Si-based photovoltaics
Introduction
In 2007, the price of oil rose steadily, as did the concerns about the security of energy supply. In the same year, some major international meetings on climate change took place such as intergovernmental panel on climate change conference and the Bali conference. They highlighted not only the increased consensus on the threat of global warming, but also the urgency of taking action to counter it. Governments all over the world have pledged to reduce their CO2 emissions, notably Europe with ambitious targets for renewable energies by 2020.
In this context, photovoltaics is seen as a key technology, as it has the potential to cover a major part of the energy needs in the long term. To achieve conditions where this potential can be tapped in a significant way, the cost of photovoltaic power has to decrease substantially every year while the photovoltaic industry has to grow at a rate of more than 30% for the next two decades.
Fortunately, this seems an achievable ambition. In 2007, just as for the previous 7 years, the photovoltaic market increased at a rate of more than 60%, limited more by the supply of Si feedstock than by the demand for photovoltaic systems.
The dominating photovoltaic technology remains the wafer-based or `bulk' crystalline Si technology, accounting for more than 90% of the photovoltaic modules produced worldwide. At IMEC, research is carried out not only on bulk-Si solar cells, but also on thin-film crystalline Si technologies, which might in the medium to long term replace the wafer-based technology because of their higher potential for cost reduction.
Next generation of Si solar cells on low-cost substrates
In order to reduce the cost of a solar cell, it is imperative to reduce the material cost associated with the Si wafer. The most obvious approach to this is to slice thinner wafers so that more wafers can be obtained from an ingot, therefore reducing the price per wafer. Already in the last few years, the thickness of wafers used in the industry has reduced steadily, the standard thickness nowadays being around 200 micrometers. It is anticipated that this trend will continue, and that the next generation of PV module will be based on two-sided cells made with very thin wafers (~100-150µm thick). This evolution is only possible if the efficiency of solar cells can be maintained at high values. To achieve this, the rear structure of the solar cell has to be modified. Instead of the conventional Al-BSF (region at the rear, which is highly doped with Al as a result of an alloying and regrowth process during a short thermal anneal), the rear surface has to be passivated with a dielectric layer, and the cell has to be locally contacted. The passivation layer not only has to provide good surface passivation, but should also have the right properties in terms of thermal stability and internal reflection.
A few years ago, IMEC introduced a promising solar cell concept featuring such dielectric surface passivation, called the `i-PERC' cell. The passivation of the rear surface is obtained with a stack of low quality oxide and PECVD nitride. For the local contacts, openings are made in the passivating layer by laser ablation and a local alloying process with Al in the openings creates the local contact with a small back-surface field region underneath, limiting recombination at the rear contacts. The front surface is textured using a plasma texturing process, particularly suited for very thin wafers.
In 2006, we had achieved a solar cell efficiency of 17.4% on a 180microns thick multicrystalline Si substrate. The Voc was as high as 630mV, and a gain of about 1% in absolute efficiency compared to the classical Al BSF cell concept had been observed C15031. In 2007, the focus was to apply the i-PERC process on very thin substrates, below 150micrometers in thickness. State-of-the-art solar cell results were obtained on 156cm2 area, 130µm thick multicrystalline Si substrates. Cells with Voc over 620mV and efficiency of 16.7% were achieved C16127. Importantly for such thin solar cells, the aluminum-induced cell bowing is inherently avoided because of the local alloy process. The surface passivation by the low-quality oxide/Si nitride stack was investigated in detail. It is clear that hydrogen plays a crucial role in the surface passivation process. During the rapid thermal treatment, hydrogen diffuses from the SiNx:H layer in the stack towards the Si-SiOx interface and passivates defects such as dangling bonds C16131. This process occurs simultaneously with the hydrogenation of the bulk from the SiNx:H anti-reflection coating, although the principles are different. Bulk passivation is optimal when a high density SiNx:H is used, out of which hydrogen diffuses in atomic form, while Si/SiOx interface passivation is most effective when molecular hydrogen is supplied, which is favoured with low density SiNx:H C16208.
Figure 1: Solar-cell results for industrial-type solar cells on very thin (130micron thick) multricrystalline Si substrates. Comparison between the i-PERC process and the conventional full Al BSF process.
Although mainstream solar cell technology makes use of B-doped p-type Si wafers, there is an increasing interest in the PV community for solar cells made with n-type material. With the trend of decreasing wafer thickness, high-efficiency (>15%) back-junction solar cells become feasible even with material with minority-carrier lifetime in the order of 100µs. N-type multicrystalline Si can fulfill this requirement at a limited cost, and has therefore a good potential for an industrial implementation. The development of the p+ emitter on the rear is essential for this type of cell, and Al alloying is one of the techniques of choice to realize it. By optimizing the emitter formation by Al-alloying and using thin wafers, we achieved an efficiency of 15.0% on n-type mc-Si. The cells displayed good values for Voc (~615mV) and FF (>80%) C16112, C16117.
In order to further reduce solar cell costs, low-cost `solar grade' wafer materials are being developed and investigated, which feature more contaminants and defects than conventional multicrystalline Si. It is crucial to understand the impact of such recombination centers and to study how process steps such as gettering and hydrogenation can deactivate them. For this, sensitive characterization techniques are needed. In 2007, IMEC developed a simple modulated photo-conductivity technique for the measurement of recombination parameters in semiconductors. Information is extracted through the analysis, in a lock-in setup, of the modulated photo-conductivity signal. Intensity modulation of the excitation provides injection-level dependent lifetime whilst frequency modulation discerns recombination and trapping rates. The new technique has led to a substantial increase of sensitivity with respect to traditional transient or steady-state photoconductance lifetime measurement techniques. This sensitivity “bonus” can be exploited for the measurement of very thin and/or very low lifetime samples, the latter being crucial for an effective advanced lifetime spectroscopy.
Back-contacted solar cells
Removing or reducing the front surface metal grid results in a higher effective semiconductor area and thus has the potential for increased cell efficiencies. Moreover, bringing the external contacts to a single surface allows for significant cost reductions in module assembly. Another benefit is that the architectural requirement for uniform appearance is fulfilled as the visual disturbance caused by the presence of the busbars and the tabbing is displaced to the rear of the module. In the past, IMEC has developed the metallization wrap through (MWT) solar cell, a technology that is now licensed to its spin-off Photovoltech. In 2007, in collaboration with Photovoltech, IMEC introduced a new process for rear-contacted thin solar cells. The process is a combination of the MWT concept, and the dielectric stack passivation scheme that is used in IMEC's i-PERC solar cell process. These two proprietary methods have a real synergetic action towards the achievement of high-efficiency industrial solar cells: the dielectric stack passivation scheme enables an improved Voc due to better passivation, an improved Jsc thanks to an increased rear-surface reflectivity, the suppression of the bowing problem observed with full Al BSF on thin cells, and an improved fill factor due to the better passivation of the critical regions in MWT cells. An average of 16.0% efficiency was achieved with a process that is readily transferable to the industry C16116.
Novel lift-off techniques
For the Si technology to remain competitive in the future, the consumption of Si will have to decrease dramatically, well below the present 9-10 g/Wp. In the past, several layer transfer (or lift-off) techniques were proposed, where a thin layer of high quality Si is peeled off from a Si wafer or ingot, and transferred onto a low-cost substrates. Most of them however relied on an epitaxial process. The cost reduction could become more substantial if the Si epitaxy could be avoided. In 2007, we proposed two novel epi-free lift-off processes.
The first one takes advantage of the property of macroporous Si to form a monocrystalline thin film over a large void by reorganisation at high temperature. The thin film created in this way can be bonded to a glass substrate for processing into a solar cell. In 2007, two essential steps were successfully developed, paving the way to a complete proof of concept. The suitability of anodic bonding to attached the layers formed through macroporous Si reorganization was demonstrated. Moreover, amorphous Si/Crystalline Si heterojunction solar cells were made on anodically bonded layers. C14755. While the experiments on layer transfer have been carried out with lithography-define pores, the use of macroporous Si by electrochemical etching was studied as an alternative. The macropores should be regular and columnar. The growth of such pores under front-side illumination on medium doped n-type Si was investigated. The various morphologies observed, with branched or columnar pores, differed from those previously reported and indicated that the mode of pore growth used was at the border between meso- and macroporous mechanism P14398.
We also presented a completely new wafering method for the production of ~50-micron-thick crystalline Si wafers. It is a lift-off process that requires only the use of a screen-printer and a belt furnace; no ion-implanted or porous layer is needed. On a thick substrate a layer is deposited with mismatched thermal expansion coefficient with respect to the substrate (for instance a metal layer). Upon cooling, the differential contraction induces a large stress field which is released by the initiation and the propagation of a crack parallel to the surface. The principle was mechanically modeled, and was demonstrated on both single and multi-crystalline Si. Films with an area of 25 cm2 and a thickness of 30-50microns were obtained. Some Si layers were further processed into solar cells. the first device had an energy conversion efficiency of 10.0% (1cm2) P16118.
Figure 2: Photograph of a wafer after the top layer has been peeled off the parent substrate (25 cm2). The parent substrate (a/) can be re-used for additional layer lift-off, and the top Si layer (/b) is still attached to the stress-inducing layer.
Figure 3: Principal stress map of the structure during crack propagation. The crack propagates parallel to the surface.
Epitaxial thin-film solar cells
A promising thin-film solar cell technology that IMEC has a large experience with is epitaxial solar cells. The idea is to reduce material cost by using low-cost highly doped crystalline Si wafers (e.g. from metallurgical grade Si or from scrap material) P14514. On these wafers, an epitaxial layer is deposited by CVD, typically 20µm thick. As the solar cell process is similar to a classical bulk-Si process, epitaxial cells are expected to be easier to implement in existing production lines than any other thin-film technology.
In 2007, we improved the stacks of porous Si layers that we introduced in the previous years to maximize internal reflection at the interface between a Si substrate and an epitaxially grown layer. The stack consists of alternating porous layers of high and low porosity defined by the quarter wavelength rule. Due to the epitaxial growth of the epitaxial layer, the porous Si reorganizes but retains its original layout. Experimental data of the total hemispherical reflectance measured from the epitaxial side was linked to the internal reflectance. An optical path length enhancement of 7 was extracted from this result. Application of these reflectors on thin film epitaxial solar cells showed an increase in short-circuit current of up to 2mA/cm2. A detailed analysis showed that the Bragg and total internal reflection (TIR) effects are additive on the cell level with the first one needed for photons striking the epi/PSi interface at a perpendicular angle while the TIR contributes to the reflection of obliquely striking photons. An internal reflectance of 80-84% was determined, in which 25% of the internal reflectance could be attributed to the Bragg effect itself C14754.
Figure 4: Reflectance of samples with a 15-layer porous Si reflector and a porous Si monolayer (after epitaxial deposition and texturing).
Another aspect of epitaxial cells that was intensively investigated in 2007 is the topic of epitaxial emitters. Even when implemented in large-throughput reactors, the CVD process is expected to remain one of the most expensive steps in an industrialized process flow for epitaxial cells. As a result, it is crucial to apply the CVD process in the most effective approach. Growing also the emitter, besides the active base, has many economical and practical advantages. The growth of the emitter profile was studied in detailed. A two-layered emitter was proposed, consisting of a moderately doped region on which a very thin but very highly doped region is grown. Such a profile minimizes the losses in the emitter while providing a strong front surface field effect and high open-circuit voltages. Moreover, dedicated solar-cell processes were established integrating both a CVD grown emitter, and concepts for optical light trapping in epitaxial cells, including texturing of the front surface and the application of the intermediate reflector at the epi-substrate interface. In this way, enhanced solar cells with efficiencies close to 15% are obtained C15060.
Thin-film polySi solar cells
Another thin-film Si solar-cell approach with high cost reduction potential is the thin-film polySi technology. Here a thin layer (only a few µm) of crystalline Si is deposited on a low-cost foreign substrate. In order to achieve a layer with a relatively low defect density, a seed layer consisting of grains with small aspect ratio is prepared using the technique of Aluminum-induced crystallization (AIC), followed by epitaxial thickening. P14865. Because we use thermal CVD to grow the active layer, a substrate has to be chosen that withstand high temperatures, such as ceramic and glass-ceramic substrates. The best efficiencies so far are 8% on ceramic substrates and 6.4% on glass-ceramic substrates, which are the best results so far worldwide for thin-film polySi cells based on a seed layer approach P14865, C14770, C16122, C16120, P16110.
A crucial part of the device in thin-film polySi solar cells is the emitter. We showed that the use of an amorphous Si/polycrystalline Si heterojunction emitter instead of a diffused homojunction emitter led to a boost in the open-circuit voltage by 90mV. This improvement turns out to be related to the absence of dopant smearing along grain boundaries. By using scanning spreading resistance microscopy, we found an enlargement of the junction area by a factor of five in case of a homojunction P15573
Figure 5: Cross-section of a diffused homojunction emitter on fine-grained polySi as measured by SSRM (Tdiffusion= 860°C; 2µm p+3µm p; p-type doping density 1017m-3 doped).
To determine the factors limiting the performance of our thin-film polySi solar cells, we carried out detailed characterization of the layers. Defect etching revealed a very large density (109cm-2) of intragrain defects in the polycrystalline Si layers obtained through AIC and epitaxy. Electron-beam-induced current measurements showed a strong recombination activity at these defects. Cathodoluminescence measurements showed the presence of two deep-level radiative transitions (0.85 and 0.93eV) with a relative intensity varying from grain to grain. These results indicate that the electronic quality in these layers is dominated by the presence of numerous electrically active intragrain defects. It also explains why the open-circuit voltage of these pc-Si solar cells is quasi-independent of the grain size P14254, C16113, C16121.
Figure 6: SEM picture of the surface of a thin-film polySi layer obtained by AIC and epitaxy, after polishing and defect etching. A very high density of dislocations can be observed within the grains.
Figure 7: Electron-beam induced current mapping of a thin-film polySi solar cell, indicating that the intragrain defects are electrically active.
Conclusion
In 2007, IMEC achieved important progress on both bulk-Si solar cells and thin-film Si solar cells. The quality and relevance of our work was recognized by experts in the field, as witness by the acceptance of papers in highly ranked journals and two awards at major international conferences.