ArticlePDF Available

Simulation of polycrystalline bismuth films Seebeck coefficient based on experimental texture identification


Abstract and Figures

Seebeck coefficient values measured for textured polycrystalline Bi films have been compared with numerically computed ones. The proposed computation procedure requires crystallographic texture mapping and values of single-crystalline transport coefficients as input data. The comparison of computed and measured Seebeck coefficients allowed determining the factors affecting films' transport properties except of crystalline texture. Two particular cases are considered: when the film properties are affected only by texture and when they depend on texture and grain size simultaneously.
Content may be subject to copyright.
Alexander S. Fedotov1, Vasiliy Shepelevich1, Sergey Poznyak2, Lyudmila Tsybulskaya2,
Alexander Mazanik1, Ivan Svito1, Sofia Gusakova1, Pawel Zukowski3,
Tomasz N. Koltunowicz3,
1 Belarusian State University, Minsk, Belarus
2 Research Institute for Physical Chemical Problems, Belarusian State University, Minsk,
3 Lublin University of Technology, Lublin, Poland
Abstract. Seebeck coefficient values measured for textured polycrystalline Bi films have
been compared with numerically computed ones. The proposed computation procedure
requires crystallographic texture mapping and values of single-crystalline transport
coefficients as input data. The comparison of computed and measured Seebeck coefficients
allowed determining the factors affecting films’ transport properties except of crystalline
texture. Two particular cases are considered: when the film properties are affected only by
texture and when they depend on texture and grain size simultaneously.
Keywords: Bismuth; Polycrystalline films; Crystal structure; Transport properties.
It is common knowledge that properties of polycrystalline materials are determined by
such factors as interface electronic states, mean free path limitations, and, in the case of
materials with a pronounced anisotropy of single-crystalline properties, by texture type [1-3].
It is a challenge usually to distinguish these physical mechanisms, especially when crystallites
are highly anisotropic and film shows a clearly defined texture.
Bismuth is known as semimetal possessing a significant anisotropy of transport
properties in the single-crystalline form. It is a basic material for many of thermoelectrical
applications [4, 5] that justifies a search for the ways of its Seebeck coefficient improvement
In the present report Seebeck coefficient of polycrystalline bismuth films with different
grain sizes and crystallographic texture has been studied. We propose a technique for
computation of polycrystalline film’s Seebeck coefficient on the basis of solution of a partial
differential equation system, considering each grain as a continuous medium with transport
coefficients corresponding to the texture. Additional factors, except of crystalline texture,
affecting the transport properties of polycrystalline Bi films are also considered.
Preparation of samples and texture identification
Two types of polycrystalline bismuth films were studied: (a) films prepared by the melt
spinning (MS) and (b) ones obtained by the electrochemical deposition (ECD). Both synthesis
techniques were well-developed and tested in our previous works [7,8].
In the MS technique, melted Bi was spilled on the room-temperature surface of a
rotating copper cylinder. Cooling rate of Bi reached appr. 106 K/s [9]. Thickness of the
fabricated MS films was about 40 µm.
ECD synthesis was performed in the galvanostatic mode from an aqueous electrolyte,
containing 0.174 M bismuth perchloride dissolved in 3 M perchloric acid. The
Corresponding author, fax: +48-81-5384575, tel: +48-81-5384713
electrodeposited Bi films were rinsed with distilled water and then picked off from the
substrate. Thickness of the electrodeposited films was about 70 µm.
Several series of samples were fabricated by both methods under the same conditions.
Five samples obtained by the MS and more than ten ECD films were investigated by scanning
electron microscopy and demonstrated good reproducibility in their structure.
For experimental determination of Seebeck coefficient, a heater element was powered
by an Agilent U3606B DC power source creating a temperature gradient about 1.5 K on the
sample investigated. Temperatures on cold and hot ends of the sample were determined using
platinum resistance thermometers connected to Agilent 34410A digital multimeters and
providing an accuracy of temperature determination about 0.05 K. Thermal EMF was
measured by an Agilent 34411A multimeter with an accuracy of about 3.5 μV.
Samples for electrophysical measurements were cut from the center of each as-
synthesized film to avoid an influence of boundaries on the structure. Seebeck coefficients
determined for different samples of each type fluctuate around the corresponding average
values within 1%.
A HZG-4M X-ray diffractometer (Cu radiation) and a LEO 1455VP scanning
electron microscope with an add-on for electron backscatter diffraction (EBSD) analysis were
used for identification of the texture.
According to the SEM data, Bi grain sizes range from 5 to 15 µm for the films
fabricated by the MS and from 0.5 to 1.5 µm for the ECD ones.
(a) (b)
Fig. 1. EBSD texture mapping for particular grains (in-film plane) of the MS (a) and ECD
films (b) and the color legend (c).
As follows from Fig. 1, two types of in-plane growth texture are observed: (001) and
(012). XRD analysis (Fig. 2a) demonstrated only the presence of reflex corresponding to
(012), but this divergence is well-known and related to the relaxation of the (001) line in
bismuth [10]. Quantitative EBSD analysis showed that the films fabricated by the MS possess
(001) and (012) textures for 39% and 54% of grains, respectively. The ECD films have (001)
texture for 36% and (012) texture for 57% of the grains. Relevant crystallographic planes are
shown in Fig. 2b.
20 40 60 80 100
Intensity, a.u.
Electrochemical deposition
Melt spinning
(a) (b)
Fig. 2. XRD patterns (a) and the corresponding crystallographic planes (b) in Bi lattice.
Mathematical model and Seebeck coefficient computation
We can use the following equation system [11] to describe the coupled processes of
heat and charge transfer:
, (1)
where J
is the current density, Tthe temperature, Vthe electric potential, S – the Seebeck
coefficient, σ the electrical conductivity, k the thermal conductivity, Cp the heat
capacity, ρ – the density, and t – the time.
We have set the computational domain consisted of 48 subdomains (according to the
SEM data, the MS films possess a column-like structure). Each subdomain is 15×15×40 µm3
parallelepiped (Fig. 3), corresponding to the measured foil thickness (40 µm) and average
lateral grain size (15 µm). The whole domain size is 120×90×40 µm3.
For computation of Seebeck coefficient, boundary conditions were set to correspond to
the open circuit regime at a small temperature gradient between two opposite faces, while the
rest boundary faces were considered as thermally insulated:
294 0
293 120
( ) 0, 0 120, ( , )
T , x , (y, z)
T , x , (y, z)
n J x
n k T x y z
, (2)
where Г is the boundary of the domain and n
is the unit vector normal to it.
For one of the faces, between which the temperature gradient was set, the potential was
considered as equal to zero. The value of thermo-EMF was obtained by averaging the voltage
through entire face with non-zero-voltage condition.
Fig. 3. Boundary conditions over the computational domain.
For quantitative predictions of the electrical conductivity, additional initial-boundary
value problems were solved on the same computational domain:
),( ,1200 ,0
),( ,120 ,101
),( ,0 ,101
),,( ,293
. (3)
The voltage drop across the rectangular sample was obtained by averaging potential
distributions on the both ends of the sample (for x=0 and x=120 boundaries) and subtracting
one from another. The conductivity of the sample was determined as a ratio of the current
density to the voltage drop, multiplied by the sample length.
All the simulations were performed for stationary cases using the finite element method.
Mesh consisted of 5∙105 tetrahedral elements. Conventional PARDISO 5.0 solver was used
with pivoting perturbation order of 10-8 [12, 13].
Transport coefficients of single crystal, such as conductivity, Seebeck coefficient and
others, are tensors [14]. To take into account the crystalline texture, it is necessary to
transform the tensors using the rotation matrix [15]. However, some geometrical relations
could simplify the procedure of transport properties assignment to the simulation domain.
Bismuth lattice cell is suitable to be described in the hexagonal coordinate system [10, 14].
Therefore it is common to distinguish two types of crystallographic directions with their own
transport coefficients. The first one coincides with C3 symmetry axis of the crystal cell and is
called c direction, whereas the directions lying in the plane normal to the C3 axis are called a
directions (Fig. 2b). If the (001) plane is parallel to the film surface, it leads to a direction
coefficients for film in-plane transport (Fig. 1b). In the case of the (012) plane parallel to the
surface, transport processes occur with the values of transport coefficients ranging between a
lower limit (values for a direction ) and an upper limit Cup, which can be approximated as
( )sin
up a c a
, (4)
where θ is the angle between the (001) and (012) planes, Ca and Cc – the transport coefficients
for corresponding directions taken from [14].
For the MS films, we excluded from the consideration 7% of the grains, which do not
show (001) or (012) dominating textures. Then, we assumed that every subdomain has the
probability to possess transport coefficients of a direction equal to 71%. The rest 29% of
grains oriented with the (012) plane parallel to the sample surface have equal probabilities to
possess transport coefficients corresponding to a direction or coefficients as given in (4)
(bimodal distribution with equal probabilities). Similar analysis was also performed for the
ECD films taking into account peculiarities of their texture.
Fig. 4 shows the dependence of Seebeck coefficient and electrical conductivity on the
volume percentage, x, of the grains oriented with c direction parallel to the temperature
gradient, whereas the rest of the grains are oriented with a direction parallel to the
temperature gradient. A linear dependence of Seebeck coefficient on the fraction of grains
oriented with c direction is obtained as a result of the calculations.
0 25 50 75 100
, Sm/m
x, %
Seebeck coefficient
|S|, V/K
Fig. 4. Calculated dependences of Seebeck coefficient and electrical conductivity for ideal bi-
textured film on the volume percentage of the grains oriented with c direction parallel to the
temperature gradient.
The performed calculation for the MS films returned the value of Seebeck coefficient
equal to 62 µV/K, whereas the experimentally measured value is 63 µV/K. For the ECD films
the divergence is rather large: 66 µV/K according to the calculations and 75 µV/K in the
As mentioned above, the grain size for the ECD films ranges from 0.5 to 1.5 µm. The
mean free path of charge carriers in bismuth at 300 K is about 250 nm [16]. It is reasonable to
suggest that the limitation of charge carrier mean free path by the scattering on grain
boundaries results in modification of the transport properties. Therefore, it is improperly to
use experimental data related to large-sized single-crystals for defining the transport
coefficients of the fine-grained ECD films.
Mobilities of electrons μn and holes μp can be calculated using Seebeck coefficient and
electrical conductivity:
)μμ(σ pn pne , (5)
. (6)
Here the concentrations of electrons, n, and holes, p, are independent on texture and can be
taken from [8]. Partial Seebeck coefficients Sn and Sp can be calculated using the Lax model
Results of the computation demonstrate (Fig. 5) that the mobility of electrons is almost
independent on the texture, while the mobility of holes decreases appreciably with x. This fact
can be explained taking into account that T-holes pocket is prolate in k-space along the c
direction, while three pockets of L-electrons are distributed symmetrically [17]. Such form of
the T-holes pocket leads to a higher effective mass of holes in the direction of prolongation as
compared with that in the other two directions and, consequently, to the reduction of holes
mobility in c direction.
0 20 40 60 80 100
, m
x, %
Fig. 5. Calculated mobilities of electrons (squares) and holes (circles) for bi-textured film
depending on the volume percentage of grains oriented with c crystallographic direction
parallel to the temperature gradient.
Two types of polycrystalline bismuth films were examined to highlight factors affecting
their transport properties depending on crystalline texture.
The proposed numerical technique predicts successfully Seebeck coefficient for the
samples synthesized by the melt spinning and possessing a clearly defined texture. The results
of calculation coincide with the experimental data with a high accuracy (about 1 %). For the
samples synthesized by the electrochemical deposition, discrepancy between the calculated
and experimental values of Seebeck coefficient reaches 15 %, indicating the presence of
additional factors affecting Seebeck coefficient in such films. The last is in agreement with
SEM analysis data showing that grain size for the ECD samples is of the order of charge
carriers’ mean free path that produces an additional impact on the transport properties.
The computation technique described can be used for determining Seebeck coefficient of
anisotropic polycrystalline (or just composite-like) materials and has to be applicable to the
films with grain sizes significantly larger than mean free path of charge carriers.
[1] U.F. Kocks, C.N. Tomé, R. Wenk, Preferred Orientations in Polycrystals and their
Effect on Materials Properties, Cambridge University Press, (2000).
[2] S. Datta, Electronic Transport in Mesoscopic Systems, Cambridge University Press,
(1995). doi:
[3] M. Lundstrom, C. Jeong, Near-Equilibrium Transport. Fundamentals and Applications,
World Scientific Publishing, (2013). doi:
[4] D.M. Rowe, Thermoelectrics Handbook: Macro to Nano, CRC Press, (2005).
[5] H.J. Goldsmid, Bismuth Telluride and Its Alloys as Materials for Thermoelectric
Generation, Materials 7 (2014) 2577–2592. doi:
[6] H.J. Goldsmid, Introduction to Thermoelectricity, Springer, Berlin, (2010).
[7] A.S. Fedotov, S.K. Poznyak, L.S. Tsybulskaya, V.G. Shepelevich, I.A. Svito, A. Saad,
A. Mazanik, A.K. Fedotov, Polycrystalline bismuth films: correlation between grain
structure and electron transport, Phys. Status Solidi B, 252 (9) (2015) 2000-2005.
[8] A.S. Fedotov, S.K. Poznyak, L.S. Tsybulskaya, I.A. Svito, V.G. Shepelevich, A. V.
Mazanik, A.K. Fedotov, T. V. Gaevskaya, Synthesis, structure and electrophysical
properties of polycrystalline bismuth films, Proceedings of International Conference
Sviridov Readings 2015, Minsk.
[9] W.J. Boettinger, J H. Perepezko, J.H. Liebermann, Rapidly Solidified Alloys:
Processes, Structures, Properties,Applications, Marcel Dekker, Inc, New York, (1993).
[10] P. Cucka, C.S. Barrett, The crystal structure of Bi and of solid solutions of Pb, Sn, Sb
and Te in Bi, Acta Crystallogr, 15 (9) (1962) 865–872.
[11] L.D. Landau, E.M. Lifshitz, Electrodynamics of Continuous Media Edition,
Butterworth Heinemann, Oxford, (1984). doi:
[12] O. Schenk, M. Bollhöfer, and R. Römer, On large-scale diagonalization techniques for
the Anderson model of localization, SIAM Review 50 (2008) 91-112.
[13] O. Schenk, A. Wächter, M. Hagemann, Matching-based Preprocessing Algorithms to
the Solution of Saddle-Point Problems in Large-Scale Nonconvex Interior-Point
Optimization, Comput Optim Appl 36 (2-3) (2007) 321-341.
[14] B.S. Chandrasekhar, The seebeck coefficient of bismuth single crystals, J Phys Chem
Solids 11 (1959) 268–273. doi:
[15] O. Engler, V. Randle, Introduction to Texture Analysis: Macrotexture, Microtexture,
and Orientation Mapping, CRC Press, (2009).
[16] D. Nakamura, M. Murata, H. Yamamoto, Y. Hasegawa, T. Komine, Thermoelectric
properties for single crystal bismuth nanowires using a mean free path limitation model,
J Appl Phys 110 (2011) 053702. doi:
[17] Z. Zhu, B. Fauque, Y. Fuseya, K. Behnia, Angle-resolved Landau spectrum of electrons
and holes in bismuth, Phys Rev B 84 (2011) 115137.
... Bismuth is a semimetal with a high mobility of charge carriers (10 2 m 2 /(V·s) or higher at 4 K), which results in high electrical conductivity, Seebeck coefficient, and relative magnetoresistance. These properties make Bi-based materials attractive for the development of magnetic field sensors [2,5,6], electrochromic devices [14], thermoelectric coolers [15,16], etc. Numerous researchers report that Bi electrodes can replace both toxic Hg electrodes in the potentiometric analysis [10] and Pb shields for the protection of integrated microcircuits from ionizing radiation [17][18][19]. ...
... Bismuth films can be synthesized by various methods, the most used of which are electrochemical deposition [2,13,16,20], pulsed laser deposition [12,21], vacuum thermal evaporation [22,23], melt spinning [13,24], etc. Pulsed laser deposition allows obtaining highquality thin films due to a high density of grains nucleation. However, due to inhomogeneous energy distribution in the laser beam profile, the inhomogeneity of the film in the center of the beam spot is usually high [25]. ...
... where σ is conductivity. According to Eq. (16), the temperature dependence of resistivity is defined by the temperature dependence of concentrations n, p and mobilities μ n , μ p . Concentration temperature dependence is similar for all the samples (Fig. 9a). ...
The study is devoted to the electronic properties and structure of bismuth films obtained by electrodeposition from high-speed perchlorate electrolyte. Polycrystalline samples were synthesized in acidic perchlorate electrolyte under the (0.18 – 70.0) mA/cm² cathode current density and annealed at 265 oC in the He gas atmosphere. The structure, microstructure and electron properties of Bi films were characterized using X-Ray diffraction analysis, scanning electron microscopy, and electron backscattered diffraction. Electrical resistance, magnetoresistance, and Hall coefficient were studied at the 5 – 300 K temperature range under magnetic field up to 8 T. A specific and unexpected behavior has been found concerning the average size and shape of Bi grains with increasing of the current density: the increase in deposition current density up to 70.0 mA/cm² contributes to Bi films formation with more isotropic and reduced average grains size. Samples synthesized under 0.18 mA/cm² current density after annealing showed electronic properties, including magnetoresistance, similar to those of flawless single-crystals. The differences in the electronic characteristics of Bi films electrodeposited under different currents and electrolyte composition were explained by changes in electronic mobility due to scattering on grain boundaries.
... In recent years, research on the preferred orientation of Bi thin films has made some progress (Marcano et al., 2010(Marcano et al., , 2013Liao et al., 2014;Fedotov et al., 2015Fedotov et al., , 2016. Wu found that Bi thin films with (111) preferred orientation could be obtained on glass substrates at temperatures ranging from À40 to 93 C via pulse laser deposition, while the orientation was random when the substrate temperature rose to 200 C (Wu & Chern, 2008). ...
Full-text available
The preferred orientation growth characteristics and surface roughness of polycrystalline bismuth (Bi) thin films fabricated on glass substrates using the molecular beam epitaxy method were investigated at temperatures ranging from 18 to 150°C. The crystallization and morphology were analyzed in detail and the polycrystalline metal film structure-zone model (SZM) was modified to fit the polycrystalline Bi thin film. The boundary temperature between Zone T and Zone II in the SZM shifted to higher temperatures with the increase in film thickness or the decrease of growth rate. Furthermore, the effect of the thickness and surface roughness on the transport properties was investigated, especially for Bi thin films in Zone II. A two-transport channels model was adopted to reveal the influence of the film thickness on the competition between the metallic surface states and the semiconducting bulk states, which is consistent with the results of Bi single-crystal films. Therefore, the polycrystalline Bi thin films are expected to replace the single-crystal films in the application of spintronic devices.
... However, the disadvantages of these alloys are limits of the number of useful alloys that can be made. The second stage is the development of complex ternary and quaternary nitride coatings, having attracted significant research and industrial interest due to the unique combination of structural characteristics and exceptional mechanical properties [5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22][23]. ...
In this study, (Zr-Ti-Nb)N, (Zr-Ti-Cr-Nb)N and (Zr-Ti-Cr-Nb-Si)N nitride coatings were obtained using a well-developed vacuum arc deposition. The systematical investigations demonstrate that the chemical composition, microstructure, and properties of the coatings intimately rely on the deposition parameters (pressure of working gas and substrate bias). Effects of Cr and Si additions on microstructure and mechanical properties of the (Zr-Ti-Nb)N coatings have been investigated using scanning electron microscopy (SEM) equipped with energy dispersive spectrum (EDS), X-ray photoelectron spectroscopy (XPS), X-ray diffraction (XRD), transmission electron microscope (TEM), hardness measurements and adhesion testing. First-principles band-structure calculations and Gibbs-Rosenbaum triangle representation have been used to investigate the elemental and phase compositions in nitride coatings. The multi-component (Zr-Ti-Cr-Nb-Si)N and (Zr-Ti-Nb)N coatings are found to be a simple face-centered cubic (FCC) solid solution. For the coatings without Si, the structure is mainly composed of TiN fcc phase and Cr2N trigonal modification. The hardness values were in the ranges (24–42 GPa). The (Zr-Ti-Nb)N, (Zr-Ti-Cr-Nb)N coatings provided the best adhesive strength in different conditions. The (Zr-Ti-Cr-Nb-Si)N coatings exhibited the worst adhesive strength, which may be attributed to the relative low hardness.
Structure and properties of (TiHfZrVNbTa)N nanostructured multicomponent coatings implanted with a very high (10¹⁸ cm–2) dose of N⁺ions have been studied. As a result of the implantation a multilayer structure has been formed in the surface layer of the coating. The structure is composed of amorphous, nanocrystalline (disperse) and nanostructured (with the initial sizes) nanolayers. In the depth of the coating two phases (with the fcc and hcp structures) having a small volume content are formed. The nitrogen concentration near the surface attains 90 at % and then decreases with the depth. In the initial state after the deposition the coating nanohardness values are from 27 to 34 GPa depending on the conditions of the deposition. As a result of the implantation the hardness is decreased approximately by the depth of the projective ions range, i.e., to 12 GPa and then increases with the depth to 23 GPa. The investigations were conducted using the Rutherford backscattering, scanning electron microscopy with the microanalysis, high resolution electron microscopy (with local microanalysis), X-ray diffraction, nanoindentation, and wear tests.
Full-text available
This paper reviews experimental research on nanocomposite protective coatings of various chemical compositions and structure. For adaptive multielement and multilayer systems with specific phase composition, structure, substructure, stress state, and high functional properties, formation conditions are considered; the behavior of such systems under extreme operating conditions and in tribological applications is examined; the structural, phase, and chemical composition are discussed as well as the hardness, friction and wear at elevated temperatures; and the adhesive strength of hierarchical protective coatings is analyzed. Finally, the adaptive behavior under different tribological test conditions of multifunctional, multilayer coatings as a function of their properties and structure is examined.
Full-text available
Bismuth telluride and its alloys are widely used as materials for thermoelectric refrigeration. They are also the best materials for use in thermoelectric generators when the temperature of the heat source is moderate. The dimensionless figure of merit, ZT, usually rises with temperature, as long as there is only one type of charge carrier. Eventually, though, minority carrier conduction becomes significant and ZT decreases above a certain temperature. There is also the possibility of chemical decomposition due to the vaporization of tellurium. Here we discuss the likely temperature dependence of the thermoelectric parameters and the means by which the composition may be optimized for applications above room temperature. The results of these theoretical predictions are compared with the observed properties of bismuth telluride-based thermoelements at elevated temperatures. Compositional changes are suggested for materials that are destined for generator modules.
Grain structure and the temperature dependences of resistivity, magnetoresistance, Hall and Seebeck coefficients measured in the range from 4 to 300 K were investigated for polycrystalline bismuth films obtained by the melt spinning (MS) and electrochemical deposition (ECD) methods. Charge-carrier concentration and mobilities were calculated assuming the carrier scattering on acoustic deformation potential as the dominant scattering mechanism, parabolicity of holes dispersion law, implying the Lax model for L-band electrons and neglecting the influence of L-band holes on conductivity. The experimental results and calculations have demonstrated that the electrical properties of the Bi films studied are strongly affected by the grain-boundary density.
1. Preliminary concepts 2. Conductance from transmission 3. Transmission function, S-matrix and Green's functions 4. Quantum Hall effect 5. Localisation and fluctuations 6. Double-barrier tunnelling 7. Optical analogies 8. Non-equilibrium Green's function formalism.
We present a mean free path limitation model to describe the temperature dependence of both resistivity and Seebeck coefficient for bismuth nanowire. Since the mobility of carriers for bismuth nanowire was limited due to dominant collision at wire boundary, the effective mobility for each carrier was estimated using cyclotron mass, appropriate band structure, and temperature dependence of Fermi energy from 4 to 300 K. Then, the resistivity and the Seebeck coefficient were calculated by using carrier density reported for bulk single crystal. In addition, an individual single-crystal bismuth nanowire sample (725 nm diameter and 2.37 mm length) grown into a quartz template was prepared to estimate the model, and the measurements were also performed. The temperature dependences of not only resistivity, but also Seebeck coefficient were quantitatively and qualitatively in very good agreement in the whole temperature region by using its crystal orientation measured from Laue measurement. We conclude that the mean free path limitation model proposed made us understand the temperature dependences of single-crystal bismuth nanowire without a finite size effect.
The Seebeck coefficient S of single crystals of bismuth has been measured between room temperature and about 250°C. The results are used to resolve a discrepancy in the data on the Seebeck coefficient of polycrystalline bismuth as given by Sato and by Savornin and Poggi; the values given by the latter workers are shown to be more reasonable. The anisotropy in S at 300°K is analyzed in terms of the current model for the carriers in bismuth. Abeles and Meiboom's results from magnetoresistance measurements and the present data are used to calculate the density-of-states effective masses of the electrons and holes respectively. The value for the electrons agrees well with that deduced from de Haas-van Alphen-effect and cyclotron resonance experiments. The value for the holes is very close to that deduced from cyclotron resonance, but much smaller than is suggested by electronic heat-capacity data.