Search

27th European Photovoltaic Solar Energy Conference and Exhibition; 1360-1363, Selective emitters for industrial-type silicon solar cells have gained increasing attention in the past years. However, the assessment of the improvement potential of a selective emitter so far mainly relies on the obtained conversion efficiency of the resulting solar cells. In this paper we employ using the dynILM method to measure the emitter saturation current densities on SiNx passivated (J0e,pass) as well as Ag screen-printed (J0e,met) phosphorus emitters on test wafers which are similar to industrial-type silicon solar cells. The J0e,met of the Ag screenprinted emitter increases from 210 fA/cm² up to 570 fA/cm² with increasing sheet resistance. In contrast, the J0e,pass decreases from 147 fA/cm² to 95 fA/cm² with increasing sheet resistance. Compared to previous work on laboratory type cell structures, our results show lower J0e,met values and higher J0e,pass values which might be explained by the different emitter profiles. The measured J0e,pass and J0e,met values allow to calculate the total emitter saturation current density J0e,total. For a homogeneously doped 60 Ω/sq emitter we predict a J0e,total of 140 fA/cm². A selective Emitter with 30 Ω/sq below the metal contacts and 100 Ω/sq below the SiNx passivation results in a J0e,total of 106 fA/cm². The reduction in J0e,total corresponds to an approx 3 to 5 mV higher Voc compared to the homogeneously doped emitter.

27th European Photovoltaic Solar Energy Conference and Exhibition; 1355-1359, The diffusion of phosphorus mediated by phosphorus oxychloride (POCl3) is extensively used in photovoltaics due to its enhanced diffusion speed and to the high controllability of the phosphorus at the wafer surface. However, the POCl3 decomposition in the gas phase in phosphorus pentoxide (P2O5) as well as the subsequent oxidation reactions induced by P2O5 and O2 at the silicon surface are not very well documented and increase the complexity of a realistic simulation of the POCl3 diffusion. It is shown in this paper that the PSG growth model of Ghoshtagore [2] should apply correctly at least in a quantitative relative way to the PSG grown during a POCl3 diffusion. This model allowed to show that the PSG contains a thin layer of 8.6 nm, presumably composed of SiO2, which acts as a diffusion barrier with a subsequent impact on the phosphorus diffusion in silicon. In order to interpret ECV profiles of emitters obtained with various POCl3 flows, a qualitative model based on the Ghoshtagore model is presented. This model emphasizes the role of O2 at the silicon/PSG interface not only as a mediator of the PSG growth, but also as a generator of self-interstitials in silicon that will enhance the extension of the ECV profile tail. It also emphasizes the role of SiP precipitates in reducing the interstitial diffusion in silicon.

27th European Photovoltaic Solar Energy Conference and Exhibition; 1285-1288, An accurate method is described for determining dopant profiles of laser-doped areas on multicrystalline silicon wafers from electrochemical capacitance-voltage measurements. The method is based on the calibration of dopant profiles using the combination of contact area and depth measurements of the etched area and is extended to uneven (i.e. textured) surfaces. The contact area is obtained by geometrical measurements and surface area factor measurements. The total etch depth is obtained by subtracting height profiles before and after etching. The calibrated dopant profiles are compared to secondary ion mass spectrometry measurements and found to be in excellent agreement.

27th European Photovoltaic Solar Energy Conference and Exhibition; 1335-1338, A newly developed characterisation method at Fraunhofer ISE allows a precise evaluation of via contact formation by determining the via resistance for metal wrap through (MWT) solar cells. Based on a previously presented via characterisation method for test structures, the new measurement setup is able to automatically characterise via-contacts directly on a solar cell. Due to its flexible design, the setup is adjustable to any cell size and any contact arrangement. The modified four point probe measurement setup with flat electrodes on the front and spiky electrodes on the rear of the cell facilitates the measurement and precise the alignment on a each MWT-via contact. The new device is an important characterisation instrument for the evaluation and optimization of via-contacts directly on cell level.

27th European Photovoltaic Solar Energy Conference and Exhibition; 1262-1264, We have investigated the potential to accelerate three-dimensional numeric simulation of silicon solar cell using thread parallelism. The device simulated is a rear side passivated cell with rear point contacts (PERC). The optical and electrical behaviour of the device was simulated with Sentaurus Device (formerly dessis). We show that the simulation run-time on a four socket Opteron 6168 machine is reduced down to 24% compared to the run-time on quad-core machine. Furthermore, limits of time reduction by varying the number of threads up to 48 are studied. Thereby, the number of threads for the optimum use of the hardware resources is determined.

27th European Photovoltaic Solar Energy Conference and Exhibition; 1343-1346, To improve PV cell efficiency and reduce costs, intensive developments have been done concerning the productivity of the furnace. The use of numerical tool gives precious information with regards to these developments. Within the SEGSI project (2007-2011), financed by French National Research Agency, a complete 3D numerical model of a 60kg crystallization silicon furnace has been developed using ANSYS 13.0 software. It is a direct solidification furnace dedicated to laboratory experiments with two distinct heating zones. The model enables to calculate the thermal of the furnace, the solidification rate and hydrodynamics of the silicon which are important in order to control and improve the quality of the ingot during crystallization process. The purpose of this work is to validate the time dependent heat transfer and hydrodynamic model with regards to experimental results. This numerical tool is now available and is integrated in the development of new crystallization process.

27th European Photovoltaic Solar Energy Conference and Exhibition; 1405-1408, Measuring current-voltage characteristics of back contacted solar cells is not trivial in the case of large area devices. Prior to the introduction of the highly efficient ZEBRA cell, which was developed on 156×156 mm2 substrates, measurement issues were of little concern as for smaller cells only one pair of Kelvin contacts was used. The results presented in this paper are relevant for the correct measurement of current-voltage characteristics of large area interdigitated back contacted solar cells. Additionally, our findings can also be applied to the measurements of back contacted solar cells with different architectures. In the present contribution we show that by fine tuning the measurement chuck alone the measured absolute energy conversion efficiency of the ZEBRA cell under steady illumination conditions increases by at least half of a percentage point, with a “gain” in the measured power exceeding three percentage points. This is due to a correct measurement of the fill factor.

27th European Photovoltaic Solar Energy Conference and Exhibition; 1230-1233, Aluminium local back surface field (Al-LBSF) solar cells are currently being investigated for the next generation of high-efficiency industrial silicon wafer solar cells. These Al-LBSF solar cells feature dielectric passivation layers with local contact openings covered by an Al layer at the rear side of the cell. These local contacts openings are typically formed by laser ablation of the dielectric layer prior to the application of Al. Laser ablation can, however, lead to damage to the underlying silicon and it is important to study the impact of this damage on the final solar cell efficiency. In this work we have fabricated homogenous-emitter Al-LBSF solar cells using ns and ps lasers for opening the rear dielectric. Cell efficiencies of up to 19.0% were achieved with both lasers. The results clearly indicate that post-laser etching using KOH is a crucial step to improve the efficiency of the cells, especially if the dielectric openings were formed by ns laser ablation. However, post-laser KOH etching can significantly affect the contact formation of Al-LBSF cells, resulting in a drop in the fill factor. The impact of post-laser KOH etching on laser-induced damage removal and contact formation for Al-LBSF solar cells was investigated in detail by means of optical microscopy, scanning electron microscopy (SEM) and photoluminescence imaging.

27th European Photovoltaic Solar Energy Conference and Exhibition; 1471-1476, There is a great interest in the testing accuracy, uncertainty and quality control of IV curve measurements and corresponding parameters extraction both at laboratory and industrial level. In this work, two different setups for the measurement of solar cells’ I-V curve in the cell calibration laboratory of CIEMAT are described. Main characteristics and differences between these systems, as the testing procedure (flash mode vs quasisteady conditions), the measurement instrumentation used and the contacting methods available, are explained. Both systems are being inter-compared in order to ensure their quality, to allow cross-reference between their results and to improve their reliability. Experimental data corresponding to a set of commercial-type multi-Si solar cells measured with both systems are presented and preliminary deductions on the role of the temperature and the contact forms are pointed out.

27th European Photovoltaic Solar Energy Conference and Exhibition; 1242-1245, This work aims to determine the origin of the high series resistance of a solar cell detected by means of an electroluminescence image. With this purpose, the main components of the series resistance, i.e. busbar, finger, contact between gridlines and silicon, and emitter sheet resistances have been determined through resistance measurements in a four-wire sensing configuration. For the spatially resolved determination of the finger, contact, and emitter sheet resistance contributions, an infrared pulsed laser was used to electrically isolate the required areas of the solar cell, meaning the destruction of the finished solar cell. The transmission line model was used for the spatially resolved measurement of both contact and emitter sheet resistance contributions. The spatially resolved series resistance data have been compared with the electroluminescence image, showing a good correlation. The solar cell series resistance has been calculated as the average of the local series resistance data, and has been compared to the value determined from the I-V curve fit providing, again, a satisfactory result.

27th European Photovoltaic Solar Energy Conference and Exhibition; 1325-1328, The understanding and therefore the optimization of n+-emitter formation in crystalline silicon using POCl3-diffusion requires a more detailed knowledge of the dopant source: the PhosphoSilicate Glass (PSG) layer. The growth of PSG during the phosphorus diffusion process depends on several process parameters, such as temperature, duration, POCl3-N2 and O2 gas flows. In this work, we compare the uncertainties in various methods for PSG thickness measurements: by an Atomic Force Microscope (AFM), a profilometer, a Scanning Electron Microscope (SEM) and a spectroscopic ellipsometer. We then quantify how the PSG thickness is influenced by the process parameters. We also measure the total amount of phosphorus (P-dose) in the PSG layer using Inductively Coupled Plasma Optical Emission Spectrometry (ICP-OES) and quantify the P dose in dependence of the process parameters as well. Finally, by combining both the measured PSG thickness and the dose, we successfully determine a lower limit for the phosphorus concentration in the PSG layer. It is, depending on the process parameters, between 7x1021 cm-3 and 1.2x1022 cm-3, which is a remarkably narrow range. These results will help to improve the phosphorus diffusion model by considering both the PSG growth behavior and PSG composition, and so will facilitate the development of a predictive model for the POCl3 diffusions process.

27th European Photovoltaic Solar Energy Conference and Exhibition; 1439-1442, In highly efficient amorphous silicon/crystalline silicon heterojunction (a-Si:H/c-Si) solar cells, the c-Si wafer is passivated by a nanometer-thin buffer layer, which is undoped amorphous silicon. Here, we report on the systematic measurement of the passivation quality (minority carrier effective lifetime) by photo-conductance decay and of the band bending in c-Si using the planar conductance technique. The thickness of the buffer layers is varied. An analytical model to calculate the band bending in c-Si is presented; it aids in understanding the influence of the buffer layer on the band bending. We find that when the buffer layer thickness increases the passivation quality increases and the band bending decreases. Therefore, we suggest that an optimum has to be found to reach good interface defect passivation and a high band bending.

27th European Photovoltaic Solar Energy Conference and Exhibition; 1429-1431, The efficiency of industrial bifacial n-type cells is mostly limited by the loss in VOC after firing of screen printed metal pastes. We have investigated the mechanism responsible for this loss by varying the metal fraction on either side of the cell and determined the actual influence of surface geometry, diffusion layers and pastes. The experimental results indicate a more detrimental influence of metal pastes on the boron emitter side. This can be attributed to more aggressive pastes and less shielded contacts as in the case of phosphorous diffused emitters.

27th European Photovoltaic Solar Energy Conference and Exhibition; 1251-1254, In this work, we investigate the surface passivation of n+ and p+ emitters by SiOx/AlOx stacks deposited by inline plasma-enhanced chemical vapour deposition (PECVD) in a fully industrial process. We show that the SiOx/AlOx stacks provide an excellent level of surface passivation on both p+ and n+ emitters with various sheet resistance values. On n+ emitters, the surface passivation by SiOx/AlOx stacks is found to be at least comparable to PECVD SiNx. These results can be explained by the relatively low interface trap density and moderate negative fixed charge density in the stack after an industrial firing process. In addition, with increasing SiOx thickness in the stack, the surface passivation quality on boron diffused p+ emitters is found to degrade, while it improves on phosphorus diffused n+ emitters. This indicates that the surface passivation by SiOx/AlOx stacks can be effectively controlled via the SiOx layer thickness. These important findings can be directly applied in various high-efficiency solar cell structures. When a simultaneous passivation is required on both p+ and n+ silicon by the same passivation film, the SiOx/AlOx stack could turn out to be an excellent candidate.

27th European Photovoltaic Solar Energy Conference and Exhibition; 1373-1375, This study described the degradation by variance of surface and quantum efficiency (QE) in multicrystalline silicon solar cell (mc-Si solar cell) after vibration test which was simulated under the condition of land & air transportation. The vibration test follows some procedures from an unleashed draft of one developing international Standard. Mechanical defects on the surface of solar cell were obviously inspected by recognizable patterns evolving, shown from the method of electroluminescence (EL) imaging. The patterns were classified into 7 types by symptoms of crack/micro-crack. On the other hand, quantum efficiency variances of cell were measured in incident photon-to-electron conversion efficiency (IPCE), i.e. IPCE performance worst-case drop was found indeed in spectrum as 23.7%. After vibration test, total breakage rate is near 1% and some specific cells had 3% power loss. EL images showed the sufficient evidences of defect’s growth due to mc-Si cells experienced vibration response during transshipment simulation. Furthermore, this applied research provided a common vibration test from a prenormative Standard, and two nondestructive test method, EL and QE/IPCE to evaluate the damaged spot on mc-Si solar cells. Result shall also be considered as one key point to improve transportation package as well in the future.

27th European Photovoltaic Solar Energy Conference and Exhibition; 1448-1452, In recent years, significant performance improvements of industrial silicon solar cells have been achieved by introducing novel structures such as selective emitters, where a lower emitter sheet resistance at contacted areas and higher sheet resistance at illuminated areas is realized, as well as by employing selective front surface fields (FSF) and local back surface fields (BSF). Fast and powerful measurement systems are necessary to assess the benefit and performance gain related to these new structures and optimize them regarding surface and bulk lifetime of the generated charge carriers, diffusion length and defect clusters. A high speed and accuracy of the measurement system enable fast in-line investigations in an industrial environment as well as obtaining a fundamental understanding of the underlying physical processes that allows for an optimization of the solar cells. To investigate locally dependent structures such as selective emitters studied in this paper, spatially resolving measurement systems are required. To this end, spatially-resolved, lock-in thermography at high temperatures is used in the present work. Selective emitters on passivated and unpassivated textured wafers are studied to demonstrate the capability of the measurement technique for investigations of structured doped regions (such as FSF or BSF) in silicon solar cells.

27th European Photovoltaic Solar Energy Conference and Exhibition; 1255-1257, Series resistance is a valuable measurement parameter for production lines as well as laboratories. It is usually described with a single lumped resistance number that is derived from the equivalent circuit for a solar cell. However, due to the distributed nature of current flow in the cell, lumped resistance is not constant over the operating region of the cell. The aim of this work is twofold: To study the behavior of lumped resistance under different operating conditions by varying irradiance intensities and cell operation point, and on the other hand, analyze the available measurement methods. The work is performed by measuring IV-curves and a Suns-Voc curves of 28 industrial 156mm c-Si solar cells under irradiance intensities ranging from 400 to 1200 W/m2 and determining series resistance values for different cell currents. The applied methods are the comparison of IV and Suns-Voc curve (Suns) and the comparison of IV curves under different intensity levels (IEC). For lumped resistances of less than 0.7 Ohm-cm2, found to be typical for most of the measured cells, the intensity dependence of resistance is negligible and does not contribute to the accuracy of the methods used. Therefore, IEC method performs better when using larger rather than smaller intensity variations. The operating current dependence of lumped resistance is more pronounced even with low resistance cells, being on average over 30% higher near maximum power point than at Voc. Comparison of the data from Suns and IEC methods showed that both methods result in similar values, typically within +/-10%, for cells with low Rs. In cells with higher Rs, the IEC method will begin to deviate from Suns method. This follows directly from the intensity dependence that starts to show up at larger series resistance values.

27th European Photovoltaic Solar Energy Conference and Exhibition; 740-743, In recent years, lasers have been widely used in fabrication processes for solar cells. One of the most promising applications of lasers in solar cell manufacturing is laser doping. Laser doping creates locally doped regions, such as those required to create selective emitters or locally doped rear-contact structures, via an industrially feasible manufacturing process. Mapping the dopant distribution of laser doped cross-sections is an effective way to rapidly qualify laser doping. However, the laser doping line width is only tens of microns, which makes characterization quite difficult. Until now, such investigation involves expensive and complex methods, for example SIMS and EBIC. This paper describes a novel cheap method with simple sample preparation to image cross-sectional dopant distribution of laser doped regions using dopant related contrast observed in secondary electron microscopy (SEM) images. To our knowledge, this is the first time that this method has been applied on delineating laser doping cross-sections. It is found that this technique can be used for different dopant sources and different doping methods under relatively wide FESEM work condition window. This method is proved to be a fast technique to investigate the influence of laser parameters on laser doping. In conjunction with alkaline etching, we also demonstrate that this method can effectively evaluate the risk of metallization shunt near the edges of dielectric film windows opened by laser.

27th European Photovoltaic Solar Energy Conference and Exhibition; 735-739, We have developed a method to qualify multi-crystalline silicon (mc-Si) wafers that are being used in a production process. An optical image of an etched wafer is made. This etching can be a standard industrial acid etching for mc-Si wafers as is commonly used for saw damage removal and simultaneous iso-texturing. Digital image processing is then applied to identify the number of dislocations and their distribution over the wafer. This information is used as input for a cell performance prediction model, where the performance is characterized by the open circuit voltage (Voc) or the efficiency. The model can include various levels of sophistication, i.e. from using an average density of dislocations to the full spatial resolution of the dislocations in a 2D simulation that includes also the metallization pattern on the cell. The predicted performance is then evaluated against pre-selected criteria. The possibility to apply this optical qualification method in an initial stage in the production enables early rejection of the wafers, further tailoring of the cell production process or identification of instabilities in the production process.

27th European Photovoltaic Solar Energy Conference and Exhibition; 1293-1295, An analysis method was presented to measure the contact resistivity of rear-side local openings to reveal the interaction of laser-ablation/metallization. Different laser-ablating conditions were compared with different pastes, firing conditions, and film stacks. No unique laser recipe can be utilized for whole metallization conditions from the contact resistivity point of view. Using an appropriate laser ablation recipe of nanosecond laser combined with a proper metallization condition, the contact resistivity of ~58.59 mΩ-cm2 for line- pattern could be comparable to the value of 40-55 mΩ-cm2 achieved by the picosecond laser.

27th European Photovoltaic Solar Energy Conference and Exhibition; 1368-1372, 120 μm thick multicrystalline Si solar cells with SiO/SiN stack rear-side passivation and Al point contacts are locally characterized using the CELLO technique. By combining different standard measurement modes—e.g. using different laser wavelengths, variation of global illumination intensity, and by measuring at different points along the I-V curve—effects due to parasitic shunting can be separated from other rear side effects like inhomogeneous surface passivation (i.e. due to stack or contacts) or series resistance. An increased series resistance at the rear side is found to improve the rear surface recombination behavior and thus induces a boost in the photocurrent response as an unexpected seemingly positive side effect. As a consequence the series resistance distribution of the rear side has to be taken into account for the correct interpretation of photocurrent data measured using light with long wavelengths.

27th European Photovoltaic Solar Energy Conference and Exhibition; 1339-1342, The performance of photovoltaic (PV) modules in the field is significantly affected by the ambient conditions it is subjected to during its operational life. The ambient conditions typically considered the most important in this respect are temperature, wind speed and moisture. In terms of temperature, the impact can be manifold: • The overall module temperature impacts directly the module performance: increased temperatures at the cell level lead to a drop in VOC which negatively impact the overall performance; • Local variations in illumination and/or temperature (temporally as well as spatially) can cause mismatch between the cells connected in series; • Thermal cycling mechanically stresses the module and may cause cracking and delamination of the different layers in the module (cell, interconnect metallization, encapsulant, glass, back sheet) over time. In short, the operational temperature has an important impact on the potential energy yield of PV modules, either through suboptimal performance of the different cells in the module, or through a reduction of the operational lifetime (reliability issues related to thermal cycling). Here, the aim is to understand by 3D FEM modeling and experimental calibration the thermal behavior in detailed spatial and fine temporal regime of modules to determine potential cell- and module-level improvements. In particular we want to build a thermal model for the MWT c-Si wafer-based modules.

27th European Photovoltaic Solar Energy Conference and Exhibition; 1319-1324, Minority charge carrier lifetimes are measured in multicrystalline silicon using a series of different techniques; photoluminescence-imaging, microwave photoconductance decay, carrier density imaging and quasi steady state photoconductance, both with front and back located coils.The different measurements tools are compared and evaluated with respect to trends through the height of an ingot and spatial characteristics within wafers. Mechanical polishing results in lower surface recombination than chemical polishing of as-sawn wafers, and offer more details using the lifetime mapping tools. The different techniques are found to have different advantages with respect to spatial resolution, tolerance for variations in surface properties and calibration of the measurements. All techniques describe the general trend in the ingot, with some variation in the absolute values.

27th European Photovoltaic Solar Energy Conference and Exhibition; 848-851, We present a small perturbation quasi-steady-state microwave detected photoconductance decay method that is significantly improved using novel quality of decay control technique. As a result, measurements of excess carrier decay lifetime, eff.d and of the steady-state effective lifetime, eff.d are unified in one parameter-free self-consistent approach. The approach shows excellent quantitative correlation with Sinton QSSPC. Advantages of the method relate to wafer mapping. The emitter saturation current, J0, maps reveal the common presence of high J0 passivation defects that can degrade cell performance. In an automated corona-charging-measuring sequence, the approach enables quantitative assessment of the state of field-effect passivation using a novel parameter; charge to saturation, Qts. It is believed that passivation defect mapping and field-effect assessment can benefit passivation engineering for high efficiency cells.

27th European Photovoltaic Solar Energy Conference and Exhibition; 1435-1438, When using photoluminescense (PL) to extract solar cell parameters, a situation that allows the simplification of the continuation equation is preferred; i.e. steady / quasi-steady-state (QSS) or free decay conditions are desirable. However, the behaviour of the cell under a periodic excitation treasures information in its modulus and in its phase that has not been explored yet by means of its PL response. This work analyses the behaviour of the PL response of a cell under periodic illuminations for a wide range of frequencies. Parameters such us the ideality factor of the diode or the effective lifetime of the cell can be extracted for several injection levels. The theoretical approach presented in the first section is then validated with simulations and, afterwards, bv experimental results.

27th European Photovoltaic Solar Energy Conference and Exhibition; 1289-1292, Two-dimensional simulations are performed on solar cell designs with locally contacted, passivated rear surfaces with various contact structures to assess the physical cause of inversion-layer shunting that has been experimentally observed in such devices. Six variations on the local rear contact structure are investigated to determine the conditions under which shunting occurs and how the shunting affects the flow of current through the device. In addition, passivation using an accumulation layer is compared to inversion-layer passivation. Since a p-type substrate is assumed in this investigation, high-quality silicon nitride (SiNx) films are used to provide inversion-layer passivation, while high-quality aluminum oxide (Al2O3) films are used for accumulation-layer passivation. It is found that when the local rear contact is insufficiently separated from the passivating dielectric layer, parasitic shunting can occur — but only if the passivating dielectric induces an inversion layer. The effect can be avoided by using a passivating dielectric that induces an accumulation layer instead. This implies that a great degree of process control in rear contact formation is required for SiNx passivation of p-type substrates and suggests that Al2O3 could bring higher production yields.

27th European Photovoltaic Solar Energy Conference and Exhibition; 1409-1412, In this work, we used a finite element model to simulate a novel nickel silicide formation process based on Excimer Laser Annealing (ELA). We evaluate the robustness of this new metallization process scheme that is selfaligned and compatible with shallow emitters. Study structures have been irradiated by ELA at several densities of energy, and also modelled into the simulation program. The results were compared and we suggest an explanation for the nickel silicidation process. The simulation results show a strong dependence of the energy density threshold to melt silicon with the thickness of the studied capping layer: Ni or SiNx. This is directly related to the reflectance of the stacks at the wavelength of the UV laser, but also to the thermal conductivity K and heat capacity cp of the materials. Phosphorus profiles after ELA can be predicted but the model still needs a calibration to match perfectly with the experience. A process path for self-aligned NiSi alloying with Excico laser system through full sheet ELA on real solar cells has been found. This has been confirmed with the realization of PERC cells with conversion efficiency up to 20%.

27th European Photovoltaic Solar Energy Conference and Exhibition; 1423-1428, We have used the Atlas device simulator within the Silvaco simulation framework to simulate high efficiency silicon based solar cells. We have focused on two different solar cell concepts that have already be shown to be compatible with large scale production and high efficiency; namely the Back-Contact Back-Junction (BC-BJ) silicon solar cell and the Double Heterojunction (DHJ) silicon solar cell. For these two solar cell designs we have investigated the device performance given different material qualities and wafer thicknesses. The performance of both solar cell designs is closely related to the silicon material quality. A low material quality will significantly reduce the efficiency of the cells. The effect is most pronounced for BC-BJ solar cells where the efficiency is completely killed for low quality wafers. The wafer thickness is also important for the cell performance. Thin wafers produce decent efficiencies even for medium quality silicon. Thicker wafers hold better potential for high efficiencies, but are more dependent on material quality. In order to realize the high efficiency potential, the thick wafers need to have effective minority carrier lifetimes (MCL) at the order of milliseconds.

27th European Photovoltaic Solar Energy Conference and Exhibition; 1234-1237, Electroplated Ni-Cu based front contacts for c-Si are deposited on patterned ARC and during the process unwanted deposition on unpatterned areas can occur which may lead to shading and possible shunt formation. In this work, study of unintended plating on ARC is carried out by surface deposition analysis and carrier life time studies wherein the front side of solar cells are exposed to electroless Ni bath and electroplating Cu bath. It has been observed that for exposure to Ni bath without any pre-treatment, deposition of miscellaneous ions present in bath occurs on the ARC which have a shading effect, but tend to recover after annealing. For samples pretreated with 1% HF and PdCl2 based solution before Ni plating, Ni deposition was observed with change in reflectance and passivation properties of the ARC. From QSSPC life time studies under UV illumination it was observed that decrease in lifetime increased from 20% to 58.5% as the bath temperature increased from 70°C to 90°C. A substantial decrease of 22mV in implied Voc resulted after exposure to Ni bath at temperature of 90°C. Dispersed copper deposits were observed for exposure to Cu bath which did not adversely impact the solar cell properties.

27th European Photovoltaic Solar Energy Conference and Exhibition; 714-718, There are numerous types of Grain Boundaries (GBs) present in multicrystalline silicon wafers, such as 3, 9, 27 and Random Angle (RA) GBs. Small-Angle Grain Boundaries (SA GBs) are also common, and often occur in locally dense networks, creating regions of high recombination. These GBs can limit the efficiency of multicrystalline silicon solar cells, especially when they are decorated with impurities. However, the electrical properties of GBs are not fully understood. Here we examine the recombination activities of different types of GBs by photoluminescence (PL) imaging, combined with Electron Back-Scattered Diffraction (EBSD) data to determine the GB types. We quantify the recombination strength of GBs with two different methods - the PL contrast at the GBs, and the GB induced recombination current. Our results indicate that SA GBs are the most recombination active GBs, followed by 27 GBs, while the majority of 3 and 9 GBs are recombination inactive. The influences of injection level, bulk lifetime and surface condition on the recombination strength of GBs are also discussed.

27th European Photovoltaic Solar Energy Conference and Exhibition; 1296-1298, The interdigitated back contact (IBC) solar cell which is commercialized by Sunpower is known for its outstanding conversion efficiency in silicon-based solar cells. It is now manufactured in AUO SunPower Malaysia in mass production. To further reduce the cost of the cell, it is necessary to keep improving the cell performance to a higher level. The purpose of the work is to improve the cell efficiency by modifying the geometry design of IBC cell with the assistance of the computational simulation. In this study, the impact of the cell geometry design such as the emitter, BSF, undoped gap, contact area and the diffusion pitch will be evaluated through the PC2D simulation [Basore PA, Cabanas-Holmen K IEEE Journal of Photovoltaics 2011;1(1): 72–77]. From the two-dimensional simulation analysis, it is recommended in order to gain higher cell efficiency, larger emitter area, with smaller BSF and adequate undoped gap are preferred for the current IBC cell design. Smaller diffusion pitch and contact area also helps the improvement of fill factor and the efficiency. The simulation results are verified by the experimental splits. From the experimental results, the efficiency of 5” n-type IBC cell is improved from 22.2% to 22.4%. It is consistent with our simulation.

27th European Photovoltaic Solar Energy Conference and Exhibition; 723-728, In this contribution two models explaining the recombination behavior of the different defect types introduced in a previous publication are discussed. For this we have applied detailed microscopic investigations, temperature-dependent EBIC measurements and root cause analysis on the nm-scale. We could separate the large amount of recombination active defects into four different classes: type-AP, type-AG, type-B and type-C. It was shown that iron precipitates are responsible for the recombination behavior at type-AP sites leading to an interface recombination. At type-AG regions, most likely decorated with iron impurities, deep defect states EM are responsible for the recombination behavior and can be described by the model of Kveder et al. The recombination processes at Type-B defect structures are either due to a low impurity level or a shallower position of the defect level EM both explaining the observed high intensity of defect luminescence. It is concluded that the recombination process at type-C defects is similar to type-AG defect structures. The temperature dependency of the recombination behavior is in good agreement with the model described by Kveder et al. for all defect types.

27th European Photovoltaic Solar Energy Conference and Exhibition; 1394-1397, The present work is related to effects originating from the spatial character and therefore referring to the distributed resistance of a silicon solar cell. Two effects are regarded: Firstly, the bending of the illuminated IVcurve particular at open-circuit-conditions towards a lower voltage due to an increased metal/emitter resistance. Secondly, enhanced fill factor losses due to lateral current paths in case of back contact solar cells. A simple approach is presented by taking into account a second one-diode-model in addition to a conventional standalone one-diodemodel. The key idea is simple: By separating these models by a resistor, diodes are forward biased differently as it is the case in real operation conditions and internal current flow leads to a distortion of the measurable global IV-curve. We show that results achieved by the so called “distributed-two-diode-model” give already comparable results to far more complex two-dimensional finite element simulations utilizing Sentaurus TCAD.

27th European Photovoltaic Solar Energy Conference and Exhibition; 1200-1203, In order to accurately model the behavior of a solar cell, the values of the series and shunt resistances must be found for the device under study. The authors in previous work proposed a new method to find these values at maximum power point using the Newton-Raphson method and current-voltage equations expressed using the Lambert W-function. This method was shown to be accurate and requires only data provided by the manufacturer. In this paper, we present the results of an experimental study carried out to further validate this method and compare values obtained against another method. Using a multi-crystalline cell exposed to outdoor illumination in Hamilton, New Zealand and a precise source measuring unit, the current-voltage curves were obtained at 1000 W/m2 of illumination. Simulated current-voltage curves generated using resistance values calculated from each of the two methods (i.e. proposed and alternative) were compared against the experimental curves. Based on a root mean square analysis over the entire current-voltage curve, the proposed method was found to yield the greatest accuracy.

27th European Photovoltaic Solar Energy Conference and Exhibition; 1413-1418, In industrial screen-printed aluminium-back-surface-field (Al-BSF) silicon solar cells the standard process for passivating the front surface and optimising its optical properties is the plasma-enhanced chemical vapour deposition (PECVD) of a silicon nitride anti-reflective coating (SiN-ARC). During PECVD of the front side SiNARC in an inline, tray-based reactor, the process plasma can wrap-around to the rear side causing a parasitic SiN deposition. We found, that the thickness of this parasitic SiN layer is in the order of 10 to 30 % of the thickness of the front side SiN-ARC. A corresponding SiN layer is investigated concerning its impact on the saturation current density of the aluminium-doped p+-region formed by two commercially available aluminium pastes. The SiN layer caused a significant decrease in thickness of the p+-region while its saturation current density increased. This implies a loss in open circuit voltage, which is calculated on the basis of a solar cell model and demonstrated experimentally.

27th European Photovoltaic Solar Energy Conference and Exhibition; 1477-1481, The charge carrier mobility is a key material parameter for photovoltaic applications. The dependences of the mobility on bulk dopant density, on injection level and on temperature are well known in the case of noncompensated crystalline silicon. Still, it remains unclear in the case of dopant compensation. A wide variety of methods, such as combined infrared radiation and voltage measurements or combination of quasi-steady-state photoconductance and quasi-steady-state open-circuit voltage measurements were employed in the past to measure the mobility. However, most of these methods require contacts and a relatively complicated structure of the sample. Recently, a new contactless method to determine the carrier mobility sum, based on a comparison between quasisteady- state photoconductance and transient photoconductance decay measurements was proposed. However, it seems that this method is currently limited to injection levels in the 1×1015 to 1×1016 cm-3 range. In this paper we present a novel method to determine the mobility sum, based on combined photoconductance and photoluminescence measurements. A mobility sum in the injection level range 1×1013 to 1×1017 cm-3 can be determined using the proposed method.

27th European Photovoltaic Solar Energy Conference and Exhibition; 861-865, We compare the degradation behaviour of rear-passivated high efficiency silicon solar cells processed from mc-, Cz-, and cast-mono Si materials. We observe that cells processed from mc-Si unexpectedly show a stronger degradation than cells processed from cast-mono or Cz-Si. Using Ga-doped mc-Si, we prove that the degradation observed on mc-Si cannot be attributed to B-O complex formation or FeB pair dissociation. For castmono, a similar degradation characteristic as for mc-Si is observed, however, the degradation is much less pronounced, making this material class a favourable candidate for processing of high efficiency rear passivated solar cells.

27th European Photovoltaic Solar Energy Conference and Exhibition; 1213-1216, We present a numerical simulation model for multicrystalline Si (mc-Si) solar cells, which predicts cell behavior from lifetime mappings of mc-Si wafers to a precision of about 3 mV in open-circuit voltage, Voc. From a scientific point of view, the model allows us to answer commonly discussed questions such as: which is a suitable method for averaging lifetimes in the quality control of mc-Si wafers, or the role of lower lifetimes in mc-Si cell performance. It is shown that mathematical averaging procedures of wafer lifetime mappings can predict cell’s Voc performance only within a rather broad range (30 mV for the arithmetic mean value, 7 mV for the geometric and 4 mV for the harmonic mean value). From an industrial point of view, the simulation model enables cell manufacturers to predict cell behavior from lifetime images of mc-Si wafers and hence help in deciding on improvement strategies.

27th European Photovoltaic Solar Energy Conference and Exhibition; 1299-1302, This paper aims at developing a design strategy of multi-layer antireflection coatings (ARCs) to improve the optical performance for multicrystalline silicon (mc-Si) solar cells. We proposed a three-dimensional (3D) ellipsoid surface model to represent the concave-like morphology formed by acidic texturization. Monte Carlo ray tracing algorithm was applied to simulate the interaction between incident light and wafer. After that, we fixed the parameters of the 3D model by fitting the simulated reflectance to the measured result of a typical industrial mc- Si wafer. The reflectance and transmittance of multi-layer ARC over 300 nm to 1100 nm were calculated by employing transfer-matrix-method (TMM). The validity of the simulation was proved by the agreement between simulated and measured reflectance of silicon nitride (SiNx:H) single-layer ARC. Finally, we studied a triple-layer stack, including bottom and intermediate layers of SiNx with high and medium refractive indexes respectively and a top layer of SiO2. Genetic algorithm was employed to optimize the thicknesses of the three layers. Calculated result showed that the designed stack can increase the short circuit current by 1.12 mA/cm2.

27th European Photovoltaic Solar Energy Conference and Exhibition; 755-758, In this paper we investigate the structure of rear local contacts of passivated emitter and rear solar cells (PERC) including the formation of voids using scanning acoustic microscopy (SAM). Whereas standard techniques as e.g. optical microscopy and scanning electron microscopy (SEM) require time-consuming sample preparation and result in localised information only, it is shown that using non-destructive SAM voids can be depicted spatially resolved over a complete cell. This is proven by comparing cross-section polishes of contact areas with SAM images. Additionally, we present the analysis of local back surface fields (BSF) using Raman microscopy. Using this technique, the thickness of a BSF can be investigated more easily compared to SEM. We found similar thicknesses of the BSF regardless of whether a filled or non-filled (void) local contact area.

27th European Photovoltaic Solar Energy Conference and Exhibition; 857-860, The root causes of breakdown of multicrystalline silicon solar cell well below the theoretical limit are investigated in this contribution. The main focus of this paper is on avalanche, or so called type-3, breakdown sites in alkaline texturized solar cells. However, the similarities and differences of the same breakdown type in isotextured solar cells are described as well. Using EBIC charge collecting contrast of p-n junctions at FIB prepared cross sections it is found that possible enhanced phosphorous diffusion at certain line defects in the material leads to spikes of the p-n junction despite of a flat surface of alkaline textured solar cells. Avalanche breakdown occurs at these spikes similar to the avalanche breakdown at etch pits in isotextured solar cells, but at slightly higher absolute reverse voltages as compared to the latter ones. It is assumed that the differences of the breakdown onset voltages are due to the slightly different p-n junction curvature shapes in alkaline and isotextured solar cells.

27th European Photovoltaic Solar Energy Conference and Exhibition; 1188-1190, Accurate physical modelling to assist and support research and development is of increasing importance for scientific institutions and for industry. Modelling is required for many different tasks, starting from fundamental questions concerning the basic working principles of photovoltaics, up to predicting the performance of a PV system. For these tasks many different approaches and methods are available. This paper gives an overview of the recent activities at the Solar Energy Research Institute of Singapore (SERIS) in this field. We discuss four different examples, related to four different techniques: Circuit modelling of series resistance losses in PV modules, device modelling of wafer solar cells to investigate passivation mechanisms, and optical modelling of scattering and periodic structures for thin-film silicon solar cells.

27th European Photovoltaic Solar Energy Conference and Exhibition; 1347-1350, To optimize the last high temperature step of a standard solar cell fabrication process (the contact cofiring step), the aluminium gettering is incorporated in the Impurity-to-Efficiency simulation tool, so that it models the phosphorus and aluminium co-gettering effect on iron impurities. The impact of iron on the cell efficiency will depend on the balance between precipitate dissolution and gettering. Gettering efficiency is similar in a wide range of peak temperatures (600-850 ºC), so that this peak temperature can be optimized favoring other parameters (e.g. ohmic contact). An industrial co-firing step can enhance the co-gettering effect by adding a temperature plateau after the peak of temperature. For highly contaminated materials, a short plateau (< 2 min) at low temperature (600 ºC) is shown to reduce the dissolved iron.

27th European Photovoltaic Solar Energy Conference and Exhibition; 1401-1404, In industrial solar cell production of c-Si solar cells rejects in the range of several percent depending on the manufacturing line are inevitable. Making use of this resource could significantly increase the yield of production leading to lower manufacturing costs. Laser post-processing of defective cells has shown a high potential for repairing cells with local defects like shunts and was proposed for large scale testing. In this work first results are presented for an automated laser repairing and post-processing system that could be placed at the end of a manufacturing line. 450 rejected standard Cz-Si solar cells from a large cell manufacturer were characterized by means of optical inspection, IV- and luminescence measurements. The common cell defects are presented and the electro- and photoluminescence technique is discussed for being used as characterization method for an automated repairing system. Results of first manually laser repaired cells are shown and a preliminary technical draft of the automated system is exposed.

27th European Photovoltaic Solar Energy Conference and Exhibition; 732-734, The measurement technique presented in this paper, provides a very sensitive detection of cracks in polycrystalline solar wafers and cells, combined with photoluminescence or electroluminescence measurements. The crack detection is based on illuminating light, entering the wafer (or cell) from the edges and being kept inside the bulk by total internal reflection. This light is scattered by cracks and inclusions, which then appear with large contrast on a dark background. An infrared line-scan camera is used as detector. Such, we are able to detect much smaller cracks than with conventional equipment. Luminescence signals can be recorded simultaneously with the same camera in the same setup. For the case of photoluminescence, only the scan line needs to be irradiated by the optical excitation source. This means that the necessary optical power can be lower and / or exposure times can be shorter. The combination of several measurement channels (crack detection, luminescence) with different light/excitation sources and a single camera is achieved by a special application software for a particular frame grabber. By this means, mechanical stability and electric quality properties can be tested simultaneously.

27th European Photovoltaic Solar Energy Conference and Exhibition; 1419-1422, During EL-imaging of PV-Modules, a typically fixed current is applied. The module heats up to a steady state. This process can take up to 30 minutes. However, local differences in thermal mass, heating power (e.g. due to local current densities), shunts and similar effects influence this process. As the current density distribution thereby changes, this changes are visible in EL imaging by taking a series of short-exposure images. Thereby not only thermal caused connections or disconnections of cracked cell segments can be exhibited, but also local variations of cell properties.

27th European Photovoltaic Solar Energy Conference and Exhibition; 1312-1318, A novel technique for imaging the recombination current pre-factor Jo of heavily doped surface regions, ubiquitous to mainstream silicon solar cells, is introduced. This technique utilises photoluminescence in a low injection regime, allowing measurement of test structures with low and moderate resistivities, which are unattainable by the conventional Kane and Swanson method [1]. The procedure is fast and simple requiring only one photoluminescence image and no photoconductance measurement (after an initial calibration). The potential of the technique is demonstrated on surface-passivated phosphorus diffusions with sheet resistances in the range of ~15 – 120 Ω/sq. A comparison is made with both high and low injection photoconductance decay (PCD) measurements and a recently proposed high injection Jo imaging technique (based on Kane and Swanson theory) [2, 3].

27th European Photovoltaic Solar Energy Conference and Exhibition; 1278-1284, We compared POCl3 and BBr3 furnace diffused dopant source to phosphorus and boron implanted dopant as well as BF3 and arsenic plasma implant dopant sources for selective emitter junction formation with laser annealing and emitter field junction formation with RTA, furnace and microwave annealing all either before or after SiN/ARC. Localized laser melt (LLM) annealing for liquid phase selective emitter dopant diffusion at 3, 5 and 6.5 J/cm2 resulted in textured wafer junction depths of 1.8um, 2.6um and 3.5um respectively and Rs as low as 7.0Ω/. LLM after SiN/ARC resulted in a 2x degradation in Rs dopant activation. Nitrogen SIMS depth profile showed high levels of N in the E19 to E20/cm3 incorporated in the melt region while oxygen incorporation in the melt region was much lower at

27th European Photovoltaic Solar Energy Conference and Exhibition; 1225-1229, For the high-efficiency type of silicon solar cells, hetero-junction with intrinsic amorphous thin-film (HIT) solar cells has been confirmed to achieve 23.7% by Sanyo Electric. From the theoretical viewpoint, how to further enhance the efficiencies of the silicon solar cells for the next generation is a most critical issue. Based on the a-Si heterojunction structure, very high Voc over 740 mV have been demonstrated. Therefore, Interdigitated back contact silicon heterojunction solar cells (IBC-SHJ) have the potential for high open voltage and high short current density due to all back contact design. Several groups reported the progress in this field, and the efficiencies ranges from 13.9% to 20%. However, IBC based on heterojunction solar cell still performs low F.F. values in the recent years. Therefore, we propose a new structure based on boron diffused emitter combined with IBC-SHJ to greatly improve the conversion efficiencies, especially for reducing series resistances. The key issue to improve efficiency is to resolve the interface resistance between c-Si and a-Si. For n-type wafer, the large valence band offset between c-Si and a-Si results in large resistive loss and lower efficiencies. Boron-doped a-Si is also difficult to have denser carrier density. Therefore, by incorporation the boron emitter between i-a-Si and c-Si can greatly reduce valence band offset and enhance the silicon solar efficiency.

27th European Photovoltaic Solar Energy Conference and Exhibition; 1329-1334, Recent trends show that screen printing technology still dominates c-Si solar cell metallization with over 90% share. Process simplicity and low cost are the major attributes to this, however screen printing has some shortcomings with respect to fine lines, precision and quality. The purpose of this paper is to develop a practical, cost effective and simple model together with evaluation procedure that can assist in reducing test time, material requirement and complexity. To achieve this, we created a mathematical model to predict screen printability in respect to screen mesh material, meshing parameters and EOM (Emulsion Over Mesh). Finally a new test procedure is introduced to evaluate screen printed metallization. Both the model and the procedure where tested on 2 commercial pastes in combination with several screen design parameters. The model proved practical and user friendly and was verified by testing on 2 different stainless steel (SS) mesh screens and vecry mesh screen (V-screen) each with different EOM. Combining the model with our proposed test procedure showed that Vscreens out performed SS screens for similar screen parameters on both pastes tested.