Methane and water vapor
Methane and water vapor were measured in the 3.27 micron region, which includes a single well-resolved water line and three partially blended methane lines in each laser scan. This frequency was obtained by mixing single frequency semiconductor lasers operating in the 1,053 and 1,553 nm regions. All three methane lines are fit, leveraging the fixed relative line strengths, all of which are for the 12CH4 isotopologue. For methane in particular, monitoring concentration via the 3 micron band is highly desirable as the associated absorption intensity is approximately 180 times stronger than those of the more commonly used 1.6 micron overtone band. Figure 2 shows the analyzer concentration data for a calibration gas at approximately 2 ppm (ambient level) over a period of 24 h. Also shown in Fig. 2 is the Allan Variance for the above time series, a metric now commonly used to quantify precision and stability of laser-based analyzers [6]. Here the first data point indicates a precision level of approximately 250 ppt (10-s average/0.1 Hz) and a minimum of about 50 ppt after approximately 9 min of signal averaging. This level of performance satisfies the World Meteorological Organization (WMO) guidelines for ambient methane monitoring, which are highlighted in the green-shaded area of the plot.
Temperature cycling and system drift
Figure 3 shows the typical analyzer performance, with temperature cycling between 15–30 °C in an environmental chamber, using a 5 °C/hr ramp rate at 7.5 °C intervals. Data are shown both for the 10-s average and 60-min smoothed result. The associated temperature profile is also shown but is slightly offset as it is recorded inside the case, where the heat from the nearby power supplies systematically elevates the value by a few degrees. In this case, a peak-to-peak variation of 0.7 ppb is obtained for the 60-min averaged data, with little or no hysteresis evidenced. This level of performance is consistent with stringent ambient research requirements for greenhouse gases. Similar performance is evidenced below for other instruments operating in the 4–5 micron spectral region.
The drift apparent in these data are primarily due to the presence of residual optical fringes in the optical train, which in the case of a DFG laser can occur at three wavelengths: the pump, signal, and idler wavelengths in the 1, 1.5, and 3.3 micron regions, respectively, within different parts of the system. In this case, the 1 micron laser is fixed in frequency, and thus any fringes including and upstream of the PPLN crystal tend to be fixed and thus primarily lead to offsets that are easily calibrated. However, tuning of the 1.5 micron laser and the associated translation into the MIR can result in fast as well as slowly varying fringes that can occur throughout the optical system. Such etalons can in some cases be isolated and minimized, yet can be strongly influenced by minute temperature changes inside the optical core.
Thermal control of the lasers as well as optical core comprising the optical bench and multipass cell are key to minimizing drift due to optical fringes, which will typically have less of an effect on the extracted concentrations as long as their period is substantially either narrower or broader than the absorption linewidth. In addition to minimizing the magnitude of the fringes, the etalons should not drift appreciably over time, which means the lasers, associated drive electronics, and physical dimensions of the core must all be as stable as possible. When subjected to relatively rapid external temperature changes as in the above tests, transitory fringing is indeed the primary source of drift in the system as all of these systems reequilibrate to the new temperature, which typically takes tens of minutes. In some cases, temperature appears to correlate well with drift, but in others it appears to be anticorrelated. This phenomenon happens as the phase can change depending on whether riding up or down an associated fringe, and thus typically these drift sources cannot be removed via single or even multiple temperature measurements within the system. This phenomena appears to be prevalent at a significant level in nearly all high precision, narrow linewidth laser-based analyzers, regardless of the technique employed as, at the most basic level, all of these instruments comprise highly sensitive optical interferometers. Given that physical changes at the nm level translate into frequency scanning at the MHz level, temperature changes at the mK level are sufficient to produce a measurable effect on the spectrum when the fidelity required is at the 5,000:1 level or better. Similarly, electronics can in some cases produce measurable effects at nA (or nV) levels of drift. As such, in addition to careful design of the optical components, validation of those components performance, and careful optical alignment, thermal testing of the entire system is presently required to validate analyzer performance for ambient GHG monitoring applications.
Methane field data: Sutro tower
Field testing of a methane analyzer (Model IRIS 5500) at Sutro Tower in the San Francisco Bay Area was performed working in collaboration with Lawrence Berkeley National Lab, which has been operating tower measurements including collaborative U.S. NOAA flask analyses of methane and other long-lived greenhouse gases. The analyzer in this test was placed inside of an enclosure that enables the system to run in an environmentally uncontrolled shed, wherein it was subjected to significant temperature variation and high moisture, including condensing fog and some rain. The monitoring shed was located at the base of Sutro tower and was relatively open to the environment with partial sheltering from the elements. The enclosure was primarily used to shelter the instrument from rain and fog, while air was circulated through the enclosure for cooling. The sample air was provided from the downstream flow of an existing flask sampling system, which was connected to a draw tube that sampled air from a height of 232 m above ground level. The instrument sample flow rate was fixed at approximately 300 sccm, and the internal cell pressure was maintained at precisely 225 mbar by controlling the small internal diaphragm pump rotational speed. System calibration was achieved using the built-in gas manifold and calibration software, wherein the duration and period of the calibration cycle is set by the user. The instrument was installed in September 2011 and remained in service continuously until approximately mid-December 2011.
Figure 4 displays a typical time series for methane dry mole fraction (DMF) obtained at Sutro tower, in this plot comprising approximately 20 days of autonomous operation. In this case, regular calibration cycles were programmed to run every 6 h, wherein the instrument drift was corrected automatically. In these tests, a total variation of only 2 ppb was noted in the data, most of which occurred in the first week of operation. Typical calibrations at only the 100 ppt level were required after the first week of continuous operation. In either case, high fidelity methane DMF data were obtained at the site, with excursions at the 100 ppb level commonly occurring. The accuracy obtained with the analyzer was at the same level evidenced in the laboratory, across wider temperature ranges than those conducted in the environmental data above.
Carbon monoxide
The wavelength agility of the MIR laser-based analyzer platform enables numerous other important species to be measured, including carbon monoxide, carbon dioxide, and nitrous oxide. Carbon monoxide was measured via discrete lines in the 4.6 micron region, in this case, mixing semiconductor lasers operating in the 1,170- and 1,574-nm regions, the former comprising a novel quantum-dot-based laser. For species such as CO, mid-infrared absorption strength is more than three orders of magnitude times stronger than the associated near-infrared bands in the 1.5 micron region, in this case, directly resulting to superior sensitivity even compared with cavity-based NIR sensors. Figure 5 shows a 20-h time series measurement from a calibration tank of CO at the 475 ppb level, in this case with data points recorded every second. Also shown is the corresponding Allan Variance, indicating a sensitivity level of <1 ppb is achieved in approximately 10 s, with an ultimate sensitivity of 140 ppt achieved after approximately 6 min of averaging. In this case, the WMO ambient monitoring target is satisfied at 1 s.
Figure 6 shows the result of ambient CO measurement in the laboratory over a period of several days. Here, the nearby local commuter traffic is clearly evident during the weekdays as both the morning and afternoon commute (9 am and 5 pm, resp.), which is approximately 1 mile away (U.S. Route 101). Additionally, decreased CO is noted during rainy periods, reducing both the average as well as peak events significantly. During the night, the air conditioning is off, and air exchange limited, leading to relatively constant CO levels in the building. Shortly after 6 am on each day the air conditioning comes on and outside air is introduced. For the late night periods when the ventilation is off, the standard deviation is approximately 440 ppt for a 10-s measurement. This level of precision suitable for high fidelity source attribution, as correlations of CO with CO2 can be used to discriminate, for example, anthropogenic versus biogenic sources CO2.
Nitrous oxide
Nitrous oxide is the third most important greenhouse gas, and it is presently not possible to measure ambient levels accurately in the near-infrared using existing methods due to the combination of low overtone absorption with typical sub-ppm concentrations. Nitrous oxide is extremely well mixed in the atmosphere, evidenced by a pole-to-pole variation of only 1 ppb. WMO guidelines for nitrous oxide measurements thus desire accuracies in the 150–200 ppt regime, presenting a significant challenge for fielded instruments. There are presently three commercial laser-based N2O analyzers that operate in the middle infrared, which are based on relatively expensive QC-DFB lasers (Los Gatos Research, Aerodyne, and Campbell Scientific), and these analyzers are both relatively large and expensive as a result. Here, nitrous oxide was measured using the DFG-based IRIS Model 4600, which measures absorption in the 4.5 micron region. In this case, the aforementioned 1,170 nm quantum-dot laser is mixed with a suitable DFB diode laser in the 1.5 micron region. In addition to N2O absorption, this instrument also simultaneously measures water vapor absorption to determine DMF. Figure 7 shows a typical time series obtained on a calibration tank for a 10-s average, together with the associated Allan Variance. In this case, a precision of 180 ppt was achieved in 10 s, with an associated typical sensitivity of 50 ppt achieved at approximately 100 s. The Allan Variance here indicates a calibration interval on the order of an hour to realize this level of accuracy routinely in the field. Presently, N2O analyzers have been located at field sites and continue to demonstrate comparable performance. Long-term field testing for methane and nitrous oxide analyzers are ongoing at field sites in California, Colorado, and Illinois.