In-situ etch depth control with precision around 1 nm via reflectance anisotropy spectroscopy during reactive ion etching of monocrystalline III/V semiconductors

点击翻译按钮获取中文摘要

The aim of this project is the improvement of the precision of in-situ etch depth control during reactive ion plasma dry-etching. Three approaches shall be employed.The first concept (for the case of common RIE etch rates of a couple of 100 nm/min) uses Fabry-Perot oscillations, i. e. periodic changes of the reflectivity of a layer for etch-related shrinking layer thickness. The oscillations are to be observed simultaneously at two different photon energies, which differ by 1/10. This relates to the use of a vernier scale of a sliding caliper. Precision of 1-2 nm in etch depth determination should be possible this way. One problem might occur, if the RAS data could not be collected and used fast enough during the etch process. Then the etch rate would have to be reduced, e. g. by reduction of the ion energy. The second approach is going to use periodic changes of the RAS signal with changes monolayer by monolayer. This effect is known from epitaxy. But the proposer-s team has already seen such oscillations in the average reflectivity for not too large etch rates during GaAs dry-etching. They are to be called monolayer oscillations here, in order to distinguish them from the Fabry-Perot oscillations of the first approach. If the second concept was going to work, the best possible precision in etch depth determination would be reached, i. e. a single monolayer. But optimal results for the second approach are only to be expected for etch rates lower than usual (like 50-100 nm/min). A third concept is offered by praxis. Discrepancies in the RAS signal or variations due to differences in doping of different layers can be used. Dopings are common in optoelectronic layer sequences. And oftentimes etching is to be stopped exactly at an interface between two layers. First results of the team are going to be extended during the project. Questions are: Under which circumstances (layer and etch parameters) can p- and n-dopings be distinguished? Can even the doping level be extracted from the signal? If so, with which precision? Can these results be used on-line (during etching)? Are data collection and calculation times small enough?