Session Th-C2

a-Si: Electronic Structure, Defects, Metastability II

Chair: Rana Biswas, Ames Laboratory and Microelectronics Center

Th-C2.1 10:30–10:50

Spatial Defect Creation Profile in a-Si:H Solar Cells Following Light-induced Degradation

R. A. C. M. M. van Swaaij, K. S. Oppedal, J. Melskens, A. H. M. Smets, and M. Zeman

Photovoltaic Materials and Devices, Delft University of Technology, P. O. Box 5031, 2600 GA Delft, the Netherlands

In this contribution we study the effects of light-induced degradation on the dark current-voltage (J-V) characteristics and the external quantum efficiency (EQE) of hydrogenated amorphous silicon (a-Si:H) solar cells, aiming at unraveling the spatial defect creation profile (DCP). It is generally accepted that the performance decrease of these solar cells results from a substantial increase of the subgap density of states associated to defect structures in a-Si:H following sustained illumination due to the Staebler-Wronski effect (SWE). These defects facilitate Shockley-Read-Hall recombination, inevitably determining the current-voltage characteristics of the solar cells. Research on the SWE has generally focused on the time evolution of the defect density and on the defect structures that result in additional energy states in the band gap. However, in order to understand the operation of light-degraded a-Si:H solar cells not only the energy position in the band gap of a defect is relevant, but also the physical position in the device where the defect is formed.

The voltage dependence of the ideality factor, n(V), of dark forward bias characteristics reflects the density-of-state distribution at the positions where the recombination rate peaks in the device. On the other hand, the dark reverse bias current is determined by thermal generation via bulk subgap states and the associated activation energy is approximately equal to half the band gap. Therefore, changes in the dark J-V characteristics reflect variations in the defect density profile.

For the degradation experiments a set of three a-Si:H solar cells with varying i-layer thickness was light soaked at a temperature of 25°C in open-circuit condition. During the degradation the performance of all solar cells was measured in situ. Solar cells were partially recovered by annealing at 130°C and the performance, EQE, and dark JV were measured at given time intervals. The dark JV characteristics were measured at temperatures ranging from 30 to 100°C. After partial recovery a stable degraded state was created that could be studied by temperature dependent dark J-V measurements.

When comparing the n(V) before degradation and in the degraded state, we found that n(V) was temperature independent and very similar for solar cells with 900 nm thick absorber layers, suggesting no large changes in the defect density. For 300 nm solar cells a large increase was observed at low forward biases. Similarly, the dark current activation energy was unaffected for the thick cell, but dropped substantially for the thin solar cell. Interestingly, EQE variations were smallest for the 300 nm, and largest for the 900 nm solar cells, mainly in the blue.

The absence of a change in n(V) after degradation indicates that the defect density is largely unaffected at the position where the dark recombination profile peaks in the device. In a-Si:H solar cells this profile is commonly assumed to have a maximum near the p-i interface. However, the EQE drop for wavelengths shorter than 600 nm suggests the defect density increases deeper in the device. We will show that the DCP is spatially inhomogeneous.

Keywords: Staebler-Wronski effect, light soaking, hydrogenated amorphous silicon, defects, spatial defect profile

Th-C2.2 10:50–11:10

Revealing the Complexity of the Staebler-Wronski Effect in Hydrogenated Amorphous Silicon Films and Solar Cells

J. Melskens, M. Schouten, S. W. H. Eijt, H. Schut, A. Mannheim, M. Zeman, and A. H. M. Smets

Photovoltaic Materials and Devices, Delft University of Technology, P. O. Box 5031, 2600 GA Delft, the Netherlands

Due to the complexity of the nanostructure of hydrogenated amorphous silicon (a-Si:H) there is no consensus on the nature of defects in a-Si:H. The limited understanding of the nanostructure and defect states has led to a wide variety of models that try to describe the Staebler-Wronski effect (SWE). However, none of these models have yet succeeded in providing convincing experimental evidence for their correctness. On the road towards SWE reduction we therefore present the latest insights in the a-Si:H nanostructure and the SWE using Doppler broadening positron annihilation spectroscopy (DB-PAS), Fourier transform infrared spectroscopy (FTIR), and Fourier transform photocurrent spectroscopy (FTPS) with in situ light soaking mode.

First, we study the as-deposited state of various a-Si:H materials. The effects of different variations in the plasma deposition conditions, such as the hydrogen-to-silane gas flow rate ratio (R), the deposition pressure and rf power, are shown to result in different dominant types of volume deficiencies. Further, it is shown that increased values of R lead to smaller sizes of the dominant volume deficiencies in the material accompanied with a higher hydrogen (H) passivation degree in vacancies. As a-Si:H deposited at higher values for R appears to be more stable against light soaking, the enhanced H passivation degree of vacancies is suggested to play a crucial role in SWE reduction. Additionally, annealing is used to study the a-Si:H nanostructure by monitoring the changes in the distribution of volume deficiencies. It is shown that divacancies agglomerate into larger vacancies from 200°C up to 400°C. The vacancy agglomeration rate is slower for R>0 a-Si:H materials. This indicates that a reduced vacancy mobility appears to be desirable when trying to reduce the SWE. Finally, FTIR spectroscopy and FTPS show that Si-H bonds are being broken above 300°C while electronic defects are being formed, which further confirms that H passivation of vacancies is linked with electronic defect passivation.

Secondly, sub gap absorption measurements using FTPS suggest the existence of at least four defect distributions in the a-Si:H band gap. This shows that bulk coordination defects, i.e. isolated dangling bonds (DBs), are unlikely to be the solely dominant defects in a-Si:H, since DBs are believed to have no more than three singly charged states (1+/0/1–). Not fully H-passivated vacancies can however have both single and double charges, which shows that nanostructure models in which defect configurations are related to vacancies and DBs might describe the dominant defect states in a-Si:H more accurately.

During light soaking the four observed defect distributions are shown to increase in density, first in a "fast" regime and then in a 'slow' regime where saturation of the defect density eventually sets in. The four light induced defect distributions increase exponentially with time ~tβ in the fast regime, where the scaling exponent β varies with R, ambient temperature and light intensity. This indicates that simplified models describing the SWE in terms of a defect density increase proportional to ~t1/3 or ~t1/2 do not reflect the real complex nature of the SWE.

Keywords: nanostructure, hydrogenated amorphous silicon, Staebler-Wronski effect, dangling bonds, vacancies

Th-C2.3 11:10–11:30

Impact on Thin Film Silicon Solar Cell Electrical Performance of Substrates Fabricated using the LaText Process: Light Trapping and Reduction of the Stabler-Wronski Effect

Erik Johnson (1), Rym Boukhicha (1), Pere Roca i Cabarrocas (1), Jean-Francois Lerat (2), and Thierry Emeraud (2)

1. LPICM-CNRS – Ecole Polytechnique, 91128 PALAISEAU, France

2. EXCICO France SAS, 13-21 Quai des Gresillons, F-92230 Gennevilliers, France

It has been previously shown that the particular structure created by using an excimer laser annealing and chemical etching process on ZnO:Al thin films can result in a remarkably high optical haze [1], and that this haze translates into improved photocurrent in hydrogenated microcrystalline silicon (μc-Si:H) thin-film silicon solar cells [2]. However, the impact on the electrical properties of the resulting cells has not been discussed in as much detail.

We present the results of work on both μc-Si:H and hydrogenated amorphous silicon (a-Si:H) solar cells employing such substrates, bringing into focus some striking results. For the μc-Si:H cells, although the photocurrent is increased by more than 15%, the open-circuit voltage (VOC) and fill factor (FF) both decrease with increased texture (by up to 20% and 10%, respectively). This issue is not directly critical, as in a "micromorph" device configuration, the μc-Si:H cells are deposited on top of the a-Si:H top-cell, so the effect will be lessened. Nevertheless, we show that the performance of the cells can be simply predicted by a shift in the Raman peak position, providing a new route to optimization.

In contrast, the initial performance of a-Si:H cells suffers much less with improved light-scattering from the LaText process, with little to no decrease in the VOC and FF. Furthermore, the robustness of the a-Si:H cells to light-induced degradation (LID), i.e. the Stabler-Wronski effect (SWE), appears to be augmented by the greater texture. Indeed, cells on the most textured substrates show no degradation at all. Even more interestingly, the stability of these cells can be directly predicted by the position and width of the Raman scattering peak. We will present a coherent explanation of the reduction of the classical SWE in such devices, and summarize why the LaText process holds so much promise for thin film silicon solar cells.

[1] E. V. Johnson, C. Charpentier, T. Emeraud, J. F. Lerat, C. Boniface, K. Huet, P. Prod'homme, and P. Roca i Cabarrocas, "Room Temperature Fabricated ZnO:Al with Elevated and Unique Light-Trapping Performance", Amorphous and Polycrystalline Thin-Film Silicon Science and Technology - 2011 (Mater. Res. Soc. Symp. Proc.) A13.5

[2] E. V. Johnson, P. Prod'homme, C. Boniface, K. Huet, and T. Emeraud, P. Roca i Cabarrocas, "Excimer laser annealing and chemical texturing of ZnO:Al sputtered at room temperature for photovoltaic applications", Sol. Energy Mater. Solar Cells 95, (2011) 2823

Keywords: microcrystalline silicon, amorphous silicon, light-induced degradation, photovoltaics, transparent conductive oxides

Th-C2.4 11:30–11:50

Crystalline Silicon Surface Passivation using Microcrystalline Silicon Oxide Layers

Kaining Ding (1), Urs Aeberhard (1), Andreas Lambertz (1), Bernd Holländer (2), Friedhelm Finger (1), and Uwe Rau (1)

1. IEK5-Photovoltaik, Forschungszentrum Jülich, Leo-Brandt-Strasse, D-52425, Germany

2. PGI-9, Forschungszentrum Jülich, Leo-Brandt-Strasse, D-52425, Germany

In a previous publication, we demonstrated the potential of doped microcrystalline silicon oxide (μc-SiOx) as a wide-gap window layer to reduce optical losses in silicon heterojunction solar cells, as this material combines high transparency and low refractive index with reasonable electrical conductivity. The highest solar cell efficiency (active area 0.67 cm2) achieved was 19.0% on non-textured p-type wafer with Voc = 667 mV, Jsc = 35.8 mA/cm2 and FF = 79.6%. The excellent Jsc for flat cells was found to mainly arise from the low optical losses in the μc-SiOx window layer.

In the present work, we show that μc-SiOx, in addition to its advantages in solar cell devices, can be used to gain more understanding on the fundamental principles of crystalline silicon (c-Si) surface passivation, as the material properties can be varied over a wide range by changing the μc-Si to a-SiOx phase ratio upon variation of, e.g., the input gas ratios during film growth. The investigation of the relationship between the c-Si surface passivation quality and a number of systematically changeable μc-SiOx film properties can provide insight into the nature of c-Si surface passivation. In particular, we studied the surface passivation properties of phosphorous and boron doped μc-SiOx layers on n- and p-type float zone wafers, respectively. The dopant and oxygen precursor gas flow ratios were systematically changed during μc-SiOx depositions giving rise to variations of the structural and optoelectronic properties. Moreover, lifetime samples with an additional thin, intrinsic a-SiOx buffer layer between c-Si and μc-SiOx were investigated to the same extent as those without buffer layer.

The quality of the μc-SiOx/a-SiOx interface was found to affect the wafer passivation depending on the strength of the field effect passivation, which is in agreement with a model assuming electrons and holes to be able to travel from the c-Si absorber through the buffer layer and to recombine at the next defect-rich interface. In the case of a strong field effect at the n/p junctions, the probability for the minority charge carriers to reach the μc-SiOx/a-SiOx interface and to recombine there is low and thus the interface quality is less important for the wafer passivation. In this case, the active doping concentration of the μc-SiOx layers determines the passivation quality. In the case of a weak field effect at the p/p or n/n junctions, the quality of the μc-SiOx/a-SiOx interface is important. The quality of the wafer passivation diminished with lower hydrogen content of the μc-SiOx layer required for dangling bonds saturation, which again is determined by the crystalline volume fraction and the oxygen content. The model implies that (i) a thicker a-SiOx layer decreases the recombination velocity at this interface by lowering the probability for the wrong type of charge carriers to penetrate this layer and (ii) the quality of the μc-SiOx/a-SiOx interface becomes more important with decreasing a-SiOx layer thickness. These predictions were additionally confirmed on device level in the present work.

Keywords: silicon heterojunction, carrier lifetime, wide-gap window, silicon oxide buffer

Th-C2.5 11:50–12:10

Solid-phase Crystallization of High Growth Rate Amorphous Silicon Films Deposited by Gas-jet Electron Beam Plasma CVD Method

E. A. Baranov (1), S. Ya. Khmel (1), A. O. Zamchiy (1, 2), and I. V. Cheskovskaya (2)

1. Institute of Thermophysics, Lavrentiev Ave. 1, 630090 Novosibirsk, Russia

2. Novosibirsk State University, Pirogova Str. 2, 630090 Novosibirsk, Russia

Thin-film crystalline silicon on glass for solar cells has to solve actual problems associated with the low cost and the large area deposition. One approach for obtaining crystalline silicon thin film on the glass is polycrystalline silicon (poly-Si) thin films are developed from a-Si:H layers by means of solid-phase crystallization.

One of the promising techniques for synthesizing thin films of a-Si:H is gas-jet electron beam plasma chemical vapor deposition method. This method provides high deposition rates of the main light-absorbing layer of silicon for solar cells with low energy consumption in a standard vacuum chamber [1].

The experiments were carried out in a vacuum chamber. During the deposition the pressure was 15 Pa. The processing gases were 5%SiH4+95%Ar mixture. The total gas flow rate was 185 sccm. The chamber was equipped with a forevacuum electron gun with a plasma cathode. It allows to generate an electron beams with 2 keV energy and a current to 70 mA. The electron beam diameter was 7 mm. The holder is equipped by heater. It allows changing substrate temperature from room to 500°C. Quartz glass plates were used as substrates for silicon film deposition at different temperatures. The thin films of a-Si:H underwent the annealing process in a vacuum (6 Pa) at 650°C for 10 hours.

Transmission spectra of synthesized films were measured with help of spectrograph equipped with photodiode arrays for optical radiation detection. Optical transmission and reflection spectra are recorded for investigation optical properties and measurement thickness of silicon thin films with help of PUMA code. The structural properties of the films were investigated by Raman spectroscopy.

The thickness of the synthesized films were about 1 μm. The deposition rate decreases from 4 to 2.9 nm/sec with increasing substrate temperature from RT to 500°C. Possible reason for decreasing of deposition rate is density increasing of the film because of reduction of voids and the reduction of hydrogen content in it. Also the growth rate decreasing with increase substrate temperature can be attributed to the temperature dependence of the reaction probability or an increase of silane desorption from the film surface. Obtained values of the optical band gap decrease from 2 to 1.8 eV with the substrate heating due to the reduction of hydrogen content in the film.

The films of polycrystalline silicon were obtained as the result of solid-phase crystallization. It is widely known the method for producing thin-film crystalline silicon on glass, but for a-Si:H obtained by gas-jet electron beam plasma chemical vapor deposition method was first used. The optical band gap of the films of polycrystalline silicon varied from 1.6 eV to 1.5 eV with substrate temperature decreasing.

The gas-jet electron beam plasma chemical vapor deposition method is used for depositing amorphous silicon thin films with high growth rates and followed by solid-phase crystallization obtained the polycrystalline silicon films.

[1] R. G. Sharafutdinov, S. Ya. Khmel, V.G.Shchukin, et al., Solar Energy Materials & Solar Cells 89, 99 (2005)

Keywords: phase crystallization, gas-jet electron beam plasma chemical vapor deposition method, amorphous silicon

Th-C2.6 12:10–12:30

Metastability Effects After Oxygen Exposure in Thick Silicon Films Deposited by VHF-PECVD on Glass Substrates Investigated by Dual Beam Photoconductivity

Gökhan Yilmaz (1), Hamza Cansever (1), Mehmet Güneş (1), Vladimir Smirnov (2), Friedhelm Finger (2), and Rudi Brüggemann (3)

1. Mugla Sitki Koçman University, Faculty of Sciences, Physics Department, Kötekli Yerleskesi, 48000 Mu─čla, Turkey

2. Forschungszentrum Jülich, IEK-5 Photovoltaik 52425 Jülich, Germany

3. Institut für Physik, Carl von Ossietzky Universität Oldenburg, 26111 Oldenburg, Germany

Multi-junction thin-film silicon solar cells are generally deposited on rough light-scattering substrates with absorber layers of undoped silicon films with thickness of 1–2 μm. The changes in the optoelectronic properties of intrinsic absorber layers due to metastability effects are the main degradation elements for solar cell degradation and are subject of the research work in the present paper

While adhesion of such thick silicon films on the rough substrates is usually good, on smooth glass substrates it is a major problem especially when the material is prepared with low silane concentration of SC= SiH4/(SiH4 + H2) < 10% for microcrystalline and amorphous silicon. For this reason, most of the published results of the metastability due to atmospheric exposures have only used thin silicon layers of typical thickness < 0.3 μm. To improve adhesion and allow investigation of material with thickness relevant for solar cell application, we use thin (10–15 nm) e-beam evaporated a-SiOx layers on the smooth glass substrate. With this interface layer adhesion was improved considerably. Silicon films thicker than 1μm were deposited using VHF-PECVD under varying SCs at 200°C. Crystalline volume fraction, IcRS, of the samples was determined from Raman measurements. Temperature dependent dark conductivity σdark, intensity dependent photoconductivity σph and sub-bandgap absorption spectra as determined from the dual beam photoconductivity (DBP) method were used to detect the changes in the metastable state. Samples were exposed to laboratory atmosphere in the dark for more than a year and to a controlled high-purity oxygen gas atmosphere for several days. All sample measurements were performed at 300K in exposing gas ambient as well as in high vacuum of 2–3x10–6 mbar. Real time monitoring of the dark conductivity has been recorded during the annealing and as samples were in the oxygen gas ambient until a steady-state of conductivity was reached.

Only small changes in conductivities and sub-bandgap absorption were found as oxygen exposed samples were characterized in oxygen gas ambient. However, substantial metastable changes occur as oxygen exposed samples were put in high vacuum at 300K. After sample placement and pump down, the final high vacuum of 2–3x10–6 mbar is established after about one hour. However, the dark conductivity of samples continues to increase and reaches a steady-state value at the end of longer period of time. As compared with the results of thin samples reported previously, changes in oxygen exposed states in thick samples are small. Depending on the sample, σdark and σph increase slightly within a factor of 2–3 and sub-bandgap absorption coefficient measured at 0.8 eV decreases by the same factor in oxygen exposed state. Following heat treatment carried out at 430K, σdark and σph remain almost unchanged but the sub-bandgap absorption coefficient recovers slightly in the annealed state. Such increase in σdark and σph can be due to small shift of the Fermi level EF towards the conduction band edge, which finally causes an increase in the density of occupied defect states below EF. As a result, the sub-bandgap absorption coefficient at lower energies should increase if no change in the true density of states occurs. However, the results of the sub-bandgap absorption coefficient spectra indicate a significant decrease in the density of occupied defect states below EF. Therefore, metastability changes in thick silicon films due to oxygen exposure should involve additional mechanisms in addition to the shift of the Fermi level, such as passivation of defects by oxygen.

Keywords: microcrystalline silicon, metastability, dark conductivity, photoconductivity, dual beam photoconductivity, sub-bandgap absorption