(285c) First Principles Calculation of the Raman Spectra of Cu2ZnSnS4, a Promising New Photovoltaic Material

Thin film solar cells based on copper zinc tin sulfide (Cu₂ZnSnS₄ or CZTS) are emerging as an alternative to cadmium telluride (CdTe) and copper indium gallium diselenide (CIGS) solar cells. Although CIGS and CdTe solar cells have already achieved impressive power conversion efficiencies (15-20%) the elements commonly used for their production are either toxic (e.g., cadmium) or rare in the earth’s crust (e.g., indium, tellurium). CZTS is made of nontoxic and abundant elements, has a band gap of ~1.4 eV and absorbs light strongly in the visible region of the electromagnetic spectrum. CZTS crystallizes in a tetragonal crystal structure where c ≈ 2a. The structure can be thought of two face centered cubic sulfur anion lattices stacked on top of each other with tetrahedral voids filled with metal cations. Half the tetrahedral voids within the sulfur anion FCC lattice are occupied by copper, zinc and tin cations in the ratio 2:1:1. Depending on the local arrangement of copper, zinc and tin within the anion lattice, CZTS can exist in three different phases: kesterite, stannite and the pre-mixed Cu-Au structure (PMCA). The unit cell for CZTS can also be viewed as two zinc blende lattices stacked on top of one another with copper and tin cations replacing some of the zinc cations until the CZTS metal cation stoichiometry is obtained. The structures of the three phases of CZTS and other metal sulfides such as ZnS and Cu₂SnS₃ (CTS) are very similar and they are practically impossible to differentiate based on X-ray diffraction (XRD). This problem becomes acute especially in the case of nanometer size crystals where the XRD peaks broaden with decreasing crystal size. Substitution of S with Se is used to make Cu₂ZnSnS_4-xSe_x (CZTSSe) and vary the band gap. Raman spectroscopy may help differentiate between CZTS, CZTSe, CTS and ZnS. However, it is still not known if Raman spectroscopy can be used to differentiate between the three phases of CZTS primarily because the three phases have not been isolated in the laboratory yet.

We have used density functional theory calculations within the generalized gradient approximation (GGA) as implemented in QUANTUM ESPRESSO to calculate the phonon frequencies and vibrational densities of states in CZTS and related compounds and compared the predictions to experimental measurements. Specifically, we compared our calculations to experimental Raman scattering due to characteristic phonon vibrations at the gamma point. We obtain excellent agreement between calculated and measured Raman scattering frequencies for CZTS.  Having validated the calculation approach and the pseudo potentials, we extended the predictions to compositions and CZTS related compounds where data is either not available or scarce. We find that the three different phases of CZTS give Raman scattering peaks which are shifted from each other by a few wavenumbers. We have also studied the Raman spectra for CZTSe and CZTSSe and found significant variations in the spectra with changes in the S:Se ratio.  Raman scattering peaks also shift with changes in the arrangements of S:Se within the lattice.  These results will be of practical use and allow one to gain deeper insights from the Raman spectra during synthesis of CZTS, CZTSe and related materials.