THREE DIMENSIONAL NUMERICAL MODELLING OF SERIES RESISTANCE LOSSES IN THE BASE OF SINGLE AND DOUBLE JUNCTION REAR POINT CONTACT SILICON SOLAR CELLS K. Kotsovos 1 , K. Misiakos 1 and D. Tsamakis 2 1 Institute of Microelectronics, NCSR Demokritos, P. O Box 60228 153 10 Aghia Paraskevi, Attiki, Greece Tel: (+30210) 6503226, Fax: (+30210) 6511723, E-mail: kotsovos@imel.demokritos.gr 2 National Technical University of Athens, School of Electrical and Computer Engineering, 157 73 Athens, Greece ABSTRACT: Rear point contact (locally diffused) solar cells with single and double collecting junctions are investigated through numerical three-dimensional simulation. The simulation method is based on the numerical solution through Fast Fourier Transform (FFT) of the minority and majority carrier transport equations and Poisson’s equation in the base of the cell. The IV curve and series resistance of the cells is evaluated through the 3D model. Experimental results are presented for the single junction solar cell, where the series resistance of the cell is measured and compared with the theoretical values. The influence of the additional collecting junction in the back surface to the cell’ s series resistance is investigated through simulation data. Keywords: Simulation; Silicon; Back Contact; Series Resistance 1 INTRODUCTION The rear point contact locally diffused (PERL) solar cell is a very advanced and efficient silicon solar cell structure. A record efficiency of 24.7% is achieved for such device [1]. The major advantage of this structure is the passivation of the back surface with high quality thermally grown oxide, while the contact with the metal is performed through very small openings in the oxide in order to minimize recombination. However, current crowding around the back contacts leads to significant ohmic losses, affecting the fill factor. For that purpose, an algorithm based on the numerical solution of the basic partial differential equations in three dimensions has been developed to investigate carrier transport effects in the base of such solar cells [2]. The simulation algorithm is based on the solution of the basic partial differential equations through Fast Fourier Transform (FFT). In this way, these equations become algebraic and can be easily solved. The basic assumptions of this method are planar geometry and low injection. Novel devices with an additional collecting junction on the back surface have also been simulated [3] using this technique. Double junction structures have also been investigated earlier [4] by means of one- and two-dimensional simulations. In the following section the outline and assumptions of the 3D model are described and in the next ones results from 3D simulations in the base of single and double junction devices are presented. Series resistance simulation data are compared with experimental measurements from fabricated single junction structures. 2 MODEL ASSUMPTIONS The base of the solar cells under consideration is considered homogeneous with thickness w and the junctions infinitesimally shallow, while photogeneration in the emitter is neglected. We set as x, y the dimensions parallel to the junctions while z is the perpendicular one as shown in fig. 1. The back diffused contacts are assumed squares with side length d, while the period length of the repeated pattern is l. The two lower parts of fig. 1 show the structure of the simulated single and double junction devices. The back emitter of the double junction solar cell covers the whole area around the diffused contact, while this region is covered with a high quality thermal oxide in the case of the single junction structure. x y z Diffusion contact d Period length l Emitter Contact Contact Front emitter Back emitter Base Base w (a) (b) (c) Figure 1: Geometry of the elementary repeated pattern (a), single junction (b) and double junction sructure (c). The mathematical model of the single junction structure with its boundary conditions and the simulation algorithm have been described with detail in [2], while the corresponding equations of the double junction device have been discussed in [3]. The simulation program calculates the device IV curve and series resistance is evaluated. 3 RESULTS The simulated devices are assumed to have base thickness w=400 µm and doping density N A =10 16 cm -3 , while the mobilities of the minority and majority carriers are taken from Klaasen [5]. The illumination spectrum is the global AM1.5 normalized to 100 mW/cm 2 and light trapping similar to the pyramidal texture scheme is