430 SESSION P: SEMICONDUCTING COMPOUNDS

Table 1. Data on eflective massesfor bismuth telluride. It has been found that the conductivity mobility
m,, m2 and m3 are the principal efective masses for current flow in the cleavage planes at room
within a valley, azd m* is the density of states temperature is about 1200 cm‘J/V-set for electrons
effective mass and about 500 for holes for specimens studied in
ZP ___-.
I I this work.
n-type
I I p-type
A further point of interest in the work on p-type
ma unknown O.O505m, material was that the ratio of the resistivities for
mllma 1.21 4.13 current flow perpendicular and parallel to the
ma/m* 0.093 764 cleavage planes and the ratio of the two Hall co-
co? 8 0.055 0.305
efficients were both practically independent of
m+ unknown 0.51 m,
temperature over the range liquid nitrogen to room
______~_
temperature, indicating that any changes in the
The relations between the conductivity mobility form of the valence band are negligibly small over
and the Hall mobility are correspondingly affected. this range.

J. Phys. Chem. Solids Pergamon Press 1959. Vol. 8. pp. 430-433. Printed in Great Britain

TRANSPORT PROPERTIES OF THE PSEUDO-BINARY

ALLOY SYSTEM Bi,Te,_,Se,
N. FUSCHILLO, J. N. BLERLY and F. J. DONAHOE
Franklin Institute, Philadelphia, Pa.

1. INTRODUCTION of bismuth is unity, so that Bi,Te,_$ey is truly a

THE alloy system Bi,Te,_$3eU has been the subject
of a number of investigations’l-*) in recent years
because of the possible application of these alloys
in thermoelectric devices. This binary system, how-
ever, presents some interesting intermetallic and
semiconductor situations which are of fundamental
interest in the study of compound semiconductors.
In this paper is presented a systematic study of the
temperature variation of the electrical conductivity,
a, and the Seebeck coefficient, S, and the room
temperature thermal conductivity, K, as selenium
is substituted for tellurium in the molecular formula
Bi,Te,_,Ser, until B&Tea is transformed into
B&Se,. Our X-ray and metallographic examina-
tions have uncovered no evidence of any phase
transformations in this system and all available
evidence points to the existence of a continuous
solid solution range between B&Tea and BiaSe,.
In particular the careful determinations of MCHUGH
and TILLER@) show that the distribution coefficient
pseudo-binary alloy system.
2. EXPERIMENTAL
The test specimens were prepared by encapsu-
lating the semiconductor grade purity materials in
evacuated Vycor tubes which were heated to above
the melting point and shaken vigorously. The melt
was frozen as rapidly as possible by plunging the
Vycor tube containing the liquid alloy into water.
The resulting ingots consisted of columnar crystals
extending to the axis of the alloy rod. Since the
c-axis of the hexagonal structure is perpendicular
to the direction of most rapid growth the c-axis will
be randomly orientated with respect to the longi-
tudinal axis of the specimen. All of the transport
properties were measured along the long axis of the
specimen and therefore represent a value averaged
over all possible orientations of the c-axis. The
Seebeck coefficient and electrical conductivity re-
spective1yweremeasuredbythestandardd.c.anda.c.
 SESSION P: SEMICONDUCTING COMPOUNDS 431

(60 c/s) methods, in a vacuum cryostat operating specific electrical conductivity of ten specimens of
between 77 and approximately 600°K.* The total different compositions. Seven of these specimens of
thermal conductivitieswere measured by the method composition y = 0 (Bi,Te,), y = 1.5, 1.8, 2.1, 2.4,
described by J0~~ti.f’) 2.7, and 3.0 (B&Se,) are clearly highly degenerate
3. RESULTS AND DISCUSSIONS semiconductors over the whole temperature range
Fig. 1 shows the temperature dependence of the studied. The natural temperature coefficient of
resistivity is positive, and with the exception of
2cCQ Bi,Te,, is constant throughout the temperature
range investigated. Moreover, both the temperature
1000 coefficient of resistivity and the magnitude of the
800 conductivity decrease regularly with increasing
600
tellurium starting from B&Se, up to y = 1.5.
Specimens of these compositions are thus behaving
like semimetals rather than semiconductors.
;
2 200 This behavior may be contrasted with the semi-
2 conductor behavior represented by the composi-
b 100 tions y = 0.72 (except below 200”K), y = 0.9,
80

60
y = 1.2 (above 450°K). The energy gap deduced
from the temperature dependence of electrical con-
ductivity is of the order of 0.13 eV. This is smaller
by a factor of the order of two from the optical band
gap.t5) This factor is not difficult to explain if the
10
maximum energy in the valence band and the
6 minimum energy in the conduction band occur in
100 2w 300 400 500 600
TPK 1 different regions of K-space. It should be noted that
the compositions having the largest optical band
FIG. 1. Electrical conductivity of Bi,Tes_$ey VS.
temperature. gap in this series of alloys are those which exhibit
near intrinsic behavior around 300°K.
* Seebeck coefficients are measured relative to copper. In Fig. 2 is shown the temperature variation of
Of course this constitutes a negligible departure from
the measurement of an absolute value except near the Seebeck coefficient for the same series of alloys.
regions where the Seebeck coefficient of the alloy is in The general features of these curves are in agree-
the process of reversing sign and has a low value. ment with the conclusions drawn from the

180

100

- I ‘j

-160
100 200 300 400 500 600

FIG. 2.ISeebeck coefficient of Bi,Te,_,Se, vs. temperature.

432 SESSION P: SEMICONDUCTING COMPOUNDS

conductivity data of Fig. 1. For example, the See-

beck coefficient of B&Se, (y = 3) increases in abso-
Conduct;vity
lute magnitude with temperature in an almost linear tsotherms of th

fashion throughout the whole temperature range.

The Bi,Te, specimen (y = 0) has a Seebeck co-
efficient which, at low temperature, indicates strong
degeneracy, but which in the neighborhood of
400°K is starting to bend back towards the S-equal-
to-zero axis at about the same temperature at which
the electrical conductivity appears to be approaching
a minimum. On the other hand the variation of S
with temperature is greatest in the specimens
(y = 0.72, O-9) which were judged to be near
intrinsic at 300°K.
It is instructive to consider various isotherms of 0 0.75 I.5 2.25 3
these results. Thus in Figs. 3 and 4 are shown the
FIG. 4. Electrical conductivity isotherms of the
Bi,Te,_$ey system.

dependence found for B&Se, (n-type).* The latter

variation is characteristic of ionic lattice type
scatteringt8) above the Debye temperature. The
T-3 variation for BisTe, has not been adequately
explained.
An isothermal (300°K) representation of s vs.
log,, o is shown in Fig. 5. The form of this curve
300
-50 I I 1
200
-100

-150

I
0 0.75 1.5 225 3
Y

FIG. 3. Seebeck coefficient isotherms of the Bi,Te,_$e,

system.

isothermal variations of S and CTrespectively at 100,

200,300 and 400°K. These results indicate a rather 1°F 1.5 2.0 2.5 30
unusual behavior at values of y in the neighborhood
of 0.9. There is a pronounced minimum in the FIG. 5. Dependence of the Seebeck coefficient on
electrical conductivity at T = 300°K.
electrical conductivity and the Seebeck coefficient
reverses sign on passing through zero near this
has the shape predicted by PRICE.@) The specimen
composition, although the former effect occurs at a
points in the intrinsic region where the graph crosses
slightly smaller value of y than does the reversal in
the axis (S = 0) were from a series of alloys of y
the Seebeck coefficient. In addition the room tem-
very close to 0.9. While this figure may be con-
perature Hall mobility is minimal at compositions
sidered interesting it is not worthwhile to apply the
in the neighborhood of Bi,Te,Se (y = 1) and varies
as T-1, which is intermediate between the Te3 * The Hall mobility of the other compositions is still
dependence found for BisTe, (p-type) and the T- * under investigation.
 SESSION P:
analysis suggested by PRICE to the present series of
alloys. To do so would involve the assumption that
the effective masses and mobilities are unchanged
in the Bi,Te,_,Sey system as y ranges from 0 to 3.
A final isotherm is shown in Fig. 6, where the
0 0.5 I I.5 2 2.5
FIG. 6. The total thermal conductivity
Bi,Te,$ey system at T = 300°K.
total thermal conductivity is plotted against y for
values at 300°K. The expected pronounced mini-
mum is found in both the total and lattice (not
plotted) thermal conductivity. This, however,
occurs in the neighborhood of y = 0.9. The
secondary minimum behavior at y = 2 to 2.5 is
as yet unexplained but is in agreement with the
results of ROSI, ABELES and JENSEN.(~)
The anomalous behavior of the transport pro-
perties in the neighborhood y = 0.9 is also accom-
panied by an optical band gap anomaly observed by
AUSTIN and SHEARIX(~)The graph of band gap
versus percentage of B&Se, added to B&Tea
exhibits a linear variation from y = 0 to y = 0.9.
At y = 0.9 a cusp-like maximum appears and the
band gap decreases sharply at first and then more
gradually to the value for B&Se,.
The full explanation of these phenomena has not
yet been determined and awaits the detailed con-
sideration of recent X-ray and Hall mobility data.
However, some preliminary observations indicate
that their explanation may lie largely in the structure
and bonding scheme for alloys of this series.
Structurally this series of alloys belongs to the
crystal class 9m and is isomorphous with the natur-
ally occurring mineral tetradymite (Bi,Te,S).“O
The atomic arrangement is a layered structure in
which the layers repeat the unit
B,-A-B,-A-B,.B,-A-B,-A-B,,
where B represents a Group VI element and A a
Group V element. DRABBLEand GOODMANhave
proposed a bonding scheme which explains the
band gap anomaly in the Bi,Te,,Se, system.
FF
SEMICONDUCTING COMPOUNDS 433
This bonding scheme indicates that B, sites are
energetically more favorable than are B, sites. Thus
a Group VI element which is more electronegative
than tellurium should prefereutially occupy B, sites.
At the nominal composition Bi,Te,Se all the B,
sites should be occupied by selenium and the
39Y
remaining B, sites by tellurium. (Entropy con-
siderations may well shift the best ordered composi-
tion to y = 0.9.) The naturally occurring mineral
tetradymite (Bi,Te,S) has this structure (lo) with
sulfur occupying B, sites. Although these structures
are ordered, no superstructure lines should appear
since the B, sites are crystallographically non-
equivalent to the B, sites.
of the This model accounts for the variation in band
gap but gives no clue to the reason why the alloy
should be intrinsic at this composition when pre-
pared from stoichiometric materials. It would be
convenient if one could point to some feature in
the liquidus-solidus curve of the pseudo-binary
system which could be used to explain the reason
for this behavior. Unfortunately the experimentally
determined liquidus-solidus curve shows no ano-
maly at this composition.(3) For the present, it
seems rather fortuitous that the critical point in the
Seebeck coefficient of this pseudo-binary system
should coincide with the energy gap maximum.
Acknowledgment~The authors are indebted to Mr.
L. C. BENNETT for his assistance, to Messrs. J. R. WIE~E
and R. PYLE for alloy preparation and to the sponsors of
the thermoelectric effects research program for their
financial support.
REFERENCES
1. SIANI S. S. and GORDYAKOVA G. N. 2%. tekh. fiz.
26, 2398 (1956).
2. AIRAPETYANTS S. V., EFIMOVA B. A., STAVITSKAYA
T. S., STILBANS L. S. and SYSOEVA L. M.
Z-h. t&h. jk 27, 2167 (1957).
3. MCHUGH J. P. and TILLER W. A. Westinghouse
Scientific Paper No. 431FD271-Pl (1958).
4. DRABBLEJ. R. and GOODMAN C. H. L, r. Phys.
Chem. Solids. 5, 142 (1958).
5. AUSTINI. G. and SHEARD,A. r. Electronics 3, 236
(1957).
6. ROSI F. D., ABELESB. and JENSENR. V. Private
communication.
7. JOFF~,A. F. Canad. J. Phys. 34, 1342 (1956).
8. KOLOMOETS N. V., STAVITSKAYA T. S. and STILBANS
L. S. Zh. tekh.jk. 27, 73 (1957).
9. PRICE P. J. Phys. Rev. 104, 1223 (1956).
10. HARKERD. Z. KristaZZogr. 89, 175 (1934).