In the paper we address a challenge how to design textures (periodic or random) that could outperform the state-of-the-art random ones. We present two approaches for systematic optimization of surface-textures: bottom-up and top-down approach. Fully 3-D optical simulations, including calibrated non-conformal layer growth model were employed to predict gains in short-circuit current densities in a micromorph solar cell (bottom-up approach) and a single-junction amorphous silicon solar cell (top-down approach). In the bottom-up approach we start with a simple sinusoidal component and change its shape systematically in the direction of broader valleys, resulting also in better conditions for the layer growth. In the top-down approach we start from a random texture (morphology fingerprint taken from Asahi U type substrate) and modify it in a spatial frequency domain. We show the role of presence/absence of different frequency regions as well as the important role of the phase spectrum on optical characteristics of the device. In both approaches improved textures have potential to outperform state-of-the-art random ones, not only from optical point of view but also in terms of conversion efficiency.
COBISS.SI-ID: 10196564