PCCP View Article Online
Published on 09 September 2014. Downloaded by Colorado College on 31/10/2014 00:28:55.
PAPER
Cite this: Phys. Chem. Chem. Phys., 2014, 16, 22222
View Journal | View Issue
Stabilization mechanism of electrodeposited silicon thin films C. Vichery,*a V. Le Nader,a C. Frantz,a Y. Zhang,b J. Michlera and L. Philippea Amorphous composite silicon thin films electrodeposited in tetrahydrofuran, containing up to 80 at% of Si and exhibiting an homogeneous dispersions of O, C and Cl in the amorphous Si matrix, have been successfully stabilized against oxidation using a post-annealing step in inert atmosphere. In order to understand the impact of the annealing step on their stabilization against oxidation, their composition and structure have been investigated upon heat treatments. It has been shown that the presence of impurities such as O, C and Cl does not have any impact on the stabilization process, which is rather linked to the presence of hydrogen in the Si composites. This conclusion has been drawn after a detailed analysis of the bonding structure of films annealed at different temperatures and dwell times by
Received 25th June 2014, Accepted 8th September 2014 DOI: 10.1039/c4cp02797c
the mean of Raman spectroscopy. It has been shown that annealing the as-deposited films at 350 1C for a couple of hours or at higher temperatures induced a hydrogen evolution, characterized by the breaking of Si–H bonds and the formation of Si–Si bonds, which stabilized the silicon network. The understanding and the reproducibility of this stabilization process of silicon thin film electrodeposited in
www.rsc.org/pccp
organic solvent paves the way for their use for many applications.
Introduction Silicon is one of the most widely used semiconductor materials for a broad range of applications in green energy (solar cells), electronics and other high-tech fields. Considering the high cost of silicon deposition via vacuum processes, significant effort has been focused on developing simple, cheap, and easily scalable alternative methods such as electrodeposition. Due to the high reactivity of silicon precursors with water and their highly negative reduction potential, silicon electrodeposition is not straightforward. It requires an inert atmosphere (typically in a glove box) and the use of perfectly anhydrous and pure reactants. In the 1980’s, electrodeposition of silicon was conducted in organic solvents such as propylene carbonate1 and tetrahydrofuran2 from silicon halides. However, pure silicon films could not be obtained and electrodeposits contained carbon, oxygen and chlorine impurities. More recently, Si has been successfully electrodeposited in ionic liquids.3,4 Unfortunately, ionic liquids are expensive, highly viscous, and not always commercially available, which may render the transfer to industry challenging. One major problem with electrodeposited silicon is the subsequent oxidation of the films when taken out of the glove box. Until now, it is not clear if the presence of oxygen is solely due to a
EMPA, Swiss Fed Labs Mat Sci & Technol, Laboratory for Mechanics of Materials & Nanostructures, CH-3602 Thun, Switzerland. E-mail: [email protected] b EMPA, Swiss Fed Labs Mat Sci & Technol, Electron Microscopy Center, ¨bendorf, Switzerland CH-8600 Du
22222 | Phys. Chem. Chem. Phys., 2014, 16, 22222--22228
this oxidation process or to the trapping of oxygen during deposition. One reported method to stabilize electrodeposited Si films, is to anneal at 350 1C for 30 min in an inert or reducing atmosphere.5 However, this has not yet proven to be reproducible.6,7 Pursuing this idea of post-synthesis heat-treatment, we successfully stabilized Si electrodeposited in dichloromethane through microwave heating at 600 1C in Ar/H2 atmosphere, although with high surface oxidation.8 Here we present a two-step synthesis, involving (i) the deposition of silicon in tetrahydrofuran (THF) and (ii) the heat treatment of the as-deposited films inside the glove box. A systematic study of the influence of the annealing parameters (temperature and dwell time) on the film composition, along with a detailed Raman analysis of the film structure will give some insight into the relationship between heat treatment and the stabilization of electrodeposited silicon films.
Experimental methods Electrodeposition Silicon thin films were synthesized at room temperature by electrodeposition in organic solvent in an argon atmosphere (H2O, O2 o 1 ppm) MBraun glove box and using an Autolab PGSTAT-101 potentiostat. Potentiostatic depositions were performed onto transparent conductive oxide (TCO) using a classical 3 electrode setup consisting of a Pt counter electrode and a Pt pseudo-reference electrode. All potentials are quoted versus Pt. The working electrode was a commercial TCO (aluminoborosilicate glass/SnO2:F) from
This journal is © the Owner Societies 2014
View Article Online
Published on 09 September 2014. Downloaded by Colorado College on 31/10/2014 00:28:55.
Paper
PCCP
Solaronix AG, Switzerland. The substrates were cleaned through 2 successive sonication steps of 15 min in acetone and isopropanol. For optimum bath stability, and film adhesion,6 a bath containing 0.3 M tetrachlorosilane (SiCl4, 99.998% Aldrich), and 0.1 M tetrabutylammonium chloride (TBACl, 97% Aldrich) as the supporting electrolyte were used in tetrahydrofuran (THF, anhydrous, inhibitor free, 99.9% Sigma Aldrich). All chemicals were used without any further purification except where mentioned. As already reported,7 we observed that only the addition of SiCl4 permitted the dissolution of TBACl in THF. The silicon films were grown under potentiostatic conditions (3.1 V vs. Pt) for 1800 s, and were afterwards rinsed in THF. Heat treatment The samples were annealed in temperatures up to 700 1C under argon atmosphere (in the glove box) with a ceramic hot plate, pre-heated at the desired temperature. Two thermocouples were used to monitor the temperature, one on the back side of the ceramic plate, and one on the top of the samples (the latter only for preliminary tests). To anneal samples at higher temperatures (750–900 1C), they were heated for a few seconds at 650 1C inside the glove box and then transferred to a tubular furnace in evacuated quartz tubes. Characterization The film morphology was characterized by scanning electron microscopy (SEM Hitachi S4800) with an acceleration voltage of 1.5 kV. Cross-sections were prepared by focused ions beam (FIB) using a dual beam FIB-SEM Tescan Vela instrument. TEM samples were prepared by scratching the film surface and dispersing the powder onto a carbon grid, and imaged using a JEOL 2200FS microscope. The films’ atomic composition was investigated via energy dispersive X-ray spectroscopy (EDX, Genesis 4000 EDAX), He-elastic recoil detection analysis (He-ERDA) and glow discharge optical emission spectroscopy (GD-OES, Horiba Jobin Yvon JY 5000 RF).9 The silicon films were also characterized by using an upright confocal Raman spectrometer (NT-MDT NTEGRA) operating with a solid state laser (532 nm) and a 100 objective. The incident light power was set to 0.5 mW in order to avoid the local crystallization of the amorphous silicon under the beam. X-ray diffraction patterns were recorded using a Bruker D8 Discovery (lCuKa = 1.5418 Å) diffractometer. In order to obtain thicker films for XRD and EDX (to minimize substrate effects), 4 cycles of deposition/annealing (650 1C, few seconds) were performed in a row on the same substrate. All the characterizations were done, at most, 30 min after taking the samples out of the glove box.
Results and discussion Silicon deposition Fig. 1a shows a cyclic voltammogram of 0.3 M SiCl4 and 0.1 M TBACl in THF recorded with a scan rate of 0.1 V s1. One single reduction peak is observed at 1.75 V vs. Pt and can be attributed to the reduction of the silicon precursor into elemental silicon. The large increase in current between 1.4 and 3.5 V
This journal is © the Owner Societies 2014
Fig. 1 Cyclic voltammogram at 0.1 V s1, (a) and chronoamperogram at 3.1 V vs. Pt (b) in THF containing 0.3 M SiCl4 and 0.1 M TBACl.
is associated with the reduction of the supporting electrolyte. Here, a deposition potential of 3.1 V vs. Pt has been chosen as a compromise between the deposition rate and the decomposition of the electrolyte. The corresponding chronoamperogram is shown in Fig. 1b. During deposition, the large initial transient recorded is due to the charging of the double layer, followed by the reduction of the initially high concentration of Si precursor at the electrode/ electrolyte interface, and the side reduction of the supporting electrolyte and solvent. However, for the rest of the transient, up to approx. 200 s, the current decrease can be explained by changes of reaction kinetics as the electrode surface is gradually covered by the electrodeposit. Indeed, Munisamy and Bard showed that the diffusion regime is never reached, even for much lower Si precursor concentrations, since the Si deposition starts to be inhibited after a Si layer of about 2 nm is formed on the substrate.6 From 200 s to approximately 800 s, the slight increase in the cathodic current might be related to the formation of cracks which thereby provide less resistive pathways through the deposit. This assumption is supported by the SEM images (Fig. 2a) which highlight a relatively high density of wide cracks. Nevertheless, no delamination was observed. Heat treatment The largest issue facing scientists working on the electrodeposition of silicon is the oxidation of the films once taken out of the glove box. Following an idea previously developed,5 electrodeposited
Phys. Chem. Chem. Phys., 2014, 16, 22222--22228 | 22223
View Article Online
Published on 09 September 2014. Downloaded by Colorado College on 31/10/2014 00:28:55.
PCCP
Paper
Fig. 3 Composition of silicon films deposited at 3.1 V for 30 min, as deposited, and annealed between 200 and 800 1C for 30 min, obtained by GD-OES, after correction from air leakage. All the values are in atomic percentage, the dashed lines are guides for the eyes.
Fig. 2 SEM (a, c) and optical images (b, d) of silicon films (B300 nm), respectively as-deposited and annealed at 700 1C for 30 min. SEM images of a top view (e) and cross-section (f) of the thick silicon film (B2 mm) annealed at 650 1C used for EDX and XRD experiments.
silicon films were annealed at different temperatures and dwell times in inert atmosphere. After annealing, it has been observed that the cracks already present in the as-deposited films extended, thus forming silicon islands (Fig. 2c). Film composition Immediately after annealing, GD-OES measurements were performed on a series of silicon films heat treated at temperatures ranging from 200 to 800 1C. The results are presented in Fig. 3. One can see that the composition stayed constant upon annealing, within the error of the measurement (4%). The high amount of carbon in the films could be explained by the decomposition of the supporting electrolyte, due to the large potential applied during the deposition process. In order to confirm these GD-OES measurements, EDX was performed. It is interesting to note (see Fig. 2e and f) that the thick film used for this characterization did not present wide cracks but a more homogeneous and porous morphology. In good accordance with GD-OES measurements, the composition obtained by EDX was (in at% 3%): 78% Si, 12% O, 6% C and 4% Cl. EDX also brought new information to light, namely the presence of about 4% Cl in the films. This is not surprising, as it is believed that the mechanism of silicon deposition involves SimCln (n/m o 4) species.10 To have a better insight in the nature of the oxygen content in the electrodeposits, a thick stabilized film was deposited onto a stainless steel substrate. Its composition has been determined by EDX before and after etching in 1% HF during
22224 | Phys. Chem. Chem. Phys., 2014, 16, 22222--22228
Table 1 Composition of a thick stabilized silicon film deposited on stainless steel, before and after HF etching, as determined by EDX
Composition (at%)
Si
O
C
Cl
Before etching After etching
70 68
14 13
12 15
4 4
45 min. As one can see in Table 1, the composition remained the same after the HF treatment. There was thus no thick silica surface layer on the silicon. The oxygen, along with the carbon and the chlorine, should therefore be homogeneously dispersed in an amorphous silicon network. To determine the origin of oxygen in the deposited films, the solvent, THF, was treated with a further drying step before deposition. This was carried in the glove box, by leaving the solvent for 4 days in a closed bottle with molecular sieves (3 Å, 20% m/v). The resulting silicon film, after heat treatment at 650 1C for 30 s, showed a lower oxygen content. From EDX, the composition was found to be: 80% Si, 7% O, 9% C and 4% Cl. A careful drying of the chemical is thus an important step if one wants to decrease the oxygen content of electrodeposited silicon films. The remaining oxygen could arise from the presence of water in the silicon precursor or in the supporting electrolyte powder and also from the decomposition of THF itself. In order to monitor the evolution of the hydrogen content upon heat treatment, He-elastic recoil detection analysis experiments were performed on 3 samples that were: (i) as-deposited, (ii) annealed at 650 1C for B30 s and (iii) annealed at 800 1C for 5 h. It was found that the atomic percentage of hydrogen decreases while increasing the heating temperature: 24 5 at% for the as deposited film, 13 3 at% for annealing at 650 1C, and 2.2 0.5 at% annealing at 800 1C. Considering the He-ERDA and
This journal is © the Owner Societies 2014
View Article Online
Published on 09 September 2014. Downloaded by Colorado College on 31/10/2014 00:28:55.
Paper
PCCP
EDX results, it is thus possible to have the approximate composition of these 3 samples: (i) Si0.59H0.24O0.09C0.05Cl0.03. (ii) Si0.68H0.13O0.10C0.05Cl0.03. (iii) Si0.76H0.02O0.12C0.06Cl0.04. The presence of such high quantity of hydrogen in the as-deposited films would likely be due to the presence of water in the electrolyte. Assuming a 4 electron process for the reduction of Si4+ to Si0, the current efficiency of the reaction (Z) can be calculated by using the following formula: Z¼
4 F mSi Q MSi
with F the Faraday constant, mSi the mass of Si in the deposited film, MSi the molar mass of silicon and Q the charge passed during electrodeposition. The current efficiency was measured using thick samples, i.e. grown over 4 cycles, in order to weight the deposit with enough accuracy, and for a composition of Si0.68H0.13O0.10C0.05Cl0.04, the efficiency was of 28%. Film stabilization The films obtained after deposition were yellow-brownish in color but turned white after a few hours when taken out of the glove box (Fig. 2b). For the heat treated samples, we observed that those annealed in the glove box at less than 400 1C remained yellowish, whereas those annealed at higher temperature turned dark brown. Only the latter samples kept their dark brown color when exposed to air (Fig. 2d). It thus seems that a heat treatment of at least 400 1C is necessary to stabilize the films against further oxidation. As the quantities of carbon, oxygen and chlorine do not change upon annealing, the stability of the films should not be linked to their presence. On the contrary, the quantity of hydrogen decreases upon heat treatment. Thus, hydrogen removal seems to be directly related to the stabilization of the films. The change in color (from yellow to brown) which goes along with the stabilization reinforces this assumption: it could be explained by the fact that the band gap of a-Si:H decreases when the hydrogen content decreases.11 As we know from He-ERDA that the quantity of hydrogen decreased upon annealing, this hydrogen evolution could lead to a reduced band gap of the silicon composite and thus to higher absorbance of visible light by the film, hence a darker color. To understand more deeply what changes happen in the structure of the films upon annealing, the series was characterized by Raman spectroscopy. The corresponding spectra are presented in Fig. 4. For crystalline silicon, the sharp strong peak observed at 520 cm1 in the spectrum is due to a triply degenerate first order transverse optical (TO) phonon.12,13 When the disorder in the silicon structure increases, the momentum selection rules is relaxed and new phonon modes are allowed. For instance in Fig. 4, the silicon thin films which were annealed at 800 and 900 1C clearly behave as nano-crystalline films, i.e. as films consisting of a dispersion of nano-sized Si crystallites (peak at 521 cm1) in an amorphous Si matrix (peak at 480 cm1).12,14,15 The presence of both amorphous and crystalline silicon can also be observed by X-ray diffraction (Fig. 5c) and high resolution
This journal is © the Owner Societies 2014
Fig. 4 Raman spectra of silicon thin films annealed at different temperatures for 30 min. Laser wavelength: 532 nm, spectral resolution: 2.8 cm1, objective magnification: 100, power B0.5 mW. In inset, a magnified view of the room temperature, 200 and 300 1C heat treated samples spectra at low energy.
Fig. 5 High resolution (a) and dark field (b) TEM images of a film annealed at 850 1C for 1 h. In inset, the corresponding selected-area diffraction pattern, a white circle indicating the position of the objective lens aperture for the latter image. (c) X-ray diffraction pattern of a thick film (B2 mm) annealed at 800 1C for 15 h and of the substrate (the peaks correspond to SnO2). Inset is a magnification to highlight the Si(111) peak, marked with a star.
TEM and dark field images (Fig. 5a and b). One can see in the XRD pattern the presence of a small peak which can be attributed to the (111) plane family of silicon. The corresponding
Phys. Chem. Chem. Phys., 2014, 16, 22222--22228 | 22225
View Article Online
Published on 09 September 2014. Downloaded by Colorado College on 31/10/2014 00:28:55.
PCCP
Paper
coherence length, as determined by the Scherrer formula, is 6.9 0.6 nm. The dark field image reveals the presence of monocrystalline nanoparticles (bright spots) between 3 and 10 nm in diameter dispersed in an amorphous matrix. High resolution images in bright field also show the presence of these nanoparticles as one can observe lattice-fringes in Fig. 5a. The size of these mono-crystalline nanoparticles is in good agreement with the coherence length of silicon. In the Raman spectra, the increase in disorder in the silicon structure leads first to an increase of the full width at half maximum (FWHM) of the peak at 521 cm1.14,16,17 Then, the break-down of the zero momentum selection rule leads to additional contributions: new vibrational modes such as higher order TO, longitudinal optic (LO), transversal acoustic (TA) and longitudinal (LA) acoustic modes contribute to the spectrum.15,18,19 As shown in Fig. 4, for the films annealed between 400 and 700 1C for 30 min, the contribution of the first order TO is not present while the contributions of the other TO, LO, TA and LA modes are observed. In these spectra, the broad peak centered around 640 cm1 corresponds to the wagging modes of Si–H2 and Si–H3 and highlights the presence of hydrogen.18,20 These films are thus clearly made of a-Si:H. In order to link the presence of hydrogen to the stability of the electrodeposited a-Si:H films, the spectra have been deconvoluted by using Gaussian functions to extract each contribution. An example of deconvolution is given in Fig. 6. In this work the hydrogen content is evaluated by the ratio AðSiHx Þ AðSiH2 Þ þ AðSiH3 Þ ¼ AðSiÞ AðSiðLAÞÞ þ AðSiðLOÞÞ þ AðSiðTOÞÞ where A(x) is the area of the peak x. Fig. 7a illustrates the evolution of the hydrogen content versus the annealing temperature and the annealing time. With respect to the temperature dependency, there is a threshold value around 350 1C for which the ratio A(SiHx)/A(Si) decreases from B1 to B0. From this first series, it is clear that there is a major change in the film structure, linked to their stabilization, after a heat
Fig. 6 Deconvolution with Gaussian functions of the Raman spectrum of a a-Si:H thin film annealed at 400 1C for 30 minutes. Before applying the deconvolution process, the fluorescence background has been subtracted.
22226 | Phys. Chem. Chem. Phys., 2014, 16, 22222--22228
Fig. 7 (a) Variation of the A(SiHx)/A(Si) ratio versus annealing temperature for samples annealed for 30 minutes. (b) Variation of the A(SiHx)/A(Si) ratio versus annealing time for an annealing temperature of 350 1C.
treatment of around 350 1C. To determine more accurately the dwell time at which this transition occurs when heat treated at 350 1C, thin films were annealed for times ranging from 10 min to 15 h. The corresponding Raman spectra are presented in Fig. 8. One can clearly observe hydrogen removal with increasing annealing duration. As before, the Raman spectra have been deconvoluted. The resulting A(SiHx)/A(Si) ratio is plotted in Fig. 7b: a dwell time of about 2 h is necessary in order to break Si–H bonds. One can see between the elemental composition evolution (Fig. 3) and the analysis of Raman spectra (Fig. 7a), that the Si–H bonds have been broken after heat treatment at temperatures higher than 350 1C. This result is consistent with what has been previously reported in the literature. It is worth mentioning that similar annealing strategies are used routinely in the semiconductor community to remove hydrogen from silicon.21 Hydrogen is considered to desorb from a-Si:H thin films by simultaneous breaking of two Si–H bonds and formation of H2, the latter getting out of the film by rapid diffusion through voids.22,23 This process of hydrogen release allows the formation of new Si–Si bonds,24 or in the case of a-SiC:H, Si–C bonds,25 inducing a longrange relaxation of the silicon network. This interpretation is also supported by the variation of the A(Si(TA))/A(Si(TO)) ratio which is related to the intermediate-range order.15 Indeed for a-Si:H thin films in which the hydrogen desorption process has not yet started, the wagging vibrational Si–Hx modes are
This journal is © the Owner Societies 2014
View Article Online
Published on 09 September 2014. Downloaded by Colorado College on 31/10/2014 00:28:55.
Paper
PCCP
successively electrodeposited in tetrahydrofuran and stabilized against oxidation through a heat treatment step in inert atmosphere. It has been shown that a heat treatment of 2 h at 350 1C in Ar induced the breaking of Si–H bonds and the formation of Si–Si bonds, leading to a better cross-linkage and a relaxation of the silicon network, and thus to the stabilization of the films. Heating the films at temperatures higher than 800 1C induced the formation of Si crystallites of less than 10 nm in the amorphous matrix. This study highlighted that the presence of impurities such as O, C and Cl was not the reason for the instability of electrodeposited silicon films, but that it was rather linked to a high amount of hydrogen. The stabilization mechanism of silicon films electrodeposited in organic solvent being now understood and reproducible, such an easy and scalable process paves the way of using these films for many applications, such as for Li-ion batteries or photovoltaics.
Acknowledgements Fig. 8 Raman spectra of silicon thin films annealed at 350 1C for different dwell times. Laser wavelength: 532 nm, spectral resolution: 2.8 cm1, objective magnification: 100, power B0.5 mW.
predominant and no TA contribution is observed in the spectra.18 In this state, the Si network is strongly disorganized. When the desorption process starts, a more organized amorphous silicon network is created because of the formation of new Si–Si bonds. Consequently the LO, TO and TA mode contributions can be observed in the spectra. One can notice in Fig. 4 and 8 that the ratio A(Si(TA))/A(Si(TO)) decreases with annealing temperature and time. As long as the hydrogen desorption proceeds, the silicon network increases its degree of order and the number of defects decreases, leading to a structural rearrangement, from amorphous to crystalline. This structural rearrangement upon annealing induces a decrease in compressive stress,25,26 which could explain the shrinking and cracking of the film observed after heat treatment. The structural relaxation and the better cross-linkage of silicon after hydrogen removal is certainly the reason of the stability of the annealed films. It is furthermore important to point out that this annealing step must be carried out in inert atmosphere. Indeed, if some oxygen is present, it will be readily decomposed to produce Si–O bonds because of the formation of highly reactive dangling bonds after the breaking of Si–H bonds. Thus, it would result in the oxidation of the films.26,27 The stability over time of the stabilized silicon thin films has been confirmed by re-measuring the Raman spectra of the samples after 3 months storage in air. The spectra remained unchanged, and even after more than 9 months in air, the samples have remained dark brown in color (Fig. 2d).
Conclusions Silicon composite films about 300 nm thick, and containing up to 80 at% Si with O, C and Cl as impurities have been
This journal is © the Owner Societies 2014
The authors would like to acknowledge Dr M. Doebeli (ETH Zurich, Switzerland) for the He-ERDA measurements, M. Mieszala (EMPA) for the FIB cuts, D. Frey (EMPA) for the GD-OES analyses and Dr P. Dunne for helping reviewing this manuscript.
Notes and references 1 A. K. Agrawal and A. E. Austin, J. Electrochem. Soc., 1981, 128, 2292. 2 J. Gobet and H. Tannenberger, J. Electrochem. Soc., 1988, 135, 109. 3 S. Z. El Abedin, N. Borissenko and F. Endres, Electrochem. Commun., 2004, 6, 510. 4 J. Mallet, M. Molinari, F. Martineau, F. Delavoie, P. Fricoteaux and M. Troyon, Nano Lett., 2008, 8, 3468. 5 E. R. Bucker and J. A. Amick, US Pat., 204/14N,60,37R,385, 1980. 6 T. Munisamy and A. J. Bard, Electrochim. Acta, 2010, 55, 3797. 7 J. P. Nicholson, J. Electrochem. Soc., 2005, 152, C795. 8 M. Bechelany, J. Elias, P. Brodard, J. Michler and L. Philippe, Thin Solid Films, 2012, 520, 1895. 9 Th. Nelis, M. Aeberhard, M. Hohl, L. Rohr and J. Michler, J. Anal. At. Spectrom., 2006, 21, 112. 10 Y. Nishimura, T. Fukunaka, C. R. Miranda, T. Nishida, T. Nohira and R. Hagiwara, ECS Trans., 2009, 16, 1. `re, A. Deneuville, A. Mini, J. Fontenille and 11 J. C. Bruye R. Danielou, J. Appl. Phys., 1980, 51, 2199. 12 Z. Iqbal and S. Veprek, J. Phys. C: Solid State Phys., 1982, 15, 377. 13 H. Richter, Z. P. Wang and L. Ley, Solid State Commun., 1981, 39, 625. 14 W. Li, D. Xia, H. Wang and X. Zhao, J. Non-Cryst. Solids, 2010, 356, 2552. 15 Z. Li, W. Li, Y. Jiang, H. Cai, Y. Gong and J. He, J. Raman Spectrosc., 2011, 42, 415.
Phys. Chem. Chem. Phys., 2014, 16, 22222--22228 | 22227
View Article Online
Published on 09 September 2014. Downloaded by Colorado College on 31/10/2014 00:28:55.
PCCP
16 S. K. Gupta and P. K. Jha, Solid State Commun., 2009, 149, 1989. 17 G. Faraci, S. Gibilisco, P. Russo, A. R. Pennisi and S. La Rosa, Phys. Rev. B: Condens. Matter Mater. Phys., 2006, 73, 033307. 18 S. Miyazaki, N. Fukuhara and M. Hirose, J. Non-Cryst. Solids, 2000, 266–269, 59. 19 C. Smit, R. A. C. M. M. van Swaaij, H. Donker, A. M. H. N. Petit, W. M. M. Kessels and M. C. M. van de Sanden, J. Appl. Phys., 2003, 94, 3582. 20 G. Lucovsky, R. J. Nemanich and J. C. Knights, Phys. Rev. B: Condens. Matter Mater. Phys., 1979, 19, 2064.
22228 | Phys. Chem. Chem. Phys., 2014, 16, 22222--22228
Paper
21 M. Stutzmann and J. Chevallier, Hydrogen in Semiconductors, Elsevier Science Publishers B.V., 1991. 22 W. Beyer and H. Wagner, J. Non-Cryst. Solids, 1983, 59&60, 161. 23 W. Beyer, Sol. Energy Mater. Sol. Cells, 2003, 78, 235. 24 Z. Remes, M. Vanecek, A. H. Mahan and R. S. Crandall, Phys. Rev. B: Condens. Matter Mater. Phys., 1997, 56, R12710. ´pin, 25 M. A. El Khakani, M. Chaker, A. Jean, S. Boily, H. Pe J. C. Kieffer and S. C. Gujrathi, J. Appl. Phys., 1993, 74, 2834. 26 D. S. Kim and Y. H. Lee, Thin Solid Films, 1995, 261, 192. 27 D. B. Mawhinney, J. A. Glass Jr and J. T. Yates Jr, J. Phys. Chem. B, 1997, 101, 1202.
This journal is © the Owner Societies 2014
Published on 09 September 2014. Downloaded by Colorado College on 31/10/2014 00:28:55.
PAPER
Cite this: Phys. Chem. Chem. Phys., 2014, 16, 22222
View Journal | View Issue
Stabilization mechanism of electrodeposited silicon thin films C. Vichery,*a V. Le Nader,a C. Frantz,a Y. Zhang,b J. Michlera and L. Philippea Amorphous composite silicon thin films electrodeposited in tetrahydrofuran, containing up to 80 at% of Si and exhibiting an homogeneous dispersions of O, C and Cl in the amorphous Si matrix, have been successfully stabilized against oxidation using a post-annealing step in inert atmosphere. In order to understand the impact of the annealing step on their stabilization against oxidation, their composition and structure have been investigated upon heat treatments. It has been shown that the presence of impurities such as O, C and Cl does not have any impact on the stabilization process, which is rather linked to the presence of hydrogen in the Si composites. This conclusion has been drawn after a detailed analysis of the bonding structure of films annealed at different temperatures and dwell times by
Received 25th June 2014, Accepted 8th September 2014 DOI: 10.1039/c4cp02797c
the mean of Raman spectroscopy. It has been shown that annealing the as-deposited films at 350 1C for a couple of hours or at higher temperatures induced a hydrogen evolution, characterized by the breaking of Si–H bonds and the formation of Si–Si bonds, which stabilized the silicon network. The understanding and the reproducibility of this stabilization process of silicon thin film electrodeposited in
www.rsc.org/pccp
organic solvent paves the way for their use for many applications.
Introduction Silicon is one of the most widely used semiconductor materials for a broad range of applications in green energy (solar cells), electronics and other high-tech fields. Considering the high cost of silicon deposition via vacuum processes, significant effort has been focused on developing simple, cheap, and easily scalable alternative methods such as electrodeposition. Due to the high reactivity of silicon precursors with water and their highly negative reduction potential, silicon electrodeposition is not straightforward. It requires an inert atmosphere (typically in a glove box) and the use of perfectly anhydrous and pure reactants. In the 1980’s, electrodeposition of silicon was conducted in organic solvents such as propylene carbonate1 and tetrahydrofuran2 from silicon halides. However, pure silicon films could not be obtained and electrodeposits contained carbon, oxygen and chlorine impurities. More recently, Si has been successfully electrodeposited in ionic liquids.3,4 Unfortunately, ionic liquids are expensive, highly viscous, and not always commercially available, which may render the transfer to industry challenging. One major problem with electrodeposited silicon is the subsequent oxidation of the films when taken out of the glove box. Until now, it is not clear if the presence of oxygen is solely due to a
EMPA, Swiss Fed Labs Mat Sci & Technol, Laboratory for Mechanics of Materials & Nanostructures, CH-3602 Thun, Switzerland. E-mail: [email protected] b EMPA, Swiss Fed Labs Mat Sci & Technol, Electron Microscopy Center, ¨bendorf, Switzerland CH-8600 Du
22222 | Phys. Chem. Chem. Phys., 2014, 16, 22222--22228
this oxidation process or to the trapping of oxygen during deposition. One reported method to stabilize electrodeposited Si films, is to anneal at 350 1C for 30 min in an inert or reducing atmosphere.5 However, this has not yet proven to be reproducible.6,7 Pursuing this idea of post-synthesis heat-treatment, we successfully stabilized Si electrodeposited in dichloromethane through microwave heating at 600 1C in Ar/H2 atmosphere, although with high surface oxidation.8 Here we present a two-step synthesis, involving (i) the deposition of silicon in tetrahydrofuran (THF) and (ii) the heat treatment of the as-deposited films inside the glove box. A systematic study of the influence of the annealing parameters (temperature and dwell time) on the film composition, along with a detailed Raman analysis of the film structure will give some insight into the relationship between heat treatment and the stabilization of electrodeposited silicon films.
Experimental methods Electrodeposition Silicon thin films were synthesized at room temperature by electrodeposition in organic solvent in an argon atmosphere (H2O, O2 o 1 ppm) MBraun glove box and using an Autolab PGSTAT-101 potentiostat. Potentiostatic depositions were performed onto transparent conductive oxide (TCO) using a classical 3 electrode setup consisting of a Pt counter electrode and a Pt pseudo-reference electrode. All potentials are quoted versus Pt. The working electrode was a commercial TCO (aluminoborosilicate glass/SnO2:F) from
This journal is © the Owner Societies 2014
View Article Online
Published on 09 September 2014. Downloaded by Colorado College on 31/10/2014 00:28:55.
Paper
PCCP
Solaronix AG, Switzerland. The substrates were cleaned through 2 successive sonication steps of 15 min in acetone and isopropanol. For optimum bath stability, and film adhesion,6 a bath containing 0.3 M tetrachlorosilane (SiCl4, 99.998% Aldrich), and 0.1 M tetrabutylammonium chloride (TBACl, 97% Aldrich) as the supporting electrolyte were used in tetrahydrofuran (THF, anhydrous, inhibitor free, 99.9% Sigma Aldrich). All chemicals were used without any further purification except where mentioned. As already reported,7 we observed that only the addition of SiCl4 permitted the dissolution of TBACl in THF. The silicon films were grown under potentiostatic conditions (3.1 V vs. Pt) for 1800 s, and were afterwards rinsed in THF. Heat treatment The samples were annealed in temperatures up to 700 1C under argon atmosphere (in the glove box) with a ceramic hot plate, pre-heated at the desired temperature. Two thermocouples were used to monitor the temperature, one on the back side of the ceramic plate, and one on the top of the samples (the latter only for preliminary tests). To anneal samples at higher temperatures (750–900 1C), they were heated for a few seconds at 650 1C inside the glove box and then transferred to a tubular furnace in evacuated quartz tubes. Characterization The film morphology was characterized by scanning electron microscopy (SEM Hitachi S4800) with an acceleration voltage of 1.5 kV. Cross-sections were prepared by focused ions beam (FIB) using a dual beam FIB-SEM Tescan Vela instrument. TEM samples were prepared by scratching the film surface and dispersing the powder onto a carbon grid, and imaged using a JEOL 2200FS microscope. The films’ atomic composition was investigated via energy dispersive X-ray spectroscopy (EDX, Genesis 4000 EDAX), He-elastic recoil detection analysis (He-ERDA) and glow discharge optical emission spectroscopy (GD-OES, Horiba Jobin Yvon JY 5000 RF).9 The silicon films were also characterized by using an upright confocal Raman spectrometer (NT-MDT NTEGRA) operating with a solid state laser (532 nm) and a 100 objective. The incident light power was set to 0.5 mW in order to avoid the local crystallization of the amorphous silicon under the beam. X-ray diffraction patterns were recorded using a Bruker D8 Discovery (lCuKa = 1.5418 Å) diffractometer. In order to obtain thicker films for XRD and EDX (to minimize substrate effects), 4 cycles of deposition/annealing (650 1C, few seconds) were performed in a row on the same substrate. All the characterizations were done, at most, 30 min after taking the samples out of the glove box.
Results and discussion Silicon deposition Fig. 1a shows a cyclic voltammogram of 0.3 M SiCl4 and 0.1 M TBACl in THF recorded with a scan rate of 0.1 V s1. One single reduction peak is observed at 1.75 V vs. Pt and can be attributed to the reduction of the silicon precursor into elemental silicon. The large increase in current between 1.4 and 3.5 V
This journal is © the Owner Societies 2014
Fig. 1 Cyclic voltammogram at 0.1 V s1, (a) and chronoamperogram at 3.1 V vs. Pt (b) in THF containing 0.3 M SiCl4 and 0.1 M TBACl.
is associated with the reduction of the supporting electrolyte. Here, a deposition potential of 3.1 V vs. Pt has been chosen as a compromise between the deposition rate and the decomposition of the electrolyte. The corresponding chronoamperogram is shown in Fig. 1b. During deposition, the large initial transient recorded is due to the charging of the double layer, followed by the reduction of the initially high concentration of Si precursor at the electrode/ electrolyte interface, and the side reduction of the supporting electrolyte and solvent. However, for the rest of the transient, up to approx. 200 s, the current decrease can be explained by changes of reaction kinetics as the electrode surface is gradually covered by the electrodeposit. Indeed, Munisamy and Bard showed that the diffusion regime is never reached, even for much lower Si precursor concentrations, since the Si deposition starts to be inhibited after a Si layer of about 2 nm is formed on the substrate.6 From 200 s to approximately 800 s, the slight increase in the cathodic current might be related to the formation of cracks which thereby provide less resistive pathways through the deposit. This assumption is supported by the SEM images (Fig. 2a) which highlight a relatively high density of wide cracks. Nevertheless, no delamination was observed. Heat treatment The largest issue facing scientists working on the electrodeposition of silicon is the oxidation of the films once taken out of the glove box. Following an idea previously developed,5 electrodeposited
Phys. Chem. Chem. Phys., 2014, 16, 22222--22228 | 22223
View Article Online
Published on 09 September 2014. Downloaded by Colorado College on 31/10/2014 00:28:55.
PCCP
Paper
Fig. 3 Composition of silicon films deposited at 3.1 V for 30 min, as deposited, and annealed between 200 and 800 1C for 30 min, obtained by GD-OES, after correction from air leakage. All the values are in atomic percentage, the dashed lines are guides for the eyes.
Fig. 2 SEM (a, c) and optical images (b, d) of silicon films (B300 nm), respectively as-deposited and annealed at 700 1C for 30 min. SEM images of a top view (e) and cross-section (f) of the thick silicon film (B2 mm) annealed at 650 1C used for EDX and XRD experiments.
silicon films were annealed at different temperatures and dwell times in inert atmosphere. After annealing, it has been observed that the cracks already present in the as-deposited films extended, thus forming silicon islands (Fig. 2c). Film composition Immediately after annealing, GD-OES measurements were performed on a series of silicon films heat treated at temperatures ranging from 200 to 800 1C. The results are presented in Fig. 3. One can see that the composition stayed constant upon annealing, within the error of the measurement (4%). The high amount of carbon in the films could be explained by the decomposition of the supporting electrolyte, due to the large potential applied during the deposition process. In order to confirm these GD-OES measurements, EDX was performed. It is interesting to note (see Fig. 2e and f) that the thick film used for this characterization did not present wide cracks but a more homogeneous and porous morphology. In good accordance with GD-OES measurements, the composition obtained by EDX was (in at% 3%): 78% Si, 12% O, 6% C and 4% Cl. EDX also brought new information to light, namely the presence of about 4% Cl in the films. This is not surprising, as it is believed that the mechanism of silicon deposition involves SimCln (n/m o 4) species.10 To have a better insight in the nature of the oxygen content in the electrodeposits, a thick stabilized film was deposited onto a stainless steel substrate. Its composition has been determined by EDX before and after etching in 1% HF during
22224 | Phys. Chem. Chem. Phys., 2014, 16, 22222--22228
Table 1 Composition of a thick stabilized silicon film deposited on stainless steel, before and after HF etching, as determined by EDX
Composition (at%)
Si
O
C
Cl
Before etching After etching
70 68
14 13
12 15
4 4
45 min. As one can see in Table 1, the composition remained the same after the HF treatment. There was thus no thick silica surface layer on the silicon. The oxygen, along with the carbon and the chlorine, should therefore be homogeneously dispersed in an amorphous silicon network. To determine the origin of oxygen in the deposited films, the solvent, THF, was treated with a further drying step before deposition. This was carried in the glove box, by leaving the solvent for 4 days in a closed bottle with molecular sieves (3 Å, 20% m/v). The resulting silicon film, after heat treatment at 650 1C for 30 s, showed a lower oxygen content. From EDX, the composition was found to be: 80% Si, 7% O, 9% C and 4% Cl. A careful drying of the chemical is thus an important step if one wants to decrease the oxygen content of electrodeposited silicon films. The remaining oxygen could arise from the presence of water in the silicon precursor or in the supporting electrolyte powder and also from the decomposition of THF itself. In order to monitor the evolution of the hydrogen content upon heat treatment, He-elastic recoil detection analysis experiments were performed on 3 samples that were: (i) as-deposited, (ii) annealed at 650 1C for B30 s and (iii) annealed at 800 1C for 5 h. It was found that the atomic percentage of hydrogen decreases while increasing the heating temperature: 24 5 at% for the as deposited film, 13 3 at% for annealing at 650 1C, and 2.2 0.5 at% annealing at 800 1C. Considering the He-ERDA and
This journal is © the Owner Societies 2014
View Article Online
Published on 09 September 2014. Downloaded by Colorado College on 31/10/2014 00:28:55.
Paper
PCCP
EDX results, it is thus possible to have the approximate composition of these 3 samples: (i) Si0.59H0.24O0.09C0.05Cl0.03. (ii) Si0.68H0.13O0.10C0.05Cl0.03. (iii) Si0.76H0.02O0.12C0.06Cl0.04. The presence of such high quantity of hydrogen in the as-deposited films would likely be due to the presence of water in the electrolyte. Assuming a 4 electron process for the reduction of Si4+ to Si0, the current efficiency of the reaction (Z) can be calculated by using the following formula: Z¼
4 F mSi Q MSi
with F the Faraday constant, mSi the mass of Si in the deposited film, MSi the molar mass of silicon and Q the charge passed during electrodeposition. The current efficiency was measured using thick samples, i.e. grown over 4 cycles, in order to weight the deposit with enough accuracy, and for a composition of Si0.68H0.13O0.10C0.05Cl0.04, the efficiency was of 28%. Film stabilization The films obtained after deposition were yellow-brownish in color but turned white after a few hours when taken out of the glove box (Fig. 2b). For the heat treated samples, we observed that those annealed in the glove box at less than 400 1C remained yellowish, whereas those annealed at higher temperature turned dark brown. Only the latter samples kept their dark brown color when exposed to air (Fig. 2d). It thus seems that a heat treatment of at least 400 1C is necessary to stabilize the films against further oxidation. As the quantities of carbon, oxygen and chlorine do not change upon annealing, the stability of the films should not be linked to their presence. On the contrary, the quantity of hydrogen decreases upon heat treatment. Thus, hydrogen removal seems to be directly related to the stabilization of the films. The change in color (from yellow to brown) which goes along with the stabilization reinforces this assumption: it could be explained by the fact that the band gap of a-Si:H decreases when the hydrogen content decreases.11 As we know from He-ERDA that the quantity of hydrogen decreased upon annealing, this hydrogen evolution could lead to a reduced band gap of the silicon composite and thus to higher absorbance of visible light by the film, hence a darker color. To understand more deeply what changes happen in the structure of the films upon annealing, the series was characterized by Raman spectroscopy. The corresponding spectra are presented in Fig. 4. For crystalline silicon, the sharp strong peak observed at 520 cm1 in the spectrum is due to a triply degenerate first order transverse optical (TO) phonon.12,13 When the disorder in the silicon structure increases, the momentum selection rules is relaxed and new phonon modes are allowed. For instance in Fig. 4, the silicon thin films which were annealed at 800 and 900 1C clearly behave as nano-crystalline films, i.e. as films consisting of a dispersion of nano-sized Si crystallites (peak at 521 cm1) in an amorphous Si matrix (peak at 480 cm1).12,14,15 The presence of both amorphous and crystalline silicon can also be observed by X-ray diffraction (Fig. 5c) and high resolution
This journal is © the Owner Societies 2014
Fig. 4 Raman spectra of silicon thin films annealed at different temperatures for 30 min. Laser wavelength: 532 nm, spectral resolution: 2.8 cm1, objective magnification: 100, power B0.5 mW. In inset, a magnified view of the room temperature, 200 and 300 1C heat treated samples spectra at low energy.
Fig. 5 High resolution (a) and dark field (b) TEM images of a film annealed at 850 1C for 1 h. In inset, the corresponding selected-area diffraction pattern, a white circle indicating the position of the objective lens aperture for the latter image. (c) X-ray diffraction pattern of a thick film (B2 mm) annealed at 800 1C for 15 h and of the substrate (the peaks correspond to SnO2). Inset is a magnification to highlight the Si(111) peak, marked with a star.
TEM and dark field images (Fig. 5a and b). One can see in the XRD pattern the presence of a small peak which can be attributed to the (111) plane family of silicon. The corresponding
Phys. Chem. Chem. Phys., 2014, 16, 22222--22228 | 22225
View Article Online
Published on 09 September 2014. Downloaded by Colorado College on 31/10/2014 00:28:55.
PCCP
Paper
coherence length, as determined by the Scherrer formula, is 6.9 0.6 nm. The dark field image reveals the presence of monocrystalline nanoparticles (bright spots) between 3 and 10 nm in diameter dispersed in an amorphous matrix. High resolution images in bright field also show the presence of these nanoparticles as one can observe lattice-fringes in Fig. 5a. The size of these mono-crystalline nanoparticles is in good agreement with the coherence length of silicon. In the Raman spectra, the increase in disorder in the silicon structure leads first to an increase of the full width at half maximum (FWHM) of the peak at 521 cm1.14,16,17 Then, the break-down of the zero momentum selection rule leads to additional contributions: new vibrational modes such as higher order TO, longitudinal optic (LO), transversal acoustic (TA) and longitudinal (LA) acoustic modes contribute to the spectrum.15,18,19 As shown in Fig. 4, for the films annealed between 400 and 700 1C for 30 min, the contribution of the first order TO is not present while the contributions of the other TO, LO, TA and LA modes are observed. In these spectra, the broad peak centered around 640 cm1 corresponds to the wagging modes of Si–H2 and Si–H3 and highlights the presence of hydrogen.18,20 These films are thus clearly made of a-Si:H. In order to link the presence of hydrogen to the stability of the electrodeposited a-Si:H films, the spectra have been deconvoluted by using Gaussian functions to extract each contribution. An example of deconvolution is given in Fig. 6. In this work the hydrogen content is evaluated by the ratio AðSiHx Þ AðSiH2 Þ þ AðSiH3 Þ ¼ AðSiÞ AðSiðLAÞÞ þ AðSiðLOÞÞ þ AðSiðTOÞÞ where A(x) is the area of the peak x. Fig. 7a illustrates the evolution of the hydrogen content versus the annealing temperature and the annealing time. With respect to the temperature dependency, there is a threshold value around 350 1C for which the ratio A(SiHx)/A(Si) decreases from B1 to B0. From this first series, it is clear that there is a major change in the film structure, linked to their stabilization, after a heat
Fig. 6 Deconvolution with Gaussian functions of the Raman spectrum of a a-Si:H thin film annealed at 400 1C for 30 minutes. Before applying the deconvolution process, the fluorescence background has been subtracted.
22226 | Phys. Chem. Chem. Phys., 2014, 16, 22222--22228
Fig. 7 (a) Variation of the A(SiHx)/A(Si) ratio versus annealing temperature for samples annealed for 30 minutes. (b) Variation of the A(SiHx)/A(Si) ratio versus annealing time for an annealing temperature of 350 1C.
treatment of around 350 1C. To determine more accurately the dwell time at which this transition occurs when heat treated at 350 1C, thin films were annealed for times ranging from 10 min to 15 h. The corresponding Raman spectra are presented in Fig. 8. One can clearly observe hydrogen removal with increasing annealing duration. As before, the Raman spectra have been deconvoluted. The resulting A(SiHx)/A(Si) ratio is plotted in Fig. 7b: a dwell time of about 2 h is necessary in order to break Si–H bonds. One can see between the elemental composition evolution (Fig. 3) and the analysis of Raman spectra (Fig. 7a), that the Si–H bonds have been broken after heat treatment at temperatures higher than 350 1C. This result is consistent with what has been previously reported in the literature. It is worth mentioning that similar annealing strategies are used routinely in the semiconductor community to remove hydrogen from silicon.21 Hydrogen is considered to desorb from a-Si:H thin films by simultaneous breaking of two Si–H bonds and formation of H2, the latter getting out of the film by rapid diffusion through voids.22,23 This process of hydrogen release allows the formation of new Si–Si bonds,24 or in the case of a-SiC:H, Si–C bonds,25 inducing a longrange relaxation of the silicon network. This interpretation is also supported by the variation of the A(Si(TA))/A(Si(TO)) ratio which is related to the intermediate-range order.15 Indeed for a-Si:H thin films in which the hydrogen desorption process has not yet started, the wagging vibrational Si–Hx modes are
This journal is © the Owner Societies 2014
View Article Online
Published on 09 September 2014. Downloaded by Colorado College on 31/10/2014 00:28:55.
Paper
PCCP
successively electrodeposited in tetrahydrofuran and stabilized against oxidation through a heat treatment step in inert atmosphere. It has been shown that a heat treatment of 2 h at 350 1C in Ar induced the breaking of Si–H bonds and the formation of Si–Si bonds, leading to a better cross-linkage and a relaxation of the silicon network, and thus to the stabilization of the films. Heating the films at temperatures higher than 800 1C induced the formation of Si crystallites of less than 10 nm in the amorphous matrix. This study highlighted that the presence of impurities such as O, C and Cl was not the reason for the instability of electrodeposited silicon films, but that it was rather linked to a high amount of hydrogen. The stabilization mechanism of silicon films electrodeposited in organic solvent being now understood and reproducible, such an easy and scalable process paves the way of using these films for many applications, such as for Li-ion batteries or photovoltaics.
Acknowledgements Fig. 8 Raman spectra of silicon thin films annealed at 350 1C for different dwell times. Laser wavelength: 532 nm, spectral resolution: 2.8 cm1, objective magnification: 100, power B0.5 mW.
predominant and no TA contribution is observed in the spectra.18 In this state, the Si network is strongly disorganized. When the desorption process starts, a more organized amorphous silicon network is created because of the formation of new Si–Si bonds. Consequently the LO, TO and TA mode contributions can be observed in the spectra. One can notice in Fig. 4 and 8 that the ratio A(Si(TA))/A(Si(TO)) decreases with annealing temperature and time. As long as the hydrogen desorption proceeds, the silicon network increases its degree of order and the number of defects decreases, leading to a structural rearrangement, from amorphous to crystalline. This structural rearrangement upon annealing induces a decrease in compressive stress,25,26 which could explain the shrinking and cracking of the film observed after heat treatment. The structural relaxation and the better cross-linkage of silicon after hydrogen removal is certainly the reason of the stability of the annealed films. It is furthermore important to point out that this annealing step must be carried out in inert atmosphere. Indeed, if some oxygen is present, it will be readily decomposed to produce Si–O bonds because of the formation of highly reactive dangling bonds after the breaking of Si–H bonds. Thus, it would result in the oxidation of the films.26,27 The stability over time of the stabilized silicon thin films has been confirmed by re-measuring the Raman spectra of the samples after 3 months storage in air. The spectra remained unchanged, and even after more than 9 months in air, the samples have remained dark brown in color (Fig. 2d).
Conclusions Silicon composite films about 300 nm thick, and containing up to 80 at% Si with O, C and Cl as impurities have been
This journal is © the Owner Societies 2014
The authors would like to acknowledge Dr M. Doebeli (ETH Zurich, Switzerland) for the He-ERDA measurements, M. Mieszala (EMPA) for the FIB cuts, D. Frey (EMPA) for the GD-OES analyses and Dr P. Dunne for helping reviewing this manuscript.
Notes and references 1 A. K. Agrawal and A. E. Austin, J. Electrochem. Soc., 1981, 128, 2292. 2 J. Gobet and H. Tannenberger, J. Electrochem. Soc., 1988, 135, 109. 3 S. Z. El Abedin, N. Borissenko and F. Endres, Electrochem. Commun., 2004, 6, 510. 4 J. Mallet, M. Molinari, F. Martineau, F. Delavoie, P. Fricoteaux and M. Troyon, Nano Lett., 2008, 8, 3468. 5 E. R. Bucker and J. A. Amick, US Pat., 204/14N,60,37R,385, 1980. 6 T. Munisamy and A. J. Bard, Electrochim. Acta, 2010, 55, 3797. 7 J. P. Nicholson, J. Electrochem. Soc., 2005, 152, C795. 8 M. Bechelany, J. Elias, P. Brodard, J. Michler and L. Philippe, Thin Solid Films, 2012, 520, 1895. 9 Th. Nelis, M. Aeberhard, M. Hohl, L. Rohr and J. Michler, J. Anal. At. Spectrom., 2006, 21, 112. 10 Y. Nishimura, T. Fukunaka, C. R. Miranda, T. Nishida, T. Nohira and R. Hagiwara, ECS Trans., 2009, 16, 1. `re, A. Deneuville, A. Mini, J. Fontenille and 11 J. C. Bruye R. Danielou, J. Appl. Phys., 1980, 51, 2199. 12 Z. Iqbal and S. Veprek, J. Phys. C: Solid State Phys., 1982, 15, 377. 13 H. Richter, Z. P. Wang and L. Ley, Solid State Commun., 1981, 39, 625. 14 W. Li, D. Xia, H. Wang and X. Zhao, J. Non-Cryst. Solids, 2010, 356, 2552. 15 Z. Li, W. Li, Y. Jiang, H. Cai, Y. Gong and J. He, J. Raman Spectrosc., 2011, 42, 415.
Phys. Chem. Chem. Phys., 2014, 16, 22222--22228 | 22227
View Article Online
Published on 09 September 2014. Downloaded by Colorado College on 31/10/2014 00:28:55.
PCCP
16 S. K. Gupta and P. K. Jha, Solid State Commun., 2009, 149, 1989. 17 G. Faraci, S. Gibilisco, P. Russo, A. R. Pennisi and S. La Rosa, Phys. Rev. B: Condens. Matter Mater. Phys., 2006, 73, 033307. 18 S. Miyazaki, N. Fukuhara and M. Hirose, J. Non-Cryst. Solids, 2000, 266–269, 59. 19 C. Smit, R. A. C. M. M. van Swaaij, H. Donker, A. M. H. N. Petit, W. M. M. Kessels and M. C. M. van de Sanden, J. Appl. Phys., 2003, 94, 3582. 20 G. Lucovsky, R. J. Nemanich and J. C. Knights, Phys. Rev. B: Condens. Matter Mater. Phys., 1979, 19, 2064.
22228 | Phys. Chem. Chem. Phys., 2014, 16, 22222--22228
Paper
21 M. Stutzmann and J. Chevallier, Hydrogen in Semiconductors, Elsevier Science Publishers B.V., 1991. 22 W. Beyer and H. Wagner, J. Non-Cryst. Solids, 1983, 59&60, 161. 23 W. Beyer, Sol. Energy Mater. Sol. Cells, 2003, 78, 235. 24 Z. Remes, M. Vanecek, A. H. Mahan and R. S. Crandall, Phys. Rev. B: Condens. Matter Mater. Phys., 1997, 56, R12710. ´pin, 25 M. A. El Khakani, M. Chaker, A. Jean, S. Boily, H. Pe J. C. Kieffer and S. C. Gujrathi, J. Appl. Phys., 1993, 74, 2834. 26 D. S. Kim and Y. H. Lee, Thin Solid Films, 1995, 261, 192. 27 D. B. Mawhinney, J. A. Glass Jr and J. T. Yates Jr, J. Phys. Chem. B, 1997, 101, 1202.
This journal is © the Owner Societies 2014
