• Energy Research

Abstract Recent progress in the electronic quality of high-performance (HP) multicrystalline silicon material is reported with measurements and modeling performed at various institutions and research groups. It is shown that recent progress has been made in the fabrication at Trina Solar mainly by improving the high excess carrier lifetimes τ due to a considerable reduction of mid-gap states. However, the high lifetimes in the wafers are still reduced by interstitial iron by a factor of about 10 at maximum power point (mpp) conditions compared to mono-crystalline Cz wafers of equivalent resistivity. The low lifetime areas of the wafers seem to be limited by precipitates, most likely Cu. Through simulations, it appears that dislocations reduce cell efficiency by about 0.25% absolute. The best predictors for PERC cell efficiency from ingot metrology are a combination of mean lifetime and dislocation density because dislocations cannot be improved considerably by gettering during cell processing, while lifetime-limiting impurities are gettered well. In future, the material may limit cell efficiency above about 22.5% if the concentrations of Fe and Cu remain above 1010 and 1013 cm−3, respectively, and if dislocations are not reduced further.

Abstract Light induced degradation can cause a severe loss of efficiency on multi-crystalline PERC solar cells (mc-LID) of more than 10%rel. In this work, the kinetics of the mc-LID annealing process and its temperature dependence is analyzed. It is shown that the initial efficiency can be partly restored by annealing the cell in the dark. However, the degradation process is not completely reversible and the degradation rates of the first and subsequent degradation cycles are different. Furthermore, lateral variations of the degradation are investigated. Four regions showing a quantitatively different degradation behavior are identified. Mc-LID of the rear contacts shows similar degradation as for standard back surface field solar cells. The degradation of grain boundaries is weaker than intra-grain degradation and thus of particular interest for root cause analysis. Decorated grain boundaries are dominated by other recombination mechanisms suppressing the appearance of mc-LID. Physical explanations for these results of a laterally different degradation behavior and an increased degradation rate after annealing are discussed.

Abstract If the ratio of two open circuit photoluminescence ( V oc -PL) images taken at two different light intensities is displayed, some grain boundaries (GBs) may show up as bright lines. This indicates that these special GBs show distinct injection intensity-dependent recombination properties. It will be shown here that this results in an apparent ideality factor of the light emission smaller than unity. The effect is reproduced with numerical device simulations using a usual distribution of defects in the band gap along grain boundaries. Quantitative imaging of this apparent luminescence ideality factor by PL imaging is complicated by the lateral horizontal balancing currents flowing at open circuit. The local voltage response of an inhomogeneous solar cell at different injection levels under open circuit is modelled by Griddler simulations, based on PL investigations of this cell. The evaluation of V oc -PL images at different illumination intensities allows us to conclude that the apparent luminescence ideality factor at the special GBs is about 0.89, whereas in the other regions it is between 0.94 and 0.95. Reverse bias electroluminescence showed no pre-breakdown sites, and hyperspectral PL imaging showed only in one of the investigated GBs particular defect luminescence. TEM investigations of two GBs, one showing distinct injection intensity-dependent recombination and the other one showing none, revealed that the investigated special GB is a large-angle GB whereas the GB not showing this effect is a small-angle GB.

Abstract The failure and loss analysis of solar cells relies on the precise determination of their electrical performance parameters. Light emitting diodes (LEDs) used as light sources in solar simulators provide a number of advantages compared to conventional Xenon-based solar simulators. In this work, we present an approach to rapidly measure the quantum efficiency and the reflectivity that takes the spectral broadening of the LEDs fully into account. Our approach does not rely on any specific shape of the spectral peaks and can thus be applied to any type of quasi-monochromatic light source. The two proposed methods yield valuable additional information on the solar cell performance and material properties that cannot be obtained by conventional solar simulators. Using LED solar simulators, they are completely in-line capable as they can be performed in less than one second.

Abstract In the last years, significant progress has been made regarding an understanding of light induced degradation at elevated temperature observed on PERC solar cells (LeTID). Nevertheless, the detailed root cause is still under discussion. Latest results show that a similar degradation occurs by annealing lifetime samples in the dark without carrier injection. In this work, we show that mc-Si PERC cells degrade and recover at high temperature without carrier injection. As the lateral appearance and the recovery behaviour agree with what is known about LeTID, it is likely that the same defect is observed. However, even after recovery the treatment in the dark does not result in LeTID stable cells. A subsequent illumination leads to a further power loss. This subsequent degradation differs from the first degradation in its kinetics and its lateral appearance. Based on these results it is concluded that two recombination active defect states are activated by LeTID. These recombination active defect states can be distinguished by annealing the samples without carrier injection before illuminating the samples. However, a high temperature anneal activates also additional defects, which might lead to a more pronounced degradation. Thus, a high temperature treatment is not recommended for LeTID testing neither as substitution nor as pretreatment prior the LeTID test.

AbstractA new generation of solar simulators is based on light emitting diode illumination sources. These measurement systems offer the opportunity to adjust the light spectrum as close as possible to the AM1.5G reference spectrum. Additionally, they provide the technical basis to combine power measurements with a spectral resolved analysis. Such an application is the determination of the quantum efficiency, which results in valuable additional cell information such as front and rear surface recombination, diffusion length, or emitter dead layer thickness. In this work, a fast method to determine the external quantum efficiency (EQE) of a solar cell using a LED solar simulator is presented. The measurement time of our LED-EQE approach could be reduced to less than half a second as no mechanical parts such as monochromators are involved. Due to the finite spectral band-width of the LEDs an adapted data analysis approach has been developed, which leads to results that show excellent agreement with standard EQE measurements.

Abstract Light induced degradation can lead to a severe efficiency loss in multi-crystalline PERC solar cells depending on bulk Si material properties and solar cell processing parameters, known as mc-LID or LeTID. Various defect models have been suggested which indicate a clear distinction to BO- and FeB-LID mechanisms. Cu is known to cause light induced degradation and therefore is one of the possible causes of mc-LID in PERC cells. However, until now, a direct microstructural proof of its presence is still missing. In this contribution, we investigate mc-LID sensitive PERC cells which show the typical lateral appearance of mc-LID where structural defects, such as grain boundaries, show a reduced degradation. Investigations of grain boundaries from front and rear side with respect to the recombination activities (LBIC) in correlation to the crystalline structure (XRD Laue mapping) indicate that gettering at grain boundaries reduces degradation. Furthermore, enhanced rear recombination at grain boundaries and scattered local spots of µm size is detected. At these regions with damaged rear passivation, Cu-containing microscopic particles are unveiled by microstructural investigations (SEM) and elemental micro-analysis (EDX). Target preparation (TEM) shows a Cu-filled channel that connects the Cu-containing microscopic particles and the silicon bulk. These observations indicate that the presence, diffusion, and precipitation of Cu might play a role in the mc-LID defect formation.

Abstract High-performance multi-crystalline silicon material (HPmc-Si) dominates the market for casted p-type silicon. Solar cells made from HPmc-Si material might suffer from light induced degradation due to the so called sponge-LID mechanism. In this work, we present a kinetic sponge-LID model showing that the degradation follows a pairing reaction involving two reactants. This implies that sponge-LID is based on a different reaction scheme compared to the well-known models for boron-oxygen- or iron-boron-degradation. Based on our model, degradation rates are investigated reading the influence of temperature and illumination on the degradation. Finally, we present statistical results implying that Sigma-3 grain boundaries are less affected of the degradation than other grain boundary types. Using a detailed spatially resolved analysis of the effective carrier diffusion length, the different behaviour of grain boundaries and intra-grain regions is quantified.