Schematics for two implementations of an chirped pulse Fourier transform microwave (CP-FTMW) spectrometer are shown. In each panel, the spectrometer consists of three main components: (1) chirped pulse microwave generation, (2) molecular beam chamber, and (3) free induction decay detection. In panel A, the chirped pulse is generated by the combination an AWG operating a and a bandwidth extension circuit . The pulse is then amplified by a (or ) traveling wave tube amplifier (TWTA) and sent to the molecular beam chamber. The molecular free induction decay (FID) is amplified by a high gain, low-noise amplifier, down converted by mixing with an PDRO and digitized by a ( hardware bandwidth) digital oscilloscope. In panel B, the chirped pulse is generated by a AWG and upconverted by mixing with an phase-locked dielectric resonator oscillator (PDRO) in a broadband mixer (Miteq TB0440LW1). After upconversion, the pulse is amplified by a TWTA and sent to the experiment. The FID is amplified and directly digitized by a ( hardware bandwidth; using digital signal processing) digital oscilloscope.
The microwave circuit used for bandwidth extension of the chirped pulse generated by the AWG at sample rate is shown in detail. The two single-frequency PDRO sources (external AWG clock and for frequency conversion) are phase locked to the output of a rubidium-disciplined crystal oscillator. After the initial upconversion mixing stage, a single sideband must be selected for bandwidth multiplication. The lower sideband of the mixed pulse is selected by a bandpass filter. Bandwidth multiplication is achieved in two stages using a frequency quadrupler and doubler . Downconversion by using the frequency doubled output of the PRDO follows each frequency multiplier and produces the final desired chirped pulse with frequency span of . Manufacturers, part numbers, and brief descriptions for the components are provided in the supplemental information.
The chirped pulse produced by the bandwidth extension circuit is shown by following amplification by the TWTA. The pulse with duration of is shown in the time domain at the top. The fast Fourier transform of the pulse is shown in the middle panel to demonstrate the frequency bandwidth and frequency-dependent electric field amplitude of the pulse. Both the oscillations observed in the frequency domain and the fall off in intensity above are caused by the TWTA and are not observed in the driving pulse from the bandwidth multiplication circuit. In the bottom panel, we show the time-frequency analysis of the pulse to demonstrate the linear frequency sweep of the microwave pulse. The spurious signals in the pulses were measured to be at least lower in power than the instantaneous sweep frequency across the full range of the pulse.
The efficiency of the chirped pulse for creating the molecular polarization is illustrated by comparing the measured signals of the 1-propyne transition for chirped (black) and Gaussian (gray) pulses to equal frequency bandwidth and peak power. For the chirped pulse, the bandwidth is varied by adjusting the sweep rate at fixed pulse duration. The variation of the signal with pulse bandwidth is shown on the left. In the graph on the right, the same data are plotted vs the inverse of the square root of the bandwidth to show the signal scaling predicted by Eq. (4). Note that the peak signal intensities for both excitation methods are the same. Reference lines for bandwidths of , , and are included in the right panel.
A small frequency range of the broadband rotational spectrum of cyclopropylacetylene is shown by using four different digital filters of the time-domain free FID. The frequency of the transition of the normal species is close to the transition frequency of a species (indicated with an asterisk). When no digital filtering is performed on the FID (rectangular), the transition is observed on the broad wings of the normal species transition. For broadband spectroscopy, the baseline resolution of the spectrum can be improved by using window functions to suppress the signal leakage. The frequency spectra for three common signal processing window functions are shown. For spectra shown in this paper, we use the Kaiser–Bessel digital filter that provides of side lobe suppression. The dramatic improvements in baseline resolution are achieved at the expense of the linewidth and the Kaiser–Bessel window increases the linewidth at half maximum by a factor of 1.72 over the rectangular window.
The signals for the 1-0 rotational transition of OCS (0.4% OCS in a 20% helium, 80% neon inert gas mixture) measured by using the direct FID detection CP-FTMW spectrometer are shown. The left panel shows a small frequency range of the broadband spectrum obtained by using a single valve pulse measurement where a signal-to-noise ratio of 3500:1 is observed. The inset shows an expanded view of the isotopomers transition and includes a reference line for the frequency measured in the NIST cavity spectrometer. The height of this line corresponds to the predicted intensity based on the natural abundance. The right panel shows the spectrum after 4000 time-domain signal averages with an inset of the isotopomer in natural abundance.
The rotational spectrum of suprane (0.1% suprane in a 20% helium, 80% neon inert gas mixture) measured by the bandwidth extended CP-FTMW spectrometer is compared to simulated -type, -type, and -type spectra which use previously measured (Ref. 40) rotational constants (, , ) and dipole moment components (, , ). Above , the intensities measured by the CP-FTMW begin to drop below the predictions. This behavior is the result of poor coupling efficiency of the horn antenna at high frequencies as well as power drop of the TWTA. The bottom panel provides an expanded scale view of the spectral region inside the box shown in top panel.
The rotational spectrum of suprane is shown following a single acquisition, 100 signal averages, and 10 000 signal averages. The intense rotational transitions are easily observed after a single acquisition. After 10 000 signal averages ( data acquisition time), weaker -type transitions as well as transitions from less abundant isotopomers are observed.
A section of the rotational spectrum of cyclopropanecarboxaldehyde (0.1% in a 20% helium, 80% neon mixture) is shown on an expanded scale as measured by the CP-FTMW spectrometer (top) and the NIST FTMW spectrometer (a “mini” Balle–Flygare–type spectrometer (Ref. 28)-bottom). Rotational transitions for all three unique isotopomers, the isotopomer, and the normal species are labeled. The CP-FTMW spectrum was recorded after 10 000 signal averages ( of data acquisition), while the cavity FTMW spectrum was recorded by averaging ten shots at each frequency position ( steps). To record the full spectrum, the CP-FTMW spectrometer requires a factor of 50 less data acquisition time.
The signals in the region of the broadband rotational spectrum suprane measured in the CP-FTMW spectrometer by using one, two, and three pulsed valve nozzles are compared. The top panel demonstrates the linear scaling in signal levels for multiple nozzle source operation. These spectra were obtained by using 25 time-domain averages for a 1 (left), 2 (center), and 3 (right) nozzle systems. The bottom panel shows the number of signal averages and total number of valve pulses required to reach equivalent signal-to-noise ratios for a one (left), two (center), and three (right) nozzle systems. The use of multiple nozzles leads to a quadratic reduction in measurement time and a linear reduction in overall sample consumption.
The ability to acquire ten signal averages for each valve pulse is illustrated for suprane. The region of the broadband rotational spectrum is shown to illustrate the performance of the spectrometer. In the top row, panels A, B, and C show the spectra obtained using: (A) three nozzles, one valve trigger event (three total valve pulse), one FID measurement, (B) three nozzles, one valve trigger event (three total valve pulses), ten FID measurements, and (C) one nozzle, 90 valve trigger events (90 total valve pulses), 90 FID measurements. Panels B and C have approximately the same signal-to-noise ratio, despite the fact that the spectrum shown in panel C represents a longer measurement time (factor of 90) and increased sample consumption (factor of 30). The bottom row shows the individual spectra of the first (D), fifth (E), and tenth (F) FIDs making up the spectrum shown in panel B.
A comparison of the isotopomer rotational transitions for OCS (1-0) as measured by the bandwidth-extended (CP1) and direct detect (CP2) CP-FTMW spectrometers with the frequency measurements of the NIST FTMW cavity spectrometer and with the intensity measurements as expected from literature values for natural abundance.
Article metrics loading...
Full text loading...