A broadband Fourier transform microwave spectrometer based on chirped pulse excitation

Abstract

Designs for a broadband chirped pulse Fourier transform microwave (CP-FTMW) spectrometer are presented. The spectrometer is capable of measuring the 7 – 18 GHz region of a rotational spectrum in a single data acquisition. One design uses a 4.2 Gsamples ∕ s arbitrary waveform generator (AWG) to produce a 1 μ s duration chirped pulse with a linear frequency sweep of 1.375 GHz . This pulse is sent through a microwave circuit to multiply the bandwidth of the pulse by a factor of 8 and upconvert it to the 7.5 – 18.5 GHz range. The chirped pulse is amplified by a traveling wave tube amplifier and broadcast inside the spectrometer by using a double ridge standard gain horn antenna. The broadband molecular free induction decay (FID) is received by a second horn antenna, downconverted, and digitized by a 40 Gsamples ∕ s ( 12 GHz hardware bandwidth) digital oscilloscope. The second design uses a simplified pulse generation and FID detection scheme, employing current state-of-the-art high-speed digital electronics. In this spectrometer, a chirped pulse with 12 GHz of bandwidth is directly generated by using a 20 Gsamples ∕ s AWG and upconverted in a single step with an ultrabroadband mixer. The amplified molecular emission is directly detected by using a 50 Gsamples ∕ s digital oscilloscope with 18 GHz bandwidth. In both designs, fast Fourier transform of the FID produces the frequency domain rotational spectrum in the 7 – 18 GHz range. The performance of the CP-FTMW spectrometer is compared to a Balle–Flygare–type cavity-FTMW spectrometer. The CP-FTMW spectrometer produces an equal sensitivity spectrum with a factor of 40 reduction in measurement time and a reduction in sample consumption by a factor of 20. The CP-FTMW spectrometer also displays good intensity accuracy for both sample number density and rotational transition moment. Strategies to reduce the CP-FTMW measurement time by another factor of 90 while simultaneously reducing the sample consumption by a factor of 30 are demonstrated.