Simulation of a-si pin solar cells with buffer
- PDF / 465,840 Bytes
- 6 Pages / 414.72 x 648 pts Page_size
- 85 Downloads / 244 Views
SIMULATION OF A-SI PIN SOLAR CELLS WITH BUFFER
PFLEIDERER H.
Siemens Corporate Research, Otto-Hahn-Ring 6, 8000 Mlnchen 83, Germany
ABSTRACT
The buffer of standard solar cells restricts the "surface recombination" at the front side. Some typical cell properties linked to the buffer are exposed by numerical simulations: The "blue snake" appearing in small-signal photocharacteristics, the electron inversion layer, the reverse field peak and the satellite space-charge dipole layer. It is possible to base a simulation on the roots of an algebraic equation. INTRODUCTION
The p-layer of amorphous-silicon (a-Si) solar cells is doped with carbon (C) in order to widen the band gap and to better exploit the blue part of the sun spectrum [1]. The p(C)-i heterojunction involves an electron-affinity variation that can contribute to the achievable open-circuit voltage [2] and hinders electrons from the i-layer to recombine with holes from the p-layer [3]. A defective junction will again promote this kind of "surface recombination", however. Now the p-layer is certainly defective, and its boundary, the p-i junction, likewise [4]. Hence only a spatial separation of p-i junction and electron-affinity junction will secure a low surface-recombination loss of photocarriers [5]. Interposed between the p-i junction and the affinity barrier then is a C-rich but otherwise undoped layer called "buffer". It may consist of a uniform section with constant C-content and a graded section with gradients of C, band gap and electron affinity.The introduction of suitable buffers has led to an elevation of the open-circuit voltage. But under prolonged illumination preferentially the buffer degrades [6]. The present contribution is restricted to phenomenological considerations. The affinity gradient of a heterojunction disturbs the local space charges and recombination rates. A buffer exposes the typical traits most distinctly. An indication of them will be given. PHYSICAL ASSUMPTIONS
The numerical simulations to be presented rest on a simple physical model. It uses a densityof-states distribution through the mobility gap consisting of 4 branches Di(E), i = 1 to 4, all of them exponential functions of the gap energy E [7]. The branches D1 and D2 (D3 and D4 ) represent donor-like tail and bottom states (acceptor-like bottom and tail states). The valence (conduction) band edge is E = Ev = E = 0 (E = EC = E4). The branches intersect at E1 = 0.3, E2 = 0.55 and E3 = 0.8 in units of the band gap E9 = E4 . The pin structure consists of 7 regions as sketched in Fig. 1.The regions 1 and 7 represent the p- and n-layers, regions 2 and 3 the buffer, and regions 4, 5 and 6 the i-layer. The band gap is E = 2 0 eV (1.7 eV) in regions 1,2 (4 to 7), and varies linearly through region 3. The density of'states Di assumes the values Di- in the different regions j. Table I gives the width of the regions together with the Diode P..__ i Ivalues D2i(E2) = D3 i(E2) at the intersection between the bottom branches. The densities EC
Regions 1 2 3 4 Sections
5
16
Fig.1
Data Loading...
