- Email: [email protected]
Low-temperature deposition of polycrystalline silicon thin films by hot-wire CVD
Solar Energy Materials and Solar Cells
ELSEVIER
Solar Energy Materials and Solar Cells 48 (1997) 269-277
Low-temperature deposition of polycrystalline silicon thin films by hot-wire CVD J.K. Rath*, H. Meiling, R.E.I. Schropp Utrecht Universi.tv, Debye Institute, Department o[Atomic and InterJhce Phvsics P.O. Box 80 000, 3508 TA Utrecht, The Netherlands
Abstract Polycrystalline silicon films have been prepared by hot-wire chemical vapor deposition (HWCVD) at a relatively low substrate temperature of 43OC. The material properties have been optimized for photovoltaic applications by varying the hydrogen dilution of the silane feedstock gas, the gas pressure and the wire temperature. The optimized material has 95% crystalline volume fraction and an average grain size of 70 nm. The grains have a preferential orientation along the (2 2 0) direction. The optical band gap calculated from optical absorption by photothermal deflection spectroscopy (PDS) showed a value of 1.1 eV, equal to crystalline silicon. An activation energy of 0.54 eV for the electrical transport confirmed the intrinsic nature of the films. The material has a low dangling bond-defect density of ~ 1017 cm 3. A photo conductivity of 1.9 × 10- 5 f U 1 cm- 1 and a photoresponse (trph/ad) of 1.4 × 10z were achieved. A high minority-carrier diffusion length of 334 nm as measured by the steady-state photocarrier grating technique (SSPG) and a large majority-carrier mobility-lifetime (laz) product of 7.1 × 10-7 cmZV i from steady-state photoconductivity measurement ensure that the poly-Si : H films possess device quality. A single junction n - i - p cell made in the configuration n+-c-Si/i-poly-Si : H/p-~tc-Si : H/ITO yielded 3.15% efficiency under 100 mW/cm 2 AM 1.5 illumination. Keywords: Polycrystalline silicon; Solar cell; Hot-wire chemical vapor deposition
1. Introduction In the d e v e l o p m e n t of p h o t o v o l t a i c (PV) devices there are two basic criteria; high efficiency a n d low cost. T h o u g h high efficiencies have been achieved by c-Si solar cells, * Corresponding author. Email: [email protected] 0927-0248/97/$17.00 ~; 1997 Elsevier Science B.V. All rights reserved PII S 0 9 2 7 - 0 2 4 8 ( 9 7 ) 0 0 1 10-4
270
J.K. Rath el al./Solar Energy Materials and Solar Cells 48 (1997) 269 277
the fabrication cost limits their widespread use. On the other hand, amorphous-silicon solar cells have a relatively low fabrication cost and large area deposition at a fast rate is possible. However. due to light-induced degradation their stabilized efficiency is limited. Therefore, a PV technology based on poly-silicon films seems to be a promising approach to combine both of the above advantages, i.e., low fabrication cost and stability. Thin-film silicon solar cells have been fabricated with efficiencies exceeding 10% by using a high-temperature process [1]. On the other hand, p - i - n microcrystalline cells using the very high-frequency (VHF) plasma CVD process at a low temperature have yielded 7.7% stable efficiency [2]. However, the deposition rate is too low ( < 2 A/s) for commercial use. To make the fabrication cost effective, a low-temperature process on cheap foreign substrates at a high deposition rate is essential. Hot-wire chemical vapor deposition (HWCVD) process satisfies all the requirements of a low-cost single-step fabrication of polycrystalline silicon thin films. Though many groups have claimed success in obtaining good-quality poly-Si films by H W C V D [3, 4], the successful demonstration of incorporation of these materials in a solar cell has not been presented so far. To that end, we have optimized deposition parameters to obtain device-quality poly-Si : H films and present here the first successful solar cell results by fabricating an n--i- p cell using poly-Si : H as the i-layer.
2. Experimental Poly-Si films were deposited on l0 cm x 10 cm Coming 7059 glass and on c-Si wafer substrates by HWCVD in one of the chambers of a high-vacuum multi-chamber system (PASTA). Two tungsten wires were used as hot filaments. The p-lac-Si : H and high band gap a-Si : H films were deposited by PECVD separately in other chambers of the same system. All the chambers are connected to a transport chamber with load-lock. The films were characterized by laser Raman spectroscopy, X-ray diffraction (XRD), electrical conductivity in the dark and photoconductivity with white light (100 mW/cm 2) illumination, steady-state photocarrier grating (SSPG), electron spin resonance (ESR) and fourier transform infrared (FTIR) measurements. The g'rproduct of the majority carrier was obtained under 700 nm monochromatic light with a flux of 1015 c m - 2 s 1. The thicknesses and deposition rates of the films were measured by Dektak profilometer and reflection/transmission measurements. Single-junction cells were fabricated on highly doped n +-c-Si wafer in the configuration n +-c-Si/i-poly-Si : H/p-~tc-Si : H/ITO. Cells were characterized by light and dark I - V and spectral response measurements.
3. Results and discussion To obtain polycrystalline Si fihns, the source gas Sill4 was diluted with hydrogen. The crystallinity depends strongly on the hydrogen dilution ratio. The crystallinity in the Si films was estimated by X-ray diffraction (XRD) and Raman spectroscopy measurements. Fig. 1 shows the XRD spectra of polysilicon films obtained at various
J.K. Rath et al./Solar Energy Materials and Solar Cells 48 (1997) 269-277
271
(111) SiHJH2=I%
(220)
SiH4/H2=5 %
sin4/I--I2= 10%
SiHdH2=I5%
60
50
40
30
20
10
20 (degree) Fig. 1. XRD spectra of poly-Si : H films at different SiHa/H2 dilutions.
SiH4/H2 ratio (r) values. The films obtained at a low r of 1% exhibit XRD peaks equivalent to (1 1 1) and (2 2 0) orientations of the powder diffraction of crystalline Si, suggesting that there is no preferential orientation of crystallites. However, at a higher r, the (2 2 0) line becomes dominant and the (1 1 1) line becomes less prominent. For an r value of 10%, only the (2 2 0) line could be observed. The grains are preferentially oriented along the (2 2 0) direction. The crystalline size, x, was calculated from the Scherrer formula [5], x = k2(A0) cos 0, where k = 0.9, 2 is the wavelength of the X-ray radiation, (A0) is the F W H M of the peaks (in units of 20) and 0 is the angular position of the peak. The crystalline volume fraction (Vf) was estimated from the Raman spectrum by the equation Vf = 1el(It + mla), where Ic and Ia are the deconvoluted intensities of the Raman spectrum at 520 c m - 1 corresponding to the crystalline part and at 480-500 cm-1 corresponding to amorphous grain boundaries. The value of m has been determined by taking into account the grain size obtained from XRD [6]. Fig. 2 shows the dependence of the crystalline volume fraction (V0 and the crystalline grain size (x) on the SiH4/H2 ratio. It is observed that Vf increases with increasing r upto a maximum of 94.6% at an r value of 10%. At higher r (15%) the Vf decreases.
272
.1.K. Rath el a/. Solar t',necW,' :l,lalertal,~ and No/ur ('ells 4~' ( 1997) 269 277
100]
-rlO0
9o 1 O rr x
.
80! 70] 60
19a
!90 ~85
x(111)
':~ ~80 ~-
x (220)
" "
x- 50 ~ dN 40J t'- 30 ~ 201 10ii
75 -70 i65
Vf ....
160 :;55 1
5
10
15
150
SiH4/Ha (%) Fig. -. ~' G r a i n sizes (.v) alone~ ( 1 1 I j a n d 1~" _" Ih o r i c n l a t i o n s c a l c u l a t e d f r o m X R D by. Schcl-rer l'ol-mtlla a n d l h e c r y s t a l l i n e v o l u m e f r a c t i o n l I;) e s l i m a t e d f r o m Rana~ln s p e c t r o s c o p y at d i f f e r e n t SiH4,"H2.
A careful o b s e r v a t i o n of the crystallites by X R D shows that the structure goes t h r o u g h d r a m a t i c changes. At a low r value of 1%, the grain sizes a l o n g the (2 2 0) and (1 l I) directions are equivalent. This suggests isotropic g r o w t h of the grains. However, with an increase of the r value, the size of the grains becomes unequal. The grain size a l o n g the (2 2 0) direction increases with a s i m u l t a n e o u s decrease along the ( 1 1 1) direction. This structural t r a n s f o r m a t i o n with change of dilution is confirmed by A F M studies [7] to be published. A F M pictures reveal that at r = 1%, the grains are spherically symmetrical. However, at higher r the grains are big and needle type with an a n i s o t r o p i c growth. The s a m p l e at r = 10% shows c o m p l e t e coalescence of grains. Features, which m a y consist of multiple grains, of the size of 0.5 lain could be observed in the A F M m i c r o g r a p h . The samples at a higher r value of 15% show similar big grains, but isolated from each o t h e r and s u r r o u n d e d by a m o r p h o u s matrix. Details of this study will be presented elsewhere [7]. Fig. 3 shows the d e p e n d e n c e of the m i n o r i t y - c a r r i e r diffusion length (L), the d e p o s i t i o n rate Rd and the activation energy IE,,) on r values. It is o b s e r v e d that Ea reaches a m i n i m u m value of 0.54eV at r = 10%. At a higher r value the Ea increases to 0.82eV and the d a r k c o n d u c t i v i t y O-d value falls to 4.6× 10- ~o ~ l cm 1 indicating a substantial a m o r p h o u s n e t w o r k in the c o n d u c t i n g path. We have o b s e r v e d this a m o r p h o u s phase between crystallites in the A F M picture. The L value increases from 37 nm at r = 1% to a m a x i m u m value of 220 nm at r = 10%. This could be a t t r i b u t e d to the increase in crystalline volume fraction that creates less g r a i n - b o u n d a r y defects. The density of g r a i n - b o u n d a r y defects has been d e d u c e d from electron spin resonance (ESR) that measures the density of neutral d a n g l i n g bonds. The defect density shows a systematic reduction from 2 . 9 x 1018 CC l at r = 1% to ~ 1017 cc - 1 at r = 10%. However, at a higher r of 15% the L value falls off to 94 nm. The h y d r o g e n content of the tihns r e m a i n e d low at 0.8 a t % for r = 1-10%, indicating a very low fraction of a m o r p h o u s network. However, it should be n o t e d that for the s a m p l e with r = 10%, the defect density remains at ~ 10 ~7 c c - x even t h o u g h the
J.K. Rathet al./Solar Energ3, Materials and Solar Cells 48 (1997) 269-277
273
3OO
0.9 0.8
0.7 0.6
...............
;:
,A,
2oo 150
t--
0.4
',,
~ 0.3 rr 0.2 0,
•
..-'
~
....
lOO ~._.1 50
A*
5
lo
15
SiH4/H2 (%)
Fig. 3. Deposition rate ( R d ) , activation energy (E.) and ambipolar diffusion length (L) at different SiH~/H2. hydrogen content is very low at 0.8 at%. This reflects the compactness of the polycrystalline structure having thin grain boundaries where a small hydrogen content of 0.8 a t % is sufficient to passivate most of the grain-boundary defects. This defect density is one order of magnitude less than for polycrystalline samples with a similar hydrogen content made in other laboratories [3]. This suggests that additional hydrogen passivation treatments are not necessary when using the present poly-Si : H in the device. The compactness of the structure at r -- 10% was confirmed by the IR spectrum where no oxygen line at 1050 c m - 1 (corresponding to SiO bonds) was observed even after several days of exposure to air. This is supported by the L values which showed no decline in the same period. The IR spectrum showed a strong Si-H stretching mode at 2000 cm -1 and a relatively weaker line at 2100 c m - 1. The absence of any bending modes at 800-900 c m - 1 suggests that the absorption at 2100 c m - ~ is not due to (Sill2), polyhydrides. It is therefore most likely due to clustered Sill bonds at the grain boundaries. However, the relative fraction of the 2100 c m - 1 mode (amount of microstructure) is much less in the present samples than in samples reported in literature where the 2100 cm 1 mode was dominant. The optical absorption was measured by photothermal deflection spectroscopy (PDS). The indirect optical gap was estimated from x/:~ versus E(eV) plot (Fig. 4) using the low-energy absorption region of PDS spectra. The optical gap of 1.1 eV for the sample obtained at r = 10% equal to crystalline silicon, correlates well with the data. The Neuchatel group [2] has found a lower optical gap for their microcrystalline samples which was attributed to the tensile strain and scattering [8]. We believe that our H W C V D poly-Si : H samples are stress free due to the larger grain size compared to the V H F C V D poly-Si : H developed by the Neuchatel group which has a grain size of typically < 20 nm. The second observation is that the activation energy of 0.54 eV for the band gap of 1.1 eV implies that the Fermi level is at the center of the gap. This is achieved without any boron compensation or use of any gas purifier. We attribute this to the compactness of the structure that prohibits in-diffusion of oxygen into the network, as confirmed by our IR studies. Moreover, the fast deposition rate (5.5 A/s) and a better vacuum in the chamber ensures less oxygen incorporation into the network.
274
J.K. Rath el al. ,Solar
EnepxvMaterials and Solar Cells 48/1997) 269 277
80.0 / 70.0 t i
60.0: 50.0tQ.
40.030.0: 20.0-
10.00.0
•
ram•
/
/
,
0.6 0.8110'1'.21'.41'.6118 2.0 Energy (eV)
Fig. 4. Square root of the absorption coefficiemI~) versus energy (E) measured by PDS.
As described in the above discussion, in the dilution series at a wire temperature of 1900~C, the poly-Si : H film with the best quality is achieved at a SiH4/H2 ratio of 10%. At this SiH4/H2 value, the wire temperature was varied to improve the material quality further, It has been observed that the deposition rate decreased slowly from Tw of 1900°C 1800~C. At a Tw < 1800:'C the deposition rate falls off drastically. We have observed that at a Tw of 1800'C, the diffusion length increased substantially to a high value of 334 nm. The photoconductivity (~rph} is 1.9 X 10 S ~ - 1 c m - 1 and aph/aa of 1.4 x 102 has been achieved. The IR spectrum showed that the 2100 cm i is completely absent and the stretching vibration is concentrated at 2000 c m - i. corresponding to Si-H mode only. This suggests that the grain-boundary defect and microstructure in this material have reduced significantly. The total hydrogen content measured from the wagging mode (640 cm -1) is only 0.47%. This value is comparable to the best I,tc-Si films made by the layer-by-layer technique by Ishihara et al. [-9]. However, the crystalline volume fraction and the average grain size and orientation of grains remained similar to the poly-Si films made at Tw of 1900°C. At Tw lower than 1800~C, the samples had a large amorphous volume fraction and the L values decreased drastically.
3.1. Cell properties We fabricated n i p cells as a test structure for the intrinsic poly-Si layers in the configuration n +-c-Si(Wafer)/i-poly-Si : H(HWCVD)/p-p.c-Si " H(PECVD)/ITO. Our approach to incorporate the polysilicon films in the above configuration is due to the
J.K. Rath et al./Solar Energy Materials and Solar Cells 48 (1997) 269 277
275
following reasons. We wanted to verify the photovoltaic quality of the poly-Si : H thin film in a solar cell without disturbing the intrinsic nature of the undoped layers that we are getting in the H W chamber by contamination with P or B. Secondly, the hot-wire chamber is part of the multi-chamber system where intrisic layers made by H W deposition can immediately be tested in solar cells using either amorphous or microcrystalline n- and p-type doped layers made in the other chambers by the P E C V D method. The properties of the p-p,c-Si : H layers have already been tested in other devices [10]. We have used heavily doped n+-c-Si wafer as substrate as well as the n-layer of the n-i-p cell to ensure that this does not contribute to the current. The above method leaves us with fewer variables that could influence the solar cell as a whole. The cell-test structure is shown in Fig. 5. For better contact, an A1 coating was made on the back side on the n +-c-Si wafer. The i-layer was made at the optimum condition of Tw = 1800°C, Ts = 430°C, Pr = 0.1 m b a r and SiH4/H2 ratio of 10%. There is a 3 nm wide band-gap buffer layer before the p-layer, which helps to improve the cell properties. We achieved an efficiency of 3.15% for an unoptimized cell. The I - V characteristics are shown in Fig. 6. A Voc of 0.45 V could be achieved. However
IF
i I
ITO (8Ohm) p-mc-Si:H(30nm)
I
o~y-Si:H(150Ohm)
t n+-c-Si (Wafer) CELL STRUCTURE
Fig. 5. Schematic representation of the n-i-p cell configuration.
'°/
'/
~.~ ~
~I........ l,=,~.~.~2-r-i.......... ]........... i........... i~l o~....... ~
...... ~.......... +.......... !........~......... .. - ~ i
i- I .......
........... i
t-----i .......... .......... i........... ........ i-..........
-~s .......... + .......... i........... i.......... .+........ i...........................................
.20~
-0.3
.........;..
..........
-0.2
-0.1
0.0
0.1 0.2 Voltage(V)
0.3
~, ..........
0.4
, ........
0.5
0.6
Fig. 6. I - V characteristics of an n-i p cell in the configuration shown in Fig. 5. Thickness of the i-layer 1.5 p.m.
276
,LK. Ralh ~I al. Solar l{netXv Material,~ and Solar CelLs' 48 (19971 269 277
the till factor (0.38~ needs a lot of improvement. A reasonably high J~c of 20 mA/cm 2 could be achieved in most cells while the i-layer was only 1.5 gm thick. It is possible due to the following reasons. (1t The poly-Si: H films have higher absorption compared to the c-Si. This is confirmed by the PDS measurements. (2) There is a good light trapping due to scattering at the textured surface of the poly-Si : H. A F M pictures reveal the roughness of the surface of the poly-Si fihns. For this material, the AZRM s (root mean square peak height) and AZ (maximum height of the peaks) are estimated to be 19.8 and 150 nm. respectively. With such roughness we would expect considerable scattering of light. In fact, optical modelling predicts such currents for this thickness of poly-Si : H films I l l ] . In an n i p cell we expect to collect all the photogenerated carriers due to drift field. The reason why we think that there is no current contribution from the highly doped n + substrate is the small minority carrier lifetime and carrier mobility at the doping level of - 101'~cm 3. Moreover, the holes froln the n + substrate would have to travel across the i-layer to be collected at the front contact. Holes being the minority carrier (as proved by the Hall measurements) and have the diffusion length (334 nm) much smaller than the thickness of the i-layer, their contribution from the bottom of the cell and the n + substrate is negligible. It should be noted that the yield was remarkably good (80%), indicating homogeneity of the cell deposition across at least 8 cm x 8 cm area. However, the spectral response measurement showed that there is still a lot of room for improvement because the external collection efficiency (ECE) with bias light was considerably lower than that in the dark. However, at negative bias < - 0.5 V, the ECE in light approaches ECE ill dark. A calculation Q E ( - I)/QE(0) at various wavelengths showed a constant value for all wavelengths. This suggests that the QE loss at 0 V could be due to a collection problem that is due to a barrier at an interface or resistive losses at external contacts. We studied the intensity dependence of the I V curves and found that indeed the efficiency increased with decreasing light intensity and reached a value of 4.8% at a light intensity of25 mW/cm 2. This was in spite of the fact that l,~,cdecreased monotonically with decrease in light intensity as expected. The improvement in efficiency was mostly due to improvement in FF and better carrier collection, it should be noted that similar characteristics have been observed for a p-lac-Si/n-c-Si heterojunction cell where such intensity dependence of spectral response and FF have been observed [12]. There are two limitations in the present polycrystalline cell. ( 1) The thin p-gc-Si : H layer with small grain sizes is a mixed-phase material and has band gap between c-Si and a-Si : H, i.e.. - 1.4 eV [13]. This leads to the band gap mismatch between p- and i-layers (Eg = 1.1 eV. shown beforej which is mostly in the valence-band offset [14]. This will result in a barrier with respect to the hole collection and holes will accumulate at the p/i interface. 12) The low mobility gap of the p-gc-Si : H results in substantial electron back diffusion [15] that leads to a decreased electric field in the i-layer. If we use high band-gap p-type amorphous silicon carbide layers, the valenceband offset will still be higher which will result in more accumulation of holes and bad collection of carriers. The use of the wide band-gap buffer layer in our cell structure removes the problem of electron back diffusion [15] to a large extent. However, the valence-band offset
J.K. Rath et al./Solar Energy Materials" and Solar Cells 48 (1997j 269 277
277
between the buffer layer a n d the p o l y - S i : H i-layer m a y form a barrier to hole collection. F u r t h e r work is in progress to address this problem.
4. Conclusions Device-quality poly-Si : H films have been o b t a i n e d by the H W C V D method. We have s h o w n that it is possible to get poly-Si : H films with the F e r m i level at the center of the gap even w i t h o u t b o r o n m i c r o - d o p i n g or the use of a gas purifier. G r a i n s showed preferential o r i e n t a t i o n s a n d the materials have ~ 95% crystalline volume fraction with grains of average size ~ 70 n m m a k i n g a c o m p a c t structure. High minority-carrier diffusion length of 334 n m makes these poly-Si : H films very promising for solar cell applications. We have d e m o n s t r a t e d the device performance of n i - p cells with poly-Si : H as active intrinsic layer. A n efficiency of 3.15% has been achieved a n d a c u r r e n t of 18,2 m A / c m 2 has been achieved in a n only 1.5 gm thick i-layer.
Acknowledgements We t h a n k D. Ruff a n d Dr. H. Mell of Philipps University, M a r b u r g , G e r m a n y for the P D S m e a s u r e m e n t . T h a n k s are due to M a r t e n van Cleef for the I T O deposition. This research was partially funded by N O V E M (Netherlands O r g a n i s a t i o n for Energy a n d E n v i r o n m e n t ) a n d is part of a film-Si project c o o r d i n a t e d by ECN, Petten.
References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15]
H. Keppner et al., Proc. 25th IEEE PVSC, Washington DC, 1996, p. 669. J. Meier et al., Proc. MRS Spring Meeting, vol. 420, San Francisco, 1996, p. 3. A.R. Middya, J. Guillet, J. Perrin, A. Lloret, E. Bourree, Proc. 13th ECPVSEC, Nice, 1995, p. 1704. J. Cifre, J. Bertomeu, J. Puigdollers, M.C. Polo, J. Andreu, A. Lloret, Appl. Phys. A 59 (19941 645. H.P. Klug, L.E. Alexander, X-ray Diffraction Procedure, Wiley, New York, 1974. E. Bustarret, M.A. Hachichia, M. Brunel, Appl. Phys. Lett 52 (1988) 1675. J.K. Rath et al., Proc. MRS Fall Meeting, vol. 452, Boston, 1996, p. 977. N. Beck, J. Meier, J. Fric, Z. Remes, A. Poruba, R. Fluckiger, J. Pohl, A. Shah, M. Vanecek, J. Non-Cryst. Solids 198 200 (1996) 903. Shun-ichi Ishihara, Deyan He, Masami Nakata, Isarnu Shimizu, Jpn. J. Appl. Phys. 32 (1993) 1539. J.K. Rath et al., Proc. 13th ECPVSEC, Nice, 1995, p. 1712. R. Krankenhagen, M. Schmidt, S. Koynov, S. Grebner, W. Henrion, I. Sieber, R. Schwarz, Proc. 131h ECPVSEC, Nice, 1995, p. 1700. M.W.M. van Cleef, J.K. Rath. F.A. Rubinelli, C.H.M. van der Weft, R.E.1. Schropp, Proc. 25th IEEE PVSC, Washington DC, 1996, p. 429. X. Xu, J. Yang, A. Bannerjee, S. Guha, K. Vasanth, S. Wagner, Appl. Phys. Lett. 67 (1995) 2323. H. Mimura, Y. Hatanaka, Appl. Phys. Lett. 50 (1987) 326. J.K. Rath, C.H.M. van der Werf, F.A. Rubinelli,R.E.I. Schropp, Proc. 25th IEEE PVSC, Washington DC, 1996, p. 1101.
ELSEVIER
Solar Energy Materials and Solar Cells 48 (1997) 269-277
Low-temperature deposition of polycrystalline silicon thin films by hot-wire CVD J.K. Rath*, H. Meiling, R.E.I. Schropp Utrecht Universi.tv, Debye Institute, Department o[Atomic and InterJhce Phvsics P.O. Box 80 000, 3508 TA Utrecht, The Netherlands
Abstract Polycrystalline silicon films have been prepared by hot-wire chemical vapor deposition (HWCVD) at a relatively low substrate temperature of 43OC. The material properties have been optimized for photovoltaic applications by varying the hydrogen dilution of the silane feedstock gas, the gas pressure and the wire temperature. The optimized material has 95% crystalline volume fraction and an average grain size of 70 nm. The grains have a preferential orientation along the (2 2 0) direction. The optical band gap calculated from optical absorption by photothermal deflection spectroscopy (PDS) showed a value of 1.1 eV, equal to crystalline silicon. An activation energy of 0.54 eV for the electrical transport confirmed the intrinsic nature of the films. The material has a low dangling bond-defect density of ~ 1017 cm 3. A photo conductivity of 1.9 × 10- 5 f U 1 cm- 1 and a photoresponse (trph/ad) of 1.4 × 10z were achieved. A high minority-carrier diffusion length of 334 nm as measured by the steady-state photocarrier grating technique (SSPG) and a large majority-carrier mobility-lifetime (laz) product of 7.1 × 10-7 cmZV i from steady-state photoconductivity measurement ensure that the poly-Si : H films possess device quality. A single junction n - i - p cell made in the configuration n+-c-Si/i-poly-Si : H/p-~tc-Si : H/ITO yielded 3.15% efficiency under 100 mW/cm 2 AM 1.5 illumination. Keywords: Polycrystalline silicon; Solar cell; Hot-wire chemical vapor deposition
1. Introduction In the d e v e l o p m e n t of p h o t o v o l t a i c (PV) devices there are two basic criteria; high efficiency a n d low cost. T h o u g h high efficiencies have been achieved by c-Si solar cells, * Corresponding author. Email: [email protected] 0927-0248/97/$17.00 ~; 1997 Elsevier Science B.V. All rights reserved PII S 0 9 2 7 - 0 2 4 8 ( 9 7 ) 0 0 1 10-4
270
J.K. Rath el al./Solar Energy Materials and Solar Cells 48 (1997) 269 277
the fabrication cost limits their widespread use. On the other hand, amorphous-silicon solar cells have a relatively low fabrication cost and large area deposition at a fast rate is possible. However. due to light-induced degradation their stabilized efficiency is limited. Therefore, a PV technology based on poly-silicon films seems to be a promising approach to combine both of the above advantages, i.e., low fabrication cost and stability. Thin-film silicon solar cells have been fabricated with efficiencies exceeding 10% by using a high-temperature process [1]. On the other hand, p - i - n microcrystalline cells using the very high-frequency (VHF) plasma CVD process at a low temperature have yielded 7.7% stable efficiency [2]. However, the deposition rate is too low ( < 2 A/s) for commercial use. To make the fabrication cost effective, a low-temperature process on cheap foreign substrates at a high deposition rate is essential. Hot-wire chemical vapor deposition (HWCVD) process satisfies all the requirements of a low-cost single-step fabrication of polycrystalline silicon thin films. Though many groups have claimed success in obtaining good-quality poly-Si films by H W C V D [3, 4], the successful demonstration of incorporation of these materials in a solar cell has not been presented so far. To that end, we have optimized deposition parameters to obtain device-quality poly-Si : H films and present here the first successful solar cell results by fabricating an n--i- p cell using poly-Si : H as the i-layer.
2. Experimental Poly-Si films were deposited on l0 cm x 10 cm Coming 7059 glass and on c-Si wafer substrates by HWCVD in one of the chambers of a high-vacuum multi-chamber system (PASTA). Two tungsten wires were used as hot filaments. The p-lac-Si : H and high band gap a-Si : H films were deposited by PECVD separately in other chambers of the same system. All the chambers are connected to a transport chamber with load-lock. The films were characterized by laser Raman spectroscopy, X-ray diffraction (XRD), electrical conductivity in the dark and photoconductivity with white light (100 mW/cm 2) illumination, steady-state photocarrier grating (SSPG), electron spin resonance (ESR) and fourier transform infrared (FTIR) measurements. The g'rproduct of the majority carrier was obtained under 700 nm monochromatic light with a flux of 1015 c m - 2 s 1. The thicknesses and deposition rates of the films were measured by Dektak profilometer and reflection/transmission measurements. Single-junction cells were fabricated on highly doped n +-c-Si wafer in the configuration n +-c-Si/i-poly-Si : H/p-~tc-Si : H/ITO. Cells were characterized by light and dark I - V and spectral response measurements.
3. Results and discussion To obtain polycrystalline Si fihns, the source gas Sill4 was diluted with hydrogen. The crystallinity depends strongly on the hydrogen dilution ratio. The crystallinity in the Si films was estimated by X-ray diffraction (XRD) and Raman spectroscopy measurements. Fig. 1 shows the XRD spectra of polysilicon films obtained at various
J.K. Rath et al./Solar Energy Materials and Solar Cells 48 (1997) 269-277
271
(111) SiHJH2=I%
(220)
SiH4/H2=5 %
sin4/I--I2= 10%
SiHdH2=I5%
60
50
40
30
20
10
20 (degree) Fig. 1. XRD spectra of poly-Si : H films at different SiHa/H2 dilutions.
SiH4/H2 ratio (r) values. The films obtained at a low r of 1% exhibit XRD peaks equivalent to (1 1 1) and (2 2 0) orientations of the powder diffraction of crystalline Si, suggesting that there is no preferential orientation of crystallites. However, at a higher r, the (2 2 0) line becomes dominant and the (1 1 1) line becomes less prominent. For an r value of 10%, only the (2 2 0) line could be observed. The grains are preferentially oriented along the (2 2 0) direction. The crystalline size, x, was calculated from the Scherrer formula [5], x = k2(A0) cos 0, where k = 0.9, 2 is the wavelength of the X-ray radiation, (A0) is the F W H M of the peaks (in units of 20) and 0 is the angular position of the peak. The crystalline volume fraction (Vf) was estimated from the Raman spectrum by the equation Vf = 1el(It + mla), where Ic and Ia are the deconvoluted intensities of the Raman spectrum at 520 c m - 1 corresponding to the crystalline part and at 480-500 cm-1 corresponding to amorphous grain boundaries. The value of m has been determined by taking into account the grain size obtained from XRD [6]. Fig. 2 shows the dependence of the crystalline volume fraction (V0 and the crystalline grain size (x) on the SiH4/H2 ratio. It is observed that Vf increases with increasing r upto a maximum of 94.6% at an r value of 10%. At higher r (15%) the Vf decreases.
272
.1.K. Rath el a/. Solar t',necW,' :l,lalertal,~ and No/ur ('ells 4~' ( 1997) 269 277
100]
-rlO0
9o 1 O rr x
.
80! 70] 60
19a
!90 ~85
x(111)
':~ ~80 ~-
x (220)
" "
x- 50 ~ dN 40J t'- 30 ~ 201 10ii
75 -70 i65
Vf ....
160 :;55 1
5
10
15
150
SiH4/Ha (%) Fig. -. ~' G r a i n sizes (.v) alone~ ( 1 1 I j a n d 1~" _" Ih o r i c n l a t i o n s c a l c u l a t e d f r o m X R D by. Schcl-rer l'ol-mtlla a n d l h e c r y s t a l l i n e v o l u m e f r a c t i o n l I;) e s l i m a t e d f r o m Rana~ln s p e c t r o s c o p y at d i f f e r e n t SiH4,"H2.
A careful o b s e r v a t i o n of the crystallites by X R D shows that the structure goes t h r o u g h d r a m a t i c changes. At a low r value of 1%, the grain sizes a l o n g the (2 2 0) and (1 l I) directions are equivalent. This suggests isotropic g r o w t h of the grains. However, with an increase of the r value, the size of the grains becomes unequal. The grain size a l o n g the (2 2 0) direction increases with a s i m u l t a n e o u s decrease along the ( 1 1 1) direction. This structural t r a n s f o r m a t i o n with change of dilution is confirmed by A F M studies [7] to be published. A F M pictures reveal that at r = 1%, the grains are spherically symmetrical. However, at higher r the grains are big and needle type with an a n i s o t r o p i c growth. The s a m p l e at r = 10% shows c o m p l e t e coalescence of grains. Features, which m a y consist of multiple grains, of the size of 0.5 lain could be observed in the A F M m i c r o g r a p h . The samples at a higher r value of 15% show similar big grains, but isolated from each o t h e r and s u r r o u n d e d by a m o r p h o u s matrix. Details of this study will be presented elsewhere [7]. Fig. 3 shows the d e p e n d e n c e of the m i n o r i t y - c a r r i e r diffusion length (L), the d e p o s i t i o n rate Rd and the activation energy IE,,) on r values. It is o b s e r v e d that Ea reaches a m i n i m u m value of 0.54eV at r = 10%. At a higher r value the Ea increases to 0.82eV and the d a r k c o n d u c t i v i t y O-d value falls to 4.6× 10- ~o ~ l cm 1 indicating a substantial a m o r p h o u s n e t w o r k in the c o n d u c t i n g path. We have o b s e r v e d this a m o r p h o u s phase between crystallites in the A F M picture. The L value increases from 37 nm at r = 1% to a m a x i m u m value of 220 nm at r = 10%. This could be a t t r i b u t e d to the increase in crystalline volume fraction that creates less g r a i n - b o u n d a r y defects. The density of g r a i n - b o u n d a r y defects has been d e d u c e d from electron spin resonance (ESR) that measures the density of neutral d a n g l i n g bonds. The defect density shows a systematic reduction from 2 . 9 x 1018 CC l at r = 1% to ~ 1017 cc - 1 at r = 10%. However, at a higher r of 15% the L value falls off to 94 nm. The h y d r o g e n content of the tihns r e m a i n e d low at 0.8 a t % for r = 1-10%, indicating a very low fraction of a m o r p h o u s network. However, it should be n o t e d that for the s a m p l e with r = 10%, the defect density remains at ~ 10 ~7 c c - x even t h o u g h the
J.K. Rathet al./Solar Energ3, Materials and Solar Cells 48 (1997) 269-277
273
3OO
0.9 0.8
0.7 0.6
...............
;:
,A,
2oo 150
t--
0.4
',,
~ 0.3 rr 0.2 0,
•
..-'
~
....
lOO ~._.1 50
A*
5
lo
15
SiH4/H2 (%)
Fig. 3. Deposition rate ( R d ) , activation energy (E.) and ambipolar diffusion length (L) at different SiH~/H2. hydrogen content is very low at 0.8 at%. This reflects the compactness of the polycrystalline structure having thin grain boundaries where a small hydrogen content of 0.8 a t % is sufficient to passivate most of the grain-boundary defects. This defect density is one order of magnitude less than for polycrystalline samples with a similar hydrogen content made in other laboratories [3]. This suggests that additional hydrogen passivation treatments are not necessary when using the present poly-Si : H in the device. The compactness of the structure at r -- 10% was confirmed by the IR spectrum where no oxygen line at 1050 c m - 1 (corresponding to SiO bonds) was observed even after several days of exposure to air. This is supported by the L values which showed no decline in the same period. The IR spectrum showed a strong Si-H stretching mode at 2000 cm -1 and a relatively weaker line at 2100 c m - 1. The absence of any bending modes at 800-900 c m - 1 suggests that the absorption at 2100 c m - ~ is not due to (Sill2), polyhydrides. It is therefore most likely due to clustered Sill bonds at the grain boundaries. However, the relative fraction of the 2100 c m - 1 mode (amount of microstructure) is much less in the present samples than in samples reported in literature where the 2100 cm 1 mode was dominant. The optical absorption was measured by photothermal deflection spectroscopy (PDS). The indirect optical gap was estimated from x/:~ versus E(eV) plot (Fig. 4) using the low-energy absorption region of PDS spectra. The optical gap of 1.1 eV for the sample obtained at r = 10% equal to crystalline silicon, correlates well with the data. The Neuchatel group [2] has found a lower optical gap for their microcrystalline samples which was attributed to the tensile strain and scattering [8]. We believe that our H W C V D poly-Si : H samples are stress free due to the larger grain size compared to the V H F C V D poly-Si : H developed by the Neuchatel group which has a grain size of typically < 20 nm. The second observation is that the activation energy of 0.54 eV for the band gap of 1.1 eV implies that the Fermi level is at the center of the gap. This is achieved without any boron compensation or use of any gas purifier. We attribute this to the compactness of the structure that prohibits in-diffusion of oxygen into the network, as confirmed by our IR studies. Moreover, the fast deposition rate (5.5 A/s) and a better vacuum in the chamber ensures less oxygen incorporation into the network.
274
J.K. Rath el al. ,Solar
EnepxvMaterials and Solar Cells 48/1997) 269 277
80.0 / 70.0 t i
60.0: 50.0tQ.
40.030.0: 20.0-
10.00.0
•
ram•
/
/
,
0.6 0.8110'1'.21'.41'.6118 2.0 Energy (eV)
Fig. 4. Square root of the absorption coefficiemI~) versus energy (E) measured by PDS.
As described in the above discussion, in the dilution series at a wire temperature of 1900~C, the poly-Si : H film with the best quality is achieved at a SiH4/H2 ratio of 10%. At this SiH4/H2 value, the wire temperature was varied to improve the material quality further, It has been observed that the deposition rate decreased slowly from Tw of 1900°C 1800~C. At a Tw < 1800:'C the deposition rate falls off drastically. We have observed that at a Tw of 1800'C, the diffusion length increased substantially to a high value of 334 nm. The photoconductivity (~rph} is 1.9 X 10 S ~ - 1 c m - 1 and aph/aa of 1.4 x 102 has been achieved. The IR spectrum showed that the 2100 cm i is completely absent and the stretching vibration is concentrated at 2000 c m - i. corresponding to Si-H mode only. This suggests that the grain-boundary defect and microstructure in this material have reduced significantly. The total hydrogen content measured from the wagging mode (640 cm -1) is only 0.47%. This value is comparable to the best I,tc-Si films made by the layer-by-layer technique by Ishihara et al. [-9]. However, the crystalline volume fraction and the average grain size and orientation of grains remained similar to the poly-Si films made at Tw of 1900°C. At Tw lower than 1800~C, the samples had a large amorphous volume fraction and the L values decreased drastically.
3.1. Cell properties We fabricated n i p cells as a test structure for the intrinsic poly-Si layers in the configuration n +-c-Si(Wafer)/i-poly-Si : H(HWCVD)/p-p.c-Si " H(PECVD)/ITO. Our approach to incorporate the polysilicon films in the above configuration is due to the
J.K. Rath et al./Solar Energy Materials and Solar Cells 48 (1997) 269 277
275
following reasons. We wanted to verify the photovoltaic quality of the poly-Si : H thin film in a solar cell without disturbing the intrinsic nature of the undoped layers that we are getting in the H W chamber by contamination with P or B. Secondly, the hot-wire chamber is part of the multi-chamber system where intrisic layers made by H W deposition can immediately be tested in solar cells using either amorphous or microcrystalline n- and p-type doped layers made in the other chambers by the P E C V D method. The properties of the p-p,c-Si : H layers have already been tested in other devices [10]. We have used heavily doped n+-c-Si wafer as substrate as well as the n-layer of the n-i-p cell to ensure that this does not contribute to the current. The above method leaves us with fewer variables that could influence the solar cell as a whole. The cell-test structure is shown in Fig. 5. For better contact, an A1 coating was made on the back side on the n +-c-Si wafer. The i-layer was made at the optimum condition of Tw = 1800°C, Ts = 430°C, Pr = 0.1 m b a r and SiH4/H2 ratio of 10%. There is a 3 nm wide band-gap buffer layer before the p-layer, which helps to improve the cell properties. We achieved an efficiency of 3.15% for an unoptimized cell. The I - V characteristics are shown in Fig. 6. A Voc of 0.45 V could be achieved. However
IF
i I
ITO (8Ohm) p-mc-Si:H(30nm)
I
o~y-Si:H(150Ohm)
t n+-c-Si (Wafer) CELL STRUCTURE
Fig. 5. Schematic representation of the n-i-p cell configuration.
'°/
'/
~.~ ~
~I........ l,=,~.~.~2-r-i.......... ]........... i........... i~l o~....... ~
...... ~.......... +.......... !........~......... .. - ~ i
i- I .......
........... i
t-----i .......... .......... i........... ........ i-..........
-~s .......... + .......... i........... i.......... .+........ i...........................................
.20~
-0.3
.........;..
..........
-0.2
-0.1
0.0
0.1 0.2 Voltage(V)
0.3
~, ..........
0.4
, ........
0.5
0.6
Fig. 6. I - V characteristics of an n-i p cell in the configuration shown in Fig. 5. Thickness of the i-layer 1.5 p.m.
276
,LK. Ralh ~I al. Solar l{netXv Material,~ and Solar CelLs' 48 (19971 269 277
the till factor (0.38~ needs a lot of improvement. A reasonably high J~c of 20 mA/cm 2 could be achieved in most cells while the i-layer was only 1.5 gm thick. It is possible due to the following reasons. (1t The poly-Si: H films have higher absorption compared to the c-Si. This is confirmed by the PDS measurements. (2) There is a good light trapping due to scattering at the textured surface of the poly-Si : H. A F M pictures reveal the roughness of the surface of the poly-Si fihns. For this material, the AZRM s (root mean square peak height) and AZ (maximum height of the peaks) are estimated to be 19.8 and 150 nm. respectively. With such roughness we would expect considerable scattering of light. In fact, optical modelling predicts such currents for this thickness of poly-Si : H films I l l ] . In an n i p cell we expect to collect all the photogenerated carriers due to drift field. The reason why we think that there is no current contribution from the highly doped n + substrate is the small minority carrier lifetime and carrier mobility at the doping level of - 101'~cm 3. Moreover, the holes froln the n + substrate would have to travel across the i-layer to be collected at the front contact. Holes being the minority carrier (as proved by the Hall measurements) and have the diffusion length (334 nm) much smaller than the thickness of the i-layer, their contribution from the bottom of the cell and the n + substrate is negligible. It should be noted that the yield was remarkably good (80%), indicating homogeneity of the cell deposition across at least 8 cm x 8 cm area. However, the spectral response measurement showed that there is still a lot of room for improvement because the external collection efficiency (ECE) with bias light was considerably lower than that in the dark. However, at negative bias < - 0.5 V, the ECE in light approaches ECE ill dark. A calculation Q E ( - I)/QE(0) at various wavelengths showed a constant value for all wavelengths. This suggests that the QE loss at 0 V could be due to a collection problem that is due to a barrier at an interface or resistive losses at external contacts. We studied the intensity dependence of the I V curves and found that indeed the efficiency increased with decreasing light intensity and reached a value of 4.8% at a light intensity of25 mW/cm 2. This was in spite of the fact that l,~,cdecreased monotonically with decrease in light intensity as expected. The improvement in efficiency was mostly due to improvement in FF and better carrier collection, it should be noted that similar characteristics have been observed for a p-lac-Si/n-c-Si heterojunction cell where such intensity dependence of spectral response and FF have been observed [12]. There are two limitations in the present polycrystalline cell. ( 1) The thin p-gc-Si : H layer with small grain sizes is a mixed-phase material and has band gap between c-Si and a-Si : H, i.e.. - 1.4 eV [13]. This leads to the band gap mismatch between p- and i-layers (Eg = 1.1 eV. shown beforej which is mostly in the valence-band offset [14]. This will result in a barrier with respect to the hole collection and holes will accumulate at the p/i interface. 12) The low mobility gap of the p-gc-Si : H results in substantial electron back diffusion [15] that leads to a decreased electric field in the i-layer. If we use high band-gap p-type amorphous silicon carbide layers, the valenceband offset will still be higher which will result in more accumulation of holes and bad collection of carriers. The use of the wide band-gap buffer layer in our cell structure removes the problem of electron back diffusion [15] to a large extent. However, the valence-band offset
J.K. Rath et al./Solar Energy Materials" and Solar Cells 48 (1997j 269 277
277
between the buffer layer a n d the p o l y - S i : H i-layer m a y form a barrier to hole collection. F u r t h e r work is in progress to address this problem.
4. Conclusions Device-quality poly-Si : H films have been o b t a i n e d by the H W C V D method. We have s h o w n that it is possible to get poly-Si : H films with the F e r m i level at the center of the gap even w i t h o u t b o r o n m i c r o - d o p i n g or the use of a gas purifier. G r a i n s showed preferential o r i e n t a t i o n s a n d the materials have ~ 95% crystalline volume fraction with grains of average size ~ 70 n m m a k i n g a c o m p a c t structure. High minority-carrier diffusion length of 334 n m makes these poly-Si : H films very promising for solar cell applications. We have d e m o n s t r a t e d the device performance of n i - p cells with poly-Si : H as active intrinsic layer. A n efficiency of 3.15% has been achieved a n d a c u r r e n t of 18,2 m A / c m 2 has been achieved in a n only 1.5 gm thick i-layer.
Acknowledgements We t h a n k D. Ruff a n d Dr. H. Mell of Philipps University, M a r b u r g , G e r m a n y for the P D S m e a s u r e m e n t . T h a n k s are due to M a r t e n van Cleef for the I T O deposition. This research was partially funded by N O V E M (Netherlands O r g a n i s a t i o n for Energy a n d E n v i r o n m e n t ) a n d is part of a film-Si project c o o r d i n a t e d by ECN, Petten.
References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15]
H. Keppner et al., Proc. 25th IEEE PVSC, Washington DC, 1996, p. 669. J. Meier et al., Proc. MRS Spring Meeting, vol. 420, San Francisco, 1996, p. 3. A.R. Middya, J. Guillet, J. Perrin, A. Lloret, E. Bourree, Proc. 13th ECPVSEC, Nice, 1995, p. 1704. J. Cifre, J. Bertomeu, J. Puigdollers, M.C. Polo, J. Andreu, A. Lloret, Appl. Phys. A 59 (19941 645. H.P. Klug, L.E. Alexander, X-ray Diffraction Procedure, Wiley, New York, 1974. E. Bustarret, M.A. Hachichia, M. Brunel, Appl. Phys. Lett 52 (1988) 1675. J.K. Rath et al., Proc. MRS Fall Meeting, vol. 452, Boston, 1996, p. 977. N. Beck, J. Meier, J. Fric, Z. Remes, A. Poruba, R. Fluckiger, J. Pohl, A. Shah, M. Vanecek, J. Non-Cryst. Solids 198 200 (1996) 903. Shun-ichi Ishihara, Deyan He, Masami Nakata, Isarnu Shimizu, Jpn. J. Appl. Phys. 32 (1993) 1539. J.K. Rath et al., Proc. 13th ECPVSEC, Nice, 1995, p. 1712. R. Krankenhagen, M. Schmidt, S. Koynov, S. Grebner, W. Henrion, I. Sieber, R. Schwarz, Proc. 131h ECPVSEC, Nice, 1995, p. 1700. M.W.M. van Cleef, J.K. Rath. F.A. Rubinelli, C.H.M. van der Weft, R.E.1. Schropp, Proc. 25th IEEE PVSC, Washington DC, 1996, p. 429. X. Xu, J. Yang, A. Bannerjee, S. Guha, K. Vasanth, S. Wagner, Appl. Phys. Lett. 67 (1995) 2323. H. Mimura, Y. Hatanaka, Appl. Phys. Lett. 50 (1987) 326. J.K. Rath, C.H.M. van der Werf, F.A. Rubinelli,R.E.I. Schropp, Proc. 25th IEEE PVSC, Washington DC, 1996, p. 1101.