Next-Generation Spectrometers for Rapid Analysis of Complex Mixtures

Spectrometers in chemical sensors are used in environmental and air quality monitoring, detection of hazardous gases and chemical warfare agents, and breath analysis in medical applications.

October 1, 2014

Office of Naval Research, Arlington, Virginia

Molecules have internal motions that are characteristic of their structure and identity. These motions can be studied by spectroscopy in different regions of the electromagnetic spectrum. Molecular spectra can be used as a “fingerprint” to unambiguously state whether or not a particular chemical is present and in what amount. As such, small portable spectrometers are frequently used as components of chemical sensors.

The 110-170 GHz spectrometer. (Left) The two-channel arbitrary waveform generator (AWG) produces both a constant frequency mixing pulse and a chirped pulse. Each pulse is filtered, frequency shifted by mixing with a 9.05-GHz local oscillator, and amplified before being sent to x12 multiplier chains. The AWG, digitizer, and phase-locked oscillator are all synchronized with a 10-MHz Rb oscillator. (Right) The 1-meter-long PVC sample cell, the vacuum system, and a few of the components, power supplies, and cables.

The spectrometers described here are used to study rotational motions of molecules. In principle, rotational spectroscopy can be used to detect any chemical compound with a dipole moment. Rotational spectra have traditionally been collected in the microwave/cm-wave region of the spectrum (2 – 50 GHz) at low temperatures (1 – 5 K). Achieving such low temperatures requires large, bulky instrumentation that is not suited for small, portable devices. Operation at room temperature significantly relaxes these requirements, but comes at the cost of signal intensity at cm-wave frequencies, meaning that optimal sensitivity will be obtained at millimeter or sub-mm frequencies, from 100 GHz to 1 THz.

A fast mm-wave spectrometer was developed that is amenable to miniaturization. The work was done in three main steps. First, a new FPGA-based digitizer that can average molecular signals in real time was used to improve data throughput to allow for the acquisition of very large numbers of averages. Second, a chirped-pulse mm-wave spectrometer was constructed using x12 amplified multiplier chains (AMCs) to generate mm-wave radiation from cm-wave sources, and to detect it via heterodyne mixing with a cm-wave local oscillator. Third, schemes to reduce the size of the spectrometer sample cell were tested.

The real-time digitizer can acquire 1 million averages of a 10 ms trace, sampled at 40 GS/s (record length of 400,000 points), in just under 12 seconds. Once the speed enhancement was confirmed, phase stability of the digitizer was tested by collecting 1 and 100 million averages of a molecular sample (nitromethane) and verifying the expected factor of 10 improvement in signal-to-noise ratios. After completing digitizer testing in the cm-wave range, the 110-170 GHz spectrometer was assembled and tested. Once chirped-pulse generation and detection were tested, methanol was used to find optimal conditions for spectrum collection.

Overall, this project had three primary goals. The first was to incorporate real-time digitizers into spectrometers to drastically increase data throughput. The second was to construct a millimeter-wave spectrometer using this technology to benefit from the natural thermal population of molecules at room temperature. The third was to then use these combined benefits to explore reductions in sample cell volume for future miniaturization efforts. The first goal was definitely met and the digitizer performance was excellent. Spectra of 1 billion averages were collected over the course of a few hours. The second goal was met, but with some caveats. Output power limitations of the amplified multiplier chain required use of relatively low bandwidth chirped pulses to sufficiently excite the molecules, which in turn meant that the full 60-GHz bandwidth could not be covered with each repetition cycle. The third goal was not met, due in part to the fragility of these devices, and possibly due to the low power handling ability of the millimeter-wave receiver, which has a damage threshold below that of the typical output power of the amplified multiplier chain. Multi-pass designs that have longer effective pathlengths, but reduce instrument footprint, are likely the best route.

This work was done by Steven Shipman of New College of Florida for the Office of Naval Research. ONR-0032