THE REVIEW OF SCIENTIFIC INSTRUMENTS VOLUME 30, NUMBER 7 JULY, 1959

Versatile and Sensitive Vibrating-Sample Magnetometer*

SIMON FONER
Lincoln Laboratory, Massachusetts Institute of Technology, Lexington, Massachusetts
(Received January 15, 1959; and in final form, May 4,1959)

A vibrating-sample magnetometer, which measures the magnetic moment of a sample when it is vibrated per-
pendicularly to a uniform magnetizing field, is described. With this instrument, changes as small as 10- 6 to 10-6 emu
have been detected, and a stability of one part in 104 has been attained. In addition to permitting convenient meas-
urements in the usual laboratory electromagnet, this instrument eliminates or minimizes many sources of error
found in other methods. It is simple, inexpensive, and versatile, yet permits precision magnetic moment measure-
ments to be made in a uniform magnetizing field as a function of temperature, magnetizing field, and crystal-
lographic orientation. The mechanical design and detailed operating characteristics are presented. Applications
and limitations of the method are outlined.

I. INTRODUCTION is that they are limited to particular phenomena which are

observable in a limited class of materials about which con-
T HE usual methods of measuring magnetic moments
can be divided into three major classes: measure- siderable prior knowledge is required. Despite many
limitations, in particular instances these indirect tech-
ment of a force. on a material in a nonuniform magnetic
field, measurement of magnetic induction in the vicinity niques are capable of extremely high sensitivity.
of the sample, and indirect measurments of phenomena All the magnetic induction measurements involve ob-
which involve the magnetic properties. The force method is servation of the voltage induced in a detection coil by a
a sensitive technique which has been employed for many flux change when the applied magnetic field, coil position,
years in the laboratory. With this method it is, of course, or sample position is changed. Many different experi-
difficult to observe the magnetization in a truly uniform mental arrangements have been employed to suit particular
magnetic field since the field gradient is essential to the investigations and some are described in reference 2. Re-
production of the force. Furthermore, this method is not cently, oscillatory coil or sample techniques have been used
easily adaptable to routine measurements of magnetization to observe the magnetization of a sample by an ac method.
versus applied field or crystallographic orientation. Ob- All of these techniques have employed an arrangement in
jections to the use of force methods for magnetic measure- which the detection coil is symmetrically distributed about
ments of highly anisotropic materials have recently been the sample with the axis of the detection coil parallel to the
raised by W 01£.1 applied magnetic field. One major disadvantage of such a
Numerous indirect techniques for measuring magnetic method is that, unless an air-core solenoid type magnet is
moments include2 measurement of the Faraday effect, used, the usual laboratory magnet must be extensively
modified. The most convenient arrangement has been to
analysis of galvanomagnetic effects such as the ferromag-
drive either the sample or the detection coil with a rod
netic Hall effect,S and microwave ferromagnetic resonance
which passes through one of the magnet pole faces; the
measurements. The advantages and disadvantages of in- laboratory magnet is then specifically adapted to the
direct measurements are illustrated by microwave ferro- magnetometer. A particularly successful vibrating-coil tech-
magnetic resonance experiments whereby the magnetiza- nique has been developed by D. O. Smith 6 using this pro-
tion may be obtained from a detailed knowledge of the cedure. The difficulties encountered here are that extremely
sample shape. 4 •5 The general problem of indirect techniques uniform magnetic fields are required, and that, further-
more, it is difficult to correct for effects produced by small
* The work reported in this paper was performed at Lincoln nonuniformities of the field. The latter problem is particu-
Laboratory, a center for research operated by Massachusetts Institute
of Technology with the joint support of the U. S. Army, Navy, and larly bothersome when the field nonuniformities are func-
Air Force.
1 W. P. Wolf, ]. App!. Phys. 28, 780 (1957).
tions of the applied field. Even in a uniform magnetic field,
2 A comprehensive review of a number of techniques for measuring for samples of small magnetic moment, large corrections
magnetization can be found in L. F. Bates, Modern Magnetism
(Cambridge University Press, New York, 1951). must be made for the magnetic effects of materials which
3 S. Foner, Phys. Rev. 101, 1648 (1956). surround the stationary sample. 7
4 Applications of shape effects to measurements of magnetic sus-
ceptibilities using nuclear resonance techniques have been described A particularly successful vibrating-sample magnetometer
by Reilley, McConnell, and Meisenheimer, Phys. Rev. 98, 264(A) is described in this paper. Since its inception, many labo-
(1955), and B. E. Holder and M. P. Klein, Phys. Rev. 98, 265(A)
(1955). 6 D. O. Smith, Rev. Sci. Instr. 27, 261 (1956), and more recently
6 Such measurements of M, to 1% have been made on thin Per- Dwight, Menyuk, and Smith,]. App!. Phys. 29, 491 (1958).
malloy films which exhibit narrow resonance line widths. See 7 The same problems would be encountered with the rotating-coil
M. H. Seavey, Jr., and P. E. Tannenwald, ]. App!. Phys. 29, 292 technique recently described by ]. Kaczer and Z. Malek, Czech.
(1958). J. Phys. 7, 481 (1957).
548

Downloaded 29 Jul 2011 to 130.65.109.155. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions
 VIBRATI~G-SAMPLE MAGNETOMETER 549

FIG. 1. Simplified form of vibrating-sample magnetom~ter: (1) loud-

speaker transducer, (2) conical paper cup support( (3) drInkIng str<l;w,
(4) reference sample, (5) sample, ~6) reference cOlIs, (7) sample cOlIs,
(8) magnet poles, (9) metal contaIner.

ratories have constructed similar instruments. The novel

features of this magnetometer are: first, sample motion
perpendicular to the applied field; and second, detectio~
coil configurations, with effective area-turns nonsymmetn-
cally distributed about the axis of vibrations, which permit FIG. 2. Detailed mechanical construction of ,vibr<l;ting-sample
magnetometer. The various parts are descnbed In the text.
this oscillating dipole field to be observed. The basic instru-
ment, briefly described earlier 8 ,9 is shown in Fig .. 1. The
amplifier linearity. The associated electronic circuits serve
sample (5) is vibrated perpendicularly to the applIed field
the function of a null detector.
by the loudspeaker assembly 0), (2), and (4). The oscil-
The measurements are made insensitive to the exact
lating magnetic field of the vibrating sample induces a
sample position by employing suitable pairs of coils. To do
voltage in the stationary detection coils, (7), and from
this for the coils (7) in Fig. 1 [a 3-dimensional view of these
measurements of this voltage the magnetic properties of
coils, (L2), is shown in Fig. 2J, the sample is first centrally
the sample are deduced. A second voltage is induced in a
positioned between the coils by visual inspection. The
similar stationary set of reference coils (6) by a reference
sample (4) which may be a small permanent magnet an 0: sample coils are then rotated about the z axis for maximum
signal output. (The output is then insensitive to small
electromagnet. Since the sample and reference are dnven
rotations of the coils about this axis). They are then trans-
synchronously by a common member, the phase and ampli-
lated in the X direction for a minimum output, along the
tude of the resulting voltages are directly related. The
Y direction (perpendicular to the paper) for a maximum
known portion of the voltage from (6), phased to balance
output, and along the Z direction for a maximum output
the voltage from (7), is then proportional to the magnetic
(if the coils are short). The sample is now located at a
moment of the sample. By this procedure the measure-
"saddle point"; the output signal is independent of small
ments can be made insensitive to changes of vibration displacements of the sample in any direction.
amplitude, vibration frequency, small magnetic field in- In addition to the above features, this vibrating-sample
stabilities, magnetic field nonuniformity, amplifier gain, or magnetometer eliminates or minimizes many of the usual
8 S. Foner, Rev. Sci. Instr. 27,548 (1956).
sources of error found in other methods. It is simple, in-
9 S. Foner, Bull. Am. Phys. Soc, Ser. II. 2, 128 (1957). expensive, and versatile, yet permits precise magnetic

Downloaded 29 Jul 2011 to 130.65.109.155. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions
 550 SIMON FONER

moment measurements to be made in a uniform magnet-

izing field as a function of temperature, magnetizing field,
and crystallographic orientation. An average stability of
the balanced signals greater than one part in 104 has been
observed. The sensitivity can be extremely high: changes
of 5 X 10- 5 to 5 X 10- 6 emu can be detected. The combina-
tion of high sensitivity and high stability permits measure-
ments on weakly magnetic materials, and differential meas-
urements of very small changes of magnetic moment. Such
sensitivities are useable because the vibrating-sample tech-
FIG. 3. Cross-sectional view of single
nique does not detect any stationary non uniformities of the glass Dewar used for low-temperature
magnetic field due to the magnet or the medium surround- measurements. The two detection coils
are shown outside the Dewar. Part of
ing the sample. to This emphasizes an important distinction the sample mount and surrounding
between the moving sample and moving coil techniques- container are indicated.
they are not equivalent unless the detection coil plus the
magnet producing the field, and the medium surrounding the
sample are moved as a unit when the sample is stationary.
This procedure is usually impractical, and generally not
employed in moving coil methods.

II. MECHANICAL DESIGN OF MAGNETOMETER

The simplified diagram of the magnetometer, Fig. 1,

includes all the basic elements of the instrument. Detailed
mechanical features of a vibrating sample magnetometer
are shown in Fig. 2. The magnetometer vacuum container field so that the magnetization is not affected appreciably
includes: an evacuation tube A soldered to a removable by external fields. Reference sample M may be replaced by
brass hat B; an O-ring vacuum seal at base plate P; and a a small coil carrying a controlled dc current when precise
threaded extension of P which joins a removable, seg- automatic recording of magnetic properties is desired.
mented, brass extension-tube S which also employs an The vibrating assembly from M to W is 28 in. long to
O-ring seal. The vacuum container is free to rotate about permit placing the magnetometer in any electromagnet
the vertical axis and rests on a counter-sunk support plate available in our laboratory. The threaded connections are
F having 3 leveling screws. The azimuth angle is given by convenient for disassembly when the magnetometer is to
the indicator scale G attached to F. The vibration system be stored in a small space, or when an element must be
is composed of: a loudspeaker C glued to a plastic element replaced.
H which provides rigid support for, and centers, a threaded The reference coils Ll and the sample signal coils L2 are
plastic rod J; a reference sample holder K in which the adjusted as described earlier-L1 is glued into place whereas
reference sample M is held; a threaded plastic connector N L2 is supported independently on a vertical screw arrange-
which is glued to a thin-walled stainless extension tube R ment for convenience in adjustment. (The supports are
to which a solid plastic rod T is attached; a threaded indicated in the figure.) Vacuum seals for electrical con-
sample holder U which permits convenient interchange of nections from Ll and to the loud speaker are also indicated.
samples W; and finally, a Teflon centering washer V press The brass extension tube S is thin-walled (0.015 in. or
fitted to U, which permits only a few thousandths of an less) to minimize heat losses. A thermocouple is attached
inch motion perpendicular to the axis of the driving as- to the smaller telescoping tube (0.25 in. o.d.) in order con-
sembly, but does not add appreciable friction to the vi- veniently to monitor the temperature at the sample.
brating system. Thermal contact between sample and outer tube is ob-
The reference sample M is a thin disk permanently mag- tained by introducing helium gas into the vacuum chamber.
netized in the plane of the disk. The magnetic material The threaded section joining S is extended to support, and
used here, uniaxial single domains of BaO· 6Fe 20g oriented allow vertical adjustment of, a Dewar cap assembly.
and compressed in a magnetic field ,11 has a large coercive A cross section of a convenient glass Dewar flask is
shown in Fig. 3. The volume of the upper section is about
10 If the sample is not saturated, only the field non uniformity over
the sample volume may be detected (see also Sec. VI B). 400 cma• The central section is minimized in diameter in
11 Such materials are now available under the trade name of order to obtain both maximum applied magnetic field and
Magnadur in Europe, and under the trade name of Ferroxdure or
Indox V in the United States. signal sensitivity. The small (80 cma) reservoir at the

Downloaded 29 Jul 2011 to 130.65.109.155. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions
 VIBRATING-SAMPLE MAGNETOMETER 551

z
bottom of the Dewar is useful for producing a slow tem-
perature change during measurements. For further control
of sample temperature, a heater may be placed in the
reservoir. With such a Dewar, a warm-up from liquid nitro-
gen (77°K) to room temperature can be extended to several
hours. Losses are sufficiently small at 77°K to allow meas-
urements for 3 to 4 hr at this fixed temperature. The same
Dewar has been used for measurements at liquid helium --+-------*------+-- x
temperatures without a surrounding liquid nitrogen Dewar.
About 20 to 30 min running time is available at 4.2°K
(sufficient for several M vs H measurements) with a fairly
rapid warmup (about 30 to 60 min warmup time from
4.2°K to 77°K).
A metal Dewar is useful for measurements over a pro-
longed period at 4.2°K, or for measurements below 4.2°K.
FIG. 4. Time varying part of dipole field in X-Z plane for
Such a Dewar (with a small tail section 0.56-in. o.d. and vibration parallel to Z and dipole moment parallel to X.
a 500-cm3 helium reservoir) has operated for 3 hr at 1.6°K.
For high-temperature work, a separate extension tube formities of the magnetizing field, which causes M to vary
similar to 5, but with a small furnace attached to the region during vibration, may be examined.
near the sample, can be used. Of course, the materials for In general, the pickup coil dimensions are not small
the sample holder and associated parts must be made of compared to their distance from the sample. Furthermore,
heat resistant materials. Weakly magnetic ceramics or the coil geometries often do not follow the symmetries of
metals may be used. the time varying field. For these reasons, the voltage in-
From the above description, it should be apparent that duced in most useful pick up coils can not be calculated
the mechanical features are sufficiently simple and flexible in closed form.
to permit numerous modifications to be made for particular One of the most convenient detection-coil arrangements
applications. is the double-coil shown in Fig. 5(a). The spatial variations
III. COIL DESIGNS of relative output signal for two typical double-coil assem-
blies are plotted in Figs. 6 and 7. These results were ob-
A. General Analysis
tained experimentally by observing the output signal due
The advantages of sample vibration perpendicularly to to the magnetic moment of a small magnetically saturated
the applied field can be realized only if suitable detection nickel sphere as it was positioned at various points in space.
coil arrangements can be devised. In practice many satis- For this measurement, the entire magnetometer was posi-
factory coil configurations can be found. This is easily seen tioned at various points in the XY plane by a suitable
if we consider the time varying part of the vibrating dipole micrometer arrangement, and the detection coils were
field. The scalar potential of a fixed dipole M at the origin moved in only the Z direction. Data were obtained to at
and pointed along the X direction is ¢=Mx/r. If M is least 0.5% accuracy with the null detection procedure de-
vibrated in the Z direction with sufficiently small ampli- scribed in Sec. IV. The experimental points were obtained
tude a, the time varying potential in the surrounding space at intervals of 0.010 to 0.020 in., and deviate from the
will be ¢1eiwt where ¢1= -a (a¢/aZ) = aMxZ/r 5 • The flux curves by less than 0.25%.
pattern of the time varying part of the field is given by An additional feature of this double-coil system is that
- grad ¢1. Its configuration in the X Z plane is shown the two coils are connected series-opposing in order to
qualitatively in Fig. 4. obtain a net output signal. This arrangement to a large
Appropriate placement of pick up coils to sense this flux extent eliminates the effects of the background noise due
may be visualized easily and interesting new configurations to magnetic field instability or mechanical vibrations of
suggested. For example, coils at 45°, or coils with hori- the magnet and coil systems. These effects are also inde-
zontal axes which would fit in a narrow pole gap could be pendently minimized by each coil since a minimum of area-
used. The latter case may require 4 coils if the instrument turns is presented to the applied field.
occupies part of the volume in the Z direction. A general The coil configuration shown in Fig. 5(a) has been em-
feature of all the useful coil configurations is that each ployed extensively for almost all our magnetic measure-
effective area-turn is nonsymmetrically distributed about ments. This arrangement has proved both easy to assemble
the axis of vibration. Pairs of coils are employed in order and most convenient in operation. Oval-shaped coils,
to minimize effects of sample position or external field shown in Fig. 5(c), have also been used extensively. Their
variations. In a similar way, the effect of slight nonuni- characteristics have the general features of Figs. 6 and 7.

Downloaded 29 Jul 2011 to 130.65.109.155. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions
 552 SIMON FONER

or Y direction so that the output signal is then insensitive

to small sample displacements in these directions. This
arrangement suffers, however, from a strong Z dependence
of the output signal. Some characteristic results for a
single-coil are shown in Fig. 8 to illustrate the dependence
observed. :i\ifeasurements have been made using such an
a b arrangement, with almost as high an accuracy as the
double-coil arrangement in Fig. 5(a), when an additional
normalizing point was obtained with the Fig. 5(a) coil
arrangement. Annoying spurious voltages, introduced by
external vibrations, can be minimized by placing a second
compensating coil of comparable area-turns in the mag-
netic field. The magnet air gap can be reduced by about a
factor of three with an appropriate field increase, when the
single-coil arrangement is employed. As shown at the top
right of Fig. 8, a gain of almost 10 in output signal can be
realized when the sample is positioned close to the top of
the coil.
f (top view) Additional examples of useful coil arrangements (all
with axes parallel to Z) are shown in Fig. 5. In particular,
Fig. Sea) may be modified by the addition of a pair of coils
coaxial with the Z axis [see Fig. Sed)]. With such an
e (top view) f (side view) arrangement, the magnitude and direction of the magnetic
moment vector in space can be determined-the Z com-
ponent is detected by the coaxial pair, and the component
in the XY plane is determined by rotating the double coil.
Figure See) shows a multiple-coil arrangement which
attempts to intercept a maximum of the sample dipole
g(side view)
field, but at the expense of additional thermal noise of the
FIG. 5. Examples of useful detection coil arrangements described
in the text. The sample, indicated by the heavy arrow, is vibrated coils. Four coils of this multi-coil array have been used for
along the Z direction. high-field configurations. An efficient modification of
Fig. See) is shown in Fig. S(f); this coil geometry, however,
An arrangement of a single coil, useful when very high is not easily fabricated. Finally, the cross section of a coil
fields are required, is shown in Fig. 5(b). The output volt- geometry which reflects most of the dipole field symmetry
age in such a case can be maximized for position in the X properties is shown in Fig. S(g). It is directly derived from
Fig. S (f), and leads to rather simple computations of output
voltage versus geometric parameters.
O' PARALLEL TO X

~MJ
o Lo.5 IN. PARALLEL TO Z

~I
I DISPLACEMENT

GROSS VERTICAL PLOT

I~
PARALLEL

~
I TOZ
~ PARALLEL TO X
PARALLEL TO Y

I I I I I I
~t
-so -60 -40 -20 o
I 20 40 60
I
so
80 -60 -40 -20 0 20 40 60 80 DISPLACEMENT (10- 3 IN. I
DISPLACEMENT (10- 3 IN .1
FrG. 7. Relative output signal from a double-coil system vs sample
FIG. 6. Relative output signal from a double-coil system vs sample position. Bottom curves show relative output vs displacement along
position. The contours of the "saddle point" are illustrated by meas- the three orthogonal directions in neighborhood of "saddle point."
urements along various directions in the XY plane. Circular detection Upper right insert shows relative output for large sample displacement
coils, each tin. diam and i in. long, with 1! in. between centers, were along the Z direction. Circular detection coils, each tin. diam and 1 in.
used. long, with 1 in. between centers, were used.

Downloaded 29 Jul 2011 to 130.65.109.155. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions
 VIBRATING-SAMPLE MAGNETOMETER 553
IV. ELECTRONIC CIRCUITS

The functions of the associated electronic circuits are:

(1) to permit accurate calibration of the signal output
obtained from the detection coils, (2) to produce a con-
venient dc output signal which is directly related to the
input and which can be recorded, (3) to provide sufficient
amplification for high sensitivity operation. These func-
tions can be performed by a variety of circuits.
A functional block diagram of the circuit used for the
magnetometer is shown in Fig. 9. The loudspeaker trans-
ducer assembly is driven at about 90 cps by a low-power FIG. 9. Block diagram of electronic system for magnetometer. Rele-
vant components of the magnetometer are indicated pictorially.
amplifier fed by an audio oscillator which is peaked at the
tuned amplifier pass-band. The combination of phase
shifter, divider, and mixer performs the function of a cali- division. The resultant voltage is then compared with the
brated ac bridge; the tuned amplifier and following circuits sample signal. In order accurately to balance the reference
then indicate any unbalance of the ac bridge, but also may signal against the sample signal, the two signals must be
incorporate a feedback circuit to permit continuous bal- carefully adjusted to be 180° out of phase. The difference
ancing. An oscilloscope is used as a convenient visual null between sample and reference signals is fed to the primary
indicator of the ac signal prior to detection-the rectified, of a well-shielded audio transformer. At balance, there is
filtered, dc output is presented on an X-V recorder. no current flow produced in this transformer primary cir-
The low-frequency and narrow band of operation permit cuit; and hence the reference divider circuit serves as an
standard circuits to be employed. For highest sensitivity accurate voltage divider of an ac potentiometer. The mag-
the input circuit must, however, be carefully constructed netic moment of the sample is then proportional to the
in order to avoid extraneous voltages of millimicrovolt reference voltage divider setting, and independent of vibra-
level. The reference permanent magnet signal furnishes the tion amplitude, frequency, or the following electronic cir-
standard comparison voltage across the reference coils. cuits which then only serve as a null-detector amplifier.
Although the amplitude of this signal may vary with vibra- The main sources of error of null measurements involve
tion amplitude, a proportional signal change is also pro- the calibration of the voltage divider network and the
duced at the sample coils, so that to high approximation standard comparison sample.
this effect can be eliminated. A calibrated, constant- A servo loop coupled to the reference voltage divider may
impedance, decade resistance-divider, incorporating a be employed to obtain null measurements of magnetic
linear (3600°, helical wire-wound) potentiometer, is moment. This complication can be avoided by balancing
employed in the reference signal circuit for precise voltage reference and sample signals at one point and recording the
amplified, rectified, unbalanced signal which is closely
proportional to the change of sample magnetic moment.
The output trace is then calibrated to compensate for any
nonlinear circuit characteristics by comparison with several
appropriate settings of the reference divider. Effects of
-5 unbalanced currents in the input circuits can be eliminated
by feeding the reference and the sample signals to high-
STANDARD-- impedance inputs of two balanced cathode-follower circuits.
o 234 5 If the reference sample is replaced by a small coil (indi-
Z (0.10IN.l-
cated by the dashed lines in Fig. 9) energized by an ad-
justable dc current, automatic recording of magnetic prop-
~I
OJ
01
erties can be made with the inherent advantages of null
measurements. The direct current is regulated by a servo-
-80 -6tl -40 -~O 6 1
20 40 loop monitoring the null output signal, and a voltage pro-
DISPLACEMENT (1O-3 IN.l
portional to the balancing direct current is recorded. This
. FIG. 8. Relative output sig,;al from a single c')il [Fig. 4(b)]. Upper system is essential for measurements of weakly magnetic
nght shows output vs Z" the dIstance between the center of the sample
and the ~01? r:f t.he cot!. Output level obtained with the standard materials because the long integrating time makes point-
double-cOli. IS I,nrhcated by the broker: lin~, Upp~r left shows output by-point balancing difficult. Such a system is ' of course ,
as ?ample IS dIsplaced along the X dIrectIOn whIle the Z position is .
mamtamed-no o~tp~t is obtained at the coil axis (X=Y=O). Ideal for magnetic measurements in general. In most cases
Bo~tom shows. sensItivIty to sa:nple ~isplaccTI?ent in the XY plane in we have found that the permanent magnet reference is less
regIOn .of ma.xlmum output. DImensIOns of smgle-coil correspond to
one COli of FIg, 6. complicated for measurements over extreme ranges of mag-

Downloaded 29 Jul 2011 to 130.65.109.155. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions
 554 SIMON FONER

netic moment, and is therefore more convenient for routine This method can be used effectively to cancel the induced
measurements. dipole moment in the sample when the axis of the cylinder
is parallel to the applied field. 12 The reference voltage cir-
v. OPERATING CHARACTERISTICS OF cuit is thus replaced by a dc current measurement for the
MAGNETOMETER equivalent canceling current shell. Of course, all the ad-
A. Calibration vantages of null measurements are still obtained. Absolute
measurements of magnetic properties as a function of
The reference voltage divider reading at null is directly crystallographic orientation can be obtained by using
proportional to the sample magnetic moment and to the several cylinders suitably oriented. The current shell
coil conversion factor. As mentioned in Sec. III, the method eliminates the need for the reference coil system;
induced voltage has not been calculated in closed form for the sample coils then function as part of the null detection
many of the useful coil arrangements. In principle this can circuit by indicating the unbalance between the current
be done so that an absolute measure of magnetic moment shell and magnetic moment. Unfortunately, the practical
may be calculated. It is doubtful, however, that the accu- limiting currents, sample shapes, and complexity, limit
racy of such a calculation will approach that of a compari- this method to special problems with weakly magnetic
son method. An expedient method of calibration involves materials.
standardizing the reference voltage divider by replacing The absolute accuracy of the instrument depends on the
the sample with a material of known magnetic properties knowledge of the magnetic properties of the calibration
and mass. In this way, one can easily calibrate the magne- standard and reproducibility of sample position. When the
tometer for any useful sample-coil configuration. The satu- substitution method of calibration is used, the major error,
uration moment of a small sphere of pure nickel (about ±1%, is introduced by the estimate of the nickel-standard
8 mg) has been used as a secondary calibration standard magnetic moment. The relative accuracy of this instrument
for most measurements. The two major reasons for choos- depends on accurate calibration of the precision resistor
ing nickel are that it can be obtained in high purity, and divider network. The total error here can be kept to less
that magnetic saturation of nickel is easily attained at low than 0.5%
magnetic fields, so that the calibration is not dependent on Repeated tests showed that when the sample coils were
the applied magnetic field. repositioned as described earlier, the output signal was
Calibration for very weakly magnetic materials is some- reproduced to 0.25%. Recalibrations of the coils, main-
what more difficult because the sample may occupy a com- tained with a fixed spacing, showed that calibration could
paratively large volume of irregular shape whereas small be maintained within ±0.5%.
samples act as simple dipoles and size effects are negligible.
Calibration for such large samples can be accomplished by
B. Sensitivity Limits
measurement of the sample susceptibility in vacuo and then
in pure oxygen gas at standard temperature and pressure. The limits of sensitivity are determined by signal to
The difference of the two measurements determines the noise ratio at the input circuit where noise is defined as any
susceptibility of the vibrating hole in the O2 gas (occupied signal not due to the sample magnetic moment. The major
by the sample) which is just the negative of the well- sources of noise arise from the Johnson noise of the wire
known susceptibility of O2• By this substitution method used for the pickup coils, and from the magnetic properties
an accurate correction for higher order shape effects of of the sample holder which, of course, superimposes an
large samples is made without recourse to tedious, less in-phase signal on the wanted signal. Use of a minimal mass
accurate computations. One expects such shape effects to of weakly diamagnetic material for a sample holder, care-
become important when the sample dimensions approach fully checked to contain no ferromagnetic impurities, is
those of the coil spacings. As an example, tests with para- essential to minimize this coherent noise contribution. Cor-
magnets of irregular shapes with linear dimensions up to rections for the small magnetic contribution of the sample
about 3 mm were made. The magnet moments agreed holder can then be made by measurements with the sample
within one percent for these cases, indicating the shape removed. Obviously, this correction is much less than the
effects were still small. equivalent case with a moving coil system.
Absolute calibration of a particular coil configuration Tests with the double-coil system [Fig. 5(a)] showed
can be made by vibrating a sufficiently small direct current that the detectable signal was within a factor of two of the
loop of known area-turns in place of, or surrounding, the theoretical thermal noise of the coil winding at room tem-
sample.
perature. Further improvements of signal to noise can be
If the sample is weakly magnetic, but of uniform cross
section (e.g., a cylinder) one can wind a coil on the sample 12 A. Arrott and J. E. Goldman, Rev. Sci. Instr. 28, 99 (1957)
employed such a technique with a moving sample. Here again, the
to produce a current shell of equivalent magnetic moment. sample motion was parallel to the applied field direction.

Downloaded 29 Jul 2011 to 130.65.109.155. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions
 VIBRATING-SAMPLE MAGNETOMETER 555

made in several ways. The thermal noise of the detection about one-tenth this size could be recorded reproducibly.
coils can be reduced substantially by immersing them in These results demonstrate that the average stability of the
a low-temperature bath, at a sacrifice of magnetizing field instrument is exceptional.
and convenience in adjustment; Some gain can be obtained
by optimizing coil geometry. As an example of an extreme D. Vibration Amplitude
case, a gain of about 10 in signal and a decrease in resistance
may be attained by using the less convenient arrangement The peak-to-peak vibration amplitude has been varied
of a single coil [see Fig. S(b) and Fig. (8)]. Finally, some from less than 0.1 mm up to 1.0 mm in order to examine
additional gain can be obtained by increasing the vibration errors caused by amplitude changes. Such tests show that
amplitude of the sample. Peak amplitudes of 1 mm can be the measured magnetic moment varied by less than ±O.S%
obtained without damaging the loudspeaker, whereas this over this range of amplitude, although a somewhat sharper
amplitude usually is kept near 0.1 mm. In principle, no balance is obtained at higher vibration amplitudes because
gain is expected by increasing the number of area-turns per of the larger signals involved. Measurements usually are
unit volume of coil if we assume impedance matching to made with vibration amplitudes of about 0.1 mm. Insen-
the input circuit and a constant space-factor are maintained. sitivity to small amplitude variation is also demonstrated
Several methods were used to estimate the ultimate sen- by results of Sec. Vc.
sitivity of the vibrating sample magnetometer with the
double-coil configuration. Because it would be difficult to E. Image Effects
insert a sample of a very small, but known, susceptibility When a magnetic material is placed near a highly per-
without also introducing background effects of the sample meable medium, a magnetic moment is induced in that
support and other impurities, differential methods were medium. This effect, called the "image effect," may be
used. The simplest measurement used the standard nickel eliminated by using the absolute current shell calibration
sample. A change in susceptibilitY,13 .:lX, of 5 X lO-9 could technique (Sec. VA), so that no net moment is present
be detected at the oscilloscope, and .:lx~5X 10-10 could be when measurements are made, by using sufficiently large
observed after synchronous phase detection (band width air gaps, or by using air-core solenoids. As mentioned
~2X lO-2 cps). The other tests used a small current carry-
earlier, the current shell technique is not useful for strongly
ing coil mounted in place of the sample. Either a direct magnetic materials because impractical currents are re-
current was passed through this coil as it was vibrated at quired to balance the induced moment. Estimates of the
90 cps, or an alternating current at 90 cps was passed image effect can be made by observing the measured mag-
through the coil which remained stationary. In both cases, netic moment when the sample and coils are positioned at
.:lx~2X lO-10 was detectable with optimum conditions and
points far from the central plane of the magnet air gap. The
synchronous detection. Probably a factor of 10 improve- largest effect should then be observed when one of the coils
ment can be expected with minor modifications. These almost touches a pole face. Such measurements with the
results compare favorably with some of the most sensitive nickel standard sphere have been made using an air gap of
force methods. A particularly high sensitivity force method, about 2 in. and fields up to 8 kilogauss. An image effect
developed by Stevens and Crawford14 permits x~2X lO-10 was not detectable and is thus estimated to be less than
to be measured. 0.2%. It is expected that an image effect will be more im-
portant for smaller air gaps, in which case such measure-
c. Stability Tests-Differential Measurements ments can be used to obtain suitable corrections.
With only the tuned amplifier (band width = 1 cps) and Image effects were also examined with a small vibrating
the oscilloscope as a null detector, it was found that the coil carrying a dc current. The image effect was no greater
8 mg Ni sample signal could be balanced reproducibly to than ± 1% for fields up to 18 kilogauss produced in an air
one part in 8000. Such reproducibility indicated that the gap of 1 t in. Undoubtedly, there is an image induced in the
long time drifts caused by the combined effects of vibration magnet poles. It appears, however, that when the sample
amplitude changes, frequency changes, varying average is vibrated, the effective image vibration is reduced by
sample position, and any other effects not considered here eddy-current shielding.
were indeed negligible. When a synchronous phase detector
was added (band width ~2X 10-2 cps) differential changes F. Vibration Frequency
13 For convenience whenever values of X are considered, the units The vibration frequency is not critical. High-frequency
of X are cgs units/g normalized to correspond to the change detectable operation is limited by the driving mechanism and capaci-
for a 1-g sample in an applied field of 10 kilogauss.
14 D. K. Stevens and J. H. Crawford, Jr., Phys. Rev. 92, 1065 tive shunting in the detection coils. Frequencies of 100 cps
(1953). Specifically they report X= 10-10 cgs units/g can be measured or less permit use of inexpensive components and minimize
for a l-g Ge sample when H = 18 kilogauss and aH /ax"'3 kilo-
gauss/em. eddy-current shielding by the vacuum chamber. The meas-

Downloaded 29 Jul 2011 to 130.65.109.155. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions
 556 SIMON FONER

urements are completely independent of eddy-currents in are mounted with the (110) plane perpendicular to the
surrounding parts if measurements and calibration are vibration direction so that measurements as a function of
made at the same temperature. A correction for changes H along the three principal crystallographic directions may
of penetration depth with temperature may be required if then be made. 16
the conducting parts are thick. The thickness of conducting
parts has been minimized, so that the temperature depend- 2. High Conductivity Materials
ence of penetration depth is less than 1%. Finally, low
If the material is strongly magnetic, usually negligible
frequencies may be desirable for measurements below 10 K
in order to minimize introduction of phonon energy. corrections need be made for the diamagnetic effect of
currents induced when the sample vibrates in a nonuniform
G. Vibration Problems field. However, many high-conductivity metals are weakly
magnetic and these induced currents will then add large
Mechanical coupling between the vibrating system and contributions to the magnetic moment. Such effects can be
the fixed detection coils must be avoided. Although the minimized by vibrating the sample in a uniform region of
coils are arranged for minimum sensitivity to external magnetic field, and, as a last resort, by laminating the
vibration, a noticeable background signal is obtained when sample. Even in a uniform field, a voltage is developed in
the vacuum chamber contacts the detection coils. Such a moving conductor. The effect is analogous to the Hall
mechanical effects are difficult to eliminate electronically effect where we replace the primary current by the moving
because the spurious background signal has the same fre- conductor. The resulting currents, perpendicular to both
quency as the sample signal and maintains a constant phase the vibration direction and the applied field, would produce
difference with respect to the sample signal. The unwanted a magnetic field configuration which to first order would
background signal may also be field-dependent if the coils be completely canceled in each of the pickup coils for the
subtend a region of nonuniform field. Usually the magne- standard configuration. Furthermore, these currents are
tometer and detection coils are both supported by the small and can be reduced further by suitable lamination of
magnet, so that some mechanical coupling may be noticed the sample if necessary. Susceptibility measurements on a
at highest sensitivity. This effect can be eliminated by solid cylinder of high purity (99.999%), oxygen-free,
shock-mounting either the magnetometer or the detection copper from 77 to 300 0 K demonstrated that these effects
coil system as required by the particular experiment. Rigid were indeed negligible. If these currents were appreciable,
clamping of the detection coils to the magnet pole faces their magnetic effects could be observed with an appro-
has also been successful. priate coil arrangement and the Hall effect deduced.

VI. APPLICATIONS B. Measurement of Magnetic Field

A. Magnetic Measurements The magnetometer may be used as a field measuring
1. Low-Conductivity Materials device by using a paramagnetic sample. If the suscepti-
bility is not too large and is field independent, the pro-
The vibrating-sample magnetometer has been in use for portionality constant, accounting for the particular coil
routine magnetic measurements as a function of tempera- geometry and the susceptibility, is obtained by calibration
ture and field of ferromagnetic, ferrimagnetic, antiferro- with the nickel standard above magnetic saturation, or by
magnetic, paramagnetic, and diamagnetic materials for an independent field measurement. The field measured is
over three years. I. The samples are first weighed, then an average value over the sample volume (which can be
glued to a standard sample mount. The magnetic moment made small). A relative accuracy of about 1% is obtained
per gram is determined, and from this result conversion to with the double-coils; the absolute accuracy depends on
other appropriate units can be made with additional infor- the calibration method.
mation. Samples can be inserted in the instrument and Extremely small field volumes can be examined by vi-
saturation moment measurements completed in about a brating a small thin flat disk such as a thin film of
minute. Spherical or approximately ellipsoidal sample Permalloy in this region. In this case, the low anisotropy,
shapes are used for magnetic moment versus field measure- high permeability, accurately known demagnetizing factor
ments of ferri- or ferromagnetic materials, so that accurate (471-), and high saturation magnetization permit measure-
corrections for demagnetizing fields can be made. Measure- ments of fields up to 10 kilogauss when the field is perpen-
ments on single crystals are generally made with spherical dicular to the film. The magnetic moment is then propor-
samples of constant demagnetizing factor, although thin tional to the applied field. Small fields may be measured
disks are also satisfactory. Crystals with cubic structure when the film is placed parallel to the applied field. A small
additional dc solenoid around the sample can then be used
15 See, for instance, S. Foner and J. O. Artman, J. Appl. Phys. 29,
443 (1958). to produce a variable field which almost balances the un-

Downloaded 29 Jul 2011 to 130.65.109.155. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions
 VIBRATING-SAMPLE MAGNETOMETER SS7

known field. Field gradients can be measured accurately form can be quite inexpensive and yet of quite high
when the sample is replaced by a small coil. accuracy.
This inexpensive version can also be useful for teaching
C. General Purpose Instrument purposes in general physics. The magnetic properties of
materials such as demagnetizing factors, anisotropy, mag-
Many of the refinements presented here for this vibrating
netization processes, and magnetic saturation, can be con-
sample magnetometer were directed toward the develop-
veniently demonstrated. Examination of various useful coil
ment of a high sensitivity, versatile, laboratory instrument.
configurations for this instrument is also very instructive.
A production line instrument, useful for quality control
measurements of ferri- or ferromagnetic materials at room
ACKNOWLEDG MENTS
temperature, can be made with the minimum elements of
Fig. 1. Since the magnetic moment will be relatively large, The author is indebted to Mr. B. Feldman and
the simplest of detection systems can be used; 60 cps can Mr. E. D. Thompson for their valuable comments, sug-
be usedJfor a driving frequency, and tolerances are not gestions and assistance during the development and use of
critical. The magnetizing field can be supplied by a small this instrument, and to Mrs. Jane Coe for assistance with
permanent magnet. A production line instrument in this numerical computations.

THE REVIEW OF SCIENTIFIC INSTRUMENTS VOLCME 30. NUMBER 7 JULY. 1959

Low-Temperature Adiabatic Calorimeter with Automatic Shield Control*

W. J. COOPER, R. H. FORSYTH, J. F. MASI, AND R. E. PABST
Callery Chemical Company, Callery, Pennsylvania
(Received April 13, 1959)

The design and operation of an adiabatic low-temperature calorimeter employing a somewhat different design
of automatic shield control is described. Thermocouple difference is sensed by a dc p,v amplifier, which is followed
in the circuit by a recorder-controller and a Leeds & Northrup Series 60 (current adjusting type) control unit. The
output of the latter is used to control the voltage across the shield heaters through a Westinghouse "Magamp"
saturable reactor. This design has been used successfully with a conventional low-temperature adiabatic calorimeter
(50-ml sample size). In most cases the recorder showed no variation from zero (±0.002°) during heating and
equilibrating periods, while momentary disturbance at "on" and "off" was less than 0.05°.

I. INTRODUCTION II. CALORIMETER

T O--:carry out our program of obtaining thermo-

dynamic properties of various materials of interest,
The calorimeter is of the conventional adiabatic vacuum
type, similar to that described by Southard and Brick-
a calorimeter capable of medium accuracy was needed for wedde. 5 The cylindrical, SO-ml sample container is con-
those materials not sufficiently pure or stable to be structed of thin-walled copper and contains eight radial
appropriate for the laboratories engaged in high-accuracy vanes for heat distribution. A false bottom of much
work. For this purpose it was decided that a conventional thinner copper acts as a radiation shield for the calorimeter
adiabatic calorimeter with automatic shield control would heater and resistance thermometer. The capsule type
best fill our needs. platinum resistance thermometer and Manganin heater
The particular design of shield control was decided upon are held in a central well by eutectic solder, which provides
after consideration of the following requirements: (1) the good thermal contact. The sample container is tinned
components should be commercially available wherever internally and gold plated externally and is suspended by
possible, and (2) the circuitry should be highly stable and nylon thread in the adiabatic shield. Both sample con-
require a minimum of maintenance. tainer and shield are surrounded by a brass vacuum
The general aspects of automatic shield control have jacket. A cryostat capable of maintaining temperatures
been thoroughly discussed in recent years by a number of from SO° to 370 0 K, is placed around the calorimeter.
authors l - 4 and will not be repeated here.
III. MEASURING CIRCUITS
* Taken frcm paper presented at Twelfth Annual Calorimetry
Conference, Portsmouth, New Hampshire (September, 1957). The circuit for measuring the electrical energy supplied
1 T. M. Dauphinee and S. B. Woods, Rev. Sci. Instr. 26, 693
(1955) . to the calorimeter and for measuring its temperature are
2 Worthington, Marx, and Dole, Rev. Sci. Instr. 26, 698 (1955).
3 Zabetakis, Craig, and Sterrett, Rev. Sci. Instr. 28, 497 (1957). 5 ]. C. Southard and F. G. Brickwedde, ]. Am. Chern. Soc. 55,
4 E. D. West and D. C. Ginnings, Rev. Sci. Instr. 28,1070 (1957). 4378 (1933).

Downloaded 29 Jul 2011 to 130.65.109.155. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions