- Email: [email protected]
Structural and optical studies on hot wire chemical vapour deposited hydrogenated silicon films at low substrate temperature
ARTICLE IN PRESS Solar Energy Materials & Solar Cells 93 (2009) 199–205
Contents lists available at ScienceDirect
Solar Energy Materials & Solar Cells journal homepage: www.elsevier.com/locate/solmat
Structural and optical studies on hot wire chemical vapour deposited hydrogenated silicon films at low substrate temperature Purabi Gogoi, Pratima Agarwal Department of Physics, Indian Institute of Technology Guwahati, Guwahati 781 039, Assam, India
a r t i c l e in fo
abstract
Article history: Received 7 December 2007 Received in revised form 18 September 2008 Accepted 24 September 2008 Available online 21 November 2008
Thin films of hydrogenated silicon are deposited by hot wire chemical vapour deposition technique, as an alternative of plasma enhanced chemical vapour deposition technique. By varying the hydrogen and silane flow rate, we deposited the films ranging from pure amorphous to nanocrystallite-embedded amorphous in nature. In this paper we report extensively studied structural and optical properties of these films. It is observed that the rms bond angle deviation decreases with increase in hydrogen flow rate, which is an indication of improved order in the films. We discuss this under the light of breaking of weak Si–Si bonds and subsequent formation of strong Si–Si bonds and coverage of the growing surface by atomic hydrogen. & 2008 Elsevier B.V. All rights reserved.
Keywords: Amorphous silicon Nanocrystalline silicon HWCVD Hydrogen dilution
1. Introduction It is well established that hydrogenated amorphous silicon (a-Si:H) and hydrogenated microcrystalline silicon (mc-Si:H) are very suitable materials for low cost and stable solar cell and other applications [1]. The extensively studied technique for the deposition of device quality a-Si:H and mc-Si:H is the plasma enhanced chemical vapour deposition (PECVD). This technique is widely used for industrial applications with RF frequency 13.56 MHz to deposit a-Si:H films with very good optoelectronic properties. However, low deposition rate, especially for mc-Si:H, offers hurdles to the cost-effective fabrication of the PECVDdeposited devices [2,3]. Another important drawback of this technique is that the bombardment by energetic ions on the growing surface creates lots of defects, which is undesirable [3]. As an alternative of plasma assisted deposition a new technology namely hot wire chemical vapour deposition (HWCVD) or catalytic CVD (Cat-CVD) is emerging to prepare device quality films [4–6]. In this method the precursor gases are decomposed with the help of hot filament by the process of catalytic cracking reaction, thus avoiding the disadvantages related to plasma assisted processes. The simplicity of design [7] is another added advantage. As this is a new technique and not as extensively studied as PECVD, there is a need of extensive research on the deposition of device quality a-Si:H, mc-Si:H, nc-Si:H, etc. by this process. Corresponding author. Tel.: +91 361 2582702; fax: +91 361 2582749.
E-mail address: [email protected] (P. Agarwal). 0927-0248/$ - see front matter & 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.solmat.2008.09.058
In this paper, we report preparation and structural studies of HWCVD silicon films deposited at low substrate temperature by varying the hydrogen and silane (SiH4) flow rate. Both X-ray diffraction (XRD) and Raman scattering studies show that the deposited films range from amorphous to nanocrystalliteembedded amorphous in nature. Scanning electron microscopy (SEM) studies on the films also reveal that the surface of these films range from smooth to that having grains. Fourier transform infrared (FTIR) spectroscopic analysis shows that the films contain 6–13% of atomic hydrogen. Further analysis shows that hydrogen is present in the network mostly in monohydride configuration, which is an important requirement for stable device applications.
2. Experimental details The films were prepared at low substrate temperature (substrate temperature 200 1C) in a load lock based HWCVD chamber by varying the hydrogen and silane flow rates (HFRS and SFRs) keeping other parameters fixed (filament temperature 1700 1C, filament to substrate distance 6 cm). The chamber is first evacuated to base pressure 3 106 Torr before the precursor gases are passed. Prior to the dissociation of the gases, the chamber pressure is set at about 55 mTorr, which decreased to about 47 mTorr when the dissociation took place except for samples #1 and 9 where the chamber pressure after dissociation became 30 and 25 mTorr, respectively. The chamber pressure decreases because of the dissociation of the gases into various
ARTICLE IN PRESS 200
P. Gogoi, P. Agarwal / Solar Energy Materials & Solar Cells 93 (2009) 199–205
Table 1 Deposition conditions, thickness and deposition rate of hydrogenated silicon films used in the present study Sample #/Series #
Sample Sample Sample Sample Sample Sample Sample Sample Sample
#1 #2 #3 #4 #5 #6 #7 #8 #9
9 > = > ;I
9 > =
9 > > > > > =
> ;III
> > > > > ;
II
Process pressure (mTorr)
SiH4 flow rate (SCCM)
H2 flow rate (SCCM)
Thickness (nm)
Deposition rate (A˚ s1)
30 47 45 50 47 49 46 45 25
4 4 4 3 2 2 2 2 2
0 10 15 15 15 10 8 5 0
483 478 302 242 174 202 325 480 315
3.22 1.77 1.20 0.89 0.64 0.75 1.20 1.78 1.75
The films are prepared by varying either the hydrogen or silane flow rate keeping substrate temperature 200 1C and filament temperature 1700 1C. Samples #1–3 form Series I (SFR 4 SCCM), samples #5–9 form Series II (SFR 2 SCCM) and samples #3–5 form Series III (HFR 15 SCCM). Prior to the dissociation of the gases the chamber pressure is set at about 55 mTorr, which decreased after the dissociation took place.
3. Results and discussions The thickness of the films as calculated from UV–Vis–NIR transmission data is found to be in the range of 175–480 nm. The pffiffiffiffiffiffiffiffi ffi band gap (EG) of the films, calculated by plotting ahn vs. hu (Fig. 1), is estimated to be in the range of 1.7–1.9 eV. There is a general trend of increase in the band gap with the increase in HFR in the two series of samples with constant SFR. In Series I, EG increases from 1.78 to 1.81 eV when HFR is increased from 0 to 15 SCCM, while in Series II, it increases from 1.7 to 1.9 eV. The change in EG with increases in HFR is more for low SFR. However, with the variation of SFR no specific trend in the band gap variation is
800
y = 846.23x-1482.46
700
sqrt (αhν) (cm-1 eV)1/2
600
Sample # 4
500 400 300 200 900 800 700
Sample # 5
y = 792.21x-1501.73
600 500 400 300 2.0
2.2
2.4
2.6
2.8
3.0
3.2
hν (eV) pffiffiffiffiffiffiffiffiffi Fig. 1. Plot of ahn vs. hu for two samples to show the variation of absorption near the band gap. The intercept of the fitted line (in red colour) in the energy axis gives the band gap of the films.
3.2
SFR2SCCM SFR3SCCM SFR4SCCM
2.8 Deposition Rate (Å/Sec)
radicals having different lifetimes. The whole set of samples were deposited on corning 7059 glass, ITO-coated glass as well as c-Si wafer substrates for carrying out different types of studies. The SFRs and HFRs are varied such that samples #1–3 form one series (Series I) and samples #5–9 form another series (Series II) with varying HFR at two different SFRs 4 and 2 SCCM, respectively; while samples #3–5 form another series (Series III) with varying SFR at constant HFR of 15 SCCM. The detailed deposition conditions are listed in Table 1. For the characterization of the films we used various tools like XRD, SEM, Raman spectroscopy, FTIR absorption spectroscopy, photoluminescence (PL) and UV–Vis–NIR transmission spectroscopy. XRD studies were done on the samples deposited on Corning 7059 using Siefert XRD 3003 TT X-ray diffractometer in 2y mode at grazing angle of incidence 2–41. Raman scattering experiments were done on the films deposited on both Corning 7059 and ITO-coated glass using Olympus BX41 Raman spectrometer with excitation wavelength of 514.532 nm in scan range 400–1000 cm1. SEM studies were performed using Leo 1430 VP scanning electron microscope on the films deposited on Corning 7059 glass. FTIR spectroscopic measurements were done in transmission mode in the wave number range 450–4000 cm1 on films deposited on c-Si substrates with the help of Perkin Elmer Spectrum I and used for the calculation of bonded hydrogen content as well as the percentage of hydrogen in the monohydride mode. UV–Vis–NIR transmission measurements in the 400–1100 nm range were used to estimate the band gap, thickness and optical constants, following Swanepoel [8]. PL measurements are done at room temperature using Aminco Bowman Series 2 Luminescence Spectrometer with excitation wavelengths 300 nm (4.13 eV), 350 nm (3.54 eV) and 375 nm (3.30 eV), while the luminescence is recorded in the range 500–900 nm (2.48–1.38 eV).
2.4 2.0 1.6 1.2 0.8 0.4 -2
0
2
4
6
8
10
12
14
16
Hydrogen Flow Rate (SCCM) Fig. 2. Deposition rate vs. hydrogen flow rate (HFR) for the films prepared by varying HFR and silane flow rate (SFR). The encircled points correspond to Series III, with SFR increasing in the direction of the arrow.
ARTICLE IN PRESS P. Gogoi, P. Agarwal / Solar Energy Materials & Solar Cells 93 (2009) 199–205
observed. It is observed that the deposition rate decreases with the increase in HFR, while it increases with increase in SFR (Fig. 2), which is quite expected. For Series I, the deposition rate decreases from 3.22 to 1.2 A˚ s1, as HFR is increased from 0 to 15 SCCM. For
HFR = 2 SCCM
intensity (a.u.)
Sample # 5
Sample # 6
Sample #8 Sample # 7 Sample #9
30
20
40
50
60
2θ (deg)
Intensity (a.u.)
Sample # 4
Sample # 3
Sample # 1
20
30
40 2θ (Deg)
50
60
Fig. 3. X-ray Diffraction pattern for the hydrogenated silicon films prepared with SFR 2SCCM (a). The films with SFR 3 and 4 SCCM do not show any crystalline peak (b).
201
Series II, it decreases from 1.78 to 0.64 A˚ s1 as the HFR is increased from 5 SCCM to 15 SCCM. In this series the films with HFR 0 SCCM has deposition rate 1.75 A˚ s1, little lower than that of HFR 5 SCCM and could be because of much lower chamber pressure during deposition (25 mTorr). It is also observed that in both Series I and II, decrease of deposition rate with increase of SFR is steeper up to HFR 10 SCCM, after which the deposition rate decreases gradually with HFR. The decrease of deposition rate with increase in HFR is expected as the atomic hydrogen etches out the growing surface and also the presence of more hydrogen in the gas mixture dilutes the film forming radicals. These results are summarized in Table 1. Fig. 3 shows the XRD pattern for different films. The nature of the XRD pattern varies from pure amorphous to nanocrystallite-embedded amorphous as the hydrogen dilution increases. Series I samples show pure amorphous nature. In the case of Series II, samples #5, 7 and 8 show weak intensity peaks at 2y value near 281, 471 and 561 corresponding to (111), (2 2 0) and (3 11) planes of crystalline silicon superimposed on the broad amorphous pattern, respectively. Sample #9 shows amorphous nature and sample #6 shows very wide humps centered near 281 and 471. The other sample in Series III, i.e., sample #4 also shows the wide humps centered near 281 and 471 corresponding to (111) and (2 2 0) plane of crystalline silicon. With the increase of HFR, the crystallinity and hence intensity of the crystalline peak should increase. For Series II samples it is observed that the intensity of the peak increases with the increase of HFR up to 8 SCCM, but when HFR is further increased to 10 SCCM, the XRD pattern does not show any sharp peak. Again when HFR is increased to 15 SCCM, we see weak crystalline peaks, but the intensity is lower than for the film with HFR 8 SCCM. This observed anomaly may be due to the low thickness of the films with HFR 10 and 15 SCCM. As the HFR is increased, the deposition rate decreases, resulting in low thickness. The crystallinty develops in the films gradually from amorphous phase along the direction of growth. So, as we increase the HFR, both the development of crystallinity due to dilution and the low thickness due to low deposition rate become the deciding factors for the resulting crystallinity. The crystallite sizes for those films are calculated using Scherrer formula and listed in Table 2. SEM images performed on the films show that some films have smooth surface morphology while others show the presence of grains (Figs. 4a and b). Series I films show smooth surface morphology. Incidentally these samples also show pure amorphous nature in XRD and Raman studies. In the case of Series II, the SEM images show variation from smooth surface morphology to the one with the presence of dispersed as well as packed grains. However, no systematic variation of surface morphology with the HFR is observed. As the SFR is decreased from 4 to 2 SCCM for fixed HFR 15 SCCM (Series III), the SEM images reveal the formation of grains.
Table 2 Estimated results of band gap, rms bond angle deviation, crystallinity fraction, crystallite size, hydrogen content and percentage of hydrogen bonded in monohydride mode Sample #
EG (eV)
Dy (deg)
Xc
dRaman (nm)
Sample Sample Sample Sample Sample Sample Sample Sample Sample
1.78 1.80 1.81 1.75 1.9 1.89 1.887 1.72 1.7
7.99 7.86 7.80 8.08 7.96 8.00 8.10 6.94 7.32
0 0 0 0.06 0.43 0.36 0.45 0.42 0.18
0 0 0 3.97 6.50 6.36 3.38 3.22 3.11
#1 #2 #3 #4 #5 #6 #7 #8 #9
dXRD (nm)
Hydrogen conc. (%)
SiH/(total H) (%)
12.0
11.5 11.8 12.8 11.6 7.7 7.3 8.7 6.2 7.5
90.91 98.29 96.39 91.02 71.65 83.33 71.59 84.75 92.10
10.0 9.3
ARTICLE IN PRESS 202
P. Gogoi, P. Agarwal / Solar Energy Materials & Solar Cells 93 (2009) 199–205
Sample # 5
Intensity (a.u.)
Sample # 6
Sample # 7
Sample # 8
Sample # 9 200
300
400
500
600
700
800
Raman Shift (cm-1)
Intensity (a.u.)
Sample # 4
Sample # 3
Sample # 2
Sample # 1
Fig. 4. SEM images for two samples deposited on Corning 7059: (a) (sample #1) smooth surface morphology and (b) (sample #8) the presence of grains in the film.
200
400
600
800
Raman Shift (cm-1)
Intensity (a.u.)
Sample # 5
Fig. 6. Raman scattering spectrum for the films under present study; for the films with SFR 2 SCCM (a) and with SFR 3 and 4 SCCM (b) for varying HFR. The deconvolution is done as described in the text.
Sample # 7
Sample # 9
1.4
1.6
1.8
2.0 Energy (eV)
2.2
2.4
Fig. 5. The figure shows the photoluminescence spectrum for a few samples deposited with SFR 2SCCM using excitation wavelength 350 nm.
Fig. 5 shows the PL emission spectrum of samples prepared with SFR 2 SCCM. We observe weak intensity PL signal in the visible regime. The PL peaks are observed at 1.7–1.8 and
2.0–2.1 eV. The peak around 1.7–1.78 eV is the most intense and comprises sub-bands. The FTIR spectrum does not show the presence of oxygen in the films. So, the possibility of the PL peak originating due to the presence of oxygen is ruled out. These peaks in the visible range are signatures of the formation of nanocrystallites. We do not observe any significant change in the PL emission spectra with the variation of excitation energy. The small intensity sharp peaks at 1.5 and 1.6 eV in the spectra are due to the instrument response. Raman scattering studies also reveal these films to be amorphous to nanocrystalline-embedded amorphous in nature (Figs. 6a and b). The crystallinity fractions differ a bit depending on the substrate; however, they show similar type of dependence on the gas flow rates. The results on the films deposited on ITOcoated substrates are discussed here. The Raman spectra for Series I show pure amorphous nature with peak occurring at 480 cm1 corresponding to transverse optic (TO) mode of amorphous silicon. The spectrum for all the samples of Series II can be best deconvoluted into three Gaussian components. The first component occurs at 480 cm1, while the third component occurs in the
ARTICLE IN PRESS P. Gogoi, P. Agarwal / Solar Energy Materials & Solar Cells 93 (2009) 199–205
Do
where Do is the shift of the nanocrystalline Raman peak corresponding to that of c-Si and B ¼ 2.0 cm1 nm2 [13]. The values of dRaman are found to be in the range 4–6.5 nm. It is observed that the crystallite size increases with the HFR (Table 2). The larger number of atomic hydrogen in the gas mixture helps in the formation of bigger crystallites in the following ways: the bonded hydrogen helps in the full surface coverage and thus facilitates the surface diffusion of the film forming precursors like SiH3, SiH2, SiH and Si. Hence the adsorbed radicals find energetically favorable sites, leading to improvement in order and also the formation and growth of crystallites [14]. Also, the atomic hydrogen acts as etchant and breaks the weak Si–Si bonds on the growing film surface and thus helps in the creation of strong Si–Si bonds [15]. The atomic hydrogen not only acts on the top surface layer but can also penetrate into few layers below the film surface where it helps in the formation of ordered structure by compensation of dangling bonds, breaking of weak bonds as well as reconstruction of strong bonds and the strain relaxation [15]. Thus higher HFR results in the bigger crystallites. There is always a distribution of crystallite sizes in the films. The crystallite size calculated from Raman data corresponds to the average size, while in XRD data the bigger crystallites dominate. Hence the crystallite size calculated from XRD data is bigger than that of Raman data. AS HFR is increased from 0 SCCM to 8 SCCM, the crystalline volume fraction (Xc) increases from 0.18 to 0.45. But for sample #6 corresponding to HFR 10 SCCM, Xc decreased to 0.36, which again increases to 0.43 when the HFR is further increased to 15 SCCM. The crystalline fraction should increase with HFR due to the improvement in order. However, for sample #6 the deposition rate becomes very low, giving less thickness (200 nm), and therefore the crystalline fraction for these samples are smaller than sample #7. In case of sample #5 the crystalline fraction again starts increasing due to improvement in order, which has thickness comparable to sample #6. From the FWHM data of the amorphous peak we have calculated the rms bond angle deviation DyB [16], which tells about the short range order in the films; the lower the DyB, the higher the order. The values of DyB for samples under present studies are found to be 6.94–8.11, which are very good for nc-Si:H and a-Si:H [10,17]. Fig. 7 depicts the variation of DyB with HFR and SFR. We have observed that for Series I, DyB decreases from 7.991 to 7.801 with the increase of HFR from 0 to 15 SCCM, which further indicates that hydrogen facilitates the
8.2 8.0 7.8 Δθ (Deg)
range from 511.83 to 518.13 cm1, which is associated with the nanocrystalline phase. The shift in the peak position from 520 cm1 corresponding to c-Si is because of the size limitation of the crystallites. The intermediate peak ranges from 490.56 to 507.77 cm1 and is assigned for bond dilation at grain boundaries by Veprek et al. [9]. The other sample in Series III, i.e., sample #4, can be best deconvoluted in to two Gaussian peaks, one at 480 cm1 and the other near 515 cm1. It is also observed that as the HFR increases, size of the nanocrystallites also increases along with the shift of the intermediate peak towards the crystalline side, which could be attributed to the improvement of medium range order [10,11]. The Raman crystallinity fraction (Xc) is calculated using the formula Xc ¼ (Ic+Iint)/(Ic+Iam+Iint) [12], where Ic is the integrated intensity component due to the crystalline phase, Iam is due to the amorphous phase and Iint is due to the intermediate peak. Sample #4 contains very less crystalline fraction (0.06) and this could be the reason for the absence of intermediate peak. Series II films have Xc ranging 0.18–0.45. The crystallite sizes are calculated by using the following formula: rffiffiffiffiffiffiffiffi B dRaman ¼ 2p
203
7.6 7.4 7.2 SFR2SCCM SFR3SCCM SFR4SCCM
7.0 6.8 -2
0
2
4 6 8 10 12 hydrogen Flow Rate (SCCM)
14
16
Fig. 7. The figure shows the variation of rms bond angle deviation with HFR. The encircled points correspond to Series III with SFR increasing in the direction of arrow.
improvement of order in the films. For Series II also, similar dependence of DyB with the HFR is observed except for samples #8 and 9. For these two samples DyB is considerably lower (6.94 and 7.32, respectively). However, if we look into the variation of DyB with SFR, i.e., Series III, no systematic variation is observed. Sample #3, which is pure amorphous, has lowest DyB (7.81), while sample #4 (Xc ¼ 0.06, beginning of nanocrystallite formation) has the highest DyB (8.081). Sample #5 having Xc ¼ 0.195 has DyB ¼ 7.961. Though the DyB is not systematically varying (Fig. 6), the values are quite low and correspond to that of device quality films. The results are summarized in Table 2. The results of XRD and Raman studies show good agreement. For Series I films, we observe pure amorphous nature in both XRD and Raman studies. For Series II samples, the most intense peak is observed in sample #7, for which Raman study also gives the highest crystallinity fraction (Xc ¼ 0.45). The intensity in XRD peaks then decreases in samples #8 and 5, for which the Xc from Raman measurements also decrease to 0.42 and 0.43. Xc for the other two samples (samples #6 and 9) further decreases to 0.36 and 0.18, where we do not observe any peak corresponding to nanocrystalline silicon in the XRD studies. In Series III, it is observed that as the SFR is decreased from 4 SCCM to 2 SCCM, the XRD pattern changes from amorphous (sample #3) to nanocrystallite-embedded amorphous (sample #5), with sample #4 having broad humps. There is a systematic change from pure amorphous to nanocrystallite-embedded amorphous in the Raman scattering spectra also. The IR absorption spectra are used for the estimation of bonded hydrogen content and also the percentage of hydrogen in monohydride configuration. For the calculation of hydrogen concentration CH absorption around 630 cm1 corresponding to the wagging mode of vibration is used in the following formula [18]: Z aðoÞ NH ¼ AW do oW
o
where NH is the total bonded hydrogen and AW ¼ 1.6 1019 cm2 [19] is a proportionality constant. For the calculation of the percentage of hydrogen in monohydride mode (CSi–H) the absorption band around 2000–2100 cm1 is deconvoluted into two peaks, one at 2000 cm1 and the other near 2100 cm1. CSi–H is calculated as the ratio between the integrated intensity around
ARTICLE IN PRESS 204
P. Gogoi, P. Agarwal / Solar Energy Materials & Solar Cells 93 (2009) 199–205
4000 sample # 3 Absorption Intensity (a.u.)
3500 3000 sample # 4 2500 2000 1500
sample # 5
1000 500 0 500
550
600 650 700 Wave No. (cm-1)
750
800
1500
films in the series. For sample #7, Xc is also the highest with the lowest amount of hydrogen in monohydride mode. For samples in Series III, CH increases with increase in SFR. The percentage of hydrogen in the monohydride mode (CSi–H) is quite high for most of the samples (Table 2). This is very much appreciated for stable device applications as the hydrogen present in polyhydride configuration is believed to be responsible for the light induced degradation of Si:H films. It is also observed that with the increase of the crystallite size the fraction of hydrogen bonded in monohydride mode decreases. We have observed that the shape of absorption band near 630 cm1 as well as 2000 cm1 changes with the development of crystallinity. As the films become more crystalline, the shape of the absorption bands starts comprising sub-bands. In the case of pure a-Si:H films the rocking/wagging mode occurs at 630 cm1 for all the configurations of silicon hydrogen bonds and the stretching mode occurs at 2000 and 2100 cm1, depending on whether hydrogen is bonded in monohydride or polyhydride configuration. However, as the crystallinity starts evolving, the wagging absorption band near 630 cm1 starts consisting of subbands towards high energy, and the band deviates from the symmetric shape [7,22]. We have observed the development of features in the absorption band with the increase of crystallinity irrespective of SFR or HFR.
Absorption Intensity (a.u.)
sample # 3 4. Conclusions
sample # 4
1000
In this paper we report on the preparation as well as structural and optical studies of hydrogenated silicon films deposited by the HWCVD method. We have deposited a series of hydrogenated amorphous silicon by varying the hydrogen and silane flow rates. Raman and XRD studies show that the films range from pure amorphous to nanocrystallite-embedded amorphous in nature. The presence of PL bands in the visible region further confirms the presence of nanocrystallites. We also thank Central Instrument Facility (CIF), IIT Guwahati for SEM studies.
500 sample # 5 0 1850
1900
1950
2000 2050 2100 Wave No. (cm-1)
2150
2200
Acknowledgments
Fig. 8. FTIR spectrum for a few samples deposited with HFR 15 SCCM. (a) Absorption band for Series III samples at 630 cm1 corresponding to wagging mode and (b) the stretching mode absorption. The spectra for samples #4 and 5 are vertically shifted for clarity.
We thank Prof. R.O. Dusane and Prof. D.S. Misra of IIT Bombay for allowing us to deposit films by the HWCVD method and use Raman set up, respectively, and Dr. Alka Kumbhar for her help during deposition.
2000 cm1 and the total integrated intensity in the range 2000–2100 cm1. Figs. 8a and b show the absorption bands for a few samples at 630 and 2000 cm1, respectively. The hydrogen concentration (CH) varies from 6.2 to 12.8 at%. The high CH compared to that of other HWCVD-deposited films [20] may be because of the low substrate temperature. Mahan et al. [21] have reported HWCVD films with high hydrogen content prepared at low substrate temperatures (17.8 and 12 at% for substrate temperature 180 and 230 1C). It is further to be noted that CH for the films deposited with SFR 2 SCCM are lower (6.2–8.7) than with 3 and 4 SCCM (11.5–12.8), which show mostly amorphous nature in Raman measurements. Series I (corresponding to SFR 4SCCM) shows monotonic increase in CH from 11.5% to 12.8%, when the HFR is increased from 0 to 15 SCCM. With the increase in HFR, the percentage of atomic hydrogen as well as band gap of the films increases, which may be associated with the relaxation of the constrained network and improvement in order. This is observed not only in HWCVD films but also in PECVD films [21]. For Series II (SFR 2 SCCM) the hydrogen content is almost same, except for sample #7 where CH (8.7%) is little higher than other
References [1] A. Matsuda, Microcrystalline silicon. Growth and device application, J. NonCryst. Solids 338–340 (2004) 1–12. [2] J. Doyle, R. Robertson, G.H. Lin, M.Z. He, A. Gallagher, Production of highquality amorphous silicon films by evaporative surface decomposition, J. Appl. Phys. 64 (1988) 3215–3223. [3] S.R. Jadkar, J.V. Sali, M.G. Takwale, D.V. Musale, S.T. Kshirsagar, The role of hydrogen dilution of silane and phosphorous doping on hydrogenated microcrystalline silicon (mc-Si:H) films prepared by hot wire chemical vapor deposition (HW-CVD) technique, Thin Solid Films 395 (2001) 206–212. [4] H. Matsumura, Summary of research in NEDO Cat-CVD project in Japan, Thin Solid Films 395 (2001) 1–11. [5] A.H. Mahan, Status of Cat- CVD (Hot Wire CVD) research in United States, Thin Solid Films 395 (2001) 12–16. [6] Y. Wang, X. Geng, H. Stiebig, F. Finger, Stability of microcrystalline silicon solar cells with HWCVD buffer layer, Thin Solid Films 516 (2008) 733–735. [7] S. Halindintwali, et al., Improved stability of intrinsic nanocrystalline Si thin films deposited by hot-wire chemical vapor deposition technique, Thin Solid Film 515 (2007) 8040–8044. [8] R. Swanepoel, Determination of the thickness and optical constants of amorphous silicon, J. Phys. E: Sci. Instrum. 16 (1983) 1214–1222. [9] S. Veprek, F.A. Sarott, Z. Iqbal, Effect of grain boundaries on the Raman spectra, optical absorption and elastic light scattering in nanometer-sized crystalline silicon, Phys. Rev. B 36 (6) (1987) 3344–3350.
ARTICLE IN PRESS P. Gogoi, P. Agarwal / Solar Energy Materials & Solar Cells 93 (2009) 199–205
[10] M. Ito, M. Kondo, Systematic study of photo degradation of tailored nanostructured Si solar cells by controlling their medium range order, Jpn. J. Appl. Phys. 45 (8) (2006) 230–232. [11] P. Gogoi, P.N. Dixit, P. Agarwal, Amorphous silicon films with high deposition rate prepared using argon and hydrogen diluted silane for stable solar cells, Sol. Energy Mater. Sol. Cells 91 (2007) 1253–1257. [12] M. Jana, D. Das, A.K. Barua, Role of hydrogen in controlling the growth of mc-Si:H films from argon diluted SiH4 plasma, J. Appl. Phys. 91 (2002) 5442–5448. [13] Y. He, C.Z. Yin, G. Cheng, L. Wang, X. Liu, G.Y. Hu, The structure and properties of nano size silicon films, J. Appl. Phys. 75 (1994) 797–803. [14] A. Matsuda, Growth mechanism of microcrystalline silicon obtained from reactive plasmas, Thin Solid Films 337 (1999) 1–6. [15] C.C. Tsai, G.B. Anderson, R. Thompson, B. Wacker, Control of silicon network structure in plasma deposition, J. Non Cryst. Solids 114 (1989) 151–153. [16] D. Beeman, R. Tsu, M.F. Thorpe, Structural information from the Raman spectrum of amorphous silicon, Phys. Rev. B 32 (2) (1985) 874–878.
205
[17] M. Stutzman, in: S. Mahajan (Ed.), Handbook of Semiconductors, vol. 3a, North-Holland, Amsterdam, 1994, pp. 657–781. [18] R.A. Street, Hydrogenated Amorphous Silicon, Cambridge University Press, Cambridge, 1992 (Chapter 2). [19] C.J. Fang, K.J. Gruntz, L. Ley, M. Cardona, F.J. Demond, G. Muller, S. Kalbitzer, The hydrogen content of a-Ge:H and a-Si:H as determined by IR spectroscopy, gas evolution and nuclear reaction techniques, J. Non Cryst. Solids 35–36 (1980) 255–260. [20] A.H. Mahan, J. Carapella, B.P. Nelson, R.S. Crandall, I. Balberg, Deposition of device quality, low H content amorphous silicon, J. Appl. Phys. 69 (1991) 6728–6730. [21] A.H. Mahan, R. Biswas, L.M. Gedvilas, D.L. Williamson, B.C. Pan, On the influence of short and medium range order on the material band gap in hydrogenated amorphous silicon, J. Appl. Phys. 96 (2004) 3818–3826. [22] S. Zhang, et al., Silicon thin films prepared in the transition region and their use in solar cells, Sol. Energy Mater. Sol. Cells 90 (2006) 3001–3008.
Contents lists available at ScienceDirect
Solar Energy Materials & Solar Cells journal homepage: www.elsevier.com/locate/solmat
Structural and optical studies on hot wire chemical vapour deposited hydrogenated silicon films at low substrate temperature Purabi Gogoi, Pratima Agarwal Department of Physics, Indian Institute of Technology Guwahati, Guwahati 781 039, Assam, India
a r t i c l e in fo
abstract
Article history: Received 7 December 2007 Received in revised form 18 September 2008 Accepted 24 September 2008 Available online 21 November 2008
Thin films of hydrogenated silicon are deposited by hot wire chemical vapour deposition technique, as an alternative of plasma enhanced chemical vapour deposition technique. By varying the hydrogen and silane flow rate, we deposited the films ranging from pure amorphous to nanocrystallite-embedded amorphous in nature. In this paper we report extensively studied structural and optical properties of these films. It is observed that the rms bond angle deviation decreases with increase in hydrogen flow rate, which is an indication of improved order in the films. We discuss this under the light of breaking of weak Si–Si bonds and subsequent formation of strong Si–Si bonds and coverage of the growing surface by atomic hydrogen. & 2008 Elsevier B.V. All rights reserved.
Keywords: Amorphous silicon Nanocrystalline silicon HWCVD Hydrogen dilution
1. Introduction It is well established that hydrogenated amorphous silicon (a-Si:H) and hydrogenated microcrystalline silicon (mc-Si:H) are very suitable materials for low cost and stable solar cell and other applications [1]. The extensively studied technique for the deposition of device quality a-Si:H and mc-Si:H is the plasma enhanced chemical vapour deposition (PECVD). This technique is widely used for industrial applications with RF frequency 13.56 MHz to deposit a-Si:H films with very good optoelectronic properties. However, low deposition rate, especially for mc-Si:H, offers hurdles to the cost-effective fabrication of the PECVDdeposited devices [2,3]. Another important drawback of this technique is that the bombardment by energetic ions on the growing surface creates lots of defects, which is undesirable [3]. As an alternative of plasma assisted deposition a new technology namely hot wire chemical vapour deposition (HWCVD) or catalytic CVD (Cat-CVD) is emerging to prepare device quality films [4–6]. In this method the precursor gases are decomposed with the help of hot filament by the process of catalytic cracking reaction, thus avoiding the disadvantages related to plasma assisted processes. The simplicity of design [7] is another added advantage. As this is a new technique and not as extensively studied as PECVD, there is a need of extensive research on the deposition of device quality a-Si:H, mc-Si:H, nc-Si:H, etc. by this process. Corresponding author. Tel.: +91 361 2582702; fax: +91 361 2582749.
E-mail address: [email protected] (P. Agarwal). 0927-0248/$ - see front matter & 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.solmat.2008.09.058
In this paper, we report preparation and structural studies of HWCVD silicon films deposited at low substrate temperature by varying the hydrogen and silane (SiH4) flow rate. Both X-ray diffraction (XRD) and Raman scattering studies show that the deposited films range from amorphous to nanocrystalliteembedded amorphous in nature. Scanning electron microscopy (SEM) studies on the films also reveal that the surface of these films range from smooth to that having grains. Fourier transform infrared (FTIR) spectroscopic analysis shows that the films contain 6–13% of atomic hydrogen. Further analysis shows that hydrogen is present in the network mostly in monohydride configuration, which is an important requirement for stable device applications.
2. Experimental details The films were prepared at low substrate temperature (substrate temperature 200 1C) in a load lock based HWCVD chamber by varying the hydrogen and silane flow rates (HFRS and SFRs) keeping other parameters fixed (filament temperature 1700 1C, filament to substrate distance 6 cm). The chamber is first evacuated to base pressure 3 106 Torr before the precursor gases are passed. Prior to the dissociation of the gases, the chamber pressure is set at about 55 mTorr, which decreased to about 47 mTorr when the dissociation took place except for samples #1 and 9 where the chamber pressure after dissociation became 30 and 25 mTorr, respectively. The chamber pressure decreases because of the dissociation of the gases into various
ARTICLE IN PRESS 200
P. Gogoi, P. Agarwal / Solar Energy Materials & Solar Cells 93 (2009) 199–205
Table 1 Deposition conditions, thickness and deposition rate of hydrogenated silicon films used in the present study Sample #/Series #
Sample Sample Sample Sample Sample Sample Sample Sample Sample
#1 #2 #3 #4 #5 #6 #7 #8 #9
9 > = > ;I
9 > =
9 > > > > > =
> ;III
> > > > > ;
II
Process pressure (mTorr)
SiH4 flow rate (SCCM)
H2 flow rate (SCCM)
Thickness (nm)
Deposition rate (A˚ s1)
30 47 45 50 47 49 46 45 25
4 4 4 3 2 2 2 2 2
0 10 15 15 15 10 8 5 0
483 478 302 242 174 202 325 480 315
3.22 1.77 1.20 0.89 0.64 0.75 1.20 1.78 1.75
The films are prepared by varying either the hydrogen or silane flow rate keeping substrate temperature 200 1C and filament temperature 1700 1C. Samples #1–3 form Series I (SFR 4 SCCM), samples #5–9 form Series II (SFR 2 SCCM) and samples #3–5 form Series III (HFR 15 SCCM). Prior to the dissociation of the gases the chamber pressure is set at about 55 mTorr, which decreased after the dissociation took place.
3. Results and discussions The thickness of the films as calculated from UV–Vis–NIR transmission data is found to be in the range of 175–480 nm. The pffiffiffiffiffiffiffiffi ffi band gap (EG) of the films, calculated by plotting ahn vs. hu (Fig. 1), is estimated to be in the range of 1.7–1.9 eV. There is a general trend of increase in the band gap with the increase in HFR in the two series of samples with constant SFR. In Series I, EG increases from 1.78 to 1.81 eV when HFR is increased from 0 to 15 SCCM, while in Series II, it increases from 1.7 to 1.9 eV. The change in EG with increases in HFR is more for low SFR. However, with the variation of SFR no specific trend in the band gap variation is
800
y = 846.23x-1482.46
700
sqrt (αhν) (cm-1 eV)1/2
600
Sample # 4
500 400 300 200 900 800 700
Sample # 5
y = 792.21x-1501.73
600 500 400 300 2.0
2.2
2.4
2.6
2.8
3.0
3.2
hν (eV) pffiffiffiffiffiffiffiffiffi Fig. 1. Plot of ahn vs. hu for two samples to show the variation of absorption near the band gap. The intercept of the fitted line (in red colour) in the energy axis gives the band gap of the films.
3.2
SFR2SCCM SFR3SCCM SFR4SCCM
2.8 Deposition Rate (Å/Sec)
radicals having different lifetimes. The whole set of samples were deposited on corning 7059 glass, ITO-coated glass as well as c-Si wafer substrates for carrying out different types of studies. The SFRs and HFRs are varied such that samples #1–3 form one series (Series I) and samples #5–9 form another series (Series II) with varying HFR at two different SFRs 4 and 2 SCCM, respectively; while samples #3–5 form another series (Series III) with varying SFR at constant HFR of 15 SCCM. The detailed deposition conditions are listed in Table 1. For the characterization of the films we used various tools like XRD, SEM, Raman spectroscopy, FTIR absorption spectroscopy, photoluminescence (PL) and UV–Vis–NIR transmission spectroscopy. XRD studies were done on the samples deposited on Corning 7059 using Siefert XRD 3003 TT X-ray diffractometer in 2y mode at grazing angle of incidence 2–41. Raman scattering experiments were done on the films deposited on both Corning 7059 and ITO-coated glass using Olympus BX41 Raman spectrometer with excitation wavelength of 514.532 nm in scan range 400–1000 cm1. SEM studies were performed using Leo 1430 VP scanning electron microscope on the films deposited on Corning 7059 glass. FTIR spectroscopic measurements were done in transmission mode in the wave number range 450–4000 cm1 on films deposited on c-Si substrates with the help of Perkin Elmer Spectrum I and used for the calculation of bonded hydrogen content as well as the percentage of hydrogen in the monohydride mode. UV–Vis–NIR transmission measurements in the 400–1100 nm range were used to estimate the band gap, thickness and optical constants, following Swanepoel [8]. PL measurements are done at room temperature using Aminco Bowman Series 2 Luminescence Spectrometer with excitation wavelengths 300 nm (4.13 eV), 350 nm (3.54 eV) and 375 nm (3.30 eV), while the luminescence is recorded in the range 500–900 nm (2.48–1.38 eV).
2.4 2.0 1.6 1.2 0.8 0.4 -2
0
2
4
6
8
10
12
14
16
Hydrogen Flow Rate (SCCM) Fig. 2. Deposition rate vs. hydrogen flow rate (HFR) for the films prepared by varying HFR and silane flow rate (SFR). The encircled points correspond to Series III, with SFR increasing in the direction of the arrow.
ARTICLE IN PRESS P. Gogoi, P. Agarwal / Solar Energy Materials & Solar Cells 93 (2009) 199–205
observed. It is observed that the deposition rate decreases with the increase in HFR, while it increases with increase in SFR (Fig. 2), which is quite expected. For Series I, the deposition rate decreases from 3.22 to 1.2 A˚ s1, as HFR is increased from 0 to 15 SCCM. For
HFR = 2 SCCM
intensity (a.u.)
Sample # 5
Sample # 6
Sample #8 Sample # 7 Sample #9
30
20
40
50
60
2θ (deg)
Intensity (a.u.)
Sample # 4
Sample # 3
Sample # 1
20
30
40 2θ (Deg)
50
60
Fig. 3. X-ray Diffraction pattern for the hydrogenated silicon films prepared with SFR 2SCCM (a). The films with SFR 3 and 4 SCCM do not show any crystalline peak (b).
201
Series II, it decreases from 1.78 to 0.64 A˚ s1 as the HFR is increased from 5 SCCM to 15 SCCM. In this series the films with HFR 0 SCCM has deposition rate 1.75 A˚ s1, little lower than that of HFR 5 SCCM and could be because of much lower chamber pressure during deposition (25 mTorr). It is also observed that in both Series I and II, decrease of deposition rate with increase of SFR is steeper up to HFR 10 SCCM, after which the deposition rate decreases gradually with HFR. The decrease of deposition rate with increase in HFR is expected as the atomic hydrogen etches out the growing surface and also the presence of more hydrogen in the gas mixture dilutes the film forming radicals. These results are summarized in Table 1. Fig. 3 shows the XRD pattern for different films. The nature of the XRD pattern varies from pure amorphous to nanocrystallite-embedded amorphous as the hydrogen dilution increases. Series I samples show pure amorphous nature. In the case of Series II, samples #5, 7 and 8 show weak intensity peaks at 2y value near 281, 471 and 561 corresponding to (111), (2 2 0) and (3 11) planes of crystalline silicon superimposed on the broad amorphous pattern, respectively. Sample #9 shows amorphous nature and sample #6 shows very wide humps centered near 281 and 471. The other sample in Series III, i.e., sample #4 also shows the wide humps centered near 281 and 471 corresponding to (111) and (2 2 0) plane of crystalline silicon. With the increase of HFR, the crystallinity and hence intensity of the crystalline peak should increase. For Series II samples it is observed that the intensity of the peak increases with the increase of HFR up to 8 SCCM, but when HFR is further increased to 10 SCCM, the XRD pattern does not show any sharp peak. Again when HFR is increased to 15 SCCM, we see weak crystalline peaks, but the intensity is lower than for the film with HFR 8 SCCM. This observed anomaly may be due to the low thickness of the films with HFR 10 and 15 SCCM. As the HFR is increased, the deposition rate decreases, resulting in low thickness. The crystallinty develops in the films gradually from amorphous phase along the direction of growth. So, as we increase the HFR, both the development of crystallinity due to dilution and the low thickness due to low deposition rate become the deciding factors for the resulting crystallinity. The crystallite sizes for those films are calculated using Scherrer formula and listed in Table 2. SEM images performed on the films show that some films have smooth surface morphology while others show the presence of grains (Figs. 4a and b). Series I films show smooth surface morphology. Incidentally these samples also show pure amorphous nature in XRD and Raman studies. In the case of Series II, the SEM images show variation from smooth surface morphology to the one with the presence of dispersed as well as packed grains. However, no systematic variation of surface morphology with the HFR is observed. As the SFR is decreased from 4 to 2 SCCM for fixed HFR 15 SCCM (Series III), the SEM images reveal the formation of grains.
Table 2 Estimated results of band gap, rms bond angle deviation, crystallinity fraction, crystallite size, hydrogen content and percentage of hydrogen bonded in monohydride mode Sample #
EG (eV)
Dy (deg)
Xc
dRaman (nm)
Sample Sample Sample Sample Sample Sample Sample Sample Sample
1.78 1.80 1.81 1.75 1.9 1.89 1.887 1.72 1.7
7.99 7.86 7.80 8.08 7.96 8.00 8.10 6.94 7.32
0 0 0 0.06 0.43 0.36 0.45 0.42 0.18
0 0 0 3.97 6.50 6.36 3.38 3.22 3.11
#1 #2 #3 #4 #5 #6 #7 #8 #9
dXRD (nm)
Hydrogen conc. (%)
SiH/(total H) (%)
12.0
11.5 11.8 12.8 11.6 7.7 7.3 8.7 6.2 7.5
90.91 98.29 96.39 91.02 71.65 83.33 71.59 84.75 92.10
10.0 9.3
ARTICLE IN PRESS 202
P. Gogoi, P. Agarwal / Solar Energy Materials & Solar Cells 93 (2009) 199–205
Sample # 5
Intensity (a.u.)
Sample # 6
Sample # 7
Sample # 8
Sample # 9 200
300
400
500
600
700
800
Raman Shift (cm-1)
Intensity (a.u.)
Sample # 4
Sample # 3
Sample # 2
Sample # 1
Fig. 4. SEM images for two samples deposited on Corning 7059: (a) (sample #1) smooth surface morphology and (b) (sample #8) the presence of grains in the film.
200
400
600
800
Raman Shift (cm-1)
Intensity (a.u.)
Sample # 5
Fig. 6. Raman scattering spectrum for the films under present study; for the films with SFR 2 SCCM (a) and with SFR 3 and 4 SCCM (b) for varying HFR. The deconvolution is done as described in the text.
Sample # 7
Sample # 9
1.4
1.6
1.8
2.0 Energy (eV)
2.2
2.4
Fig. 5. The figure shows the photoluminescence spectrum for a few samples deposited with SFR 2SCCM using excitation wavelength 350 nm.
Fig. 5 shows the PL emission spectrum of samples prepared with SFR 2 SCCM. We observe weak intensity PL signal in the visible regime. The PL peaks are observed at 1.7–1.8 and
2.0–2.1 eV. The peak around 1.7–1.78 eV is the most intense and comprises sub-bands. The FTIR spectrum does not show the presence of oxygen in the films. So, the possibility of the PL peak originating due to the presence of oxygen is ruled out. These peaks in the visible range are signatures of the formation of nanocrystallites. We do not observe any significant change in the PL emission spectra with the variation of excitation energy. The small intensity sharp peaks at 1.5 and 1.6 eV in the spectra are due to the instrument response. Raman scattering studies also reveal these films to be amorphous to nanocrystalline-embedded amorphous in nature (Figs. 6a and b). The crystallinity fractions differ a bit depending on the substrate; however, they show similar type of dependence on the gas flow rates. The results on the films deposited on ITOcoated substrates are discussed here. The Raman spectra for Series I show pure amorphous nature with peak occurring at 480 cm1 corresponding to transverse optic (TO) mode of amorphous silicon. The spectrum for all the samples of Series II can be best deconvoluted into three Gaussian components. The first component occurs at 480 cm1, while the third component occurs in the
ARTICLE IN PRESS P. Gogoi, P. Agarwal / Solar Energy Materials & Solar Cells 93 (2009) 199–205
Do
where Do is the shift of the nanocrystalline Raman peak corresponding to that of c-Si and B ¼ 2.0 cm1 nm2 [13]. The values of dRaman are found to be in the range 4–6.5 nm. It is observed that the crystallite size increases with the HFR (Table 2). The larger number of atomic hydrogen in the gas mixture helps in the formation of bigger crystallites in the following ways: the bonded hydrogen helps in the full surface coverage and thus facilitates the surface diffusion of the film forming precursors like SiH3, SiH2, SiH and Si. Hence the adsorbed radicals find energetically favorable sites, leading to improvement in order and also the formation and growth of crystallites [14]. Also, the atomic hydrogen acts as etchant and breaks the weak Si–Si bonds on the growing film surface and thus helps in the creation of strong Si–Si bonds [15]. The atomic hydrogen not only acts on the top surface layer but can also penetrate into few layers below the film surface where it helps in the formation of ordered structure by compensation of dangling bonds, breaking of weak bonds as well as reconstruction of strong bonds and the strain relaxation [15]. Thus higher HFR results in the bigger crystallites. There is always a distribution of crystallite sizes in the films. The crystallite size calculated from Raman data corresponds to the average size, while in XRD data the bigger crystallites dominate. Hence the crystallite size calculated from XRD data is bigger than that of Raman data. AS HFR is increased from 0 SCCM to 8 SCCM, the crystalline volume fraction (Xc) increases from 0.18 to 0.45. But for sample #6 corresponding to HFR 10 SCCM, Xc decreased to 0.36, which again increases to 0.43 when the HFR is further increased to 15 SCCM. The crystalline fraction should increase with HFR due to the improvement in order. However, for sample #6 the deposition rate becomes very low, giving less thickness (200 nm), and therefore the crystalline fraction for these samples are smaller than sample #7. In case of sample #5 the crystalline fraction again starts increasing due to improvement in order, which has thickness comparable to sample #6. From the FWHM data of the amorphous peak we have calculated the rms bond angle deviation DyB [16], which tells about the short range order in the films; the lower the DyB, the higher the order. The values of DyB for samples under present studies are found to be 6.94–8.11, which are very good for nc-Si:H and a-Si:H [10,17]. Fig. 7 depicts the variation of DyB with HFR and SFR. We have observed that for Series I, DyB decreases from 7.991 to 7.801 with the increase of HFR from 0 to 15 SCCM, which further indicates that hydrogen facilitates the
8.2 8.0 7.8 Δθ (Deg)
range from 511.83 to 518.13 cm1, which is associated with the nanocrystalline phase. The shift in the peak position from 520 cm1 corresponding to c-Si is because of the size limitation of the crystallites. The intermediate peak ranges from 490.56 to 507.77 cm1 and is assigned for bond dilation at grain boundaries by Veprek et al. [9]. The other sample in Series III, i.e., sample #4, can be best deconvoluted in to two Gaussian peaks, one at 480 cm1 and the other near 515 cm1. It is also observed that as the HFR increases, size of the nanocrystallites also increases along with the shift of the intermediate peak towards the crystalline side, which could be attributed to the improvement of medium range order [10,11]. The Raman crystallinity fraction (Xc) is calculated using the formula Xc ¼ (Ic+Iint)/(Ic+Iam+Iint) [12], where Ic is the integrated intensity component due to the crystalline phase, Iam is due to the amorphous phase and Iint is due to the intermediate peak. Sample #4 contains very less crystalline fraction (0.06) and this could be the reason for the absence of intermediate peak. Series II films have Xc ranging 0.18–0.45. The crystallite sizes are calculated by using the following formula: rffiffiffiffiffiffiffiffi B dRaman ¼ 2p
203
7.6 7.4 7.2 SFR2SCCM SFR3SCCM SFR4SCCM
7.0 6.8 -2
0
2
4 6 8 10 12 hydrogen Flow Rate (SCCM)
14
16
Fig. 7. The figure shows the variation of rms bond angle deviation with HFR. The encircled points correspond to Series III with SFR increasing in the direction of arrow.
improvement of order in the films. For Series II also, similar dependence of DyB with the HFR is observed except for samples #8 and 9. For these two samples DyB is considerably lower (6.94 and 7.32, respectively). However, if we look into the variation of DyB with SFR, i.e., Series III, no systematic variation is observed. Sample #3, which is pure amorphous, has lowest DyB (7.81), while sample #4 (Xc ¼ 0.06, beginning of nanocrystallite formation) has the highest DyB (8.081). Sample #5 having Xc ¼ 0.195 has DyB ¼ 7.961. Though the DyB is not systematically varying (Fig. 6), the values are quite low and correspond to that of device quality films. The results are summarized in Table 2. The results of XRD and Raman studies show good agreement. For Series I films, we observe pure amorphous nature in both XRD and Raman studies. For Series II samples, the most intense peak is observed in sample #7, for which Raman study also gives the highest crystallinity fraction (Xc ¼ 0.45). The intensity in XRD peaks then decreases in samples #8 and 5, for which the Xc from Raman measurements also decrease to 0.42 and 0.43. Xc for the other two samples (samples #6 and 9) further decreases to 0.36 and 0.18, where we do not observe any peak corresponding to nanocrystalline silicon in the XRD studies. In Series III, it is observed that as the SFR is decreased from 4 SCCM to 2 SCCM, the XRD pattern changes from amorphous (sample #3) to nanocrystallite-embedded amorphous (sample #5), with sample #4 having broad humps. There is a systematic change from pure amorphous to nanocrystallite-embedded amorphous in the Raman scattering spectra also. The IR absorption spectra are used for the estimation of bonded hydrogen content and also the percentage of hydrogen in monohydride configuration. For the calculation of hydrogen concentration CH absorption around 630 cm1 corresponding to the wagging mode of vibration is used in the following formula [18]: Z aðoÞ NH ¼ AW do oW
o
where NH is the total bonded hydrogen and AW ¼ 1.6 1019 cm2 [19] is a proportionality constant. For the calculation of the percentage of hydrogen in monohydride mode (CSi–H) the absorption band around 2000–2100 cm1 is deconvoluted into two peaks, one at 2000 cm1 and the other near 2100 cm1. CSi–H is calculated as the ratio between the integrated intensity around
ARTICLE IN PRESS 204
P. Gogoi, P. Agarwal / Solar Energy Materials & Solar Cells 93 (2009) 199–205
4000 sample # 3 Absorption Intensity (a.u.)
3500 3000 sample # 4 2500 2000 1500
sample # 5
1000 500 0 500
550
600 650 700 Wave No. (cm-1)
750
800
1500
films in the series. For sample #7, Xc is also the highest with the lowest amount of hydrogen in monohydride mode. For samples in Series III, CH increases with increase in SFR. The percentage of hydrogen in the monohydride mode (CSi–H) is quite high for most of the samples (Table 2). This is very much appreciated for stable device applications as the hydrogen present in polyhydride configuration is believed to be responsible for the light induced degradation of Si:H films. It is also observed that with the increase of the crystallite size the fraction of hydrogen bonded in monohydride mode decreases. We have observed that the shape of absorption band near 630 cm1 as well as 2000 cm1 changes with the development of crystallinity. As the films become more crystalline, the shape of the absorption bands starts comprising sub-bands. In the case of pure a-Si:H films the rocking/wagging mode occurs at 630 cm1 for all the configurations of silicon hydrogen bonds and the stretching mode occurs at 2000 and 2100 cm1, depending on whether hydrogen is bonded in monohydride or polyhydride configuration. However, as the crystallinity starts evolving, the wagging absorption band near 630 cm1 starts consisting of subbands towards high energy, and the band deviates from the symmetric shape [7,22]. We have observed the development of features in the absorption band with the increase of crystallinity irrespective of SFR or HFR.
Absorption Intensity (a.u.)
sample # 3 4. Conclusions
sample # 4
1000
In this paper we report on the preparation as well as structural and optical studies of hydrogenated silicon films deposited by the HWCVD method. We have deposited a series of hydrogenated amorphous silicon by varying the hydrogen and silane flow rates. Raman and XRD studies show that the films range from pure amorphous to nanocrystallite-embedded amorphous in nature. The presence of PL bands in the visible region further confirms the presence of nanocrystallites. We also thank Central Instrument Facility (CIF), IIT Guwahati for SEM studies.
500 sample # 5 0 1850
1900
1950
2000 2050 2100 Wave No. (cm-1)
2150
2200
Acknowledgments
Fig. 8. FTIR spectrum for a few samples deposited with HFR 15 SCCM. (a) Absorption band for Series III samples at 630 cm1 corresponding to wagging mode and (b) the stretching mode absorption. The spectra for samples #4 and 5 are vertically shifted for clarity.
We thank Prof. R.O. Dusane and Prof. D.S. Misra of IIT Bombay for allowing us to deposit films by the HWCVD method and use Raman set up, respectively, and Dr. Alka Kumbhar for her help during deposition.
2000 cm1 and the total integrated intensity in the range 2000–2100 cm1. Figs. 8a and b show the absorption bands for a few samples at 630 and 2000 cm1, respectively. The hydrogen concentration (CH) varies from 6.2 to 12.8 at%. The high CH compared to that of other HWCVD-deposited films [20] may be because of the low substrate temperature. Mahan et al. [21] have reported HWCVD films with high hydrogen content prepared at low substrate temperatures (17.8 and 12 at% for substrate temperature 180 and 230 1C). It is further to be noted that CH for the films deposited with SFR 2 SCCM are lower (6.2–8.7) than with 3 and 4 SCCM (11.5–12.8), which show mostly amorphous nature in Raman measurements. Series I (corresponding to SFR 4SCCM) shows monotonic increase in CH from 11.5% to 12.8%, when the HFR is increased from 0 to 15 SCCM. With the increase in HFR, the percentage of atomic hydrogen as well as band gap of the films increases, which may be associated with the relaxation of the constrained network and improvement in order. This is observed not only in HWCVD films but also in PECVD films [21]. For Series II (SFR 2 SCCM) the hydrogen content is almost same, except for sample #7 where CH (8.7%) is little higher than other
References [1] A. Matsuda, Microcrystalline silicon. Growth and device application, J. NonCryst. Solids 338–340 (2004) 1–12. [2] J. Doyle, R. Robertson, G.H. Lin, M.Z. He, A. Gallagher, Production of highquality amorphous silicon films by evaporative surface decomposition, J. Appl. Phys. 64 (1988) 3215–3223. [3] S.R. Jadkar, J.V. Sali, M.G. Takwale, D.V. Musale, S.T. Kshirsagar, The role of hydrogen dilution of silane and phosphorous doping on hydrogenated microcrystalline silicon (mc-Si:H) films prepared by hot wire chemical vapor deposition (HW-CVD) technique, Thin Solid Films 395 (2001) 206–212. [4] H. Matsumura, Summary of research in NEDO Cat-CVD project in Japan, Thin Solid Films 395 (2001) 1–11. [5] A.H. Mahan, Status of Cat- CVD (Hot Wire CVD) research in United States, Thin Solid Films 395 (2001) 12–16. [6] Y. Wang, X. Geng, H. Stiebig, F. Finger, Stability of microcrystalline silicon solar cells with HWCVD buffer layer, Thin Solid Films 516 (2008) 733–735. [7] S. Halindintwali, et al., Improved stability of intrinsic nanocrystalline Si thin films deposited by hot-wire chemical vapor deposition technique, Thin Solid Film 515 (2007) 8040–8044. [8] R. Swanepoel, Determination of the thickness and optical constants of amorphous silicon, J. Phys. E: Sci. Instrum. 16 (1983) 1214–1222. [9] S. Veprek, F.A. Sarott, Z. Iqbal, Effect of grain boundaries on the Raman spectra, optical absorption and elastic light scattering in nanometer-sized crystalline silicon, Phys. Rev. B 36 (6) (1987) 3344–3350.
ARTICLE IN PRESS P. Gogoi, P. Agarwal / Solar Energy Materials & Solar Cells 93 (2009) 199–205
[10] M. Ito, M. Kondo, Systematic study of photo degradation of tailored nanostructured Si solar cells by controlling their medium range order, Jpn. J. Appl. Phys. 45 (8) (2006) 230–232. [11] P. Gogoi, P.N. Dixit, P. Agarwal, Amorphous silicon films with high deposition rate prepared using argon and hydrogen diluted silane for stable solar cells, Sol. Energy Mater. Sol. Cells 91 (2007) 1253–1257. [12] M. Jana, D. Das, A.K. Barua, Role of hydrogen in controlling the growth of mc-Si:H films from argon diluted SiH4 plasma, J. Appl. Phys. 91 (2002) 5442–5448. [13] Y. He, C.Z. Yin, G. Cheng, L. Wang, X. Liu, G.Y. Hu, The structure and properties of nano size silicon films, J. Appl. Phys. 75 (1994) 797–803. [14] A. Matsuda, Growth mechanism of microcrystalline silicon obtained from reactive plasmas, Thin Solid Films 337 (1999) 1–6. [15] C.C. Tsai, G.B. Anderson, R. Thompson, B. Wacker, Control of silicon network structure in plasma deposition, J. Non Cryst. Solids 114 (1989) 151–153. [16] D. Beeman, R. Tsu, M.F. Thorpe, Structural information from the Raman spectrum of amorphous silicon, Phys. Rev. B 32 (2) (1985) 874–878.
205
[17] M. Stutzman, in: S. Mahajan (Ed.), Handbook of Semiconductors, vol. 3a, North-Holland, Amsterdam, 1994, pp. 657–781. [18] R.A. Street, Hydrogenated Amorphous Silicon, Cambridge University Press, Cambridge, 1992 (Chapter 2). [19] C.J. Fang, K.J. Gruntz, L. Ley, M. Cardona, F.J. Demond, G. Muller, S. Kalbitzer, The hydrogen content of a-Ge:H and a-Si:H as determined by IR spectroscopy, gas evolution and nuclear reaction techniques, J. Non Cryst. Solids 35–36 (1980) 255–260. [20] A.H. Mahan, J. Carapella, B.P. Nelson, R.S. Crandall, I. Balberg, Deposition of device quality, low H content amorphous silicon, J. Appl. Phys. 69 (1991) 6728–6730. [21] A.H. Mahan, R. Biswas, L.M. Gedvilas, D.L. Williamson, B.C. Pan, On the influence of short and medium range order on the material band gap in hydrogenated amorphous silicon, J. Appl. Phys. 96 (2004) 3818–3826. [22] S. Zhang, et al., Silicon thin films prepared in the transition region and their use in solar cells, Sol. Energy Mater. Sol. Cells 90 (2006) 3001–3008.