Efficient and sustainable production, recovery and recycling phases of semiconductors (SC) life cycles require noninvasive, inline methods able to identify their composition in material streams. Ideally, the sensor system should be fast and incorporated into conveyor-belt operations. Rapid identification as well as spatial distribution maps would allow for real-time monitoring and quality control of the material stream. Considering these requirements, we suggest the sequential use of fast hyperspectral reflectance imaging (HSI) and Raman spectroscopic sensors for the identification of SC types in a sensor network configuration. We propose spectral proxies based on electronic properties derived from HSI-reflectance (i.e. absorption edge linked to the band gap values) and Raman sensors (i.e. Raman-active phonon modes) for SC identification. We identify potential limitations of each proxy on identifying undoped/doped SC materials, and discuss which process workflows enable optimized SC classification. We demonstrate the multi-sensor approach with SC standards (GaAs, GaSb, InP, 4H-SiC, and Borosilicate) which are relevant for both opto- and power-electronic devices, and showcase the potential of sequential data acquisition by fast HSI-reflectance sensors in the visible to shortwave-infrared (integration times: (4.5–18) ms) and Raman scattering (excitation laser: 532 nm, acquisition times: (0.5–10) s).
|