Effective and Noneffective Recombination Center Defects in Cu2ZnSnS4: Significant Difference in Carrier Capture Cross Sections

By combining the electron–phonon coupling effect and the static coupling formalism, we calculate, through the first-principles methods, the carrier capture cross sections of the three possible nonradiative recombination center (NRRC) defects in Cu2ZnSnS4. These values are currently unavailable but critical for understanding the limiting factors of the minority carrier lifetime and simulating the photovoltaic devices. We show that the cross sections for SnZn2+ capturing one electron (a (+2/+1) transition) and for [CuZn–SnZn]+ capturing one electron (a (+1/0) transition) are both very large, whereas for SnZn+ capturing one electron (a (+1/0) transition) is much smaller by several orders of magnitude. The minority carrier lifetime will be limited to below 1 ns if the concentrations of SnZn2+ and [CuZn–SnZn]+ are higher than 1015 cm–3, so they are effective NRRCs, whereas the lifetime can be as long as 10 μs with the same concentration of SnZn+, so SnZn+ is a noneffective NRRC. The phonon mode analysis shows that the cross section is strongly correlated with the vibration mode of Sn–S bonds around the defects and its coupling with the localized wavefunction on the defect state. SnZn2+ and [CuZn–SnZn]+ have a short and strong Sn–S bond with a high-frequency vibration mode, and these two defects undergo a large structural distortion after capturing an electron, which decreases the barrier for carrier capture and thus produces a large cross section. In contrast, SnZn+ has a softer Sn–S vibration mode and thus much higher barrier for electron capture. Our calculations not only identify two effective NRRCs, which provide the mechanism behind why the Cu-poor, Zn-rich, Sn-poor growth condition, were widely adopted for fabricating high-efficiency Cu2ZnSnS4 solar cells but also show that a very large difference can exist in the carrier capture cross sections for the same defect in different charge states (SnZn2+ vs SnZn+). We propose that the deep-level defects may have large carrier capture cross sections if they are surrounded by strong bonds and undergo considerable structural relaxations after capturing a carrier, which can be used as an empirical criterion for the quick identification of effective NRRCs.