Controlled Etching Of Single-nanometer Precision Silicon Nanopore Arrays

Silicon nanopores have the advantages of high chemical stability,good reusability,high mechanical load-bearing capacity and good compatibility with existing CMOS fabrication technologies.This makes it show great potential and development in the fields of DNA sequencing,protein profiling and solar power generation.However,fabrication of sub-10 nm silicon pore arrays with high efficiency,high quality and low cost is still very challenging,thus new fabrication methods need to be developed urgently.Therefore,this paper proposes a new idea of using a combination of metal-assisted chemical etching and machine learning to achieve controllable processing of sub-10 nm silicon nanopore arrays.A special platform for etching processing was designed and built,which established the hardware foundation for subsequent experiments;the self-assembly of silica-covered gold nanoparticle arrays on the silicon wafer surface by spin-coating method,and the influence of the homogenizer parameters(rotational speed,acceleration,working procedure and working time)on the self-assembly of nanoparticle arrays were studied,and an ordered single-layer dense array of silica-covered gold nanoparticle arrays was prepared,which provided a regular gold nanoparticle template for subsequent etching processing.By studying the morphological evolution of silicon nanopores during the etching process,the influence of the main parameters affecting the etching process(type of silicon doping,doping concentration,volume of hydrofluoric acid,volume of hydrogen peroxide and etching time)on the morphology of silicon nanopores was revealed;using density flooding theory(DFT)modeling was used to point out that the silicon nanopore formation is mainly determined by the bending energy band at the interface between gold nanoparticles and silicon wafers and the carrier concentration;on this basis,a machine learning approach was used to model the complex relationships among etchant formulation,etching time,silicon wafer doping type and doping concentration to obtain a narrow window of process parameters for processing regular silicon nanopore arrays on different silicon wafers;finally,the Finally,the regular sub-10 nm silicon hole arrays were processed on various Si wafers with different doping types and doping concentrations.The proposed machine-learning-assisted guided etching of sub-10 nm Si nanopore arrays will provide a feasible and cost-effective way to process high-quality single-nanometer precision nanostructure arrays and devices.