F 2 p2 ð3:4:116Þ
a¼x;y;z v0 6¼v
Ev;C Ev0 :C0
In Eq. 3.4.116 C represents symmetry species: C ¼ A or Esep for case C1 ðG3 Þ; and
C ¼ A1 ; A2 or E for case Cs ðG6 Þ: The C0 in the summation of the last term is equal
to C for C1 ðG3 Þ case. For case Cs ðG6 Þ; C0 ¼ A2 ðA1 Þ for C ¼ A1 ðA2 Þ and C0 ¼ E
for C ¼ E: The wavefunction jv; Ci appearing in Eq. 3.4.116 is an eigenfunction of
a pure internal-rotation problem described by a Hamiltonian operator
Hinternalrotation ¼ Fp2s þ VðsÞ: ð3:4:117Þ
For C to be a non-degenerate symmetry species (C ¼ A; A1 or A2 , etc.), jv; Ci can

be chosen to be real. Then we have

hv; Cjps jv; Ci ¼ hv; Cjpys jv; Ci

¼ hv; Cjpy
s jv; Ci ð3:4:118Þ
¼ hv; Cjps jv; Ci
¼ 0;
where we used pys ¼ ps and p

s ¼ ps (y and
denote Hermitian conjugation and
complex conjugation, respectively). That is,
hv; Cjps jv; Ci ¼ 0; ð3:4:119Þ
for a non-degenerate species. From Eqs. 3.4.116 and 3.4.119 we have the fol-
lowing expressions for the effective rotational Hamiltonian operator in the specific
vibrational state after suitable coordinate axis rotations for removing cross terms
such as Jx Jy þ Jy Jx :
Heff ðv; AÞ ¼ W0 ðv; AÞ þ Aeff ðv; AÞJz2 þ Beff ðv; AÞJx2 þ Ceff ðv; AÞJy2 ; ð3:4:120Þ
Heff ðv; EÞ ¼ W0 ðv; EÞ þ Aeff ðv; EÞJz2 þ Beff ðv; EÞJx2 þ Ceff ðv; EÞJy2
þ qeff ðvÞJz þ reff ðvÞJx þ seff ðvÞJy ; ð3:4:121Þ
where the symmetry species notation A is used for A in case C1 ðG3 Þ; for A1 or A2
in case Cs ðG6 Þ; or for A01 ; A02 ; A001 or A002 in case Cs ðG12 Þ; and E for Esep in case
C1 ðG3 Þ; for E in case Cs ðG6 Þ; or for E0 or E00 in case Cs ðG12 Þ:
Coefficients appearing in Eqs. 3.4.120 and 3.4.121 are given by
W0 ðv; C ¼ A or EÞ ¼ hv; CjFp2s þ VðsÞjv; Ci ð3:4:122Þ

!
X jhv; Cjps jv0 ; C0 ij2
Aeff ðv; C ¼ A or EÞ A þ Fq2z 1 þ 4F ; etc. ð3:4:123Þ
v0 6¼v
Ev;C Ev0 :C0
qeff ðvÞ 2Fqz hv; Cjps jv; Ci; etc. ð3:4:124Þ
In Eqs. 3.4.123 and 3.4.124, is used in place of = because we take small changes
caused by the coordinate axis rotation for removing the cross terms into
consideration.
3.4.1.8 Tunneling Matrix Formulation of the Torsional–Rotational Problem
We describe here the tunneling matrix formulation for the torsional–rotational

problem. As mentioned in Sect. 3.4.1.3 above for the pure internal-rotation
problem, this formulation is valid for a high barrier case. This formalism was
developed first by Hougen [48] for analyzing a microwave spectrum of the
hydrazine molecule N2 H4 ; which has an umbrella (inversion) motion of the two
NH2 group and an internal rotation of the two NH2 groups about the N–N bond as
large-amplitude vibrations. Characteristics of the tunneling matrix formalism are
summarized as follows:
1. It is useful for describing a vibrational–rotational problem involving large-
amplitude vibrations with high potential barriers.
2. It is available only for treating vibrational–rotational energy levels within a
specific vibrational state.
3. Large-amplitude-vibrational functions are constructed from wavefunctions
localized at potential minima. In the one-top internal-rotation case, we have
three functions localized at the three potential minima, and we express them as
j1i; j2i and j3i:
4. A kinetic model for describing large-amplitude-vibrations involves more loose
restriction than that for the traditional formulation does; only the symmetry
requirements should be satisfied. For a one-top internal-rotation problem, for an
example, S1 ð0; 0; sÞai ðai : a constant vector) is replaced by ai ðsÞ in Eqs. 3.4.1
and 3.4.2, and ai ðs þ 2p=3Þ ¼ ai ðsÞ, etc. is imposed.
5. An effective rotational Hamiltonian operator used in a spectral analysis is
written from a phenomenological consideration.
6. Quantities determined from the spectral analysis are the tunneling matrix ele-
ments expressed as h1jAj2i; for an example.
7. The molecular symmetry group which is used in the group theoretical con-
sideration is the permutation–inversion group proposed by Longuet-Higgins
[35].
We will describe below (a) the basis set functions for constructing the Ham-
iltonian matrix, (b) parameterized expressions of the Hamiltonian matrix elements,
(c) a note on a least square analysis with the tunneling matrix formulation, and (d)
comparison with the perturbation treatment of the traditional formalism. For

simplicity, we restrict the present description mainly to case C1 ðG3 Þ since no
essential changes appear for other cases.
Basis Set Functions
Three localized vibrational functions are defined using permutation operations as

j1i; j2i ¼ ð123Þj1i; and j3i ¼ ð132Þj1i; ð3:4:125Þ
where j1i is a vibrational wavefunction localized at potential minimum 1 in a
specific vibrational level. The functions j1i; j2i; and j3i thus defined are frame-
work functions localized at framework 1–3, respectively. Symmetrized basis set
functions belonging to symmetry species A; Eþ ; and E of the PI group G3 (see
Table 3.5) are constructed from these three framework functions:
1
jA; torsioni ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ðj1i þ j2i þ j3iÞ; ð3:4:126Þ
3ð1 þ 2SÞ
1
jEþ ; torsioni ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ðj1i þ x
j2i þ xj3iÞ; ð3:4:127Þ
3ð1 SÞ
1
jE ; torsioni ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ðj1i þ xj2i þ x
j3iÞ; ð3:4:128Þ
3ð1 SÞ
where
h1j1i ¼ h2j2i ¼ h3j3i ¼ 1; ð3:4:129Þ
h1j2i ¼ h2j1i ¼ h1j3i ¼ h3j1i ¼ h2j3i ¼ h3j2i ¼ S: ð3:4:130Þ
When we use PAM coordinate system written in Eqs. 3.4.1 and 3.4.2, Eulerian
angles ðh; /; vÞ are unchanged under the permutation operation (123), and
accordingly the symmetric top rotational function jJ; Ki is also unchanged.
Therefore, a symmetrized torsional–rotational wavefunction belonging to a sym-
metry species C is given by
jC; J; Ki ¼ jC; torsionijJ; Ki: ð3:4:131Þ
In passing, let us consider case Cs ðG6 Þ briefly. When we put
j1i ¼ UðsÞ; ð3:4:132Þ
we have
ð23Þ
j1i ¼ ð23Þ
UðsÞ ¼ UðsÞ: ð3:4:133Þ
(Note: UðsÞ ¼ UðsÞ for the v ¼ 0 torsional state, UðsÞ ¼ UðsÞ for the v ¼ 1
torsional state, etc.) On the other hand, from Eq. 3.4.125
ð23Þ
ðj1i þ j2i þ j3iÞ ¼ ð23Þ
j1i þ ð132Þð23Þ
j1i þ ð123Þð23Þ
j1i: ð3:4:134Þ
From Eqs. 3.4.133 and 3.4.134, we have
ð23Þ
ðj1i þ j2i þ j3iÞ ¼ þðj1i þ j2i þ j3iÞ; forð23Þ
j1i ¼ þj1i ð3:4:135Þ
ð23Þ
ðj1i þ j2i þ j3iÞ ¼ ðj1i þ j2i þ j3iÞ; for ð23Þ
j1i ¼ j1i ð3:4:136Þ
Accordingly, jA; torsioni shown in Eq. 3.4.126 is the A1 or A2 vibrational wave-
function for case Cs ðG6 Þ according to ð23Þ
j1i ¼ þj1i or ¼ j1i; respectively.
Vibrational functions jEþ ; torsioni and jE ; torsioni shown in Eqs. 3.4.127 and
3.4.128 for case C1 ðG3 Þ become pair functions belonging to the degenerate E
species for case Cs ðG6 Þ:
Expressions of Hamiltonian Matrix Elements
Without a somewhat subtle discussion for eliminating cross terms of angular

momentum component operators, we write simply a phenomenological Hamilto-
nian operator as
H ¼ hv þ AJz2 þ BJx2 þ CJy2 þ iqJz þ irJx þ isJy : ð3:4:137Þ
In Eq. 3.4.137, coefficients hv ; A; B; C; q; r; and s are operators (functions) of the

variable describing the internal rotation. Jx ; Jy ; and Jz are all invariant under the G3
group operations, and the coefficient operators belong all to A symmetry species.
Hermitian conjugation (y) and time-reversal operation (
) are important for
obtaining expressions for Hamiltonian matrix elements. The time-reversal opera-
tion is operationally equivalent to the complex conjugation. From the fact that the
Hamiltonian H is invariant under Hermitian conjugation and time-reversal
(complex conjugation), and
Jay ¼ Ja ; Ja
¼ Ja ; ð3:4:138Þ
where a ¼ x; y; and z; we have
U y ¼ U; U
¼ U; for U ¼ hv ; A; B; C; ð3:4:139Þ
and
V y ¼ V; V
¼ þV; for V ¼ q; r; s: ð3:4:140Þ
Since the coefficients are of A-symmetry, we obtain the following relations
between tunneling matrix elements by using the relations (3.4.139, 3.4.140)
h1jUj1i ¼ h2jUj2i ¼ h3jUj3i;

ð3:4:141Þ
h1jUj2i ¼ h2jUj1i ¼ h1jUj3i ¼ h3jUj1i ¼ h2jUj3i ¼ h3jUj2i;
and
h1jVj1i ¼ h2jVj2i ¼ h3jVj3i ¼ 0;
ð3:4:142Þ
h1jVj2i ¼ h2jVj1i ¼ h1jVj3i ¼ h3jVj1i ¼ h2jVj3i ¼ h3jVj2i:
Writing more in detail, for example,
h1jVj1i ¼ ðh1jV y j1iÞ

¼ h1jV y
j1i ¼ h1jVj1i ¼ 0;
h1jVj2i ¼ ðh2jV y j1iÞ
¼ h2jV y
j1i ¼ h2jVj1i;
h1jVj2i ¼ ð132Þh1jVj2i ¼ h3jVj1i ¼ h1jVj3i;
where we assume j1i to be real.

Using Eqs. 3.4.141 and 3.4.142, we have
1
hA; J; KjHjA; J; K 0 i ¼ ðh1jhv j1i þ 2h1jhv j2iÞdK;K 0 þðh1jAj1i
1 þ 2S
þ 2h1jAj2iÞK 2 dK;K 0 þ ðh1jBj1i þ 2h1jBj2iÞhJ; KjJx2 jJ; K 0 i
þ ðh1jCj1i þ 2h1jCj2iÞhJ; KjJy2 jJ; K 0 i; ð3:4:143Þ
1
hEþ ; J; KjHjEþ ; J; K 0 i ¼ ðh1jhv j1i h1jhv j2iÞdK;K 0 þ ðh1jAj1i
1S
h1jAj2iÞK 2 dK;K 0 þ ðh1jBj1i h1jBj2iÞhJ; KjJx2 jJ; K 0 i
pffiffiffi
þ ðh1jCj1i h1jCj2iÞhJ; KjJy2 jJ; K 0 i þ 3 h1jqj2iKdK;K 0

þ h1jrj2ihJ; KjJx jJ; K 0 i þ h1jsj2ihJ; KjJy jJ; K 0 iÞ:
ð3:4:144Þ
We omit to write the expression for the E species since it produces the same
energies as the Eþ species does. We call h1jAjniðn ¼ 1; 2; 3Þ, etc. the tunneling
matrix elements. We note that there is no contribution from the s term of (3.4.137),
in case Cs ðG6 Þ (the symmetry plane is assumed to be the xz-plane).
Note on a Least Square Analysis with the Tunneling Matrix Formulation
For the purpose of using Eqs. 3.4.143, 3.4.144 for the spectral analysis, it is
convenient to use A11 ; A12 ; . . .; q12 ; r12 ; and s12 defined in Eq. 3.4.145 below as
adjustable parameters in the least squares analysis since Eqs. 3.4.143, 3.4.144
include the overlap integral S in addition to tunneling matrix elements. Although
various ways are possible, the definition shown below is considered to be favorable
since they are approximately equivalent to the tunneling matrix elements because
of the small S value.
1
ðh1jAj1i þ 2h1jAj2iÞ ¼ A11 þ 2A12
1 þ 2S
1
ðh1jAj1i h1jAj2iÞ ¼ A11 A12
1S
1
ðh1jBj1i þ 2h1jBj2iÞ ¼ B11 þ 2B12
1 þ 2S
1
ðh1jBj1i h1jBj2iÞ ¼ B11 B12
1S
1
ðh1jCj1i þ 2h1jCj2iÞ ¼ C11 þ 2C12 ð3:4:145Þ
1 þ 2S
1
ðh1jCj1i h1jCj2iÞ ¼ C11 C12
1S
1
h1jqj2i ¼ q12
1S
1
h1jrj2i ¼ r12
1S
1
h1jsj2i ¼ s12
1S
For examples, A11 and A12 to be determined in the least squares analysis are
connected with the tunneling matrix elements by

1 1 2 1 2 2
A11 ¼ þ h1jAj1i þ h1jAj2i;
3 1 þ 2S 1 S 3 1 þ 2S 1 S
ð3:4:146Þ
1 1 1 1 2 1
A12 ¼ h1jAj1i þ þ h1jAj2i:
3 1 þ 2S 1 S 3 1 þ 2S 1 S
Another way for avoiding inconvenient non-zero overlap integrals among the
framework functions is to use newly defined framework functions jn : newi
ðn ¼ 1; 2; 3Þ which are orthogonal with each other. The new framework functions
jn : newiðn ¼ 1; 2; 3Þ are obtained by an appropriate transformation from the
framework functions jniðn ¼ 1; 2; 3Þ defined above as the functions localized at
potential minima. In the formalism using jn : newi in place of jni; tunneling matrix
elements h1jAjni, etc. are placed simply by h1 : newjAjn : newi, etc. and all the
overlap integrals vanish in the Hamiltonian matrix element expressions given in
Eqs. 3.4.143, 3.4.144. The transformation from jni to jn : newi is given briefly in
the Appendix (Sect. 3.4.5) for the general case of the tunneling matrix formalism.
Comparison with the Perturbation Treatment in the Traditional Formulation
The expressions Eqs. 3.4.143, 3.4.144 in the tunneling matrix formulation are
equivalent to the expressions Eqs. 3.4.122–3.4.124 for the effective rotational
Hamiltonian in the traditional treatment. Comparing Eqs. 3.4.143, 3.4.144 with
Eqs. 3.4.122–3.4.124, we have relations
2h1jhv j2i hv; AjFp2s þ VðsÞjv; Ai;

h1jhv j2i hv; EjFp2s þ VðsÞjv; Ei;
!
X jhv; Ajps jv0 ; Aij2
2h1jAj2i Fq2z 1 þ 4F ;
v0 6¼v
Ev;A Ev0 ;A
!
X jhv; Ejps jv0 ; Ej2 ð3:4:147Þ
h1jAj2i Fq2z 1 þ 4F ; etc.;
v0 6¼v
Ev;E Ev0 ;E
pffiffiffi
3h1jqj2i 2Fqz hv; Ejps jv; Ei;
pffiffiffi
3h1jrj2i 2Fqx hv; Ejps jv; Ei;
pffiffiffi
3h1jsj2i 2Fqy hv; Ejps jv; Ei:
The following relations will be practically useful for obtaining information on

the internal-rotation potential:
X jhv; Ajps jv0 ; Aij2 X jhv; Ejps jv0 ; Eij2
3A12 ¼ 4F 2 q2z 4F 2 q2z ; etc.;
v0 6¼v
Ev;A Ev0 ;A v0 6¼v
Ev;E Ev0 ;E
pffiffiffi
3q12 ¼ 2Fqz hv; Ejps jv; Ei; etc.
ð3:4:148Þ
3.4.1.9 Transition Selection Rules
We mention briefly the overall selection rule of the electric dipole transition.
When the transition jUi i ! jUj i occurs, the electric dipole transition moment
defined by
Z
Mij ¼ hUi jlZ jUj i ¼ U
i lZ Uj dV ð3:4:149Þ
has a non-zero value. In Eq. 3.4.149, lZ is the laboratory-fixed Z-component of the

electric dipole moment operator. According to a theorem of group theory, U
i lZ Uj
involves the totally symmetric property for non-zero Mij :
For a molecule of C1 ðG3 Þ symmetry, lZ is of A-symmetry species. Thus, the
overall selection rules are
A $ A; E $ E ðEþ $ Eþ ; E $ E Þ: ð3:4:150Þ
(Usually E $ E is used since the Eþ and E states are degenerate.)

For a molecule of Cs ðG6 Þ symmetry, lZ is of A2 -symmetry species, and the
overall selection rule are
A1 $ A2 ; E $ E: ð3:4:151Þ
It should be noted that A E splittings cannot be obtained directly from any usual
spectral observation.
3.4.2 Two-Top Internal-Rotation Problem
Considering a possibility that signals from molecules with two methyl groups are
observed astronomically, we describe torsional–rotational problem of molecules
with two methyl internal rotors in this section. In the present article only asym-
metric top molecules with two methyl internal rotors are treated, and symmetric
top molecules such as ethane (C2 H6 ) are not treated. Molecules with two
inequivalent methyl tops and those with two equivalent methyl tops will be dis-
cussed below separately.
3.4.2.1 Case of a Molecule with Two Inequivalent Methyl Internal Rotors
Group Theoretical Treatment
We treat here two cases: a molecule of C1 symmetry and of Cs symmetry. In Figs.

3.30 and 3.31, a C1 molecule and a Cs molecule with two inequivalent methyl
groups are shown, respectively. The PI group for the molecule of C1 symmetry
with two inequivalent methyl internal rotors is G9 : The symmetry of this kind of
molecule is denoted as C1 ðG9 Þ: Similarly, the symmetry of the molecule of Cs
symmetry with two inequivalent methyl internal rotors is denoted as Cs ðG18 Þ:
Character Table. We denote protons in one methyl group by 1, 2, 3 and the
other methyl protons by 4, 5, 6 (see Figs. 3.30 and 3.31). In Table 3.11, the
character table of the permutation–inversion group G9 for the C1 molecule is
given, and the character table of G18 for the Cs molecule is given in Table 3.12.
The two symmetry species in each of four pairs, ðE1þ ; E1 Þ; ðE2þ ; E2 Þ;
ðE3þ ; E3 Þ and ðE4þ ; E4 Þ; appearing in Table 3.11 are group theoretically of
different symmetry species but degenerate energetically (separable degeneracy),
Fig. 3.30 2-Butanol, an

example of a C1 molecule
with inequivalent two methyl
groups
Fig. 3.31 Ethylmethylether,

an example of a Cs molecule
with inequivalent two methyl
groups
Table 3.11 Character table for the PI group G9 of a C1 molecule with two inequivalent methyl
tops
E (123) (132) (456) (465) Pa Qa Ra Sa
A 1 1 1 1 1 1 1 1 1
E1
E1þ 1 x x
1 1 x x x
x
E1 1 x
x 1 1 x
x
x x
E2
E2þ 1 1 1 x x
x x
x x
E2 1 1 1 x
x x
x x
x
E3
E3þ 1 x x
x x
x
1 1 x
E3 1 x
x x
x x 1 1 x
E4
E4þ 1 x x
x
x 1 x
x 1
E4 1 x
x x x
1 x x
1
x ¼ expði2p=3Þ
a
P ¼ ð123Þð456Þ; Q ¼ ð123Þð465Þ; R ¼ ð132Þð456Þ; S ¼ ð132Þð465Þ
and the four pairs are combined to E1 ; E2 ; E3 and E4 ; respectively. Nine operations
ð23Þð56Þð78Þ
; ð13Þð56Þð78Þ
, etc. in Table 3.12 correspond to the reflection in the
plane of symmetry at an equilibrium configuration of the ethylmethylether mol-
ecule, which is an example of a molecule of Cs symmetry having inequivalent two
methyl groups.
Transition Selection Rules. Overall selection rules are as follows: electric dipole
transitions occur between states with same symmetry species except for A1 $ A2
in a Cs molecule. Accordingly, splittings produced by internal-rotation motion are
not obtained directly from any usual spectral observation.
Coordinate System and Transformation Properties of Variables. Next, let us
consider the coordinate system for the two-methyl top case and transformation
properties of variables appearing in the equations which represent the coordinate
system. We consider only the PAM system since the RAM and IAM systems are
impossible for the two-top problem except for very rare cases. The equations for
the two-top case corresponding to Eqs. 3.4.1, 3.4.2 for the one-top case are
given by
Table 3.12 Character table for the PI group G18 of a Cs molecule with two inequivalent methyl tops
E (123) (456) (123)(456) (123)(465) Pa
(132) (465) (132)(465) (132)(456)
A1 1 1 1 1 1 1
A2 1 1 1 1 1 1
E1 2 1 2 1 1 0
E2 2 2 1 1 1 0
E3 2 1 1 1 2 0
E4 2 1 1 2 1 0
a
P ¼ fð23Þð56Þð78Þ
ð23Þð46Þð78Þ
;
ð13Þð46Þð78Þ
ð23Þð45Þð78Þ
g

Ri ¼ R þ S1 ðh; /; vÞ S1 ðhP1 ; /P1 ; vP1 ÞS1 ð0; 0; s1 ÞSþ1 ðhP1 ; /P1 ; vP1 Þ

ai þ di Atop1 þ Atop1 ;
ð3:4:152Þ
for protons in methyl group 1 (top1: i ¼ 1; 2; 3),

Ri ¼ R þ S1 ðh; /; vÞ S1 ðhP2 ; /P2 ; vP2 ÞS1 ð0; 0; s2 ÞSþ1 ðhP2 ; /P2 ; vP2 Þ

ai þ di Atop2 þ Atop2 ;
ð3:4:153Þ
for protons in methyl group 2 (top2: i ¼ 4; 5; 6),
Ri ¼ R þ S1 ðh; /; vÞðai þ di Þ: ð3:4:154Þ
for the other nuclei, where ai and di ; represent the equilibrium positions and small-
amplitude displacements with vector components in the PAM x; y; z-axis system.
Atop1 and Atop2 represent centers of mass of three protons in methyl group 1 and
methyl group 2, respectively. We consider the axis system ðx1 ; y1 ; z1 Þ fixed to the
methyl top 1 where the origin is in the center of mass of the methyl top 1 and the
z1 -axis is along the C3 -axis of the methyl top 1. The rotation matrix
Sþ1 ðhP1 ; /P1 ; vP1 Þ rotates the ðx1 ; y1 ; z1 Þ-axes so that the ðx1 ; y1 ; z1 Þ-axes are par-
allel to the PAM ðx; y; zÞ-axes. The rotation matrix Sþ1 ðhP2 ; /P2 ; vP2 Þ is the cor-
responding one for the methyl top 2. We assume C3 symmetry of the methyl group
and therefore

S1 ð0; 0; 2p=3ÞSþ1 ðhP1 ; /P1 ; vP1 Þ a1 Atop1 ¼ Sþ1 ðhP1 ; /P1 ; vP1 Þ a2 Atop1 ;





ð3:4:155Þ
Table 3.13 Transformation properties of variables in the right of Eqs. 3.4.152–3.4.154 under
generating operationsa of G9 and G18
R (translation) h; /; v (rotation) s1 (internal rotation) s2 (internal rotation)
ð123Þ R h; /; v s1 þ 2p=3 s2
ð456Þ R h; /; v s1 s2 þ 2p=3
ð23Þð56Þð78Þ
R p h; p þ /; p v s1 s2
a
Generating operations of G9 are ð123Þ and ð456Þ; and generating operations of G18 are
ð123Þ; ð456Þ; and ð23Þð56Þð78Þ
Fig. 3.32 Internal-rotation

energy level diagram in the
ground vibrational state (the
J ¼ K ¼ 0 level) is illustrated
schematically for the inequiv-
alent-two-top case
For a Cs molecule, we set vP1 ¼ /P1 ¼ 0 and vP2 ¼ /P2 ¼ 0; assuming the plane
of symmetry to be the xz-plane. Transformation properties of variables in Eqs.
3.4.152–3.4.154 under generating operations of the PI group G9 and G18 are shown
in Table 3.13. The transformation properties under the generating operations (123)
and (456) are the same for G9 and G18 : We drop small-amplitude displacements di
in Table 3.13 since we consider here only the large-amplitude internal rotation.
3.4.2.2 Energy Level Splitting Pattern
A qualitative energy level diagram for the two-top problem is illustrated for J ¼
K ¼ 0 in Fig. 3.32. The J ¼ K ¼ 0 level at the left is first split into an A and E
component by internal-rotation motion of top 1. Each of these levels is then split
again into A and E component by internal-rotation motion of top 2. Finally, the
apparently fourfold degenerate level indicated as EE is split into two degenerate
levels ðE3 ; E4 Þ by top–top interaction.
3.4.2.3 Hamiltonian Operator
In the same way that we have obtained Eq. 3.4.9 for a one-top case, we have the
following expression for a PAM classical kinetic energy of a molecule with two
methyl internal rotors
1 X 1 X 1 X
T¼ Ia x2a þ Is1 s_ 21 þ Is1 s_ 1 k1a x1a þ Is2 s_ 22 þ Is2 s_ 2 k2a x2a ;
2 a¼x;y;z 2 a¼x;y;z
2 a¼x;y;z
ð3:4:156Þ
where Ia ða ¼ x; y; zÞ are the principal moments of inertia of the molecule, Is1 and
Is2 are the inertial moments of methyl top 1 and top 2 about their C3 axes,
respectively, and k1a and k2a (a ¼ x; y; z) are the direction cosines of the C3 axes of
top 1 and top 2, respectively.
By the same procedure employed in the one-top case, Eqs. 3.4.9–3.4.14, we
have the following classical kinetic energy expression written in terms of
the momenta conjugate to the variables representing the overall rotation and the
methyl internal rotations:
X 1
T¼ Ja2 þ F1 ðp1 p1 Þ2 þF2 ðp2 p2 Þ2
a¼x;y;z
2I a
þ F12 ½ðp1 p1 Þðp2 p2 Þ þ ðp2 p2 Þðp1 p1 Þ ð3:4:157Þ
where
X
p1 ¼ q1a Ja ;
a¼x;y;z
X
p2 ¼ q2a Ja ;
a¼x;y;z
R2
F1 ¼ ; ð3:4:158Þ
2ðR1 R2 Q2 ÞIs1
R1
F2 ¼ ;
2ðR1 R2 Q2 ÞIs2
Q
F12 ¼ pffiffiffiffiffiffiffiffiffiffi ;
2 ð R1 R2 Q 2 Þ I s 1 I s 2
with
q1a ¼ k1a Is1 =Ia ; ða ¼ x; y; zÞ

q2a ¼ k2a Is2 =Ia ; ða ¼ x; y; zÞ
X
R1 ¼ 1 q1a k1a ;
a¼x;y;z
X ð3:4:159Þ
R2 ¼ 1 q2a k2a ;
a¼x;y;z
X pffiffiffiffiffiffiffiffiffiffi
Q¼ k1a k2a Is1 Is2 =Ia ;
a¼x;y;z
The quantum mechanical PAM Hamiltonian operator is obtained by replacing

Ja ða ¼ x; y; zÞ; p1 and p2 in Eq. 3.4.157 with the quantum mechanical angular
momentum operators
o
Jz ¼ i
h ; etc.;
ov
o
pi ¼ i
h ði ¼ 1; 2Þ;
osi
and by adding the potential function Vðs1 ; s2 Þ to T: That is
H ¼ AJz2 þ BJx2 þ CJy2 þ F1 ðp1 p1 Þ2 þF2 ðp2 p2 Þ2

þ F12 ½ðp1 p1 Þðp2 p2 Þ þ ðp2 p2 Þðp1 p1 Þ þ Vðs1 ; s2 Þ;
ð3:4:160Þ
where A ¼ 1=ð2Iz Þ; B ¼ 1=ð2Ix Þ; C ¼ 1=ð2Iy Þ:

The kinetic energy part of the Hamiltonian operator Eq. 3.4.160 includes in
principle nine independent parameters: A; B; C; I1 ; I2 ; two of ðk1x ; k1y ; k1z Þ; and two
of ðk2x ; k2y ; k2z Þ: On the other hand, from the standpoint of a practical spectral
analysis it may be convenient to use more phenomenological Hamiltonian operator
shown below although it involves more parameters:
H ¼ A0 Jz20 þ B0 Jx20 þ C 0 Jy20 þ f1 p21 þ f2 p22 þ 2f12 p1 p2

þ q1 Jz0 p1 þ q2 Jz0 p2 þ r1 Jx0 p1 þ r2 Jx0 p2 þ s1 Jy0 p1 þ s2 Jy0 p2 þ Vðs1 ; s2 Þ;
ð3:4:161Þ
Relations between the kinetic energy parameters appearing in Eqs. 3.4.160 and
3.4.161 can be obtained by comparing coefficients of corresponding quadratic
terms of angular momenta in the two equations, after performing a rotation which
relates the x; y; z-principal axes in Eq. 3.4.160 to the x0 ; y0 ; z0 -axes in Eq. 3.4.161
according to the equation
0 1 0 1
x0 x
@ y0 A ¼ S1 ðhn ; /n ; vn Þ@ y A; ð3:4:162Þ
z0 z
The relations between three pairs of parameters in Eqs. 3.4.160 and 3.4.161 are
independent of hn ; /n ; and vn :
f1 ¼ F1 ; f2 ¼ F2 ; f12 ¼ F12 : ð3:4:163Þ
For the rest parameters we find

0 1 0 1
A0 0 0 Axx Axy Axz
B C B C
@0 B0 0 A ¼ S1 ðhn ; /n ; vn Þ@ Axy Ayy Ayz ASþ1 ðhn ; /n ; vn Þ;
0 0 C0 Axz Ayz Azz

½r1 ; s1 ; q1 ¼ 2ðF1 q1x þ F12 q2x Þ; 2 F1 q1y þ F12 q2y ; 2 F1 q1z þ F12 q2z
Sþ1 ðhn ; /n ; vn Þ;

½r2 ; s2 ; q2 ¼ 2ðF2 q2x þ F12 q1x Þ; 2 F2 q2y þ F12 q1y ; 2 F2 q2z þ F12 q1z
Sþ1 ðhn ; /n ; vn Þ;
ð3:4:164Þ
with
Aaa ¼ ð1=2Ia Þ þ F1 q21a þ F2 q22a þ 2F12 q1a q2a

ða ¼ x; y; zÞ;

ð3:4:165Þ
Aab ¼ F1 q1a q1b þ F2 q2a q2b þ F12 q1a q2b þ q2a q1b
ða; b ¼ x; y; y; z; x; zÞ:
When a spectral analysis is made with Eq. 3.4.161, the quantities on the left of Eqs.
3.4.163 and 3.4.164 will be determined in a least squares fit. From the point of
view of the PAM quantities on the right, Eqs. 3.4.163–3.4.164 represent 15 non-
linear equations in the 15 unknowns: F1 ; F2 ; F12 ; A; B; C; q1x ; q1y ; q1z ; q2x ; q2y ; q2z ;
hn ; /n ; and vn : By substituting values of A; B; C; q1x ; q1y ; q1z ; q2x ; q2y ; q2z ; in the
six equations of the first line of Eq. 3.4.159 and by using equality,
k21x þ k21y þ k21z ¼ k22x þ k22y þ k22z ¼ 1; ð3:4:166Þ
we can obtain values of eight quantities k1x ; k1y ; k1z ; k2x ; k2y ; k2z ; Is1 ; and Is2 : Since
we can calculate values of F1 ; F2 and F12 by using these eight quantities and Eqs.
3.4.158 and 3.4.159, a consistency check is possible, and we can estimate to what
extent the PAM model is reproduced by the phenomenological Hamiltonian.
The jet-cooled Fourier-transform microwave spectrum of N-methyl-acetamide
(CH3 NHCð¼ OÞCH3 ), which contains two inequivalent methyl tops with
relatively low barriers to internal rotation, was analyzed successfully with the use
of the phenomenological Hamiltonian operator of type of Eq. 3.4.161, adding
higher order terms of rotation, torsion, and torsion-rotation terms [49]. Since the
N-methyl-acetamide has a plane of symmetry at an equilibrium, the term s1 Jy p1 þ
s2 Jy p2 is eliminated in Eq. 3.4.161, and we can set vn ¼ /n ¼ 0 in Eqs. 3.4.162
and 3.4.164.
In Fig. 3.33, examples of observed transitions of N-methyl-acetamide
(CH3 NHCð¼OÞCH3 ), [49], are shown in a schematic torsional–rotational
energy level diagram in its ground vibrational state.
Fig. 3.33 The torsional–rotational energy level diagram of N-methyl-acetamide in the ground
vibrational state is shown with observed transitions [49]. In order to designate rotational levels,
quantum numbers J; Ka ; Kc for an asymmetric top molecule are used for levels of A- and
E3 -symmetry species, and J and K are used for E1 -, E2 -, and E4 -symmetry levels
3.4.2.4 Basis Set Wavefunctions and Hamiltonian Matrix Elements
In order to construct the Hamiltonian matrix, we use wavefunctions of a form
jm1 ð1Þijm2 ð2ÞijJ; Ki; ð3:4:167Þ
i.e., a product of a two-top free rotor wavefunction jm1 ð1Þijm2 ð2Þi and a sym-
metric top rotational wavefunction jJ; Ki: Here, free rotor wavefunctions jm1 ð1Þi
and jm2 ð2Þi are given by
1
jm1 ð1Þi ¼ pffiffiffiffiffiffi expðim1 s1 Þ; ð3:4:168Þ
2p
1
jm2 ð2Þi ¼ pffiffiffiffiffiffi expðim2 s2 Þ; ð3:4:169Þ
2p
Basis Set Functions for the C1 ðG9 Þ Case
By noting that
ð123Þjm1 ð1Þi ¼ xm1 jm1 ð1Þi; ð123Þjm2 ð2Þi ¼ jm2 ð2Þi; ð123ÞjJ; Ki ¼ jJ; Ki;
ð456Þjm1 ð1Þi ¼ jm1 ð1Þi; ð456Þjm2 ð2Þi ¼ xm2 jm2 ð2Þi; ð456ÞjJ; Ki ¼ jJ; Ki;
ð3:4:170Þ
where x ¼ expði2p=3Þ; we have the following basis set functions belonging to

each of symmetry species of G9 :
jm1 ; m2 ; J; Ki for m1 ¼ 3n; m2 ¼ 3n0 : A symmetry species,
jm1 ; m2 ; J; Ki for m1 ¼ 3n þ 1; m2 ¼ 3n0 : E1þ symmetry species;
jm1 ; m2 ; J; Ki for m1 ¼ 3n 1; m2 ¼ 3n0 : E1 symmetry species;
jm1 ; m2 ; J; Ki for m1 ¼ 3n; m2 ¼ 3n0 þ 1 : E2þ symmetry species,
jm1 ; m2 ; J; Ki for m1 ¼ 3n; m2 ¼ 3n0 1 : E2 symmetry species,
jm1 ; m2 ; J; Ki for m1 ¼ 3n þ 1; m2 ¼ 3n0 þ 1 : E3þ symmetry species,
jm1 ; m2 ; J; Ki for m1 ¼ 3n 1; m2 ¼ 3n0 1 : E3 symmetry species,
jm1 ; m2 ; J; Ki for m1 ¼ 3n þ 1; m2 ¼ 3n0 1 : E4þ symmetry species,
jm1 ; m2 ; J; Ki for m1 ¼ 3n 1; m2 ¼ 3n0 þ 1 : E4 symmetry species,
ð3:4:171Þ
where jm1 ; m2 ; J; Ki jm1 ð1Þijm2 ð2ÞijJ; Ki; and n ¼ 0; 1; 2; . . .; n0 ¼ 0; 1;
2; . . .:
Basis Set Functions for Case Cs ðG18 Þ
For case Cs ðG18 Þ; A1 and A2 species basis functions are given by

1
jA1 ; m1 ; m2 ; J; Ki ¼ pffiffiffi ½jm1 ð1Þijm2 ð2ÞijJ; Ki
2 ð3:4:172Þ

þð1ÞJK j m1 ð1Þij m2 ð2ÞijJ; Ki ;
1
jA2 ; m1 ; m2 ; J; Ki ¼ pffiffiffi ½jm1 ð1Þijm2 ð2ÞijJ; Ki
2 ð3:4:173Þ

ð1ÞJK j m1 ð1Þij m2 ð2ÞijJ; Ki ;
where m1 ¼ 3n; m2 ¼ 3n0 ðn ¼ 0; 1; 2; . . .; n0 ¼ 0; 1; 2; . . .Þ: It should be

pffiffiffi
noted that if m1 ¼ m2 ¼ K ¼ 0 in Eqs. 3.4.172 and 3.4.173, 1= 2 is replaced by
1=2:
Degenerate Ei ði ¼ 1; 2; 3; 4Þ basis functions for case Cs ðG18 Þ are equal to
ðEiþ ; Ei Þ in Eq. 3.4.171 for case C1 ðG9 Þ:
Hamiltonian Matrix Elements
The Hamiltonian matrix elements hm1 ; m2 ; J; KjHjm01 ; m02 ; J; K 0 i can be written

down rather easily after deciding what we adopt as the potential function Vðs1 ; s2 Þ
in Eq. 3.4.161. When adopting a simple potential function
V31 V32
Vðs1 ; s2 Þ ¼ ð1 cos 3s1 Þ þ ð1 cos 3s2 Þ þ V12 cos 3s1 cos 3s2 ;
2 2
ð3:4:174Þ
three non-zero off-diagonal matrix elements hm1 ; m2 ; J; KjHjm01 ; m02 ; J; K 0 i; related
to the potential function, are given by
V31
hm1 ; m2 ; J; KjHjm1 3; m2 ; J; Ki ¼ ;
4
V32
hm1 ; m2 ; J; KjHjm1 ; m2 3; J; Ki ¼ ;
4 ð3:4:175Þ
V12
hm1 ; m2 ; J; KjHjm1 3; m2 3; J; Ki ¼ ;
4
V12
hm1 ; m2 ; J; KjHjm1 3; m2 3; J; Ki ¼ ;
4
The Hamiltonian matrix for each symmetry species is constructed by combining
basis set functions given by Eqs. 3.4.171–3.4.173 with the matrix elements
hm1 ; m2 ; J; KjHjm01 ; m02 ; J; K 0 i:

3.4.3 Case of a Molecule with Two Equivalent Methyl

Internal Rotors
3.4.3.1 Group Theory
We treat two cases: a case that a molecule is of Cs symmetry at an equilibrium

configuration, and another case that a molecule is of C2v symmetry at an
equilibrium configuration. We call the former and the latter case Cs ðG18 Þ and case
C2v ðG36 Þ; respectively, since the PI group for the former case is G18 and that for
the latter case is G36 : In Figs. 3.34 and 3.35, dimethylamine and acetone are shown
as examples of case Cs ðG18 Þ and case C2v ðG36 Þ; respectively.
Character Table
The character tables for Cs ðG18 Þ and for C2v ðG36 Þ are given in Tables 3.14 and
3.15, respectively. In Table 3.14, E1sep ; E2sep and Gsep represent combined pairs
ðA1þ ; A1 Þ; ðA2þ ; A2 Þ and ðEþ ; E Þ of separably degenerate irreducible repre-
sentations, respectively.
Transition Selection Rules
Since the laboratory-fixed component lZ of the electric dipole moment operator

belongs to A2 symmetry species in the G18 and A3 in the G36 PI group, selection
Table 3.14 Character table for a molecule of Cs ðG18 Þ symmetry with two equivalent methyl
internal rotors
E (123)(456) Pa (132)(456) (123) Qa (123)(465) (456) Ra
(132)(465) (465) (132)
A1 1 1 1 1 1 1 1 1 1
A2 1 1 1 1 1 1 1 1 1
E 2 1 0 2 1 0 2 1 0
E1sep
A1þ 1 1 1 x x x x
x
x
A1 1 1 1 x
x
x
x x x
E2sep
A2þ 1 1 1 x x x x
x
x
A2 1 1 1 x
x
x
x x x
Gsep
Eþ 2 1 0 2x x 0 2x
x
0
E 2 1 0 2x
x
0 2x x 0
x ¼ expði2p=3Þ:
a
P ¼ fð14Þð26Þð35Þð78Þ
; ð15Þð24Þð36Þð78Þ
g;
Q ¼ fð162534Þð78Þ
; ð142635Þð78Þ
; ð152436Þð78Þ
g;
R ¼ fð153624Þð78Þ
; ð163425Þð78Þ
; ð143526Þð78Þ
g
Table 3.15 Character table for a molecule of C2v ðG36 Þ symmetry with two equivalent methyl
internal rotors
E (123)(456) Pa (123)(465) (123) (142635)(78)
Qa (142536)(78) (23)(56)
(132)(465) (132)(456) (132) (143526)(78)

(143625)(78) (23)(45)
(456) (152436)(78)
(152634)(78) (23)(46)
(465) (153624)(78)
(153426)(78) (13)(56)
(162534)(78)
(162435)(78) (13)(45)
(163425)(78)
(163524)(78) (13)(46)
(12)(56)
(12)(45)
(12)(46)
A1 1 1 1 1 1 1 1 1 1
A2 1 1 1 1 1 1 1 1 1
A3 1 1 1 1 1 1 1 1 1
A4 1 1 1 1 1 1 1 1 1
E1 2 2 2 1 1 1 0 0 0
E2 2 2 2 1 1 1 0 0 0
E3 2 1 0 2 1 0 2 1 0
E4 2 1 0 2 1 0 2 1 0
G 4 2 0 2 1 0 0 0 0
a
P ¼ fð14Þð26Þð35Þð78Þ
g;
Q ¼ fð14Þð25Þð36Þð78Þ; ð15Þð26Þð34Þð78Þ; ð16Þð24Þð35Þð78Þg
rules on the overall torsional–rotational symmetry species for allowed electric

dipole transitions are given by
A1 $ A2 ; E1sep $ E2sep ; E $ E; Gsep $ Gsep ; ð3:4:176Þ
for case Cs ðG18 Þ; and

A1 $ A3 ; A2 $ A 4 ; E1 $ E 2 ; E3 $ E3 ; E4 $ E4 ; ð3:4:177Þ
for case C2v ðG36 Þ:
Coordinate System and Transformation Properties of Variables
The description depends on the orientations of the methyl groups to the PAM axes.
We will take the dimethylamine molecule shown in Fig. 3.34 as an example of
case Cs ðG18 Þ and the acetone molecule shown in Fig. 3.35 as an example of case
C2v ðG36 Þ: The PAM axes ða; b; c ! z; x; yÞ are shown in Figs. 3.34 and 3.35 (the
y-axis is perpendicular to the zx-plane). The transformation properties are obtained
from Eqs. 3.4.152–3.4.154 representing the relation between the laboratory-fixed
coordinates and the internal variables, together with the following structural
constraint:
hP1 þ hP2 ¼ p; /P1 ¼ /P2 ; vP1 ¼ vP2 ; ð3:4:178Þ
Fig. 3.34 Dimethylamine, an

example of a Cs molecule
with equivalent two methyl
groups
Fig. 3.35 Acetone, an

example of a C2v molecule
with equivalent two methyl
groups
Table 3.16 Transformation properties of variables in the right of Eqs. 3.4.152–3.4.154 under
generating operations of the group G18 for dimethylamine and G36 for acetone
R h; /; v (rotation) s1 (internal s2 (internal
(translation) rotation) rotation)
ð123Þ R h; /; v s1 þ 2p=3 s2
ð456Þ R h; /; v s1 s2 þ 2p=3
ð14Þð26Þð35Þð78Þ
R h; /; v þ p s2 s1
ð23Þð56Þ
R p h; p þ /; p v s1 s2
Generating operations of G18 are ð123Þ; ð456Þ and ð14Þð26Þð35Þð78Þ
; generating operations of
G36 are ð123Þ; ð456Þ; ð14Þð26Þð35Þð78Þ
and ð23Þð56Þ
for dimethylamine (case Cs ðG18 Þ), and

hP1 þ hP2 ¼ p; /P1 ¼ /P2 ¼ 0; vP1 ¼ vP2 ¼ 0; ð3:4:179Þ
for acetone (case C2v ðG36 Þ).

Transformation properties under generating operations of G18 and G36 are given
together in Table 3.16.
3.4.3.2 J ¼ K ¼ 0 Energy Level Splitting Pattern
In Fig. 3.36, an energy level diagram for J ¼ K ¼ 0 is illustrated qualitatively for

a molecule with equivalent two methyl groups. The schematic splitting pattern
Fig. 3.36 An internal-

rotation energy level diagram
is illustrated schematically
for J ¼ K ¼ 0 in the ground
vibrational state of a
molecule with equivalent two
methyl internal-rotors
depicted in Fig. 3.36 will be understood by making A=E splittings equal in top 1
and top 2 in Fig. 3.32. Levels of symmetry species A2 ; E2sep for case Cs ðG18 Þ and
A2 ; A3 ; A4 ; E2 ; E4 for case C2v ðG36 Þ; which do not appear for the J ¼ K ¼ 0 in the
ground vibrational state, appear for K 6¼ 0 levels and/or v 6¼ 0 vibrationally
excited states.
3.4.3.3 Hamiltonian Operator, Basis Set Functions

and Hamiltonian Matrix
Hamiltonian Operator
The torsional–rotational Hamiltonian operator for a molecule with two equivalent

methyl internal-rotors is obtained by imposing the following relations in Eqs.
3.4.158–3.4.160:
F1 ¼ F2 ð FÞ; k1z ¼ k2z ; k1x ¼ k2x ; k1y ¼ k2y ð3:4:180Þ
for dimethylamine of case Cs ðG18 Þ; and
F1 ¼ F2 ð FÞ; k1z ¼ k2z ; k1x ¼ k2x ; k1y ¼ k2y ¼ 0 ð3:4:181Þ
for acetone of case C2v ðG36 Þ: The resultant Hamiltonian operator is given by

H ¼ A0 Jz2 þ B0 Jx2 þ C0 Jy2 þ Dbc Jx Jy þ Jy Jx
þ qðp1 p2 ÞJz þ rðp1 þ p2 ÞJx þ sðp1 þ p2 ÞJy

þ F p21 þ p22 þ 2F12 p1 p2 þ Vðs1 ; s2 Þ; ð3:4:182Þ
where
A0 ¼ A þ 2ðF F12 Þq21z ; B0 ¼ B þ 2ðF þ F12 Þq21x ; C0 ¼ C þ 2ðF þ F12 Þq21y ;

Dbc ¼ 2ðF þ F12 Þq1x q1y ; q ¼ 2ðF F12 Þq1z ; r ¼ 2ðF þ F12 Þq1x ;
s ¼ 2ðF þ F12 Þq1y ;
ð3:4:183Þ
for dimethylamine of case Cs ðG18 Þ; and
A0 ¼ A þ 2ðF F12 Þq21z ; B0 ¼ B þ 2ðF þ F12 Þq21x ; C 0 ¼ C; Dbc ¼ 0;

q ¼ 2ðF F12 Þq1z ; r ¼ 2ðF þ F12 Þq1x ; s ¼ 0;
ð3:4:184Þ
for acetone of case C2v ðG36 Þ:
The simplest potential function will be
1 1
Vðs1 ; s2 Þ ¼ V3 ð1 cos 3s1 Þ þ V3 ð1 cos 3s2 Þ þ V12 cos 3s1 cos 3s2 :
2 2
ð3:4:185Þ
Basis Set Functions
Basis set functions of each symmetry species for case Cs ðG18 Þ are give as follows:
A-symmetry:
pffiffiffi
jA1 i ¼ ð1= 2Þ jm1 ; m2 ijJ; Ki þ ð1ÞK j m2 ; m1 ijJ; Ki ;
pffiffiffi
jA2 i ¼ ð1= 2Þ jm1 ; m2 ijJ; Ki ð1ÞK j m2 ; m1 ijJ; Ki ; ð3:4:186Þ
ðm1 ; m2 ¼ 0 mod 3 0 and J K JÞ
where jm1 ; m2 i ¼ jm1 ð1Þijm2 ð2Þi ¼ ð1=2pÞ expðim1 s1 Þ expðim2 s2 Þ: When m1 ¼

pffiffiffi
m2 ¼ 0 the factor 1= 2 in the equations above should be replaced by 1/2.
E-symmetry:
jEa i ¼ jm1 ð1Þijm2 ð2ÞijJ; Ki;

ð3:4:187Þ
jEb i ¼ ð14Þð26Þð35Þð78Þ
jEa i ¼ ð1ÞK j m2 ð1Þij m1 ð2ÞijJ; Ki;
with m1 ¼ 3n þ 1ðn ¼ 0; 1; 2; . . .Þ; m2 ¼ 3n0 þ 1ðn0 ¼ 0; 1; 2; . . .Þ, and
J K J:
E1sep -symmetry:
pffiffiffi
jA1þ i ¼ ð1= 2Þ jm1 ; m2 ijJ; Ki þ ð1ÞK j m2 ; m1 ijJ; Ki ;
ðm1 ¼ 3n þ 1ðn ¼ 0; 1; 2; . . .Þ; m2 ¼ 3n0 1ðn0 ¼ 0; 1; 2; . . .Þ;
and J K JÞ
pffiffiffi
jA1 i ¼ ð1= 2Þ jm1 ; m2 ijJ; Ki þ ð1ÞK j m2 ; m1 ijJ; Ki ;
ðm1 ¼ 3n 1ðn ¼ 0; 1; 2; . . .Þ; m2 ¼ 3n0 þ 1ðn0 ¼ 0; 1; 2; . . .Þ;
and J K JÞ:
ð3:4:188Þ
pffiffiffi
When m1 ¼ m2 ; the factor 1= 2 in the equations above should be replaced by
1=2:
E2sep -symmetry:
pffiffiffi
jA2þ i ¼ ð1= 2Þ jm1 ; m2 ijJ; Ki ð1ÞK j m2 ; m1 ijJ; Ki ;
ðm1 ¼ 3n þ 1ðn ¼ 0; 1; 2; . . .Þ; m2 ¼ 3n0 1ðn0 ¼ 0; 1; 2; . . .Þ;
and J K JÞ
pffiffiffi
jA2 i ¼ ð1= 2Þ jm1 ; m2 ijJ; Ki ð1ÞK j m2 ; m1 ijJ; Ki ;
ðm1 ¼ 3n 1ðn ¼ 0; 1; 2; . . .Þ; m2 ¼ 3n0 þ 1ðn0 ¼ 0; 1; 2; . . .Þ;
and J K JÞ:
ð3:4:189Þ
pffiffiffi
When m1 ¼ m2 ; the factor 1= 2 in the equations above should be replaced by
1=2:
Gsep -symmetry:
jEþa i ¼ jm1 ; m2 ijJ; Ki;
ðm1 ¼ 3nðn ¼ 0; 1; 2; . . .Þ; m2 ¼ 3n0 þ 1ðn0 ¼ 0; 1; 2; . . .Þ;
and J K JÞ;
jEþb i ¼ ð14Þð26Þð35Þð78Þ
jEþa i ¼ ð1ÞK j m2 ; m1 ijJ; Ki;
jEa i ¼ jm1 ; m2 ijJ; Ki;
ðm1 ¼ 3n þ 1ðn ¼ 0; 1; 2; . . .Þ; m2 ¼ 3n0 ðn0 ¼ 0; 1; 2; . . .Þ;
and J K JÞ;
jEb i ¼ ð14Þð26Þð35Þð78Þ
jEa i ¼ ð1ÞK j m2 ; m1 ijJ; Ki ð3:4:190Þ
Basis functions of E1 ; E2 and G for case C2v ðG36 Þ are equal to those for
E1sep ; E2sep and Gsep in Eqs. 3.4.186–3.4.190 for case Cs ðG18 Þ; respectively. Other
basis functions for case C2v ðG36 Þ are as follows:

jA1 i ¼ ð1=2Þ jm1 ; m2 ijJ; Ki þ ð1ÞK j m2 ; m1 ijJ; Ki

þ ð1ÞJK j m1 ; m2 ijJ; Ki þ ð1ÞJ jm2 ; m1 ijJ; Ki

jA2 i ¼ ð1=2Þ jm1 ; m2 ijJ; Ki þ ð1ÞK j m2 ; m1 ijJ; Ki

ð1ÞJK j m1 ; m2 ijJ; Ki ð1ÞJ jm2 ; m1 ijJ; Ki
ð3:4:191Þ
jA3 i ¼ ð1=2Þ jm1 ; m2 ijJ; Ki ð1ÞK j m2 ; m1 ijJ; Ki

ð1ÞJK j m1 ; m2 ijJ; Ki þ ð1ÞJ jm2 ; m1 ijJ; Ki

jA4 i ¼ ð1=2Þ jm1 ; m2 ijJ; Ki ð1ÞK j m2 ; m1 ijJ; Ki

: þ ð1ÞJK j m1 ; m2 ijJ; Ki ð1ÞJ jm2 ; m1 ijJ; Ki ;
where m1 ¼ 3n ðn ¼ 0; 1; 2; . . .Þ; m2 ¼ 3n0 ðn0 ¼ 0; 1; 2; . . .Þ, and 0 K J:

When (m1 ¼ m2 ) or (m1 ¼ m2 and K ¼ 0), the factor 1=2 should be replaced by
pffiffiffi
1=ð2 2Þ in the above equations, and when m1 ¼ m2 ¼ K ¼ 0 the factor 1=2
should be replaced by 1=4:
pffiffiffi
jE3a i ¼ ð1= 2Þ jm1 ; m2 ijJ; Ki þ ð1ÞJ jm2 ; m1 ijJ; Ki
jE3b i ¼ ð14Þð26Þð35Þð78Þ
jE3a i
pffiffiffi
¼ ð1= 2Þð1ÞK j m2 ; m1 ijJ; Ki þ ð1ÞJ j m1 ; m2 ijJ; Ki
pffiffiffi
jE4a i ¼ ð1= 2Þ jm1 ; m2 ijJ; Ki ð1ÞJ jm2 ; m1 ijJ; Ki
jE4b i ¼ ð14Þð26Þð35Þð78Þ
jE4a i
pffiffiffi
¼ ð1= 2Þð1ÞK j m2 ; m1 ijJ; Ki ð1ÞJ j m1 ; m2 ijJ; Ki ;
ð3:4:192Þ
where m1 ¼ 3n þ 1 ðn ¼ 0; 1; 2; . . .Þ; m2 ¼ 3n0 þ 1 ðn0 ¼ 0; 1; 2; . . .Þ, and
pffiffiffi
0 K J: When m1 ¼ m2 and K ¼ 0; the factor 1= 2 should be replaced by 1=2
in the above equations.
Hamiltonian Matrix Elements
It is easy to obtain expressions of Hamiltonian matrix elements hm1 ; m2 ; J;

KjHjm01 ; m02 ; J; K 0 i; corresponding to Eqs. 3.4.78–3.4.81 for one-top problem. The
Hamiltonian matrices for each symmetry species of G18 and G36 are constructed by
combining basis set functions given by Eqs. 3.4.186–3.4.192 with the matrix
elements hm1 ; m2 ; J; KjHjm01 ; m02 ; J; K 0 i:
3.4.4 Tunneling Matrix Formulation for Molecules

with Two Methyl Internal Rotors
We describe the tunneling matrix formulation in the case of two methyl internal
rotors. As mentioned in the one-top problem, the tunneling matrix formulation is
effective in the case of a high potential barrier to internal rotation. The description
is somewhat different depending on geometrical symmetry of a molecule. Case
C1 ðG9 Þ of a molecule with inequivalent two methyl tops and case Cs ðG18 Þ of a
molecule with equivalent two methyl tops will be treated below.
3.4.4.1 Case C1 ðG9 Þ of a Molecule with Inequivalent Two Methyl Tops
Considering 2-butanol as an example, we will describe the formulation.
Vibrational Framework Functions and Basis Functions Belonging to Each

Symmetry Species
When j1i represents the vibrational framework function at framework 1, nine

vibrational framework functions jniðn ¼ 1; 2; 3; . . .; 9Þ are defined as
jni ¼ On j1i; ð3:4:193Þ

where operations On ðn ¼ 1; 2; 3; . . .; 9Þ are as follows:
O1 ¼ E;
O2 ¼ ð123Þ; O3 ¼ ð132Þ; O4 ¼ ð456Þ; O5 ¼ ð465Þ;
O6 ¼ ð123Þð456Þ; O7 ¼ ð123Þð465Þ; O8 ¼ ð132Þð456Þ; O9 ¼ ð132Þð465Þ:
ð3:4:194Þ
We note again that the concept of framework function is effective in a high barrier
case. Basis functions belonging to each of symmetry species of G9 ; constructed
from framework functions, are given as follows using the theorem well known in
the group theory
1 X 9
jCðtorsionÞi ¼ vC ðOn Þ jni; ð3:4:195Þ

NðCÞ n¼1
where vC ðOn Þðn ¼ 1; 2; . . .; 9Þ are characters given in Table 3.11, and

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
NðAÞ ¼ 9ð1 þ 2S2 þ 2S4 þ 2S6 þ 2S7 Þ;
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
NðE1þ Þ ¼ NðE1 Þ ¼ 9ð1 S2 þ 2S4 S6 S7 Þ;
NðE2þ Þ ¼ NðE2 Þ ¼ 9ð1 þ 2S2 S4 S6 S7 Þ; ð3:4:196Þ
NðE3þ Þ ¼ NðE3 Þ ¼ 9ð1 S2 S4 S6 þ 2S7 Þ;
NðE4þ Þ ¼ NðE4 Þ ¼ 9ð1 S2 S4 þ 2S6 S7 Þ;
where Sn ¼ h1jniðn ¼ 1; 2; . . .; 9Þ; note S1 ¼ 1:
Phenomenological Hamiltonian Operator and Hamiltonian Matrix Element

Expressions
Since rotational functions jJ; Ki (J K J) are invariant under symmetry

operations of the PI group G9 ; torsional–rotational basis functions belonging to
each symmetry species are given by
jC; J; Ki ¼ jCðtorsionÞijJ; Ki: ð3:4:197Þ
We adopt the following phenomenological Hamiltonian operator
H ¼ hv þ AJz2 þ BJx2 þ CJy2 þ iqJz þ irJx þ isJy : ð3:4:198Þ
The last three terms in Eq. 3.4.198 represent the Coriolis interaction between the
overall rotation and the methyl internal rotation. All the coefficients,
hv ; A; B; C; q; r; and s; are of A-symmetry. The following transformation properties
of the coefficients under the combined operation of Hermitian conjugation and
time reversal (indicated by z) are also useful:
Table 3.17 Numerical factors fn and gn in Eq. 3.4.200

n¼1 n¼2 n¼4 n¼6 n¼7
fn gn fn gn fn gn fn gn fn gn
A 1 0 2 0 2 0 2 0 2 0
E1þ 1 0 1 1 2 0 1 1 1 1
E2þ 1 0 2 0 1 1 1 1 1 1
E3þ 1 0 1 1 1 1 1 1 2 0
E4þ 1 0 1 1 1 1 2 0 1 1
hz ¼ þh for hv ; A; B and C;
z
ð3:4:199Þ
h ¼ h for q; r and s:
Based on the description given above, parameterized expressions for the torsional–
rotational Hamiltonian matrix elements are given as
(
9 X
0
hC; J; KjHjC; J; K i ¼ 2
f ðCÞn h1jhv jnidKK 0 þ h1jAjniK 2 dKK 0
NðCÞ n¼1;2;4;6;7
i
þ h1jBjnihJ; KjJx2 jJ; K 0 i þ h1jCjnihJ; KjJy2 jJ; K 0 i
X pffiffiffi
þ 3gðCÞn ½h1jqjniKdKK 0 þ h1jrjnihJ; KjJx jJ; K 0 i
n¼1;2;4;6;7

þ h1jsjnihJ; KjJy jJ; K 0 i ð3:4:200Þ
where f ðCÞn and gðCÞn are numerical factors given in Table 3.17. In Eq. 3.4.200,
the symmetry species C takes A and Eiþ ði ¼ 1; 2; 3; 4Þ. The Hamiltonian matrix for
Ei ði ¼ 1; 2; 3; 4Þ produces the same eigenvalues as that for Eiþ does since we
have a relation

hEi ; J; KjHjEi ; J; K 0 i ¼ hEi ; J; K 0 jH y jEi ; J; Ki

ð3:4:201Þ
¼ hEiþ ; J; K 0 jHjEiþ ; J; Ki:
When obtaining numerical factors fn and gn in Table 3.17 the following
equalities are used:
h1jhj2i ¼ h1jhj3i; h1jhj4i ¼ h1jhj5i;

ð3:4:202Þ
for h ¼ hv ; A; B; and C; and
h1jhj1i ¼ 0;
h1jhj2i ¼ h1jhj3i; h1jhj4i ¼ h1jhj5i; ð3:4:203Þ
for h ¼ q; r; and s:
The description given above is also available for case Cs ðG18 Þ of inequivalent-
two-top molecules except for that the term is Jy in Eq. 3.4.198 does not contribute
to any Hamiltonian matrix element and that symmetry species A for case C1 ðG9 Þ
corresponds to A1 and A2 for case Cs ðG18 Þ; which distinguish the two K-type
doubling levels (the J ¼ K ¼ 0 level belongs to either A1 or A2 symmetry species
depending on the symmetry property of the torsional state under the symmetry
operation ð23Þð56Þð78Þ
; the reflection in the plane of symmetry in point group).
3.4.4.2 Case Cs ðG18 Þ of a Molecule with Equivalent Two Methyl Tops
We treat the tunneling matrix formulation for a dimethylamine-like molecule. We

obtain the formulation for this case by making suitable changes in some of Eqs.
3.4.193–3.4.203 and Table 3.17. Using the character table of G18 for a dimeth-
ylamine-like molecule shown in Table 3.14, it is shown that nine vibrational
framework functions, given by Eq. 3.4.193 with Eq. 3.4.194, produce symmetrized
torsional functions with the following symmetry species:
A1 þ E þ E1sep þ Gsep for ð14Þð26Þð35Þð78Þ
j1i ¼ þj1i;
ð3:4:204Þ
or A2 þ E þ E2sep þ Gsep for ð14Þð26Þð35Þð78Þ
j1i ¼ j1i:
For simplicity, we will discuss the former case of Eq. 3.4.204 in detail. The
symmetrized torsional basis functions belonging to each of symmetry species are
given by
A1 : jA1 ðtorsionÞi ¼ NðA1 Þ½1; 1; 1; 1; 1; 1; 1; 1; 1;

E : jEa ðtorsionÞi ¼ NðEÞ½1; x
; x; x
; x; x; 1; 1; x
;
jEb ðtorsionÞi ¼ NðEÞ½1; x; x
; x; x
; x
; 1; 1; x;
E1sep : jE1sep;a ðtorsionÞi jA1þ ðtorsionÞi ¼ NðE1sep Þ½1; x
; x; x; x
; 1; x; x
; 1;
: jE1sep;b ðtorsionÞi jA1 ðtorsionÞi ¼ NðE1sep Þ½1; x; x
; x
; x; 1; x
; x; 1;
Gsep : jGsep;a ðtorsionÞi jEþa ðtorsionÞi ¼ NðGsep Þ½1; x; x
; 1; 1; x; x; x
; x
;
: jGsep;b ðtorsionÞi jEþb ðtorsionÞi ¼ NðGsep Þ½1; 1; 1; x
; x; x
; x; x
; x;
: jGsep;c ðtorsionÞi jEa ðtorsionÞi ¼ NðGsep Þ½1; x
; x; 1; 1; x
; x
; x; x;
: jGsep;d ðtorsionÞi jEb ðtorsionÞi ¼ NðGsep Þ½1; 1; 1; x; x
; x; x
; x; x
;
ð3:4:205Þ
P9
where ½a1 ; a2 ; . . .; a9 ¼ n¼1 an jni; and
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
NðA1 Þ ¼ 1= p 9ð1 þ 4S2 þ 2S6 þ 2S7 Þ;
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi NðEÞ ¼ 1= 9ð1 2S2 S6 þ 2S7 Þ;
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
NðE1sep Þ ¼ 1= 9ð1 2S2 þ 2S6 S7 Þ; NðGsep Þ ¼ 1= 9ð1 þ S2 S6 S7 Þ;
ð3:4:206Þ
Since transformation properties of rotational functions under generating operations
of G18 are given as
ð123ÞjJ; Ki ¼ jJ; Ki;

ð456ÞjJ; Ki ¼ jJ; Ki; ð3:4:207Þ

K
ð14Þð26Þð35Þð78Þ jJ; Ki ¼ ð1Þ jJ; Ki;
torsional–rotational basis functions belonging to each symmetry species are given

as
jA1 (overall); J; Ki ¼ jA1 (torsion)ijJ; K ¼ eveni;

jA2 (overall); J; Ki ¼ jA1 (torsion)ijJ; K ¼ oddi;
jEa (overall); J; Ki ¼ jEa (torsion)ijJ; Ki; ða ¼ a; bÞ
ð3:4:208Þ
jE1sep a (overall); J; Ki ¼ jE1sep a (torsion)ijJ; K ¼ eveni; ða ¼ a; bÞ
jE2sep a (overall); J; Ki ¼ jE1sep a (torsion)ijJ; K ¼ oddi; ða ¼ a; bÞ
jGsep a (overall); J; Ki ¼ jGsep a (torsion)ijJ; Ki ða ¼ a; b; c; dÞ:
In the phenomenological Hamiltonian operator given also by Eq. 3.4.198,

coefficients hv ; A; B; C and q are of A1 symmetry, and r and s are of A2 symmetry
since Jz ; Jx ; and Jy are of A1 ; A2 and A2 symmetry, respectively (see Table 3.16).
Using these symmetry properties of the coefficients and their transformation
properties under the combined operation (z) of Hermitian conjugation and time
reversal shown in Eq. 3.4.199, we have the following relations:
h1jhj2i ¼ h1jhj3i ¼ h1jhj4i ¼ h1jhj5i;

h1jhj6i ¼ h1jhj9i; ð3:4:209Þ
h1jhj7i ¼ h1jhj8i;
for h ¼ hv ; A; B; and C;
h1jqj1i ¼ 0;
h1jqj2i ¼ h1jqj3i ¼ h1jqj4i ¼ h1jqj5i;
ð3:4:210Þ
h1jqj6i ¼ h1jqj9i ¼ 0;
h1jqj7i ¼ h1jqj8i;
and
h1jhj1i ¼ 0;
h1jhj2i ¼ h1jhj3i ¼ h1jhj4i ¼ h1jhj5i;
ð3:4:211Þ
h1jhj6i ¼ h1jhj9i;
h1jhj7i ¼ h1jhj8i ¼ 0;
for h ¼ r and s: From Eqs. 3.4.209–3.4.211, we obtain the following parameter-

ized expressions of Hamiltonian matrix elements for each of symmetry species:
hAðoverallÞ; J; KjHjAðoverallÞ; J; K 0 i
X X
¼ 9NðA1 Þ2 aðAÞn h1jhk jnihJ; KjRk jJ; K 0 i;
n¼1;2;6;7 k¼1;2;3;4
hEa ðoverallÞ; J; KjHjEa ðoverallÞ; J; K 0 i
"
2
X X
¼ 9NðEÞ aðEÞn h1jhk jnihJ; KjRk jJ; K 0 i
n¼1;2;6;7 k¼1;2;3;4
pffiffiffi
þ 3ð2h1jrj2i h1jrj6iÞhJ; KjJx jJ; K 0 i
pffiffiffi i
þ 3ð2h1jsj2i h1jsj6iÞhJ; KjJy jJ; K 0 i ;
hEsep a ðoverallÞ; J; KjHjEsep a ðoverallÞ; J; K 0 i
"
2
X X ð3:4:212Þ
¼ 9NðEsep Þ aðEsep Þn h1jhk jnihJ; KjRk jJ; K 0 i
n¼1;2;6;7 k¼1;2;3;4
pffiffiffi i
þ 3ð2h1jqj2i h1jqj7iÞKdKK 0
hGsep a ðoverallÞ; J; KjHjGsep a ðoverallÞ; J; K 0 i
"
2
X X
¼ 9NðGsep Þ aðGsep Þn h1jhk jnihJ; KjRk jJ; K 0 i
n¼1;2;6;7 k¼1;2;3;4
pffiffiffi
3ðh1jqj2i þ h1jqj7iÞKdKK 0
pffiffiffi
3ðh1jrj2i þ h1jrj6iÞhJ; KjJx jJ; K 0 i
pffiffiffi i
3ðh1jsj2i þ h1jsj6iÞhJ; KjJy jJ; K 0 i
where
X
hk Rk ¼ hv þ AJz2 þ BJx2 þ CJy2 ;
k¼1;2;3;4

aðAÞ1 ; aðAÞ2 ; aðAÞ6 ; aðAÞ7 ¼ ð1; 4; 2; 2Þ;

aðEÞ1 ; aðEÞ2 ; aðEÞ6 ; aðEÞ7 ¼ ð1; 2; 1; 2Þ;

aðEsep Þ1 ; aðEsep Þ2 ; aðEsep Þ6 ; aðEsep Þ7 ¼ ð1; 2; 2; 1Þ;

and aðGsep Þ1 ; aðGsep Þ2 ; aðGsep Þ6 ; aðGsep Þ7 ¼ ð1; 1; 1; 1Þ:
In Eq. 3.4.212, an unsubscripted A is used to indicate either of A1 or A2 since A1

and A2 discriminate merely between K ¼ even and K ¼ odd levels, and Esep is
similarly used for either of E1sep ðK ¼ evenÞ or E2sep ðK ¼ oddÞ:
We will mention briefly changes occurring in the treatment of the tunneling
matrix formulation for acetone-like molecule of case C2v ðG36 Þ: When we assume
ð14Þð26Þð35Þð78Þ
j1i ¼ þj1i and ð23Þð56Þ
j1i ¼ þj1i; ð3:4:213Þ
the symmetrized torsional functions, jA1 ðtorsionÞi; jEðtorsionÞi; jE1sep ðtorsionÞi
and jGðtorsionÞi; shown for the case of a dimethylamine-like molecule in Eq.
3.4.205 corresponds to jA1 ðtorsionÞi; jE3 ðtorsionÞi; jE1 ðtorsionÞi and jGðtorsionÞi;
respectively, for an acetone-like molecule of case C2v ðG36 Þ: Rotational functions

are classified in the case of an acetone-like molecule as follows:
pffiffiffi
jA1 ðrotationÞ; J; Ki ¼ ð1= 2Þ jJ; Ki þ ð1ÞJK jJ; Ki ; ðK ¼ even; K 0Þ
pffiffiffi
jA2 ðrotationÞ; J; Ki ¼ ð1= 2Þ jJ; Ki ð1ÞJK jJ; Ki ; ðK ¼ even; K 0Þ
pffiffiffi
jA3 ðrotationÞ; J; Ki ¼ ð1= 2Þ jJ; Ki ð1ÞJK jJ; Ki ; ðK ¼ odd; K [ 0Þ
pffiffiffi
jA4 ðrotationÞ; J; Ki ¼ ð1= 2Þ jJ; Ki þ ð1ÞJK jJ; Ki ; ðK ¼ odd; K [ 0Þ:
ð3:4:214Þ
pffiffiffi
When K ¼ 0; the factor 1= 2 should be replaced by 1=2 in the above equations.
Another change is that contributions from the term isJy should be removed in
the parameterized expressions Eq. 3.4.212 of Hamiltonian matrix elements.
3.4.5 Appendix to 3.4
3.4.5.1 Mathematical Relation 1

eivRAM Jz =h Jx iJy eivRAM Jz =h ¼ eivRAM Jx iJy ;
eihRAM Jy =h ðJz iJx ÞeihRAM Jy =h ¼ eihRAM ðJz iJx Þ; ð3:4:215Þ

ei/RAM Jz =h Jx iJy ei/RAM Jz =h ¼ ei/RAM Jx iJy :
Proof By putting

U ¼ eiaJz =h Jx iJy eiaJz =h ; ð3:4:216Þ
we have
oU i i
¼ Jz U þ UJz
oa
h
h
i
¼ ðJz U UJz Þ

h
i
¼ eiaJz =h Jz ; Jx iJy eiaJz =h ð3:4:217Þ

h
i

¼ eiaJz =h i hJy i2 hJx eiaJz =h

h

¼ ieiaJz =h Jx iJy eiaJz =h
¼ iU:
That is
oU
¼ iU: ð3:4:218Þ
oa
Solving this differential equation, we have U ¼ u expðiaÞ: Since U ¼ Jx Jy at

a ¼ 0; we obtain

U ¼ eia Jx iJy : ð3:4:219Þ
The first relation of (3.4.215) is thus derived. h

Similarly, the second and the third relations of (3.4.215) can be derived.
2 3
Jx
6 7
ei/RAM Jz =h eihRAM Jy =h eivRAM Jz =h 4 Jy 5eivRAM Jz =h eihRAM Jy =h ei/RAM Jz =h
Jz
3 2
Jx
6 7
¼ Sþ1 ðhRAM ; /RAM ; vRAM Þ4 Jy 5 ð3:4:220Þ
Jz
Proof From the first relation of (3.4.215) we have
1
ivRAM i

eivRAM Jz =h Jx eivRAM Jz =h ¼ e þ eivRAM Jx þ eivRAM eivRAM Jy
2 2
¼ ðcos vRAM ÞJx ðsin vRAM ÞJy ;
1
ivRAM 1

eivRAM Jz =h Jy eivRAM Jz =h ¼ e eivRAM Jx þ eivRAM þ eivRAM Jy
2i 2
¼ ðsin vRAM ÞJx þ ðcos vRAM ÞJy ;
Therefore, we have
2 3 2 3
Jx Jx
eivRAM Jz =h 4 Jy 5eivRAM Jz =h ¼ Sþ1 ð0; 0; vRAM Þ4 Jy 5: ð3:4:221Þ
Jz Jz
Similarly by using the second and the third relations of (3.4.215), we have
2 3
Jx
6 7
ei/RAM Jz =h eihRAM Jy =h eivRAM Jz =h 4 Jy 5eivRAM Jz =h eihRAM Jy =h ei/RAM Jz =h
Jz
3 2
Jx
6 7
¼ ei/RAM Jz =h eihRAM Jy =h Sþ1 ð0; 0; vRAM Þ4 Jy 5eihRAM Jy =h ei/RAM Jz =h
Jz
3 2
Jx
6 7
¼ Sþ1 ð0; 0; vRAM Þei/RAM Jz =h eihRAM Jy =h 4 Jy 5eihRAM Jy =h ei/RAM Jz =h
Jz
3 2
Jx
6 7
¼ Sþ1 ð0; 0; vRAM ÞSþ1 ðhRAM ; 0; 0Þei/RAM Jz =h 4 Jy 5ei/RAM Jz =h
Jz
3 2
Jx
6 7
¼ Sþ1 ð0; 0; vRAM ÞSþ1 ðhRAM ; 0; 0ÞSþ1 ð0; /RAM ; 0Þ4 Jy 5
Jz
2 3
Jx
þ1 6 7
¼ S ðhRAM ; /RAM ; vRAM Þ4 Jy 5:
Jz
ei/RAM Jz =h eihRAM Jy =h eivRAM Jz =h S1 ðh; /; vÞeivRAM Jz =h eihRAM Jy =h ei/RAM Jz =h
¼ S1 ðh; /; vÞS1 ðhRAM ; /RAM ; vRAM Þ ð3:4:222Þ
Proof Using Eq. 3.4.220 and the relation between the molecule-fixed components
Jx ; Jy ; Jz and the laboratory-fixed components JX ; JY ; JZ of the angular momentum
operator
2 3 2 3
Jx JX
4 Jy 5 ¼ Sþ1 ðh; /; vÞ4 JY 5; ð3:4:223Þ
Jz JZ
together with the commutation relations ½Ja ; Ji ¼ 0ða ¼ x; y; z; i ¼ X; Y; ZÞ; we

have
ei/RAM Jz =h eihRAM Jy =h eivRAM Jz =h Sþ1 ðh; /; vÞeivRAM Jz =h eihRAM Jy =h ei/RAM Jz =h
¼ Sþ1 ðhRAM ; /RAM ; vRAM ÞSþ1 ðh; /; vÞ;
By taking the inverse operation in both sides of the above equation, we obtain the
relation (3.4.222). h
3.4.5.4 Permutation–Inversion Group
The permutation–inversion group proposed by Longuet-Higgins [35] will be

mentioned below briefly.
In order to introduce the permutation–inversion group, we take an exact copy of
some sentences from the paper of Longuet-Higgins [35] with permission from
Taylor & Francis.
‘‘Let P be any permutation of positions and spins of identical nuclei, or any
product of such permutations. Let E be the identity, E
the inversion of all particle
positions, and P
the product PE
¼ E
P: Then the molecular symmetry group is
the set of
1. all feasible P; including E;
2. all feasible P
; not necessarily including E
’’.
The molecular symmetry group used above is equivalent to the permutation–
inversion group. In the above description, the word ‘‘feasible’’ involves very
important concept. ‘‘feasible’’ transformations are those which can be achieved
within the time scale of a given laboratory experiment without passing over an
insuperable energy barrier. Although any P and any P
keep the Hamiltonian
operator of a free molecule invariant surely, the molecular symmetry group
(permutation–inversion group) is composed of feasible P and feasible P
: Such an
idea will be natural, considering that experimental spectra reflect phenomena
occurring only within the time scale of observation.
Let define the group operations in the permutation–inversion group, taking the
case of the CH3 OH molecule as a simple example. Let call methyl protons 1, 2 and
3. If the internal rotation is not feasible, the permutation–inversion group elements
are E and ð23Þ
: ð23Þ
corresponds to a plane of symmetry. As shown in Table 3.8,
a configuration produced by the operation ð23Þ
can be achieved by an overall
rotation about molecule-fixed y-axis. When the internal rotation is feasible, we
have six group operations E; ð123Þ; ð132Þ; ð23Þ
; ð13Þ
; ð12Þ
; which are generated
from two elements ð123Þ and ð23Þ
(such two elements are called generating
operations). We call the permutation–inversion for the methanol G6 for a simple
reason that it has six elements. When FðR1 ; R2 ; R3 ; ra ð1Þ; rb ð2Þ; rc ð3ÞÞ be a
function or an operator regarding to nuclei 1, 2, and 3, the permutation–inversion
operations are defined as follows, obeying Hougen’s method described in
Ref. [48]:

ð23Þ
F R1 ; R2 ; R3 ; ra ð1Þ; rb ð2Þ; rc ð3Þ ¼ F R1 ; R3 ; R2 ; ra ð1Þ; rb ð3Þ; rc ð2Þ ;

ð123ÞF R1 ; R2 ; R3 ; ra ð1Þ; rb ð2Þ; rc ð3Þ ¼ F R2 ; R3 ; R1 ; ra ð2Þ; rb ð3Þ; rc ð1Þ ;

ð23Þ
ð123ÞF R1 ; R2 ; R3 ; ra ð1Þ; rb ð2Þ; rc ð3Þ ¼ ð23Þ
F R2 ; R3 ; R1 ; ra ð2Þ; rb ð3Þ; rc ð1Þ

¼ F R3 ; R2 ; R1 ; ra ð3Þ; rb ð2Þ; rc ð1Þ ;

¼ ð13Þ
F R1 ; R2 ; R3 ; ra ð1Þ; rb ð2Þ; rc ð3Þ :
ð3:4:224Þ
3.4.5.5 Framework Functions Defined so as to be Orthogonalized

with Each Other
Let us put N framework functions localized at each of N potential minima as

follows:
jni ¼ On j1i; ðn ¼ 1; 2; 3; . . .; NÞ; ð3:4:225Þ
where On ðn ¼ 1; 2; 3; . . .; NÞ stand for permutation–inversion group operations

corresponding to feasible large-amplitude vibrations. For the case of a molecule
with one methyl top executing internal rotation, O1 ¼ E (identity transformation),
O2 ¼ ð123Þ; and O3 ¼ ð132Þ: The framework functions jniðn ¼ 1; 2; 3; . . .; NÞ as
defined above in Eq. 3.4.225 have non-zero overlap integrals Dij ðDij ¼ hijjiÞ and it
is inconvenient in the viewpoint of spectral analysis. For avoiding the inconve-
nience, we introduce framework functions defined newly as follows:
N
X
ji : newi ¼ D1=2 jmi; ði ¼ 1; 2; 3; . . .; NÞ ð3:4:226Þ
mi
m¼1
where D1=2 Sd 1=2 Sy ; and d Sy DS is the diagonal matrix of positive eigen-

values obtained by applying the unitary transformation S to D (D is a matrix
composed of elements Dij ði; j ¼ 1; 2; 3; . . .; NÞÞ: For the new framework functions
ji : newiði ¼ 1; 2; 3; . . .; NÞ; it can be shown that
hi : newjj : newi ¼ dij ;

ð3:4:227Þ
ji : newi ¼ Oi j1 : newi;
assuming h1j1i ¼ 1:
For a simple case of a molecule with one methyl top, Eq. 3.4.226 leads to
0 1 2 1 1 30 1
pffiffiffiffiffiffiffiffi þ pffiffiffiffiffiffi
2 pffiffiffiffiffiffiffiffi
1
pffiffiffiffiffiffi
1 pffiffiffiffiffiffiffiffi
1
pffiffiffiffiffiffi
j1 : newi 1þ2S 1S 1þ2S 1S 1þ2S 1S j1i
1
@ j2 : newi A ¼ 6 1 7@
4 pffiffiffiffiffiffiffiffi pffiffiffiffiffiffi pffiffiffiffiffiffiffiffi þ pffiffiffiffiffiffi pffiffiffiffiffiffiffiffi pffiffiffiffiffiffi j2i A
1 1 1 2 1
3 1þ2S 1S 1þ2S 1S 1þ2S 1S 5
j3 : newi pffiffiffiffiffiffiffiffi pffiffiffiffiffiffi pffiffiffiffiffiffiffiffi pffiffiffiffiffiffi pffiffiffiffiffiffiffiffi þ pffiffiffiffiffiffi
1 1 1 1 1 2 j3i
1þ2S 1S 1þ2S 1S 1þ2S 1S
ð3:4:228Þ
where S ¼ h1j2i ¼ h2j1i ¼ h1j3i ¼ h3j1i ¼ h2j3i ¼ h3j2i:

References
1. L.I. Schiff, Quantum Mechanics, 2nd edn. (McGraw-Hill, New York, 1969)
2. G. Herzberg, Atomic Spectra and Atomic Structure (Dover Publications, Inc., New York,
1944)
3. S. Yamamoto, S. Saito, Astrophys. J. 370, L103 (1991)
4. H. Klein, F. Lewen, R. Schieder, J. Stutzki, G. Winnewisser, Astrophys. J. 494, L125 (1998)
5. C.H. Townes, A.L. Schawlow, Microwave Spectroscopy (McGraw- Hill, New York, 1955)
6. M. Born, J.R. Oppenheimer, Ann. Phys. 84, 457 (1927)
7. T. Oka, J. Chem. Phys. 47, 5410 (1967)
8. M.R. Aliev, J.K.G. Watson, in Chapter 1 in Molecular Spectroscopy: Modern Research, vol.
III, ed. by K. Narahari Rao (Academic Press, Orland, 1985)
9. J.L. Dunham, Phys. Rev. 41, 721 (1932)
10. E. Hirota, J.M. Brown, J.T. Hougen, T. Shida, N. Hirota, Pure Appl. Chem. 66, 571 (1994)
11. P.R. Bunker, R.E. Moss, J. Mol. Spectrosc. 80, 217 (1980)
12. W. Gordy, R.L. Cook, Microwave Molecular Spectra (Wiley, New York, 1970)
13. B.T. Sutcliffe, J. Tennyson, Int. J. Quantum. Chem. 39, 183 (1991)
14. P.R. Bunker, Molecular Symmetry and Spectroscopy (Academic Press, New York, 1979)
15. E.B. Wilson, J.C. Decius, P. Cross, Molecular Vibrations (McGraw-Hill, New York, 1955)
16. D. Papoušek, M.R. Aliev, Molecular Vibrational–Rotational Spectra (Elsevier, Amsterdam,
1982)
17. C. Eckart, Phys. Rev. 47, 552 (1935)
18. E.B. Wilson, J. Chem. Phys. 7, 1047 (1939)
19. H.H. Nielsen, Rev. Mod. Phys. 23, 90 (1951)
20. J.K.G. Watson, Mol. Phys. 15, 479 (1968)
21. A.R. Hoy, I.M. Mills, G. Strey, Mol. Phys. 24, 1265 (1972)
22. P. Jensen, P.R. Bunker (eds.), Computational Molecular Spectroscopy (Wiley, Chichester,
2000)
23. E.B. Wilson, J.B. Howard, J. Chem. Phys. 4, 260 (1936)
24. B.T. Darling, D.M. Dennison, Phys. Rev. 57, 128 (1940)
25. P.R. Bunker, P. Jensen, Molecular Symmetry and Spectroscopy, 2nd edn. (NRC Press,
Ottawa, 1998)
26. S. Maes, G. Amat, Cah. Phys. 11, 25 (1957)
27. J.K.G. Watson, J. Chem. Phys. 46, 1935 (1967)
28. J.K.G. Watson, in Vibrational Spectra and Structure, vol. 6, ed. by J. Durig (Elsevier,
Amsterdam, 1977)
29. K. Sarka, D. Papoušek, J. Demaison, H. Mäder, H. Harder, in Vibration-Rotational
Spectroscopy and Molecular Dynamics, ed. by D. Papoušek. Advanced Series in Physical
Chemistry, vol. 9 (World Scientific, Singapore, 1997)
30. K. Sarka, J. Demaison, in Computational Molecular Spectroscopy, eds. by P. Jensen,
P.R. Bunker (Wiley, Chichester, 2000)
31. P.R. Bunker, P. Jensen, Fundamentals of Molecular Symmetry (IOP Publishing, Bristol,
2004)
32. P. Jensen, P.R. Bunker, in Encyclopedia of Chemical Physics and Physical Chemistry, eds. by
J.H. Moore, N.D. Spencer (IOP Publishing, Bristol, 2001)
33. R.G.A. Bone, T.W. Rowland, N.C. Handy, A.J. Stone, Mol. Phys., 72, 33 (1991)
34. V. Špirko, S. Civiš, M. Ebert, V. Danielis, J. Mol. Spectrosc. 119, 426 (1986)
35. H.C. Longuet-Higgins, Mol. Phys. 6, 445 (1963)
36. V. Špirko, W.P. Kraemer, J. Mol. Spectrosc. 133, 331 (1989)
37. C.D. Esposti, G. Cazzoli, D. Damlani, P.G. Favero, F. Strumia, Infrared Phys. 28, 21 (1988)
38. W.J. Lafferty, F.J. Lovas, J. Phys. Chem. Ref. Data 7, 441 (1978)
39. A.J. Alexander, H.W. Kroto, D.R.M. Walton, J. Mol. Spectrosc. 62, 175 (1976)
40. M.B. Bell, P.A. Feldman, M.J. Travers, M.C. McCarthy, C.A. Gottlieb, P. Thaddeus,
Astrophys. J. 483, L61 (1997)
41. R. Boquet, J. Demaison, L. Poteau, M. Liedtke, S. Belov, K.M.T. Yamada, G. Winnewisser,
C. Gerke, J. Grippm, Th. Koöhler, J. Mol. Spectrosc. 177, 154 (1996)
42. G. Herzberg, Molecular Spectra and Molecular Structure, vol. III (D. van Nostrand Co, Inc.,
Princeton, 1967)
43. K. Yamada, J. Mol. Spectrosc. 81, 139 (1980)
44. F.W. Loomis, R.W. Wood, Phys. Rev. 32, 223 (1928)
45. B.P. Winnewisser, J. Reinstädtler, K.M.T. Yamada, J. Behrend, J. Mol. Spectrosc. 136, 12
(1989)
46. J.T. Hougen, I. Kleiner, M. Godefroid, J. Mol. Spectrosc. 163, 559 (1994)
47. D.R. Herschbach, J. Chem. Phys. 31, 91 (1959)
48. J.T. Hougen, J. Mol. Spectrosc. 89, 296 (1981)
49. N. Ohashi, J.T. Hougen, R.D. Suenram, F.J. Lovas, Y. Kawashima, M. Fujitake, J. Pyka, J.
Mol. Spectrosc. 227, 28 (2004)
Chapter 4
Laboratory Experimental Methods
Thomas F. Giesen and Karl Jacobs
4.1 Introduction
The vast majority of molecules detected in space has been identified by rotational
and ro-vibrational spectra in the millimeter- and sub-millimeter wavelength
region. Laboratory studies at sub-millimeter wavelengths or terahertz frequencies
require sophisticated spectroscopic techniques, which have been developed over
the last few decades by worldwide leading spectroscopy groups. In this section
spectroscopic tools for high resolution Terahertz spectroscopy of gas phase mol-
ecules are presented and illustrated by examples from the Cologne laboratories.
4.1.1 Molecules Relevant to Laboratory Astrophysics
The class of molecules relevant to astronomy is quite extensive and it contains

well known stable molecules like carbon monoxide, water, and ammonia, as well
as highly reactive, unstable species, e.g., unsaturated carbon hydrides and a
number of molecular ions. Hydrogen is by far the prevailing element of gas phase
interstellar matter, and molecular and protonated molecular hydrogen are the first
molecules to be formed in interstellar space. In reactions with more heavy atoms,
e.g., carbon, oxygen, nitrogen, sulfur, fluorine and chlorine, light molecular
hydrides and their cations are formed. Up to date CHþ , CH, CH2 , NH, NH2 ; NH3 ,
OH, OHþ ; H2 O; H2 Oþ ; H3 Oþ , HCl, H2 Clþ , HF, SHþ , and H2 S have been detected
in diffuse clouds and cold environments of the interstellar medium. With the center
T. F. Giesen (&) and K. Jacobs

Universität zu Köln, I. Physikalisches Institut, Zülpicher Str. 77, 50937 Köln, Germany
e-mail: giesen@ph1.uni-koeln.de

180 T. F. Giesen and K. Jacobs
of mass almost at the central heavy atom, light hydrides have small momenta of
inertia and accordingly large rotational constants. As a consequence their rota-
tional spectra are predominantly found in the THz region. For example, the ground
state rotational constant of methylidynium, CHþ , is B = 418 GHz. Only recently
accurate laboratory data of the lowest rotational transition J ¼ 1 0 at 835 GHz
were reported by Amano in 2010 [1], and following this CHþ was astronomically
detected in the direction of massive star forming regions by Falgarone et al. [2]
using the Heterodyne Instrument for the Far Infrared (HIFI) onboard the Herschel
satellite [3].
Due to the lack of a permanent electrical dipole moment, molecular and pro-
tonated molecular hydrogen, H2 , and Hþ 3 , have no pure rotational spectra to be
observed in the millimeter- and sub-millimeter-wavelength region, but their deu-
terated isotopologues HD, H2 Dþ ; D2 Hþ posses rotational transitions in the Tera-
hertz region thanks to a dipole moment pointing along the axis of symmetry.
Because of their relevance to astronomy they are prime targets for laboratory
investigations. The lowest rotational transitions of para-H2 Dþ at 1.370 THz and of
ortho-D2 Hþ at 1.477 THz were recently revisited by Asvany et al. [4] applying
most accurate laser induced reaction spectroscopy at sub-millimeter wavelengths.
From the list of molecules detected in space, see tables in Chap. 5, it becomes
evident that carbon containing species contribute mostly to the diversity of
interstellar chemistry. Carbon chain molecules terminated by nitrogen, oxygen or
hydrogen groups are present as well as pure carbon chains and carbon hydrogen
rings. Most of them are highly reactive and unstable under terrestrial conditions.
Laboratory studies of carbon species therefore rely on advanced preparatory
methods such as electrical discharge cells and supersonic jet laser ablation sources.
Provided a permanent electrical dipole moment their rotational spectra spread over
the frequency range of a few Terahertz, but at low temperatures, as predominate in
cold interstellar clouds, the maximum of line intensities is shifted to lower fre-
quencies of tens to hundreds of GHz. Pure carbon chain molecules have no per-
manent dipole moment and thus no pure rotational spectra, but thanks to their low
lying vibrational modes they can be identified by ro-vibrational spectra in the
1–5 THz region.
Figure 4.1 shows the stick spectra of a few example molecules relevant to
astrophysics. It is clearly to be seen that H2 Dþ and H2 O have only few rotational
transitions in the Terahertz region. In opposite, pure rotational spectra of heavier
molecules, e.g., the C3 H carbon chain molecule, are more dense and mostly below
1 THz whereas ro-vibrational transitions of lowest bending modes of carbon chain
molecules occur in the Terahertz region, as for example in case of the v ¼ 1 0
transitions of C3 H. Figure 4.1 also shows that spectra of larger, asymmetric top
molecules such as dimethyl ether (DME) are rather complex with spectral line
densities often close to the confusion limit. Beside DME, other organic molecules
have been found mostly in warm regions of the interstellar medium with conse-
quently complicated and dense spectra. It is one of the challenges of laboratory
astrophysics to record and analyze spectra of complex molecules in the Terahertz
4 Laboratory Experimental Methods 181
region and to give highly accurate frequency positions for a clear assignment of
spectra from astrophysical observations.
So far laboratory studies led to the large amount of spectroscopic data currently
available in molecular data bases, such as the Jet Propulsion Laboratories (JPL)
data base [5] and the Cologne Data Base for Molecular Spectroscopy [6]. The
number of molecules under laboratory investigation is still growing and hence the
number of entrances in data bases do as well. Especially the demand for Terahertz
spectroscopic data is tremendous due to the advent of observatories operating at
high frequencies, such as the APEX telescope [7] or the HIFI instrument onboard
the Herschel satellite [3]. No doubt, we currently live in a very exciting and
challenging time for both laboratory astrophysics and astronomical observations.
4.1.2 Spectral Line Widths and Intensities
Absorption spectroscopy is widely used to study spectra of gas phase molecules at

highest spectral resolution. The general setup of an absorption spectrometer con-
sists of a monochromatic tunable radiation source, a sample of molecules under
investigation, and a sensitive detection system. As has been pointed out above, the
Terahertz spectral region is most relevant for astrophysical observations and,
therefore, appropriate tunable radiation sources with highest spectral resolution for
Fig. 4.1 Calculated stick spectra of specific molecules in the Terahertz region
laboratory investigations are a prerequisite. At low sample pressures the spectral

line width is given by the molecules velocity distribution, which leads to Doppler
line profiles with full width at half maximum DmD ,
rffiffiffiffiffi
7 T
DmD ¼ m 7:1 10 : ð4:1:1Þ
M
Here m denotes the transition frequency, T the temperature in Kelvin (K) and M the
mass of the molecule in atomic mass units (amu). At 1 THz and room temperature
the Doppler width of a molecule like C3 H, with M = 37 amu, is DmD ¼ 2 MHz,
which requires a spectrometer with a spectral resolution Dm of better than 10-6.
Radiation sources with monochromaticity Dm of typically 107 to 108 are needed
to measure molecular line contours and, in sophisticated sub-Doppler experiments,
the underlying substructure of hyperfine lines.
4.2 Monochromatic Radiation Sources for High Resolution

Absorption Spectroscopy
4.2.1 Overview
Until recently a major constraint in high resolution Terahertz spectroscopy was the
lack of appropriate monochromatic radiation sources. High frequency electronic
devices, such as Klystrons, Gunn diodes and electron tubes [8] have been used to
generate monochromatic radiation up to roughly 100 GHz and subsequent fre-
quency multiplication techniques by Schottky multiplier devices generate radiation
in the THz region, although with a significant drop in output power. In the 1980th
backward wave oscillators (BWOs) were developed for Terahertz spectroscopy.
They produce radiation from a guided electron beam traveling along a high voltage
periodic structure placed in a resonator tube. BWOs above 1 THz, which are
tunable over a spectral range of a few hundred GHz, and with output powers of
several mW, were developed and successfully applied to spectroscopy. For two
decades, BWOs became workhorse radiation sources in many laboratories until
recently, when multiplier based all solid state sources became more favorable. For
several reasons BWOs did not prevail in the spectroscopic community, major ones
being constraints in handiness, lifetime, and commercial availability. Instead,
commercially available cascaded Schottky multiplier chains fed by highly mono-
chromatic radiation of a synthesizer at typically 10–30 GHz are nowadays widely
in use to cover the spectral range up to 2 THz. Although the optical quality of their
radiation is better in comparison to BWO tubes, the low output power at the end of
the multiplication process is still of concern. Amplifiers for frequencies above
100 GHz and multipliers with smaller losses may ease this constraint and allow to
further extend the spectral range above 2 THz. A promising approach to frequency
multiplication are superlattice devices and first results have been reported recently
by Endres et al. [9]. Because they generate only odd numbered harmonics the
power-drop at higher harmonics is less drastically compared to Schottky type
multipliers, although at 3.1 THz the output does not exceed a few 10-9 W. While
for frequencies in the mid-infrared region and above quite powerful high resolu-
tion sources are available, the range from 2 to 10 THz is still a technical challenge.
Efforts to bridge the Terahertz gap from the high frequency side have been
reviewed by Siegel [10] and new solutions, such as quantum cascade lasers for
Terahertz radiation, became commercially available recently.1
Lasers have also been used for both frequency down- and up-conversion to
generate Terahertz radiation. Far-infrared gas lasers at fixed frequencies are mixed
with radiation from tunable sources on appropriate mixing devices. By this
manner, tunable laser-sideband radiation is generated at Terahertz frequencies,
which can be used for high resolution spectroscopy. Finally, laser-mixing of
far-infrared and near infrared lasers were used for difference frequency generation
in the Terahertz range. First commercially available systems are offered by Toptica
Photonics AG.2
In cases when less spectral resolution is sufficient, Fourier Transform spec-
troscopy proofs to be a useful tool to record broad scan spectra. The constraint of
commonly used black body radiation sources (globars)—emitting only few pho-
tons in the Terahertz region with low optical beam quality—has recently been
overcome by collimated synchrotron radiation sources for the far infrared, which
emit 2–3 orders of magnitudes more photons than globars do. The French SOLEIL
and the Canadian Light Source synchrotron facilities both have far infrared beam
lines combined with high resolution Fourier-Transform spectrometers. Although
Fourier Transform spectroscopy is of relevance to Terahertz spectroscopy, in the
following section the focus is on spectrometers which make use of monochromatic
radiation sources.
4.2.2 Terahertz Backward-Wave Oscillators
4.2.2.1 Principle and Performance of BWOs
One of the most successfully used sources for monochromatic THz radiation is the
BWO. The device consists of a periodic metallic structure placed inside a reso-
nator tube. Electrons leaving a heated filament are accelerated in a strong electrical
field of several kV. An axial homogenous magnetic field of order of 1 T confines
the electrons to a narrow beam with current densities as high as 300 A/cm2. By
passing the low voltage periodic structure (typically at 20 V) the electron phase
1
Alpes Lasers, http://www.alpeslasers.ch/.
2
http://www.toptica.com/.
velocity is modulated and energy is transferred to eigenmodes of the resonator,

thereby stimulating electromagnetic modes to oscillate.
The oscillation frequency depends on the electron velocity and the dimension of
the periodic structure. Higher electron velocities or reduced grid spacings both
lead to higher frequencies. Figure 4.2 shows a schematic drawing of the BWO
tube and a typically frequency–voltage curve. At -5 kV cathode voltage the
electron velocity reaches 10% of the speed of light. A detailed theoretical
description of BWOs can be found in articles by Levush [11, 12] and references
therein. Tubes with different periodic structures have been designed and fabricated
by the Russian company ISTOK Research and Production in Fryazino. They cover
the spectral range from 36 GHz to 1.4 THz (see [13]) with output powers of some
tens of mW at low and medium frequencies (300–600 GHz) and one to a few mW
at the high frequency end. The line width of BWO radiation is of the order of few
MHz. As an example the upper trace of Fig. 4.3 shows the rQ3 branch spectrum of
H2 S2 at 978 GHz which has been recorded with a BWO spectrometer [14, 15]. For
comparison the lower trace shows the same spectrum measured with the high-
resolution Fourier-Transform Spectrometer at Giessen university [16]. The full
Doppler width at half maximum of H2 S2 lines at room temperature and at 1 THz is
1.5 MHz, which is easily resolved by a BWO spectrometer. On the other hand an
FT-spectrometer with two meter optical path length has a spectral resolution of
only 2:5 103 cm1 or 75 MHz. The line width of BWOs is mainly attributed to
phase noise, instabilities of the voltage supply and inhomogeneity of the magnetic
field. The frequency accuracy and stability of BWOs can significantly be improved
by at least three orders of magnitude when the output radiation is phase-locked to a
highly accurate frequency standard. Rubidium standards with frequency accuracy
of Dm=m ¼ 1011 or the GPS time standard are well suited to stabilize the BWO
frequency in a phase-locking loop (PLL) mode.
Radiation Out
Window
Backward Wave Oscillator OB-44
U/f characteristic
1.1
1
Magnetic Field
f [THz]
H e 0.9
0.8
0.7
Slow Wave Structure
0.6
2 2.5 3 3.5 4 4.5 5 5.5 6 6.5 7
Filament Filament + Cathode
Slow Wave Structure Voltage U [kV]
Fig. 4.2 Schematic drawing of a backward wave oscillator (BWO) and typical frequency versus
voltage curve
Fig. 4.3 rQ3 branch of H2S2 recorded with a BWO spectrometer (upper trace) and with a high
resolution FT-Spectrometer (lower trace). (Modified version of Fig. 4.2 with permission from
[14], Ó 1993, Elsevier)
4.2.2.2 Principle of Phase-Lock Loops
The technique of phase-locking is widely used to reduce and stabilize radiation in

the cm- and mm-wavelength region (see, e.g., Bester et al. [17]). Phase-locking
applied to a BWO at 170 GHz has been reported as early as 1970 [18]. Later Goy
et al. used a phase-lock loop to stabilize the output of a carcinotron, which is a
West-European type of a BWO [19]. In 1982 van Vliet reported on a frequency
doubled 230 GHz PLL-BWO which was used as accurate local oscillator in a
heterodyne receiver setup at 460–500 GHz [20]. Goy et al. and Bogey et al. used
phase-locked Thomson-CSF carcinotrons up to 480 GHz [21], and Bava et al. a
600 GHz PLL-BWO [22]. In 1993 the Cologne spectroscopy group reported the
first PLL-BWO at 1 THz (see [15, 23, 24]). Later Bogey used Russian PLL-BWOs
up to 900 GHz [25]. PLL-BWOs were also used in laboratory groups in Japan,
namely by Yamada at 1 THz [26] and Amano at 600 GHz [27]. Thanks to
Terahertz BWOs an extensively large number of molecules has been measured and
most reliable spectroscopic data in the submm-wavelength region were achieved.
A PLL compares the frequency of a monochromatic radiation source with a
most accurate frequency standard. Deviation from the reference standard generates
an error signal in the PLL which is fed back to the monochromatic radiation source
to correct the output frequency. Frequency accuracies of Dm=m ¼ 109 and better
are feasible which corresponds to an absolute frequency accuracy of 1 kHz at
1 THz.
To phase-lock the frequency (mBWO ) of a BWO to the frequency (mREF ) of a
reference standard the BWO frequency is superimposed on a fast harmonic mixer
with the mth harmonics of a phase locked synthesizer signal (m mMW ). The
generated intermediate frequencies mIF ¼ mBWO m mMW is amplified and used in
a feed back loop to control the voltage supply of the BWO. A schematic diagram
of a PLL unit is shown in Fig. 4.4. An IF matching unit separates the intermediate
frequency (IF), which is typically in the order of a few 10–350 MHz, from other
products of the mixer unit and the IF signal is amplified by a low noise HEMT
amplifier (For details see [28]). The low frequency intermediate signal enters a
Fig. 4.4 Schematic diagram of a PLL unit for locking the frequency of a BWO to a reference
frequency. (Reproduction of Fig. 4.3 with permission from [28]. Ó 1998, American Institute of
Physics)
phase detector and is compared to the precise phase of a reference signal (mREF ),
which can be either a rubidium time standard, the GPS time reference signal, or a
signal of comparable accuracy. The Cologne spectrometers are synchronized to
rubidium standards of either 5 or 10 MHz. The phase comparator detects slightest
deviations of the intermediate signal from the reference and generates an error
signal voltage which is amplified in a high gain fast DC amplifier and fed to the
anode of the BWO to correct the BWO frequency. A review on phase locked BWO
has been published by Krupnov [29].
The phase lock is monitored by a frequency analyzer, which displays the IF
signal (see Fig. 4.5). The quality of the PLL depends on the stability and sufficient
radiation power of the BWO and the synthesizer signal as well. For stable oper-
ation the peak maximum of the IF signal has to be at least 30 dB above the side
lobes. Additional filters can partly reduce the levels of side lobes.
4.2.3 BWO Terahertz Spectrometers
Phase locked BWOs have been implemented in high resolution spectrometers to

record spectra of gas phase molecules in simple absorption cells, electrical dis-
charge cells and in supersonically cooled molecular jets.
One of the successfully operating instruments is the Cologne Terahertz spec-
trometer which uses several BWO tubes to cover the spectral range from 200 GHz
to 1.2 THz. Figure 4.6 shows the experimental setup with a standard absorption
cell 2–3 m long. To avoid collisional line broadening the pressure in the sample
cell is kept typically at a few tens to hundreds lbar. The signal is detected on a
Fig. 4.5 Intermediate

frequency (IF) signal of a
phase locked BWO
monitored by a frequency
analyzer (logarithmic plot).
The noise sidebands are
approximately 48 dB at
10 kHz resolution bandwidth
below the carrier.
(Reproduction of Fig. 4.6
with permission [28]. Ó
1998, American Institute of
Physics)
4 K liquid helium cooled InSb hot electron bolometer (QMC Instruments, Cardiff)
pffiffiffiffiffiffi
with sensitivity of 1011 W= Hz. The detector peak sensitivity can be tuned from
typically 600 GHz up to 2 THz with an inhomogeneous magnetic field. This
allows sensitive detection from 100 GHz up to 2.3 THz. To increase the sensitivity
of the spectrometer the BWO output is frequency modulated at typically 50 kHz to
record signals in second derivative using a fast Lock-In amplifier. A PC records
the output signal of the Lock-In and controls the frequency tuning of the BWO by
continuously stepping the output frequencies of a synthesizer. For the setup shown
in Fig. 4.6 a 110 GHz and alternatively a 18 GHz synthesizer was used. Tuning
the synthesizer frequency the PLL error signal corrects the BWO input voltage
until the error signal is zero. For example, to record spectra at 1 THz the 9th
harmonic of a 110 GHz synthesizer signal is used to generate an IF signal of
350 MHz and slowly scanned to record spectra. By this manner a number of stable
molecules, such as carbon monoxyde (CO), water (H2 O), methanol (CH3 OH),
hydrogen cyanide (HCN), methyl cyanide (CH3 CN), formic acid (HCOOH),
dimethyl ether (CH3 OCH3 ), ethylmethylether(C2 H5 OCH3 ), and many others,
including isotopologues, have been recorded with the Cologne THz spectrometer.
Beside stable molecular species a large number of radicals and unstable molecules,
produced in discharge cells, have been studied with the aid of BWO spectrometers
such as light hydrides, SH, PH, NH, CH2 , chain molecules like C3 ; HC3 ; HC3 N and
HC5 N, the sulfur containing species HSOH and HSSH, and molecular ions, for
example, COþ and HCOþ . A list of recently investigated molecules at Cologne is
given in [30]. Figure 4.7 shows as an example the rQ4 branch spectrum of HSSH at
1.258 THz.
4.2.4 Sub-Doppler Spectroscopy
At pressures low enough that collisional broadening of spectral lines is negligible

molecular line shapes are Gaussian and the line width is determined by the
molecular mass, the center frequency, and the temperature of the gas (see
Eq. 4.1.1). The Gaussian profile results from an ensemble of molecules with
Fig. 4.6 Schematic diagram of the Cologne terahertz spectrometer. (Reproduction of Fig. 4.1
with permission from [28]. Ó 1998 American Institute of Physics)
Fig. 4.7 Doppler limited

spectrum of HSSH at
1.258 THz recorded in
second derivative. In
comparison to the main
isotopomer, the bandhead of
the 34S isotopomer is
displaced toward higher
frequencies. (Reproduction of
Fig. 4.4 of [85], Ó 1995 with
permission from Elsevier)
Maxwell–Boltzmann velocity distribution which absorb electromagnetic radiation

at Doppler shifted frequencies m ¼ m0 þ DmD . Here DmD ¼ vz m0 =c denotes the
frequency shift due to the Doppler effect for molecules with velocity vz propa-
gating parallel to the incident beam. For vz [ 0; DmD is negative and the absorption
frequency m is shifted to lower frequencies. m0 is the rest frequency at vz = 0, and
c is the speed of light.
Fig. 4.8 Sub-Doppler

spectrum of the J ¼ 4 3
transition of DCN in its first
excited m2 bending state at
290 GHz including hyperfine
components (lower trace).
For comparison the Doppler
limited spectrum is shown in
the upper trace. (Reproduced
from [34], Ó 2004, with
permission from Elsevier)
To record sub-Doppler line profiles part of the incident beam is reflected back by
a mirror or at the detector surface and passes the gas cell a second time but in
opposite direction, thereby depopulating the lower states of molecules with velocity
component -vz. When the spectrometer is tuned to the rest frequency m = m0,
radiation of both incident and back reflected beam is absorbed by the same group of
molecules with velocity vz = 0. Radiation of the back reflected beam now
depopulates the lower state of molecules with vz = 0 and thus less radiation of the
incident beam gets absorbed. As a consequence a narrow dip at rest frequency m0
occurs and in favorable cases underlying hyperfine structure transitions can be
resolved. The high-resolution possibilities of BWO applications have been dem-
onstrated by sub-Doppler spectra of C17 O at 449 GHz [31], DCN up to 942 GHz
[32], H13 CN at 604 GHz and D13 C15 N at 764 GHz [33, 34]. Figure 4.8 shows the
sub-Doppler spectrum of the J ¼ 4 3 transition of DCN in its first excited m2
bending state at 290 GHz. The line width of hyperfine components is 30 kHz with a
corresponding accuracy of 1–5 kHz.
4.2.5 Multiplier-Spectrometer
Above 1 THz the output power of BWOs drops drastically. Electrical and
mechanical constraints restrict the operational range of BWOs for spectroscopic
applications to slightly above 1 THz. To reach frequencies above 1 THz multipliers
are nowadays widely in use to generate higher harmonic output frequencies from
low fundamental input frequencies. The development of multiplier configurations is
subject of current research [10, 35–39]. Multiplier devices make use of the non-
linear current–voltage response of semiconductor materials to generate higher
harmonic frequencies. Today multiplier devices and cascaded multiplier chains are
Fig. 4.9 Schematic drawing of a multiplier spectrometer. The setup uses a planar Schottky
multiplier. (Reproduced from [44], Ó 2000, with permission from Elsevier)
commercially available for frequencies up to 2 THz with lW output power [40].3

The use of multiplier chains up to frequencies above 2 THz in molecular spec-
troscopy has recently been described thoroughly by Drouin et al. [41].
Often planar Schottky diodes based on GaAs semiconductors are used for
frequency multiplication. Semiconductor doping concentrations, break down
voltage, heat conductivity, anode diameter size and the epitaxial layer thickness
are parameters which need to be optimized to allow for high-conversion effi-
ciencies. The diodes are placed in state-of-the art cross guide or inline waveguide
blocks which are either mechanically milled or electro-formed. The mechanical
and electrical design of the multiplier device is crucial to achieve broadband
response, low standing wave ratios, and good idler matching within the operation
band. For an extension to higher frequencies low series resistances, low parasitic
capacities and high cut-off frequencies of the semiconductor material is pre-
requisite. For details on solid state multipliers see for example [40, 42]. As fun-
damental frequency sources for input of multiplier devices BWOs, Gunn diodes,
and commercial synthesizers with sufficient output are used. Figure 4.9 shows the
Cololgne spectrometer setup with a phase-locked BWO as radiation source and a
planar Schottky multiplier device to produce 1.4 [43] and 1.6 THz [44] radiation.
Different types of multipliers such as superlattice devices [45] and monolithic
membrane diodes (MOMEDs) [39] have been developed and tested in recent years
to generate frequencies up to 3 THz.
3
Virginia Diodes, Inc., 979 Second Street SE, Suite 309, Charlottesville, VA 22902; Radiometer
Physics GmbH, http://www.radiometer-physics.de/rpg/html/home.html
4.2.6 Planar Schottky Multipliers
Figure 4.10 shows a planar Schottky frequency multiplier designed for broad-band
spectroscopy at 1.38 THz. The multiplier generates harmonics of a fundamental
frequency of 110–240 GHz. A planar low-capacitance resistive mixer diode is
adapted to broad-band impedance matching. The planar Schottky diode provides a
junction capacitance of 3 fF, a series resistance of 11 X, an ideality factor of 1.3, a
breakdown voltage of 8.4 V, and additionally a small parasitic capacitance (less
than 1 fF). With these specifications the calculated cutoff frequency is 4 THz. The
anode size of the diode is less than 2 1 lm2 . The anode beam lead is made of
gold and is about 300 lm in length. This offers a very low ohmic resistance at
Terahertz frequencies and a mechanically flexible anode. To minimize the para-
sitic capacitance of the anode, an air bridge approx. 20 lm in length was etched
below an Au-anode finger. To achieve a mechanically stable contact, the volume
of the air bridge is filled with polyimide (r ¼ 2:7) and a part of the finger structure
is additionally fixed with a polyimide layer on the top. The GaAs chip is etched
down to a size of 170 100 lm2 . A GaAs-substrate (doping concentration, n ¼
2 3 1018 cm3 ) with a 2 lm buffer layer (n ¼ 5 1018 cm3 ) and a 50 nm-
thin epilayer (n ¼ 5 1017 cm3 ) was employed. The ohmic contact is made on
the surface of the GaAs crystal and spaced 8 lm from the junction area, thus
reducing effectively the ohmic resistance of the device. The diode definition and its
fabrication were carried out in the Laboratory of Semiconductor Devices, State
University of Nishnii Novgorod, Russia [42]. The polyimide reinforced film-diode
is quite rugged and reliable, with enhanced power handling capability compared to
whisker-contacted diodes. The input waveguide (WR 10) of the multiplier is lin-
early tapered from a full-size rectangular waveguide to a single-ridge waveguide,
which forms a small gap of 300 lm.
Figure 4.11 shows the multiplier block with the planar Schottky diode incor-
porated. The planar diode is mounted across the gap, with the cathode connected to
the planar filter structure. An E - H tuner is mounted on the input waveguide
Fig. 4.10 Design of a GaAs Schottky planar film diode. The overall size of the diode is
500 lm 9 150 lm; the anode size is less than 2 lm 9 1 lm. (Reproduced with permission from
[43], Ó 1999, IEEE)
Fig. 4.11 Schematic

diagram of the multiplier
block, the channel cover, and
the rectangular horn antenna.
The drawing is not to scale.
(Reproduced with permission
from [43]. Ó 1999, IEEE)
flange of the multiplier to improve the input power matching. The planar filter and
the channel cover forms a suspended substrate transmission line. The horn antenna
is a scaled version of a standard-gain rectangular horn antenna and was milled in
four single-plane triangular parts. For a particular backshort setting, a bandwidth
of 1 GHz at the second harmonic is achieved. Sufficient power for molecular
spectroscopy beyond 1 THz can be produced. Compared to whisker contacted
devices planar diodes are easy to handle.
4.2.7 Monolithic Membrane Diodes
For frequencies above 50 GHz GaAs is the III–V semiconductor material of

choice to fabricate solid state multipliers with response times of sub-picoseconds,
corresponding to Terahertz frequencies. However, short response times require
fabrication of rather thin semiconductor microstrip circuits, which are difficult to
machine. An alternative option for low capacity devices is the production of
MOMED, which are made of ultra thin layers. MOMEDs are used as mixers in
various applications up to 2.5 THz. One such device fabricated at the JPL in
Passadena has been used as frequency tripler with adjustable backshorts. The
device has been tested with a high power (10 mW) BWO at 780 GHz as input
radiation source to generate output frequencies of 2.3–2.7 THz. Figure 4.12 shows
a D2 O transition at 2.33 THz recorded in Cologne.
4.2.8 COSSTA: Cologne Sideband Spectrometer for THz

Applications
An alternative approach to generate Terahertz radiation is the FIR-laser sideband

technique. Radiation from an optically pumped FIR gas laser is superimposed with
radiation of a tunable monochromatic microwave source on a fast, non-linear
mixing device to generate sidebands at the sum and difference frequencies of the
two input frequencies. FIR gas lasers, when operated with different gain media,
such as CH3 F, CH3OH, and HCOOH have thousands of laser lines in the range of
Fig. 4.12 A 2.3 THz

spectrum of D2O recorded
with a MOMED multiplier
spectrometer
0.2 to over 10 THz. Strong THz laser lines at a few tens mW mixed with
microwave radiation from either a YIG oscillator (2–4 GHz) [46], a frequency
multiplied synthesizer (up to 75 GHz) [47], or a klystron (22–114 GHz) [48]
have been used to generate tunable laser sideband signals up to 3 THz on a
GaAs-Schottky barrier diode. The microwave is either coupled to the mixer
coaxially or via waveguides. The separation of sidebands from the fundamental
laser radiation is achieved by a diplexer in combination with a Farbry–Perot
interferometer or by a grating monochromator. Frequency tuning of the laser
sidebands is achieved by tuning the frequency of the microwave radiation source.
In 1997 the Cologne group presented a FIR sideband laser system using a PLL
locked BWO (280–380 GHz) as tunable microwave source [49]. The high fre-
quency of the BWO leads to sidebands well separated from the carrier frequency
which can easily be extracted by a grating monochromator. Both the optical
coupling of BWO radiation onto the mixer as well as the clearly separated side-
bands have substantially simplified the use of laser sideband techniques for
Terahertz applications. A schematic overview of the layout of the Cologne side-
band spectrometer for Terahertz applications (COSSTA) is presented in Fig. 4.13.
Aside from the traditional absorption cell and the hot carrier liquid-He-cooled
magnetically tuned InSb detector COSSTA consists of three essential units: (a) the
frequency-stabilized FIR laser system, (b) the evacuated optics and mixer arrange-
ment, which serves to superimpose the BWO and FIR beams, and (c) the BWO phase-
lock assembly. The radiation of a frequency-stabilized FIR ring-laser system is
mixed on an open-structure Schottky barrier diode pumped with phase-stabilized
power output from a BWO for generating the sidebands. The tunability of the
system is basically determined by the frequency coverage of the BWOs, which is
about 100 GHz at 330 GHz center frequency. An example of the wide tuning
range and the high frequencies accessible with COSSTA is shown in Fig. 4.14.
The J ¼ 16 15 rotational transitions of CO in different excited vibrational states
Fig. 4.13 Cologne sideband spectrometer for Terahertz applications—COSSTA. (Reproduction

of Fig. 4.1 of [51], Ó 2000, with permission from Elsevier)
Fig. 4.14 The J ¼ 16 15

transitions are shown for
v = 0–3. The lines appear as
second derivatives of a
Gaussian line-shape because
of the 2f-modulation.
Reproduced from Fig. 4.2 of
[86] with permission Ó 2009,
ESO
of quantum numbers v = 0, 1, 2, and 3 in the range from 1791 to 1841 GHz is

depicted.
The frequency stability of this sideband spectrometer is limited solely by the
instability of the FIR laser frequency, which is typically in the range of 15 MHz.
To overcome this limitation, an analog automatic frequency control (AFC) loop
was developed to synchronize the FIR laser frequency of 1.626 THz with the 15th
harmonic of a Gunn oscillator (InP Gunn, 108 GHz), which is phase-locked to a
stable reference frequency [50]. The IF of the harmonic mixer is amplified by an
ultralow noise HEMT operating at 350 MHz. The frequency error signal drives a
piezo actuator to move the grating of the CO2 pump laser whose frequency is
corrected within the gain profile of a CO2 laser line. By this manner long term
frequency stability (AFC loop error) of 2 kHz and absolute frequency calibration
of the FIR laser was achieved. The BWO submillimeter and FIR laser radiation are
coupled quasi-optically onto a whisker-contacted Schottky diode for generating
laser sidebands. Separation of laser frequency and tunable sidebands are achieved
with a low-loss blazed grating (see Fig. 4.13). Finally the tunable sideband beam
passes the absorption cell and is focused onto a liquid-helium-cooled InSb
bolometer. With a 1T6 Schottky diode (University of Virginia), sideband power
levels of 1:6 lW were reached. To avoid power losses introduced by water vapor
absorption, a vacuum box for the optical system is used to cover the entire side-
band beam and the main part of the laser beam. Various applications of COSSTA
to molecular spectroscopy have been published (see, e.g., [32, 51–55].
4.3 Spectroscopy in Supersonic Jets
4.3.1 General Remarks
Supersonic jets are nowadays widely used for spectroscopic investigations of

reactive, short-lived molecules such as molecular radicals and ions [56], weakly
bound van der Waals complexes [57], and large bio-molecules [58]. In the adia-
batic expansion of a buffer gas the rotational temperature of seeded molecules
drops to a few K and the molecular partition function is strongly reduced compared
to standard gas cell experiments at room temperature. Adiabatically cooled mol-
ecules have significantly simplified rotational spectra with strongly enhanced line
intensities which compensate for the usually short path lengths of the probing
beam through a jet.
4.3.2 SuJeSTA
A Supersonic Jet Spectrometer for Terahertz Applications (SuJeSTA) has been

developed at the Cologne laboratories to study reactive molecules by means of
high resolution mm- and submm-wavelength spectroscopy at lowest temperatures
(see Fig. 4.15). For this purpose a supersonic jet apparatus has been combined with
a phase stabilized sub-millimetre-wave spectrometer. Highly reactive molecules
are produced either in an electrical discharge, which is placed in the throat of the
nozzle, or by laser ablation of solid targets such as graphite and silicon. Both
techniques are well established and have been described in the literature (see, e.g.,
[59, 60]).
Fig. 4.15 SuJeSTA Spectrometer is presented as a block diagram. (Reproduction of Fig. 4.1 of
[87], Ó 2009, with permission from Elsevier)
4.3.2.1 Electrical Discharge Sources
A pulsed jet discharge source at typically 20–50 Hz repetition rate and pulse
duration times of a few hundreds of ls is used to produce reactive molecules from
a mixture of precursor gases (see Fig. 4.15). Flow controller units allow for stable
and reproducible gas flow conditions and constant mixing ratios during the mea-
surements. Fast background subtraction during the off-times of the jet leads to a
substantial reduction of the noise level and allows for corrections of baseline
instabilities due to etalon effects. The amount of radicals and ions strongly depends
on the source conditions such as the length of the discharge pulse compared to the
length of the precursor gas pulse and the time delay between them. A pumping unit
with a two stage roots blower keeps the pressure in the chamber at about
5 9 10-2 mbar.
The jet is probed by radiation of tunable BWOs operating between 300 and
1200 GHz. The BWO spectrometer can be used in two different modes, a phase
stabilized mode for highly precise measurements over a small range of about
500 MHz, and in a free running fast scanning mode, which covers a large fre-
quency range of several GHz but with less accuracy. The fast scanning mode is
used for searching absorption lines of little or even completely unknown species,
whereas the phase stabilized mode is used for highly precise measurements
(10 kHz accuracy) on selected, identified lines and for long integration times to
record spectra of weak lines. Instead of BWOs, a synthesizer combined with

frequency multiplier chains can be used to cover most of the frequency range from
100 to 2000 GHz.
The probing THz beam (see Fig. 4.15) is oriented perpendicular to the jet.
Multipass optics increase the path length through the jet by an order of magnitude.
Alternatively, the radiation can be oriented in line with the supersonic jet propa-
gating in a double pass configuration. Figure 4.16 shows the 13 C16 OJ ¼ 4 3
transition at 440.8 GHz recorded with a probing beam parallel to the propagation
of the jet. Two Doppler shifted components appear at 1:25 MHz off the center
frequency. From these data the CO velocity has been calculated to be 850 m/s. The
line width of the CO transition is reduced by a factor of 3 compared to CO
measured in a stagnation cell at 300 K. For phase sensitive detection the BWO
output frequency was modulated with 57 kHz and demodulated by a lock-In
amplifier to record absorption spectra in second derivative.
4.3.2.2 Laser Ablation Sources
An alternative way of producing short-lived molecules in supersonic jets is the

ablation of solid targets by pulsed lasers. A schematic drawing of a laser ablation
source is shown in Fig. 4.17. In our setup light of a Q-switched Nd:YAG laser
operating at either the fundamental mode (1064 nm) or at the 2nd; 3rd, or 4th
harmonic, is focused by a 30-cm focal length CaF2 lens onto the surface of a solid
rod of graphite, silicon, or other target materials. The vaporized products are diluted
in a pulse of helium and collisionally cooled in a supersonic expansion. Highest
yields and stable conditions are usually achieved at Nd:YAG laser power of 10–
40 mJ per pulse at 10 ns pulse duration. Molecules are preferably formed in the
Fig. 4.16 13CO in a

supersonic jet expansion. The
two 340 kHz wide Doppler
components are shifted
by ±1.25 MHz. For
comparison a three times
broader line at the center has
been measured in a cell at
room temperature
Fig. 4.17 The laser ablation

source used for producing C3,
TiO, TiO2, and AlO in
Cologne
high pressure regime of a 6 mm-long reaction channel before adiabatically

expanding in supersonic jet propagation into the vacuum chamber. Largest yields
are usually obtained for helium backing pressures of 2 bar at a background pressure
of 50 lbar in the vacuum chamber. Operation of the pulsed gas valve, timing of the
ablation laser, and signal recording is controlled and synchronized by a TTL timer
unit at 30 Hz repetition rate. Laser ablation sources have been successfully used in
our laboratory to study gas phase spectra of C3 , TiO, TiO2 , and AlO.
4.3.3 OROTRON
The absorption of electromagnetic radiation in a gas probe is significantly

enhanced when resonant cavities are used instead of single pass absorption cells.
The gain in effective optical path length is of the order Q k=2p, where Q is the
cavity quality factor and k is the free-space wavelength. Since Q can be quite
large, cavity spectrometers are very sensitive, which is advantageous for mea-
surements in molecular jets or beams where absorption lengths are rather short.
The major constraints in the application of an external resonant cavity absorption
cell is that the long effective path length occurs only when the microwave source
frequency and the cavity resonant frequency are exactly synchronized. Further-
more, application of source frequency modulation, normally used to increase the
sensitivity, gives rise to baseline fluctuations, which are substantially stronger
compared to those of a simple spectrometer operated with a free space absorption
cell. These technical problems are principally avoided in an intracavity resonator
with a tunable coherent source incorporated [61, 62].
A sensitive intracavity spectrometer was developed for investigation of gener-
ally very weak absorption lines in the millimeter-wavelength region. Key element
of the spectrometer is a tunable oscillator, called OROTRON, which generates

millimeter-wave radiation through the interaction of an electron beam with the
electromagnetic field of a high quality (Q & 104) Fabry–Pérot resonator cavity.
The cavity consists of a movable spherical mirror and a fixed planar mirror with a
periodic structure imprinted on its surface. Focused by the field of strong permanent
magnets the electron beam travels along the periodic structure of the plane mirror,
thus stimulating cavity modes to oscillate. A typical value for the anode voltage is
in the range of 500–1200 V with current of the electron beam of 15–40 mA. A thin
mica foil separates this ultra high vacuum region of the OROTRON from the
probing region. A molecular jet is injected by a pulsed valve into the probing region
perpendicular to the resonator axis. Absorption in the jet attenuates the electro-
magnetic field of the resonator which is sensitively detected by measuring the
electric current of the electron beam. The movable spherical mirror is furnished
with mechanical screws for adjustment of the cavity length and with piezoceramic
transducer units for fine tuning of the resonator in the process of line search and
recording. A small portion of the resonator radiation leaks out through an orifice in
the center of the spherical mirror and is used to determine the resonance frequency
during the measurements. Figure 4.18 shows the setup of the OROTRON.
The OROTRON is a very stable intracavity radiation source which is tunable in
a wide spectral range. The frequency stability dm/m is proportional to the voltage
harmonic spectrum
mixer analyser
OROTRON
MW frequency
synthesizer meter
spherical mirror
mm-wave
radiation nozzle pulse
pump modulation
molecular
beam
pulsed
nozzle
boxcar
integrator
collector
lock-in
cathode
amplifier
plane mirror (anode)
with periodical structure
accelerating modulation PC: data acquisition

voltage ~ 25kHz and processing
Fig. 4.18 Block diagram of an OROTRON spectrometer: a highly sensitive intracavity

spectrometer. (Reproduced from [69], Ó 2003, with permission from Elsevier)
stability dU/U applied to the periodic structure (anode) and scales with the inverse
of the cavity quality factor Q:
dm dU k
2 ; ð4:3:1Þ
m U Q
where k ^ 100 is the number of ridges in the periodic structure. The high quality
of the Fabry–Pérot resonant cavity (typically about 104) leads to high frequency
stability.
At a reasonable voltage stability of 10-5 the frequency stability is dm / m & 10-7
which corresponds to a radiation line width of 10–15 kHz at 100–150 GHz.
Thanks to high Q values the stability of the frequency is not seriously affected by
the instability of the voltage U and therefore active frequency stabilization is not
necessary. In addition simple tunability of the OROTRON spectrometer makes it a
very efficient tool to search for weak transitions of molecules or for highly diluted
trace gases. Initially designed for studies of weakly absorbing stable molecules
[63] and trace gases at concentrations of 0.01–1 ppm in air and water vapor [64], a
most promising new application of the OROTRON is the application to molecular
jet spectroscopy [65]. The potential of the new technique for observation of very
weakly bound van-der-Waals complexes of type R–CO, with R = Ar, Ne, He,
H2 ; D2 ; N2 , and CO have been investigated by Surin and co-workers (see [66] and
references therein). As an example, which demonstrates the sensitivity of the
OROTRON spectrometer, the 107 GHz J ¼ 1 0 transition of 13 C17 O in natural
abundance (4 106 ) is shown in Fig. 4.19a. The observed linewidth of the
hyperfine structure components of 300–400 kHz is caused by Doppler broadening
in a free expanding supersonic jet. The effect of Doppler broadening can partly be
reduced in cases of power saturation. Figure 4.19b shows three saturation dips of
(a) (b)
13 17 17
C O C O
J 0 J 0
5/2-5/2
7/2-5/2
F=3/2-5/2
107286 107288 107290 107292 112358 112359 112360 112361

Frequency, MHz Frequency, MHz
Fig. 4.19 Spectra of 13 C17 O and sub-Doppler spectra of 12 C17 O. (Reproduction with permission
from [65], Ó 2001, American Institute of Physics)
the J ¼ 1 0 transition of 12 C17 O. The hyperfine structure is caused by the 17O

nuclear quadrupole moment [31, 65]. The dips exhibit line-widths of about
100 kHz.
At present three OROTRON tubes operating at 90–130, 110–160, and
130–180 GHz at anode voltages of 350–700, 550–1100, and 700–1300 V,
respectively, are available at Cologne, which have been designed and constructed
in Russia. OROTRONs with higher anode voltages and higher output frequencies
are feasible but the increase of dissipative heat in the periodic structure is a major
concern. The heat load can be reduced when operating the OROTRON in a pulsed
current mode. As a first approach to sub-millimeter wavelengths a pulsed ORO-
TRON with current of 50–100 mA for frequencies up to 380 GHz was developed
in Russia in close collaboration with the Institute of Spectroscopy (Troitsk,
Moscow region) and the Institute for Metrology and Space (Mendeleevo, Moscow
region). Switching the power on and off was accomplished by tuning the low
potential on a grid placed between emitting cathode and periodical structure and
the output pulse power was found to be 20–40 mW [67, 68].
4.3.3.1 Two-Photon-Pump-Probe Experiments
To extend the accessible frequency range of OROTRON jet spectrometers beyond

the millimeter wavelength region and into the microwave range a new double
resonance technique has been developed and successfully applied. The setup uti-
lizes the 90–160 GHz OROTRON as probe beam for signal detection, and an
amplified 18–43 GHz beam of a microwave synthesizer (HP 86350) as pump
beam. The pump beam radiation is emitted from a broadband horn-antenna and
focused by two lenses into the sample region inside the OROTRON cavity. The
molecular jet is oriented perpendicular to both pump beam and probe beam which
intersects the radiation in the middle of the OROTRON cavity. In a first appli-
cation this technique has been used to measure new transitions of the CO dimer in
the microwave frequency range at 18–43 GHz [69]. As an example Fig. 4.20
shows the P(5) transition of the f d subband of the CO-dimer at 38.275 GHz
which is indicated as heavy single ended arrow in the inlet. To detect this new
transition the OROTRON probe beam is set to the previously known Q(4) tran-
sition of the f a subband at 153890.572 MHz (double ended arrow). The
microwave pump radiation is slowly scanned while the OROTRON frequency is
fixed to the Q(4) center frequency. For phase sensitive detection the microwave
output power is amplitude modulated at 50 kHz (square-wave modulation) and the
double resonance signal was recorded in the OROTRON collector circuit by a
lock-in amplifier at 50 kHz reference frequency. Because the jet is pulsed the
output signal of the lock-in has to be recorded with a boxcar integrator synchro-
nized to the pulse of the jet.
From these experiments it becomes evident that this OROTRON-detected
double resonance technique can be extended to higher frequencies provided high
Fig. 4.20 12CO double

resonance experiment. _ (CO)
(Reproduction from [69], f:A 2
J = 4, K = 1
Ó 2004, with permission
from Elsevier)
+
d:A
J = 5, K = 1
+
a:A
J = 4, K = 0
38274.5 38275.5 38276.5

Frequency / MHz
power radiation sources as pumps in the submillimeter wave region

(200–1000 GHz) are available. In case of weak output powers of the pump,
multipass optics or cavity arrangements can be introduced to ensure more effective
coupling to the molecular sample.
4.3.3.2 Two-Photon Absorption Spectroscopy
The possibility of using two-photon absorption in a standing electromagnetic wave

by molecular or atomic gases for removing Doppler broadening of the absorption
lines was first discussed by Vasilenko et al. [70]. The idea of the method is that, for
a particular velocity group of the absorbing particles in a gas, the first order Doppler
shift has an opposite sign for two counterpropagating electromagnetic waves.
Therefore, if one quantum of radiation is absorbed from each of the two counter-
propagating waves, the resulting frequency of the two-photon transition does not
depend on the velocity of the particle; i.e., the Doppler effect will not influence the
absorption linewidth. This effect has been used for Doppler-free laser spectroscopy
in the infrared or optical regions, involving electronic or vibrational transitions of
atoms or molecules (see, for example [71]). Observations of Doppler-free two-
photon absorption lines in the microwave or millimeter wave regions on pure
rotational molecular transitions became possible using highly sensitive intracavity
OROTRON spectrometer. As an example two-photon absorption on rotational
transitions of OCS molecules was measured [72]. A recording of the J ¼ 12 10
transition is shown in Fig. 4.21. The width FWHM of the central peak is about
40 kHz, 7–8 times smaller than the normal Doppler width of a single-photon
absorption line. The center frequencies of the recorded transitions coincide within
Fig. 4.21 OCS two photon

transition with sub-Doppler J=12 OCS
line width. (Reproduced with J = 12 10
permission from [72], Ó 2001
by the American Physical
Society)
J=11
J=10
40 kHz
100kHz
139866,0 139866,2 139866,4 139866,6 139866,8
Frequency, MHz
the estimated uncertainty to the values obtained in the previous Lamb dip and
molecular beam studies.
4.4 Heterodyne Mixers for High Resolution Spectroscopy

of Interstellar Molecules
4.4.1 Heterodyne Detection
To measure doppler shifts and line profiles of interstellar molecular lines, some-
times resolutions as high as 106 are needed. Whereas moderate resolution spec-
troscopy can use direct detectors behind grating- or Fabry–Pérot spectrometers,
instruments with resolutions of 103 or higher typically use the heterodyne prin-
ciple. Heterodyne receivers take the RF signal at mRF picked up by an antenna and
combine it with a, usually much stronger, sinusoidal ‘‘local oscillator’’ signal at
mLO . In a nonlinear device known as mixer, the two frequencies are multiplied,
resulting in, among others, the difference frequency mIF ¼ msig mLO which is
called the intermediate frequency or IF. The IF frequency is an exact replica of the
RF frequency. Because mIF mRF , it is easy to achieve high frequency resolution
at the IF with multichannel spectrum analyzers such as a filter bank, an acousto-

optical spectrometer or a digital Fast Fourier Transform spectrometer.
The frequency- and phase conservation of the heterodyne principle also allows
aperture synthesis interferometry, where the IF output of several receivers can be
cross-correlated, thus allowing a much higher spatial resolution than would be
achievable with a single dish telescope. This is exploited, e.g., in the Atacama
Large Millimeter Array (ALMA).
At frequencies up to the 100 GHz range, it is possible to use a low noise
amplifier behind the antenna, so that simple nonlinear mixing elements such as
semiconductor Schottky diodes can be used. Their rather large noise and low
conversion efficiency thus does not determine the input noise performance of the
system.
However, at frequencies in the mm-wave band up to the FIR, low noise
amplifiers are nonexistent. The mixing element is the first element in the signal
chain, so that its noise performance and conversion efficiency determine the sys-
tem performance. The extreme sensitivity requirements of radio astronomy can
only be achieved using superconducting devices as mixers. Superconductor–Iso-
lator–Superconductor (SIS) tunnel junctions for frequencies in the mm-bands up to
about 1.5 THz and superconducting Hot Electron Bolometer mixers are the
workhorses of astronomical receiver systems today. With their outstanding sen-
sitivities they are the building blocks of cornerstone instruments for submm/THz
spectroscopy such as the HIFI on the Herschel Space Observatory, the ALMA or
the Stratospheric Observatory for the Far-Infrared (SOFIA).
After an early phase using Schottky diode mixer receivers at the KOSMA
Gornergrat observatory in the mm-wavelength bands, KOSMA has started to
develop and fabricate superconducting mixers in the late 1980s and has used them
subsequently in all receivers operated in the observatory, including a 490/880 GHz
dual color multi-pixel receiver, now in operation at the NANTEN2 observatory
(Atacama).Building on this expertise, we were selected to contribute one of the
mixer channels for HIFI on the Herschel Space Observatory. We also developed
SIS and Hot-Electron Bolometer (HEB) mixers for several other projects, such as
AST/RO at South Pole Station, the GREAT instrument for SOFIA, the SMA
interferometer, the Stratospheric Terahertz Observatory and atmospheric physics
projects.
A good overview over the field of superconducting detectors can also be found
in [73].
4.4.2 SIS Mixers
4.4.2.1 Physical Principles and Frequency Limits
The characteristic energy scale of (low-temperature) superconductivity is in the

few meV range, corresponding to frequencies in the submillimeter-wave or
Terahertz range (at 1 THz, hm = 4 meV). It can therefore be expected that

superconductors can be used for detection of photons in this energy range.
Superconductor–Insulator–Superconductor mixers use the sharp current nonlin-
earity of quasiparticle excitations (approximately equivalent to normal conduction
electrons) tunneling across an insulating barrier between the two superconductors
(Fig. 4.22).
Biasing one side of the tunnel junction to the ‘‘gap voltage’’ 2D=e results in a
current flow across the barrier. Due to the singularity in the density of states at the
energy gap of the superconductor, the nonlinearity is almost as sharp as a switch,
which would be the ideal frequency mixer. Under irradiation with photons,
quasiparticles can gain enough energy to enable tunneling even at bias voltages
below the gap voltage. This process is called photon assisted tunneling and leads to
the characteristic ‘‘photon steps’’ in the DC I/V curve. Figure 4.23 shows a typical
example of the irradiated or ‘‘pumped’’ DC-I/V curve of an SIS tunnel junction
mixer at 630 GHz together with the IF output power response to hot (296 K) and
cold (77 K) blackbody radiation.
The fact that photon quanta become visible in a simple DC tunneling I/V is a
remarkable feature. Mixers showing this behaviour, i.e., where hm/e is larger than
the voltage scale of their DC nonlinearity, are called quantum mixers. As it turns
out, such a mixer can reach the quantum limit of heterodyne detection sensitivity
given by the Heisenberg uncertainty relation between amplitude and phase.
An additional advantage of SIS mixers is their rather low local oscillator power
requirement. This is especially true for the submillimeter/Terahertz range where
local oscillator power is very scarce. The required LO power can be roughly
estimated by PLO ¼ ðhm=eÞ2 =RN , with RN as the normal resistance of the tunnel
junction at bias voltages beyond the gap. With RN of 10–50 X; typical LO powers
are in the few hundred nanowatt range. Note that the required power scales with m2.
The gap energy 2D, which is the binding energy of the Cooper pairs, is about
Fig. 4.22 ‘‘Semiconductor’’

model of the density of states
in SIS
Fig. 4.23 SIS junction under

photon irradiation and mixer
output power
2.8 meV for Niobium, a standard superconductor for SIS mixers, corresponding to
a ‘‘gap frequency’’ of 2D=h ¼ 700 GHz. A fundamental frequency limit for the
tunneling process, and therefore for frequency mixing with SIS elements, occurs at
frequencies above twice the gap frequency at 4D=h, 1400 GHz for Niobium. At
this frequency, the photon energy is high enough to produce equal tunneling
currents in both directions of the barrier for all DC bias voltages, effectively
canceling each other. It turns out in practice that above 1200 GHz the DC-bias
range for mixer operation is getting too small for practical mixers. This fre-
quency limit can only be surpassed with tunnel junctions made of superconductors
other than Niobium with higher energy gaps, corresponding to higher critical
temperatures.
The intrinsic mixing properties of an SIS mixer are described by the Quantum
Theory of Mixing introduced by John Tucker in 1979 [74, 75]. This theory in its
three-frequency approximation turns out to be a very powerful instrument for the
practical design of SIS mixers because it only needs the DC I/V characteristic and
the embedding impedance of the junction at the signal-, local oscillator- and IF to
completely predict its RF properties. All higher harmonics can usually be safely
ignored due to the high shunting capacitance of the junction. The theory takes the
non-classical response of the tunnel currents to external fields into account, pre-
dicting the mixer gain and input noise. The non-classical effects include negative
differential resistance with non-classical mixer gain [1 and the possibility of zero
added mixer noise, thus achieving quantum limited heterodyne detection with a
minimum noise temperature of the order of hm/k, corresponding to 4.8 K/100 GHz.
The theory does not address any effects from the tunneling of the Cooper pairs
(Josephson effects), effectively expecting their complete suppression by an external
magnetic field.
As will be shown below, practical SIS mixers have to use high-Q integrated RF
transmission line matching circuits to exploit their full sensitivity. Up to fre-
quencies close to the gap frequency, the same superconductors as for the tunnel
junctions can be used as lossless conductors for these transmission lines.
Above, the photon energy is high enough to destroy the superconductivity, giving
rise to large losses in the now normal-conducting lines. The loss in the junction
electrodes themselves can become important if the junction area cannot be
assumed to be small against the wavelength in the junction so that the junction
itself is a lossy transmission line. These phenomena lead to the fact that Niobium-
based SIS mixers have a strong decrease in sensitivity above about 750 GHz and
other materials have to be used for the circuits above this frequency.
4.4.3 SIS Mixer Device Design and Fabrication
4.4.3.1 SIS Tunnel Junctions
After the first experience with Pb–Oxide–Pb (lead) based SIS junctions (a spin-off
of the IBM superconducting computer project) in the early 1980s [76], Nb junc-
tions were developed to solve the problems of the mechanically and chemically
unstable lead junctions. The initially achieved tunnel characteristics however were
disappointingly rounded and leaky due to the unfavorable characteristics of
Niobium oxide as the insulating barrier. The discovery of Gurvitch et al. [77] that
a very thin layer of aluminum perfectly wets the Niobium surface and can serve as
an artificial tunneling barrier after oxidation was a breakthrough and resulted in
nearly ideal Nb–Al–Al2 O3 –Nb–SIS junctions with extremely low leakage currents
and high current densities. Figure 4.24 shows a cross section of a typical SIS
tunnel junction.
High quality Nb thin film layers are fabricated by sputtering from a high purity
target. The background pressure must be in the low 10-8 mbar range to avoid the
incorporation of oxygen which reduces the critical temperature. This dictates the
need for a load-locked vacuum system. Care has also to be taken that the Nb
surface has to cool down before sputtering the thin barrier aluminum layer to avoid
diffusion into the grain boundaries of the Nb film. Thermal oxidation produces a
tunnel barrier of Al2 O3 . The thickness and therefore the tunnel current density is
approximately proportional to the product of oxygen pressure and exposure time.
After blanket-depositing the tunnel layers onto the substrate, the (sub)-micron
sized tunnel junctions are defined by either photolithography or direct electron
beam writing, followed by an anisotropic Reactive Ion Etching process using
reactive gases such as SF6 in a plasma discharge. Silicon monoxide or silicon
Fig. 4.24 Cross section of an

SIS tunnel junction
dioxide layers are evaporated or sputtered to insulate the top wiring layer con-
necting the tunnel junctions to the circuit. The insulation layer also serves as the
dielectric layer for integrated RF circuits described below.
4.4.3.2 RF Properties of SIS Junctions
To exploit the quantum limited sensitivity of the tunnel junctions, they have to be
very effectively coupled to the radiation of the telescope. As the typical tunnel
junction size is 1 lm, which is much smaller than the wavelength being received,
an antenna is needed. One approach is the conversion of the free space wave from,
e.g., a telescope to a guided wave using a horn connected to a single-mode
waveguide. A probe then couples the waveguide mode into a thin-film transmis-
sion line on the detector substrate. At sub-millimeter wavelengths, the challenge is
that the device substrate has to be ultra-thin, and only a few groups have the
technology available. The second widely used approach is to integrate the detector
into a planar antenna on the same chip and use a dielectric lens to focus the
radiation onto the antenna. The advantage is that the device substrate can now be
thick, but the coupling to a Gaussian beam is limited in efficiency, whereas cor-
rugated horns can reach 98%.
A major complication to the impedance match of an SIS tunnel junction to
either the waveguide probe or the planar antenna impedance is their large intrinsic
capacitance due to the very thin barrier acting as a dielectric between the two
superconductors. Typical values for Nb-based junctions are 80–90 fF/lm2 ,
effectively shorting out the tunnel resistance even for micron-sized junctions. The
capacitance has to be compensated at the operating frequency by an external
circuit. This limits the intrinsic RF bandwidth of the detector to about
B ¼ 2=xRN CJ , with RN the tunnel junction normal resistance, x the operating
angular frequency, and CJ the junction capacitance. Note that the RN C product is
independent of junction size but is inverse proportional to the current density of
the junction, i.e., for large bandwidths one needs high current density devices. As
RN is approximately the RF impedance of the nonlinear junction characteristic, the
requirement to have RN in a range of 10–50 X for easy matching to waveguides or
other transmission lines leads to junction sizes of microns or submicrons. The
present state of the art allows producing junctions of 0.6 lm2 to 0.25 lm2 with
current densities up to 15 kA/cm2 with Aluminum Oxide as the barrier. At 1 THz,
this translates into a fractional bandwidth of 25% with RN = 20–60 X. Higher
current densities can be achieved with Aluminum Nitride (AlN) barriers.
4.4.3.3 Integrated Matching Circuits
Early experiments with SIS mixers used waveguide circuits with tunable backshorts
to compensate the capacitance. This approach leads to low coupling efficiencies at
submm wavelengths due to the waveguide and backshort losses, which are severe
due to the high-Q of the circuit. The breakthrough for SIS junctions at submm
wavelengths came with the integration of superconducting transmission lines,
mostly as microstrip lines, as resonant and/or impedance-transforming elements
together with the tunnel junctions. To meet the desired impedance match conditions
between junction and waveguide of course requires detailed knowledge about the
superconducting transmission lines [78]. The determining parameter is the complex
surface impedance of the superconductor, which can be calculated with the Mattis–
Bardeen theory [79, 80]. The imaginary part of the surface impedance contains a
‘‘kinetic’’ inductance part (caused by the pair current component), the real part
corresponds to power loss. Below the gap frequency of the superconductor, the
losses are extremely small, allowing to design the necessary high-Q circuits with
negligible loss. This is the main reason for the very low noise performance of SIS
receivers up to about 700 GHz. Figure 4.25 shows a microscope photograph of a
typical junction circuit with an integrated tuning structure developed at KOSMA
[81], consisting of a series inductive line next to the junction followed by two k/4
impedance transformer sections.
The SiO2 dielectric layer for the Nb–SiO2 –Nb microstrip line can be the same
as the junction isolation layer, adding no processing steps.
Above the gap frequency of the superconductor, the surface resistance and
subsequently the RF loss increases dramatically, so that even conventional normal
conductors are less lossy than the superconductor. The crossover point is slightly
higher than the gap frequency due to the higher phase velocity in normal metal
microstrip line, leading to longer line lengths. Figure 4.26 shows calculations of
the power loss in a quarter wavelength microstrip transmission line for several
combinations of Nb, NbTiN and Al as studied for the Herschel HIFI band-2 mixer.
It can be seen that Niobium deteriorates above 650 GHz and has a crossover with
Fig. 4.25 Microphotograph

of SIS mixer circuit
Fig. 4.26 Transmission loss

in superconducting microstrip
lines
the normal metal Aluminum at about 800 GHz. In principle, NbTiN films of
practical quality would allow lossless tuning structures up to 1150 GHz. Above, a
good normal conductor (at 4.2 K) such as Aluminum has to be used.
Very important for low-noise operation of SIS mixers is the complete sup-
pression of the AC-Josephson effects with a (preferably superconducting) magnet
producing a flux of at least one flux quantum in the barrier (0.014 T for a 1 lm2
Nb junction). The AC-Josephson effect produces noise in the vicinity of the
Shapiro steps, occurring at bias voltages of nhm/2e corresponding to harmonics of
the local oscillator frequency. Especially at submillimeter wavelengths, the first
order step is within the bias voltage range of the first photon step, where optimum
mixer performance is usually achieved. Incomplete suppression leads to a smaller
range of available bias voltages and/or additional noise. Trapping of external flux
can be a source of instability.
4.4.3.4 Waveguide Mixers
In the early attempts to match to the small impedance of the capacitively shunted
junctions, waveguide mixers with two adjustable mechanical backshorts had to be
used. For applications in radio astronomy, especially for remote observing or space
instruments, this is very inconvenient, and mechanical elements operating at 4 K
tend to be unreliable. With the incorporation of the junctions into integrated
impedance matching structures, it became possible to use only one backshort, and
with the modern three-dimensional electromagnetic simulation software, it is
possible to design mixers with a fixed backshort for instantaneous RF bandwidths
of 35%. Figure 4.27 shows a rendered drawing of the inside of a typical waveguide
mixer block used at KOSMA.
The backshort cavity behind the mixer device is stamped into a copper block
using a steel stencil. The junction is mounted in a channel across the waveguide,
and the horn antenna is flanged to the front surface. The mixer also incorporates
the superconducting magnets to suppress the unwanted Josephson current. As a

further example, the mixer developed at KOSMA for the Herschel space obser-
vatory is shown in Fig. 4.28. Several versions of such fixed-tuned broadband
mixers with excellent noise performance in the 345, 490, 660 and 800 GHz bands
have been built and have been successfully used as mixers in receivers for ground-
based astronomical telescopes such as the Gornergrat Observatory or AST/RO at
the Amundsen–Scott South Pole Station.
The integration of matching circuits with the SIS tunnel junction resulted in a
breakthrough in performance over the entire mm- and submm wavelength range.
The receivers in today’s instruments (Herschel, ALMA, NANTEN, APEX, etc.)
typically achieve noise temperatures between 3 and 5 hm/k up to the Niobium gap
frequency, increasing to 10–15 hm/k at 1200 K.
Fig. 4.27 SIS waveguide

mixer
Fig. 4.28 SIS waveguide mixer for Band 2 on HIFI/Herschel

Typical local oscillator sources are Schottky diode frequency multiplier chains,
multiplying either a Gunn oscillator source signal at around 100 GHz or the power
of a 10–20 GHz synthesizer multiplied to about 100 GHz, where power amplifiers
are recently available. These chains now operate up to 1900 GHz with output
powers in the few hundred microwatts at low submillimeter wavelengths to less
than 10 lW at THz frequencies. THz sources under development are laser
photomixers and quantum cascade lasers.
With single pixel mixers operating close to their theoretical noise limit, an
effective increase in sensitivity can only be achieved with a combination of
multipixel ‘‘array’’ receivers, an increase of the instantaneous IF bandwidth, and
adding more functionality to the mixer circuit, such as balanced mixers or side-
band separating mixers. Smaller arrays such as the KOSMA/NANTEN dual color
receiver for 490/880 GHz (Sub-Mm Array Receiver for Two frequencies) can be
built by essentially stacking the available single pixel detectors. Present devel-
opments concentrate on introducing new concepts such as SIS mixers on Silicon
membranes with integrated beam lead contacts, making the assembly of mixers
substantially easier. Figure 4.29 shows a micrograph of such a novel SIS mixer
fabricated at KOSMA.
Achieving larger instantaneous IF bandwidth than the present 4 or 8 GHz is
especially desirable at THz frequency because of the increased linewidths, and
certainly for the large Doppler linewidth of extragalactic sources. Matching
between the SIS mixer and the first cryogenic amplifier is the main obstacle here.
A solution needs low-inductance contacting (beam leads) and possibly the inte-
gration of mixer and first amplifier stage.
Fig. 4.29 SIS mixer device

on Silicon membrane with
beam lead contacts
Another field of present research is pushing the limit of SIS mixers to at least
2 THz, beyond the demonstrated present limit of 1200 GHz. SIS mixers at THz
have the advantage of better sensitivity, wider instantaneous IF bandwidth and
better power stability over their only available alternative at the low THz fre-
quency band: superconducting Hot Electron Bolometers (see below). This devel-
opment mainly needs good quality high-current density SIS tunnel junctions where
at least one of the Nb electrodes is replaced by a compatible material with higher
critical temperature, such as NbN or NbTiN. There is little hope that materials can
be found that could push the limit higher. The high-Tc superconductors found until
today cannot be used to make good hysteretic tunnel barrier junctions, and also
their RF properties at THz are not favourable.
4.4.4 Superconducting Hot-Electron Bolometers
Superconducting HEBs rely on the electron heating and the resulting phase tran-
sition to normal conductor of a small superconducting bridge. With the device
biased at the transition temperature, the local oscillator and signal power heat up
the device. The beat frequency between the two signals consequently causes a
large resistance variation which causes a current at the difference frequency at
constant bias. The HEB mixer is therefore an example of a ‘‘square law’’ mixer,
responding to the square (i.e., power) of the sum of the incident fields but unable to
follow the instantaneous field. The speed or maximum IF of a bolometer is
determined by the heat capacitance and the thermal conductance of the cooling
mechanism. Superconducting bolometers can be very fast, as the radiation power
is directly absorbed by the electrons in the superconductor rather than in a separate
absorber. Using very thin (\ 10 nm) films with the superconductor in the ‘‘dirty
limit’’, only the electron gas is heated up and the heat capacitance is low. Also, the
devices are very small with a device length in the 100 nm range.
In the initial proposals by Gershenzon et al. [82] cooling by phonons (i.e., via the
substrate to the thermal bath) was assumed as the cooling mechanism. For Nb, the
electron–phonon relaxation time is in the nanosecond (100 MHz) range, which is not
fast enough for astronomical applications. However, materials such as Niobium
Nitride (NbN) have small enough electron–phonon and phonon escape time con-
stants, so that IF’s up to several GHz became possible. In 1993, D. Prober proposed a
new type of hot-electron bolometer [83], which uses the diffusion of the hot electrons
into a large (normal conducting) heat sink at the ends of the microbridge as cooling
mechanism. This avoids the slow electron–phonon relaxation time and leads to very
fast bolometers if the bridge is short enough. The IF frequency scales as (length)-2
and is about 2 GHz for a 200 nm long bridge. IF bandwidths up to 8 GHz have been
demonstrated at 20 GHz LO frequency, but practical difficulties in biasing the
devices stably made the phonon-cooled variant more successful up to date.
Using only the heating of the electron gas in the bridge, the mixing response is
not limited by the superconductor gap frequency, so that HEB mixers can be used
far into the THz range. An enormous advantage from the RF engineering view-
point is that the device is almost purely resistive, thus not requiring any high-
Q (lossy) tuning circuit. Also, there is no need for a magnetic field. The local
oscillator power requirement is below 1 lW coupled power and is not frequency
dependent. At these power levels, solid state local oscillator sources become
feasible even at THz frequencies.
The dominant noise source of the device is the thermal fluctuation noise, a
smaller contribution is from Johnson noise of the device resistance. As the thermal
fluctuation noise has the same roll-off frequency as the mixer gain (thermal
response time), the noise bandwidth is actually larger than the gain bandwidth. The
input noise temperature of the device is frequency independent except for the
quantum limit.
Practical mixers have reached a noise temperature of less than 1000 K at
1.9 THz with an IF bandwidth of over 3 GHz. This can be regarded as a break-
through for high-resolution spectroscopy in the Far Infrared region which came
just in time for the new observatories such as SOFIA or Herschel. Although most
groups have to rely on the technologically less demanding quasioptical techniques
for coupling the radiation to the bolometer, waveguide technology with its
advantages in coupling to a Gaussian beam can even be used at 1.9 THz, if
micromachining technology is available. Figure 4.30 shows a photograph of a
1.9 THz mixer using a NbTiN Hot Electron Bolometer on a Silicon Nitride
membrane fabricated at KOSMA [84].
The waveguide dimension is only 60 9 120 lm. The device is mounted in
‘‘flip-chip’’ technique with the device facing the waveguide, so that it is seen
through the almost transparent Silicon Nitride substrate. The bolometer size is
0.4 9 4 lm and is fabricated with electron-beam lithography. A similar mixer for
1.4 THz was successfully used at the APEX telescope (MPIfR Bonn, Atacama) to
take astronomical spectra. Similar mixers for 1.4 and 1.9 THz were delivered to
the German receiver at terahertz frequencies (GREAT), which is waiting to be
flown on the SOFIA aircraft starting in 2011.
Fig. 4.30 1.9 THz HEB mixer for SOFIA

References
1. T. Amano, Astrophys. J. 716, L1 (2010)

2. E. Falgarone, V. Ossenkopf, M. Gerin, P. Lesaffre, B. Godard, J. Pearson, S. Cabrit,
Ch. Joblin, A.O. Benz, F. Boulanger, A. Fuente, R. Güsten, A. Harris, T. Klein, C. Kramer,
S. Lord, P. Martin, J. Martin-Pintado, D. Neufeld, T.G. Phillips, M. Röllig, R. Simon,
J. Stutzki, F. van der Tak, D. Tryssier, M. Yorke, N. Erickson, M. Fich, W. Jellema,
A. Marston, C. Risacher, M. Salez, F. Schmülling, Astron. Astrophys. 518, L118 (2010)
3. T. de Graauw et al. (with 165 authors), Astron. Astrophys. 518, L6 (2010)
4. O. Asvany, O. Ricken, H.S.P. Müller, M.C. Wiedner, T.F. Giesen, S. Schlemmer, Phys. Rev.
Lett. 100, 233004 (2008)
5. H.M. Pickett, R.L. Poynter, E.A. Cohen, M.L. Delitsky, J.C. Pearson, H.S.P. Müller,
J. Quant. Spectrosc. Radiat. Transf. 60, 883 (1998)
6. H.S.P. Müller, F. Schlöder, J. Stutzki, G. Winnewisser, J. Mol. Struct. 742, 215 (2005)
7. R. Güsten, L.Å. Nyman, P. Schilke, K. Menten, C. Cesarsky, R. Booth, Astron. Astrophys.
454, L13 (2006)
8. A. Staprans, E.W. McCune, J.A. Ruetz, Proc. IEEE 61, 299 (1973)
9. C.P. Endres, B.J. Drouin, J.C. Pearson, H.S.P. Müller, F. Lewen, S. Schlemmer, T.F. Giesen,
Astron. Astrophys. 504, 635 (2009)
10. P.H. Siegel, IEEE Trans. Microw. Theory Tech. 50, 910 (2002)
11. B. Levush, T. Antonsen Jr., A. Bromborsky, W.R. Lou, D. Abe, S. Miller, Y. Carmel,
J. Rodgers, V. Granatstein, W. Destler, in Electron Devices Meeting 1991, Technical Digest
(1991), p. 775
12. B. Levush, T.M. Anonsen, A. Bromborsky, W.R. Lou, Y. Carmel, IEEE Trans. Plasma Sci.
20, 263 (1992)
13. A.N. Korolev, S.A. Zaitsev, I.I. Golenitskij, Y.V. Zhary, A.D. Zarkurdayev, M.I. Lopin, P.M.
Meleshkevich, E.A. Gelvich, A.A. Negirev, A.S. Podedonostsev, V.I. Poognin, V.B. Homich,
A.N. Gargin, IEEE Trans. Electron. Devices 48, 2929 (2001)
14. M. Liedtke, R. Schieder, K.M.T. Yamada, G. Winnewisser, S.P. Belov, A.F. Krupnov,
J. Mol. Spectrosc. 161, 317 (1993)
15. S.P. Below, M. Liedtke, T. Klaus, R. Schieder, A.H. Saleck, J. Behrend, K.M.T. Yamada,
G. Winnewisser, A.F. Krupnov, J. Mol. Spectrosc. 166, 489 (1994)
16. G.M. Plummer, G. Winnewisser, M. Winnewisser, J. Hahn, K. Reinartz, J. Mol. Spectrosc.
126, 255 (1987)
17. M. Bester, K. Jacobs, B. Vowinkel, in Proceedings of 13th European Microwave Conference,
Nürnberg (1983), p. 308
18. A.F. Krupnov, L.I. Gershtein, Prib. Tekh. Eksp. 6, 143 (1970)
19. P. Goy, C. Fabre, M. Gross, S. Haroche, J. Phys. B 13, L83–L91 (1980)
20. A.H.F. van Vliet, T. de Graauw, S. Lindholm, H. van de Stadt, Int. J. Infrared Millimeter
Wave 3, 817 (1982)
21. P. Goy, J.M. Raimond, G. Vitrant, S. Haroshe, Phys. Rev. A 26, 2733 (1982)
22. E. Bava, A. Godone, G.D. Rovera, Infrared Phys. 23, 157 (1983)
23. G. Winnewisser, A.F. Krupnov, M.Yu. Tretyakov, M. Liedtke, F. Lewen, A.H. Saleck,
R. Schieder, A.P. Shkaev, S.V. Volokhov, J. Mol. Spectrosc. 165, 294 (1994)
24. G. Winnewisser, Vib. Spectrosc. 8, 241 (1995)
25. M. Bogey, S. Civi, B. Delcroix, C. Demuynck, A.F. Krupnov, J. Quiguer, M.Yu. Tretyakov,
A. Walters, J. Mol. Spectrosc. 182, 85 (1997)
26. I. Morino, M. Fabian, H. Takeo, K.M.T. Yamada, J. Mol. Spectrosc. 185, 142 (1997)
27. T. Amano, Astrophys. J. 531, L161 (2000)
28. F. Lewen, R. Gendriesch, I. Pak, D.G. Paveliev, M. Hepp, R. Schieder, G. Winnewisser, Rev.
Sci. Instrum. 69, 32 (1998)
29. A.F. Krupnov, Int. J. Infrared Millimeter Waves 22, 1 (2001)
30. S. Schlemmer, T. Giesen, F. Lewen, G. Winnewisser, in Frontiers of Molecular Spectroscopy,

ed. by J. Laane (Elsevier, Amsterdam, 2009), p. 241
31. G. Klapper, L. Surin, F. Lewen, H.S.P. Müller, I. Pak, G. Winnewisser, Astrophys. J. 582,
262 (2003)
32. S. Brünken, E.A. Michael, F. Lewen, T. Giesen, H. Ozeki, G. Winnewisser, P. Jensen,
E. Herbst, Can. J. Chem. 82, 676 (2004)
33. U. Fuchs, S. Brünken, G.W. Fuchs, S. Thorwirth, V. Ahrens, F. Lewen, S. Urban, T. Giesen,
G. Winnewisser, Z. Naturforsch. 59, 861 (2004)
34. S. Brünken, U. Fuchs, F. Lewen, S. Urban, T. Giesen, G. Winnewisser, J. Mol. Spectrosc.
225, 152 (2004)
35. P. Siegel, R. Smith, M. Graidis, S. Martin, IEEE Trans. Microw. Theory Tech. 47, 59 (1999)
36. J. Bruston, E. Schlecht, A. Maestrini, F. Maiwald, S. Martin, R. Smith, I. Mehdi, P. Siegel,
J. Pearson, in Proceedings of the SPIE-International Symposium on Astronomical Telescope
and Instrumentation (2000), p. 22
37. G. Chattopadhyay, E. Schlecht, J. Ward, J. Gill, H. Javadi, F. Maiwald, I. Medhi, IEEE Trans.
Microw. Theory Tech. 52, 1538 (2004)
38. A. Maestrini, J. Ward, J. Gill, H. Javadi, E. Schlecht, G. Chattopadhyay, F. Maiwald,
N. Erickson, I. Mehdi, IEEE Microw. Wirel. Compon. Lett. 14, 253 (2004)
39. F. Maiwald, S. Martin, J. Bruston, A. Maestrini, T. Crawford, P.H. Siegel, in Microwave
Symposium Digest, 2001 IEEE MTT-S International, vol. 3 (2001), p. 1637
40. H. Xu, G.S. Schoenthal, J.L. Hesler, T.W. Crowe, R.M. Weikle, IEEE Trans. Microw. Theory
Tech. 55, 648 (2007)
41. B.J. Drouin, F.W. Maiwald, J.C. Pearson, Rev. Sci. Inst. 76, 093113 (2005)
42. F. Lewen, D.G. Paveljev, B. Vowinkel, J. Freyer, H. Grothe, G. Winnewisser, in Proceedings
of 4th International Workshop on Terahertz Electronics, Erlangen (1996)
43. F. Maiwald, F. Lewen, B. Vowinkel, W. Jabs, D.G. Paveljev, M. Winnewisser, G. Winnewisser,
IEEE Microw. Guided Wave Lett. 9, 198 (1999)
44. F. Maiwald, F. Lewen, V. Ahrens, M. Beaky, R. Gendriesch, A.N. Koroliev, A.A. Negirev,
D.G. Paveljev, B. Vowinkel, G. Winnewisser, J. Mol. Spectrosc. 202, 166 (2000)
45. C.P. Endres, F. Lewen, T.F. Giesen, S. Schlemmer, D.G. Paveliev, Y.I. Koschurinov,
V.M. Ustinov, A.E. Zhucov, Rev. Sci. Instrum. 78, 043106 (2007)
46. G. Piau, F.X. Brown, D. Dangoisse, P. Glorieux, IEEE J. Quantum Electron. QE-23, 1388
(1987)
47. G.A. Blake, K.B. Laughlin, R.C. Cohen, K.L. Busarow, D.H. Gwo, C.A. Schmuttenmaer,
D.W. Steyert, R.J. Saykally, Rev. Sci. Inst. 62, 1701 (1991)
48. P. Verhoeve, E. Zwart, M. Versluis, M. Drabbels, J.J. ter Meulen, W.L. Meerts, A. Dymanus,
D.B. Mclay, Rev. Sci. Inst. 61, 1612 (1990)
49. F. Lewen, E. Michael, J. Stutzki, G. Winnewisser, J. Mol. Spectrosc. 183, 207 (1997)
50. E. Michael, F. Lewen, R. Gendriesch, J. Stutzki, G. Winnewisser, Int. J. Infrared Millimeter
Waves 20, 1073 (1999)
51. R. Gendriesch, F. Lewen, G. Winnewisser, J. Hahn, J. Mol. Spectrosc. 203, 205 (2000)
52. R. Gendriesch, F. Lewen, G. Winnewisser, H.S.P. Müller, J. Mol. Struct. 599, 293 (2001)
53. R. Gendriesch, K. Pehl, T. Giesen, G. Winnewisser, F. Lewen, Z. Naturforsch. 58, 129 (2003)
54. S. Brünken, H.S.P. Müller, F. Lewen, G. Winnewisser, Phys. Chem. Chem. Phys. 5, 1515
(2003)
55. S. Brünken, H.S.P. Müller, S. Thorwirth, F. Lewen, G. Winnewisser, J. Mol. Struct. 780(81),
3 (2006)
56. H. Linnartz, T. Motylewski, F. Maiwald, D.A. Roth, F. Lewen, I. Pak, G. Winnewisser, Chem.
Phys. Lett. 292, 188 (1998)
57. M. Hepp, R. Gendriesch, I. Pak, Y.A. Kuritsyn, F. Lewen, G. Winnewisser, M. Brookes,
A.R.W. McKellar, J.K.G. Watson, T. Amano, Mol. Phys. 92, 229 (1997)
58. T. Imasaka, D.S. Moore, T. Vo-Dinh, Pure Appl. Chem. 75, 975 (2003)
59. T. Motylewski, O. Vaizert, T.F. Giesen, H. Linnartz, J.P. Maier, J. Chem. Phys. 111, 6161
(1999)
60. T.F. Giesen, A. Van Orden, H.J. Hwang, R.S. Fellers, R.A. Provencal, R.J. Saykally, Science
265, 756 (1994)
61. B.S. Dumesh, V.P. Kostromin, F.S. Rusin, L.A. Surin, Meas. Sci. Technol. 3, 873 (1992)
62. B.S. Dumesh, L.A. Surin, Rev. Sci. Instrum. 67, 3458 (1996)
63. L.A. Surin, Vib. Spectrosc. 24, 147 (2000)
64. B.S. Dumesh, V.D. Gorbatenkov, V.G. Koloshnikov, V.A. Panfilov, L.A. Surin, Spectrochim.
Acta A 53, 835 (1997)
65. L.A. Surin, B.S. Dumesh, F. Lewen, D.A. Roth, V.P. Kostromin, F.S. Rusin, G. Winnewisser,
I. Pak, Rev. Sci. Instrum. 72, 2535 (2001)
66. A.V. Potapov, L. Surin, V.A. Panfilov, B.S. Dumesh, Opt. Spectrosc. 106, 183 (2009)
67. V.L. Bratman, B.S. Dumesh, A.E. Fedotov, Yu.A. Grishin, F.S. Rusin, Int. J. Infrared
Millimeter Waves 23, 1595 (2002)
68. Yu. Grishin, M. Fuchs, A. Schnegg, A. Dubinski, B. Dumesh, F. Rusin, V. Bratman,
K. Mobius, Rev. Sci. Instrum. 75, 2928 (2004)
69. L.A. Surin, D.N. Fourzikov, F. Lewen, B.S. Dumesh, G. Winnewisser, A.R.W. McKellar,
J. Mol. Spectrosc. 222, 93 (2003)
70. L.S. Vasilenko, V.P. Chebotaev, A.V. Shishaev, Pis’ma, Zh. Eksp. Teor. Fis 12, 161 (1970)
71. V.S. Letokhov, V.P. Chebotaev, Springer Series in Optical Sciences, vol. 4 (Springer, Berlin,
1970)
72. L.A. Surin, B.S. Dumesh, F.S. Rusin, G. Winnewisser, I. Pak, Phys. Rev. Lett. 86, 2002
(2001)
73. J. Zmuidzinas, P.L. Richards, Proc. IEEE 92, 1597 (2004)
74. J.R. Tucker, IEEE J. Quantum Electron. QE-15(11), 1234 (1979)
75. J.R. Tucker, M.J. Feldman, Rev. Mod. Phys. 57(4), 1055 (1985)
76. K.-H. Gundlach, S. Takada, M. Zahn, H.J. Hartfuss, Appl. Phys. Lett. 41, 294 (1982)
77. M. Gurvitch, M.A. Washington, H.A. Huggins, Appl. Phys. Lett. 42, 472 (1983)
78. R.L. Kautz, J. Appl. Phys 49(1), 308 (1978)
79. D.C. Mattis, J. Bardeen, Phys. Rev. 111, 412 (1958)
80. R. Pöpel, J. Appl. Phys. 66, 5950 (1989)
81. K. Jacobs, U. Kotthaus, B. Vowinkel, Int. J. Infrared Millimeter Waves 13, 15 (1992)
82. E.M. Gershenzon, M.E. Gershenzon, G.N. Goltsman, A.D. Semenov, A.V. Sergeev, IEEE
Trans. Magn. 27, 1317 (1991)
83. D.E. Prober, Appl. Phys. Lett. 62, 2119 (1993)
84. P.P. Munoz, S. Bedorf, M. Brandt, T. Tils, N. Honingh, K. Jacobs, IEEE MWCL 16(11), 606
(2006)
85. S.P. Belov, F. Lewen, T. Klaus, G. Winnewisser, J. Mol. Spectrosc. 174, 606 (1995)
86. R. Gendriesch, F. Lewen, G. Klapper, K.M. Menten, G. Winnewisser, J.A. Coxon,
H.S.P. Müller, Astron. Astrophys. 497, 927 (2009)
87. M. Caris, T.F. Giesen, C. Duan, H.S.P. Müller, S. Schlemmer, K.M.T. Yamada, J. Mol.
Spectrosc. 253, 99 (2009)
Chapter 5
List of Molecules Observed in Interstellar
Space
Koichi M. T. Yamada and Gisbert Winnewisser
By the end of July 2010, the number of observed interstellar molecules listed in
CDMS (Cologne Database for Molecular Spectra)1 has reached 163.
Many of them are organic molecules. Among them 27 molecules are composed
only of carbon and hydrogen, which are listed in Table 5.1. Table 5.2 lists 51
organic molecules, which contain one hetero atom. Although the populations of
hetero-atoms are much less than that of the carbon atom in the universe, organic
molecules which contain two hetero-atoms have been detected (14 molecules) in
the interstellar space as listed in Table 5.3.
In addition to those organic molecules, fairly many inorganic molecules have
been detected in the interstellar space (Tables 5.4, 5.5, 5.6). Among them 30
molecules contain one or more carbon atoms, as listed in Table 5.4. Table 5.5 lists
23 inorganic molecules which contain one or more hydrogen atoms so far detected.
The rest 18 molecules contain neither a carbon nor a hydrogen atom.
K. M. T. Yamada (&)
National Institute of Advanced industrial Science and Technology (AIST), EMTech,
Onogawa 16-1, Tsukuba, Ibaraki 305-8569, Japan
G. Winnewisser
Alte Str. 8A, 79249, Merzhausen, Germany
1
http://www.astro.uni-koeln.de/cdms/.
220
Table 5.1 Organic molecule composed of carbon and hydrogeny

2 atoms 3 atoms 4 atoms 5 atoms 6 atoms 7 atoms 8 atoms 9 atoms 10 atoms 11 atoms 12 atoms
CH CH2 CH3 * CH4 *
C2H l-C3H C4H C5 H C6 H C7H C8H
c-C3H C4H- C6H- C8H-
l-C3H2 l-C4H2 C6H2
c-C3H2 C6H6 *
C2H4*
HCCH* HC4H HC6H *?
*
CH3C2H CH3C4H CH3C6H
CH2CHCH3
CH+
y
The mark ‘‘*’’ is used for vibration-rotation transitions and ‘‘?’’ for tentative assignments. The symbols ‘‘l’’ and ‘‘c’’ indicates linear and cyclic isomers, respectively
K. M. T. Yamada and G. Winnewisser
Table 5.2 Organic molecules containing one hetero atomy
3 atoms 4 atoms 5 atoms 6 atoms 7 atoms 8 atoms 9 atoms 10 atoms 11 atoms 12 atoms 13 atoms
HCO H2CO H2CCO c-H2C3O
CH3OH
H2CS CH3SH
H2CN H2CCN CH3CN CH3C2CN C2H5CN CH3C5N n-C3H7CN
CH3NC C2H5OH H2CNH H2CCNH?
H2NCN
HCN HCCN HCCCN HC4N HC5N HC7N HC9N HC11N
HCCNC CH2CHCN
HC2CHO CH3CHO CH2CHCHO? C2H5CHO C2H5OCHO C2H5OCH3
c-C2H4O
5 List of Molecules Observed in Interstellar Space
CH2CHOH CH2CCHCN (CH2OH)2

CH3NH2 (CH3)2O (CH3)2CO
HNC HNC3
HCP
HCNH+ HC3NH+
+ +
HCO H2COH
HCS+
HOC+
y
The symbol ‘‘n’’ is used for the normal isomer
221
Table 5.3 Organic molecules containing two hetero atoms

4 atoms 5 atoms 6 atoms 7 atoms 8 atoms 9 atoms
HNCO NH2CHO CH3OCHO
HCNO CH3COOH CH3CONH2
HOCN CH2OHCHO
HNCS H2NCH2CN
HSCN
HC(O)CN
HOCO+ HCOOH
Table 5.4 Inorganic 2 atoms 3 atoms 4 atoms 5 atoms 6 atoms

molecules containing one or
C2 ** C3 * C5 *
more carbon atomsy
C3N C5N
C3N- C5N-
CO CO2 *
OCS
C2O C3O
C2P C3S
CO+
CN AlNC
MgCN
MgNC
NaCN
SiCN
SiNC
CP
CS C2S
CSi c-C2Si c-C3Si C4Si
CF+
y
The mark ‘‘*’’ is used for rovibrational transitions, and ‘‘**’’ for
rovibronic transitions. The symbol ‘‘c’’ is used for cyclic
isomers.
5 List of Molecules Observed in Interstellar Space 223
Table 5.5 Inorganic 2 atoms 3 atoms 4 atoms 5 atoms

molecules containing one or
more hydrogen atoms H2 H+3 * H3O+
HD H2D+
HD2+
H2O
H2O+
H2S
NH NH2 NH3
PH3 ?
HNO
HN2+
OH
OH+
AlOH
SH *
SH+
SiH ? SiH4 *
HCl
HF
y
The mark ‘‘*’’ is used for vibration-rotation transitions, and ‘‘?’’
for tentative assignments.
Table 5.6 Inorganic 2 atoms 3 atoms

molecules without a carbon
and a hydrogen atom AlF
AlCl
AlO
ClK
ClNa
FeO
NO N2O
NS
NSi
NP
O2
PO
SO SO2
SO+
SiO
SiS
Index
A ðKÞ
Beff , 114
21 cm, 27 Big bang, 16
Acetone, 162 Biogenic, 15
AFC, 194 Bohr radius, 46
Alanine, 126 Boltzmann plot, 35
Alchemy, 4 Born-Oppenheimer
AlCl, 31 approximation, 59, 71
AlF, 31 parameter, 60
AlH, 31 Brahe, 4
ALMA, 37, 204 Bunsen, 7, 17
Almagest, 2 Butanol, 150
AlNC, 31 BWO, 182, 183
AlO, 198
Amino acetonitrile, 37
Angular momentum C
orbital, 47 CI, 11
spin, 48 CII, 11
total, 48 C1, 150
Angular velocity, 122 C1(G3), 126, 130, 135
Anharmonicity constant, 64 C1(G9), 150, 159, 166
APEX, 33, 181 C2, 7
Ar, 51 C2H4O, 36
Array receiver, 212 C2v, 160
Astrology, 4 C2v(G36), 160
Asymmetric top, 88, 90 C2v(M), 104
near oblate, 91 C3, 7, 15, 186, 198
near-prolate, 91 C3H, 36, 180, 187
asymmetry parameter C3H2, 31
bo, bp, 110 C3N-, 34
j, 91 C3v, 123
C3v(M), 106
C4H-, 34
B C5, 15
B matrix, 82 C5N-, 34
B Star, 10 C6H6, 36
Balmer lines, 10 CCCCCCC-, 15
Barnard, 5 C8H-, 34
225
226 Index
C (cont.) D
CaII, 7 D/H ratio, 38
CaH, 31 D2H+, 180
Carbon ortho, 180
atom, 52 D2O, 192
ionized, 10 D3h(M), 107
neutral, 11 DCN, 38, 189
Casimir’s function, 71 Deferent, 2
CDMS, 28, 181, 219 Degenerate vibration, 96
Centrifugal distortion, 64, 99 d-Orionis, 7
CH, 13, 29, 32, 179 Difference frequency generation, 183
CH+, 13, 29, 179 Dimethyl ether, 180, 187
CH2, 179, 187 Dimethylamine, 162, 169
CH2NH, 29 Dirac equation, 48
(CH3)2O, 29 Dissociation energy, 61
CH3BF2, 129, 139 Doppler
CH3C6H, 39 shift, 188
CH3CCH, 35 sub-, 187
CH3CN, 35, 187 width, 181
CH3F, 192 Double resonance, 201
CH3OH, 29, 36, 124, 128, 137, Dunham coefficient, 64
187, 192
Character, 104
Character table, 102, 104 E
Charles IV, 4 Eckart equation, 79
Circumstellar, 30 Electric quadrupole, 55, 71
Cloud Energy level
cold, 5 two-top, 153
dark, 5, 17, 35 Epicycle, 2
dense, 12, 34 Equilibrium configuration, 78
diffuse, 13, 30 Esep, 127
giant molecular, 30 Ethylmethylether, 151, 187
CN, 13 Euler angles, 76
CnH, 35 EVLA, 39
CNPI group, 102
CnS, 35
CO, 12, 15, 16, 30, 32, 38, 187, 193, 197 F
solid, 15 F matrix, 82
CO+, 187 FeH, 31
Contact transformation, 98 FeO, 31
Coordinate Fine structure, 8
laboratory-fixed, 120 Fortrat diagram, 114, 115
Coriolis coupling, 80 Framework function, 132, 145, 166
Coriolis interaction, 119 Fraunhofer, 6, 17, 21
COSSTA, 193
CRL2688, 31
CS, 16 G
Cs, 150, 160 G matrix, 83
Cs(G12), 126, 134, 139 G12, 104, 126, 129
Cs(G18), 150, 159, 160, 166 G18, 153
Cs(G6), 126, 127, 130, 137 G3, 126
cyanopolyynes, 13 G6, 126
Index 227
G9, 128 HCOOH, 29, 187, 192

GaAs, 190 HD, 32, 38, 180
Gap frequency, 206 HDO, 32
Generating operation, 127, 128, 130, 153, 162 He, 51
GF calculation, 83 HEB, 213
Glycine, 37 helium, 8
GREAT, 204, 214 HEMT, 194
Group axioms, 102 Herschel, 5, 22
Group theory, 102 space observatory, 32, 36, 39, 204, 211
HF, 32, 179
HIFI, 181, 209
H High barrier limit, 119, 132
H, 43 Hipparchus, 1
H109a, 28 HNCO, 112
H2, 8, 16, 30 HII, 10
H2CCHCN, 29 Hulst, 9
H2Cl+, 34, 39, 179 Hund
H2CO, 13, 28, 29, 36, 110 case (a), 67
H2CS, 29 case (b), 68
H2D+, 32, 179 coupling case, 65
para, 180 rule, 11, 53
H2O, 13, 29, 32, 103, 111, 179 Hydrazine, 144
H2O+, 39, 179 Hydrogen, 8
H2S, 179 Hyperfine structure, 49, 54
H2S2, 184
H3O+, 32, 179
Hamiltonian I
atom, 49 IAM, 120, 128
diatomic molecule, 57 IF, 203
effective rotational, 142, 143, 148 Internal coordinates, 73
free-internal-rotor, 131 Internal rotation, 119
hydrogen atom, 43 Internal rotor
IAM, 125 one methyl, 119
internal-rotation, 143 two equivalent methyl, 160
linear rotor, 93 two methyl, 150
matrix element, 136, 159, 166, 167 Interstellar thermometer, 35
non-linear molecule, 84 IRC+10216, 31, 34, 36
PAM, 123, 142, 155 ISO, 15, 36
phenomenological, 155, 167
RAM, 123, 124
rigid rotor, 87 J
two equivalent methyl, 163 Jansky, 9
Harmonic wavenumber, 84 jet
Hartmann, 7 pulsed, 196
HC11N, 14, 35 supersonic, 195
HC3N, 29, 108, 187 Josephson effect, 206
HC5N, 29, 108, 187 AC, 210
HC7N, 108 JPL catalog, 39
HC9N, 35, 109 JPL catalogue, 181
HCl, 32, 179
HCN, 108, 187
HCnH, 35 K
HCO+, 28, 34, 187 K-type doubling, 110
HCOOCH3, 29 Karl IV, 4
228 Index
K (cont.) NH2CHO, 29
KCl, 31 NH3, 13, 29, 36, 106, 179
Kepler, 4 Normal coordinates, 82
Kirchhoff, 7 dimensionless, 83
Nuclear kinetic energy, 74
classical, 121
L Nuclear spin
Laboratory fixed axis, 71 statistics, 111
Laguerre polynomial, 47
Lamb shift, 48
K-type doubling, 69
O
Laplace equation, 56
OI, 11
Laser
O star, 10, 13
CO2, 194
OCS, 202
FIR, 192
Odin, 36
sideband, 193
OH, 13, 27, 29, 32, 179
Legendre polynomial, 46
OH+, 33, 179
Li, 51
Orbit
LiH, 31
atomic, 48
linear rotor, 87
Orbital, 48
LMH, 37
Orion, 12, 16, 38
Loomis–Wood diagram, 115
Orion-KL, 32
LTE, 37
OROTRON, 199
Lyman lines, 10
ortho, 111
Oxygen
neutral, 11
M
Magnetic interaction, 70
Mathieu equation, 130
Mattis–Bardeen, 209 P
Meteors, 3 PAH, 36
MgCN, 31 PAM, 120, 128, 151
MgH, 31 para, 111
MgNC, 31 Pauli exclusion principle, 50
Molecular symmetry, 101 Permutation-inversion group, 119
Molecule fixed axis, 76 Perturbation theory, 142
MoMeD, 192 PGOPHER, 109
MS Group, 102 PH, 187
Multiplier, 189 PH3, 105
Phase-locking loop, 185
Photo dissociation region, 11
N Planets, 2
N-methyl-acetamide, 156 Potential energy, 59
Na, 51 Potential energy function, 73
NaCl, 31 Principal axis system, 81
NaCN, 31 Ptolemaeus, 1
Na-D, 6 Ptolemy, 1
NaH, 31
NANTEN2, 204
Ne, 51 Q
Neutral element E , 103 Quantum defect, 51
Newton, 4 Quantum number
NH, 179, 187 magnetic, 46
NH2, 32, 179 principal, 46
Index 229
R Spin-orbit coupling, 11, 53

RAM, 120, 123, 128, 136 spin-orbit interaction, 48, 66
Reduced mass, 44, 60 Spin-rotation coupling, 68
Representation, 102 Spin-spin coupling, 68
irreducible, 102 Star formation region, 11
Resonance, 101 SuJeSTA, 195
Rotation, 75 SWAS, 36
Rotational constant, 63, 88 SWS, 15
Rotational spectra Symmetric top
a-type, 112 oblate, 88, 90
asymmetric top, 112 prolate, 88, 89
b-type, 112 Symmetry group, 102
linear molecule, 108 Symmetry operations, 102
Rovibrational spectra inversion, 102
a-type, 115 permutation, 102
b-type, 117 rotational, 102
Russel–Saunders coupling, 52, 65 Synchrotron
Rydberg CLS, 183
constant, 10 SOLEIL, 183
formula, 46
T
S
Taurus, 5, 13
Schottky
TMC 1, 13
multiplier, 182
Terahertz spectrometer
planar, 191
Cologne, 186
Schördinger equation
Term symbols, 53
diatomic molecule, 57
TiO, 198
electronic, 72
TiO2, 198
harmonic oscillator, 62
TMC, 5
internal rotation, 130
TMC-1, 35
molecular, 72
Totally symmetric representation, 107
nuclear, 72
Transition
vibrational, 61
electric dipole, 107, 149, 151, 161
Selection rule, 101, 149, 151, 160
Translational coordinates, 73
Sgr B2, 29, 32, 37
Tunnel junction, 205
SH, 187
Tunneling matrix, 132, 144, 167, 169
SH+, 33 Two-dimensional vibration, 95
Shell, 49 Two-photon transition
closed, 50 Doppler free, 202
sub-, 50
SiC2, 36
SiH, 32
SiO, 30 U
SIS, 204, 211 Unfeasible, 106
SOFIA, 204 Unit antisymmetric tensor, 85
Solar specrum, 6 Unitary operator, 97
Spectroscopy
Fourier transform, 183
Terahertz, 181 V
Spherical top, 87, 89 V3, 130
Spin V6, 135
electron, 48 van-der-Waals complex, 200
nuclear, 49 Vector model, 49, 67
230 Index
V (cont.) S-reduced, 99
Vibration, 75 Wavefunction
Vibrational coordinates, 81 angular, 45
Vibrational energy electron spin, 48
asymmetric top, 91 free-internal-rotor, 131
linear rotor, 94 harmonic oscillator, 62, 131
molecular, 72
nuclear, 72
W radial, 45, 47
Wang basis, 91, 139 rotation–vibration, 76
Watsonian rotational, 93
A-reduced, 99 symmetric top, 88

Yamada Winnewiser Interstellar Molecules

Uploaded by

Copyright:

Available Formats

Yamada Winnewiser Interstellar Molecules

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Yamada Winnewiser Interstellar Molecules

Uploaded by

Copyright:

Available Formats

Springer Tracts in Modern Physics

For further volumes:

Managing Editor Solid-State Physics, Editors

ISSN 0081-3869 e-ISSN 1615-0430

ISBN 978-3-642-16267-1 e-ISBN 978-3-642-16268-8

Springer Heidelberg Dordrecht London New York

Ó Springer-Verlag Berlin Heidelberg 2011

Cover design: eStudio Calamar, Berlin/Figueres

Printed on acid-free paper

Springer is part of Springer Science+Business Media (www.springer.com)

August, 2010 Koichi M. T. Yamada

2 Molecules Detected in Interstellar Space . . . . . . . . . . . . . . . . . . . . 27

2.3 Metal-bearing Molecules in the Interstellar Medium. . . . . . . . . 30

3 Fundamentals of Spectroscopy for Astrophysics . . . . . . . . . . . . . . 43

3.4.3 Case of a Molecule with Two Equivalent Methyl

4 Laboratory Experimental Methods . . . . . . . . . . . . . . . . . . . . . . . . 179

5 List of Molecules Observed in Interstellar Space. . . . . . . . . . . . . . 219

Koichi M. T. Yamada and Gisbert Winnewisser

1.1.1 The Ancient Mechanical Universe

K. M. T. Yamada and G. Winnewisser (eds.), Interstellar Molecules, 1

observations. Ptolemy presented his astronomical observations in a convenient

1.1.2 The Dark Ages

superstition, mysticism, and magic continued to obstruct the development of the

1.1.3 The Renaissance of Science

1.1.4 Discovery of Dark Nebulae

Fig. 1.1 This is the Taurus

1.1.5 Dawn of Spectroscopy

on the solar spectrum was followed in 1823 by a second catalog, summarizing

1.2 Interstellar Medium (ISM)

of lm-particles of rather uncertain chemical composition. A smaller fraction of the

1.2.1 Atomic and Molecular Hydrogen

1.2.2 The 21-cm Line of Neutral Hydrogen

gas in large regions of our

1.2.3 Recombination Lines

1.2.4 Atomic Fine- and Hyperfine-Structure Lines

increasing energy, respectively. This ‘‘normal’’ term order in atomic carbon—the

1.3 Interstellar Molecular Lines

In this section, we present a brief general overview of interstellar molecular lines;

In addition to the gas phase molecules, there is additional ambiguous low-

1.3.1 Small Molecules

interstellar molecules is not complete. Missing may be others due to a variety of

1.3.2 Large Molecules

1.3.3 Non-polar Molecules

problems, mainly technical and by exhibiting transitions in the Terahertz region

1.3.4 Atomic Carbon and CO Molecule

1.3.5 Molecular Cloud and Star Formation

1.4 Molecular Spectroscopy: An Irreplaceable Tool

1.4.1 Scope of Interstellar Atomic and Molecular Spectroscopy

relationship between laboratory development and interstellar progress, determined

1.4.2 Terahertz Region: Far–InfraRed Domain

1.5 Episodical Appendix

1.5.1 Life Story of Fraunhofer

1.5.2 Life Story of Herschel

Friedrich Wilhelm Herschel (1738–1822), generally known as Sir William

reports. After years of profound research, he brought out an important article on

Karl M. Menten and Friedrich Wyrowski

In this section, an overview is given of molecules detected in the interstellar

2.1 Detection of the First Interstellar Molecules

K. M. Menten and F. Wyrowski (&)

K. M. T. Yamada and G. Winnewisser (eds.), Interstellar Molecules, 27

with s ¼ 12 and ms ¼ 12 : The spin eigen function n is often expressed as a and b

hso ¼ Aso ð^l ^sÞ:

hHF ¼ að^j ^IÞ: