SiliconPV 2018 Awards presented to Dr. Felix Haase and Nils Folchert

We present a novel model for yield analysis of tandem devices, e.g., Si cells with perovskites, III-V or other top cells designs. Inputs are the IV and QE of the individual subcells and the irradiance-dependent module temperature of the bottom cell. Our model calculates the temperature of the tandem module by taking into account the performance and spectral working range of the different tandem devices, enabling an irradiance- and weather-dependent yield analysis for these modules. We apply the model to compare two types of tandem model, a GaInP top cells on GaAs and Si bottom cells. When the subcells are series connected both technologies perform similarly well within 0.3%. The performance of the GaInP/Si can be significantly improved by 5.6% using 3T devices with a back-contacted bottom cell instead of a 2T configuration. For GaInP/Si, the 3T-device works similarly well as the 4T-device, enabling the integration of monolithic tandem cells into modules at comparable high efficiencies.

This contribution focuses on improving the fundamental understanding of the carrier lifetime regeneration in boron-doped p-type Czochralski-grown silicon (Cz-Si) due to the deactivation of the boron-oxygen complex. We examine passivated emitter and rear cells (PERCs) fabricated on boron-doped p-type Cz-Si in darkness under constant forward bias voltage (Ubias) in order to isolate the impact of excess electrons on the regeneration kinetics from the possible direct impact of photons. In the solar cell base Ubias defines the electron concentration. Based on these dark regeneration experiments, we address the existing inconsistency regarding the dependence of the regeneration rate constant Rde on the average excess carrier density Δnavg: If the regeneration is performed under illumination, Rde ~ Δnavg has been reported by several research groups [1−2]. A proportional dependence of Rde on Δn seems to be inconsistent with the strictly mono-exponential decay of the effective defect concentration during the regeneration process, as Δn changes strongly during the regeneration. One possible explanation would be that the lifetime is in fact constant at the increased temperature applied during regeneration [3], which would result in a constant Δn value at regeneration temperature. Using the method of dark regeneration by current injection into solar cells, we are able to measure the total recombination current of the cells at the actual regeneration temperature under Ubias, i.e., at the physically relevant conditions. We observe that the regeneration at practically constant Δn can be described by a double-exponential process, featuring a fast (Rde,fast) and a slow (Rde,slow) component, with only Rde,fast increasing with increasing Ubias at the regeneration temperature. Hence, we identify Rde,fast with the standard regeneration process performed under illumination at elevated temperature. The direct comparison of Rde,fast values measured in the dark under electron injection and under illumination clearly proves that the regeneration is purely electronically stimulated mechanism.

Passivating carrier-selective contacts are key elements for high-efficiency silicon solar cells. Besides a-Si:H heterojunctions, polycrystalline silicon-rich films on a thin interfacial oxide combine excellent surface passivation and low contact resistivity for both electrons and holes. In a simple, non-patterned cell poly-Si would provide the front side emitter and the rear contact. However, absorption losses limit the thickness of poly-Si films used as front contacts to less than 20 nm. In order to achieve sufficient lateral conductivity in the front side emitter contact despite this limitation, transparent conductive films are imperative.

Transparent conductive oxides (TCOs) like indium-doped tin oxide (ITO) or aluminium-doped zinc oxide (AZO) are prominent candidates. The temperature stability of polycrystalline silicon on oxide (POLO) junctions allows high-temperature annealing of the AZO/POLO combination resulting in excellent junction passivation as well as enhanced transparency and conductivity of the TCO. Besides conductivity and transparency, a low contact resistance between the TCO and the adjacent layers is required to translate the low contact resistance of the POLO junctions into high solar cell efficiency. To that end, both the metal/TCO interface as well as the interface between the TCO and the carrier-selective contact material underneath should form stable contacts without any barrier to carrier transport.

We investigate the contact resistance and the structural properties of the interface between sputtered AZO and poly-Si. We identify the formation of an amorphous interfacial layer as a possible cause for high contact resistance. The high spatial resolution of transmission electron microscopy (TEM) combined with energy-dispersive X-ray spectroscopy (EDXS) and electron energy loss spectroscopy (EELS) allows for an analysis of the composition of the interfacial region. It turns out that the interfacial layer is silicon oxide like. Based on the results of this investigation we develop a modified AZO processing sequence to overcome this challenge.

After applying phosphorus diffusion gettering, the bulk lifetime of epitaxially grown silicon wafers increases significantly. However, the typically strong diffusion is not ideal for highest solar cell efficiencies. We find that n-type POLO junctions also lead to an increase of the bulk lifetime of epitaxial wafers that is as large as the increase achieved with phosphorus-diffusion gettering. Furthermore, the POLO junctions offer an improved recombination behaviour. This offers an alternative post-wafering treatment to increase the charge carrier lifetime of epitaxial wafers that can be effectively integrated into the processing of solar cells with POLO junctions. We measure an effective lifetime of 8 ms on a 3 Ω cm n-type epitaxial wafer at an injection level of Δp = 1015 cm-3.

Within this contribution, we examine the lifetime degradation in multicrystalline silicon (mc-Si) under illumination at elevated temperature, an effect sometimes denoted “LeTID”. Our lifetime analysis of belt-furnace-fired high-performance mc-Si wafers shows that the lightinduced degradation is most pronounced on samples with Al2O3/SiNx-stack passivation. In contrast to that, the degradation on samples coated with Al2O3 single-layers is negligibly small. We identify the presence of SiNx to be a key component to trigger the defect activation process. Our measurements suggest that hydrogen released during the high-temperature firing from the hydrogen-rich PECVD-deposited SiNx into the silicon bulk might play a major role in the defect activation process. Additionally, we find that the magnitude of lifetime degradation increases exponentially with increasing temperature during fast-firing. Comparing this result with recently published results from the literature, we conclude that hydrogen-metal complexes are a likely root cause of LeTID.

Unusually high carrier lifetimes of 23.7 ms are measured by photoconductance decay on 1.4-Ohm cm n-type Czochralski silicon wafers passivated using plasma-assisted atomic-layer-deposited Al2O3 on both wafer surfaces. The measured effective lifetimes significantly exceed the intrinsic lifetime limit previously reported in the literature (“Richter limit”). Several important prerequisites have to be fulfilled to allow the measurement of such high lifetimes on Al2O3-passivated n-type silicon wafers: (i) large-area wafers are required to detach the impact of edge recombination via the Al2O3-charge-induced inversion layer, (ii) an exceptionally homogeneous Al2O3 surface passivation is required, and (iii) very thick silicon wafers are needed. Based on our lifetime measurements on n-type silicon wafers of different doping concentrations we introduce an adapted parameterization of the intrinsic lifetime for n-type crystalline silicon and discuss the implications concerning the maximum reachable efficiencies in n-type silicon solar cells, which are higher than previously assumed.

The passivating contacts with poly-silicon on oxide (POLO) are a promising technique to suppress the recombination at the interface between metal and silicon. In order to apply the POLO contacts at the cell front side, however, it is important to make the poly-Si transparent enough to avoid significant parasitic absorption losses. We show in this paper that the transparency of poly-Si films can be significantly improved if they are deposited at temperatures below 550 °C. We therefore investigate the dependence of the transparency of undoped poly-Si films on deposition temperature. The optical parameters (n and k) of those films are determined with spectroscopic ellipsometry and applied in ray tracing simulations. Our study shows that the poly-Si deposition at 530 °C leads to an increase of the photo-generated current density up to 0.8 mA/cm² compared to poly-Si films deposited at 610 °C.

Atomic-layer-deposited titanium oxide (TiOx) is examined for the application as electron-selective full-area contact for n-type silicon solar cells. Although the surface passivation quality of TiOx-passivated n-type silicon wafers is quite poor directly after deposition of the TiOx, we demonstrate that annealing in ambient environment at only 250°C reduces the surface recombination velocity to values below 10 cm/s over the entire cell-relevant injection range. Application of a hydrogen remote-plasma leads to a further reduction in the surface recombination. By combining lifetime measurements with X-ray diffraction (XRD) characterization we demonstrate that the degradation of the passivation by TiOx during annealing at increased temperature (>350°C) is due to the crystallization of the amorphous TiOx into the anatase phase. Finally, we implement our optimized ALD-TiOx as electron-selective full-area rear contacts into n-type silicon solar cells and reach efficiencies up to 18.4% after low-temperature annealing in our first batch. The extracted Srear from IQE measurements of 110 cm/s is well compatible with efficiencies >20%.