| Popis: |
Polycrystalline Cu(In,Ga)Se2 (CIGS) based thin-film solar cells achieve power-conversion efficiencies of almost 23% on the laboratory scale, one of the highest among thin-film solar cells. The aim of further CIGS research and development is to reach conversion efficiencies of 25%, which is currently the efficiency of the best single-crystalline Si based solar cells. To reach this goal, the factors limiting efficiency, e.g. non-radiative recombination of charge carriers, should be minimized. Such recombination processes may occur at line or planar defects present in the CIGS absorbers (among other interfaces, such as absorber and buffer layer). In the present study, the structure and composition of several defects as well as their evolution during the growth were investigated for an enhanced understanding. Highest efficiencies in CIGS solar cells are achieved, when the absorber is fabricated with a three-stage co-evaporation process. During the second stage of this process, Cu and Se are evaporated on the initially formed (In,Ga)2Se3 layer. The composition of the absorber becomes Cu-rich ([Cu]/([In] + [Ga]) > 1) during this stage. The change in composition leads to recrystallization, i.e. grain growth and defect annihilation, thus enabling higher conversion efficiencies. Therefore, it is crucial to investigate the recrystallization and the evolution of the microstructure at the second-stage of the CIGS growth. In the literature, two methods were suggested for this purpose: i) investigating the microstructural evolution of diffusion couples during a heating study; ii) ex-situ comparison of a growth-interrupted and a growth-finished sample. In the first part of this study, a Cu-poor ([Cu]/([In] < 1) CuInSe2 (CIS) precursor layer with a Cu2-xSe capping layer was prepared and heated in a scanning transmission electron microscope (STEM) to mimic the recrystallization. During the Cu diffusion from the Cu-rich Cu2-xSe phase into the Cu-poor CIS phase, the growth of defect-free grains towards the grains with closely-spaced planar defects (PDs) was monitored by low-angle annular dark-field (LAADF) imaging, whereas elemental depth profiles were analyzed by energy-dispersive X-ray spectroscopy (EDXS) before and after heating. The substantial impact of the Cu excess on the recrystallization was also indicated by an in-situ heating experiment of a Cu-poor CIS film without a Cu2-xSe layer on top, in which neither grain growth nor defect annihilation was detected. Monitoring of the recrystallization within the CIS absorber layers was performed for the first time by means of STEM and provided direct evidence for the currently accepted theory of the grain growth mechanism. In the second part, a CIGS absorber grown via co-evaporation was analyzed. During the growth, one piece of the sample was removed before the recrystallization at the second stage. For the remaining piece, the three-stage process was completed. The defect concentrations as well as the in-depth elemental analysis were performed by STEM-LAADF imaging and EDXS, respectively. Similar to the in-situ heating results, much larger grains with reduced linear/planar defect concentrations were detected in the absorber layer for which the growth had been completed. Although most of the structural defects were annihilated after the recrystallization, few structural defects were detected by LAADF imaging after the recrystallization, and even after the completion of the three-stage growth process. Further analyses were performed via aberration-corrected, high-resolution STEM (HR-STEM) in combination with electron energy-loss spectroscopy (EELS) to elucidate the nature of individual microstructural defects from various stages of the growth. HR-STEM and EELS results revealed the structure and chemistry of defects that were present in both growth-interrupted and growth-finished samples: Σ3-twin boundaries and stacking faults with stoichiometric elemental distribution; grain boundaries, tilt boundaries and dislocations with cation redistribution, i.e. Cu enrichment and In depletion. Stoichiometric inversion boundaries, Cu enriched ‘complex’ PDs and an extrinsic Frank partial dislocation were detected only in the growth-interrupted Cu-poor samples, whereas a ‘Cu2-xSe secondary phase’ was detected only in the growth-finished absorber layer. The present work provided direct insight into the recrystallization of CIGS absorbers and evolution of structural defects, as well as a thorough investigation of individual defects in CIGS absorbers. |
