Thin-film silicon solar cells offer the advantages of low material and manufacturing costs. In order to enhance the absorptance of incident light in the optically thin silicon absorber layer, this technology requires advanced light-trapping concepts. Conventional devices apply randomly textured transparent conductive oxide substrates serving as light-scattering front contacts as well as reflective light-scattering metal back contacts. In recent years, several novel light-trapping concepts based on periodic nanostructures and periodically nanotextured interfaces, such as grating couplers, photonic crystals or plasmonic reflection gratings, have been suggested and prototyped. For these concepts the absorption of incident light in the solar cells is enhanced by light-coupling to waveguide modes which are supported by the silicon absorber layer of the solar cells but can be excited at the same time by incident light.
In this contribution, our recent progress on light-trapping in periodically structured prototype thin-film silicon solar cells made of hydrogenated amorphous silicon and hydrogenated microcrystalline silicon is presented. The prototype solar cells show a superior light-trapping effect compared to solar cells applying the conventional random texture for light-trapping. To better understand this improved light-trapping effect, the coupling of incident light to waveguide modes in periodically nanostructured thin-film silicon solar cells is analysed in-depth. Therefore, the shape of the grating structure and the geometry of the unit cell of the two-dimensional periodic grating structure of the thin-film silicon solar cells are varied systematically and the excitation of the waveguide modes is studied. To characterize the coupling of incident light to individual waveguide modes, advanced characterization techniques, i.e. angular and polarization dependent spectral response measurements of resolution below 3 nm as well as near-field scanning optical microscopy, are developed and employed. Finally, based on our study new routes for improved designs of the periodic nanostructure of thin-film silicon solar cells will be outlined.

In the past few years a large amount of work has been performed on thin film a-Si:H and nc-Si solar cells with periodically textured back reflectors, composed of photonic and plasmonic crystals. These periodic back reflectors were predicted to i) introduce strong diffraction of light resulting in wave-guiding modes and ii) enhance light concentration through generation of plasmons. This light trapping is particularly needed at long wavelengths (lambda;> 600 nm) where photon absorption is low. Rigorous simulations predicted very large gains (>40%) in absorption and photo-current, with the absorption predicted to approach the Lambertian limit. It is an open question whether periodic back reflectors can exceed the performance of randomly textured back reflectors. However experimental solar cells fabricated with periodic back reflectors exhibit photo-currents considerably lower than rigorous predictions. Our measured external quantum efficiency (EQE) for a 1000 nm nc-Si cell with a periodic array of nano-cones is more than a factor of 1.5 to 2 lower than the simulated values in the long wavelength region (above 600 nm). Similar results are found for solar cells fabricated in other laboratories.
We analyze systematically the sources of these differences between experiment and prediction. We evaluate the parasitic losses in the ITO anti-reflection coating, and find only small losses (<5%) that cannot account for this discrepancy. There are considerable parasitic losses in the metal back reflector that partly contribute to the lower values measured in experiment. We evaluate this loss by simulations with periodic dielectric back reflectors. Another factor that has previously been overlooked is the loss of phase coherence of incoming waves within the solar cell, caused by random variations of thickness and structural inhomogeneity. This incoherence can severely deplete the wave-guided modes and can account for the much lower EQE and photo-currents observed experimentally. The loss of this phase coherence is apparent in the difference between measured and simulated absorption even for flat solar cells. We evaluate the loss of phase coherence in flat cells from experimental measurements, and will discuss how this can be a limiting factor for light trapping in periodic solar cells.

Black silicon (b-Si) is a material with an optically graded boundary which exhibits a reflectivity as low as 1 - 5 % in the whole range of Si absorption. A cone-like nano-texture, with a depth of less than 1 mu;m, leads to a density graded surface with a smooth variation of the refractive index. Nano-texturing of thin silicon wafers and films also results in an additional enhancement of the optical absorption due to extremely efficient light trapping [1]. The unique optical properties of b-Si, as well as the opportunity to produce nano-textures on different silicon substrates (mono- or polycrystalline wafers, a-Si:H and µc-Si:H films and others) are of significant interest for photovoltaic applications, in particular for thin-film solar cells [2].
Heterojunction solar cells manufactured from large grained polycrystalline silicon (poly-Si) films, prepared by electron-beam-crystallization, exhibit promisingly high open-circuit voltages (V_oc), but suffer from a low short-circuit current (J_sc) due to the lack of an appropriate light-trapping scheme [3]. We applied a b-Si nano-texture in order to increase the absorption and, thus, to improve the J_sc. The nano-texturing of the 10 mu;m thick poly-Si absorber was achieved using metal-catalyzed wet-chemical etching. The process parameters such as metal particle size, etch time and etch solution were optimized in order to maximize the light trapping effect in the poly-Si film. With an additional wet-chemical post-treatment we control the final morphology and improve the electronic properties of the nano-texture [4]. UV-Vis spectroscopy reveals a reflectivity below 5 % in the spectral range important for Si solar cell operation and excellent light trapping in the film also for wavelengths above 700 nm. The passivation of the nano-textured Si surface with hydrogenated amorphous silicon (a-Si:H) was optimized using monocrystalline silicon (c-Si) wafers as a model system. Injection-level dependent lifetime measurements reveal excellent passivation after the (i)a-Si:H deposition, showing effective carrier lifetimes above 1 ms.
The good optical properties of the nano-textured thin-film poly-Si solar cells result in a significant increase of J_sc. However, due to some reduction in the V_oc the overall energy conversion efficiency did not improve in the same way. Further optimization of the doped a-Si:H layer stack on the nano-textured Si surface is needed in order to benefit from the excellent optical properties and to transform them into higher efficiencies.
[1] S. Koynov, M. S. Brandt, M. Stutzmann, J. Appl. Phys. 110, 043537 (2011).
[2] S. Koynov, M. S. Brandt, M. Stutzmann, Appl. Phys. Lett. 88, 203107 (2006).
[3] D. Amkreutz, J. Müller, M. Schmidt, T. Hänel, T. F. Schulze, Prog. Photovolt: Res. Appl.19, 937 (2011).
[4] M. Algasinger, J. Paye, F. Werner, J. Schmidt, M. S. Brandt, M. Stutzmann, S. Koynov, Adv. Energy Mater. 3, 1068 (2013).

Hydrogenated nanocrystalline silicon (nc-Si:H) is widely used as absorber layer in the bottom cell of multijunction thin-film silicon solar cells. The nano-textured substrates with morphology features in the order of 1 mu;m are commonly used to achieve efficient light trapping in the state-of-the-art devices. However, the nano-textured substrates deteriorate (by the incorporation of defect rich filaments) the electrical performance of solar cells by decreasing the open-circuit voltage (Voc) and fill factor (FF) compared to flat substrates. In addition, the Voc and FF of nc-Si:H solar cells show considerable drop as the cell thickness (and defect rich filaments) increases. As a result the performance of solar cell with absorber layers with typically thicknesses of 2-4 mu;m do not benefit from the higher generated current densities. Therefore, textured substrates which can provide efficient light trapping and maintain high Voc and FF for thick absorber layers under high deposition rates are highly desirable.
Recently, Sai et al. [AIST,Japan] have shown that periodically textured substrates with a large period resulted in higher Jsc and improved Voc and FF in 3-mu;m-thick nc-Si:H solar cells compared to substrates with a small period. Their work suggests that substrates with an even larger period, which we here refer to as micro-textures (feature size >5 mu;m), might have the potential to maintain high Voc and FF in thick cells by growing nc-Si:H material free from defective filaments. In this contribution, we explore the behavior of nc-Si:H solar cells deposited on the microtextured glass substrates. The micro-textures on glass were generated by wet-etching with In2O3:Sn as catalyst in a solution composing of HF and H2O2 for 30 min. The 1.5 microns AZO layers were fabricated by RF magnetron sputtering. Nano-textured AZO was obtained by wet-etching. This procedure allowed to obtain micro-textures with large opening angles and smooth U-shape. The advantages of the micro-textures for nc-Si:H solar cells were systematic studied by using three different textured substrates: (i) nano-textured ZnO:Al (AZO) on flat glass, (ii) micro-textured glass coated with as-deposited ZnO:Al, and (iii) modulated surface textured substrate by superposing nano-textured ZnO:Al on micro-textured glass.
The micro-textured substrates result in higher Voc and FF than nano-textured substrates. For thick solar cells, high Voc and FF are maintained. Particularly, the Voc only drops from 564 to 541mV as solar cell thickness increases from 1 to 5 mu;m. The improvement in electrical performance of solar cells is ascribed to the growth of dense nc-Si:H layers free from defective filaments on micro-textured substrates. Thereby, micromorph tandem solar cells with an initial efficiency of 13.3%, Voc = 1.464 mV and FF = 0.759 are obtained, indicating the high potential of micro-textures for high-efficiency multi-junction thin-film silicon solar cells.

Light trapping is an important issue for hydrogenated amorphous silicon (a-Si:H) thin-film solar cells. Texturing the back reflector by a periodic structure has emerged as an effective method to improve the cell performance, because the periodic back reflector can couple the incident light into guided modes and Bloch-wave surface plasmon polaritons (BW-SPPs) which can propagate along the absorber layer. In this study, an anti-ring array, a novel nanostructure, was fabricated by monolayer of polymer/nanosphere hybrid and applied as a back reflector in substrate-type a-Si:H thin-film solar cells.
The anti-ring structure comprises a nanodome centered inside a nanohole. Several dips were observed in the specular reflectance spectra of Ag-coated anti-ring array. These dips were caused by BW-SPPs and the corresponding wavelengths were related to the period of the nanostructure. When a period of 1000 nm was used, the cell exhibits a Voc of 0.81 V, a Jsc of 11.11 mA/cm², a FF of 66.01% and a power conversion efficiency of 5.94%, which is 39% improvement in comparison with that of the cell fabricated on a flat back reflector. Compared to the flat counterpart, 17.9% enhancement in Jsc and 39% enhancement in power conversion efficiency are obtained without degradation of the FF. The enhancement can be attributed to the BW-SPPs effect and the diffuse scattering of light.
We also found that the cell fabricated on an anti-ring back reflector outperform that fabricated on a nanohole back reflector (i.e. without center nanodome) when a period of 1000 nm was used. The angle-resolved reflectance spectrum shows that the Ag-coated anti-ring array can scatter more light into large angle than the Ag-coated nanohole array, which is consistent with the simulation result based on finite-difference time-domain (FDTD) method.

Nano-textured “black” silicon (b-Si) exhibits a reflectivity of only a few percent in the visible range as well as excellent light scattering properties [1]. It is therefore especially well-suited to enhance the short circuit current (jsc) of silicon solar cell concepts like silicon hetero-junction (SHJ) solar cells on thin wafers [2] or polycrystalline (poly-Si) thin film solar cells [3]. However the nano-textured surface is prone to high defect densities and thus a major challenge is the development of suitable passivation layers. Up to now solar cell concepts comprising b-Si rely on surface passivation with Al2O3 [4], or thermal silicon oxides [5]. Such dielectric passivation layers must be opened locally for contact formation. An alternative but up to now unexplored route is the passivation with hydrogenated amorphous silicon. This approach is advantageous since passivation and contacting of the nanostructured surface can be achieved with a single layer and furthermore the amorphous-crystalline SHJs excellent passivation enables the highest open circuit voltages of any silicon wafer-based technology [2].
In this communication we present nano-textured amorphous-crystalline silicon hetero-junctions which exhibit excellent passivation quality and low reflection values. Minority carrier lifetimes above 1.3 ms, and implied open circuit voltage above 700 mV on nano-textured silicon surfaces with reflectivity below 5 % are reached using plasma-enhanced chemical vapor deposition of approximately 5 to 6 nm thick intrinsic amorphous silicon layers for the passivation of b-Si.
Nano-textured silicon has been implemented in wafer-based and thin film amorphous-crystalline SHJ solar cells. For both cell types the reflectivity was strongly reduced and an effective reflectivity below 5% was obtained. For the thin film-based solar cell an increase in jsc by 30% has been obtained. For wafer-based solar cell jsc was increased as compared to a planar reference, however due to inferior blue response jsc is lower than for a conventional textured reference. Numerical simulations indicate that this loss is due to a photo inactive region inside the nanotexture. Furthermore these early prototype cells experience a drop of the implied voltage during cell processing, which leads to a final open circuit voltage of 614 mV. Options to preserve the excellent initial passivation of the intrinsic amorphous silicon layers during the whole process and to improve the poor blue response of these solar cells will be discussed. Utilizing these improvements may open a viable route towards high-efficiency b-Si solar cells with thin absorbers.
[1] S. Koynov et al., Appl. Phys. Lett. 88 (2006) 203107
[2] M. Taguchi et al., IEEE J. of Photovoltaics (2013) in press
[3] D. Amkreutz et al., Prog. Photovolt: Res. Appl 19 (2011) 937
[4] P. Repo et al., Energy Procedia 38 (2013) 866-871
[5] J. Oh et al., Nature Nanotechn. 7 (2012) 743-748

Poor internal reflection of IR light at the lossy metal rear reflector of crystalline silicon solar cells limits the spectral response of the cells near the silicon bandgap, even for highly reflective metals. Here, we present a thorough experimental and theoretical investigation of the optical properties of Si/dielectric/metal structures that are representative of the rear of rear-passivated crystalline silicon solar cells, and extend the analysis to “absorbing dielectrics” that represent the rear transparent conductive oxide (TCO) layer in silicon heterojunction solar cells. By calculating the electric field intensity at the surface of the metal, we find that for thin dielectric layers, p-polarized light arriving at the back surface above the Si/dielectric critical angle is strongly absorbed in plasmonic modes that cause significant internal reflection losses. We employ a ray tracer to calculate the total reflectance of a solar cell with random pyramids, using as input the internal reflectance values as a function of incident angle determined from the field intensities. The results reveal that—consistent with our measurements—dielectric layers at least 150 nm thick minimize absorption in the metal reflector; the rear passivation layers in solar cells can thus serve an optical as well as electrical role if properly designed. Finally, we demonstrate this in silicon heterojunction solar cells by measuring record IR internal quantum efficiency with a thick, low-refractive index dielectric buffer layer sandwiched between the rear of the cell and the metal reflector.

Improving the conversion efficiency of thin-film silicon solar cells is a delicate interplay between spectral utilization, materials processing and light management. For enhancing the photo-current density generated by the solar cell, the spectral absorptance of the absorber layer must be enlarged. This means that the spectral losses due to reflectance and supporting layers have to be minimized. In this respect, light management techniques play an important role. In particular, combining light scattering at textured interfaces with efficient rear reflector enables enhanced light coupling in the absorber layers.
In this contribution we report our recent studies on light management applied to nano-crystalline silicon solar cells. We analysed the effect of decoupled texturization between front and rear side on the absorptance of the absorber layer. At the front side, we considered high aspect ratio pyramidal features for anti-reflective effect and enhanced light in-coupling, while at the rear side we took into account shallower pyramidal features for effective light scattering. The novelty of our study lies in the optical modelling of the complete p-i-n solar cell structure based on nc-Si:H and in the analysis of its spectral performance by means of the excited wave-guided modes. A 3-D Maxwell solver based on finite element method was used for the simulation of absorptance and reflectance spectra, while an iterative method based on the calculation of poles of the Fresnel coefficients was deployed for sampling the excited wave-guided modes.
Our study comprised three simulation phases, in which the thickness of the intrinsic nc-Si:H layer was kept constant to 2 µm. Firstly, an ample parameters space was investigated. Varying period, height and duty cycle of both front and back side textures, we could rank the three most promising structures to be used later on. In the second phase, we especially focussed on the metallo-oxide interface at the back side. Regardless the simulated geometrical structure, we found that the insertion of doped layers based on nc-SiO_x:H with the concurrent usage of an appropriate back transparent conductive oxide (TCO) resulted in the minimization of plasmonic losses in the silver rear reflector. In the third phase we finally optimized the thickness of the front TCO.
Our best simulated solar cell structure showed an implied photo-generated current density equal to 35.65 mA/cm², which is +3.40% higher than the value predicted by Tiedje-Yablonovitch limit calculated for the same thickness and in the wavelength range between 300 nm and 1200 nm. This enhancement was ascribed to the optimized decoupled front and back side texturization, which increased the light in-coupling at long wavelengths. In fact, by sampling the resonance peaks of the nc-Si:H absorptance on a dispersion-relation diagram, we found that all of them were related to wave-guided modes concurrently excited by the front and the back textures.

Widespread implementation of silicon-based photovoltaic (PV) technology is still limited in no small measure by high materials costs. One interesting approach to reduce materials consumption is shown in recent reports from our group by utilizing thin, ribbon-like, micro-scale Si solar cells, devices generated using precise methods of micromachining Si(111) wafers in conjunction with transfer-printing techniques that allow their deterministic assembly on secondary substrates. The design of our earliest generations of microcells, however, requires crucial improvements to address fundamental criteria for optimal carrier collection efficiency, as a passivation layer (e.g., thermal oxide, silicon nitride) is not incorporated, leading to substantial losses from surface recombination.
In this presentation, we describe a simple, yet robust approach for fabrication of silicon solar microcells that integrates passivation. By incorporating a thermal oxide as etching/diffusion mask, which also doubles as an effective electrical passivation and anti-reflection layer, we demonstrate significant advance in microcell design that enhances process reliability, energy conversion efficiencies, and modes of module assembly. We report a best cell efficiency of 11.7% under an AM1.5D solar spectrum for an optically thin (30 µm thick) device measured on a non-reflective substrate, which is a substantial improvement over previously reported results. External quantum efficiency measurements specifically show a marked improvement in the blue response that results from mitigating losses due to surface recombination in these high-surface-area, micro-scale devices.
Light management strategies explored for these microcells include integrating backside reflectors (BSR) and luminescent materials, which redirect light to the otherwise unilluminated sidewall and bottom surfaces of these thin devices to compensate the inherent low optical absorption. Utilizing a diffuse BSR for a device embedded in a transparent waveguiding polymer matrix would double its power output (The efficiency in this scenario reaches around 20%). Additionally, doping the polymer matrix with fluorescent inorganic nanorods (CdSe/CdS) or organic dye molecules (DCM), which absorbs incident light and emits into total internal reflection (TIR) modes inside the waveguide, boosts concentration ratio to around 3 and 4 (both with a diffuse BSR), respectively, furthering lowering down the power-referenced materials consumption of these lightweight devices.

Utilizing thin (~50 mu;m) crystalline Si (c-Si) has recently been a promising candidate for reducing material costs in solar cell applications. Insufficient light absorption by employing a thin c-Si wafer requires a nanostructured surface for reducing optical reflectance as low as 5 % in solar spectrum. However, surface evolution of higher-index crystalline planes as well as a higher surface-to-volume ratio of c-Si seriously degrade the internal quantum efficiency (IQE) especially for a blue region (wavelengths of 400~600 nm) due to the increased surface recombination.
We present that the OH-terminated surface of nanostructured (nanoholes) Si improved the passivation performance of atomic-layer-deposited (ALD) Al2O3 in comparison to a conventional H-terminated surface, which then resulted in the improvement in blue response. The OH-termination of nanostructured surface prior to ALD Al2O3 process has been carried out by inserting O2 plasma followed by HF treatment. A decreased amount of dangling bonds as well as the improved quality of interfacial SiOx between Al2O3 and Si were confirmed using Raman spectroscopy. Surface recombination velocity (SRV, cm/s) extracted by the effective carrier lifetime (mu;s) also revealed a superior performance of the OH-terminated surfaces on the nanostructured silicon. The improvement in SRV results was developed more evidently with increasing the depths of nanoholes from 300 to 800 nm. As a result, our new approach improved the IQE in wavelengths of 300~470 nm so that the short circuit current density of 21.5 mA/cm2 was obtained, which was 11.4 % higher than that of conventional H-terminated surface.

Enhancing light absorption in ultrathin film silicon solar cells is important for reducing costs and improving performance. In this work, we integrate a genetic algorithm with finite difference time domain simulations to determine the optimum silicon nanostructure with maximum solar absorption for some fixed equivalent thickness. Different single-sided (top or bottom) and double-sided gratings structures are evaluated with building blocks of nanowires, nanoholes, nanocones, and tapered nanohole structures. We compare the performance of these structures to that of a thin film with an optical path length of twice the thickness as well as the Lambertian limit. We find that double-sided grating structures improve light absorption over the entire solar spectrum compared to single-sided structures. This work demonstrates the ability of the genetic algorithm optimization technique to quickly search through a large parameter space to determine nanostructures with maximum solar absorption.

Although high-efficiency (~14-15%) a-Si:H/mu;c-Si:H tandem solar cells have been demonstrated in the initial state, the stabilized efficiencies of such devices after long-term illumination are limited to ~12% due to the light-induced degradation of a-Si:H, known as the Staebler-Wronski effect [1]. Therefore, the suppression of the light-induced degradation is crucial in order to make significant progress of thin-film silicon solar cells.
In our previous studies [2, 3], we have demonstrated that high-efficiency and low-degradation a-Si:H solar cells can be obtained when the a-Si:H absorber layer is deposited by a remote plasma process using a triode PECVD technique. Although the deposition rate is relatively low (0.01-0.03 nm/s) compared to the conventional diode-type PECVD process (~0.2 nm/s), the light-induced degradation in conversion efficiency (Δeta;/eta;_ini) of single-junction solar cell is substantially reduced (e.g., Δeta;/eta;_ini~11% and ~19% for the cells deposited by triode and diode PECVD, respectively, at an absorber thickness of 250 nm). As a result, a stabilized efficiency of 9.6% has been attained using a commercially-available TCO substrate.
In this contribution, we report on the further progress of a-Si:H single-junction solar cells realized by optimizing whole device design. We show that stabilized efficiencies of as high as 10% can be attained even when the solar cell is thickened up to ~400 nm. Results of the material characterization such as microstructure parameters and light-induced metastable defects in the a-Si:H layers and devices are also presented.
[1] D. L. Staebler and C. R. Wronski, Appl. Phys. Lett. 31, 292 (1977) . [2] S. Shimizu et al., J. Appl. Phys. 97, 033522 (2005). [3] T. Matsui et al., Prog. Photovolt: Res. Appl. 21, 1363 (2013).

Light-induced degradation (LID) of amorphous silicon (a-Si:H) solar cells due to the Staebler-Wronski effect has been widely discussed in literature. However, it has been most often discussed with respect to the degradation of the intrinsic absorber layer.
In the present study, LID of a-Si:H solar cells is studied with respect to the amorphous silicon carbide (p-(a-SiC:H)) layer that is part of the window layer of high efficiency solar cells. We have deposited solar cell series varying the p-(a-SiC:H) thickness and the substrate roughness of single junction solar cells in superstrate configuration. The solar cell design is state-of-the-art using low-pressure chemical vapor deposition zinc-oxide for front and back contacts that are in contact with p- and n-doped silicon oxide layers. Plasma-enhanced chemical vapor deposition (PECVD, 40 and 13 MHz) has been used for all silicon layers, using a cluster tool with dedicated chambers for p-doped, intrinsic, and n-doped layers.
During light soaking, a systematic open-circuit voltage (Voc) increase could be observed for thin p-layers, while Voc decreases for thick p-layers. This effect is more pronounced for rough than for smooth substrates: The critical p-(a-SiC:H) thickness, at which light soaking has no effect on the Voc, increases with increasing substrate roughness. These Voc changes have a strong impact on the conversion efficiency of the solar cells. First, the optimum p-(a-SiC:H) thickness depends on the substrate roughness. Second, highest stabilized cell efficiencies are obtained using thinner p-(a-SiC:H) layers than what is optimum in initial state. Different contributions of short-circuit current, fill factor, and Voc to LID of the conversion efficiency are discussed. All trends could be reproduced using different cell designs in three different PECVD systems.
To discriminate the effect of effective p-layer thickness on rough substrates, the nominal thicknesses are corrected by the effective surface as determined from AFM measurements.
Different mechanisms could lead to the observed Voc changes. These are investigated by bias light and bias voltage dependent EQE measurements and by analyzing the degradation /annealing kinetics of the solar cells. The changes are related to layer properties as measured by ellipsometry, photothermal deflection spectroscopy, and conductivity.
Finally, we will briefly discuss our latest tandem and triple junction solar cells where we incorporated these a-Si:H cells as top cells.

The a-Si:H solar cells having high open-circuit voltage (Voc>950 mV) are highly desirable for the top junction in thin-film silicon based tandem and triple-junction solar cells. Except for the high Voc, the top cell should also have high spectral response (between 350-600 nm wavelength range) to allow thinner absorber layer in order to reduce the light-induced degradation. Therefore, doped layers with high transparency are required. Commonly the a-Si:H solar cells deploy a-SiC:H as p-layer and nc-Si:H or a-Si:H as n-layer. Those doped layers will inevitably lead to high parasitic absorption losses, and thus make it difficult to achieve sufficient photocurrent with a thin absorber layer. In this contribution, we will first discuss how to achieve high Voc by processing a-Si:H at high-pressure and high-power regime. Then deployment of highly transparent silicon oxide (SiOx) doped layers will be discussed to obtain better spectral response (or external quantum efficiency) and higher Voc than the conventional doped layers. Specifically, following key points will be presented in the conference.
1. The a-Si:H is deposited at high-pressure (>5 mbar ) and high-power (>0.1 W/cm2) regime, which results in larger bandgap than materials commonly processed at low-pressure and low-power regime. The bandgap of a-Si:H can be tuned by H2/SiH4 dilution, power and pressure. High performance device-grade a-Si:H can be obtained over wide deposition window, in contrast to the narrow window at low-pressure regime.
2. Highly transparent p-SiOx:H with sufficient conduction is investigated. Firsly, it should have good ohmic contact with front ZnO TCO. Second, the control of crystallinity of p-SiOx:H is a critical point to obtain high Voc for a particular absorber layer. Finally, a very thin layer of i-SiOx:H inserted in the p/i interface can significantly reduce the boron diffusion during the deposition of i-layer, and thus considerably improve the blue spectral response. Consequently, EQE higher than 70% at lambda;=400 nm is achievable.
3. Low absorption n-SiOx:H to replace absorptive n-aSi:H is necessary to achieve high spectral response over 500-700 nm wavelength range. Furthermore, the low refractive-index n-SiOx:H layer can also function as intermediate reflection layer in multijunction devices. Control of the i/n interface is crucial to achieve high FF. The insertion of an ultra-thin (<3 nm) n-aSi:H or n-aSiOx layer can significantly increase the FF of solar cells and result in FF comparable to cells with a-Si:H n-layer, without reduction of spectral response compared to single n-SiOx layer.
After optimization of the p-SiOx:H and n-SiOx:H doped layers, a-Si:H solar cells with high Voc, high FF and excellent spectral response is obtained (Voc>960 mV, FF>74%, and efficiency>10%). The light-induced degradation of solar cells with SiOx:H doped layers are investigated, and will be compared to the solar cells with conventional doped layers.

The manufacturing cost of thin film Si based tandem and triple junction cells and modules is at present too high to meet current module market prices. Conventionally, microcrystalline silicon is used as the low-bandgap absorber in micromorph solar cells (a-Si/µc-Si tandem cells). However, due to the considerable thickness needed for the µc-Si:H absorber, it takes three to four times as many deposition reactors compared to single junction cells to produce tandem cells, leading to high cost of ownership. One of the approaches to reduce processing time of the low-bandgap layer(s) in multijunction silicon-based solar cells is the use of hydrogenated amorphous silicon germanium (a-SiGe:H). In general however, a-SiGe:H has not been considered a viable option because of (i) the high defect density for PECVD a-SiGe:H, at band gaps < 1.4 eV, and (ii) the cost of GeH4. On the other hand, due to its direct gap nature, the thickness of an a-SiGe:H absorber layer can be kept 10 times smaller than that of µc Si:H.
We are investigating whether a-SiGe:H can be reconsidered for inexpensive production of multijunction thin film Si based solar cells if HWCVD is used as the deposition method. HWCVD is a simple and low cost deposition technique allowing high deposition rates while maintaining good defect passivation. Early results reported by NREL include the achievement of material with a band gap close to that of µc Si:H (1.2 eV) with an equivalent photoresponse (in excess of two orders of magnitude). Their work has led to 8.64% single junction cells without any band-gap profiling in the absorber layer.
We now continued this development to provide a novel thin film alloy for the struggling micromorph technology. We have produced a-SiGe:H materials with Tauc band gaps ranging from 1.6 eV down to 1.2 eV. Due to the efficient dissociation of silane and germane gases at the hot filament, a high deposition rate is achieved. Moreover, the dissociation rate of germane is three times faster than that of silane. The deposition rate for the lowest-gap material is 0.7 nm/s and is always higher than 0.5 nm/s. With this deposition rate, an active absorber layer (i-layer) of 150 nm is readily deposited within 5 minutes. This should be compared to the roughly one-hour long deposition time needed to deposit a 2-µm thick µc Si:H film at the commonly used deposition rate of 0.5 nm/s. Using a GeH4/SiH4 ratio of 1, we deduce from Raman spectroscopy that the films already contain 60-70% Ge, showing that Ge is preferentially incorporated in the film. We will report an extensive microstructure analysis and will present our first cells.

In thin-film silicon solar cells, the transport of photoexcited carriers in the active layer plays an essential role in device operation. Thus, improving the carrier transport directly leads to higher device performance. So far, the carrier transport was extensively studied for as-grown films at room temperature. Nevertheless, it is not well studied during growth. Here, we characterize the transport of photoexcited carriers during growth for further understanding the growth process and improving the device performance.
We performed in-situ measurement of the photocurrent during a-Si:H growth by plasma-enhanced chemical vapor deposition [1]. The device-grade intrinsic a-Si:H film growing on a glass substrate was illuminated with the intensity modulated probe laser. The photocurrent excited by this illumination was collected by dc biased interdigit contacts on the glass substrate. Because the photocurrent was superimposed on various other currents such as plasma associated currents and a leakage current, we used a lock-in technique to distinguish it. The measurement setup was placed in 60 MHz very-high-frequency discharge in a parallel plate configuration [2].
We found that during the growth, the photocurrent gradually increased as the film grew with time. Such a time evolution of the photocurrent indicates the existence of a defect-rich surface layer in the early stage of growth and the formation and growth of a bulk layer at a later stage. We also found a significant improvement of the photoconductivity after the growth, i.e., during postgrowth annealing. The photoconductivity was increased by one order of magnitude at the temperatures of 433 - 513 K. The characteristic time for the postgrowth annealing was obtained to be of the order of 100 - 10000 s, depending on the temperature.
[1] S. Nunomura, I. Sakata and M. Kondo, Appl. Phys. Express. 6, 126201 (2013). [2] S. Nunomura, I. Yoshida and M. Kondo, Appl. Phys. Lett. 94, 071502 (2009).

Silicon dangling bonds (db's) are the dominant defects in hydrogenated amorphous silicon (a-Si:H). Their creation by light is of great importance for the performance of a-Si:H based solar cells. Yet, the underlying processes are not understood in detail. Not even the structural and energetic characteristics of the defect - a prerequisite for modeling - are well established due to the variety of possible realizations of a db defect the amorphous network. To make any progress, theoretical models must be developed that reproduce the statistical distribution of the experimental findings.
Within the German EPR Solar project, we have calculated electron paramagnetic resonance (EPR) parameters of dangling bonds obtained by removing a hydrogen atom from a random Si-H bond. These calculations agree well with experiment on the g-tensor distribution. However, the experimentally observed red-shift in the Si hyperfine couplings compared to dangling bonds in a crystalline environment cannot be fully reproduced [1]. This indicates that the theoretical defect ensemble deviates from the experimental one, either due to an unknown selection mechanism in experiment, or due to an unintended bias in theory. One such bias might be the dangling-bond creation by H abstraction.
In my talk, I will compare the H-abstraction ensemble to an alternative set of defects created by inserting H into a Si-Si bond. The obtained models fall into two classes. In the first one, H indeed breaks the bond by attaching itself to one of the Si atoms, thereby creating a dangling bond at the other atom or -- after network reconstruction -- in a different place in the network. These dangling-bond realizations confirm our previous conclusion that the amorphous network has an intrinsic propensity to form dangling-bond defects upon perturbation of the network [2]. In the other cases, H assumes a bond-center position in a +1 charge state. The additional spin-polarized electron is then trapped at a suitable site nearby, without breaking any bonds. The calculated EPR parameters for the latter are distinctly different.
[1] Phys. Rev. B 84, 193304 (2011).
[2] Phys. Rev. B 87, 125308 (2013).