Understanding the Beneficial Role of Grain Boundaries in Polycrystalline Solar Cells from Single-Grain-Boundary Scanning Probe Microscopy


  • We thank C. Ferekides (USF, Tampa) for PX CdTe and CdTe/CdS cells, A. Fahrenbruch (Stanford Univ.) for CdTe crystals, I. Bar-Yosef (WIS) for use of his Dimension microscope, L. Kronik (WIS) for many enlightening discussions, P. De Wolf (DI, Veeco) for guidance with SCM, S. Richter (TAU) for guidance with CP-AFM, and V. Kaydanov (Col. School of Mines) for some early discussions. Financial assistance from the Weizmann Institute, via its Levin fund, the Feinberg Grad. School and the Philip M. Klutznick Research Fund, and, for the initial stages of this work, from USDOE is gratefully acknowledged. DC holds the Schaefer Chair in Energy Research. Earlier versions of this work, including the basic model that is given here in toto, were presented at the Spring 2003 MRS meeting (S. Francisco, Symp. B; cf. also MRS Bull., July 2003, 28(7), p. 521) and at the 38th IUVSTA and ISF workshop on Electronic Processes and Sensing on the Nano-Scale (Eilat, May 2003). Short communications dealing with various aspects of this work have appeared (cf. refs. 7, 8, 40). Supporting Information is available online from Wiley InterScience or from the author.


The superior performance of certain polycrystalline (PX) solar cells compared to that of corresponding single-crystal ones has been an enigma until recently. Conventional knowledge predicted that grain boundaries serve as traps and recombination centers for the photogenerated carriers, which should decrease cell performance. To understand if cell performance is limited by grain bulk, grain surface, and/or grain boundaries (GBs), we performed high-resolution mapping of electronic properties of single GBs and grain surfaces in PX p-CdTe/n-CdS solar cells. Combining results from scanning electron and scanning probe microscopies, viz., capacitance, Kelvin probe, and conductive probe atomic force microscopies, and comparing images taken under varying conditions, allowed elimination of topography-related artifacts and verification of the measured properties. Our experimental results led to several interesting conclusions: 1) current is depleted near GBs, while photocurrents are enhanced along the GB cores; 2) GB cores are inverted, which explains GB core conduction. Conclusions (1) and (2) imply that the regions around the GBs function as an extension of the carrier-collection volume, i.e., they participate actively in the photovoltaic conversion process, while conclusion (2) implies minimal recombination at the GB cores; 3) the surface potential is diminished near the GBs; and 4) the photovoltaic and metallurgical junction in the n-CdS/p-CdTe devices coincide. These conclusions, taken together with gettering of defects and impurities from the bulk into the GBs, explain the good photovoltaic performance of these PX cells (at the expense of some voltage loss, as is indeed observed). We show that these CdTe GB features are induced by the CdCl2 heat treatment used to optimize these cells in the production process.