Tailored ultrafast supercontinuum generation in an inverse-designed waveguide based on machine learning and 3D nanoprinting technology

点击翻译按钮获取中文摘要

Supercontinuum sources, with the unique combination of broad bandwidth, spatially coherent radiation and extreme spectral brightness, are powerful tools for spectroscopic applications in the field of life science. However, the spectra delivered by supercontinuum sources usually greatly overspan the spectral region of interest, leaving most of the energy unused. Therefore, tailored supercontinuum generation with well-defined spectral properties is highly attractive for spectroscopy. It is well known that the key to tailored supercontinuum generation is either dispersion management of the nonlinear waveguides or direct control of the input ultrafast pulses. Since the latter method has limited controllable variables and only influences the initial stage of supercontinuum generation, the dispersion management strategy is more promising for tailored supercontinuum generation. The main objective of the project “Tailored ultrafast supercontinuum generation in an inverse-designed waveguide based on machine learning and 3D nanoprinting technology” is to achieve sophisticated dispersion management of a waveguide and thus tailored supercontinuum generation. The waveguide is homogeneously made of glass or polymer, or a hybrid of glass/liquid and polymer. Note that the longitudinal dispersion profile is inverse-designed from the user-defined supercontinuum spectrum via machine learning technology (based on supervised neural network and/or genetic algorithm) under the condition that all parameters of the input pulses are fixed. Moreover, this well-designed dispersion landscape will be experimentally implemented by axial cross-section modulation of the waveguide on the sub-micrometer scale, which is otherwise difficult to be implemented, thanks to the high precision of the 3D nanoprinting technology. Due to the design and fabrication flexibility offered by this proposed photonic platform, much more complex scenarios of supercontinuum generation can be expected. The concept of inverse-design will be evaluated with respect to finding the most suitable dispersion configuration for a certain desired output, such as covering desired waveband(s), or exhibiting arbitrary user-defined spectral features. Overall, the project will develop a straightforward-to-implement tailored supercontinuum generation methodology that can achieve desired spectra with optimal spectral shapes and densities, resolving the demands in many spectroscopy-related applications.