Lessons from X-Ray Absorption Spectroscopy Studies of Dopants in CdTe Solar Cells—Where in the CdTe Lattice Do Cu, As and Se Reside?

Cadmium telluride (CdTe) is the leading polycrystalline thin-film photovoltaic material in the global market. Its record efficiency (18.6% for modules, and 22.6% for lab cells) is substantially governed by defect engineering of the absorber through processes such as chlorine (Cl) treatment, selenium (Se) alloying, and p-type doping (e.g., copper (Cu_Cd), arsenic (As_Te)). The incorporation of Cu, As, and Se in the absorber dictates the optoelectronic properties of the devices: from carrier density, bandgap, lifetime of minority carriers, to long-term stability. Although Density Functional Theory (DFT) has been used to predict and elucidate the atomic and electronic structures, it is unclear how, across the absorber, these atomic species evolve structurally from the fabrication to operation—specifically inside the CdTe lattice (if they enter the lattice at all). This understanding is crucial since polycrystalline CdTe devices suffer from low dopant activation, limiting the open circuit voltage to less than 1 V, and keeping the efficiency far from the theoretical limit (~32%).

Herein, to probe the atomistic evolution of Cu, As, and Se, we used synchrotron-based/nanoprobe fluorescence X-ray absorption near edge structure (XANES) due to its high spatial resolution and chemical sensitivity (to arrangements of neighboring atoms and electronic structures). To study the structutal distributions across the absorber, XANES spectra were measured from cross-sectional CdTe cells and devices. By fitting the measured spectra against reference spectra simulated from extensive libraries, which include point defects, grain boundaries, and secondary phases, we were able to extract atomistic information from the solar cells.

On p-type dopants: XANES analysis of various device architectures reveals structural inhomogeneities across the absorbers from point defects in bulk to secondary phases. In the case of conventional Cu doping, Cu atoms are very sensitive to their atomic surrounding and the majority of them, rather than entering the targeted lattice site (Cd) and contributing to the hole density, tend to react with impurities and form segregated phases—both near the back contact and inside the absorber. Cu rarely enters the Cd site, but when it does, it complexes with Cl (Cu_Cd-Cl_i), behaving as a detrimental defect. For the case of the state-of-the-art As doping, on the other hand, As atoms tend to enter the targeted lattice site (Te) more, explaining the higher hole density achieved in As-doped CdTe. Furthermore, As_Te may form neutral charged complexes with Cl (less detrimental than Cu_Cd-Cl_i), which may be responsible for the compensation of hole density and thus low dopant activation. Notably, compared to XANES spectra measured from single crystal CdTe—without Cl treatment, we confirm that the spectral features of As-Cl complexes are unique to polycrystalline CdTe. Surprisingly we also observe structural changes under applied bias and illumination, suggesting the rearrangement of atomic environment and changes in electronic structure under operation.

Finally, through XANES studies on Se alloying, we find evidence of Se present not only substituting on the anionic site of the CdTe lattice but also as interstitials and segregated at dislocation cores. Under environmental stressors (heat and light), Se migrates into the anionic site, with a decrease in Se-Cl co-passivation at the dislocation cores, co-occurring with the device degradation.

Altogether, the lessons from the XANES analyses have shed light on the complexity of the defect chemistry in CdTe, and filled some gaps in our fundamental understanding of the materials, which, in turn, can help pushing CdTe toward the theoretical limit.