Amorphous germanium layers prepared by UV-photo-induced chemical vapour deposition

Amorphous germanium layers prepared by UV-photo-induced chemical vapour deposition

surface ELSEVIER sciel3ce Applied Surface Science 106 (1996) 75-79 Amorphous germanium layers prepared by UV-photo-induced chemical vapour depositi...

353KB Sizes 6 Downloads 17 Views

surface ELSEVIER

sciel3ce

Applied Surface Science 106 (1996) 75-79

Amorphous germanium layers prepared by UV-photo-induced chemical vapour deposition S. Chiussi, P. Gonzfilez, J. Serra, B. Le6n *, M. P~rez-Amor Dpto. Ffsica Aplicada, Unizersidade de Vigo, Lagoas Marcosende 9, 36280 Vigo, Spain Received 17 September 1995; accepted 31 January 1996

Abstract

Hydrogenated amorphous germanium (a-Ge:H) films have been deposited via laser induced chemical vapour deposition (LCVD) by irradiating germane/helium mixtures with an ArF excimer laser beam passing parallel above Si(100) wafers, metal plates and glass substrates. The analysis of the film thicknesses by profilometry and the characterisation of the material properties by scratch testing, FFIR spectroscopy, Raman spectroscopy, and hydrogen effusion measurements, showed significant dependencies of the growth rate and of the material properties on various processing parameters such as laser power, total pressure, and substrate temperature. Very homogeneous, adherent, and low impurity a-Ge:H films with thicknesses from 10 to 100 nm have been obtained at laser power around 1 W, total pressure of 40 Torr and substrate temperature of 250°C. Higher laser power, total pressure and especially the enhancement of the substrate temperature above the threshold for the thermally induced deposition process lead to higher growth rates but also to the formation of less adherent films and powdery deposits with microcrystalline components.

1. Introduction

Over the last decade, photo-induced chemical vapour deposition (photo-CVD) has proved to be a suitable technique for the deposition of a large variety of thin films with well defined thicknesses and material properties. The possibility of obtaining high quality thin films, with dielectric, semiconducting or corrosion resistant properties at low substrate temperatures [1-4], can offer great benefits especially in the preparation of complex multilayer structures with high interface quality due to the minimisation of thermally-induced interdiffusion [5].

* Corresponding author. Tel.: + 34-86-812216; fax: + 34-86812201.

One of the most efficient photo-CVD techniques for depositing amorphous thin films is the laser induced chemical vapour deposition (LCVD) in the parallel configuration, which causes a decomposition of the precursor gases near the substrate surface without irradiating it, and therefore avoiding additional substrate heating and thus a modification of the growing film and the underlying substrate [6]. This technique shows a considerable enhancement of the deposition rates, as compared to techniques using other photon sources, especially when excimer lasers are used, which provide selectable highly energetic photons and intense power densities capable of photolytically decomposing a wide range of precursor gases [6]. In this work, the 193 nm radiation of ArF excimer lasers has been used to obtain thin amorphous hydro-

0169-4332/96/$15.00 Copyright © 1996 Elsevier Science B.V. All rights reserved. PII S0 1 6 9 - 4 3 3 2 ( 9 6 ) 0 0 4 4 1 -2

76

S. Chiussi et al. /Applied Surface Science 106 (1996) 75 79

genated germanium (a-Ge:H) layers, which can find use not only as a low bandgap material in superlattice structures for microelectronic devices, such as amorphous thin film solar cells [7-9], but are especially useful as overlayers for subsequent preparation of epitaxial Si~Ge I x alloys by pulsed laser induced epitaxy (PLIE) [7-13]. The deposition process presented in this work has been particularly optimised to combine with the PLIE technique [12,13], answering to the necessity of producing very homogeneous low impurity a-Ge:H films. The incorporation of the LCVD technique into an integrated L C V D / P L I E process should constitute a technologically interesting alternative for producing a-Ge:H films and finally also epitaxial layers on temperature sensitive substrates or superlattices.

2. Experimental The deposition of a-Ge:H films was carried out in a self made HV stainless steel chamber connected to a turbomolecular pump system (Balzers TPH 060) for reaching base pressures below 2 × 10 - 6 mbar and to a rotary pump (Balzers Duo 016 B), which allowed high gas flows (up to 1100 sccm) during the deposition process. Different substrates, such as Si(100) wafers, metal plates and glass (Coming 7059) were placed on top of a heated stainless steel substrate holder. The laser beam of 193 nm emitted by an ArF excimer laser (Questek 2000 or Lambda Physik LPX 220i) was passed parallel above the substrate at a distance of 1 mm. The substrate temperature was measured by a thermocouple and controlled by an electronic device (Omron E6CS). As shown in Fig. 1, the reaction gas germane 4.6 (GeH 4) (Praxair) and the buffer gas helium 6.0 (He) (Praxair) entered the chamber through a nozzle, which was located perpendicular to the laser beam. During the deposition process, the flow rates of the processing gases (GeH 4 and He), and the purging gas (He), used to avoid deposition on the entrance and exit windows (Suprasil) for the laser beam, were kept constant by mass flow controllers (Bronkhorst HiTech). Total pressure was controlled by an electronic butterfly valve (MKS 252) connected to a Baratron (MKS 122 B), that measured the absolute pressure in the reaction chamber. Residues of toxic

MassFlow Germane Controllers

ReactionGases

MassFlow Controllers Silane

~

Argon

Helium

Ar

Exhaust

aser eam

Exhaust ~ , , ~ )~k BaratronI~- ~ "*~Thermocouple Ro~a~ T'~o

Butterfly

| [

O~en Ro~ary

Fig. 1. Experimental setup for the ArF LCVD of a-Ge:H films.

germane in the exhaust gases were decomposed at 800°C in a self-made decomposition oven. The characterlsation of the film thickness was carried out by profilometry (Sloan Dektak 3), the film adherence measured by scratch testing (MST CSEMEX), and the film structure analysed by FT Raman spectroscopy (Bruker RFS 100), FTIR spectroscopy (Bomem MBI00), and hydrogen effusion measurements. The temperature dependence of the hydrogen effusion was determined by a self-made system consisting of a heated quartz tube connected to a turbomolecular pumping system (Balzers T P H I 7 0 ) and a mass spectrometer (Balzers Q M G / Q M E 064), which gave strong indications about the incorporation of hydrogen in the films. Additional characterisation of the films by elastic recoil detection (ERDA), Rutherford backscattering spectroscopy (RBS), X-ray photoelectron spectroscopy (XPS) and X-ray diffraction (XRD) were used to corroborate the systematically obtained results.

3. Results and discussion The growth of homogeneous adherent a-Ge:H layers by ArF LCVD was found to be strongly affected by several process parameters such as laser power, total pressure, and substrate temperature but independent from the nature of the substrate. The variation of the most significant process parameters showed the following results:

S. Chiussi et al. /Applied Surface Science 106 (1996) 75-79

3.1. Dependence on the laser power The variation of the incident laser power between 1 and 15 W, obtained by increasing the repetition rate of the 270 mJ laser pulses up to 60 Hz exhibited a linear enhancement of the growth rate with the laser power measured at the exit window of the reaction chamber. Therefore, by dividing the growth rate by the laser power, the laser contribution to the rate of deposition could be separated from the other process parameters influencing the deposition of the 10-100 nm thick a-Ge:H films. A further enhancement of the laser power up to 45 W leads to the formation of powdery deposit which can be attributed to homogeneous gas phase nucleation of the film forming species due to an excess of photolytic products in the reaction volume [4]. 3.2. Dependence on germane flow and total pressure The variation of the germane flow rate from 1 to 5 sccm does not affect the deposition rate significantly, whereas the total pressure was found to be one of the most critical process parameters for obtaining homogeneous and adherent amorphous films. As seen in Fig. 2, the variation of the germane flow leads to only insignificant fluctuations of the growth rate of the homogeneous amorphous layers whose values were comparable within the limits of

25--

2O

J

15

10

5

0 ~ 0

I 20

'

J ' J ~ I 40 60 80 Total Pressure [Torr]

'

I 100

Fig. 2. P r e s s u r e d e p e n d e n c e o f the a - G e : H g r o w t h rate at 1 s c c m ( * ) and 5 s c c m ( D ) G e H 4 o b t a i n e d at 2 5 0 ° C with max. 15 W incident laser p o w e r .

77

experimental error of around 20%, which is due to the limitation of the film thickness measurements for films thinner than 50 rim. On the other hand, an increase of the total pressure from 8-80 Torr produces a strong enhancement of the growth rate. On reaching an upper pressure threshold of ~ 60 Torr, gas turbulence started, leading to the growth of non-uniform films with variations of the film thickness up to 100%. On exceeding 80 Ton', in addition to the gas turbulences, that lead to inhomogeneous film growth, a powdery deposit was detected similar to the deposition occurring at high laser power. This can be accounted for the occurrence of a homogeneous gas phase nucleation of the film forming species due to a pressure dependent enhancement of the collision rate in the reaction volume. Moreover, a low pressure threshold exists due to the need to purge the windows efficiently and whose value is specific for the experimental setup. In our system a 300 sccm He flow for each window was necessary to avoid any deposition on the windows of the reaction chamber. Lowering the total pressure below 20 Tort did not diminish the homogeneity and adherence of the films, but led to a deposition of powdery components on the windows and consequently perturbed the experimental conditions. 3.3. Dependence on the substrate temperature The increase of the substrate temperature from 50-450°C led first to a slight enhancement of the growth rate until reaching a temperature threshold of ~ 300°C. As shown in Fig. 3, above this threshold a strong enhancement of the film growth took place, which implies an alteration of the reaction kinetics, due to an additional pyrolitic contribution caused by thermal decomposition of GeH 4 known to start above 280°C [14]. This thermally driven deposition has not been observed at substrate temperatures < 250°C, so that at low temperatures the LCVD process is predominant. The clear contribution of thermal CVD at temperatures > 300°C led not only to extremely high growth rates, but also to a considerable change in the film structure, which could be observed by Raman spectroscopy. While at low temperatures, films showed the typical Raman shift around 270 cm-J for Ge-Ge bonds in an amorphous structure, an additional sharp peak around 300 cm -j was

S. Chiussi et al. /Applied Surface Science 106 (1996) 75-79

78

o4t

0.08 --

0.3

0.06

£

300 cm "1

270 cm "1

,~f~

-

g ~.,

0.2

~= 0.04

0.1

0.02

0.o 50

.

*

r

'

150

.

* I

250

'

r 350

'

! 450

0.00

I

'

500

Substrate Temperature [°C]

I

'

400

I

'

1

300

'

I

200

100

Wavenumbers [em-~] Fig. 3. Temperature dependence of the a-Ge:H growth rate at l sccm GeHa and 40 Tort total pressure.

found for the high temperature films, which can be attributed to crystalline germanium [15]. Characterisation of these films by XRD spectroscopy showed the typical diffraction peaks for microcrystalline germanium.

3.4. Optimised amorphous films and powdery, deposit The typical optimised experimental parameters for obtaining homogeneous low impurity a-Ge:H fihns on a 2.5 × 2.5 cm 2 area, which are suitable for a subsequent PLIE process, were found to be: 300 sccm He for purging each window, 50 sccm He as the buffer gas and 1 sccm GeH 4 as the reaction gas, 40 Torr as total pressure and 250°C as the substrate temperature. However, outside the optimum parameter range for obtaining homogeneous films, several threshold values were found. Powdery deposits were obtained when any one of the following threshold values were reached: incident laser power > 15 W, total pressure > 80 Torr and substrate temperature < 200°C. The Raman spectroscopy analysis of the powdery deposits (which clearly showed a lower adherence by scratch testing than the amorphous thin films) demonstrates, that beside the G e - G e peak typical for amorphous films ( ~ 270 c m - l ) , there exists the one for the crystalline material ( ~ 300 cm -~) (Fig. 4). The variation of the process parameters can therefore be used to tailor different degrees of microcrys-

Fig. 4. Raman spectra of a typical adherent amorphous film (b) obtained at optimum conditions and of a non-adherent powdery deposit (a) obtained at high total pressures.

tallinity in the thin germanium films, which has also been corroborated by XRD spectroscopy. The FTIR spectra for 100 nm thin films did show only a weak peak at 1870 cm -1 which can be attributed to GeH bonds typical for hydrogenated amorphous films [16], whereas ERDA analysis indicated a high hydrogen content up to 14 at% in the powdery deposit and 3 - 6 at% in the amorphous films. These discrepancies in the results, lay strong grounds for assuming that void formation occurs during the deposition, which incorporate hydrogen

2E-7 t H

(bulk)

& d~

1~7 -

5E-8 -

0

' 100

t ' I ' I 200 300 400 Effusion T ~ a t u r e

'

I 500 [°C]

'

I 600

Fig. 5. Hydrogen effusion spectra of a non-adherent film obtained at high total pressures.

S. Chiussi et al. / Applied Surface Science 106 (1996) 75-79

into the amorphous matrix with only a low amount of GeH bonds. This assumption has been corroborated by hydrogen effusion of the powdery deposit, showing additionally to the high temperature peak at ~ 450°C typical for adherent homogeneous material, a low temperature peak ~ 250°C effusion temperature, which can be attributed to hydrogen incorporated in voids (Fig. 5) [17]. The additional characterisation of the films by RBS and XPS analyses were made in order to check the impurity content of the obtained films. These results showed only a negligible carbon and oxygen contamination of the film surface [ 12,13].

4. Conclusions LCVD has been used to produce very homogeneous and low impurity a-Ge:H films on different substrates. The films obtained at optimum conditions show an amorphous structure and hydrogen content of 3 - 6 at%, whereas outside of this optimum parameter, a powdery deposit with microcrystalline components has been detected. The results show that the obtained films are suitable for a subsequent PLIE treatment and that a tailoring of microcrystallinity can be carried out.

Acknowledgements The authors thank J. Castro for his collaboration during the experimental work, F. Lusquifios for building up the hydrogen effusion system and measuring the samples, R. Larciprete for the XPS, S. Martelli for the XRD, M.G. Grimaldi for the RBS analyses, and N. Banerji for helpful discussions. The thickness measurements by profilometry were car-

79

ried out in the Central Analytical Facility (C.A.C.T.I.) of the University of Vigo. Financial support by the HCM EEC contract ERBCHRXCT 930355, XUGA 32107B92DOG211, and grant M E C - S B 9 3 A0742819D is gratefully acknowledged.

References [1] P. Gonzfilez, D. Fernandez, J. Pou, E. Garcla, J. Serra, B. Le6n, M. P~rez-Amor and T. SziSr~nyi, Appl. Phys. A 57 (1993) 181. [2] P. Gonzfilez, D. Fernandez, J. Pou, E. Garcfa, J. Serra, B. Le6n and M. P6rez-Amor, Thin Solid Films 218 (1992) 170. [3] J. Pou, P. Gonzalez, E. Garcfa, D. Fernandez, J. Serra, B. Le6n, S.R.J. Saunders and M. PErez-Amor, Appl. Surf. Sci. 79/80 (1994) 338. [4] T.R. Dietrich, S. Chiussi, H. Stafast and F.J. Comes, Appl. Phys. A 48 (1989) 405. [5] M. Konagai, H. Takei, W.Y. Kim and K. Takahashi, in: Proc. PVSEC-2, Peking, People's Republic of China, 1986, p. 437. [6] I.P. Herman, Chem. Rev. 89 (1989) 1323. [7] C.J. Kiely, V. Tavitian and J.G. Eden, J. Appl. Phys. 65 (1989) 3883. [8] D.H. Lowndes, D.B. Geohegan, D. Eres, S.J. Pennycook, D.N. Mashburn and G.E. Jellison, Jr., Appl. Surf. Sci. 36 (1989) 59. [9] R.H. Bube, Annu. Rev. Mater. Sci. 20 (1990) 19. [10] J.R. Abelson, T.W. Sigmon, K.B. Kim and K.H. Weiner, Appl. Phys. Lett. 52 (1988) 230. [11] A. Slaoui, C. Deng, S. Talwar, K.J. Kramer, T.W. Sigmon, J.P. Stoquert and B. Prevot, Appl. Surf. Sci. 86 (1995) 346. [12] R. Larciprete, P. Willmott, S. Martelli, C. Cesile, E. Borsella, S. Chiussi, P. Gonzalez and B. Le6n, Appl. Surf. Sci. 106 (1996) 179. [13] S. Chiussi, P. Gonzalez, B. Le6n, R. Larciprete, P. Willmott, S. Martelli, C. Cesile and E. Borsella, Appl. Surf. Sci. 102 (1996) 42. [14] L.H. Hall, J. Electrochem. Soc. 119 (1972) 1593. [15] P. Evrard, J.L. Stehle, C. Pickering and R.T. Carline, Thin Solid Films 222 (1992) 73. [16] T.P. Driisedau, A. Annen, B. Schr~Sder and H. Freistedt, Phil. Mag. B 69 (1994) 1. [17] M. Stutzmann, J.-B. Chevrier, C.P. Herrero and A. Breitschwerdt, Appl. Phys. A 53 (1991) 47.