Cadmium Telluride and Grain Boundaries: a Preliminary Study
- Author(s): Liao, Michael Evan
- Advisor(s): Goorsky, Mark S
- Duan, Xiangfeng
- et al.
The efficacy of the CdCl2 treatment on polycrystalline CdTe-based solar cells was discovered over a quarter of a century ago; and yet, the exact mechanism of this treatment is still not fully understood to this day. In fact, the lack of understanding stems from a debate on the exact role of grain boundaries in CdCl2-treated CdTe solar cells. Some hypothesize that the CdCl2-treatment causes grain boundaries to become beneficial to solar cell performance while others disagree and claim that the treatment simply mitigates the harmful effects of grain boundaries via passivation. A future goal of this project is to determine which, if either, hypothesis is correct by direct wafer bonding single crystalline CdTe. Direct wafer bonding of single crystalline materials would create only one grain boundary at the bonded interface. This approach allows the orientation and surface chemistry of interfaces to be controlled in order to study the chemistry of grain boundaries methodically. However, before any direct wafer bonding can be done, a preliminary study of single crystalline CdTe is necessary. High-quality direct wafer bonding can only be achieved if the surfaces of each wafer satisfy certain requirements. Additionally, analyzing single crystalline CdTe materials prior to bonding is crucial in order to make any insightful connections between results found from direct bonding of single crystalline CdTe and what is observed in polycrystalline CdTe.
First, the surface of an (001) CdTe layer epitaxially grown on an (001) InSb substrate is studied using atomic force microscopy. Stacking faults on the CdTe surface are observed and the thickness of the grown CdTe epilayer is calculated by considering the interplanar angles between the (001) and (111) crystallographic planes as well as the dimensions of the stacking faults. While the stacking faults will inhibit successful wafer bonding, the roughness of the regions outside the stacking faults is 0.9 nm, which is an acceptable roughness for direct wafer bonding.
High resolution x-ray diffraction is used to study the strain of the CdTe epilayer at the epilayer-substrate interface by generating reciprocal space maps of the (004), (115), and (335) crystallographic planes. It is found that CdTe grown on an (001) InSb substrate at a low growth temperature exhibits nearly 0% relaxation. As a result, the in-plane lattice parameter of the CdTe layer is maximally strained to match the smaller lattice parameter of the InSb substrate. Consequently, the CdTe lattice is tetragonally strained normal to the substrate surface, which causes the out-of-plane lattice parameter of CdTe to be larger than its intrinsic value.
Lastly, a CdCl2-treated CdTe-CdS (p-type CdTe on n-type CdS) solar cell structure is simulated using a semiconductor-heterojunction simulation program. In literature, it has been reported that chlorine atoms from the treatment segregate along grain boundaries in polycrystalline CdTe and cause the formation of local p-n junctions by inverting the grain boundaries to n-type. The simulated structure includes one grain and 2 grain boundaries. The grain/bulk CdTe material is p-type while the grain boundaries are made to be n-type with varying doping concentrations. Both the conduction band and valence band energy exhibit downward sloping from the CdTe surface to the CdTe-CdS interface. This structure assumes that the grain boundaries are parallel to the CdTe-CdS interface. While these simulations do not prove the existence of the local type-inversion hypothesis, they do entertain a novel possibility for future devices fabrication methods.