Nitrous oxide (N2O) is an important long lived greenhouse gas (GHG) with a global warming potential (GWP) of 265 compared to carbon dioxide (CO2) (GWP = 1) and methane (CH4) (GWP = 28)  . Unlike CO2 and CH4, the inventory of N2O emissions remains largely uncertain because some of the N2O emission sources have not yet been systematically quantified, such as those from agricultural sources, landfills, and waste water treatment systems. More importantly, some N2O sources may not have even been discovered yet.
The California Global Warming Solutions Act of 2006 (AB 32) requires California to reduce its GHG emissions to 1990 levels by 2020  . In order to achieve this goal, accurate GHG emission budgets are needed, in which an improved estimation of N2O emissions from California agriculture is essential. Various gas emissions, except for N2O, from silage have been investigated by various laboratory and field experiments    . The N2O emissions from silage corn cropland, but not from the corn silage, have been estimated worldwide    . Wang and Burries monitored nitrogenous gases produced by silage in both field and laboratory silos using a mass spectrometer  . They reported a maximum N2O concentration of 4.35% in the farm silo gas at about 54 hours after ensiling. Because N2O and CO2 both have a mass of 44, the N2O signals were obtained by differentiating them from CO2 peaks in the mass spectra, which may have introduced measurement uncertainty. Little information has become available on N2O produced by silage since the Wang and Burries measurements, because in the interim silage gases produced during the early phase of ensiling were studied mostly for toxicity  . Insufficient attention has been given to N2O as an important GHG gas emitted during the silage making process.
The present paper demonstrates that FTIR (Fourier Transform Infrared) spectroscopy can be used to measure N2O concentration in the complex mixture that comprises silage gas, because some of the N2O absorption peaks are isolated from other gas absorptions in the FTIR spectra. Since the FTIR spectroscopic technique has been used for real-time measurements of N2O and other gas species   , it is an excellent tool for real time monitoring of N2O in silage gas and may also be suitable for measuring N2O emissions in other anthropogenic emissions mixtures.
2. Materials and Methods
2.1. Fourier Transform Infrared (FTIR) Spectrometer
An FTIR spectrometer (Bruker Optics, Inc., Tensor 27, Germany) with spectral resolution of 1 cm−1 was used in the present study, which covers the spectral range from 850 to 7000 cm−1 with a mid-infrared (MIR) source, a KBr beam-splitter, and a Cryo-cooled MCT detector. A White Cell with variable optical path length between 3.2 - 64 meters and a volume of 16 liters was coupled to the FTIR spectrometer to increase gas absorptions in the White Cell.
2.2. LINEFIT Program
An instrumental line shape retrieval program, called LINEFIT, developed by Hase  at the Karlsruhe Institute of Technology in Germany was used to analyze the FTIR spectra to derive the gas concentrations inside the White Cell. The LINEFIT code is in wide use within the FTIR group of the NDACC (Network for Detection of Atmospheric Composition Change, http://www.ndsc.ncep.noaa.gov) and TCCON (Total Carbon Column Observing Network)  for the analysis of reference gas cell spectra recorded with high-resolution   , but has also been used for the analysis of spectra observed with low-resolution spectrometers  .
2.3. Laboratory Bucket Silos
3.1. FTIR Spectra of the Silage Gas and Verification of N2O Peaks
Gas samples were analyzed with Bruker Tensor 27 FTIR spectrometer coupled with a long-path White Cell. When switching gas samples in the White Cell, the White Cell was flushed three times by alternatively evacuating the chamber down below 1 hPa and filling with pure nitrogen (N2) to 1013 hPa. When analyzing pure N2 and standard N2O gas samples, the White Cell was filled with either one to 1013 hPa. To analyze the silage gas, 5 liters of silage gas from one 5-liter Tedlar bag plus 11 liters of high purity N2 were added to the White Cell to bring the pressure up to 1013 hPa in the 16-liter White Cell. The White Cell was operated at ambient pressure to prevent any contamination to gas samples due to possible leaks. Therefore, the air samples from the bucket silos had been diluted by a factor of 3.2 in the White Cell.
The optical path length of the White Cell was adjusted close to its maximum of 64 meters in order to maximize instrument sensitivity and therefore to detect weak absorption signals. First, the FTIR spectrum of the high purity N2 was recorded as background. Then, the FTIR spectrum of the silage gas plus N2 mixture was recorded. Significant N2O absorption peaks were found in the FTIR spectra of the silage gas. In order to confirm that N2O absorptions were identified correctly, the FTIR spectrum of a 5-ppm standard N2O gas was also recorded.
Figure 1 shows four spectra in the 1850 - 2850 cm−1 spectral region at 1 cm−1 spectral resolution: 1) FTIR spectrum of air samples from the bucket silo; 2) FTIR spectrum of 5-ppm standard N2O gas in N2 balance; 3) FTIR spectrum of pure N2; and 4) simulated spectrum of N2O, CO2, CO, and H2O gases generated by the LINEFIT program and Hitran 2008 spectral lines parameters  . The first three spectra in Figure 1 were measured
Figure 1. FTIR spectra of air samples from the bucket silos (green), 5-ppm standard N2O with N2 balance (red), and pure N2 gas (blue) in white cell recorded by a Bruker tensor FTIR spectrometer. A simulation spectrum of gases N2O, CO2, CO, and H2O by LINEFIT program is also shown (black) and upward shifted.
using the Bruker FTIR spectrometer coupled with a White Cell. The simulated spectrum takes into account not only the transmittance of all four gases but also the FTIR instrumental parameters.
The N2O absorption bands can be identified in the FTIR spectrum of the 5-ppm N2O standard gas and the simulation spectrum. The N2O absorption bands in the region of 2130 - 2270 cm−1 are the fundamental ν3 and associated “hot bands”, which are strong absorptions  . The N2O absorption bands located between 2400 - 2600 cm−1 include a harmonic band of 2ν1 and a combination band of ν1 + 2ν2  which are relatively weak absorptions. These N2O absorption bands cannot be seen in the spectrum of the pure N2 gas. Figure 1 shows that the absorption of the silage gas in the region of 2130 - 2270 cm−1 was saturated while the absorption of the 5-ppm N2O in the same region was not, indicating that the N2O concentration in the silage gas were much higher than 5 ppm. On the other hand, the absorption of the silage gas in the region of 2400 - 2600 cm−1 were not saturated and, by comparison, the absorption of the 5-ppm N2O gas in this region were much weaker, further confirming that N2O concentration in the bucket silos was much higher than 5 ppm.
A CO2 fundamental ν3 absorption band  located at 2349 cm−1 is shown in Figure 1 in the spectra of both the 5-ppm N2O gas and the pure N2 gas. Because CO2 should not be present in the White Cell in either case, this CO2 absorption came from CO2 gas outside the White Cell between 1) the FTIR MIR source and the White Cell, and 2) the White Cell and the FTIR MCT detector, because these two areas were exposed to ambient conditions. In the spectrum of the silage gas, the saturation between 2200 and 2400 cm−1 was due to the combined strong absorptions of CO2 and N2O. The CO2 3ν2 harmonic band at 2077 cm−1 is a weak absorption  that can be seen in the spectrum of the silage gas but not presented in the spectra of the 5-ppm N2O gas and the pure N2 gas.
3.2. Retrievals of FTIR Spectra Using LINEFIT Program
3.2.1. Selection of the Spectral Windows
The 5-ppm standard N2O spectrum in Figure 1 showed that 1) N2O absorption at the fundamental ν3 and associated “hot bands” was not saturated between 2130 and 2270 cm−1 and 2) the absorption of CO2 at the fundamental ν3 band located at 2349 cm−1 was not saturated either. Therefore, three spectral windows can be used to retrieve the 5-ppm N2O spectrum: i) 2140.0 - 2285.0 cm−1, ii) 2130.0 - 2650.0 cm−1, and iii) 2510.0 - 2615.0 cm−1.
In the spectrum of the silage gas, two spectral windows at 2140 - 2615 and 2510 - 2615 cm−1 were selected to retrieve the N2O concentrations. When fitting the spectral window at 2410 - 2615 cm−1, a special technique in the LINEFIT program, called “de-weighting”, was applied to the sub-window between 2300 and 2510 cm−1 because of the complexity of CO2 absorptions in this region.
3.2.2. Retrieval of FTIR Spectra of the 5-ppm Standard N2O Gas
Retrieval of the 5-ppm N2O spectrum was to confirm that the selected spectral windows can be used to accurately derive the N2O gas concentrations from the FTIR spectra. The retrieval of the 5-ppm N2O spectrum was also used to calibrate the optical path length of the White Cell. The retrieval of FTIR spectra using the LINEFIT program does not need any calibration when the total columns (in m−2) of the species are derived, but in order to convert from total columns to gas concentrations, the optical path length of the White Cell is needed. Because it is difficult to accurately estimate the White cell’s path length, calibration of the White Cell’s optical path length was performed using the 5-ppm N2O standard gas mixture.
Figures 2-4 show the retrieval results of the 5-ppm N2O spectrum at three spectral windows. Both FTIR and fitted spectra are shown on the low portion of the figures in red and blue, respectively. The differences between FTIR and fitted spectra, called “Residuals”, are shown on the upper portion of these figures. In principle, the LINEFIT program simulates a transmittance spectrum of all species at the given spectral window and fits it to the observed FTIR spectrum by scaling the total column of each species until the residuals fall in a preset threshold. The N2O columns in the White Cell along the optical path derived from the three spectral windows were 6.99 × 1021, 6.69 × 1021, and 7.18 × 1021 m−2, respectively. The optical path length of the White Cell was calibrated to be 57.6 m. The converted N2O concentrations were 4.94, 4.91, and 5.08 ppm respectively, which were −1.2%, −1.8%, and 1.7% different from the expected values of 5 ppm, resulting in an average of 4.98 ± 0.09 ppm. The ±0.09 ppm error bar was the standard deviation of the retrieved values from the three spectral windows. Uncertainties caused by general spectroscopic error (line intensities and broadening coefficients) and experimental error (temperature, pressure, instrument line shapes, etc.) were not accounted in the error bar; these contribute less than 10% to the error estimate for
Figure 2. Retrieval of the 5-ppm N2O spectrum at window 2140 - 2285 cm−1. The FTIR spectrum (red) and the fitted spectrum (blue) are shown on the bottom panel. The differences between FTIR and fitted spectra (residuals) are shown in top panel in black.
Figure 3. Same as Figure 2 except for window at 2140 - 2650 cm−1.
open-path FTIR measurements  . Similar uncertainty estimates apply to the error bars of the retrieval results in the rest of this paper so will not be repeated.
3.2.3. Retrieval of FTIR Spectra of the Silage Gas
Figure 4. Same as Figure 2 except for window at 2510 - 2615 cm−1.
Figure 5. Retrieval of the bucket silos samples spectrum at window 2510 - 2615 cm−1. The FTIR spectrum (red) and the fitted spectrum (blue) are shown on the bottom panel. The differences between FTIR and fitted spectra (residuals) are shown in top panel in black.
Figure 6. Same as Figure 5 except for at 2140 - 2615 cm−1 window with a sub-window at 2300 - 2510 cm−1 de-weighted (not shown).
Similar to Figures 2-4, both FTIR and fitted spectra are shown on the low portion of the figures in red and blue, respectively. Their residuals are shown on the upper portion of the figures. In Figure 5, the N2O column in the White Cell retrieved from 2510 - 2615 cm−1 window was 8.11 × 1023 m−2, equivalent to 573.8 ppm N2O in the White Cell. Figure 6 shows the 2140 - 2615 cm−1 window without the 2300 - 2510 cm−1 sub-window because this region was de-weighted. In the de-weighting spectral regions, the LINEFIT program neglects the residuals between observed and simulated spectra in order for the residuals from other spectra regions to fall below the pre-set threshold. This technique is usually used when there is a particular absorption feature that the LINEFIT program cannot handle correctly. The N2O column in the White Cell retrieved from this window without the sub-window was 7.98 × 1023 m−2, or 564.5 ppm. Because the N2O concentration in the Tedlar bag was diluted by a factor of 3.2 when the samples were introduced into the White Cell, the N2O concentrations in the silage gas are estimated to be 1836 and 1806 ppm, respectively, retrieved from the two spectral windows, resulting in an average of 1821 ± 21 ppm. These two estimates agree within 1.7%, similar to the uncertainty in retrieving concentrations from the 5-ppm N2O standard gas mixture spectrum.
FTIR spectroscopy was used to analyze N2O in a complex gas mixture produced in corn silage. The corn silage was made in laboratory bucket silos and silage gas samples were collected during the early phase of ensiling. A 5-liter Tedlar bag of the silage gas was analyzed with a Bruker FTIR spectrometer at 1 cm−1 spectral resolution coupled with a long-path White Cell. N2O absorption peak fingerprints were identified in the FTIR spectra of the silage gas using both standard N2O gas spectra and simulated spectra from the IR spectral database. This observation verified that N2O was produced in the bucket silos at the early phase of ensiling. To quantify the N2O concentration, the FTIR spectra were retrieved using the LINEFIT program. N2O concentrations retrieved from three spectral regions of the 5-ppm N2O spectrum were 4.98 ± 0.09 ppm confirming that N2O can be accurately retrieved from FTIR spectra using LINEFIT program. An N2O concentration of 1821 ± 21 ppm was derived from two spectral regions of the silage gas FTIR spectra indicating that N2O concentration in the silage gas was very high. Wang and Burries  measured N2O concentrations in the farm silos using mass spectroscopy and reported that N2O varied from 10,000 to 43,500 ppm in the first 66 hours and peaked at about 54 hours after ensiling, which was about 5.5 - 23.9 times greater than the N2O concentration in the laboratory bucket silos measured in the present study. The difference between the two measurements could be due to several factors: 1) N2O and CO2 both have a mass of 44, so the N2O signals were obtained by differentiating them from CO2 peaks in the mass spectra, which might have introduced uncertainty in the mass spectroscopic measurements, 2) The differences in sampling locations and methods between the two measurements, and 3) The N2O concentrations between farm silos and laboratory bucket silos might be truly different.
The primary goal of this work was to 1) verify that N2O was present in silage gas during the early phase of ensiling based on FTIR spectroscopy, 2) demonstrate that N2O concentrations in silage gas can be accurately retrieved from FTIR spectra using the LINEFIT program, and 3) confirm that FTIR spectroscopy is a viable method for measuring N2O concentrations in complex gas mixtures, such as the silage gas. Since only one bag of silage gas was analyzed by FTIR for this study, the concentrations measured should be considered preliminary. Due to limited amounts of silage and silage gas sampled, the concentrations measured in this work should not be construed a quantitative estimate of N2O emissions from silages. In order to quantitatively investigate the N2O emissions from silage, an extended study is needed, including measurements of representative silage mixtures and ages of gas samples during silage making process. Most importantly, research is needed on methodology to reduce the N2O emissions during the silage making process. This work is intended as a pilot work to demonstrate the efficacy of using FTIR to measure N2O concentrations in complex mixtures such as emissions from silage.
This work was partly supported by California Air Resources Board (CARB) under research contract #11-325 and the Academic Federation Professional Development Award 2015-2016 by University of California-Davis. Y. Z would like to thank the Karlsruhe Institute of Technology for hosting a 2-week visit.
 Myhre, G., et al. (2013) Anthropogenic and Natural Radiative Forcing. In: Stocker, T.F., Qin, D., Plattner, G.-K., Tignor, M., Allen, S.K., Boschung, J., Nauels, A., Xia, Y., Bex, V. and Midgley, P.M. Eds., Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change, Cambridge University Press, Cambridge, UK and New York, USA, p. 659.
 California Air Resources Board (CARB) (2006) California Global Warming Solutions Act of 2006 (AB32).
 Hafner, S.D., Howard, C., Muck, R.E., Franco, R.B., Montes, F., Green, P.G., Mitloehner, F, Trabue, S.L. and Rotz, C.A. (2013) Emission of Volatile Organic Compounds from Silage: Compounds, Sources, and Implications. Atmospheric Environment, 77, 827-839.
 Hafner, S.D., Franco, R.B., Kung Jr., L., Rotz, C.A. and Mitloehner, F. (2014) Potassium Sorbate Reduces Production of Ethanol and 2 Esters in Corn Silage. Journal of Dairy Sciences, 97, 7870-7878.
 Linn, D.M. and Doran, J.W. (1984) Effect of Water-Filled Pore Space on Carbon Dioxide and Nitrous Oxide Production in Tilled and Nontilled Soils. Soil Science Society of America Journal, 48, 1267-1272.
 Van Groenigen, J.W., Kasper, G.J., Velthof, G.L., van den Pol-van Dasselaar, A. and Kuikman, P.J. (2004) Nitrous Oxide Emissions from Silage Maize Fields under Different Mineral Nitrogen Fertilizer and Slurry Applications. Plant and Soil, 263, 101-111.
 Liu, C., Yao, Z., Wang, K. and Zheng, X. (2014) Three-Year Measurements of Nitrous Oxide Emissions from Cotton and Wheat-Maize Rotational Cropping Systems. Atmospheric Environment, 96, 201-208.
 Wang, L.C. and Burries, R.H. (1960) Mass Spectrometric Study of Nitrogenous Gases Produced by Silage. Agricultural and Food Chemistry, 8, 239-242.
 Zhao, Y., Strong, K., Kondo, Y., Koike, M., Matsumi, Y., Irie, H., Rinsland, C.P., Jones, N. B., Suzuki, K., Nakajima, H., Nakane, H. and Murata, I. (2002) Spectroscopic Measurements of Tropospheric CO, C2H6, C2H2, and HCN in Northern Japan. Journal of Geophysical Research: Atmospheres, 107, ACH 2-1-ACH 2-16.
 Wunch, D., Taylor, J.R., Bernath, D. Fu, P., Drummond, J.R., Midwinter, C., Strong, K. and Walker, K.A. (2007) Simultaneous Ground-Based Observations of O3, HCl, N2O, and CH4 over Toronto, Canada by Three Fourier Transform Spectrometers with Different Resolutions. Atmospheric Chemistry and Physics, 7, 1275-1292.
 Hase, F., Blumenstock, T. and Paton-Walsh, C. (1999) Analysis of the Instrumental Line Shape of High-Resolution Fourier Transform IR Spectrometers with Gas Cell Measurements and New Retrieval Software. Applied Optics, 38, 3417-3422.
 Hase, F. (2012) Improved Instrumental Line Shape Monitoring for the Ground-Based, High-Resolution FTIR Spectrometers of the Network for the Detection of Atmospheric Composition Change. Atmospheric Measurement Techniques, 5, 603-610.
 Hase, F., Drouin, B.J., Roehl, C.M., Toon, G.C., Wennberg, P.O., Wunch, D., Blumenstock, T., Desmet, F., Feist, D.G., Heikkinen, P., De Mazière, M., Rettinger, M., Robinson, J., Schneider, M., Sherlock, V., Sussmann, R., Té, Y., Warneke, T. and Weinzierl, C. (2013) Calibration of Sealed HCl Cells Used for TCCON Instrumental Line Shape Monitoring. Atmospheric Measurement Techniques, 6, 3527-3537.
 Frey, M., Hase, F., Blumenstock, T., GroB, J., Kiel, M., Mengistu Tsidu, G., Schafer, K., Sha, M.K. and Orphal, J. (2015) Calibration and Instrumental Line Shape Characterization of a Set of Portable FTIR Spectrometers for Detecting Greenhouse Gas Emissions. Atmospheric Measurement Techniques, 8, 3047-3057.
 Johnson, H.E., Merry, R.J., Davies, D.R., Kell, D.B., Theodorou, M.K. and Griffith, G.W. (2005) Vacuum Packing: A Model System for Laboratory-Scale Silage Fermentations. Journal of Applied Microbiology, 98, 106-113.
 Rothman, L.S., et al. (2009) The HITRAN 2008 Molecular Spectroscopic Database. Journal of Quantitative Spectroscopy &. Radiative Transfer, 110, 533-572.
 Plyler, E.K., Tidwell, E.D. and Maki, A.G. (1964) Infrared Absorption Spectrum of Nitrous Oxide (N2O) From 1830 cm-l to 2270 cm-l. Journal of Research of the National Bureau of Standards-A. Physics and Chemistry, 68A, 79-86.
 Plyler, E.K. and Barker, E.F. (1931) The Infrared Spectrum and the Molecular Configuration of N2O. Physical Review, 38, 1827-1836.
 Martin, P.E. and Barker, E.F. (1932) The Infrared Absorption Spectrum of Carbon Dioxide. Physical Review, 41, 219-303.
 Smith, T.E.L., Wooster, M.J., Tattaris, M. and Griffith, D.W.T. (2011) Absolute Accuracy and Sensitivity Analysis of OP-FTIR Retrievals of CO2, CH4 and CO over Concentrations Representative of “Clean Air” and “Polluted Plumes”. Atmospheric Measurement Techniques, 4, 97-116.