Search

In High-Performance mc-Si [1] random grain boundaries, although being recombination active, often enhance material quality by reducing dislocations. With this work, we take a step towards statistical large-scale investigations of crystal defects via a combined analysis of different inline-measurements on as-cut wafers: photoluminescence images for the extraction of recombination-active structures and reflection and infrared transmission images for the extraction of the grain structure. The combined extraction of recombination-active structures and grain structures allows isolating dislocations from grain boundaries for all material types. To discern dislocations from other recombination-active defect structures, an image-processing-based analysis technique has been developed. By applying this separation on wafers from various bricks of our material set, typical developments of grain structure and dislocations can be identified. As a particular application, we investigate the correlation between the development of dislocations in higher parts of the brick and grain size in the lower parts. The results support theory quantitatively: Dislocation ratio in the upper brick part shows a correlation with the square root of the weighted median of the grain size in the lower brick part (R≈0.85). However, the results also show that grain size distribution, in particular grain size homogeneity, has to be considered to account for a stronger distinction between materials.

In this paper we present a new update to PC1Dmod, which extends the original PC1D program by implementing Fermi-Dirac statistics and a range of state-of-the-art models in order to improve the accuracy of c-Si device simulation. In PC1Dmod 6.2 the list of models is further expanded to include a parameterization of incomplete ionization of dopants (Altermatt et al., J. Appl. Phys.100, 113715, 2006), with parameters for phosphorus, boron, arsenic, gallium and aluminum. The results have been verified against a previous implementation of the model. We show that the inclusion of incomplete ionization is of particular importance for moderately doped surface regions with doping densities in the range ∼5 × 10 17 to ∼5×10 19 cm -3 , resulting in a deviation of up to 15% in the simulated recombination current density and sheet resistance. The effect of incomplete ionization is also shown to be more pronounced in devices made from compensated Si material. Furthermore, the default solar spectrum has also been updated in PC1Dmod 6.2, taking advantage of the increased maximum file size to include a more realistic representation. The generation profiles calculated using PC1Dmod 6.2 together with input from OPAL 2 agree well with results from Sentaurus TCAD for both planar and pyramidally textured surfaces, thus facilitating a more direct comparison between PC1Dmod 6.2 and other simulation programs utilizing standard spectra. The simulation comparison is also extended to the electrical performance of the modeled devices, showing good agreement in the calculated short-circuit current density for a range of planar and textured solar cells with varying emitter doping. Finally, several new output options, including emitter saturation current and injection-dependent lifetime curves have been added, enabling a simpler and more direct way for users to access and plot important device properties.

The upgrade of state of the art p-type silicon solar cell production lines to passivated rear side technology (PERC) will be one of the major trends in the next years and new production processes for further cost reduction will continuously gain relevance. In 2007, we have introduced the laser based foil metallization technology “FolMet”: the rear electrode of p-type PERC devices as well as the local contact is fabricated by attaching conventional aluminum foil during the so-called laser fired contact process to the silicon wafer. This process features improved internal optical properties, a huge cost saving potential and a simplified cell production process. In this publication we focus on the acceleration of the laser process, which is together with module assembly issues a remaining challenge towards industrialization. We carried out comprehensive simulations, to better understand the correlation between different laser parameters on melting- and evaporation depth of the 8 μm thin aluminum foil. We determined lower limits for crucial laser pulse parameters to successfully attach the foil onto the substrate and validated these parameters experimentally. According to these results, we set up a system based on a pulsed high power laser featuring repetition rates Frep ≤ 2 MHz with an unique ultrafast polygon scanning system, allowing for scan-speeds vscan ≤ 1000 m/s. Thereby, we demonstrate processing times tpro < 0.8 s for industrial wafer, which corresponds to a reduction in laser process time by the factor of 20 compared to state of the art laser scanning technology.

For the extraction of spatially resolved solar cell parameters, a variety of methods has been introduced in the past years. Nearly all methods use the fact that the local luminescence intensity can be calibrated to local junction voltage. The different methods however use different approaches for the calibration. In this work we discuss the different approaches used throughout literature. We investigate the key assumption of two approaches which assumes that there are no lateral junction voltage gradients at sufficiently low excitation conditions. Our investigation is based on circuit simulations and experimental results on an example cell. We give an explanation for the remaining contrasts in luminescence images taken at these low excitation conditions. A third approach which does not rely on the key assumption is discussed with respect to measurement time in comparison to the other approaches. Finally the role of the short circuit current is discussed and a modification to the approach as known from literature is proposed.

The BOSCO (“ BO th S ides CO llecting and CO ntacted”) solar cell features a double-sided emitter and contact grids on both surfaces. The emitter region on the rear is connected to the front side by diffused vias. The structure allows the use of standard module interconnection technology and favours the use of silicon substrates with low to medium diffusion length and low resistivity for maximum benefit towards other structures, such as Al-BSF and PERC. Within this work, we summarize the latest results on multi-crystalline silicon (mc-Si). Monofacial efficiencies of 17.4% on large-area wafers from 9N (99.9999999% pure) block-cast mc-Si and 16.9% for low-quality 5N upgraded metallurgical-grade mc-Si have been achieved. These values represent a gain of 0.6 to 0.7% abs compared to Al-BSF cells processed in parallel. First tests of bifacial operation under outdoor conditions yield a gain in output power of 13% compared to monofacial operation. This effect makes the BOSCO solar cell concept a promising candidate to enable bifacial operation, even for low-cost wafers of low to medium diffusion length material.

In this paper we present a one-dimensional numerical simulation study concerning the electrical properties of the TCO / a-Si:H(p) / a-Si:H(i) / c-Si(n) hole contact in amorphous / crystalline silicon heterojunction (SHJ) solar cells. Simulations where performed with Sentaurus TCAD from Synopsys considering the implemented barrier tunneling model. The modification of the a-Si:H(p) / a-Si:H(i) / c-Si(n) p/n junction by the opposing TCO / a-Si:H(p) junction and its influence on the device recombination for the relevant working conditions are investigated. We demonstrate the relevance of the effective screening length in the amorphous silicon, which depends on the effective space-charge in the a-Si:H(p) and the work function mismatch at the TCO / a-Si:H(p) interface. We highlight the importance of an improved work function matching as this presents an additional degree of freedom for the design of the SHJ and is an important parameter to reach both a very high Voc and FF.

Significant improvements in the experimental setup of Micro-Raman (μRS) and Micro-Photoluminescence-Spectroscopy (μPL) for solar cell characterization are reported. The lateral resolution of doping density mapping with μRS is improved to below 200 nm. A highly resolved measurement on an aluminum BSF demonstrates the increase in resolution. Furthermore, the characterization of thin silicon layers within the same experimental setup is presented. An excitation source with low penetration depth is used to screen out background signal from the substrate. Finally, a surface topography mapping technique is implemented to extend μRS and μPL to uneven surfaces. With this extension micro cracks in conventionally textured silicon solar cells could be successfully characterized.

Due to boron being present in compensated n-type silicon, minority carrier lifetime degrades under illumination. Lifetime reduction by light-induced degradation up to a factor of 16 was observed for illumination with 100 mW/cm2. In contrast to p-type, the degradation process in n-type does not follow a simple exponential trend. So this degradation process is time dependently investigated in this contribution. It is known that this process depends on a combination of light intensity and sample temperature, hence degradation is investigated independently and a separation of the influences was possible: a relation between light intensity and saturation value of normalized Cz defect concentration Nt*, and between sample temperature during illumination and defect generation rate will be presented. Furthermore it is shown that due to degradation the diffusion length of compensated n-type silicon in the degraded state is as low as in degraded p-type silicon. Finally it is shown that solar cells made from a compensated n-type crystal suffer from efficiency degradation.

Recently, several novel methods have been proposed to image short-circuit current density j sc based on diverse physical principles. This work compares these methods and points out physical limitations, advantages and drawbacks of each approach. One method based on photoluminescence (PL) imaging and two methods based on dark and illuminated lock-in thermography (DLIT / ILIT) are discussed. As a versatile reference technique for j sc mapping, spectrally-resolved light-beam induced current (SR-LBIC) is applied. Experimental results for crystalline silicon solar cells with varying substrate properties, rear-side passivation schemes and process-induced defects are presented. Investigated parameters are quantitative accuracy of local j sc , spatial resolution, measurement time, spectral excitation dependency and calibration. Furthermore, robustness towards locally increased series resistance R s and injection-dependent recombination is discussed along with proneness to artefacts due to local shunts, spatially varying optics and photogeneration, and fitting algorithm artefacts.

Annealing induced silicidation of plated nickel contacts can severly lower the solar cell performance due to deep nickel silicide spikes penetrating the space charge region. This work summarizes several attempts to characterize performance limiting deep silicide structures and determines the influence of different passivation layer structuring technologies on the silicide growth. Reverse biased electroluminescence measurements revealed that the deep nickel silicides occur after an anneal and arelocated along the structured and plated passivation layer openings. Cross-section studies of the plated contacts demonstrated that deep silicide growth is present independently of the applied passivation layer structuring technology. While for laser-ablation the critical silicide structures could be mostly attributed to accelerated silicide growth at laser induced defects, there are also deep silicide structures that appear without any obvious correlation even in the case of defect free wet chemical passivation structuring of the contact openings. By varying the emitter doping profile and the annealing temperature after plating the distance of the space charge region and the surface was found to influence the annealing-induced solar cell performance decrease. Considering the presented results there are three ways to overcome annealing induced degradation of Ni plated solar cells. Either optimized process designs with sufficient contact adhesion and contact resistivity without the need of thermal silicide formation, improved silicide depth control (e.g. Ni source limited growth) or the implementation of a selective emitter design with pn-junction depths well above 1 μm within the contacted area.

This work investigates the influence of physical parameters and of different methods on the analysis of QSSPC measurements with the aim to extract J 0 -values of diffused surfaces. We show that the right choice of not only the parameter-set and physical models but also the method is crucial. The comparison with injection dependent simulation data leads to a deviation of up to 34 % from the absolute J 0 -value for parameters found in literature. In contrast, the recently implemented method in the WCT-120 lifetime tester software [1] which accounts for band-gap narrowing in the substrate agrees within 3 % over the full range of assumed base doping. Furthermore we show that the carrier-lifetime in the substrate shows a significant influence on J 0 -values obtained using the absolute measured recombination and leads to overestimation of up to 50 % even for high base-lifetimes of τ SRH = 1 ms. The implemented parameter-set and analysis method is based on common accepted models and consistent with various simulation tools, and thus allows for the extraction and comparison of injection- and substrate-independent J 0 -values.