Monitoring of Porous Silicon Layers Used for Epitaxial Wafer Production With Inline Reflectance Spectroscopy

Vahlman, Henri; Al-Hajjawi, Saed; Haunschild, Jonas; Wöhrle, Nico; Richter, M.; Jablonka, Lukas; Schremmer, Hans; Rein, Stefan

doi:10.1109/jphotov.2022.3176429

Cited by 2 publications

(15 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It cannot be excluded that bulk or interface scattering effects influence the results of our work. However, as stated in [7], the simulated reflectance curves show still a good overall agreement with the measured ones. Also, the SEM measured thicknesses correlate well with the results from the fitting approach (see Fig.…”

Section: Physical Modelsupporting

confidence: 72%

“…For our reflection model, we add a physical model to the classic model. The physical model is taken from [7] and calculates reflectance spectra R predicted with given layer parameters. We changed the implementation of the physical model to pytorch [13] to make parallel computing of multiple reflection curves and backpropagation possible.…”

Section: Reflection Modelmentioning

confidence: 99%

“…The reflection and the hybrid model consist not only of a machine learning part, but also of a physical model, which is taken from [7]. In this work, a numerical fitting approach is used to determine the layer thicknesses and porosities of two porous layers on top of a thick silicon wafer from reflectance spectra.…”

Section: Physical Modelmentioning

confidence: 99%

“…The interface between two layers, as can be seen in Figure 5, is not smooth and due to a measurement spot size of roughly 0.78 cm 2 this can influence the measured reflectance spectra. According to [7] the physical model is tuned by introducing a new parameter, the relative standard deviation (RSD) of the thickness to tackle this uncertainty. A variation of multiple thicknesses for the porous layers are used to calculate reflection curves and the results are then weighted by the probability function of a standard distribution to determine the actual output.…”

Section: Physical Modelmentioning

confidence: 99%

“…Our work is based on the physical model described in [7], that uses spectroscopic reflectance data and a stepwise optimization procedure to predict the same layer parameters. In the mentioned work, the physical model is modified such that it can be fitted optimally to actual inline measurement data.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

A self-consistent hybrid model connects empirical and optical models for fast, non-destructive inline characterization of thin, porous silicon layers

et al. 2023

Self Cite

View full text Add to dashboard Cite

Epitaxially-grown wafers on top of sintered porous silicon are a material-efficient wafer production process, that is now being launched into mass production. This production process makes the material-expensive sawing procedure obsolete since the wafer can be easily detached from its seed substrate. With high-throughput inline production processes, fast and reliable evaluation processes are crucial. The quality of the porous layers plays an important role regarding a successful detachment. Therefore, we present a fast and non-destructive investigation algorithm of thin, porous silicon layers. We predict the layer parameters directly from inline reflectance data by using a convolutional neural network (CNN), which is inspired by a comprehensive optical modelling approach from literature. There, a numerical fitting approach on reflection curves calculated with a physical model is performed. By adding the physical model to the CNN, we create a hybrid model, that not only predicts layer parameters, but also recalculates reflection curves. This allows a consistency check for a self-supervised network optimization. Evaluation on experimental data shows a high similarity with Scanning Electron Microscopy (SEM) measurements. Since parallel computation is possible with the CNN, 30.000 samples can be evaluated in roughly 100 ms.

show abstract

Section: Physical Modelsupporting

confidence: 72%

Section: Reflection Modelmentioning

confidence: 99%

Section: Physical Modelmentioning

confidence: 99%

Section: Physical Modelmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations