A New Method for Measuring Dielectric Constant and Loss in the Range of

Centimeter Waves
S. Roberts and A. Von Hippel

Citation: Journal of Applied Physics 17, 610 (1946); doi: 10.1063/1.1707760

View online: http://dx.doi.org/10.1063/1.1707760
View Table of Contents: http://aip.scitation.org/toc/jap/17/7
Published by the American Institute of Physics

Articles you may be interested in

Method of measuring dielectric constant using an oscilloscope
American Journal of Physics 83, 782 (2015); 10.1119/1.4922233

Measurement of the Dielectric Constant and Loss of Solids and Liquids by a Cavity Perturbation
Method
Journal of Applied Physics 20, 817 (2004); 10.1063/1.1698535

Microwave Dielectric Measurements

Journal of Applied Physics 18, 789 (2004); 10.1063/1.1697843

Dispersion and Absorption in Dielectrics I. Alternating Current Characteristics

The Journal of Chemical Physics 9, 341 (2004); 10.1063/1.1750906
 ANew Method for Measuring Dielectric Constant and Loss in the
Range of Centimeter Waves*
S. ROBERTS AND A. VON HIPPEL
Laboratory for Insulation Research, Massachusetts Institute of Technology, Cambridge, Massachusetts
(Received April 29, 1946)

In 1940, dielectric measurements in the centimeter range were considered as difficult and not
very accurate. The authors, therefore, developed a "hollow-pipe" method which overcame these
objections and required only a weak oscillator and small amounts of the dielectric material.
The theory and its practical applications, as perfected by March 1941, are presented in this
paper.

T HEtionalcapacitance of a condenser is propor-

to the dielectric constant or per-
The "lumped" circuit aspect used above be-
comes invalid in the ultra-high frequency range,
mittivity E* of the dielectric material which it but the concept of a complex dielectric constant
contains. is not affected. It describes the physical fact
(1) that a dielectric in a periodic electromagnetic
field carries an electric current of the density
where Co is called the vacuum capacitance and
EO is the permittivity of free space. When sub- J = uE+aD / at;= (E' - jE")jwE, (5)
jected to a sinusoidal voltage V, a condenser consisting of two components, a charging current
passes a current I which normally does not lead of the density
the voltage by 900 but only by (900-~) (Fig. 1).
(6a)
If the current is defined in the general way
storing electric energy by the reversible displace-
E*
1= jwCV = jwCo- V, (2) ment of charge carriers, and a loss component
EO
(6b)
the existence of the loss angle ~ can be repre- measuring the dissipation of energy in heat
sented by introducing a complex permittivity caused by traveling charge carriers as well as
E*=E'-jE". (3) by the time lag between the electric field E and
the dielectric flux density D.
The charging current is proportional to E', the The dielectric ~onstant and loss can be found
loss current to E", and the loss tangent indirectly in the microwave range by transmis-
tan ~ = E" / E' (4) sion line or optical methods. In the former E* is
derived from the intrinsic wave impedance Zo or
is given by the ratio of loss current to charging propagation function 'Yo of the dielectric medium,
curren t. E' and E" or E' and tan ~ are the paratn- while in optical methods it is evaluated from the
eters normally used for describing the frequency complex index of refraction n* equal to n(l- jK).
response of a dielectric material. The loss tangent Standing or traveling waves ean serve for meas-
is identical with the power factor cos () for small
angles ~.

* This
research, about which a preliminary report was
I J~WOOVl-
!' __
given at the meeting of the American Physical Society in
April, 1940, was ready for publication by March, 1941, but
had to be classified and withheld from general circulation.
The method, now one of common knowledge, has been
refined further in our laboratory during the war and is
being used with coaxial lines and hollow wave guides in
the range from one meter to one centimeter. A survey of
O
v '" 0
to
0
e
t
I
I
I

to WCov
V

this newer development and of the extensive dielectric FIG. 1. Relation between current and voltage in a condenser
information obtained will be given elsewhere. with loss.

610 JOURNAL OF APPLIED PHYSICS

uring anyone of these interrelated sets of to its axis. The dielectric is inserted in the closed
parameters. end of the pipe opposite to the transmitter, filling
The normal optical procedure employs travel- the volume to a height d. Above it the wave
ing waves. The intensities of the reflected and pattern is measured in air. E' and E" are calculated
transmitted beam are measured and nand K from the ratio of the field strength in node and
evaluated by Fresnel's equations. Esau, with his
co-workers Baz 1 and Kebbel,2 has carried this
method over into the ultra-high frequency 1... - ...
(I
range, but here it becomes definitely inferior. _1-_ MAGNETRON

Serious boundary problems arise because the :'" OSCILLATOR

wave-length is of the dimensions of the sample, "~

'- ~
and standing waves are set up between trans- I +_ _ SLOT
mitter and receiver, producing intensity fluctua-
tions by interference. . x
The macroscopical wave-length, on the other
hand, proves of great advantage if standing
waves, that is interference optics, are applied
because a detector may travel directly through
the profile of the wave pattern. Drude's two
classical methods 3 utilize this possibility and
have since been employed in many variations. 4
However, a handicap of the Lecher system re- 4-,1<--1--- MEDIUM I
mained: an empirical calibration of the con- AIR

denser system was necessary, and extreme care

had to be taken to avoid perturbation of the ~~~<rt+f'I-~-_ Y
waves by the detector system and to shield the "2 ~o
2
t
0 :...c;!'~-- MEDIUM 2
detector against stray fields. These limitations w w ~~ DIELECTRIC

have been overcome by the development of a '---_ _ _ PIPE CLOSED

"hollow-pipe" method, to be described in the AT END

following paragraphs. FIG. 2. Diagram of the apparatus showing principle of

the method.
PRINCIPLE OF THE METHOD
By limiting the electromagnetic field to the antinode Emin/Emax and the distance Xo of the
enclosure of a hollow pipe or coaxial line, all first node from the surface of the dielectric.
boundary and stray effects disappear auto- It is a known procedure to calculate the ter-
matically and small amounts of any dielectric minating impedance of a transmission line from
can be measured with precision. With this con- the voltage ratio V mini Vmax and the distance XO,5
cept in mind we developed the apparatus shown or an acoustic impedance, by the location and
schematically in Fig. 2. A transmitter radiates relative magnitude of the sound-pressure maxima
waves of a given frequency into one end of a and minima. 6 But measurements of the terminal
closed wave guide; they are reflected by the impedance of a short wave guide filled with a
metallic boundary at the other end. Standing dielectric material have not been applied, so
waves are set up and can be measured by a probe far as we know, to the evaluation of E' and E".
detector traveling along a slot in the pipe parallel We are giving, therefore, in the following pages
a mathematical theory of the method, the de-
1 G. Baz, Physik. Zeits. 40, 394 (1939).
2 W. Kebbel, Hoch:tech. u. Elek:akus. 53, 81 (1939). scription of our apparatus, and some results
3 P. Drude, Wied. Ann. 55, 633 (1895); 61, 466 (1897). illustrating its performance.
4 See, for instance, G. Potapenko, Zeits. f. Physik 20,
26 (1923); M. Seeberger, Ann. d. Physik 16, 77 (1933)' H. O. Rosenstein, Zeits. f. Hochfrequenztechnik 36,
R. King, Rev. Sci. Inst. 8, 201 (1937); K. SleYogt, Ann: 81 (1930); N. N. Maloy, Physik. Zeits. Sowjetunion 11,
d. Physik 36, 141 (1939); H. Slatis, Ann. d. Physik 36 539 (1937).
397 (1939). ' 6 W. M. Hall, J. Acous. Soc. Am. 11, 141 (1939).

VOLUME 17, JULY, 1946

 MATHEMATICAL THEORY and their ratio is
A standing wave in the pipe can be repre- Em in l-e- 2p
sented as the sum of two traveling waves pro- tanh p. (12)
gressing in opposite directions parallel to the
axis of symmetry. The transverse field com-
The first minimum above the dielectric bound-
ponents E(x)l and H(x)l in air (medium 1) at a
ary occurs at a distance Xo where the incident
variable distance x from the dielectric surface
and reflected electric waves bave a phase dif-
along a line parallel to the axis are given below.
ference of 7r radians, that is,
E(xh=Ailexp ('Ylx)+Arlexp (-'YIX) 27rxo 27rxo
-=-2if;---7r, (13a)
=Ail[exp ('Ylx)+rO exp (-'YIX)], Al Al
(7) or
(13b)

The wave impedance Z(O) can now be expressed

in terms of the directly measurable parameters
xu, AIt and Emin/Emsx. Expanding coth cp in Eq.
(9) one obtains

Ail and Arl are the amplitudes of the incident tanh p- j cot if;
and reflected waves at the dielectric surface. Z(O)=Zll------
1 - j tanh p cot if;
Their ratio
E min 27rxo
(8) ---jtan--
where E ma" Al
=Zll------- (14)
Emin 27rXo
defines a reflection coefficient which character- I-j--tan--
Emax Al
izes the terminating pipe section containing the
dielectric medium 2. The wave impedance Z(O) The impedance Z(O) is determined by the
of this section, given by the ratio of the electric length d and the propagation function 'Y2 of the
to the magnetic field intensities at the dielectric short section of pipe containing the dielectric,
surface, is related to Yo and cp. or conversely, if Z(O) and d are known, 'Y2 can
be derived from them.
E(O) 1 +ro 1+e- 2</1 The standing wave in the dielectric is
Z(O)=--=Zl--=Zl =Zlcothcp. (9)
H(O) l-ro l-e- 2</1
(15)
Normally the attenuation in the air-filled part
of the pipe can be neglected, that is, the propaga-
tion 'YI, is reduced to its phase factor
(10)
and at x= -d (Fig. 2) the pipe is terminated by
a metal plate, introducing at this boundary a
where Al is the wave-length in air inside the' voltage node. .
wave guide. Since 'Yl is imaginary, that is,
O=Ai2 exp (-'Y~)+Ar2 exp ('Y2d). (16)
lexp('Ylx )I ==1, it follows from (7) that the
maximum and minimum field amplitudes of the Solving (16) for Ar2/Ai2 and substituting this
standing wave are value in (15) gives
E ma" = lAd (1+ Irol) = IAil (l+e- Zp ), E(Oh
(11) Z(O) =--=Z2 tanh wi. (17)
Emin IAil (1-1 rol) = IAil (1-e- 2p
), H(O)2

612 JOURNAL OF ApPLIED PHYSICS

 I U 101 1--\ Il- .! I}" ~ ~1.61t:l_t1.l1-\ Ie Ltll-.,j
1/1 f.l 1)( ('f \\1'. ~ fX.. :'r"D\ ~ ~ I)( ~J:2C Ii
I~) I '" .Y .m. ~ l"J r\ 1'\ ~ u.. f>() r-. J c: IT" ~ D4
II
\I
~ Di ~
v
jIl~~ ~II
1)\
iJIIlV r-,.. ~ 1\ 1'1- !all v
\V fI'I!: H- ~ If.... P\' ) r- Io-" ~
\
) F" V t- ~
V F! i-" IF
P<- '"
I
I
" ]
~I\
loX IV
f'.. eYl\
~
v
'\ /<IIA V
I)
1-

JIO' V
f\..
)

II
I'k f' \ II \ "'~\
,J Isr' '\. [\., I 1\
1\ \j IY ] \
1\ \ V 19' \ 1/ '\~
1:' 1\ v / I~ I~I
I\~ ..,./1\ I 1\
/
/[\
- II! \~
".... I-"'
\
\
leo-

/ \{ ..,
i\ '\
II \
1/ ~ 100"
"V
V , 1/
\ j
Iter. l-

l-

o 2 3 4 T 5 6 7

FIG. 3. Chart of the function Ceil =tanh Te iT /Te iT , with! and T expressed in degrees.

The characteristic impedance Z2 and the pro- Equating (19) and (14) yields
pagation function 1'2 of any TEM or TE wave
Emin 211"xo
in a dielectric are related in the following manner: ---jtan--
(18a) tanh 'Y~ - JXI Emax Xl
---=--. (20)
where Jl.2 is the permeability of the dielectric 'Y~ 211"d Emin 211"XO
I-j--tan--
medium. The permeability of dielectric materials Emax Xl
is generally Jl.o of free space; therefore
(18b) The function Ceii can be found, as shown
and above, by measuring the thickness of the sample
1'1 d, the wave-length Xl in the air-filled pipe, the
Z(O) =Zl- tanh 'Y2d. (19) ratio of minimum to maximum field strength
1'2
Emin/ Em ax , and the distance Xo of the first mini-
010 !
mum from the dielectric surface.
," ;
7 0.001 The next step is the determination of Te iT
008

= 'Y~ from charts or from a series approximation

006 3 of the function
tanh Te iT
004 2
----=Ce il (21)

.002
In Fig. 3 the argument T is the ordinate, the
absolute value T the abscissa, while C and ~ are
025 075
parameters of intersecting curves. T and ~ are
expressed in degrees. The hyperbolic functions
FIG. 4. Emin/Emax and XO/AI, versus the ratio of thickness
d of the sample to wave-length A2 in the dielectric. are multivalued so that measurements with a

VOLUME 17~ JULY, 1946 613

 geometry of the pipe and the mode of propaga-
tion. The frequency of plane waves with a wave-
length Ac in the medium to which "}' refers is
called the cut-off frequency. The values of Ac for
the lowest order waves in different wave guides
are
rectangular pipe7 Ac=2Xwidth
round pipes Ac = 1.71 X diameter
coaxial line A<= 00
The value of "}'1 is given by

(24)
w

Bearing in mind that J.' for dielectrics is generally

of free space, one obtains
J.'o

FIG. 5. Constructional details of the apparatus. For the coaxial line this equation reduces to
I (26) .
single thickness d of the dielectric may leave the
value of "}'2 in doubt. But if two different thick- The determination of E2* may now be sum-
nesses are used, only one set of values T/d and marized as follows. The ratio Em in/E max , the
T, will satisfy both experimental results. distances Xo and d and the wave-length Al are
For low loss materials it is desirable to use a measured. From these measured values C and
thickness near A2/4 or higher odd multiples of t are calculated with the help of Eq. (2.0). From
A2/4. As Fig. 4 shows, Emin/Emax reaches maxima REFLECTOR
for these values of thickness. The apparatus used
in 194.0 could measure the loss if Emin/ Emax
were larger than .0 ..0.03. For d=Ad4 in Fig. 4,
the loss would be below this limit, while for a
thickness d=3Ad4 or T=31f/2, the loss could
be determined easily. In handling dielectric
materials with low loss one can obtain more ac-
curate results than are obtainable from the
chart, by using the following series approxima-
tion, valid in the range C> 1 when T~3/21f.
.0.212 .0 ..0.096 )
TeiT=j ( 4.71---eii----ei2i . (22)
C C2
FIG. 6. Details of magnetron.
The complex dielectric constant can be found
from "}'2 by the general relation C and t the values of T and T are found from the
chart or by a series approximation. "}'2 is found
(23)
7 L. J. Chu and W. L. Barrow, Proc. I.R.E. 26, 1520
(1938).
8 G. L. Southworth, Bell Sys. Tech. J. IS, 284 (1936);
The cut-off wave-length Ac is determined by the W. L. Barrow, Proc. I.R.E. 24, 1298 (1936).

614 JOURNAL OF APPLmD PHYSICS

by dividing re
iT by d. Finally E2* is determined of hole in the diaphragm! inch. The resonator
from '"Y2 by Eq. (25) or (26). R is inserted for part of its length in the pipe
section holding the magnetron. The main pipe,
APPARATUS AND PROCEDURE which slides over the free end of the resonator,
The apparatus used in 1940 is shown in Fig. 5. is adjusted to resonance and then clamped in
It is now obsolete in many details and will position.
therefore be described only briefly.
A round wave guide P is mounted vertically
and is fitted at the bottom with a detachable
end piece in which either solid or liquid dielec-
trics may be inserted. A magnetron oscillator T
with its magnet M is located at the upper end
of the pipe. The output of the magnetron is
monitored by a bolometer detector B. The stand-
ing wave above the dielectric is measured by
means of the traveling crystal detector E which
is driven by the lead screw W while its position
is indicated to an accuracy of .001 inch by the
revolution counter C.
The magnetron oscillator is of a special design
shown in Fig. 6. An open anode construction,
which serves to radiate the energy into the pipe,
avoids the tuned parallei"transmission line which,
FIG. 8. Details of traveling crystal detector.
in conventional constructions,9 feeds the energy
through the glass envelope to a dipole antenna. The construction and circuit diagram of the
The principal advantage of this design is its bolometer detector B are shown in Fig. 7. This
simplicity of construction. It has an efficiency of arrangement proved very stable and troublefree,
about one percent and requires a critical adjust- but of course less sensitive than a crystal de-
ment of the magnetic field intensity and the tector of the silicon-tungsten type.
orientation in the field.
The slotted pipe maintains a wave which is
The output of the magnetron is filtered and
polarized so that. the electric field is most in-
stabilized by means of a resonator R, a cylindrical
tense and the longitudinal magnetic field zero at
tube fitting in the pipe and closed on both ends
the slot. The electric field is shown in Fig. 8
by diaphragms. The approximate dimensions of
together with a more detailed drawing of the
this filter for a wave-length of 6.0 cm are: length
detector construction. Good metallic contact be-
3 inches, diameter of tube 1t inches and diameter
tween detector housing and pipe is required and
is provided by the slider L held firmly with the
phosphor-bronze spring A.
In principle, one measurement with the di-
electric material in the pipe would be sufficient.
After tuning to resonance by adjusting the
plunger D (Fig. 5), Xo, Emin/Emax and Al are
found by the rectified current and position of the
traveling detector. In practice, a second meas-
urement is carried through with the empty pipe
for calibrating the detector. Here the field
!>oon.
strength is known. It should go down to zero at
FIG. 7. Bolometer circuit and method of construction.
the minimum and increase towards the maXl-
9 E. G. Linder, Proc. I.R.E. 27, 732 (1939). mum proportionally to the sine of the phase

VOLUME 17, JULY, 1946 615

angle 8, where TABLE I. Dielectric constant and loss at=6.0 cm.
(Temperature 252C except where noted.) .
0= 2?r(x-xo)/}'l' (27)
Then Material e'/Eo tan 6
I E(X)ll = Em,ax 1 sin 0 I (28) Solids
Polystyrene 2.53 .0005 to .002
The calibration curve of the detector is found by Lucite 2.57 .005 to .OOS
noting the detector current for several positions Plexiglas 2.60 .006
Cellulose acetate 3.24 .033
on both sides of the minimum. The amplitude Polyethylene 2.25 .0002 to .0012
at 'the minimum is an indication of attenuation Glass
Soda-lime-silica te 6.7 .014
in the pipe and frequency instability of the os- Pyrex No. 774 4.8 .009
cillator, which should be very low for proper Pyrex No. 707 3.9 .0012
Fused quartz, 3.80 .0001
operation. The wave-length is determined by Paraffin wax 2.25 .0002
measuring the positioRs of two minima. Plywood (Bakelite bonded) 1.9 0.060
Poplar, grain II electric field 2.05 0.075
With a low loss dielectric in the pipe, the Poplar, grain 1- electric field 1.75 0.065
measurements are carried out only in the neigh- Liquids
DistiIled water (28SC) 71 0.22
borhood of the minimum, because the maximum Glycerine U.S.P. (2SC) 5.6' 0.60
Benzene (thiophene free) 2.25 <.001
is flatter and the detector is likely to be over-
loaded. in this case a simple formula can be de-
rived for calculating the ratio of minimum to smaller values of this ratio than could be done
maximum field strength. The absolute value of directly.
electric field strength is found from (7).
RESULTS
IE(x)l! = !A,l!!l+e-2('Y,x+</!
Table I gives the values of E' and tan ~ at
=:= !A,l!ll-e- 2 (p+i8)1. (29) A= 6.0 cm for a number of common materials
including solids and liquids. The first measure-
Then the ratio of this field strength to the mini-
ments on most of these substances were made in
mum, given by (11), is
1940 on the apparatus described above. Since
1 ep+i8-e-(p+i~=lsinh. (p+ jO)I, that time extensive and more accurate data at
A = 3 cm, A = 10 cm and at longer wave-lengths
Emin eP-e- P smh p have been obtained for 'a wide variety of di-
electrics in this Laboratory and summarized in
= ISinh p cos 0.+ j cosh p sin 01. (30)
the two volumes Tables of Dielectric Materials
smh p prepared under O.s.R.D. Contract OEMsr-191.
The square of this ratio is Table I has been edited to some extent on the
basis of these more recent data.
IE(x)112 sin 2 0
cos 2 0+--- (31) ACKNOWLEDGMENT
Emin2 tanh 2 p
The authors are indebted to the International
Solving for tanh p, one obtains the desired result. Telephone and Telegraph Company for financial
assistance in this research, and to Dr. H. H.
Emin sin (1
--=tanh p= Buttner, its Vice President, for his active interest
Emax (IE(X)112
cos 2 (J
)l . (32)
in the progress of the work.
- Initially, the study was to be in cooperation
Emin 2
with Professor J. A. Stratton and Dr. L. J.
The use of this indirect method of determining Chu, of the Department of Physics, whose advice
Emin/ Emax makes it possible to measure much on theoretical questions was of great value.

616 JOURNAL OF ~PPLIED PHYSICS