Search

Further increasing the conversion efficiency of silicon solar cells by advanced solar cell concepts like SHJ (Silicon Hetero-Junction), PERT (Passivated Emitter Rear Totally diffused), IBC (Interdigitated Back-Contact), requires a highly controlled silicon surface in terms of metal impurity density. The goal of this study is investigating the deposition property of different metals in acidic cleaning solutions and consequently determining the impact of these impurities on the effective minority lifetime of the crystalline silicon solar cell material. The deposition behaviour of impurities is mathematically described with an adsorption isotherm theory. Once attached at the silicon surface, the metal impurity can diffuse into the silicon bulk material during a high temperature process step like diffusion, passivation or contact annealing (firing). First, we have studied the adsorption isotherm behaviour of prominent metal impurities (Fe, Cu, Ti, Co, Cr and Zn) in acidic liquid onto a silicon surface. Moreover, the impact of metal impurities on effective minority carrier lifetime in commercially available n-type Cz-silicon (3–5 Ω cm, 239 cm2) is determined. From controlled contamination experiments we can extract an exchange volume, that provides the minimum amount on wet chemical cleaning solution that is required to clean a silicon surface from a given high metal impurity level to a specified low metal impurity level. This exchange volume plays a crucial role by defining feed and bleed recipes on wet chemical cleaning tanks in solar cell manufacturing. A second parameter extracted is the effective minority carrier lifetime degradation of the metal impurities. After deposition on the surface, the impurities are driven into the silicon bulk by high temperature processes. Depending on the lifetime degradation capacity, a relation of effective minority carrier lifetime versus surface contamination level could be identified for each metal impurity. We found Co, Fe and Cu having a strong impact on effective minority carrier lifetime degradation, e.g. 2e10/cm2 Fe surface concentration can reduce the effective minority carrier lifetime by 75%, while Ti, Zn and Cr are more moderate and only high surface concentrations above 1e12/cm2 lead to effective minority carrier lifetime degradation of more than 20% compared to the non-contaminated wafers.

In this work, we improved the performance of the MWT-PERT solar cells focusing on increasing their V-oc by combining the MWT concept with n-PERT technology. The impact of different post-laser treatments on the via surface morphology and the via passivation was investigated. KOH texturing can partially remove the laser damage in the via and reduce the via SRV to 1000 cm/s. With optimized post-laser treatment and via passivation, an average V-oc of 685mV was achieved for our large-area n-type MWT-PERT cells. Front Ni/Cu plating and rear Ag and Al screen-printing were used for the metallisation, the compatibility of this hybrid metallisation scheme was studied. Using industrial solder-through interconnection technology, the cell was integrated in a laminate reaching a one-cell module efficiency at 20%. The reliability of the one-cell modules was preliminarily investigated in an extended reliability test. (C) 2017 The Authors. Published by Elsevier Ltd. The authors gratefully acknowledge the financial support of the PV4FACADES project (Solar-Era.net) and Imec's industrial affiliation program for Photovoltaics (IIAP-PV).

We have studied the effect of different cleanings at device level for nPERT (passivated Emitter, Rear Totally diffused) cells with a rear epitaxial emitter. Cleanings based on sulfuric acid and peroxide mixtures (SPM), which are effective but also expensive, were used as reference. We found that simplified cleanings such as those based on sulfuric acid and ozone mixtures (SOM), and those based on water and ozone (IPV), are giving similar results at device level as the expensive SPM cleaning, leading to similar J0e, J0, FSF, and lifetime values. The implied VOC values of corresponding test devices fabricated using the simplified cleanings are at the same high level as the test devices using more expensive SPM cleanings. Actual nPERT devices with moderately doped epitaxially-grown emitters with a constant doping profile (box profile) showed average efficiencies above 22.0% for both SOM and IPV cleanings. Very high VOC values of up to 695 mV and implied VOC values of up to 730 mV were shown for cells with a flat rear surface. Finally, the best nPERT cells with rear epitaxial emitter and SOM cleaning showed efficiencies up to 22.5%, as externally confirmed by ISE CalLab.

Recently, large area hybrid silicon heterojunction cells (SHJ) with nickel/copper (Ni/Cu) plated front contacts were shown to benefit from a simpler processing sequence than n-PERT cells featuring a boron diffused rear emitter. Hybrid refers here to the combination of a diffused front surface field (FSF) and a SHJ rear emitter. In this work, further process simplifications for such hybrid SHJ cells are evaluated. Based on reflectance and adhesion considerations, stacks of tin-doped indium oxide and copper (ITO/Cu) and chromium/silver (Cr/Ag) are tested as alternative to the previously chosen ITO/Ag rear contact stack. Using ITO/Cu/Al as rear contact stack, average efficiencies of 20.8±0.2% are presented on 6-inch wafers. We also demonstrate that the excimer laser annealing step previously employed to form nickel silicides at the front side can be skipped without comprising solder tab adhesion quality.

Tunneling aluminum oxide (Al2O3) layers are proposed as a candidate for hole collecting passivated contacts. These ultra-thin Al2O3 films ($\approx 1.5\text{nm}$) are deposited on boron diffused ($113\Omega/\square$) hydrophilic surfaces to act as metal-insulator-semiconductor (MIS) contacts. Record values of emitter recombination current densities (35–40 fA/cm2) and contact resistivity ($\approx 15\mathrm{m}\Omega\cdot \text{cm}^{2}$) were measured for this type of contact. Most importantly, all of these processes are carried out using industrially relevant tools on Cz wafers with non-polished surfaces.

This paper presents the different strategies which have been studied at imec in order to reduce the front surface recombination of our monofacial high-efficiency rear-junction (RJ) n-PERT diffused cells. These devices feature: a laser doped selective FSF on a lightly POCl 3 diffused front side; front contacts by Ni/Cu plating followed by immersion Ag plating; a diffused rear boron emitter passivated with an Al 2 O 3 /PECVD SiO x stack and sputtered rear AISi contacts. Each of the three main parameters contributing to the front recombination has been tackled, starting from the reduction of the metal front recombination (by improving the laser doping process), then the reduction of the contact fraction with the implementation of busbar free cells and finally, a further reduction in the recombination of the passivated areas, by improving the thermal treatment of the dielectrics. This study shows that an average 23% efficiency is within reach for large-area double-side contacted RJ PERT cells, with voltages over 700 mV and an average FF of 80.7 % .

Hydrogen passivation of laser-induced defects (LasID) is shown to be essential for the fabrication of laser-doped solar cells. On first-generation laser-doped selective emitter solar cells where open-circuit voltages were predominately limited by the full-area back surface field, a 10-mV increase and 0.4% increase in the pseudo-fill factor were observed through hydrogen passivation of defects generated during the laser doping process, resulting in an efficiency gain of 0.35% absolute. The passivation of such defects becomes of increasing importance when developing higher voltage devices and can result in improvements in implied open-circuit voltage on test structures up to 50 mV. On n-type PERT solar cells, an efficiency gain of 0.7% absolute was demonstrated with increases in open-circuit voltage and pseudo-fill factor by applying a short low-temperature hydrogenation process using only hydrogen within the device. This process was also shown to improve the rear surface passivation, increasing the short-circuit current of approximately 0.2 mA/cm 2 of wavelengths from 950 to 1200 nm compared with that achieved using an Alneal process. Subsequently, an average efficiency of 20.54% was achieved.

In this paper, we evaluate p-type passivated emitter and rear locally diffused (p-PERL) and n-type passivated emitter and rear totally diffused (n-PERT) large area silicon solar cells featuring nickel/copper/silver (Ni/Cu/Ag) plated front side contacts. By using front emitter p-PERL and rear emitter n-PERT, both cell structures can be produced with only a few adaptations in the entire process sequence because both feature the same front side design: homogeneous n+ diffused region with low surface concentration, SiO2/SiNx:H passivation, Ni/Cu/Ag plated contacts. Energy conversion efficiencies up to 20.5% (externally confirmed at FhG-ISE Callab) are presented for both cell structures on large area cells together with power-loss analysis and potential efficiency improvements based on PC1D simulations. We demonstrate that the use of a rear emitter n-PERT cell design with Ni/Cu/Ag plated front side contacts enables to reach open-circuit voltage values up to 676 mV on 1–2 Ω cm n-type CZ Si. We show that rear emitter n-PERT cells present the potential for energy conversion efficiencies above 21.5% together with a strong tolerance to wafer thickness and bulk resistivity. Copyright © 2014 John Wiley & Sons, Ltd.

We investigate passivation of the aluminum-silicon interface by thin aluminum oxide (Al 2 O 3 ) layers grown by thermal atomic layer deposition (ALD) on HF-last silicon surfaces. We first report effective lifetimes of Al 2 O 3 -passivated n- and p-type silicon wafers as a function of the number of ALD cycles. Then, we present saturation current density and contact resistance measurements of aluminum contacts on n/n + and n/p + junctions, passivated with a selection of the investigated layers. Our results show that aluminum contacts on n + silicon can be successfully passivated with thin Al 2 O 3 layers without compromising contact resistance. However, we did not observe significant contact passivation for acceptable contact resistance in the case of Al 2 O 3 passivated aluminum contacts on p + silicon. We explain our experimental results from the asymmetry between conductance and valence band offsets of Al 2 O 3 on silicon.

In this paper, we investigate the mechanisms leading to rear reflectance losses in i-PERC type solar cells by means of reflectance measurements, scanning spreading resistance microscopy (SSRM), and transmission electron microscopy (TEM). Using 2 μm sputtered (PVD) AlSi12.7% (eutectic composition) it is found that the interaction between Si present in the sputtered mixture and the rear dielectrics upon contact firing leads to a large drop in rear reflectance even without any laser ablated contact openings. On the other hand, in the case of local Al-BSF formation with pure PVD Al, it is shown that the presence of Si on top of the rear dielectrics, coming from Si diffusion into the Al layer with alloying, leads to parasitic absorption which contributes almost entirely to the measured drop in rear reflectance.

In order to minimize the considerable loss of Si material during the processing of a silicon photovoltaics (PV) substrate into a solar cell, the optimization of the saw damage (SD) removal is an important step. The PV cell process therefore requires a routine metrology solution suitable to probe the SD present in the substrate. Though cross-sectional Transmission Electron Microscopy is typically used for this purpose, this technique is not suited for in-line measurements. It is furthermore destructive, time-consuming and limited to very small sampling volumes. In this paper, we demonstrate the capabilities of the Photomodulated Optical Reflectance (PMOR) method for the in-line measurement of SD. PMOR is a laser-based pump-probe technique, wherein the probe laser measures the reflectance of the sample (DC reflectance) as well as the change in reflectance (AC reflectance) which the pump laser has induced by optically injecting excess carriers and heat. As PMOR is a local and nondestructive technique, it offers a very promising approach for the simultaneous characterization of roughness and SD. We demonstrate the versatility of this approach and the response of the DC and AC reflectances when studying wafers which have undergone successive etchings to remove the SD sequentially (© 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim)

Recently, the interest for the implementation of local ion implantation for emitter formation has been renewed, as it results in high cell efficiencies [1] , while the amount of process steps is reduced. After implantation, a thermal diffusion (annealing) is necessary for the removal of the ion-induced damage, the activation of dopants and the formation of the desired profile shape. The goal is to change the implanted retrograde near surface (10-100 nm) inactive profile, into a damage free, box-like shaped, and fully active emitter (or BSF) with a low saturation current density (J 0e ). Here, we study different post-implantation annealing schemes, to achieve high performance boron emitters for n-type IBC solar cells. We show experimentally, that the use of oxygen gas during the boron anneal has an impact on the electrical performance (J0e) of the emitter. On the other hand, the obtained J0e values are higher than those obtained by the standard diffusion approach. Hence, in order to understand these results and optimize the implant/anneal process, the modeling of the profile shape and the damage evolution would be extremely valuable. In this work, we report on our initial simulation results, and we demonstrate the impact of oxygen on the evolution of the profile shapes of the implanted B-emitters after anneal, taking into account boron diffusion kinetics mechanisms such as Transient Enhanced Diffusion (TED) and Oxidation Enhanced Diffusion (OED). Moreover, we monitor the residual damage and we make a first effort to correlate the increase on the J0e values with the residual damage levels. Throughout this work, we adopt the process simulator tool Sentaurus Process.

In this work we investigate the potential of inline single side wet-etch processing for rear side emitter removal and edge isolation for large area p-type Passivated Emitter and Rear (PERC) silicon solar cells. We focus on the integration of this process in flows leading to shallow emitters and screen printed front contacts and to deep, highly ohmic emitters in association with Cu contacts. The inherent advantage of this process, beside high throughput, is the external gettering performed by full emitter etch at the rear side. This is particularly relevant in the case of two-step emitter formation process, where removal of phosphorus diffused layers is most effective when performed prior to the high temperature drive-in step. Another effect that we make use of is the front-side emitter etch back performed by the etch bath vapors, which can be used to tune the phosphorus surface concentration and thus the emitter sheet resistance. Using this process, we report efficiency up to 19.9% for CZ large-area, homogeneous-emitter, screen printed cells, and 20.4% for CZ, large-area Cu plated cells.

This paper investigates by means of finite-elements and SPICE simulation the further improvement of solar cell performance of PERC-type cells by introduction of industry-compatible solutions, such as a 120 sq. phosphorous emitter with a lower surface concentration than that of current 60 sq. standard emitters and the replacement of the silver screen-printed front-side metallization by narrow, dense electro-plated copper lines on lowly resistive Ti or NiSi contacts.

Excellent front surface passivation is one of the key issues for interdigitated back contact (IBC) silicon solar cells. To create a better understanding of passivation properties and find the most important issues when developing passivation schemes for various injection conditions, an advanced simulation model is needed. The software package Sentaurus TCAD is applied in this work. The main input parameters of the model are the interface state density, the amount of fixed charges, the capture cross sections and the activation energies which have been determined experimentally as reported previously. As an example, simulation results of the advanced model of the recombination at the Si-SiNx interface were presented and compared to the measured data.

We examine the effectiveness of hydrogen passivation as a function of defect type and microstructure at grain boundaries (GBs) in multicrystalline silicon. We analyze a solar cell with alternating mm-wide bare and SiNx-coated stripes using laser-beam-induced current, electron backscatter diffraction, X-ray fluorescence microscopy, and defect etching to correlate pre- and post-hydrogenation recombination activity with GB character, density of iron-silicide nanoprecipitates, and dislocations. A strong correlation was found between GB recombination activity and the nature/density of etch pits along the boundaries, while iron silicide precipitates above detection limits were found to play a less significant role. Copyright © 2010 John Wiley & Sons, Ltd.

We present large-area n-type PERT solar cells featuring a rear boron emitter passivated by a stack of ALD Al2O3 and PECVD SiOx. After illustrating the technological and fundamental advantages of such a device architecture, we show that the Al2O3/SiOx stack employed to passivate the boron emitter is unaffected by the rear metallization processes and can suppress the Shockley-Read-Hall surface recombination current to values below 2 fA/cm2, provided that the Al2O3 thickness is larger than 7 nm. Efficiencies of 21.5% on 156-mm commercial-grade Cz-Si substrates are demonstrated in this study, when the rear Al2O3/SiOx passivation is applied in combination with a homogeneous front-surface field (FSF). The passivation stack developed herein can sustain cell efficiencies in excess of 22% and Voc above 685 mV when a selective FSF is implemented, despite the absence of passivated contacts. Finally, we demonstrate that such cells do not suffer from light-induced degradation.

This Letter discusses an important difference between positively charged SiO2 and negatively charged Al2O3 rear-passivated p-type Si solar cells: their illumination level dependency. For positively charged SiO2 rear-passivated p-type Si solar cells, a loss in short circuit current (JSC) and open circuit voltage (VOC) as a function of illumination level is mainly caused by parasitic shunting and a decrease in surface recombination, respectively. Hence, the relative loss in cell conversion efficiency, JSC, and VOC as a function of the illumination level for SiO2 compared to Al2O3 rear-passivated p-type Si solar cells has been measured and discussed. Subsequently, an exponential decay fit of the loss in cell efficiency is applied in order to estimate the difference in the energy output for both cell types in three different territories: Belgium (EU), Seattle and Austin (US). The observed trends in the difference in energy output between both cells, as a function of time of the year and region, are as expected and discussed. (© 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim)

The ongoing trend to decrease the wafer thickness in the fabrication of silicon solar cells increases the number of possible cracks. Therefore, an in-depth understanding of the influence of cracks on solar cells is necessary. In this work, we investigate silicon solar cells with cracks by employing three camera-based spatially-resolved techniques and use the data to quantitatively calculate local-current voltage parameters. Cracks mainly influence the recombination current density of a silicon solar cell. This fact is clearly shown by drastically increased recombination in the space charge region. This recombination reduces parameters like fill factor, open circuit voltage and the cell efficiency. The comparison of the solar cell data before and after the formation of cracks shows a 0.2 % absolute efficiency loss in the global current-voltage parameters and up to 1 % absolute efficiency loss in the local current-voltage analysis at the crack positions.

This work characterizes p-type Silicon surface passivation using a high-k material (Al 2 O 3 or HfO 2 ) combining capacitance voltage (CV) and lifetime measurements.

This paper presents the results of the integration of a high sheet resistance (120 Ω/sq) homogeneous emitter with an industrial feasible process flow on large area p-type CZ-Si PERC solar cells with plated Ni-Si/Cu front contacts. Focus is put on the junction isolation performed by either applying a mask prior to POCl 3 diffusion or by selectively removing the rear side emitter in an inline 1-sided wet-chemical etching tool after diffusion. A metallization sequence [1,2] applying Ni and Cu plating, is performed to contact the moderately doped emitters. The best experimental split in this comparison resulted in excellent average conversion efficiency of 20.2%.

This work proves that blistering is the partial delamination of a thick enough Al 2 O 3 layer caused by gaseous desorption in the Al 2 O 3 layer upon thermal treatments above a critical temperature: the Al 2 O 3 layer acts as a gas barrier and bubble formation occurs. First, using an atmospheric pressure rapid thermal processor with an atmospheric pressure ionization mass spectrometry, desorbing species upon heating of Si/Al 2 O 3 samples are identified: evident desorption peaks are observed around 400 °C for all spectra. The spectrum for m/e = 18, an indication of H 2 O, illustrates that gaseous desorption from Al 2 O 3 and from the Si substrate itself continues up to 600 °C and 700 °C, respectively. Also, it is shown that in the case of a 30 nm Al 2 O 3 layer, blistering starts at same annealing temperatures as gaseous desorption begins. In the case of a thin enough ( 2 O 3 film, blistering does not show. To complete the proof, elastic recoil detection measurements clearly show that after annealing a thick Al 2 O 3 film above 400 °C the H content is higher near the c-Si interface as compared to the near surface. Fortunately, effective lifetime and capacitance voltage measurements show that 5 to 10 nm Al 2 O 3 layers can still be adequate passivation layers after being annealed in N 2 environment at temperatures up to 500–700 °C: (i) interface trap densities (D it ) can remain below 1×1011 cm−2 and (ii) fixed charge densities (Q f ) stay negative and in the order of −3×1012 cm−2 Random local Al back surface field (BSF) solar cells, fabricated using a blistered film as rear surface passivation and no additional contact opening step, clearly show that random local BSFs are created upon firing of a blistered rear passivation layer covered by metal. Therefore, it is clear that blistering should be avoided, since it will reduce the overall rear surface passivation. The key to avoid blistering is using 5 to 10 nm Al 2 O 3 passivation layers and performing an annealing step prior to capping and co-firing. Al 2 O 3 /SiN x passivated local Al BSF p-type Si solar cells are made using an out-gassing step with temperatures up to 700 °C. For these cells, the reduction in blistering and hence improvement in rear surface passivation is clearly reflected in the gain in average Voc as a function of out-gassing temperature.

26th European Photovoltaic Solar Energy Conference and Exhibition; 1052-1057, We observe a relationship between bias dependence and the process flow applied to produce passivated emitter rear contacted (PERC) cells. The first process flow utilizes a deposited silicon oxide layer, for both diffusion mask and passivation purposes. The second process flow uses the same silicon oxide diffusion mask, but it is removed after diffusion and a freshly deposited silicon oxide layer is used for passivation purposes. Cells that have been passivated with fresh silicon oxide result in a large bias dependence at ultra low bias levels below 10-1 sun, but saturate performance beyond 10-1 sun. Cells that have been passivated with the silicon oxide diffusion mask, an increase bias increases performance. For this process, we also observe a relationship between illumination level dependence and the diffusion recipe which densifies the oxide. With a 90 ohm/sq emitter there is an increase in collection in the UV wavelength range compared to a 60ohm/sq emitter. What is unexpected is that, in the case of the 90 ohm/sq device, there is also an increase in collection of wavelengths above 700nm under no bias compared to the 60ohm/sq device. The cells studied in this paper are fabricated using 148.25 cm2, 160μm p-type Cz-Si wafers with screen printed Ag front contacts and are rear-side passivated with a deposited rear silicon oxide/silicon nitride stack and area locally contacted with Al. The highest efficiency of the cells studied is 19.2 % with a Voc of 637mV, Jsc of 38.2 mA/cm2 and a fill factor of 79.1%

Firing impacts on surface passivation provided by a SiO2 and SiO2/SiNx stack with evaporated Al films are studied by capacitance-based techniques on MIS capacitors. For devices with insulator layers consisting solely of as-deposited SiO2, the densities of either interface states (Dit) or fixed charges (Qfc) are hardly influenced by firing. Capping the SiO2 layer with a SiNx layer results in a shift of the peak activation energy of Dit toward the valence band (Ev) of Si. Firing this SiO2/SiNx stack leads to an increase of Qfc, a reduction of Dit, and a moderate shift of peak activation energy of Dit toward Ev. Co-firing with the Al film on top significantly reduces the Qfc, Dit, and Dit peak activation energy, which is resulting from the atomic hydrogen passivation. These results are of particular interest for the development of solar cells with rear surface passivation and local contacts.

In this work we examine the effectiveness of hydrogen passivation at grain boundaries as a function of defect type and microstructure in multicrystalline silicon. We analyze a specially prepared solar cell with alternating mm-wide bare and SiN x -coated stripes using laser beam-induced current (LBIC), electron backscatter diffraction (EBSD), synchrotron-based X-ray fluorescence microscopy (μ-XRF), and defect etching to correlate pre- and post-hydrogenation recombination activity with grain boundary character, density of iron-silicide nanoprecipitates, and dislocations. This study reveals that the microstructure of boundaries that passivate well and those that do not differ mostly in the character of the dislocations along the grain boundary, while iron silicide precipitates along the grain boundaries (above detection limits) were found to play a less significant role.

We describe a high sensitivity measurement setup for the determination of recombination parameters in semiconductors at low levels of carrier injection. The setup is based on a lock-in amplifier and on a commercially available contactless conductivity detector. The information on recombination is extracted through the analysis, assuming quasi-steady-state conditions, of the low frequency, sinusoidally modulated photoconductivity signal induced by the illumination of a 950 nm light emitting diode array. Experimental results show a substantial increase in sensitivity with respect to traditional transient or quasi-steady-state techniques based on the same detection principle. The sensitivity bonus can be exploited for the extension of the carrier injection range for which effective recombination lifetime is measurable, both in the case of p-type and n-type wafers.

QSSPC (quasi-steady-state photo-conductance) is a well-known technique for lifetime measurement, but up to now its use has been limited to silicon. In this paper we describe a simple method to extend this technique to germanium samples. The significant differences between silicon and germanium are separately analyzed. For each of these differences the correction to be made is described. With this procedure the distinctive features that made this technique so successful and widespread are maintained.

Impurity decoration of grain boundaries (GB) introduces deep levels and enhances their recombination activity. In this study we use the standard DLTS technique to investigate the influence of the GB structure on the electrical properties of mc-Si. The detected DLTS spectra have an opposite signal compared to the usually known majority carrier peaks of deep levels, but standard DLTS analysis can yield trap energies and capture cross sections. By comparing the properties between GB and intra-grain regions with similar minority carrier lifetime, it is considered that the DLTS signals might be due to traps generated in the extended defects present both at GBs and grains. It is observed that ∑3 GB has deep level traps at the same energy level as ∑9 GB, while the trap concentration in the former one is lower than the latter. Furthermore, a phosphorous gettering treatment can reduce the recombination activities both in the grains and at the GBs.