Effect of hydrogen passivation on polycrystalline silicon thin films

Thin Solid Films 487 (2005) 152 – 156 www.elsevier.com/locate/tsf

Effect of hydrogen passivation on polycrystalline silicon thin films S. Hondaa,T, T. Matesa, M. Ledinskya, J. Oswalda, A. Fejfara, J. Kocˇkaa, T. Yamazakib, Y. Uraokab, T. Fuyukib b

a Institute of Physics, Academy of Sciences of the Czech Republic, Cukrovarnicka´ 10, 162 53 Prague 6, Czech Republic Graduate School of Materials Science, Nara Institute of Science and Technology, Takayama 8916-5, Ikoma, Nara 630-0192, Japan

Available online 2 March 2005

Abstract Hydrogen passivation is essential for improving the properties of polycrystalline silicon thin films. We have observed that remote plasma hydrogenation with duration up to 30 min effectively passivated the defects and improved the Hall mobility, trap density and photoluminescence intensity. Over 60 min of hydrogenation caused the photoluminescence intensity to decrease. It seems that excessive hydrogenation not only passivated defects but also created new defects (Si–H2 bonds and hydrogen molecules) in the grains. Raman spectroscopy detected that hydrogen formed Si–H2 bonds in the poly-Si up to 100 nm from surface. Creation of these defects corresponded to a decrease of the photoluminescence intensity. These defects might be harmful to poly-Si-based devices. D 2005 Elsevier B.V. All rights reserved. Keywords: Hydrogen passivation; Polycrystalline silicon; Photoluminescence; Raman spectroscopy; Si–H2 bonding; Hydrogen molecules

1. Introduction Polycrystalline silicon (poly-Si) is a material widely used in the semiconductor industry. It is particularly interesting, for example, for the photovoltaic industry as an attractive base material for solar cell applications. Performance of polycrystalline silicon thin film solar cells largely depends on transport properties dominated by grain boundaries. Hydrogen passivation can reduce the grain boundary defects and it is considered essential for improving the solar cell efficiency.

2. Experimental Thin film poly-Si grown by atmospheric pressure chemical vapour deposition (APCVD) was deposited with SiH2Cl2 and BCl3 as source and doping gases, respectively, directly on SiO2 film fabricated by wet oxidation process of P+ c-Si wafer with low resistivity ~0.002 Vcm. Holes with 80-Am diameter were opened by photolithography in the T Corresponding author. E-mail address: [email protected] (S. Honda). 0040-6090/$ - see front matter D 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.tsf.2005.01.056

SiO2 at interval of 450 Am in order to contact the backside electrode for getting map of local currents measurement. In order to obtain poly-Si films with high crystallinity and large grain size, we deposited poly-Si films using the twostep method [1]. Polycrystalline silicon fabricated by this method has a columnar structure with (220) orientation and average grain size of 2 Am on average. The thickness of the poly-Si layer was 12 Am. Hydrogen passivation of poly-Si thin films was performed with the remote plasma system in order to prevent surface damage by ions. The flowing gas was a mixture of H2: 5 sccm and Ar: 0.3 sccm, the pressure was set to 1.010 2 Pa, the substrate temperature to 300 8C and the RF-power to 200 W. Hydrogen passivation time varied from 5 min to 60 min. In order to investigate the hydrogen passivation effect atomic force microscopy (AFM) combined with the mapping of local conductivity was chosen [2]. The combined AFM system was located inside the UHV chamber (10 8 Pa) to prevent the oxidation of surface. The van der Pauw technique was used to measure the Hall effect and its temperature dependence. Photoluminescence measurements at 7 K were performed on sample immersed in liquid He. PL spectra were analysed with the Ge detector

S. Honda et al. / Thin Solid Films 487 (2005) 152–156

102

Hall Mobility [cm2/Vs]

and lock in amplifier. PL was excited by the Ar laser with the wavelength of 488 nm and the power density on the sample 5 mW/cm2. Raman spectroscopy was used to observe local vibration modes (LVM) of Si–H2 bonds in a stretching mode near the 2100 cm 1 line and hydrogen molecular (H2) vibrations near the 4160 cm 1 lines. The Raman scattering spectra were taken in the backscattering geometry using excitation by the Ar laser at 514 nm.

153

101

3.1. Map of local currents Fig. 1 shows the map of local currents of both asdeposited poly-Si and hydrogenated poly-Si (60 min). The as-deposited poly-Si showed contrasted map of local currents due to different conductivities of various grains isolated by grain boundaries. On the other hand, the map of local currents of hydrogenated poly-Si did not show the contrast comparable to that of as-deposited poly-Si. This finding agrees with another observation of the same behaviour by different method [3]. To our knowledge, this is the first time the changes in local currents due to hydrogen passivation are reported.

103

102

101 1013

Nt [cm-2]

3. Results and discussion

Activation Energy [meV]

100

1012

1011

3.2. Hall mobility

0

10

20

30

40

50

60

70

80

Hydrogenation Time [min]

Results of the carrier mobility, its activation energy and the trap state density measured and estimated by Hall effect are shown in Fig. 2 [4]. It can be seen that after only 5-min hydrogen passivation, the electronic properties of poly-Si were dramatically improved compared to that of as-deposited poly-Si. Improvement of poly-Si properties was saturated for hydrogenation times over 30 min. Hydrogen effectively passivated the defects at grain boundaries and suppressed the band bending at grain boundaries, which act as barriers for majority carriers and recombination sites for minority carriers. Properties of our poly-Si were dominated by grain boundaries and the

Fig. 2. Effect of hydrogenation on electronic properties of poly-Si observed by Hall measurements.

decrease of the trap density at grain boundaries by hydrogenation led to the suppression of the band bending and the improvement of carrier mobility. However, many researchers reported that hydrogenation could influence Si in two ways: passivates the defects by hydrogen bonding and also generates new defects [5]. Such a surface damage region must be very thin compared to the thickness of our poly-Si thin films (12 Am) because it cannot be seen by Hall measurement that averages over the whole poly-Si thickness. 3.3. Photoluminescence

3000nm

3000nm

Fig. 1. Map of local currents measured by AFM with conductive cantilever on as-deposited (left) and on hydrogenated (right) poly-Si samples.

Photoluminescence spectra (see Fig. 3) revealed a peak around 0.98 eV attributed to band tail-to-tail luminescence [6]. No sharp excitonic transition around 1.1 eV as in c-Si was observed. The grain boundaries give rise to localized states in the band gap such as Si dangling bonds and strained Si–Si bonds, the latter of which are widely believed to be the origin of exponential band tails [3]. PL spectra of poly-Si show a broad peak at an energy somewhat smaller than the energy of the optical gap. In LPCVD grown poly-Si the

154

S. Honda et al. / Thin Solid Films 487 (2005) 152–156

5.2 as deposited

Luminescence intensity

15 min

5.0

tail to tail

-1

520 cm

FWHM [cm-1]

30 min 60 min

O1 '

4.8 4.6 4.4 4.2 c-Si

4.0 0 0.80

0.85

0.90

0.95

1.00

1.05

1.10

Energy (eV) Fig. 3. Photoluminescence spectra of poly-Si as a function of hydrogen passivation time.

maximum of the PL peak occurred at 0.9 eV at a temperature of 4.3 K and has been assigned to radiative recombination of electrons and holes trapped in band tail states [3]. PL spectra of our poly-Si reveal a PL band centred at 0.98 eV. It is conceivable that the larger value of PL peak energy in poly-Si by APCVD and laser crystallized poly-Si compared to that of poly-Si by LPCVD is due to steeper band tails [7]. Fig. 3 shows the dependence of PL spectra on hydrogenation time. With an increase of hydrogenation time, PL intensity increased because of the defect passivation and resulting suppression of non-radiative recombination. However, after 60-min hydrogenation, the PL intensity decreased again, possibly due to surface damage caused by hydrogenation. In this paper, we discuss only the main peak around 0.98 eV. O1 peaks are related to oxygen. These peaks are not discussed in this paper because correlation between oxygen and remote plasma hydrogen passivation is still unknown. 3.4. Raman spectroscopy We employed Raman scattering spectroscopy in order to characterize deterioration of properties caused by a long hydrogen passivation. Fig. 4 shows that FWHM of the Raman peak (TO-LO 520 cm 1) increased from 4.4 cm 1 to 4.9 cm 1 with an increase of hydrogen passivation time up to 60 min (FWHM of the peak of c-Si is 4.0 cm 1 as a reference) [5]. The increase of the FWHM indicates creation of defects or disorder in poly-Si thin films after hydrogen passivation. We have also the observed local vibration mode (LVM) of Si–H2 bonds (2100 cm 1 band), attributed to atomic vacancies in the grains of poly-Si after hydrogen passivation [8]. This peak also indicated the degradation of carrier mobility in crystallized poly-Si by excimer-laser annealing [8]. Another band at 2000 cm 1 (Si–H bonds) related to

10

20

30

40

50

60

70

80

Hydrogen passivation time (min) Fig. 4. FWHM of the Raman peak (TO-LO 520 cm 1) as a function of hydrogen passivation time.

passivated dangling bonds at grain boundaries was not present for our poly-Si as it depends on the grain size [5]. The 2100 cm 1 band started to appear in poly-Si with hydrogen passivation over 60 min in Fig. 5. In Fig. 5 Si– H2 peak was slightly shifted in our poly-Si (from 2100 cm 1 to 2125 cm 1). Our other results showed that this peak is affected by grain size: with an increase of the grain size (from 2 Am to 15 Am), the peak position gets close to 2100 cm 1. This finding confirmed that hydrogen not only

Si-H Si-H2

Si-H3

60 min

(Raman intensity (a. u.))

0.75

30 min

15 min

5 min

as-deposited

1700 1800 1900 2000 2100 2200 2300 2400

Wavenumber (cm-1) Fig. 5. Raman scattering spectra of poly-Si as a function of hydrogen passivation time.

passivated defects but also caused new defects with long time hydrogen passivation, are in good agreement with the results of PL study. Hydrogen formed not only Si–H2 bonds but also hydrogen molecules (H2) inside the thin films. Fig. 6 shows peak corresponding to H2 around 4160 cm 1 in another part of the Raman spectra [9]. Two kinds of H2 have been observed in Si, i.e., H2 in plateles (4158 cm 1) and in Td interstitial sites (3601 cm 1) [10]. In our poly-Si, hydrogen formed only as H2 in plateles and it also started to appear only at 60-min hydrogenation. We performed chemical etching of the surface layer with sequential rinses in HNO3 followed by HF. Fig. 7 shows normalised Raman intensity of Si–H2 and H2 peak as a function of etching distance from surface. Si–H2 bonds were formed not only at the surface because Si–H2 peak was still apparent after etching away 50 nm from the surface. Si–H2 and H2 peak disappeared after etching different thickness from the poly-Si surface and so the Si– H2 bonds formation extended deeper into the layer than the H2 formation. We checked that c-Si wafer subjected to hydrogen passivation at the same conditions showed the same behaviour as poly-Si. It is important to note that SiH Raman signal always accompanies the H2 signal. Hydro-

H2 molecules peak

(Raman intensity (a. u.))

60 min

Normalised Raman Intensity

S. Honda et al. / Thin Solid Films 487 (2005) 152–156

155

1.0

-1

2100 cm

-1

4160 cm

0.8 0.6 0.4 0.2 0.0 0

20

40

60

80

100

120

etching distance from surface (nm) Fig. 7. Depth profiles of the integrated in intensity of 2100 cm 4160 cm 1.

1

and

gen-related platelets act as sinks to which H2 diffusion. [10]. Once formed these Si–H2 bonds become effective trap for any further hydrogen (H2) being introduced into the material during hydrogen passivation. Because of this behaviour, Si–H2 showed the difference depth profile compared to H2. FWHM of TO-LO 520 cm 1 also got close to FWHM of as-deposited poly-Si, after etching away 100 nm. These kinds of defects were formed in very thin region compared to the whole thickness (thickness of our polySi is 12 Am), so they did not affect the electrical properties measured by Hall effect, but caused the degradation of the PL intensity. We can conclude that there is a correlation between the decrease of PL and Si– H2 and H2 formation in poly-Si.

4. Summary

30 min

Hydrogenation passivated the defects at grain boundaries and improved the electronic properties of APCVD grown poly-Si this films. PL intensity indicated that hydrogen passivated the defects and suppressed the non-radiative recombination. Excessive hydrogen passivation (over 60 min) caused a decrease of PL intensity. This degradation is explained by the results that the hydrogenation led also to the creation of subsurface defects ( Si–H2 bonds and H2 ) up to depth of 100 nm from the surface. These defects must be considered as harmful to any poly-Si-based devices due to decrease of PL.

15 min

5 min

as-deposited

Acknowledgment 3900

4000

4100

4200

4300

Wavenumber (cm-1) Fig. 6. Another part of the Raman scattering spectra of poly-Si as a function of hydrogen passivation time, showing the presence of hydrogen molecules after longest hydrogenation times.

The authors would like to thank V. Vorlicek, I. Gregora, J. Pokorny´ for help with Raman spectroscopy measurement. This research was supported by AVOZ 10100521, VaV/300/ 01/03, VaV SN/172/05, GAAV IAA1010316, IAA1010413 and GA CˇR 202/05/H003 projects.

156

S. Honda et al. / Thin Solid Films 487 (2005) 152–156

References [1] Y. Ishikawa, K. Nishioka, Y. Yamamoto, Y. Uraoka, T. Fuyuki, Proc. 29th IEEE Photovoltaic Spec. Conf., IEEE, New York, p. 1294. [2] B. Rezek, J. Stuchlik, A. Fejfar, J. Kocˇka, Appl. Phys. Lett. 74 (1999) 1475. [3] A.U. Savchouk, S. Ostapenko, G. Nowak, J. Lagowski, L. Jastrzebski, Appl. Phys. Lett. 67 (1995) 82. [4] J.Y.W. Seto, J. Appl. Phys. 46 (1975) 5247. [5] K. Kitahara, K. Ohnishi, Y. Kato, R. Yamaguchi, T. Kurosawa, Jpn. J. Appl. Phys. 11 (2003) L6742.

[6] R.A. Street, Hydrogenated Amorphous Silicon, Cambridge University Press, 1991. [7] K. Brendel, N.H. Nickel, K. Lips, W. Fuhs, Solid State Commun. 126 (2003) 499. [8] K. Kitahara, S. Murakami, A. Hara, K. Nakajima, Appl. Phys. Lett. 72 (1998) 2438. [9] N. Fukata, S. Sakai, K. Murakami, K. Ishioka, K. Nakamura, M. Kitajima, S. Fujimura, J. Kikuchi, H. Haneda, Jpn. J. Appl. Phys. 36 (1997) L1456. [10] A.W.R. Leitch, J. Weber, V. Alex, Mater. Sci. Eng, B 58 (1999) 6.