Impact of copper on light-induced degradation in Czochralski silicon PERC solar cells

Abstract

Both multicrystalline and Czochralski (Cz) silicon substrates are known to suffer from various mechanisms of light-induced degradation (LID), including copper-related LID (Cu-LID). Past studies on Cu-LID have mostly been performed on unprocessed wafers, omitting the impact of the solar cell process on the copper distribution. Here, we carefully contaminate Cz-substrates of different quality with different amounts of copper and process the substrates into complete industrial Cz-Si PERC solar cells, reaching a comprehensive mapping of the impact of Cu-LID for the PV industry. The results show that both the copper contamination level and Cz crystal quality are critical factors affecting the extent of Cu-LID. Most importantly, we show that copper can result in significant concentrations in the bulk of the finished PERC cells after being exposed to only trace surface contamination. Consequently, even a small local copper contamination area (~ 3-4 cm 2) is sufficient to induce strong LID in the full-sized (156×156 mm 2) cell parameters, resulting e.g. in ~7% relative efficiency loss during light soaking. The corresponding short circuit current density decreases by up to a factor of two in the contaminated areas.

Low-T anneal as cure for LeTID in Mc-Si PERC cells

“…The starting wafers for both sample types were commercial 156×156 mm 2 boron-doped highperformance multicrystalline silicon wafers, the initial bulk lifetime was around 490 s and the substrate material was known to suffer from typical LeTID. All process steps followed typical industry standards for PERC cells [11], [15]. The implied-VOC were otherwise identical to the cells but no metal contacts were deposited.…”

“…The maximum temperature is limited to 300 °C in order to have a minimal impact on the junction or contact properties as the anneal times of interest are relatively long. Finally we will study if the degradation and regeneration during DA resemble metal precipitation kinetics, since earlier studies indicate that some metals may still be present in the bulk after the last high temperature step [4], [12], [13], [14], [15] although a strong evidence has been reported for hydrogen playing a crucial role as well [9], [16].…”

Vertically integrated modeling of light-induced defects: Process modeling, degradation kinetics and device impact

“…As a proof of concept for our modeling approach, we use Cz-substrates intentionally contaminated with Cu processed into PERC devices. Details of the contamination process and device results can be found in [26]. We use the background lifetime, i.e., the lifetime limit imposed by defects besides the modeled light-induced defects, as a simulation parameter to match the initial efficiency (absolute efficiencies not shown due to confidentiality).…”

“…For comparison, we also simulated the expected BO-kinetics, using the parametrization used by Schön et al [28] for compensated n-type silicon that was also recently used by Vahlman et al for p-type silicon [19] The parametrization is based on the so called fast and slow recombination centers (FRC and SRC, respectively), with their properties modeled as in [29][30][31][32]. The used boron and interstitial oxygen concentrations were 7•10 15 cm -3 and 9.5•10 17 cm -3 , respectively, as measured from the as-grown wafers prior to cell processing [26]. Fig.…”

“…In contrast, the simulated BO-LID is a less suitable candidate for the main LID-mechanism, because the modeled simulation extent is less than 1% rel , although the experimentally observed degradation is approximately 7% rel . Note that the [B] and [O i ] values used were chosen to be the highest values within the measurement uncertainty [26], suggesting that the actual BO-LID is more likely to be less than more severe than what simulated here.…”