Solar Cells: From Sunlight into ElectricityView this Special Issue
Development of Hydrogenated Microcrystalline Silicon-Germanium Alloys for Improving Long-Wavelength Absorption in Si-Based Thin-Film Solar Cells
Hydrogenated microcrystalline silicon-germanium (μc-:H) alloys were developed for application in Si-based thin-film solar cells. The effects of the germane concentration and the hydrogen ratio on the μc-:H alloys and the corresponding single-junction thin-film solar cells were studied. The behaviors of Ge incorporation in a-:H and μc-:H were also compared. Similar to a-:H, the preferential Ge incorporation was observed in μc-:H. Moreover, a higher significantly promoted Ge incorporation for a-:H, while the Ge content was not affected by in μc-:H growth. Furthermore, to eliminate the crystallization effect, the 0.9 μm thick absorbers with a similar crystalline volume fraction were applied. With the increasing , the accompanied increase in Ge content of μc-:H narrowed the bandgap and markedly enhanced the long-wavelength absorption. However, the bias-dependent EQE measurement revealed that too much Ge incorporation in absorber deteriorated carrier collection and cell performance. With the optimization of and , the single-junction μc-:H cell achieved an efficiency of 5.48%, corresponding to the crystalline volume fraction of 50.5% and Ge content of 13.2 at.%. Compared to μc-Si:H cell, the external quantum efficiency at 800 nm had a relative increase by 33.1%.
Hydrogenated amorphous silicon (a-Si:H) has been widely studied [1, 2] and employed as an absorber in silicon thin-film solar cells  because of its high absorption coefficient in the visible range of the solar spectrum and the feasibility of large area deposition. However, the solar spectrum is distributed from ultraviolet to near-infrared (IR) region. The bandgap of approximately 1.75 eV  for a-Si:H limits the absorption in IR region. On the concept of light absorption, only the photons having the energies larger than the bandgap of absorbers can contribute to photoexcited carriers . For effective use of the low-energy photon in the solar spectrum, the development of a lower-bandgap material is important. Accordingly, the integration of lower-bandgap material and the concept of spectrum splitting have been applied as multijunction thin-film solar cells for allowing more efficient use of solar spectrum. Compared to single-junction solar cell, the multijunction cell generally has a broadened and effective spectral response. The more efficient light absorption is attributed to the component cells with different bandgap absorbers, which leads to a higher cell efficiency. Yunaz et al. have demonstrated a potential efficiency over 20% by using AMPS-1D simulation for the Si-based multijunction thin-film solar cell . Other groups have integrated a-Si:H and hydrogenated microcrystalline silicon (μc-Si:H) absorbers into tandem structure cells with a stabilized efficiency over 10% [7–9]. Moreover, Yan et al. have reported an a-Si:H/a-SiGe:H/μc-Si:H triple-junction cell reached a recorded efficiency of 16.3% .
Due to a lower bandgap of 1.1 eV , μc-Si:H has been utilized as an absorber for IR absorption [11–14]. In addition, μc-Si:H has a minor Staebler-Wronski effect (SWE) , which has less impact on the long term film quality and cell performance than amorphous material. Nevertheless, the indirect bandgap nature of μc-Si:H leads to a low absorption coefficient. Therefore, a thick μc-Si:H absorber is usually needed to obtain adequate IR absorption. Matsui et al. have reported that the Ge incorporation in microcrystalline silicon network led to a bandgap narrowing and an increase in IR absorption, with the consequence of a thinner μc-:H absorber in the cells [15–17]. The μc-:H consists of an amorphous-crystalline mixed phase of binary SiGe alloys, which are affected by the deposition parameters including the hydrogen ratio and the germane concentration . The addition of Ge to Si network not only lowers the bandgap, but could also reduce the crystallization of the films. The crystalline volume fraction can not only influence the electrical properties including bandgap and carrier collection, but also change the optical absorption. The trade-off between crystallization and Ge incorporation of μc-:H alloys should be carefully manipulated for the requirement of IR absorption.
Previous works on μc-:H alloy [18, 19] have reported the effect of Ge incorporation by varying but have not yet considered the accompanied variation of crystallization. In this work, to eliminate the effect of different degree of crystallization, the μc-:H absorber with a similar crystalline volume fraction was applied to indeed discuss the effect of Ge content on cell performance. Furthermore, we compared the behaviors of the Ge incorporation in a-:H and μc-:H alloys. The effects of and on Ge incorporation were discussed.
2. Experimental Detail
Silicon thin films including μc-:H were deposited by a single-chamber process in a multichamber plasma-enhanced chemical vapor deposition (PECVD) system equipped with 27.12 MHz rf power, NF3 in situ plasma cleaning, and a load-lock chamber. The films were prepared on Corning EAGLE XG glass substrate at approximately 200°C. A gas mixture of highly H2-diluted SiH4 and GeH4 was introduced to deposit μc-:H thin films. The , defined as [H2]/[SiH4], was varied from 71.4 to 123.0. The , defined as [GeH4]/[GeH4 + SiH4], was changed from 0 to 6.8%. In contrast, the lower varied from 0 to 6 and the varied from 8.3% to 16.7% were employed for a-:H deposition. The film Ge content was calculated by the integrated intensities of Ge3d and Si2p core lines using the quantitative X-ray photoelectron spectroscopy (XPS) analysis [20–22]. A presputtering was conducted to eliminate contaminations and native oxides on the film surface. We have found in our previous work that the Ge content would have variation in the incubation layer. This incubation region (approximately 0.1 μm) occupied only small part of the absorbing layer (~0.9 μm). The measured Ge content shown in the paper should be representative for the absorbing layer. The crystalline volume fraction was estimated from Raman spectra, which were obtained from a high-resolution confocal Raman microscope with an excitation laser at a wavelength of 488 nm. The dark and photocoplanar conductivities of the prepared films were obtained by an - measurement system equipped with an AM1.5G illumination. A spectrophotometer was used to determine the transmittance and the reflectance of the films. The optical bandgap was obtained when the absorption coefficient is 104 cm−1.
The commercial textured SnO2:F-coated substrates were utilized for preparing superstrate p-i-n μc-:H cells. A 0.9 μm thick μc-:H absorber was employed in single-junction solar cells with a p-type μc-Si:H layer and an n-type hydrogenated microcrystalline silicon oxide (μc-:H) layer. The cell was characterized by an AM1.5G solar simulator. The area of the device for measurement was 0.25 cm2 which was defined by the silver electrode. A measuring system having monochromator, chopper, lock-in amplifier, and - meter was applied to measure the external quantum efficiency (EQE).
3. Results and Discussion
3.1. Ge-Incorporation in Amorphous and Crystalline Silicon-Germanium Alloys
The dependence of Ge content ([Ge]) on with different in amorphous and microcrystalline SiGe alloys is shown in Figure 1(a). As can be seen, the Ge content in a-:H alloys rapidly increased as increased from 0 to 2 at a fixed and tended to saturate as was larger than 2. The phenomenon suggested that the hydrogen atoms promoted Ge incorporation in the amorphous network . One possible reason may relate to the sticky nature of GeH3 species more than the SiH3 species. The diffusion length of GeH3 species is less than SiH3 species during the growth of SiGe alloy , which makes it more difficult to reach the energetically favorable sites on the film surface. As a result, Ge is easier to form weak bonds than Si in SiGe binary network. When the atomic hydrogen is sufficient in plasma, a high H-coverage growth surface and local heating lead to well-relaxed network [25–27]. Thus, rigid Ge-related bonds increase as increasing hydrogen. Accordingly, more Ge atoms can be left in the films.
In high hydrogen-containing gas mixture with over 2, the saturation of Ge content was observed for a-:H alloys. Presumably, the sufficient hydrogen atoms promote rigid Ge bonding in the films. Compared to a-:H alloys, a much higher hydrogen diluted gas mixture is needed for the crystallization of the μc-:H. When the was over 85 at a fixed , Ge content was not significantly changed, suggesting that the effect of hydrogen for Ge incorporation in the μc-:H films has less impact. The resulting Ge content in the μc-:H film with increasing was kept at approximately 13 and 16.7 at.%, with of 5.0% and 7.1%, respectively.
In addition to the Ge content, the incorporation efficiency of Ge was also discussed. The incorporation efficiency represents the ratio of the transformation from GeH4 to film Ge content, defined as [Ge]/. As shown in Figure 1(b), the tendency of incorporation efficiency of a-:H and μc-:H films was similar to that of the film Ge content with the increasing . The Ge incorporation efficiency was larger than one in both amorphous and microcrystalline SiGe alloys. This suggests that Ge was preferentially incorporated into films more than Si. The incorporation efficiency over 1 also indicates that the change of alters the Ge content significantly, as well as the film characteristics. One of the reasons was the less dissociation energy of GeH4 compared to SiH4. The more efficient decomposition of GeH4 was known from SiH4-GeH4-H2 discharge plasma field . However, adding more GeH4 decreased the Ge incorporation efficiency. More produced sticky GeH3 precursors led to an increase in the weak Ge-related bonds [29, 30]. Consequently, under the hydrogen-containing atmospheres, the probability of the SiH3 replacement on a weak Ge-bonded site may be enhanced, which reduced the effective Ge incorporation.
In short, the preferential incorporation of Ge in SiGe alloys was observed. Compared to high environment, the Ge content in SiGe alloys was affected by the hydrogen significantly in low environment. More Ge content can be achieved by adding more GeH4 in the gas mixture. Nevertheless, with increasing Ge content, the incorporation efficiency of Ge into solid phase decreased with increasing .
3.2. Effect of the Hydrogen Ratio on Film Properties and Cell Performance
The microstructure of μc-:H films deposited with different at of 5% was studied by the Raman spectroscopy. Figure 2 shows the resulting Raman spectra, where the transverse optical (TO) modes mainly consisted of amorphous, intermediate phase and crystalline Si-Si networks . The TO mode of amorphous Si-Si network is distributed as a Gaussian function at 480 cm−1. This is attributed to the Si-Si network in short-range order. The full width of half maximum and the Raman shift of a-Si phase are related to the variation of bonding angle of a-Si network [32, 33]. For the narrow c-Si Lorenzian peak, the TO mode is at 520 cm−1. When the c-Si grain becomes as small as few nanometers in a crystalline-to-amorphous transition region, the Raman shift of c-Si peak decreases because of momentum conservation [34, 35]. The peak of intermediate phase is in a Raman shift ranging approximately from 490 to 510 cm−1. This is ascribed to the defective part of the Si-Si crystallines, which include small size crystallite, bond dilation at grain boundaries, or a silicon wurtzite phase consisting of twins [36, 37]. When the increased from 83.5 to 120.3, more crystalline phase is accompanied with less amorphous phase. However, the resulting c-Si peak constantly appeared near 512 cm−1 as increasing . In previous work [17, 38, 39], when Ge presents nearby the crystallites, the c-Si peak has a red-shift. In addition, the increased Ge content was in a linear correlation with decreasing c-Si peak. As mentioned in Section 3.1, Ge content was unchanged in the μc-:H films at a fixed . The higher degree of crystallization at a higher is contributed to more crystallites in the films. In addition, there was no significant difference in Raman spectra at approximately 300 cm−1 for μc-SiGe:H samples. This may be due to a low Ge content used in this study, which contributed to negligible Ge-Ge TO mode signal from the crystal phase .
Effect of on and optical bandgap is shown in Figure 3. The crystalline volume fraction is defined by , where , , and were the integrated intensities of crystalline, intermediate, and amorphous phase, respectively [41, 42]. With a kept , the increased with increasing . More H2 in the gas mixture promoted the crystallization of μc-:H growth. Moreover, given the same , the required for μc-:H was much larger than that for μc-Si:H. This suggests that adding GeH4 significantly suppressed crystalline growth. This should be due to the distorted Si network by incorporating Ge, and more Ge-induced defects in the film, which needs more H-atom to be eliminated. When was varied from 83.5 to 124.1 and was kept at 5%, the increased from 25.2% to 70.6%, corresponding to the decreased from 1.93 to 1.87 eV. The more crystalline phase led to a narrower bandgap, which shifted light absorption to IR. To investigate the effect of of μc-:H absorbers on cell performance, we further employed different μc-:H alloys as absorbers by changing the .
Figure 4 shows the cell structure and the - characteristics of μc-:H p-i-n single-junction cells using absorbers prepared with different . This cell performance is shown in Table 1. Accompanied with the increasing from 88.6 to 124.1, the resulting bandgap narrowing of the absorber influenced the internal electric field and decreased the from 485 to 430 mV. On the contrary, the was significantly enhanced from 17.17 to 19.25 mA/cm2. More crystalline phase in the film contributed to more photocurrent in the cells due to the lower bandgap. When the was 94.9, the corresponding of the absorber was 50.5% which led to an optimal cell efficiency of 5.48%.
3.3. Effect of the Germane Concentration on Film Properties and Cell Performance
In Section 3.1, we have shown that the significantly changed the Ge content in the film. To reveal the effect of on cell performance is therefore important for improving long-wavelength absorption. The μc-:H absorbers in single-junction solar cells were prepared with different of 0, 3.7%, 5.0%, and 6.8%. In addition, the μc-:H absorber with a similar of approximately 55% was applied to eliminate the effect of the crystallization of absorber on the cell performance. When the increased from 0 to 5.0%, the film Ge content increased from 0 to 13.2 at.%, as shown in Table 2. As a result, the bandgap decreased from 1.96 to 1.85 eV, corresponding to a reduction in of 90 mV. The worsened FF from 71.0% to 59.3% may be due to the more Ge-related defects created in the absorber with increasing Ge incorporation. With more Ge incorporation which reduced the bandgap of the absorber, the significantly increased from 17.38 to 18.50 mA/cm2 due to more optical absorption. When the was 6.8%, the film Ge content further went up to 18.0 at.%, which resulted in the degraded cell performance. The , FF, and decreased to 370 mV, 53.0%, and 17.27 mA/cm2, respectively.
The improvement of according to the change of Ge content can be revealed by the EQE measurement. As shown in Figure 5, no significant drop in spectral response in short-wavelength region was observed as the increased from 0 to 5%, while the spectral response in the range of 600–1100 nm was enhanced. The external quantum efficiency at 800 nm increased from 26.6% to 35.4%. This relative increase of 33.1% in spectral response suggested that Ge incorporation effectively enhances the optical absorption in the infrared region. However, the red-to-IR response reduced as the absorber was prepared with of 6.8%. Too much Ge incorporation could degrade the transport of carriers generated in the long-wavelength region, which will be discussed in the next section. Besides, when the was 6.8%, the μc-:H absorber near p/i interface may preferentially grow in amorphous phase. Compared to microcrystalline phase, amorphous phase generally has higher short-wavelength absorption. As a result, the increase in the spectral response range of 300–500 nm was observed.
The results of EQE measurement for the μc-:H cells having absorber prepared with of 5.0% and 6.8% were presented in Figure 6. The spectral response was measured under 0 and −2 bias voltages to reveal the difference in carrier transport. If a reverse voltage bias of −2 V was applied to the device, the electric built-in field can be enlarged and the photogenerated carriers trapped by the defects can be driven out. If the cell having defects was measured with the reverse bias, the spectral response would be enlarged. For the μc-:H cell employing the absorber prepared by of 6.8%, the difference of as measured by EQE with 0 and −2 bias voltages was 1.05 mA/cm2. In comparison, the difference of for μc-:H cell employing absorber prepared with of 5.0% under the same bias voltages was less than 0.25 mA/cm2. The result indicates that too much Ge incorporation would lead to the degraded carrier collection and worsen cell performance. Moreover, in contrast to the photogenerated electrons, the holes generated by long-wavelength photons near back contact would drift toward longer distance. The change in spectral response was presumably due to the degraded hole collection .
The effects of and on μc-:H alloys and the corresponding single-junction cells were studied. Similar to a-:H, the preferential Ge incorporation was observed in μc-:H. Moreover, a higher significantly promoted Ge incorporation for a-:H, while the Ge content was not affected by in μc-:H growth. To eliminate the crystallization effect, the 0.9 μm thick absorbers with a similar crystalline volume fraction were applied. With the increasing , the accompanied increase in Ge content of μc-:H narrowed the bandgap and edly enhanced the long-wavelength absorption. When the increased from 0 to 5%, the spectral response at 800 nm was significantly improved from 26.6% to 35.4%, which was a relative increase by 33.1%. However, the bias-dependent EQE measurement revealed that too much Ge incorporation in absorber deteriorated carrier collection and cell performance. With the optimization of and , the single-junction μc-:H cell achieved an efficiency of 5.48%, corresponding to the crystalline volume fraction of 50.5% and Ge content of 13.2 at.%. Future work will include the application of μc-:H absorbers in the tandem cell structure.
Conflict of Interests
The authors declare that there is no conflict of interests regarding the publication of this paper.
This work was sponsored by National Science Council in Taiwan under Contract no. NSC-102-3113-P-008-001 and no. NSC-2221-E-009-122.
M. A. Green, Solar Cells: Operating Principles, Technology and System Applications, University of New South Wales, Sydney, Australia, 1998.
J. Meier, S. Dubail, R. Flückiger, D. Fischer, H. Keppner, and A. Shah, in Proceedings of the 1st World Conference on Photovoltaic Energy Conversion (WCPEC '94), p. 409, Waikoloa, Hawaii, USA, 1994.
B. Yan, G. Yue, L. Sivec, J. Yang, S. Guha, and C.-S. Jiang, “Innovative dual function nc-SiOx:H layer leading to a16% efficient multi-junction thin-film silicon solar cell,” Applied Physics Letters, vol. 9, Article ID 113512, 2011.View at: Google Scholar
T. Matsui, M. Kondo, K. Ogata, T. Ozawa, and M. Isomura, “Influence of alloy composition on carrier transport and solar cell properties of hydrogenated microcrystalline silicon-germanium thin films,” Applied Physics Letters, vol. 89, Article ID 142115, 2006.View at: Google Scholar
T. Matsui, C. W. Chang, T. Takada, M. Isomura, H. Fujiwara, and M. Kondo, “Microcrystalline Si1−xGex solar cells exhibiting enhanced infrared response with reduced absorber thickness,” Applied Physics Express, vol. 1, no. 3, Article ID 031501, 2008.View at: Google Scholar
W. Paul, D. K. Paul, B. von Roedern, J. Blake, and S. Oguz, “Preferential attachment of H in amorphous hydrogenated binary semiconductors and consequent inferior reduction of pseudogap state density,” Physical Review Letters, vol. 46, no. 15, pp. 1016–1020, 1981.View at: Publisher Site | Google Scholar
R. L. C. Vink, G. T. Barkema, and W. F. Van Der Weg, “Raman spectra and structure of amorphous Si,” Physical Review B, vol. 63, no. 11, Article ID 115210, pp. 1152101–1152106, 2001.View at: Google Scholar
M. Luysberg, P. Hapke, R. Carius, and F. Finger, “Structure and growth of hydrogenated microcrystalline silicon: Investigation by transmission electron microscopy and Raman spectroscopy of films grown at different plasma excitation frequencies,” Philosophical Magazine A, vol. 75, no. 1, pp. 31–47, 1997.View at: Google Scholar
T. Kaneko, M. Wakagi, K. Onisawa, and T. Minemura, “Change in crystalline morphologies of polycrystalline silicon films prepared by radio-frequency plasma-enhanced chemical vapor deposition using SiF4+H2 gas mixture at 350°C,” Applied Physics Letters, vol. 64, p. 1865, 1994.View at: Google Scholar