Electrical conductivity and thermoelectric power in amorphous As2Se3Tlx semiconductors

Journal of Non-Crystalline Solids, 12 (1973) 168-176. © North-Holland Publishing Company

ELECTRICAL CONDUCTIVITY AND THERMOELECTRIC I N A M O R P H O U S As 2 Se 3 T1x S E M I C O N D U C T O R S *

POWER

R. STRUNK Institut f~r Werkstoffe der Elektrotech nik, Rheinisch- Westfdlische Technische Hochschule Aachen, Aachen, Germany

Received 18 December 1972

Dc conductivity and thermopower measurements as a function of temperature have been performed on solid As2 Se3T1x glasses with x ranging from 0 to 1.99. Investigations have been extended to low temperatures. By substitution of T1 the conductivity can be increased by more than 5 orders of magnitude without changing the type of transport mechanism. Band-like conduction is observed with holes being the dominant charge carriers. The three-dimensional network of AszSe3 is assumed to be modified by the T1 atoms; this leads mainly to a decrease in the activation energy.

1. Introduction In order to test present models [ 1 - 3 ] of electronic states and charge transport in amorphous semiconductors, both structural analysis and experiments concerned with electrical and optical properties have been performed. However, in spite of various experiments on a large variety of semiconducting materials there is rather little information about systematic analysis of one definite system. Thus, an uncertainty may often arise as to whether a phenomena observed is merely due to the special composition used, or can be interpreted as a general property of amorphous semiconductors. Moreover the variation of structural parameters corresponding to compositional change may give further information about the influence of the type and the degree of disorder on electronic properties. We have studied electronic transport phenomena o f solid amorphous semiconductors in the ternary system As-Se-T1, which exhibits a large glass-forming region [4] with compositions ranging from medium to high resistivity material. This paper deals with dc conductivity and thermoelectric power measurements on As2Se3T1x glasses. Such measurements are suitable for estimating the electrical energy gap and its temperature coefficient, the sign of dominant charge carriers and the kind of conduction mechanism [3]. Further interpretations concerning the * Work supported by Deutsche Forschungsgemeinschaft, Sonderforschungsbereich 56 Festk6rperelektronik.

R. Strunk, Amorphous As2Se3Tlx semiconductors

As

20

40 ~ at%TI

80

169

f[

Fig. 1. Composition of probes studied in the system As-Se-TI. Glass forming region after Flaschen et al. [4]. origin of the conduction process may be somewhat dubious. Separate calculations of both carrier density and mobility and their respective dependence upon composition are not straightforward and additional investigations are needed.

2. Samples Four different compositions of the system As2Se3T1x have been prepared with x = O, 0.38, 0.97 and 1.99; i.e. the parent glass pure arsenic selenide and with three different amounts of 'impurities' of T1 (fig. 1). The elementary constituents (purity 99.995%) of the specific glass compositions were mixed together and sealed in evacuated quartz ampoules. The ampoules were heated at 900°C for 8 h and air quenched afterwards. Flat cylindrical samples of 2 to 5 mm thickness and of about 20 mm in diameter were prepared from the raw material by melting and pressing the glass under an Ar atmosphere near its softening temperature. The plane surfaces were polished and for conductivity measurements gold contacts were evaporated on both sides. The homogeneity of the specimen was checked by electron probe microanalysis. X-ray diffraction measurements were performed to be sure that the material was amorphous (in that sense), before and after melting and pressing.

3. Apparatus For conductivity measurements an evaporating cryostat with a supplementary electronically regulated heater was used. Thus a temperature range from -190°C (cooling with liquid nitrogen) to 200°C could be covered. The probe chamber was filled with helium in order to ensure thermal equilibrium soon after altering the

170

R. Strunk, Amorphous As2 SeaT1x semiconductors

temperature. The actual lower and upper temperature limits for a certain measurement were determined by the resistance and softening temperature, respectively, of the specimen under test and they depended on the specific composition. The electrical conductivity was determined by two-terminal measurements using a Keithley model 640 vibrating capacitor electrometer. Current measurements were possible down to about 10 -15 A with electrical field strengths up to 104 V/cm. Thermoelectric power measurements were performed under vacuum (10 -2 Torr) using a modified Tettex conductivity measuring cell. Details of the apparatus are described in ref. [5]. Similar arrangements for measuring Seebeck coefficients of high resistivity material are described in ref. [ 6 - 8 ] . The thermopower of As2Se3TIx semiconductor probes was studied in the temperature range from room temperature to 110°C; temperature differences used were between 2 and 8 deg C. The total insulating resistance between the electrodes was greater than 1014 [2 for all temperatures applied. This proved to be satisfactory even for the compositions with lower conductivities. However, large time constants appeared due to the low thermal conductivity of the samples and bad thermal contact between the electrodes and the necessarily insulated heating equipment. In order to obtain a constant temperature gradient across the specimen the temperature difference and - as far as possible - the reference temperature were kept constant for at least 2 h. With the high ohmic electrometer input connected to the 'hot' electrode (the 'cold' electrode was connected to circuit ground) the voltage between the electrodes was recorded by a line recorder until no systematic increase or decrease of meter deflection could be stated. These quasi-terminal values were averaged graphically and plotted for different values of temperature and temperature difference. Seebeck coefficients were determined by extrapolating at zero temperature differences.

4. Experimental results 4.1. Reliability o f measurements During the investigation of electrical properties of semiconducting material several errors may arise, due to blocking contacts (and other non-ohmic effects), diffusion of contact material or annealing. The following experimental procedure was performed to check the reliability of results discussed below: For each probe the current-voltage characteristic was measured in the whole temperature range. No field dependence could be observed up to 104 V/cm, the upper limit of field strengths applied. Comparative measurements with evaporated A1, In and Ni contacts gave identical results within a few per cent. The samples could be heated and cooled down again without any irreversible change in conductivity and thermopower. The repro° ducibility of thermoelectric power measurements was especially examined by reversing cold and hot electrodes and measuring with increasing and decreasing temperature. The deviations did not exceed the normal scatter of values.

R. Strunk, Amorphous As2 Se3Tlx semiconductors

102 ! ~ G

o As2Se3 ~

10-16

n

.....

~ As2SeaT/'~

~\\~

I

I

2

3

~

1

5

171

6"10-31( -~

l/f ~ Fig. 2. Temperature dependence of dc conductivity of different As2SeaT1x semiconducting glasses. Dashed lines denote extrapolation to T = =.

~

2

~

I

10-14

0

tO

20

30

~0

at%T~Fig. 3. Dc conductivity at room temperature and extrapolated values at T = = versus Tlconcentration in the system As2 SeaTlx. ix(=) corresponds to pre-exponential factor of eq. (1). + values extrapolated at 20°C by Kolomiets [ 13 ].

We conclude that our results of conductivity and thermoelectro power measurements can be interpreted as bulk properties o f the material. 4.2. Conductivity Fig. 2 shows the electrical conductivity as a function of reciprocal absolute temperature. For all compositions studied, the conductivity is a negative exponential function of temperature. The thermal activation energies AE a are derived from the constant slopes of the semilogarithmic plots, assuming a temperature dependence of conductivity of the form o(T)

=

C e x p ( - AEa/kT).

(1)

With increasing T1 content the conductivity increases by a factor of more than 105. Fig. 3 shows the room temperature conductivity as a function of T1 content and the extrapolated values (see fig. 2) at T = ~¢, corresponding to the pre-exponential factor of eq. (1). The conductivity for T = oo is nearly independent of the T1 concentration, indicating that a change in composition mainly affects the activation energy. In fig. 4 the conductivities of different compositions are plotted against the corresponding band gaps 2AEa, showing that the increase of conductivity with increasing

172

R. Strunk, Amorphous As2SeaT1x semiconductors

~5 10.7

10-s

"~

~se3nQ~

10"9 I0-~°

r=20°C

10-11 i0-12 10-~31 09

~XxAszSe~ q, 11

~3

1.5

17 eV 19

2~Ea= Fig. 4. Dc conductivity at room temperature of different compositions as a function of corresponding electrical bandgap.

TI-content is due to a decrease of activation energy. Only for the highest amount of TI does an additional decrease of C appear. Thus, compositional independence of C, observed for many amorphous semiconductors [9], holds for the As2Se3Tlx system as well, except for compositions near the edge of the glass-forming region.

4.3. Thermopower The thermoelectric power measurements indicate p-type conduction for all compositions studied. Fig. 5 shows the Seebeck-coefficient plotted as a function of lIT so that activation energies AE s can be calculated using the relation s(z3 =

In accordance with conductivity measurements the thermopower decreases with increasing T1 content (fig. 6), i.e. with increasing conductivity. In addition we have plotted the Seebeck-coefficients extrapolated at T = ~, which corresponds to (k/lel)A of eq. (2). For all compositions the constant A is approximately zero. Thus, like conductivity the thermopower depends on activation energy only. Fig. 7 shows the activation energies Z~7 deduced from conductivity (z~?o) and thermopower measurements (AEs) for the different compositions. Both energies decrease exponentially with T1 concentration. The activation energy of conductivity is always greater than ~ E s and the quotient seems to be independent of the T1 content. Unfortunately, the error in thermopower measurement is rather high so that details of interpretation become somewhat ambiguous. •

R. Strunk, Amorphous As2 Se3TIx semiconductors

173

T I

I

I

I

3

mV

/

-k-

S

I

r

3

~ ,

~

S e

25

27

-

-

i m--K V'k,,

-

l ./~///

~

""~

,.zo'c -

As~Se3l U[

29

3340qK

31

I0

I --T

/ " " T',

40

Fig. 6. Seebeck coefficient at 20°C and the extrapolated values at T = ~ versus TI concentration in the system As2Se3T1x. S(oo) corresponds to (k/lel)A of eq. (2).

f ~

30 atzr7

Fig. 5. Temperature dependence of Seebeck coefficients of different As2 Se aT1x semiconducting glasses. For comparison, the results of Edmond [10] for liquid As2Se a (a) and of Callaerts et al. [12] for liquid and solid As2Se 3 at higher temperatures (b) are plotted.

"°L

20

]

-

i ~r~'~'~=

of%T/"~ Fig. 7. Activation energy determined by different methods versus TI concentration. AE a and AE s are inferred from the temperature variation of conductivity and of the Seebeck coefficient respectively. The additional marks refer to conductivity (+) and photoconductivity (×) measurements by Kolomiets [13].

174

R. Strunk, amorphous As2SeaTlx semiconductors

5. Discussion Liquid and solid As2Se 3 have been studied by many workers [7-14] although not yet in a low-temperature range. Our results of conductivity and thermopower measurements are in good agreement with those reported in literature. From conductivity data one can conclude that for As2Se 3 no change in transport mechanism appears over the whole temperature range where conductivity measurements are possible. In fig. 5 our results of thermopower measurements on As2Se 3 near room temperature are compared with those of Edmond [10] on liquid As2Se 3 and of Callaerts et al. [12] on liquid and solid As2Se 3 at higher temperatures. The Seebeck coefficient is proved to increase continuously with decreasing temperature from the liquid to the solid state, in agreement with conductivity. The conductivity of the ternary system As2Se3T1 x has been investigated by Kolomiets et al. [13] and for small amounts of T1 by Legal et al. [14]. Both authors found a decrease of conductivity and an increase in activation energy with increasing concentration of T1 up to 0.6 at % and 7 at % respectively (region of dashed lines in fig. 3). This is assumed [14] to be due to a diminishing short-range order at small T1 contents, deduced from microstructural X-ray analysis. For greater amounts of T1, increasing conductivities and decreasing activation energies are likewise reported. However, these results are only in qualitative agreement with one another and with our measurements. For comparison the results of Kolomiets et al. (measured in a higher temperature range and extrapolated at room temperature) are marked in fig. 3 and fig. 7. We assume that in As2Se3T1 x glasses with x ~> 0.6 the network of As2Se 3 is modified rather than simply broken up by the substituted T1 atoms. The continuous change of electrical and thermoelectric properties with composition can hardly be explained by the assumption that the T1 atoms act as !mpurity centers. The uniform log o - 1/T and S - 1/T dependence of all compositions studied suggests that bandlike conduction is the dominating transport mechanism in these glasses, i.e., in terms of current models [1-3] of non-crystalline semiconductors: mainly carriers in nonlocalized states participate in charge transport. Using the expressions valid for crystalline, non-degenerate semiconductors [15 ] the positive sign of the Seebeck coefficients together with the linear S - 1I T relations suggest either unipolar conduction with exceeding hole concentrations (p >2> n), or ambipolar, intrinsic conduction with exceeding hole mobilities (n = p, ]~h >/'re)" The first case would lead to AE o = AxEs,which does not agree with our experimental results, except for a thermally activated mobility [3, 16] of the form AE,

This leads to ~ E s = AE o - AE u < AE a. From fig. 7 one finds (AE o - AEs) ranging from ~ 0.1 eV for As2Se 3 to ~ 0.04 eV for As2Se3Tll.99. These values seem to be

R. Strunk, Amorphous As2 Se3Tlx semiconductors

175

rather low compared with results of drift mobility measurements on As2Se 3 reported in the literature [3, 17] and our own results of mobility and photoconductivity investigations on As2Se 3Tlx glasses [ 18]. For the second case ~Ts AEa

/ah -- #6

- - - < 1 . /'th + ~e

From fig. 7 one finds AEs/AE a ~- const. ~- 0.9 which gives gth//2e ~-- 20, independent of composition. A linear temperature dependence of the activation energy such as AE = AE 0 - t3T leads to s(oo) - ,

_ n/k,

where 6 depends on the scattering process [15]. Thus, with S(o°) ~ const. ~ 0 (fig. 6):/~ ~- k6, independent of composition. For in-band conduction in amorphous semiconductors ~ should be of order unity [19] so that t3 ~- k = 8.6 × 10 -5 eV/K. The temperature coefficient of the band gaps deduced from optical absorption and photoconductivity measurements is about 3 × 10 -3 eV/K [3, 13, 18]. However, as long as optical and electrical band gaps are not identical (perhaps not even correlated [20] ) different temperature coefficients may exist as well.

Acknowledgement The author is grateful to Prof. A. Hersping for several ideas concerning this work and wishes to thank Mr. J. Brunk and Mr. N. Pinkert for their help with the performance of protracted experiments.

References [1] E.A. Davis and N.F. Mott, Phil. Mag. 22 (1970) 903. [2] M.H. Cohen, H. Fritzsche and S.R. Ovshinsky, Phys. Rev. Lett. 22 (1969) 1065. [3] N.F. Mott and E.A. Davis, Electronic Processes in Non-Crystalline Materials (Clarendon Press, Oxford, 1971). [4] S.S. Flaschen, A.D. Pearson and W.R. Northover, J. Amer. Ceram. Soc. 42 (1959) 450. [5 ] W. Biicker, Untersuchung der elektrischen Eigenschaften des organischen Halbleitersystems Phenol-Formaldehyd-Harz - amorpher Kohlenstoff (Thesis, T.H. Aachen, 1972). [6] H. Mette and C. Loscoe, Rev. Sci. Instr. 37 (1966) 1537. [7] H.L. Uphoff and J.H. Healy, J. Appl. Phys. 32 (1961) 950. [8] B.T. Kolomiets and E.M. Raspopova, Soy. Phys.-Semicond. 5 (1972) 1346. [9] J. Stuke, J. Non-Crystalline Solids 4 (1970) 1. [10] J.T. Edmond, Brit. J. Appl. Phys. 17 (1966) 979.

176 [11] [12] [13] [14] [15] [16] [17] [18] [19] [20]

R. Strunk, Amorphous As2 SeaTlx semiconductors J.T. Edmond, J. Non-Crystalline Solids 1 (1968) 39. R. Callaerts, P. Nagels and M. Denayer, Phys. Lett. 38A (1972) 15. B.T. Kolomiets, Y.V. Rukhlyadev and V.P. Shilo, J. Non-Crystalline Solids 5 (1971) 402. D. Legal, V. Trkal, I. SRB, S. Dokoupil, V. Smid and V. Rosick~i, Phys. Stat. Sol. 12(a) (1972) K39. O. Madelung, in: S. Fliigge, Encyclopedia of Physics, Vol. XX, Berlin (1957). H.K. Rockstad, R. Flasck and S. Iwasa, J. Non-Crystalline Solids 8 - 1 0 (1972)326. J.M. MarshM1 and A.E. Owen, Phil. Mag. 24 (1971) 1281. K.H. Klapheck, A. Klein and R. Strunk, Inst. f. Werkst. d. E-Technik, TH Aachen, to be published. M. Cutler and N.F. Mott, Phys. Rev. 181 (1969) 1336. H. Fritzsche, J. Non-Crystalline Solids 6 (1971) 49.