Simulation and process development for ion-implanted N-type silicon solar cells
MetadataShow full item record
As the efficiency potential for the industrial P-type Al-BSF silicon solar cell reaches its limit, new solar cell technologies are required to continue the pursuit of higher efficiency solar power at lower cost. It has been demonstrated in literature that among possible alternative solar cell structures, cells featuring a local BSF (LBSF) have demonstrated some of the highest efficiencies seen to date. Implementation of this technology in industry, however, has been limited due to the cost involved in implementing the photolithography procedures required. Recent advances in solar cell doping techniques, however, have identified ion implantation as a possible means of performing the patterned doping required without the need for photolithography. In addition, past studies have examined the potential for building solar cells on N-type silicon substrates, as opposed to P-type. Among other advantages, it is possible to create N-type solar cells which do not suffer from the efficiency degradation under light exposure that boron-doped P-type solar cells are subject to. Industry has not been able to capitalize on this potential for improved solar cell efficiency, in part because the fabrication of an N-type solar cell requires additional masking and doping steps compared to the P-type solar cell process. Again, however, recent advances in ion implantation for solar cells have demonstrated the possibility for bypassing these process limitations, fabricating high efficiency N-type cells without any masking steps. It is clear that there is potential for ion implantation to revolutionize solar cell manufacturing, but it is uncertain what absolute efficiency gains may be achieved by moving to such a process. In addition to development of a solar specific ion implant process, a number of new thermal processes must be developed as well. With so many parameters to optimize, it is highly beneficial to have an advanced simulation model which can describe the ion implant, thermal processes, and cell performance accurately. Toward this goal, the current study develops a process and device simulation model in the Sentaurus TCAD framework, and calibrates this model to experimentally measured cells. The study focuses on three main tasks in this regard: Task I - Implant and Anneal Model Development and Validation This study examines the literature in solar and microelectronics research to identify features of ion implant and anneal processes which are pertinent to solar cell processing. It is found that the Monte Carlo ion implant models used in IC fabrication optimization are applicable to solar cell manufacture, with adjustments made to accommodate for the fact that solar cell wafers are often pyramidally textured instead of polished. For modeling the thermal anneal processes required after ion implant, it is found that the boron and phosphorus cases need to be treated separately, with their own diffusion models. In particular, boron anneal simulation requires accurate treatment of boron-interstitial clusters (BICs), transient enhanced diffusion, and dose loss. Phosphorus anneal simulation requires treatment of vacancy and interstitial mediated diffusion, as well as dose loss and segregation. The required models are implemented in the Sentaurus AdvancedModels package, which is used in this study. The simulation is compared to both results presented in literature and physical measurements obtained on wafers implanted at the UCEP. It is found that good experimental agreement may be obtained for sheet resistance simulations of implanted wafers, as well as simulations of boron doping profile shape. The doping profiles of phosphorus as measured by the ECV method, however, contain inconsistencies with measured sheet resistance values which are not explained by the model. Task II - Device Simulation Development and Calibration This study also develops a 3D model for simulation of an N-type LBSF solar cell structure. The 3D structure is parametrized in terms of LBSF dot width and pitch, and an algorithm is used to generate an LBSF structure mesh with this parametrization. Doping profiles generated by simulations in Task I are integrated into the solar cell structure. Boundary conditions and free electrical parameters are calibrated using data from similar solar cells fabricated at the UCEP, as well as data from lifetime test wafers. This simulation uses electrical models recommended in literature for solar cell simulation. It is demonstrated that the 3D solar cell model developed for this study accurately reproduces the performance of an implanted N-type full BSF solar cell, and all parameters fall within ranges expected from theoretical calculations. The model is then used to explore the parameter space for implanted N-type local BSF solar cells, and to determine conditions for optimal solar cell performance. It is found that adding an LBSF to the otherwise unchanged baseline N-type cell structure can produce almost 1% absolute efficiency gain. An optimum LBSF dot pitch of 450um at a dot size of 100um was identified through simulation. The model also reveals that an LBSF structure can reduce the fill factor of the solar cell, but this effect can be offset by a gain in Voc. Further efficiency improvements may be realized by implementing a doping-dependent SRV model and by optimizing the implant dose and thermal anneal. Task III - Development of a Procedure for Ion Implanted N-type LBSF Cell Fabrication Finally, this study explores a method for fabrication of ion-implanted N-type LBSF solar cells which makes use of photolithographically defined nitride masks to perform local phosphorus implantation. The process utilizes implant, anneal, and metallization steps previously developed at the UCEP, as well as new implant masking steps developed in the course of this study. Although an LBSF solar cell has not been completely fabricated, the remaining steps of the process are successfully tested on implanted N-type full BSF solar cells, with efficiencies reaching 20.0%.