Authors' response:
We recently reported1 on solar cells fabricated using low-bandgap platinum polyyne polymer (P1). The solar-cell characterization had been repeated many times and independently verified, hence we believe that the photovoltaic characterization data are correct. However, spectroscopic ellipsometry measurements had been performed only once, and surface roughness had not been taken into account in the fitting. Therefore, we repeated the fitting of the spectroscopic ellipsometry data with surface roughness correction using both our own software (modified Lorentz oscillator model)2 and W-VASE software (J. A. Woollam, Lorentz model and point-by-point fitting). Surface roughness was taken into account using the Brugemann effective medium approximation (EMA)3. However, fitting attempts resulted in multiple solutions with similar fit quality, and the parameter uncertainties estimated by W-VASE software were very large (especially the thickness of active layer and rough layer when using EMA). One possible reason for this is that EMA theory is applicable only if the dimensions of inclusions in the composite material are smaller than λ/10, and the dielectric function of phases is independent of the size and shape3. Although the height of the islands satisfies this condition, this is clearly not the case for lateral size, therefore different treatment of the scattering needs to be implemented4. Based on atomic force microscopy (AFM) images of blend films on Si, geometric limits of the validity of the Rayleigh–Rice theory4 are satisfied for wavelengths above ∼400 nm, but one possible problem is that the variation in composition of surface features cannot be excluded for blend films. Thus, the optical functions of the P1:PCBM blend films cannot be reliably determined by spectroscopic ellipsometry due to the intrinsic experimental and fitting limitations (inapplicability of EMA). For that reason, the experimentally determined optical functions data for P1:PCBM and P3HT:PCBM blends can only be used to compare their optical properties because they have been derived under the same assumptions, but cannot be used for accurate prediction of device performance.
This is a preview of subscription content, access via your institution