Sb alloys for thin film thermoelectric devices

Sb alloys for thin film thermoelectric devices

Nuclear Instruments and Methods in Physics Research B7/8 (1985) 566-570 North-Holland, Amsterdam 566 ION BEAM MIXING TO PRODUCE Bi/Sb ALLOYS FOR T...

682KB Sizes 0 Downloads 41 Views

Nuclear Instruments and Methods in Physics Research B7/8 (1985) 566-570 North-Holland, Amsterdam

566

ION BEAM MIXING TO PRODUCE Bi/Sb

ALLOYS

FOR THIN

FILM

THERMOELECIRIC

DEVICES A.M. IBRAHIM and D.A. THOMPSON Department

of Engineering

Physics and institute for Materials

Research. MeMaster

University,

Hamilton.

Ontano,

Canada US

4MI

Ion beam mixing using 80 keV Ar+ and 120 keV Kr+ has been used to produce thin films of Bi-Sb with various compositions of Sb ranging from 5% to 49%. In addition, n- and p-type doping of the films was attempted by the addition of Se and Sn during the mixing process. Rutherford backscattering analysis indicated that uniform composition films were formed after doses of 1.5-3 x 10”’ Ar+crn-* and 7 x lOI Kr+cm-*. TEM measurements show that grain growth occurs during the mixing. Thermo-electric measurements indicate that the Seebeck coefficient, S, has a maximum value of 66 pVUV/Kfor an Sb concentration of 13%. This alloy also exhibits a 30% increase in the figure of merit over that of a pure Bi film. Attempts to dope the Bi-Sb films using Sn and Se resulted in no change in S but an increase of 35% in the Seebeck voltage, Vs.

resistivity p. and Seebeck voltage temperature.

1. Introduction

Thin film thermoelectrics have a wide range of applications which include power generation [l], radiation detection [2] and surface temperature measurements (31. Thin film thermoelectric generators are of increasing interest due to the growing applications in microelectronics that require light weight power generators. Recently, ion-beam mixing has been reported [4] as a potential technique for preparing thin film thermoefectries. One criterion for determining the usefulness of an alloy for thermoelectric applications is based on the figure of merit of the alloy, Z, which is defined as [S]: Z = S=o/K,

(1)

where, S is the Seebeck coefficient = dV,/dT (V, being Seebeck voltage), T is the temperature, u is the electrical conductivity, and K is the sum of the electronic (K,) and lattice (K,) conductivities. Therefore, for a high Z, the alloy should have a large S and a/K ratio. The requirements of high u and low K implies a high carrier mobility, g, and a low lattice thermal conductivity, K,. Hence, the requirement of high S necessitates a material with a positive energy gap. Bi-Sb alloys, with low Sb concentrations have been shown to satisfy these requirements [a]. In the present work, ion-beam mixing using Kr+ and Ar+ beams was employed to prepare thin films of Bi-Sb alloys with various Sb concentrations. Rutherford backscattering (RBS) was used in situ to monitor the mixing. Structural characterization was performed using transmission electron microscopy (TEM). Electrical and thermoelectrical characterization was carried out by measurements of Hall coefficient R,, electrical 0168-583X/85/$03.30 0 Elsevier Science Publishers B.V. (North-Holland Physics Publishing Division)

V, as funtion

of

2. Experimental Samples of two basic configurations were prepared vacuum (- 1 X 10v4 Pa) evaporation of Bi and Sb thin glass substrates: A layer of Bi on top of a layer of Sb; A layer of Sb between two layers of Bi. The rate of deposition was 0.5-l rim/s and the overall thickness of a sample was 50 nm. Doping elements were introduced by the deposition of Se (donor) of Sn (acceptor) between the Sb and Bi layers. The evaporated samples were bombarded, at room temperature, with 80 keV Ar+ or 120 keV Kr+ beams. The beam energy was selected such that the maximum in the elastic energy deposition occurs near the interface of the Bi/Sb layer. Ion beam current densities were - 300 nA cm-* for Ar+ or - 180 nA cm-* for Kr+, otherwise severe sputtering occurs. For Ar+ current densities lower than 150 nA cm-*, carbon build-up was observable. The mixing beam was rastered over the sample to ensure a uniformly bombarded area of about 0.3 cm*. The RBS technique using 2 or 2.9 MeV He+ and a scattering angle of 160” was used to analyse each sample in situ before and after each bombardement. Spectra were recorded for both normal and 60” from normal incident beams. Structural characterization by TEM was carried out on the ion beam mixed and as evaporated samples. For these measurements the Bi-Sb films were evaporated on NaCl substrates. Electrical resistivity p, Hall coefficient R,, measurements were carried out by on (i) (ii)

A.M.

Ibrahim,

D.A. ntompson / Ion beam mixing to produce Bi/Sb

using the van der Pauw technique [7] over the temperature range 4.8-300 K using a m&d&d veduh 6f a previously reported system [8]. Electrical leads were attached to the samples with silver paint. A magnetic field of 1.5 T was used for the Hall coefficient measurements. Thermoelectrical characterization of the mixed samples was carried out by measuring the Seebeck voltage as a function of temperature. Two copper electrodes, pressure contacting the sample, were used to measure the thermal voltage. One contact was heated while the other was maintained at room temperature. Evaluation of the figure of merit, Z, as given by eq. (1) would normally require an absolute measurement of the thermal conductivity which is difficult to accomplish for a thin film sample. Instead, the Harman technique [9] was used to get an estimation of the change in Z due to mixing. This technique is based on measuring the DC and AC resistivities of the sample. The DC measurement simulates isothermal conditions while the AC measurement, with frequency higher than the inverse of the diffusivity, simulates adiabatic conditions. These measurements lead to a value of (ZT) through the following relationship [lo]; ZT=P.,,/PT.X

-

1,

(2)

where, pAc and pDc are the AC and DC resistivities respectively, and T is the temperature in K. However, to obtain absolute values for Z from such measurements several correction factors are required. These have been discussed by Trefny [lo].

3. Results and discussion The result of RBS analysis, using 2 MeV He” at normal incidence for a sample of configuration (i) is shown in fig. 1 before and after mixing with 80 keV Ar+ . The Sb surface channel is indicated by the arrow. A dose of 3 X lOi Ar+cmT2 was required for complete mixing. RBS analysis, using 2.9 MeV He+ at 60* incidence, for a sample of configuration (ii) is shown in fig. 2. After a dose of 1.5 x 1016 Ar+cme2, the dip between the Bi peaks disappeared and Sb is seen to extend to the surface. For Kr” bombardment, the same degree of mixing occurs after a dose of 7 X 10” KrTlcmm2. The results of TEM studies are given in fig. 3. The diffraction pattern obtained shows the film to be polycrystalline. The complexity of the diffraction pattern before mixing is due to the existence of separate layers of Bi and Sb. The complexity disappeared after forming the alloy. An increase of about 30% in the grain size was observed as a result of Kr+ bombardment. Similar observations have been reported previously [1l-1.51. Results of measuring the Seebeck voltage, L’s, as a function of temperature for Bi-Sb alloys with Sb con-

567

alloys

oxx 4 ,

,

360

370

360

390 ,400 4lO CHANNEL NUMBER

420

--. Fig. 1. Backscattering spectra using 2 MeV He’ for a sample composed of 25 nm of Bi on top of 25 nm of Sb.

centrations - 49%, - 23% and - 13% are shown in fig. 4. The Seebeck coefficient has a maximum value of 66 pV/K for an alloy containing - 13% Sb and decreases to 60 pV/K at 5% Sb and 26 pV/K at - 49% Sb. The polarity of V, is negative indicating n-type conductivity which was also confiied by Hall effect rn~~ern~ts. The behaviour with respect to Sb concentration can be attributed to the dependance of the energy gap on the Sb concentration, X. Bi-Sb alloys have been reported [ 16-201 to be semiconductors in the range of 0.04 s x < 0.22. To confirm semiconductor type behaviour, the electrical resistivities of Bi. and Bi-Sb alloy were mea-

d P 2.4 ii! F

CHANNEL NUMBER Fig. 2. Backscattering spectra using 2.9 MeV He+ for a sample composed of 5 nm Sb between two 25 nm layers of Bi. VIII. ION BEAM MIXING

568

Fig. 3. TEM micrographs,

A.M.

Ibrahim,

(a) before

D.A.

Thompson / Ion beam mixing to produce Bi/Sb

olioys

mixing, (b) after mixing with 7X 10’5Kr+cm-2.

sured as a funtion of temperature. The results are shown in fig. 5. For pure Bi, the resistivity decreases monotonitally with temperature which is consistent with the fact

that thin film Bi may behave like a semiconductor 1211, whilst bulk Bi is semimetallic. The Bis7-Sb,3 alloy exhibited a constant resistivity 01rrer the range 4.8 K to

A.M.

0

Ibrahtm,

D.A.

Thompson / Ion beam mixmg to produce Bi/Sb

569

Bi/Sb implanted wlth 3 80

lye fW$/$nplanted . o

alloys

CA 2(*--A&

AA Ref. (1) 0 Ref. IlO) \

Bi/Sb/Bi implanted wlth 1.5~10” Ak’/cm* (-23% Sb)

I2

ATOMIC PERCENT OF Sb Fig. 6. Variation of the corrected Seebeck coefficient content in undoped Bi-Sb thin film alloys.

ELECTRODE TEMPERATURE (K) Fig. 4. Seebeck voltage versus temperature for three alloys with different Sb concentrations.

K. Above 200 K the resistivity decreases with temperature. Similar behaviour has been reported [22] for 1% Sb doped Bi films of various thicknesses. To compare the present results of S with those for Bi-Sb alloys prepared by different techniques [l], one has to take into account the difference in sample thick200

versus Sb

ness between the present work and that previously reported. In the present work, the alloy thickness is about 50 nm while that reported in ref. [l] had a thickness of 1 pm. Assuming that the thickness dependence of S for Bi-Sb alloys is the same as that of pure Bi, the correction for the thickness can be carried out making use of the results of Favemrec et al. [23] who measured S versus thickness for pure Bi films. The corrected data are shown in fig. 6 along with those of Trefny [l,lO]. The possibility of a change in S due to annealing by the repeated heat cycles was investigated by repeated measurements of S. The results for up to ten cycles of measurement show no variation in the value of S.

-1.51

“St 50

Fig. 5. Resistivity Bis,-Sb,, alloy.

100 150 200 250 Temperature (K) versus

temperature

4-

2300

350

OI

‘1

300 for pure

Bi and

a

I

I

I

I

I

I

I

I

I

,I

340 380 420 460 500 ELECTRODE TEMPERATURE (K)

Fig. 7. Effect of doping elements on V, versus temperature. VIII. ION BEAM MIXING

570

A.M. Jbrahrm, D.A. Thompson / Ion beam mixing to produce Bi/Sb

References

Table 1

Comparison of ZT for Bi and Bis,-Sb,,

thin films

Thin film

Current (A) (DC or RMS)

ZT

Bi

9.6X10_7 6.4~ IO-’

0.068 0.1

Bts,-Sb,,

alloys

Subsequent RBS analysis indicated no change in composition or redistribution after the repeated heating cycles. The effect of doping elements on the absolute value of V, and S is shown in fig. 7. This shows that there is a slight enhancement in the absolute value of V’; however, there is no change in the value of S, in agreement with the results reported [24] for bulk Bi-Sb alloy doped with Pb (acceptor). The variation in ZT due to mixing was probed by comparing the results for a mixed Bi,-Sb,, alloy and pure Bi. The results are given in table 1. These measurements predict 30% increase in ZT of the alloy over that of pure Bi.

4. summary and conelusiona Uniform thin film Bi-Sb alloys of various compositions can be prepared by ion beam mixing Using multilayer samples allows better control over the final alloy composition and requires lower ion doses to achieve films with uniform composition compared to bi-layer samples. The ~e~~l~t~c properties have’been found to depend on the final composition of the thin film with the best performance being achieved with a Bi,,-Sb,, alloy. It showed an increase of about 30% in ZT over that of pure Bi. Doping with Se and Sn did not affect the absolute value of S. We wish to thank Dr J.A. Davies for many useful discussions during the progress of this research. This work was funded by the Natural Sciences and Engineering Research Council of Canada.

[I] J.U. Trefny, Proc. 16th Intersociety Energy Conversion Engineering Conference, Atlanta, GA (1981) p. 2019. [2] K.L. Chopra and D.K. Pandya, Thin Solid Films 50 (1978) 81. [3] K.G. Cooper and J.P. Lloyd, J. Sci. Instr. 42 (1965) 79. [4] R.E. Benenson. V.K. Tikku, D. Kollewe and H.-S. Jin, Nucl. instr. and Meth. 209/210 (1983) 185. [S] SW. Angrist, Direct Energy Conversion (Aliyn and Bacon, Boston 1982). [6] R.B. Horst and L.R. Williams, Proe. 3rd Intern. Conf. on Thermoelectric Energy Conv. (1980) p. 139. [7] L.J. van der Pauw, Philips Tech. Rev. 13 (1958) 1. [8] J. Shewchun, K.M. Ghanekar, R. Yager, H.D. Barber and D.A. Thompson, Rev. Sci. Instr. 12 (1971) 1797. [9] T.C. Harman, J.H. Cahn and M.J. Logan, J. Appl. Phys. 30 (1959) 1351. [lo] J.U. Trefny, Proc. 4th Intern. Conf. on Thermoelectric Energy Conversion (sponsored by IEEE), Arlington,Texas (1982) p. 79, [ll) B.Y. Tsaur, S.S. Lau, L.S. Hung and J.W. Mayer, Nuci. ltnstr. and Meth. 182/183 (1981) 67. (121 B.Y. Tsaur and N. Mile@&, J. Appl. Phys. 52 (1981) 728. [13] B.Y. Tsaur, S.S. Lau and J.W. Mayer, Phil. Mag. 44 (1981) [14] FY. Tsaur, S.S. Lau and J.W. Mayer, Appl. Phys. Lett:36 (1980) 823. [15] S.W. Chiang, T.P. Chow, R.F. Reihl and K.L. Wang, J. Appl. Phys. 52 (1982) 4027. [16] G.A. Mironova, M. Sudakova and Ya G. Ponomorev, Sov. Phys. JEF’T 51 (1980) 2124. 1171 G.A. Mironova, M. Sudakova and Ya G. Ponomorev, Sov. Phys. Sol. State 22 (1980) 2124. [18] N.A. Rodionov, N.A. Red’koand G.A. Ivanov, Sov. Phys. Sol. State 21 (1979) 1473. [lQ] N.A. Rodionov, G.A. Ivanov, K.G. Ivanov and N.A. Red’ko, Sov. Phys. Sol. State 23 (1981) 1987. 1201 N.B. Brandt, R. Hermann, G.I. Golysheva, L.I. Devyatkova, D. Kusnik, W. Kraak and Ya.G. Ponomarev, Sov. Phys. JEPT 56 (1982) 1274. 1211 V. Damodara Das and N. Jayaprakash, Vacuum 31 (1981) 841. 1221 V. Damodara and MS. Jagadeesh, J. Vat. Sci. Teehnol. 19 (1981) 89. [23] M.M.E. Favennec, M. LeContellec and J.Y. LeTraon, Thin Solid Films 13 (1972) 73. [24] G.E. Smith and R. Wolfe, J. Appl. Phys. 33 (1962) 841.