You are on page 1of 13

Solid-State Electronics Pergamon Press 1966. Vol. 9, pp. 339-351.

Printed in Great Britain

THE HALL EFFECT AND RELATED PHENOMENAt


A. C. BEER
Battelle Memorial Institute, Columbus Laboratories, Columbus, Ohio, U.S.A.

A b s t r a c t - - A review is presented of the theory of the Hall effect and related phenomena in solids.
Basic principles are illustrated by application of the Lorentz force to moving charges in a free-electron
system. Application to a real solid is indicated through a consideration of the effects of band structure
and charge-carrier scattering mechanisms. Association is also made with the off-diagonal elements of
the conductivity tensor, and the influence of certain symmetries in the medium is discussed. Special
mention is made of specific parameters useful in characterizing a material for Hall-effect applications,
including the dependence of the parameters on temperature and impurity content.
T h e question of the variation of Hall coefficient with magnetic field is examined in some detail, with
particular attention to band structure and crystal anisotropy, scattering, inhomogeneities and multi-
band effects. Differentiation is made between isothermal and adiabatic phenomena, and the import-
ance of various thermomagnetic contributions is pointed out. Related galvanomagnetic phenomena
which are discussed include magnetoresistance effect and also the magnetic-field dependence of the
skin depth in the case of the transmission of a.c. fields through a disk.

R 6 s u m 6 - - - O n pr6sente une revue de la th~orie de l'effet Hall et des ph6nom~nes alli6s dans les
solides. Les principes fondamentaux sont illustr~s par l'application de la force Lorentz aux charges
mobiles dans u n syst~me ~t ~lectron mobile. L'application h u n solide r~el est indiqude par une
consideration des effets des structures de bande et des mdcanismes d'~parpillement des porteurs de
charge. U n e association est aussi faite avec les 6Mments non-diagonaux du tenseur de conductivit~
et l'influence de certaines sym&ries dans le milieu est discut~e. Une mention sp~ciale est faite des
param~tres spficifiques utiles pour caract6riser u n mat~riau pour applications h effet Hall, comprenant
la d6pendance des param~tres sur la temperature et la quantit~ d'impuretC
La question de la variation du coefficient Hall avec le champ magn~tique est examin6e en d&ail,
et en particulier la structure de bande et l'anisotropie de cristaux, l'#parpillement, les inhomo-
g~neit~s et les effets de bandes multiples. O n diffdrentie entre les phdnom~nes adiabatiques et
isothermes et l'importance de diverses contributions thermomagn~tiques est indiqu6e. Les phdno-
m~nes galvanomagn~tiques alli6s qui sont discut~s comprennent l'effet de magn~toresistance et aussi
la dfipendence d u champ magn~tique sur la profondeur de peau dans le cas de la transmission de
champs h courant alternatif h travers u n disque.

Zusammenfassung--Eine/]bersicht fiber die Theorie des Hall-effektes u n d verwandter Erschein-


ungen in Festkfrpern wird gegeben. Z u r Veranschaulichung der grundlegenden Gesetzm~issigkeiten
wird die Wirkung der Lorentzkraft auf die bewegten Ladungen in einem System freier Elektronen
dargestellt. Die Anwendung auf reale Festk6rper wird angedeutet, indem die Wirkungen von
Bandstruktur u n d Ladungstr~gerstreumechanismen beriieksidhtigt werden. Auch wird der Zusam-
menhang mit den nichtdiagonalen Elementen des LeitfLlaigkeitstensors hergestellt u n d der Einfluss
gewisser Kristallsymmetrien besprochen. Ausdriickliche E r w ~ m u n g finden spezifische Parameter,
welche geeignet sind, ein Material zu kennzeichnen, das flit Anwendungen des Hall-effektes in
Frage kommt, einschliesslich der Abhiingigkeit dieser Parameter yon Temperatur u n d Verunreinig-
ungsgehalt.
Auf die Frage der Veriinderung des Hallkoeffizienten mit dem Magnetfeld wird etwas eingegangen.
Besondere Beachtung finden dabei Bandstruktur u n d Kristallanisotropie, Streuung, Inhomogeni-
t~ten u n d Vielbandeffekte. Zwischen isothermen u n d adiabatischen Erseheinungen wird unter-
schieden u n d die Bedeutung einer Reihe yon thermomagnetischen Effekten ldargemacht. Die hier
diskutierten verwandten galvanomagnetischen Effekte sehliessen den Widerstand im Magnetfeld
und auch die Magnetfeldabh~ngigkeit der Eindringtiefe ein, auf welche es bei der Transmission yon
Wechselfeldern d u t c h eine Scheibe ankommt.

t Partial support by Air Force Office of Scientific Research through Grant No. 525-66 is gratefully acknowledged.
339
340 A. C. B E E R

1. PRINCIPLES INVOLVED An interesting and important consequence


THE HALLEFFECTin a solid is a direct manifestation results from the particular form of the force term
of the Lorentz force on moving charge carriers.* due to the magnetic field. It will be recalled from
The total force on a charged particle in an electric elementary physics that the rate at which work is
and magnetic field is given by the vector equation done on a particle, that is, the rate of change of
F = e[E+ (1/c)v x H ] (1) energy, is given by
where F is the force, e the charge and v the velocity dW/dt = F.v (3)
of the particle; E is the electric field and H the
magnetic field intensity. A permeability of unity It is now apparent from equation (1) that the
is assumed, and for the Gaussian system of units magnetic field does not change the energy of the
c is the speed of light. charge carriers, since the scalar triple produce
To illustrate the physics involved, consider a v . v x H vanishes. Thus a quasi-static magnetic
confined stream of free particles, each having field does not supply energy to the system. This
charge e and initial velocity vx. A magnetic field statement does not, of course, take account of
in the z direction produces initially a deflection of high-frequency dissipative effects.
charge along the y direction. Thus a charge un- It will also be recalled that the electric current
balance is created, and this results in an electric density J~ can be expressed in terms of the density
field E v. This field, which is the Hall field, builds of charge carriers by the relation
up until the force it exerts on a charged particle
J~ = ~ n~evx<o (4)
counterbalances that resulting from the magnetic
field. Then subsequent particles of the same charge
and velocity are no longer deflected, and a steady where n, is the density of charge carriers having
state exists. A pictorial representation of the situa- velocity vx(o. In the case of our illustrative model,
tion is shown in Fig. 1. where all particles were assumed to have the same
velocity, one obtains
Jl+ Jx Yz = nevx, where n = ]~ nf. (5)
E¥~g HZ
e EX With the use of equation (5), expression (2) for the
FIG. 1. Hall field due to action of magnetic field on Hall field becomes
positive charge carriers (holes).
E v = J~-Iz/nec = - (1/nec)(J x H)v
T h e magnitude of the Hall field follows at once
from a consideration of the force balance in the = (1/nec)(Hx J)y (6)
y direction, namely Now the Hall coeflicient is defined by the ratio
Ev = - - ( l / c ) [ v x H ] u = (1/c)vxHz (2) E v / J x H z , namely
If, as in a real solid, all particles do not have a
R =-- E v / J z H z = 1/nec, (7)
constant velocity vz, but there exists a distribution
of velocities, then vx must be replaced by a certain where the equality follows from equation (6). Its
"average" velocity. Thus it is only particles of this derivation therefore assumed vx to be constant.
average velocity which will traverse the solid un- We note from equation (7) that R is negative for
deflected. The other particles undergo deflection conduction by electrons and positive for conduc-
---some in the positive y direction and others in tion by holes.
the negative y direction, depending on the relative
magnitude of their velocities as compared to this 2. APPLICATION TO A REAL SOLID
average. This point will be discussed later in con- In a real solid our idealized free particle treat-
nection with magnetoresistance. ment no longer applies, and it must be replaced
* For a more detailed presentation of galvomagnetie by a theory that takes into account the distribution
phenomena and of the theory of solids, or for more in- of velocities and the interaction of the charge
formation on semlcondueting materials, consult Refs. 1-6. carriers with impurities, defects and lattice
T H E H A L L E F F E C T AND R E L A T E D P H E N O M E N A 341

thermal vibrations of the solid--i.e, the scattering, This angle is called the Hall angle, namely
as well as the band structure of the solid. The
latter consideration arises from the fact that the 0 - tan -:t Ey/Ex. (9)
charge carriers are not really free, but exist in a Thus the Hall effect can be described as a rotation
potential energy field having the periodicity of the of the electric field--a concept which is very useful.
lattice. As a result of these constraints, only At zero Hall field, the equipotential lines are
certain energy states, or bands, are allowed for the perpendicular to J, but when the Hall field appears,
charge carriers. Furthermore, the relationship they are oblique, so that a Hall voltage exists across
between energy and velocity of the carriers is not the specimen in a direction normal to the current.
the simple ½mvZ of the free electron, but is more The rotation aspect of the Hall effect is also
complex. As an approximation, one frequently brought out by considering the components of the
characterizes the charge carriers by an effective conductivity tensor, which relates electric current
mass, m*. In general, m* is anisotropic, i.e. it is densities and fields. For the boundary condition
a tensor. It may also vary with energy of the that Jz = 0, we may write
carrier.
Taking into account most of the complexities J~ = zzxEx+ ~rzyEu
just mentioned, one still obtains--when one type Jz=0, H=Hz
of charge carrier is present--an expression similar
to equation (7), namely Ju = - CrxuEx+ ~r¢xEy" (10)

R = r/nec. (8) The above equations hold for media of sufficient


symmetry that azx = avy and that crxz = ayz = O.
The parameter r, called the Hall coefficient factor, This is obviously satisfied in isotropic media, and
depends on the nature of the scattering, the band it also holds for cubic systems when the coordinate
structure, the magnetic field strength, and on the axes are along the cube axes. The fact that for
statistics characterizing the distribution of veloci- certain conditions of symmetry one can obtain the
ties of the carriers. Its functional dependence relation
on these variables is usually slight, and in most ~zy(H) = - ~yz(H), (11)
cases its value differs from unity by less than + 50
per cent. The fact that r approaches unity in the as is implicit in equations (10), indicates that a
limit of strong magnetic fields, provides a means Hall effect unit can be used as a non-reciprocal
for determining it experimentally if measure- electrical element, which has been called a
ments can be made into the strong-field limit. "gyrator".(7)
Another point to note is that r is unity for very Although the validity of relation (11) is depen-
large carrier concentrations--i.e, large degeneracy. dent on a certain symmetry in the solid, it should
This might be expected since in such cases only be pointed out that the general Onsager relation-
those charge carriers with velocity in the neigh- ships( s ) are applicable to the elements of the
borhood of the Fermi velocity contribute signifi- conductivity tensor, regardless of the degree of
cantly to conduction, and we have the character- isotropy of the material, and they yield the general
istics of the single-velocity picture discussed relation
cartier. The variation of r with magnetic field is, *,k(H) = *k,(--H). (12)
of course, of interest in connection with Hall effect
applications, and this point will be discussed in Two useful items of information follow from the
more detail later. above Onsager relation, namely

3. ELECTRIC FIELDS, H A L L A N G L E A N D (a) cq~(H) = ~ u ( - H ) (13)


H A L L MOBILITY
We note that the electric current density Jz and provided relation (11) holds,
exists as a result of an applied electric field Ex. (b) crzy(H) = - ~zu(-I-I). (14)
With the Hall field present, the resultant electric
field in the solid lies at some angle 0 to the x axis. That is, the diagonal elements of the conductivity
342 A . C . BEER
tensor contain only even powers of H; the off- ality is the Hall coefficient, as was defined in
diagonal elements, in cases of adequate symmetry, equation (7) or (16). Thus we may write.
contain only odd powers of H.
Since the boundary conditions for Hall coeffi- EH = R t t × J, (19)
cient measurement require that Jv vanish, we note
where EH is an odd function of H. The same
from equations (9) and (10) that
relationship, of course, follows from the free-
~zv = ¢rxxtan O. (15) particle picture and equations (6) and (7). The
operational definition of R is obviously a quantity
Also, from the definition of the Hall coefficient as which is odd in H, since its measurement involves
given in equation (7), we see from equation (10) taking half the difference of E(H) and E ( - H ) along
with Jy = 0 that the Hall probes. Any contributions which are even
in H therefore cancel.
R = ~rxv/Hcrxx~ (16) For isotropic media, or for certain cases where
in weak magnetic fields. H is along a direction of high symmetry, the total
An important attribute of charge carriers is their electric field measured along the direction H x J is
mobility, that is, in the case of conductivity precisely EH, as given above. In a more general
mobility, the drift velocity per unit electric field. case, however, if H is along a lower symmetry
The conductivity mobility is thus related to the direction in an anisotropic crystal, the field across
conductivity ¢r by the Hall probes will also have a component which
is even in H. T o see how this comes about, we
i~ = ~r/ne, (17) note that the resistivity tensor [the inverse of the
conductivity tensor of equations (10)] can be
in the case of conduction by a single type of carrier. written as a sum of antisymmetric and symmetric
It is also customary to define a Hall mobility tzH, parts. Thus we have:
again for conduction by a single type of charge
carrier, by the relation E = p(H)J = pa(H)J+ps(H)J. (20)
That is,
I~HH/C = tan 0, or/ZH = Rac. (18)
p,k(H) = p,~a(H)+pf~8(H), (21)
With the additional use of equations (8) and (17),
where, by definition,
it can be seen that the ratio of Hall and conductivity
mobilities is precisely the Hall coefficient factor r. m~a(H) = --pk,a(H), (22)
From equation (18), it also follows that the Hall
angle is proportional to the mobility as well as e,ks(H) = pk,8(H). (23)
to the magnetic field. This is an important con- Since, by equation (22), pua(H) = 0, it follows that
sideration, since it tells us that to have a large
all diagonal elements are symmetric. Now by
Hall effect--or magnetoresistance effect, for that Onsager's relations--cf, equation (12)--the above
matter--at reasonable magnetic fields we must
yields:
favour those materials with large charge-carrier
mobilities. As a matter of fact, the condition p~ka(H) = - p , . k a ( - H ) , i.e. p~,~a(H) is odd in H
~HH[c ~ 1 is usually used as a criterion to express
(24)
the weak magnetic field region, while ixnH/c >> 1
denotes the strong field region. p~,~s(H) = p ~ S ( - H ) , i.e. p~S(H) is even in H.
(25)
4. VECTOR REPRESENTATION OF HALt, FIELD
Since the Hall field, as the term is used here, is The field across the Hall probes is determined by
by definition in a direction perpendicular to H and the off-diagonal elements--that is, i # k in equa-
to J, and is proportional to the magnitudes of tions (24) and (25)--of the resistivity tensor, and
these quantities, it must be proportional to the we have
vector product H ×J. The constant of proportion- EH× J = R H × J+(psJ)H× J (26)
T H E H A L L E F F E C T AND R E L A T E D P H E N O M E N A 343

In going from equations (20) to (26) we have resistance in the Hall circuit as
expressed the antisymmetric components of the
resistivity tensor by the vector product R H x J . Z~n~ = Kpb/Lt, (27)
This is a special case, adapted to our geometry, where p is the resistivity and b, L, t are respectively
of a more general relation given by CASIMIa.* the width, length and thickness of the Hall element.
Now the first term on the right hand side of According to S~ER and coUaborators(lS), the value
equation (26) results from the off-diagonal terms of K is usually between 2 and 5.
of the resistivity tensor which are antisymmetric The Hall voltage V~ is given, in practical units,
in their indices, and which, we saw, are odd in H. by
This is the familiar Hall field. The second term, VH = MO-aRHI/t (volts) (28)
which results from off-diagonal terms which are
symmetric and therefore even in H, is associated where ?, is essentially unity, differing only for
with a magnetoresistance effect. The term vanishes geometries where contact shorting is significant.
under conditions of sufficient symmetry, as was Now for maximum power transfer, the external
illustrated by equations (10). In the absence of resistance must equal Z~nt, so that the expression
adequate symmetry, however, it follows that the for the power in the load becomes
Hall probes can register an admixture of Hall
Wo = VH2/4Z~nt. (29)
voltages (odd in H) and magnetoresistance voltages
(even in H). Since, under such conditions, V ( H ) The power supplied to the specimen is
is no longer equal to - V ( - H ) , it follow that for
most Hall effect devices, such a situation must be W~ = I~pL/bt (30)
avoided. An example of the admixture of Hall and and hence, in practical units,
magnetoresistance voltages is given in Section 6.
Wo = ( 1 0 - 1 6 a ~ / ~ ( R ~ / p ~ ) W ~ ~ ~ 2 ~ W ~ .
5. OPTIMIZATION OF MATERIALS PARAMETERS
A number of considerations enter into the criteria (31)
for the characteristics of the best material for a Thus for maximum power, we are interested in
given Hall effect application. Many of these will highest mobility materials.
be discussed in subsequent papers. Here we shall For case (b), where maximum output voltage is
devote our attention only to the Hall element-- desired, we have, from equations (28) and (30):
excluding the magnetic circuit--and shall consider
two contrasting examples which illustrate some of V~ = AIO-SRIII/t
the major factors to be considered. These cases = IO-SA(b/Lt)II2(RH/pll2)W,I/~ ,,~ (I~R)ll2HW, ll2.
are (a) maximum power output and (b) maximum
voltage. In both cases, the constraint on the input (32)
current to the element is that imposed by a certain For maximum output voltage, the figure of merit
maximum heat generation rate. for the material is therefore the quantity R[~/p
An exact calculation of maximum output voltage or V'(/~R). Representative values of these para-
or power from the Hall element involves, of course, meters of merit for various materials are given in
the solution of the field equations for the geometry Table 1. It is apparent from the table that if the
of the system which includes the active element operating requirement is maximum output power
and the contacts. Various aspects of the problem --without further limitations--then InSb is the
have been considered by a number of investigators obvious choice; if the requirement is the highest
using, for the most part, conformal transforma- no-load output voltage, then silicon is the fore-
tions.(9-n) Several conference papers are to be front contender J15,16) In practice, other constraints
devoted to this subject.(lg) For our purposes here, are usually imposed. An important one is that the
we shall follow the procedure adopted by SAKER temperature coefficient be small. This means--as
and co-workers(13) and absorb these geometry can be seen from the last column in Table 1 -
factors in a parameter K, writing the internal that intrinsic or nearly intrinsic materials must be
* See, for example, pp. 41-43 and 83-85 of Ref. 1. avoided. Thus, in the case of InSb, the specimen
344 A . C. B E E R

Table 1. Representative values of materials parameters at 3 0 0 ° K t

o,
Material cm3/C f~-cm cm-s cm2/V-sec R/p1~2 klt~2§ kz(R/pl/a)§ . ~o]°C

n -- Si 360,000 200 2 x 10la 1500 25,000 0.09 120 <0.01


7000 4.5 1 × 10t5 1300 3300 0.07 16 <0.01

n-Ge 87,000 25 5 x 10la 36001] 17,000 0'5 85 4


21,000 5 3"5 × 1014 3600 9000 0.5 45 0.3
4250 1 1 '7 x 1015 3500 4000 0'5 20 <0"01

GaAs 1700 0-2 3"7 x 10*5 8500 3800 3 17 0.1


15 0.0045 4.2 x 1017 3300 220 0"4 1 '1 0.01

InAs 3700 0.1 1 '7 x 10x5 35,00011 12,000 50 60 3


570 0.02 1 "1 × 1016 28,000 4000 30 20 0.1
115 0.005 5"5 x 1016 23,000 1600 20 8 <0"01

InSb 380 0.005 2 × 1016 75,00011 5400 225 27 2


19 0.0006 3 "3 x 10t7 32,000 770 40 3 "8 0.01

Bi 5-10 ,~0.002 ,~6 × 1017 ~5000 200 1 1 ~-~0.4

t Data from Ross et al.( TM and from unpublished results at Battelle.


Hall coefficient factors of approximately 1 "18 were used for Si and Ge; lower values for the I I I - V materials.
§ The normalization factors kl and k2 are chosen so that the values of the respective figure of merits are unity in
the case of bismuth.
II Denotes that specimen is intrinsic or nearly intrinsic.

must be doped into the moderate 10 t7 region.(17) a decrease by a factor of 2½ from good bulk values, t
T h e mobility is then degraded by impurity scat- F r o m T a b l e 1 it is apparent that because of the
tering to a value which is not much different from higher resistivity of InAs, the material could be
that in InAs of the required doping. Another almost an order of magnitude thicker to yield the
consideration, which is especially pertinent in the same resistance. T h e mobility in the element would
case of I n S b , is the resistance of the Hall unit. I n be essentially the bulk value. F o r these reasons,
most circuits this needs to be the order of ohms I n A s is often preferred to InSb(20), in spite of the
rather than fractions of ohms. T o achieve such fact that the m a x i m u m figure of merit is higher
values with a material of resistivity as low as that for the latter material.
in the doped InSb, requires very thin s p e c i m e n s - -
perhaps the order of a few microns. I f deposition 6. M A G N E T I C FIELD D E P E N D E N C E OF
HALL COEFFICIENT
techniques are used to produce such thin films,
Factors which can influence how the Hall
the mobility may be further degraded. For example,
coefficient varies with magnetic field include the
the I n S b in Hall elements of respective resistances
nature of the charge-carrier scattering, the degree
3 and 15 f~ produced by MARAIS etal.(ls) possessed
of degeneracy (i.e. the location of the Fermi level
electron mobilities of 34,000 and 18,000cm2/
with respect to the band edge), the band structure,
V-see, respectively. Film thicknesses were between
and the degree of inhomogeneity of the specimen.
10 and 16/~. T h e deposits were polycrystalline,
and although it appears that recrystallization ~"Further improvements are being reported (see the
conference papers cited in Ref. 19), and it appears that
techniques(19) might produce films as thin as 1-2/~ the mobility gap between vapour deposited films and bulk
with mobilities of 30,000, the result still represents crystals of InSb is being substantially reduced.
THE HALL EFFECT AND RELATED PHENOMENA 345

(a) Charge-carrier scattering and degeneracy a set of ellipsoids of revolution arranged so as to


For scattering by acoustic mode lattice vibra- possess cubie symmetry. Figure 3 shows a cal-
tions, the weak magnetic field Hall coefficient culated magnetic-field dependence of the Hall
factor is 3rr/8 or 1-18 for the case where the carrier
concentration is low enough that classical statistics I'O
apply; under similar conditions, that for scattering J
by ionized impurities only is 315rr/512 or 1.94. If, 0-9 ~ J
however, just a small amount of impurity scattering
is mixed with acoustic mode scattering, the Hall o.s
coefficient factor is actually less than 37r/8.(21) As
one approaches the strong magnetic field limit, the
Hall coefficient factor approaches unity. A similar H: d: III
0-6
behaviour occurs in the case of large degeneracy. I I I I I I ! I
It is to be noted that the scattering mechanisms 0 I0 20
cited above--and also other commonly encountered CUlT
isotropic scattering processes--lead to a monotonic Fxo. 3. Theoretical Hall coefficient factor resulting from
decrease in R with H. voltages across Hall probes as a function of
oJ±'r( ~ eHrlm±*c ~ t~±'r)
(b) Band structure for J along a [111] direction (after BULLIS and KRaG(S3)).
The quantity m~* is the longitudinal mass (that associated
Anisotropy. Anisotropy in the surfaces of with the axis of revolution of the energy spheroid) and
constant energy can lead to complex variations of m±* is the transverse mass.
R with field. One may encounter increases, de-
creases, or combinations of both--leading to coefficient factor, as determined from voltages
maxima or minima. Any general theoretical across the Hall probes. Besides depicting the effect
development would be complicated even for the of the anisotropy, this figure illustrates another
case of cubic symmetry, and for details the reader important consideration, namely that for field and
is referred to the literature. Discussion here will current directions of relatively lower symmetry--
be confined to several illustrative examples. Figure here [~10] and [111], respectively--one can have an
2 shows the anisotropy factor in r0 resulting from intermixing of Hall and magnetoresistance volt-
ages, even in crystals of cubic symmetry. This
1.2 point was discussed in Section 4, where we noted
that under such conditions a Hall element would
no longer exhibit V(H) = - V ( - H ) .
1.o , f ~
Although the existence of anisotropy renders
more difficult any theoretical calculations, it does
o.8! / --~
provide substantial flexibility for the design
engineer to exploit in order to realize a specified
..y 0"6 dependence of Hall voltage on H. The changes
~s in this behaviour which result from different direc-
tions of magnetic field with respect to the crystal-
, lographic axes are illustrated by data taken on
J Oblate Prolate 3 ~2-cm n-type germanium presented in Fig. 4 for
0.2 / energy energy
surfaces surfaces current in the [100] direction and in Fig. 5 for
I I I I current in the [110] direction.
0'01 002 00501 0"2 0"5 I'0 2 5 I0 20 50 Multiband effects. The participation of two or
K more non-equivalent energy bands in conduction
can have a very conspicuous effect on the magnetic
FIG. 2. Dependence of anisotropy factor in/xrr//~ on the
anisotropy of the effective mass for a family of spheroidal
field dependence of the Hall coefficient and on the
energy surfaces arranged so as to possess cubic symmetry other galvanomagnetie effects. T o generalize equa-
(After HrnmNG(22)). The quantity K i s the ratio mF*[m±*. tions (8) or (16) for the multiband case, we need
346 A. C. B E E R

to sum the contributions from each band in obtain- materials depends only on the carrier densities,
ing the elements of the conductivity tensor shown 1
in equation (10), noting that the antisymmetric Rtt-,oo ]elc(p_n), n ~ p, (34)
terms bear opposite signs for conduction by holes
and does not involve the mobilities. Thus in the
example just cited, we may expect a significant
37 >
~. , , ~ o l o o g [oo0 ox,~T ~ change in Hall coefficient as the magnetic field is
-~ 3"6 varied. This is precisely what one finds, for
3-5 example, in p-type germanium, where the valence
o
3-4 I] I r band is degenerate and bands of light- and heavy-
"3 53
mass holes exist. Experimental data for such a
specimen are shown in Fig. 6. T h e decrease in r0
3'2
at lower temperatures is attributed to the influence
2 3'1 of increased ionized impurity scattering. Warping
o 3-0 (non-sphericity) of the heavy-mass band exists,
2 4 6 8 I0 12 14 [6 18 20 22
and in silicon the competition of effects from the
Mognefic F}eld, H, ira kHogouss
two bands, from warping, and from mixed scatter-
Fro. 4. Ratio of Rn]po for a [100] cylinder for two direc- ing can result in substantially different magnetic
tions of the magnetic field (after MAsoN et a/.(7)). field dependences at low temperatures in specimens
of different purity. This is illustrated in Fig. 7.

-- 11 _ x_ (c) Inhomogeneities in specimens


.~ 3.6 ~ ,,--x---'~o,o~g [oo01 | We shall note later that any mechanism that
3-5 I I~'-~-~. Hz2"~°eromloo*] disturbs the Hall field--be it contact shorting,
surface conditions, multiband effects, or inhomo-
geneifies--can strongly affect the magnetoresis-
tance. In the case of inhomogeneities--if the
o
a 5,2 d ialong, ~lO]' H a'long ~ ' T O
J ' ~ / ~ ~'<L,~_
[ gradients are in the right direction--an admixture
LII ' II of magnetoresistance and Hall effect can occur.
-£ 3"0 This can substantially affect the field dependence
~. 0 2 4 6 S I0 12 14 16 18 20 2~'
Megnefic Field, H, in kilogouss
of the measured Hall coefficient--as can be seen
from the data in Fig. 8.
FIO. 5. Ratio of Rn/po for a [110] cylinder for five direc- T h e effect of inclusions in thin films can be
tions of the magnetic field (after MASONet al.(7)). especially striking. For example WIENER and DAVIS
found after etching out free indium from InSb
or electrons. In the case of an electron and a hole films (of thicknesses between 1 and 1"5/z) that the
band, the expression for the weak-field Hall magnetoresistance at 5 kG fell to one-fifth its
coefficient becomes former value and the Hall voltage increased almost
four times.(19) T h e case of specimen inhomogeneity
Ro = -- (rel*e2n-- rnffn~P)/[] e IC(l~en+/~hp)2], (33) as well as the related situation of non-uniform
magnetic field, is also discussed in several con-
where n, p are the densities of the electrons and ference papers.(27)
holes, respectively;/~e, fin the respective mobilities;
and re, rn, the respective Hall coefficient factors. 7. T H E R M A L EFFECTS
We note from the numerator in equation (33) There are a number of thermal effects in a mag-
that the carrier concentrations are weighted by the netic field, which can produce transverse voltages
squares of the mobilities. I f therefore the mobilities or temperature gradients.~" These result from the
are substantially different in the two bands, one
1"A more thorough excursion into these phenomena
can expect large effects. In contrast, in the strong- can be*found in Refs. 1-6, and, in yet further detail, in
field limit, the Hall coefficient for non-intrinsic Refs. 28 and 29.
T H E H A L L E F F E C T AND R E L A T E D PHENOMENA 347

velocity separation of charge carriers by the effects are important---or if the so-called "thermo-
Lorentz force--the energetic ones going to one electric materials"* are involved~then considera-
side, the slower ones to the other. Temperature tion of thermomagnetic effects may be advisable.
gradients are produced, and also electric fields. In Thermomagnetic phenomena are responsible in
the Righi-Leduc effect, a longitudinal temperature first approximation, for differences between the
gradient produces a transverse temperature so-called isothermal and adiabatic Hall coeffi-
gradient (thermal analog of the Hall effect); in cients.Ca0-32) T o reduce any such differences, in
the Nernst effect, it produces a transverse electric addition to choice of materials having relatively
field. In the Ettingshausen effect, a longitudinal large thermal conductivities, one may also employ
electric current produces a transverse temperature the expedient of making a good thermal attachment

~z. _ -- - - E95
Ge Specimen 2-C
NA- No~lO'3 cm -3
Experimental data at temperature
7, I
indicated : v•z~oD
• • • •, - -o I
p

Values for theoretical curves :


. . . . . + - - . Z ~ IIO X \,~ F~ = 6 5 0 0 0 cma/volt-sec
A A "~ xtK
8 I. L
\\ ,u-3//~2 = 8
\ \~ P3/Pz = 0 0 4
\ \ Ordinates normolizedat H =20 kilo-
~\gauss
:,< i --
~ \ r
cu

(D

-c

. I Eo,o]
i
0"1 I'0 IO

Magnetic Field H, kilogauss

FIG. 6. Hall coefficient in p-type germanium. Theoretical curves (solid


lines) are based on /~2L = 65,000 cm2]V-sec, representative of the lattice
mobility at 77°K, and on normalization of ordinates at 20 kG. Subscripts
3 and 2 refer to light- and heavy-mass bands, respectively (after BEER(24)).

gradient. The latter effect, if large, can modify the of the Hall element to a block of material having
Hall field, since the potential probes and leads are high thermal conduction.
seldom made of the same material as the specimen. Independent of thermomagnetic effects are, of
Therefore the Ettingshausen temperature gradient course, any temperature gradients and resuking
can produce a thermoelectric voltage which adds thermoelectric voltages produced by uneven heat-
to the Hall voltage. ing in the specimen. These factors must always
T h e thermomagnetic effects depend inversely be considered in design of a Hall element.
on the thermal conductivity of the material and
8. RELATED P H E N O M E N A
therefore can usually be negleeted where there is
an appreciable lattice component of thermal A widely studied galvanomagnetic effect is
conductivity. If, however, the application of the transverse magnetoresistance. This phenomenon can
Hall effect is sufficiently critical that higher order • See, for example, p. 112 of of Ref. 29.
348 A. C. B E E R

be illustrated by the pictorial scheme in Fig. 1. of a certain "average" velocity will be undeflected.
I f a charge carrier is deflected by the Lorentz T h e remaining carriers, having velocities either
force so as to traverse a longer path, it will contri- larger or smaller than the "average" will traverse
bute less to the conductivity, and there will be a larger paths. T h u s the resistance is increased. A
positive magnetoresistance. It was noted, however, similar situation obtains if more than one type of
that if all charge carriers have the same velocity carrier is present. It is also apparent that any
none will be deflected at steady state, since then mechanism which shorts out the Hall voltage--

1.24

g
i.z0

N ['16
Symbol
z~
v
o
o
T= 7 7 oK

Sample
162
2A
IB
8A
/
o H2

..a

u
~ ,.08
c

o
I

1.00

0.96
IO00 IO,O00 IO0,OOO
Magnetic Field , gauss

FIG. 7. Magnetic-field dependence of relative Hall coefficient in several p-type silicon samples of different purity
(after BRmtt25)).The ultra-pure specimens 8A and 1B have room-temperature resistivities of 2700 and 5400 ~-cm,
respectively. Sample 1B appears to have more compensation than does 8A, a postulate borne out by the Rcr data
which are 5 to 10 per cent lower.

the Hall field will exactly cancel the v x H force. for example, special geometry, shorting contacts,
This is the case of the isotropic metal, where the inhomogeneities in the specimen--wiU increase
velocity of all electrons is essentially the Fermi the magnetoresistance.
velocity. It is also true for electron scattering T h e technique of shorting the Hall field is used
mechanisms described by a relaxation time which to advantage in the Corbino disk,(33) shown in
does not vary with energy. If, however, there is a Fig. 9. As a result, the measured magnetoresistance
distribution of electron velocities--as in a semi- continues as/z2H 2 and does not saturate at strong
conductor--then it is clear that only those electrons magnetic fields. Another procedure for fabricating
THE HALL EFFECT AND RELATED PHENOMENA 349

2"0

n-type InSb
77 ° K

Sample B
Ro=4'O x I03 crn3/coulomb

~6

09
1.5

1'4

J.2

H ~"
SQmple A
e =3'4xlO2ohm-cm
rL~L 0 1,0 rr
o
R,,,~= 1"4~7x104 cm3/coulomb
I0-e F H =I

"90
/ 90
I0 102 103 104
H in Gouss
FIG. 8. Magnetic-field d e p e n d e n c e of t h e n o r m a l i z e d H a U c o e ~ c i e n t of two h i g h - p u r i t y n - t y p e I n S b
s a m p l e s . Carrier c o n c e n t r a t i o n s are: s a m p l e zq: 4-2 × 1014 c m - 8 ; s a m p l e B ( n o n - l m i f o r m ) : 5 x 1014 to
5 x 10 is c m -18 (after B A ~ et al.{Zo)).
350 A. C. B E E R

magnetoresistance elements, involves the use of 2. W. C. DUNLAP,JR., An Introduction to Semiconduc-


shorting bars.(34) I n these types o f elements, the tors, Wiley, New York (1957).
3, C. HILSUM and A. C. ROSE-INNES, Semiconducting
p h e n o m e n o n which is exploited is magnetoconduc- I I I - V Compounds (International Series of Mono-
tivity, i.e. the diagonal d e m e n t s of the conductivity graphs on Semiconductors), Pergamon Press,
tensor.05) Magnetoresistanee elements are dis- Oxford (1961).
cussed in a n u m b e r of conference papers. (36) 4. O. MADELUNO,Physics of I I I - V Compounds, Wiley,
New York (1964).
II q 5. E. H. PUTLEY, The Hall Effect and Related Phenomena
(Semiconductor Monographs), Butterworths,
London (1960).
6, M. RODOT, Les Materiaux Semi-Conductors, Dunod,
Paris (1965).
7. W. P. MASON, W. H. HEWITT and R. F. WICK,
J. Appl. Phys. 24, 166 (1953).
8. A. C. BEER, op. cit., p. 6.
9. R. F. WICK, J. Appl. Phys. 25, 741 (1954).
Corbino Disk Electrode Shorting 10. H. J. LIPPMANN and F. KUHRT, Z. 2Vaturforsch.
(a) (b) 13a, 462, 474 (1958).
11. Additional literature is cited in Ref. I, p. 56-59.
FIR. 9. Mechanisms to produce shorting of the Hall 12. M. G. BUEHLERand G. L. PEARSON,Paper II-4; S.
voltage. GROTZMANN,Paper II-5 ; J. HAEUSLER,Paper II-6.
13. E. W. SAKER,F. A. CUNNELL and J. T. EDMOND,
A variant of the magnetoresistance element in- Brit. J. Appl. Phys. 6, 217 (1955).
volves use of the magnetic field to modulate the 14. I. M. Ross, E. W. SAKERand N. A. C. THOMPSON,
skin depth b y means of the magnetoconductivity J. Sci. Instrum. 34, 479 (1957).
effect. F o r example, the r.f. transmission through a 15. L. L6FCR~N, J. Appl. Phys. 29, 158 (1958).
16. O. J. MENCALI and T. S. SmLLIDA% Solid-State
disk of a h i g h - p u r i t y semimetal such as b i s m u t h Electron. 7, 379 (1964).
at low temperatures is effectively controlled by a 17. R. K. WmLARDSONand A. C. BEER, Elec. Mfg. 57,
magnetic field of a few gauss. T h e choice of a 79 (1956).
semimetal was m a d e because the equality of 18. M. MARAIS,M. MENORETand A. PERIGAUD,L'Onde
J~Iectrique 44, 1005 (1964).
electron and hole concentrations insures pre- 19. H. H. WIEDER and N. M. DAVIS, Solid-State
dominance of the diagonal elements in the con- Electron. 8, 605 (1965). In regard to mobility in
ductivity tensor over the nondiagonal elements, thin films, see also the following conference
even at strong fields. U n d e r such conditions the papers: H. H. WIEDER, Paper II-1; J. CARROLL
expression in cgs units, for the skin d e p t h in a and J. F. SPIVAK,Paper II-2a.
20. C. HILSUM, Electron. Engr. 30, 664 (1958).
m e d i u m of unit permeability is given b y 21. V. A. JOHNSONand K. LARK-HOROVITZ,Phys. Rev.
C 82, 977 (1951). The relationship is also shown in
Ref. 1, p. 133.
= ~/(2~rcoaxz)" (35) 22. C. HERmNC, Bell Syst. Tech. J. 34, 237 (1955).
23. W. M. BULLISand W. E. KRA%Phys. I~ev. 101, 580
T h e above relation is merely the familiar zero- (1956).
magnetic-field expression with the magneto- 24. A. C. BEER,J. Phys. Chem. Solids 8, 507 (1959).
conductivity azx in place of the zero-field con- 25. A. C. BEER, op. cit., p. 209.
ductivity ao. 26. R. T. BATE, J. C. BELL and A. C. BEER, J. Appl.
Phys. 32, 806 (1961).
I f the other condition prevails, namely that 27. H. H. WIEDER, Paper II-2; M. RODOT, C. VERIE
~xv >~ Crxx, then one encounters the so-called and F. BAILLY,Paper II-3 ; H. WEISS,Paper II-7 ;
"helicon" wave transmission, which can occur in I. HLASNIK,Paper IVA-Sb.
media where the Hall angle is large.(a7) 28. I. M. TSlDIL'KOVSKII,Thermomagnetic Phenomena in
Semiconductors (in Russian). Fizmatgiz, Moscow
F u r t h e r details on the modulation of skin depth
(1960); English Translation by A. TYBOLEWlCZ,
are given in a conference paper.(as) edited by H. J. GOLDSMID,Academic Press, New
York (1962).
REFERENCES 29. J. R. DRABBLE and H. J. GOLDSMID, Thermal
1. A. C. BEER, Galvanomagnetic Effects in Semiconduc- Conduction in Semiconductors (International Series
tors (Supplement 4 to Solid State Physics Series), of Monographs on Semiconductors), Pergamon
Academic Press, New York (1963). Press, Oxford (1961).
THE HALL EFFECT AND RELATED PHENOMENA 351

30. A. G. SANIOILOWITSCHand L. L. KORENBLIT, 35. For a discussion of the distinction between magneto-
Fortschr. Physik 1, 486 (1954). conductivity and magnetoresistanee, consult p. 229
31. ¥. A. JOHNSONand F. M. SHIPLEY,Phys. Rev. 90, of Ref. 1.
523 (1953). 36. H. Wvlss, Paper II-7; L. T. Y u ~ , Paper 111-4;
32. L. S. STIL'BANS,Zh. Tekh. Fiz. 22, 77 (1952). R. BONN~ILLE,Paper III-9.
33. H. WEISSand H. W~.L~a, Z. Physik 138, 322 (1954) ; 37. For a discussion of helicon wave phenomena consult
O. MADELU~G,Naturwissenschaften 42, 406 (1955). R. BOWERSand M. C. STEELE,Proc. Inst. Elect.
See also p. 71-76 of Ref. 1. Electron Engrs. 52, 1105 (1964); also R. G.
34. H. WELV~ERand H. WEISS in Solid State Physics CHAMBERSand B. K. JONES,Proc. Roy. Soc. A270,
(F. SEXTZand D. TUR~mULL,eds.), Vol. 3, p. 1. 417 (1962); JACQUESGRI~MILLET,Comptes 2~endus
Academic, New York (1956). 257, 2431 (1963).
38. P. F. HILLE, F. J. REID and A. C. BEER,Paper II-8.

You might also like