Robust Bayesian parameter estimation for spectral encoded single-shot THz spectroscopy

Area of Science:

• Terahertz spectroscopy
• Plasma physics
• Nonlinear optics

Background:

• Terahertz time-domain spectroscopy (THz-TDS) uses spectral encoding with chirped probe pulses.
• Bandwidth-chirp tradeoffs in THz-TDS cause spectral nulls and phase nonlinearities, complicating analysis.
• Accurate refractive property extraction requires models accounting for the THz-TDS detection process.

Purpose of the Study:

• Develop and validate a dispersion-detection forward model for THz-TDS.
• Incorporate detection process, dielectric dispersion, and chirped probe pulse properties into the model.
• Improve parameter estimation for plasma diagnostics and nonlinear phenomena.

Main Methods:

• Developed a dispersion-detection forward model for THz-TDS.
• Employed a Bayesian framework for parameter estimation, yielding posterior probability distributions.
• Validated the model by measuring refractive properties of an inductively coupled argon plasma.

Main Results:

• The validated model reduced peak residuals by approximately 64% compared to direct fitting.
• Accurate ranges for electron density ([1.55, 2.39] × 10^19 m^-3) and collision frequency ([2.00, 2.46] × 10^11 rad/s) were obtained.
• The framework was applied to quantify refractive index changes during nonlinear bistability in plasma sources.

Conclusions:

• The developed dispersion-detection forward model accurately characterizes plasma refractive properties using THz-TDS.
• The Bayesian framework provides robust parameter estimation, accounting for spectral encoding effects.
• This approach enhances the analysis of complex plasma dynamics and nonlinear optical phenomena.