Manufacturing and Testing of Laser-Induced Blackbody Emitters

Optical properties of surfaces are important for a wide variety of energy applications. Emissivity is one property particularly relevant for applications involving radiative energy transport, and measures the amount of light a surface can emit compared to its maximum. Oftentimes, the emissivity spectrum, or emissivity as a function of wavelength, must be tuned for different applications. For example, thermophotovoltaic emitters should have high emissivity above the bandgap of the photovoltaic cell to ensure maximum power output from the cell, and low emissivity below the bandgap to prevent parasitic heating.<br/><br/>We have pioneered a laser processing technique to tune the optical properties of surfaces. In this technique, laser pulses are used to ablate away material, creating metasurfaces of microscale hills and valleys dotted with nanoparticles. From this processing technique, we can achieve a variety of possible emissivity spectra. However, it is difficult to know a priori what emissivity spectrum will result from a given set of laser processing parameters. This is because the laser ablation is a highly multiphysical process, so it is difficult to predict the surface profile, and it is further difficult to predict the optical properties of a given surface profile due to the expensive finite-difference-time-domain simulations required.<br/><br/>In this work, we directly map the laser processing parameters to the emissivity spectrum, bypassing the surface profile. There are three laser processing parameters: power, speed, and spacing. In order to create this mapping, we developed a high-throughput experimentation technique involving varying the laser processing parameters, creating 35,280 unique combinations. We then used each combination of processing parameters to engineer the surface of stainless steel, and measured the emissivity spectrum of each combination. Using this dataset, we develop a mapping between laser processing parameters and emissivity spectra, first with an forward machine learning model combined with global optimization to determine which laser parameters were ideal for a desired spectrum, and second with an inverse model to directly output the laser parameters for a desired spectrum.<br/><br/>Using the laser processing parameters identified, we manufacture a near-blackbody metasurface on tantalum (Ta) to increase its emissivity (~0.98 between 0.2 and 10 microns). We then use this surface as an emitter for thermophotovoltaic (TPV) power conversion. We used an experimental apparatus featuring an actively-cooled 1.2/1.0 eV tandem TPV cell with an Au back-reflector spaced 5mm away from the emitter. We show that the Ta metasurface more than doubles the TPV power density from 2.18 to 4.91 W/cm2 at 2100C while retaining the high efficiency of 28%.<br/><br/>We then demonstrate the thermal stability of the metasurface with both high temperature and long timescale annealing. For high temperature testing, we conduct step heating from 2100 to 1500C in increments of 100C for 10 minutes each, noting the above-bandgap emissivity drops from 0.98 to 0.95. For long-timescale annealing, we conduct heating at 1500C for 100 hours, noting the broadband emissivity decreases from 0.98 to 0.92. We supplement these measurements with an annealing model considering both evaporation through vapor pressure and surface diffusion.<br/><br/>Overall, we have developed a end-to-end methodology for designing metasurfaces for thermophotovoltaic applications. We first determine which laser processing parameters result in a desired emissivity spectrum of a metasurface. Then, we use the metasurface as an emitter for TPV, demonstrating the benefits of increased emissivity. Lastly, we demonstrate thermal stability with long-duration, high-temperature testing and modeling. This work demonstrates a promising way forward for high-performing metasurfaces for TPV applications.