Hydrophilic silicon surfaces become hydrophobic without microroughening after 2 00°C low energy hydrogen plasma cleaning. The fully hydrogen-terminated silicon surfaces do not bond to each other, not even by the application of external pressure. A subsequent 400 to 600°C, 4 mm thermal treatment in ultrahigh vacuum converts the wafer surfaces to hydrophilic and bondable which can be attributed to desorption of hydrogen from the surfaces. Hydrophobic silicon surfaces prepared by a dip in HF (without subsequent water rinse) are terminated by H and a small amount of F, or by H and a small amount of OH (after subsequent water rinse). Hydrogen bonding of Si-F (HF) H-Si or Si-OW (HOH) OH-Si across the two mating surfaces appears to be responsible for room temperature spontaneous hydrophobic or hydrophilic wafer bonding, respectively.

No abstract

LSTM based language models are an important part of modern LVCSR systems as they significantly improve performance over traditional backoff language models. Incorporating them efficiently into decoding has been notoriously difficult. In this paper we present an approach based on a combination of onepass decoding and lattice rescoring. We perform decoding with the LSTM-LM in the first pass but recombine hypothesis that share the last two words, afterwards we rescore the resulting lattice. We run our systems on GPGPU equipped machines and are able to produce competitive results on the Hub5'00 and Librispeech evaluation corpora with a runtime better than real-time. In addition we shortly investigate the possibility to carry out the full sum over all state-sequences belonging to a given wordhypothesis during decoding without recombination.