Lawrence Berkeley National Laboratory

In our paper, we reported a previously unobserved Raman-like band centered at around 550 cm-1 [e.g., Figs. 2(a) and 3(a)] and found that it diminished after the material degraded. Our latest study has revealed that the assignment of the 550 cm-1 band as a Raman band was erroneous, and this Raman-like peak was in fact the result of stray light of the unusually strong photoluminescence (PL) signal, peaked at around 770 nm, in the pristine sample. The strongest band-edge PL intensity was on the order of (0.5-1.0) × 105 cps [e.g., Figs. 2(b) and 3(d)] at the same or comparable power as the Raman measurements. Assuming a typical stray-light suppression ratio of 10-4 to 10-5 for a single grating spectrometer (an exact value is not available from the manufacturer of the instrument), the stray-light level would be comparable to that of the observed "Raman" peak typically on the order of 1 cps. In our new study, this broad band was eliminated after adding an appropriate short-pass filter to block the band-edge PL signal. The broad band at around 1340 cm-1, as shown in Fig. 5(a), has the same origin as the 550 cm-1 band. The observed anticorrelation between the signals of around 100 cm-1 and 550 cm-1 was the result of structure degradation that enhanced the Raman signal near 100 cm-1 and, in the meantime, reduced the PL signal near the perovskite bandgap at around 770 nm but appeared at 550 cm-1 as stray light. The wrong interpretation of 550 cm-1 does not affect the primary conclusion, i.e., the multiple-stage structure transformation and the interpretation of other results.We regret this error and intend to write a paper to report the latest findings in detail in the near future. The effort of Jose F. Castaneda, a new Ph.D. student in Zhang's group, has led to the finding of this error.