The Systematic Study Of Reactive Ion Etching Technology On Semiconductor Fabrication And The Micro-fabrication-based Metamaterials On Terahertz Modulation Devices

With the rapid development of the integrated circuit(IC)industry,the micro-fabrication technologies have been paid unprecedented attention recently.As one of the most crucial micro-fabrication technologies,the reactive ion etching(RIE)technique plays an essential role in IC fabrication due to its high etching rate and capability of anisotropic etching.At the same time,the interdependence of multiple etching parameters remains the challenge to build a clear-cut relation among them.Therefore,to reach the best performance of such technology,for instance,realize a perfect vertical sidewall,the role of each parameter needs to be clarified.Machine learning,as a practical approach to dimension reduction,provides an ideal tool to resolve this multivariable problem.Besides that,other applications based on micro-fabrication technology has also attracted much attention,especially the metamaterial based on the artificial unit cells.Nowadays,modern micro-fabrication makes it possible to fabricate structures in micro/nanoscale,which strongly support the application of metamaterials in the terahertz regime.Limited by the shortage of terahertz modulation devices,the study on terahertz technology is still in the beginning.Therefore,it is crucial to carry out the study of terahertz devices by using metamaterials.In the first part of this thesis,a systematical study of the reactive ion etching both experimentally and numerically through machine learning is carried out.The importance and influences of each etching parameter,including RF power,the ratio of gases,gas flow rate,and chamber pressure,have been investigated.The relations among them have also been discussed.Here,a silicon substrate is the targeting system,aiming to obtain a perfect vertical trench using a mixture gas of SF₆and O₂ in a capacitively coupled plasma reactive ion etching system(CCP-RIE).Furthermore,an artificial neural network(ANN)model is composed according to the machine learning algorithm,which is finally used to reveal the relations between parameters.The relative weight of parameters on etching rate and etching profile is calculated by using a Random Forest Algorithm.The second part of this thesis mainly focuses on the application of the micro-fabrication techniques upon terahertz metamaterials.Firstly,a new dual-frequency tunable narrow bandpass filter in the THz regime is proposed.By rotating the relative angle of the polarization orientation of incident THz waves to the device,it realizes a switchable center frequency of filter from 0.548 THz to 0.725 THz with the minimum transmittance of 86.5%and less than 77 GHz bandwidth(-3 d B).This high-quality bandpass performance is attributed to the excitation of trapped modes,which is excited by the induced antiphase current excitation between two metal layers.In addition,two aspects of dynamic modulation,including frequency modulation and transmission modulation,are concerned.Respectively,the strategy of resonance frequency modulation based on complementary metal layer structures and the strategy of dynamic transmission modulation based on graphene field-effect transistor is proposed.In the third part,a new-designed rapid thermal annealing(RTA)system with a novel double nested vacuum chambers feature.This RTA system can supply a high vacuum environment with a vacuum system.The annealing temperature can reach 1200?by using two silicon nitride heat devices.The annealing process is steady and stable,thanks to a full PID control system.The structure of the dual-chamber has been widely used in eliminating contact resistance and improving the quality of dielectric layers.