X-ray photoemission and bremsstrahlung isochromat spectroscopy of bulk single crystalline SixGe1−x alloys

X-ray photoemission and bremsstrahlung isochromat spectroscopy of bulk single crystalline SixGe1−x alloys

JOURNAL OF ELECTRON SPECTROSCOPY and Related Phenomena ELSEVIER Journal of Electron Spectroscopy and Related Phenomena 88-91 (1998) 395-398 X-ray p...

331KB Sizes 0 Downloads 24 Views

JOURNAL OF ELECTRON SPECTROSCOPY and Related Phenomena

ELSEVIER

Journal of Electron Spectroscopy and Related Phenomena 88-91 (1998) 395-398

X-ray photoemission and bremsstrahlung isochromat spectroscopy of bulk single crystalline SixGel_x alloys Y. Saito a'*, S. Fujimori a, S. Suzuki a, S. Sato a, T. Honda h, M.

Suezawa b

aDepartment of Physics, Tohoku University, Sendai 980-77, Japan blnstitute for Materials Research, Tohoku University, Sendai 980-77, Japan

Abstract The electronic structures of the valence and the conduction bands of bulk single crystalline SixGe l-x alloys (x = 0, 0.05, 0.135, 0.18, 0.47, 1) are studied by X-ray photoemission spectroscopy and bremsstrahlung isochromat spectroscopy. The valence band maxima monotonically shift to a lower binding energy side as the composition x increases. The conduction band minima rapidly shift to the high binding energy side in the small composition range x = 0-0.18, and then slowly shift on the composition in the region above x = 0.18. As a result, the energy gaps increase as a quasi-linear function of composition x with a slope discontinuity at x = 0.18. We can quantitatively see from the results that the discontinuity of the band gap obtained by the optical absorption measurement mainly originates from the shifts of the conduction band minima. © 1998 Elsevier Science B.V. Keywords: SixGel x alloy; X-ray photoemission spectroscopy; Bremsstrahlung isochromat spectroscopy; Shifts of conduction band minima

1. Introduction Semiconductor alloys have attracted much interest in technological applications o f electronic and electro-optical devices because their physical properties change depending on the alloy constituents. In recent years, there has been a renewed interest in SixGe l-x alloys and superlattices. Because silicon is the most technologically advanced semiconductor, the results of experiments on S i - G e systems have many potential applications. In particular, the bulk phase behavior of SixGe x-x alloys has not only importance on practical applications but also more fundamental and physical problems as a prototype o f other semiconductor alloys. Therefore, many theoretical studies have been * Corresponding author.

reported for SixGe l-x alloys [1-5]. These alloys form a substitutional solid solution for whole composition. Since the lattice constant of Ge is about 4% larger than that o f Si, a strain can affect both the band structure and the transport properties. The band structure calculations such as virtual crystal approximation (VCA), coherent potential approximation (CPA), and more improved C P A have been reported already [2,4]. However, experimental reports are few to date. F r o m the optical absorption measurement [6], the band gap versus the composition of SixGel_x alloys relation has shown an interesting feature. This relation is roughly expressed by the linear function with different slopes in the regions around both ends of composition, which has a discontinuity at the composition of about x = 0.15. According to the band calculations, this is interpreted as the conduction band m i n i m a

0368-2048/98/$19.00 © 1998 Elsevier Science B.V. All rights reserved PI1 S0368-2048(97)00187-4

396

E Saito et al./Journal of Electron Spectroscopy and Related Phenomena 88-91 (1998) 395-398

change from L point of pure Ge to X(A) point of pure Si at this crirical composition. In this paper, we report the electronic structures of SixGex_x alloys by means of X-ray photoemission spectroscopy (XPS) and bremsstrahlung isochromat spectroscopy (BIS). Compared with optical measurements which give information on the joint density of states, these experimental methods have been used to study valence and conduction bands independently. Thus, we have observed the shifts of the edges of the valence and conduction bands. We discuss the effect of the shifts on the relation of the band gap versus the composition.

iV a l e h c e i ~ i i i i i i i i i i i i A i i i }

i

:

of ~|X~I~

:

:

:

~ ili~ .....

iiii!/'~iiil :::::::::: .....

= = = = = = = = = = = = = = = = = = = = = = = = = = = =



:

:

:

:

:

:

2. Experimental details Bulk single crystalline Si~Ge~_x alloys were grown by the traveling solvent method [7]. The composition x was determined by an electron probe X-ray microanalyzer (EPMA). We prepared the samples with the composition of x = 0.05, 0.135, 0.18, 0.47. They exhibit an n-type conductivity with a carrier concentration of the order of ~ 1015cm -3 at room temperature. Because the electric resistivity was about 1 9cm, the charging effect due to the X-ray irradiation is negligible. Both XPS and BIS measurements have been carried out by using an ESCALAB MK-II (VG Scientific Co.) photoelectron spectrometer. The pressures during XPS and BIS measurements were below 5.0 x 10 -9 and 6.0 x 10 -8 Pa, respectively. For XPS measurement, the exciting photon source used was M g K a (h~ -- 1253.6 eV) radiation. The binding energy was determined by comparison with a binding energy of the gold 4f7/2 peak of 84.0 eV. It is in agreement with the binding energy determined from the core levels of pure Si and Ge. This also proves to have no charging effect. For BIS measurement, the detected photon energy was set at 1486.6 eV. The Fermi level was determined by comparison with a gold BIS spectrum. The total energy resolutions for XPS and BIS measurements were about 0.8 eV. Clean surfaces of samples were obtained by in situ scraping with a diamond file. Contamination was checked by the presence of O ls and C ls signals. The surfaces were kept clean during the measurements through occasional filing.

~.,~-.~:: i :: ': :::::::::::::::::::::::::::::::::::::::: : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : :

r":~

::

:: i :: :: ::

:

:: : : :: :::: :: :: :: : :: : :: :: :: :: :: :: :: :: :: :: :: :: :: :: :::::::

"?,~:"71111111iiiiii!iiiiiiiiiiii i i i

-15

i :: i]i

-10

:: Iii

*~1~ .... i i

i i i i ] i

-5

:

!,'~"*~:::

iii i iii iiiii iii ii iii iiii ii

: i i i i i i i i i ! i iii

:

;: ~ ~: ~;

0

Binding Energy (eV) Fig. 1. Valence band XPS spectra of four SixGe l-x alloys (x = 0.05, 0.135, 0.18, 0.47), pure Ge (x = 0) (top) and Si (x = 1) (bottom) measured by MgKa radiation (hg = 1253.6 eV) at room temperature. These spectra are obtained by the subtraction of a secondary electron background and MgK(~3,4,5,6 satellites from the raw data. Normalizations are carried out at the maximum height. The main three structures are designated as A, B, and C.

3. Results and discussion Fig. 1 shows the valence band XPS spectra of four SixGe l_x alloys (x = 0.05, 0.135, 0.18, 0.47), Ge (x = 0) (top), and Si (x = 1) (bottom). Binding energy is taken as a minus sign for XPS and a plus sign for X-BIS. The present data shown in Fig. 1 were obtained by subtracting a secondary electron background and MgKot3,4,5, 6 satellites from the raw spectra. In Fig. 1, the main structures are designated as A, B, and C. These spectra were normalized at the height of the peak A. The valence bands of pure Si and Ge are formed from the states produced by the electron configuration of 3s3p 3 and 4s4p 3, respectively. According to the band calculations [8], it is explained that the structures

E Saito et aL/Journal of Electron Spectroscopy and Related Phenomena 88-91 (1998) 395-398

0

5

Binding Energy (eV)

10

Fig. 2. Conduction band BIS spectra of four SixGel_x alloys (x=0.05, 0.135, 0.18, 0.47), pure Ge (x = 0) (top) and Si (x=l) (bottom) measured by a detection of the photon energy (h~,= 1486.6 eV) at room temperature. The spectrum of pure Si is taken from Ref. [9]. Normalizationsare carried out at the maximum height. The least squares smoothing lines are also drawn by the solid lines. The main three structures are designated as D, E, and F. near - 2.5 eV (A) and - 10.5 eV (C) are mainly formed from p and s derived states, respectively. The middle structure near - 8 eV (B) is originated from the mixed states between s and p electrons. Although spectral shapes of pure Si and Ge resemble each other, the energy position and the relative intensity of each structure are different. As shown in Fig. 1, the valence band XPS spectra of alloys also show three structures. It is important point that each peak position in alloys shows no change even if the composition x changes. They remain at the same position as those of pure Si and Ge. It is seemed that this observation is consistent with the theoretical prediction. According to CPA, the density of states shows no shift in the average position between those of the consistuent elements, though the relative intensity changes as the concentration of constituent elements in an alloy is changed.

397

Fig: 2 shows BIS spectra for the same alloys, Si and Ge. The spectrum of pure Si is taken from Ref. [9]. Normalization is carried out at the maximum heights of the spectra. The least squares smoothing lines are also drawn by a solid line. The main three structures are designated as D, E, and F. The band calculations [10] predict that the density of states of the conduction bands of pure Si and Ge will resemble each other, due to their similar electron configurations. As shown in Fig. 2, however, the spectral shapes of pure Si and Ge are different from each other. The structures D of mainly s derived states are observed for both pure Si and Ge, and the structure F of mainly p derived states are observed for only pure Ge. This is because the photoionization cross-section of Ge 4p electron is about four times larger than that of Si 3p electrons in this energy range. These three structures are also observed for alloys, but their positions shift toward the higher binding energy side as the concentration of Si increases. This tendency is quite contrasted to the case of the present valence band spectra. In order to elucidate the shifts of the conduction band edges in the present alloy system, we tried to determine the energy positions of the valence band maxima (Ev) and the conduction band minima (Ec) by means of a simple extrapolating method. This method is schematically explained in Fig. 3(a). Linear lines by a least squares method (for pure Si) were extrapolated to the leading edges of the valence and the conduction band spectra, respectively. The cross positions of extrapolated lines and the background levels were regarded as the tentative values of Ev and Ec. In spite of uncertainty of the extrapolation method, the present analysis provides only the qualitative information for the shifts of the band edges. In Fig. 3(b), the analytical results as described above are indicated as a function of composition x. In order to compare the shifts of the edges with the band gap, the ordinate is used both as the band gap energy and binding energy. The band gap measured by the optical absorption measurement [2] at room temperature and calculated results by CPA [6] at 0 K are shown together as (i) and (ii), respectively. A slight decrease of Ev as composition x increases quantitatively indicates that the valence band maxima gradually go away from the Fermi level (EF). In contrast to this, Ec once goes rapidly away from EF as

a'iI"

Y. Saito et al./Journal of Electron Spectroscopy and Related Phenomena 88-91 (1998) 395-398

398

-"

'

I

XPS

'

""~'~

,--'-o

X

I

o

X-BIS..

'

'1

E - Ev(in Ge) (eV)

discontinuity of the slope at x = 0.18. Comparing the value of Ec - Ev with the band gap energy obtained by the optical absorption measurement and CPA calculation, the present value shows a discrepancy due to the uncertainty of the method for determing Ev and Ec. However, the composition of the discontinuity and the change of the slopes for the present results are qualitatively in agreement with (i) and (ii).

4. Conclusion 1

,i~'--............ -'~"......................... ÷

.S 0.5

]

/

,,/0 ;

C

@

!

We have measured the valence band XPS and BIS spectra of bulk single crystalline SixGel-x alloys. These spectra have shown similar structures to those of pure Si and Ge. However, the position of the valence band structures has shown no change, while the conduction band structures have shifted. Present experiments have revealed that the discontinuity of the band gap mainly originates from the shifts of the conduction band minima.

Acknowledgements -0.5

I

0

i

I

J

I

I

I

.t

I

I

[

0.5

Composition x of SixGel_x

Fig. 3. (a) The extrapolating method for determing the valence band maxima Ev and the conduction band minima Ec. (b) Ev (0) and Ec (O) estimated by the extrapolating method (Ec of pure Si is taken from Ref. [9]). Ev of pure Ge is taken as zero. The ordinate is used both as the band gap energy and binding energy. Ev - Ec ( + ) is shown to see the trend of changes of the band gap. Eyes are guided by the broken line: (i) the band gap derived from the optical absorption measurement at room temperature [2]; (ii) calculated results by CPA at OK [6].

composition x increases, and then slowly shifts in the composition region above x = 0.18. This trend of the shifts is qualitatively in agreement with the theoretical calculation [1]. It predicts that EF, which is the P impurity level for the present alloys, linearly increases as a function of x and the change is a few hundred meV [11]. Therefore, it is considered that the shift of Ev corresponds to the shift of EF and the discontinuous change of E c is not due to the shift of EF, but the intrinsic change of the conduction band. As shown in Fig. 3(b), the Ec - Ev ( + ) curve clearly indicates the

The authors would like to thank Dr. Akinori Tanaka for reading the manuscript and contributing valuable comments.

References [1] M.-Z. Huang, W.Y. Ching, Superlattices and Microstructures 1 (1985) 137. [2] S. Krishnamurthy, A. Sher, Phys. Rev. B33 (1986) 1026. [3] Z.-Z. Xu, Solid State Commun. 91 (1994) 563. [4] M. Fewhat, A. Zaoui, B. Khelifa, H. Aourag, Solid State Commun. 91 (1994) 407. [5] M. Laradji, D.P. Landau, Phys. Rev. B51 (1995) 4894. [6] R. Braunstein, A.R. Moore, F. Herman, Phys. Rev. 109 (1958) 695. [7] T. Honda, M. Suezawa, K. Sumino, Jpn. J. Appl. Phys. 35 (1996) 5980. [8] V.V. Nemoshkalenko, V.G. Aleshin, Yu.N. Kucherenko, Solid State Commun. 20 (1976) 1155. [9] W.B. Jackson, S.-J. Oh, C.C. Tsai, J.W. Allen, Phys. Rev. Lett. 53 (1984) 1481. [10] J. Chelikowsky, D.J. Chadi, M . L Cohen, Phys. Rev. B8 (/973) 2786. [11] K.E. Newman, J.D. Dow, Phys. Rev. B30 (1984) 1929.