Deconvolution of high-resolution depth profiling data with sputtering induced roughness for reconstruction of nano-layer structure

In this paper, a deconvolution method is proposed to obtain the original nano-layer structure directly from measured high-resolution depth profiling data with sputtering induced roughness. This method combines the Mixing-Roughness-Information (MRI) depth resolution function and the Total Variation (TV)-Tikhonov regularization. The regularization parameter of α that influences strongly the error and stability of the deconvolution result is optimized based on the L-curve criterion by the simulated annealing algorithm. The simulation results show that this method can be used to effectively obtain the original nano-layer structure even if the depth profiling data contains noise and/or sputtering induced roughness. Subsequently, as an example, the original multilayer structure of 4 × Si(15 nm)/Al (15 nm) is obtained directly from measured AES depth profiling data and the obtained roughness values accordingly at the depths of the middle of the first Al sublayer and of the last Al sublayer agree well with the ones measured by AFM. •Method is proposed for deconvoluting depth profiling data with sputtering induced roughness.•Nano-layer structure could be reconstructed by the proposed deconvolution method.•L-curve and simulated annealing method are used for optimizing the regularization parameters.•The deconvoluted result meets the requirements of accuracy, stability and timesaving.