Carbon Monoxide Emissions from Corn Silage

Carbon monoxide (CO) emissions from corn silage were observed using Fourier Transform Infrared (FTIR) spectroscopy and laser spectroscopy. In the first experiment, corn silage was produced using laboratory bucket silos. Air samples were collected from the bucket silos during the first week of en-siling and analyzed using a low-resolution Bruker FTIR spectrometer coupled with a long optical path length White Cell. The CO concentration in the bucket silo gas, derived from the FTIR spectra using the LINEFIT program, was as high as 48.0 ppm. In the second experiment, air samples were collected through a flux chamber from an Ag-Bag silage pile on a commercial dairy that was opened several months after ensiling. The Ag-Bag air samples were analyzed using a high-resolution Bruker FTIR spectrometer, and CO concentrations were retrieved to be 6.83 ppm, corresponding to an area emission rate of 33.7 mg/(hour∙m 2 ). An LGR N 2 O/CO gas analyzer based on infrared laser spectroscopy was also used to measure the CO concentrations from the same flux chamber. Elevated CO concentrations were observed from these si-lage sources. The present study revealed that CO was emitted by corn silage during different phases of ensiling. Annual CO emissions from the corn silage were estimated to be much lower than those from the well-known emission categories in San Joaquin Valley (SJV) and California, but comparable to those from food and agricultural sources in the SJV. It is also confirmed that FTIR spectroscopy is a viable method for measuring CO concentrations in complex gas mixtures, such as silage gas.


Introduction
Carbon monoxide (CO) is a toxic criteria pollutant.Because it has about 280 How to cite this paper: Zhao, Y.J., Wexler, A.S., Hase, F., Pan, Y. and Mitloehner, F.M. (2021) Carbon Monoxide Emissions from times affinity for hemoglobin, inhaling high concentrations of CO can result in asphyxiation.At lower concentrations, CO can lead to headaches, lack of concentration and other health effects.The OSHA standards prohibit worker exposure to CO concentrations of more than 50 ppm in air averaged during an 8-hour time period [1].
Like volatile organic compounds (VOCs), CO forms ozone in the troposphere by reacting photochemically with nitrogen oxides (NO x ) (e.g.[2]).Although the O 3 forming potential of CO is very low in comparison with most VOCs (e.g.[3] [4]), the contributions of CO to the tropospheric O 3 budget cannot be neglected because CO concentrations may be substantially higher.Whereas tropospheric VOC concentrations are typically in the tens of ppb levels, CO concentrations vary from hundreds ppb in clean areas to ppm levels in polluted areas and are much higher near CO emission sources (e.g.[5] [6]).
CO has been classified as one of the six criteria pollutants in the US together with O 3 , lead, sulfur oxides (SO 2 ), nitrogen oxides (NO x ), and particulate matter (PM) [7].Emission sources of CO have been widely studied and most of them are well known.A very common emission source of CO is the incomplete combustion of carbonaceous materials, such as gasoline, oil, coal, and wood, in natural and anthropogenic processes such as motor vehicles equipped with internal combustion engines, wildfires and biomass burning.However, there may be some unexpected CO emission sources that warrant further investigation.
Silo gas is produced during ensiling, particularly in the initial few weeks of ensiling.Because silo gas had been reported to cause serious respiratory disease and death of humans working near or in silos ( [8] and reference therein), it is a well-known poisonous gas mixture.The majority of silo gas is nitrogen dioxide (NO 2 ), which can cause human respiratory disease and even be lethal.In addition to NO 2 , other chemical components were also observed in the silo gas, such as CO 2 , N 2 O, VOCs [9] [10] [11] [12] [13].Because silo gas is a complex mixture, it may contain constituents that have not been identified yet and might be of relevance for atmospheric chemistry, air pollution, and public health.The present paper reports our observations of CO emitted by corn silage, which constitutes a new CO emission source and therefore a potential concern for both air quality and public health.
The San Joaquin Valley (SJV) in central California has the worst air quality designation in the US in terms of ground-level O 3 and PM.At the same time, the density of dairy farms in the SJV is the highest in the US, which feeds more than 16% of US milk cows in 2019 and 2020 [14] [15].Because silage is a major feed for dairy cows and is produced locally from fresh corn crops, production and consumption of silage in the SJV may contribute to O 3 production through the CO + NO x O 3 cycle, as well as the VOC + NO x O 3 cycle.According to USDA statistical reports, 10.1 million tons of corn for silage were produced in California in 2020, which was about 7.3% of total production in the US and placed California the second place in corn silage production [16] [17].
Although many studies have been conducted on silage emissions (e.g.[11] [12] [13] [18] [19]), thorough studies using new monitoring techniques would identify more unknown components emitted by silage, determine their emission rates, and provide more useful data to help with reducing silage emissions.The present paper also demonstrates that FTIR spectroscopy can be used to investigate complex mixtures in air emissions, such as those from silages.Because the FTIR spectroscopic technique has been used for real-time measurements of trace gaseous species (e.g.[20] [21]), it is an excellent tool for real-time monitoring of air emissions from silages and other complex sources.

Laboratory Bucket Silos, Ag-Bag Silage, and Air Sample Collections
The present paper contains data from two different projects.Because the infrared spectroscopic measurements of CO in silage emissions were not included in the original two project plans, the present paper reports these additional findings.
Various types of laboratory silos had been used for decades to simulate the conditions of large-scale silos for different purposes (e.g.[22] [23] and references therein).Making corn silage in laboratory bucket silos and collecting gas samples from the buckets have been described by Hafner et al. [12].Twelve kilograms of corn forage were manually compressed in 18.9-liter buckets.Excess silage gases from the bucket silos filled 5-liter Tedlar bags (SKC-West Inc., Fullerton, CA) during the first week after ensiling.Gas samples were then analyzed by various analyzers.Because the bucket silo experiment was originally designed to evaluate the effects of silage additives on the production of volatile organic compounds (VOCs) within corn silage using gas chromatography, only one bag of silage gas was analyzed using the FTIR spectrometer.In addition to reporting VOC measurements, Hafner et al. [12] presented typical properties of the silage produced in the bucket silos and concluded that they were within or close to the recommended values.Therefore, the corn silage produced in the bucket silos was normal, and the gas samples collected from the bucket silos should be representative of the regular ensiling processes.
Making and storing silage in so called Ag-Bags is a relatively new technique to produce silage under commercial conditions.Unlike the traditional silage piles or silos, large elongated plastic bags were filled with chopped corn forage, compressed, and sealed for ensiling.Excessive gas produced in the initial phase of ensiling was periodically released by small vent holes during the first 1 -2 weeks of ensiling.Then the plastic bags were sealed for several months until the silage inside the Ag-Bag was ready for feeding.
Collecting air samples from emission surfaces using flux chambers has been widely reported (e.g.[24] [25] and references therein).In the present study, a portion of the silage face surface was covered by a flux chamber as soon as the Ag-Bag was opened and defaced for feeding, as shown in Figure 1 (the dome shaped transparent device was the flux chamber).A 20 lpm of ambient air was supplied to the flux chamber as sheath flow through a Teflon tubing, and about 10 lpm of air samples was taken out from the flux chamber through another Teflon tubing for analysis.Air samples from the flux chambers were either filled into Tedlar bags for offline analysis or fed to gas analyzers for real-time measurements.Figure 1 also shows other equipment and devices, but these are not relevant to the current discussion.

Analytical Equipment
Two FTIR spectrometers were employed in the present study.One instrument was a Bruker Tensor 27 (Bruker Optics, Inc., Germany) with a spectral resolution of 1 cm −1 , which covered 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 16 liters volume was coupled to the Bruker Tensor FTIR spectrometer.Gas samples in the Tedlar bag from the laboratory bucket silo were analyzed using this FTIR spectrometer.The second instrument was a Bruker IFS 125HR FTIR spectrometer (Bruker Optics, Inc., Germany) with a high-resolution greater than 0.001 cm −1 and broad spectral region from ultraviolet (UV) to far infrared (far-IR).In

LINEFIT Program
An instrumental line shape retrieval program, called LINEFIT, developed by ing Network [21]) for the analysis of reference gas cell spectra recorded with high-resolution spectrometers [27] [28], but has also been used for the analysis of spectra observed with low-resolution spectrometers [29].

Low Resolution FTIR Spectra and Verification of CO in the Bucket Silo Gas
Recording of the infrared spectra from silo gas using the Bruker Tensor 27 FTIR spectrometer was described in detail in the previous paper for N 2 O detection in silo gas [13].The only exception is that no standard CO spectrum was recorded because CO was not in the original project plan and therefore not noticed when the air samples were analyzed by FTIR spectrometer.Due to the well-known physical/optical properties of CO gas, its infrared absorption spectra can be accurately calculated and identified without comparing to its pure gas spectra.One important note from our previous paper is that the air samples from the bucket silos were diluted by a factor of 3.2 in the White Cell when their FTIR spectra were recorded.program.The simulation procedure was described in detail in Zhao et al. [13] and thus will not be repeated here.Because the gas concentrations in the bucket silo gas were unknown when the simulation spectra were generated, to better identify the individual absorptions of the five gases in the silo gas spectra, their concentrations were scaled so that their absorptions close to the intensities of the  Because of less interference of N 2 O absorption in the FTIR spectrum of the Ag-Bag air samples, CO absorptions lines can still be seen beyond 2180 cm −1 , as shown in Figure 3, which is different from the FTIR spectrum in Figure 2.

High Resolution FTIR Spectra and Verification of CO in the Ag-Bag Silage Gas
Therefore, a spectral region from 2090 to 2210 cm −1 was selected for CO retrieval from high-resolution FTIR spectra.

Retrieval of Low Resolution FTIR Spectra of the Bucket Silo Gas
The CO concentration in the bucket silo gas was retrieved from the spectral the estimated uncertainty in retrieving N 2 O concentrations was less than 2%.
The uncertainty of CO retrieval from FTIR spectra is estimated to be in the order of 3% based on CO line intensities and instrument line shape.
To our knowledge, CO concentrations in silage gas have never been reported in the literature.Our results show that CO concentration was as high as 48.0 ppm in the bucket silo gas.Considering the OSHA standards level prohibiting worker exposure to CO more than 50 ppm in the air averaged during an 8-hour

Retrieval of High Resolution FTIR Spectra of the Ag-Bag Silage Gas
Two bags of air samples were collected from the Ag-Bag silage through a flux chamber and analyzed using a high resolution FTIR spectrometer.CO concentrations were retrieved from the FTIR at a spectral window between 2090 -2210 cm −1 .The spectral fitting results are shown in

Real-Time CO Concentrations from Ag-Bag Silage
Figure 6 shows the real-time raw CO data from Ag-Bag silage samples measured by a high spectral resolution LGR N 2 O/CO analyzer.The blue dash line indicates sampling locations.There were eight sampling locations in the sampling system for different experiments and four of them were used for sampling from Ag-Bag silage.Because of the pre-configurations of data collection, the sampling locations in Figure 6 were not selected consecutively.Air samples were drawn by a Teflon coated diaphragm pump at 10 lpm from locations #7 (Flux Chamber inlet, ambient air) and #3 (Flux Chamber outlet, silage samples) and then fed to various gas analyzers for real-time measurements.Only raw CO data measured by the LGR N 2 O/CO analyzer are presented here.The red line shows CO measurements and the green line covers the period when the two Tedlar bags were filled for FTIR analysis.Although data from locations #1 and #2 are shown in Figure 6 for data continuity, they are not discussed here because they were not relevant to this study.It can be seen that CO concentrations were low in the flux chamber inlets (#7, ambient air) and elevated in the flux chamber outlet (#3, silage air samples) by about 3 -5 ppm. Figure 6 presents raw CO data of the LGR analyzer to confirm that high CO concentrations were also observed with an independent monitoring device.There are some CO spikes when sampling locations were switched.
Because the switch was achieved using a mechanical rotary valve, it took a longer time to switch to or away from #7, which might have caused a pressure fluctuation in the air samples and influenced the LGR analyzer.CO data from the LGR analyzer are not available when the two Tedlar bags were filled, as shown in Figure 6.

Discussion
California is a leading dairy state in the nation.In 2019, about 18.5% of US dairy cows were farmed, and 18.6% of US milk/cream was produced in California [14] [15] [30], ranking California the #1 state in the US.The number of dairy cows in the SJV was about 87.3% of the state total in 2019 [14].Because silage is a major feed for dairy cows, California and the SJV dairy farms might be an additional contributor to the state's CO budget due to the CO emissions from silage.Considering that CO is an important O 3 precursor, CO emissions from silage would also have contributed to the production of O 3 in California, particularly in the SJV where high ground-level O 3 concentrations are observed.However, CO emitted from silage has not been included in the O 3 budget yet because this CO emission source had not been previously reported.CO is also a gas hazardous to human health.Due to the continuous emission from silage during various ensiling phases, CO is present and potentially at concentrations that might be healthrelevant.
Because no information was available on CO emission from silage, the importance of its contribution to the CO budget has been unclear.To understand its importance, approaches similar to estimating the N 2 O contribution in Zhao et al. [13] were applied here to estimate the CO contribution.Although this work cannot provide enough data to quantitatively determine the CO emissions from silage, we may be able to roughly estimate it based on our limited observational data and several assumptions.The volume of the bucket silos was 18.9 liters and corn forage compressed in the bucket silos was 12 kg.Assuming all CO gas emitted from one bucket silo was captured in one 5-liter Tedlar bag, in which the CO concentration was measured at about 48 ppm, the CO mass emission from one bucket silo was about 300 μg or the mass emission rate was about 25 μg per kg corn forage.This assumption could have underestimated the CO emission rate from the bucket silos because: 1) the silo gas that remained in the bucket silo container in the initial ensiling phase was not included to determine the CO emissions, and 2) CO was continuously emitted for several months after the initial ensiling.
We took two approaches to estimate the impact of these CO emissions.The first approach was based on silage production and CO mass emission rates from silage.In 2019, about 9.83 million tons of corn silage were produced in the SJV and 10.7 million in California [31].The CO emission from the corn silages would be about 0.25 tons per year in the SJV and 0.27 tons per year in California in 2019.The second approach was based on the silage consumptions by milk cows.
The calculated yearly need for corn silages by milking cows was 10.9 tons per cow per year [32].The number of milk cows in 2020 was reported 1.51 million heads in the SJV and 1.72 million in California [15] [30] [33].Therefore, total CO emissions from the corn silages in 2020 would be 0.41 tons per year in the SJV and 0.47 tons per year in California.Despite large uncertainty in estimating the annual CO emissions from silages based on silage production and consumption, both estimates suggest that CO emissions from the corn silages could range 0.25 -0.41 tons per year in the SJV and 0.27 -0.47 tons per year in California.
The CO Emission Projections by Summary Category in California counties are available online [34].The projected CO emissions by some major categories in the SJV and California in 2020 are summarized in Table 1.The estimated Table 1.CO emission projects by category in the SJV and Calfirona in 2020 [34] and CO emissions from the corn silages estimated in this study.

Conclusions
Elevated CO concentrations above ambient level were detected from two types of corn silages in two ensiling phases using three infrared spectrometers and three different sampling methods.An air sample collected in one Tedlar bag directly from the laboratory bucket silos during the early phase of ensiling was analyzed with a low-resolution FTIR spectrometer.Absorptions of CO were identified in the low-resolution FTIR spectrum at window 2025 -2170 cm −1 .CO concentration of 48.0 ppm in the bucket silo was derived at this spectral window using a LINEFIT program.Air samples were collected with two Tedlar bags through a flux chamber at an Ag-Bag silage pile when it was opened several months after ensiling.These two air samples were analyzed using a high-resolution FTIR spectrometer.Absorptions of CO were identified at the high-resolution spectral window 2090 -2210 cm −1 and CO concentrations of 6.83 ppm were derived.The area emission rate of CO from the Ag-Bag silage pile was estimated at about 29.0 mL/(hour•m 2 ) in volume or 33.7 mg/(hour•m 2 ) in mass.It may be an occupational safety concern if space containing silage, such as a silo, is not well vented because the CO concentration would continuously increase.Air samples collected through a flux chamber at Ag-Bag silage piles were also monitored in real-time by a high-resolution LGR analyzer and CO concentrations between 3 -5 ppm were recorded (Figure 6).Despite the differences between offline FTIR analysis and real-time LGR measurements, both results confirmed that CO concentrations were elevated in the emissions of the Ag-Bag silage piles.
Because N 2 O emissions from laboratory bucket silo were observed in the initial phase of ensiling [13], several attempts were made to detect N 2 O in the Ag-Bag silage piles, including a traditional N 2 O gas analyzer based on broadband infrared absorptions, a high spectral resolution LGR N 2 O/CO gas analyzer, and FTIR spectrometers.Air samples were collected with Tedlar bags from the Ag-Bag silage through a flux chamber and analyzed with a high-resolution FTIR spectrometer, but no N 2 O above the ambient level was observed.Surprisingly, noticeable CO absorptions were identified in the FTIR spectra of the air samples from the Ag-Bag silage.A high-resolution spectral window at 2090 -2210 cm −1 suitable for CO retrieval was selected; see Figure 5, to derive the CO concentrations in the air samples from the Ag-Bag silage.Then, the low-resolution FTIR spectrum recorded earlier for bucket silo gas [13] was re-examined, and a lowresolution spectral window suitable for CO retrieval at 2025 -2170 cm −1 was selected; see Figure 4, to derive the CO concentrations in the bucket silo gas.Reviewing the LGR data also revealed that CO concentrations were elevated in the emissions of Ag-Bag silage.This is how the high CO concentrations were discovered in air samples from both bucket silo and Ag-Bag silages.As mentioned earlier, CO measurements in the silage emissions came from two independent experiments and three spectrometers, provided more evidence supporting the observations of high CO concentrations in the silage emissions during different ensiling phases.
this case, a 40-cm long absorption cell was used to analyze air samples.Air samples from two Tedlar bags from the Ag-Bag samples were analyzed using this FTIR spectrometer.A high spectral resolution LGR N 2 O/CO gas analyzer based on Laser spectroscopy (Los Gatos Research, San Jose, CA) was used to simultaneously monitor CO and N 2 O concentrations.LGR analyzers utilize a unique laser absorption technology called Off-Axis Integrated Cavity Output Spectroscopy (OA-ICOS) with high precision (0.1 ppb in 1 second) and a wide dynamic range (0 -10 ppm).Air samples collected from the Ag-Bag silage through the flux chamber were analyzed in real time by this LGR analyzer.

Figure 1 .
Figure 1.Flux chamber covering the surface of an Ag-Bag silage face for air sampling.

Figure 2
Figure 2 shows an infrared spectrum of the bucket silo gas (red) in the 2000 -2250 cm −1 spectral region at 1 cm −1 spectral resolution, together with the simulated

FTIR
spectrum in the following absorption regions: 1) H 2 O between 2000 -2030 cm −1 , 2) CO 2 between 2030 -2080 cm −1 , 3) CO between 2080 -2150 cm −1 , and 4) N 2 O between 2150 -2250 cm −1 .It should be mentioned that the final gas concentrations were accurately retrieved by the LINFIT program as described later, not during the simulation.Due to the overlapping absorptions of CO, N 2 O and CO 2 above 2150 cm −1 and by H 2 O, CO 2 , and CO below 2080 cm −1 , these spectral regions are not optimal for the quantification of CO.By contrast, CO lines are prominent features in the silo gas spectra between 2080 -2150 cm −1 , clearly indicating the presence of CO in the silo gas.

Figure 3 Figure 3 . 4 . 4 . 1 .
Figure 3 shows an infrared spectrum of the air sample from Ag-Bag silage (red) in the 2090 -2210 cm −1 spectral region with 0.001 cm −1 resolution, together with the simulated infrared spectra of 3 gases whose absorption fall in this spectral region: 1) H 2 O, 2) N 2 O, and 3) CO generated by the LINEFIT program.In order to compare these spectra in the figure, the silage spectrum was downward shifted by 0.02.Because the absorption lines of individual gases are well separated in the figures in black.Four narrow subregions indicated by green dotted lines with higher residuals are de-weighted in the spectral fitting to minimize the propagation of these residuals onto the CO results.In Figure4, the CO column in the White Cell retrieved from the 2025 -2175 cm −1 window was 2.35 × 10 22 m −2 , equivalent to 15.0 ppm CO in the White Cell.Because the CO concentration in the Tedlar bag was diluted by a factor of 3.2 when the samples were introduced into the White Cell, the CO concentrations in the bucket silos gas are estimated to be 48.0 ppm.According to the N 2 O retrieval from silo gas in Zhao et al.[13],

Figure 4 .
Figure 4. Retrieval of the bucket silo air samples from a low-resolution spectrum between 2025 -2175 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.The de-weighing spectral regions are shown in green dotted lines on both top and bottom panels.

Figure 5 .Figure 5 .
Figure 5. Retrieval of the Ag-Bag silage samples from high resolution FTIR spectrum at window between 2090 -2210 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 .
Figure 6.Real-time measurements of CO from Ag-Bag silage by LGR N 2 O/CO analyzer.The blue dash line indicates the sampling locations, the red line shows raw CO concentrations, and the green line covers the period when the two Tedlar bags were filled-up for FTIR analysis.
CO emissions from corn silage in this paper were much lower than the projected CO emissions from well-known categories in both the SJV and California in 2020, but comparable to the emissions in a sub-category of food and agricultural sources in the SJV (0.24 -0.41 tons per year from corn silage vs. 1.10 tons per year from food and agricultural sources).Because this paper is the first report of CO emissions from silage and has very likely underestimated the CO emissions due to limited observational data, it is necessary to improve the esti- a. T/D = Tons per day; b.T/Y = Tons per year.annual