Low-temperature plasma and massively parallel simulations
Fierro, Andrew Steven
MetadataShow full item record
Quantifying the effects of vacuum ultraviolet (VUV) radiation on nanosecond pulsed plasma discharges is difficult due to the short timescales involved and difficulty in detecting radiation at these wavelengths as they are readily absorbed by molecular oxygen and many optical materials. Regardless, an experiment capable of measuring wavelengths in the VUV regime (120 nm < λ < 180 nm) was constructed. Complete time-resolved characterization of the emission spectra between 120 nm and 800 nm was carried out. Results indicated the presence of VUV emission before voltage collapse with a non-thermalized electronic energy level distribution. As the plasma transitions from the streamer to spark phase, the plasma becomes thermalized taking on the characteristic electronic temperature of 3.1 eV assuming a Maxwell-Boltzmann distribution. Further analysis showed that most emission from the discharge results from the nitrogen second positive system, although it becomes heavily quenched at high densities. Plasma measurements are often accompanied by simulation in assisting with quantifying many plasma parameters. However, the large range in densities encountered in simulation present a significant challenge to overcome due to the computational complexity. As such, many plasma simulation codes are accelerated through the use of parallel computing architectures (e.g. clusters, threads). Here, a fully three-dimensional particle-in-cell code with Monte Carlo collisions was implemented and accelerated using the NVIDIA CUDA environment for use on graphics processing units (GPU). The entire simulation was verified by comparison to known Boltzmann equation solvers. Results showed the influence of space charge on the breakdown process as it causes a non-thermalized electron energy distribution function. Additionally, a full photon suite including individual wavelengths was implemented into the simulation. This allowed for simulation of emission spectra from a non-thermalized energy state distribution and incorporation of wavelength dependent photo-proceeses.