Isocyanic acid in the atmosphere and its possible link to smoke-related health effects

Abstract

We measured isocyanic acid (HNCO) in laboratory biomass fires at levels up to 600 parts per billion by volume (ppbv), demonstrating that it has a significant source from pyrolysis/combustion of biomass. We also measured HNCO at mixing ratios up to 200 pptv (parts-per-trillion by volume) in ambient air in urban Los Angeles, CA, and in Boulder, CO, during the recent 2010 Fourmile Canyon fire. Further, our measurements of aqueous solubility show that HNCO is highly soluble, as it dissociates at physiological pH. Exposure levels > 1 ppbv provide a direct source of isocyanic acid and cyanate ion (NCO^-) to humans at levels that have recognized health effects: atherosclerosis, cataracts, and rheumatoid arthritis, through the mechanism of protein carbamylation. In addition to the wildland fire and urban sources, we observed HNCO in tobacco smoke, HNCO has been reported from the low-temperature combustion of coal, and as a by-product of urea-selective catalytic reduction (SCR) systems that are being phased-in to control on-road diesel NO_x emissions in the United States and the European Union. Given the current levels of exposure in populations that burn biomass or use tobacco, the expected growth in biomass burning emissions with warmer, drier regional climates, and planned increase in diesel SCR controls, it is imperative that we understand the extent and effects of this HNCO exposure.

Every day billions of people are exposed to smoke: from tobacco, from biomass, or low-temperature coal combustion used for cooking and heating, and from wildfires (1, 2). Extensive research on the pyrolysis of biomaterials has shown that numerous volatile and semivolatile organic compounds are produced (3–5). It is important to understand the impacts of these emissions from the global scale down to the personal level. Significant human health effects have been associated with smoke exposure, including cataracts, cardiovascular impairment, and chronic diseases such as rheumatoid arthritis (2, 6). A number of studies have shown that protein carbamylation and associated inflammatory response is a common biochemical pathway causing these effects (7–9). In vivo isocyanic acid (HNCO, H─N═C═O) and its aqueous anion, cyanate (NCO^-), have been identified as biochemical intermediates in this protein carbamylation (8). In this work we show that smoke from biomass, including tobacco, contains HNCO at concentrations that cause carbamylation at physiologically significant levels. Thus smoke-related HNCO exposure is strongly linked to several classes of major negative health effects.

Isocyanic acid has been known since Liebig and Wöhler (10), but it has not previously been measured in the atmosphere. The compound is moderately acidic (pK_a = 3.7) and unstable in pure form as it readily polymerizes (11). However, it is volatile (BP = 23.5 °C estimated) and relatively stable at dilute (several ppmv) concentrations in the gas-phase (12). Isocyanates are toxic at high concentrations, as was demonstrated after the accidental release of methyl isocyanate, CH₃NCO, in Bhopal, India, when thousands of people suffered injury and death (13). The work-place exposure to isocyanates is of concern and has been linked to a number of health effects. Consequently, quite low limits for occupational exposure, 0.5 ppbv for methyl isocyanate, and 5 ppbv for total isocyanates have been established in some jurisdictions (14, 15).

We have recently developed a negative-ion proton-transfer chemical ionization mass spectrometer (NI-PT-CIMS) for sensitive [5 pptv (parts-per-trillion by volume) detection limit] and fast response (1 sec) measurement of HNCO and other acids in air (12) (see Materials and Methods below). The NI-PT-CIMS was used to measure HNCO in the emissions from laboratory biomass fires (4, 12, 16), in the Los Angeles (LA) urban area during May and June, 2010 (17), and during a period when Boulder, CO was impacted by emissions from the 2010 Fourmile Canyon wildfire. In addition we measured the Henry’s Law solubility of HNCO at pH = 3, and using the expression for solubility of a weak acid, estimate HNCO to be highly soluble at physiologic conditions, pH = 7.4. Finally we show that the HNCO levels observed in our combustion/pyrolysis measurements, combined with its solubility and reported in vitro biochemical studies, imply that HNCO makes a significant contribution to smoke-related health impacts that are a major societal concern.

Results and Discussion

Sources.

An example emission-time profile from measurements made at the US Forest Service Fire Sciences Laboratory in Missoula, MT (hereafter referred to as the Firelab), Fig. 1A, shows that HNCO and CO are highly correlated during flaming combustion with an HNCO/CO ratio between 0.1% and 0.6%. Smoldering combustion usually produced a second peak of CO emissions, with values of HNCO/CO that were factors of 5–10 lower. The two emission regimes are shown roughly as colored regions in Fig. 1B along with data from LA and the Fourmile Canyon wildfire. HNCO had an ambient “background” concentration, on the order of 10 pptv or less; and ranged up to 100 pptv at the LA site, and up to 200 pptv during the Boulder measurements. The laboratory biomass burning smoke was measured close to the source hence had much higher levels of HNCO and CO with up to 600 ppbv of HNCO. Laboratory biomass burning ratios of HNCO to other well known CN-containing biomass burning marker species ranged from 1 to 1.6 for HNCO/acetonitrile (CH₃CN), and from 0.33 to 0.5 for HNCO/hydrogen cyanide (HCN) with one fuel as low as 0.16 (4).

Fig. 1.

(A) Time line for HNCO and CO emissions from a laboratory burn of California sage brush. (B) Measured HNCO vs. CO for the Fire Lab emissions measurements (red circles), CalNex LA ground site (black diamonds), and the Boulder Fourmile Canyon fire (blue triangles), with the general flaming stage and smoldering stage relationships highlighted.

Other laboratory studies have reported HNCO in concentrated emissions from pyrolysis and low-temperature combustion of biomass (18) and coal (19), and pyrolysis of tobacco ingredients (20). HNCO has also been observed as a by-product of urea-selective catalytic reduction systems that are being instituted to control NO_x (nitric oxide and nitrogen dioxide) emissions from diesel engines (21). A number of studies show that biomass combustion and pyrolysis produce HNCO from amide or polyamide functionalities (18, 22, 23) {-HN-C(O)-}. Reduced nitrogen coproducts include HCN, CH₃CN, and NH₃, and the oxidized nitrogen coproducts include N₂O, NO, HNO₂, NO₂, and HNO₃. The temperatures involved in natural convection combustion and pyrolysis are low enough that NO_x is not formed from N₂ and O₂, so the emitted nitrogen comes solely from the fuel (24).

Coal is a common fuel used directly for cooking and heating especially in rural areas of developing countries (25). Low-temperature combustion of coal having sufficient nitrogen, has been shown to be a source of HNCO (19). In this case the precursors are likely to be nitrogen heterocyclic compounds, although specific precursors have not been identified. Further work on the combustion of coal char at 600 °C measured the HNCO emission as 12 ± 4.5% of original fuel nitrogen (26).

Tobacco smoking presents an obvious source because, in this case, HNCO is produced from both the plant material, mostly from proteins (polyamides) (18) as well as from pyrolysis of urea, an additive in some cigarettes:

[1]

Apparently there are no reports of HNCO measurements in actual tobacco smoke, but a surrogate pyrolysis study found that 93% of added urea (4 mg/g tobacco) was pyrolyzed to HNCO, resulting in an estimated emission rate of 1.9 mg/g tobacco smoked (20). The amount of HNCO delivered to the average smoker, from this source alone, can be estimated using the parameters given in Baker and Bishop (20). If the lower ranges of values are assumed for: cigarette size (0.7 g), fraction of tobacco burned in puffing (0.3), fraction of HNCO transferred to mainstream smoke (0.3), and the fraction transmitted through the filter (0.3), then a smoker consumes 36 μg (filtered) to 108 μg (unfiltered) of HNCO per cigarette. The mixing ratio of HNCO in mainstream cigarette smoke can be estimated from these factors and an average volume of mainstream smoke of 470 mL (27), to be 40 to 140 ppmv. We observed HNCO in cigarette smoke in a brief laboratory test, however the levels were too high for us to quantify with the CIMS instrument configured in the ambient measurement mode. A more detailed study of tobacco smoke was beyond the scope of the current work. We conclude that tobacco-derived HNCO needs to be measured more extensively and potential exposure to it quantified. This measurement is especially important because HNCO is not currently included in the FDA “Proposed Initial List of Harmful/Potentially Harmful Constituents in Tobacco Products, including Tobacco Smoke” (28).

Diesel urea-selective catalytic reduction (SCR) exhaust systems represent a source of emerging interest, because HNCO is a recognized intermediate in this chemistry. These systems work by injecting a small flow (1–3% by volume urea/fuel ratios) of urea solution (32% by weight) into a catalyst system. Conditions and materials are optimized to induce not just reaction [1], but also provide for the complete catalytic hydrolysis of HNCO:

[2]

While extensive measurements of HNCO emissions from actual working diesel systems are lacking, there are reports of HNCO emissions from model systems that show that up to 5–10% of injected urea N is measured in exhaust as HNCO, mostly at lower temperatures and with older catalysts (> 1,000 h operation). The observed HNCO mixing ratios in exhaust streams ranged up to 50 ppmv (21). Understanding this source must be given a high priority considering the expected growth in SCR-controlled diesel engines in the European Union (EU) and United States.

Ambient Measurements.

The impact of HNCO emitted by a wildfire on the ambient air of an urban area can be seen in the close correlation of the HNCO and CO levels we measured in Boulder, Colorado during the recent 2010 Fourmile Canyon fire (Fig. 2A). In this case there is essentially minute-by-minute correlation in the levels of the two species. The ratio of HNCO/CO during the three enhancement events is consistent with smoldering in the laboratory biomass fires (Fig. 1B). Concurrent measurements of CH₃CN made by gas chromatography/mass spectrometry (GC/MS) (see Materials and Methods below), are also shown in Fig. 2A, and also show close correlation with HNCO and CO, given that the GC/MS sampling time was the first 5 min of each period. The HNCO/CH₃CN ratios in these measurements were quite a bit lower than those observed in the laboratory burns described above. This lower ratio implies either very different source ratios for the fuels involved in this fire, or a much shorter lifetime for HNCO.

Fig. 2.

(A) Mixing ratios of HNCO, CO, and acetonitrile from measurements made in Boulder, CO. during the 2010 Fourmile Canyon fire, with HNCO 30-s averages shown in green circles, CO 1-min averages shown by the red line, and CH₃CN shown as blue bars representing the 5-min sample time. (B) shows the individual 20-s HNCO measurements (blue dots), and the 5-min averaged HNCO (blue line) and ± 1σ (light blue) vs. time of day for the measurements at the CalNex LA ground site.

There are no other reports of measurements of HNCO in the ambient atmosphere with which to compare, but we can contrast our HNCO observations with those of other reduced nitrogen species known to have strong biomass burning sources: CH₃CN and HCN (29). The lowest HNCO values we observe were below 10 pptv, much lower than corresponding background levels of HCN and CH₃CN, which are in the range of 50 to 100 pptv. Because biomass burning is a far larger source of trace gases than diesel engines or tobacco smoke (5), it is almost certainly the largest global source of HNCO, thus the disparity in background values implies a shorter lifetime for HNCO relative to the other species.

A clear diurnal variation in HNCO was observed in the LA basin (Fig. 2B) at the Pasadena ground site during the CalNex 2010, a study with combined air quality and climate goals (30). Wildfires were not observed during this period and are therefore not responsible for the observed HNCO. As the diurnal variation of HNCO coincides with those of other photochemical products, such as ozone and oxygenated VOCs, which peaked midday when processed air from the downtown LA area arrived at the site, we suggest that HNCO in LA is also formed photo-chemically, potentially from monomethyl amine, formamide, and acetamide (31, 32) a source that has not been previously considered in urban areas. Moreover, a surface or vehicle source of HNCO would show low midday values when the boundary layer mixing is maximized and higher values during rush hours, particularly in the morning. Our LA measurements give a baseline urban value, in the absence of local wildfires or a possible future increase from vehicle sources.

Atmospheric Removal Processes.

Hydroxyl radicals are generally responsible for removing trace organic species from the atmosphere (33). HNCO is relatively stable against reaction with OH radical [k ≈ 10^-15 cm³/molec-s extrapolated from high temperature data (34)], yielding an atmospheric lifetime of several decades for this process. It is possible that there are mechanisms, such as adduct formation, that are important at low temperatures, but we are not aware of any studies of that phenomenon. HNCO is expected to have a very low absorption at near-UV to visible wavelengths that constitute the solar actinic region based on measurements at wavelengths shorter than 280 nm (35, 36). The major dissociation channel (HNCO⇒H + NCO) has a threshold of 261 nm which corresponds to a bond energy D_o(H-NCO) = 109.6 ± 0.4 kcal/mole (37), and a channel that forms triplet NH has a 332 nm threshold (38), however the absorption cross-section in that region is quite low. These absorption features result in an HNCO lifetime against photolysis of months.

The major loss processes of HNCO in the lower troposphere are likely the heterogeneous uptake to aerosols or liquid water (fog, clouds, precipitation, and the ocean) and subsequent reactions. HNCO is a moderately weak acid in aqueous solution (pK_a = 3.7) (11) and exhibits relatively slow hydrolysis that is pH-dependent (39). The partitioning of HNCO to aqueous solution at low-concentration, i.e., Henry’s Law solubility, H (M/atm), has apparently not been measured previously. We have measured H for isocyanic acid in an aqueous buffer at pH = 3.0 ± 0.1, and room temperature (t = 25 ± 1 °C). The decrease in gas-phase HNCO, exiting a saturated liquid sample, was measured for a range of volume flow-rate to liquid volumes (Fig. 3A). For a system in which the mass transfer between the liquid and gas phases is rapid, the following relationship holds (40):

[3]

Where C_t/C_o is the relative concentration in the gas phase exiting the reactor, φ is the volumetric flow rate, V is the liquid volume, R is the ideal gas constant, T is temperature, k_l is the first order loss rate in solution, and t is the time. The slopes of these exponential curves vs. the flow-rate/liq.-volume yielded the Henry’s Coefficient (Fig. 3B). The resulting value is given in Table 1 and is commensurate with Henry’s Coefficients of some related species (41). HNCO is only slightly soluble at pH = 3, however, it is a weak acid, pK_a = 3.7 ± 0.2, hence its effective Henry’s Constant can be calculated as a function of pH with the following relationship (42):

[4]

where H* is the intrinsic Henry’s Coefficient independent of any liquid-phase equilibria. This relationship is plotted in Fig. 4 for HNCO using H* derived from the H_eff measured at pH = 3, and the known pK_a.

Fig. 3.

(A) shows the concentration decay curves data for some of the individual equilibration experiments at the flow rates shown, and (B) shows the resulting fit to Eq. 3 and the points for the individual experiments from (A) along with points from additional experiments (solid dots), which are not shown in (A) for the sake of clarity.

Table 1.

Henry’s Law constants of HNCO and related compounds

Compound	H*, M/atm	Notes
HNCO	21	based on Heff at pH3
HONO*	49	for the undissociated acid, pKa 3.3
HCN*	12	for the undissociated acid, pKa 9.2
CH3CN*	50

^*From ref. 41.

Fig. 4.

The Henry’s Law constant (blue) from our measurement at pH = 3 and the weak acid equilibrium relationship Eq. 4, first-order loss rate due to hydrolysis from the rate constants measured by Jensen (39) (solid red), and corresponding aqueous phase lifetime (dashed red) of HNCO vs. pH. The open red circle is our measurement of the first-order loss rate at pH = 3 (± 1σ) and the open blue square is our measurement of H_eff at pH = 3. Also shown is the Henry’s Law constant for HCN (green). The yellow band indicates the range of pHs most characteristic of ambient aerosol, and the pink band indicates physiological pH, and the error bar at pH = 7.4, is the estimated uncertainty based on the uncertainties in H_eff measured at pH = 3, (± 3 M/atm) and, pK_a (± 0.2 pH units).

The rate of hydrolysis of HNCO has been measured as a function of pH and found to have several mechanisms, one set that is direct (i.e., first order), and one that is acid-catalyzed (39):

[5a]

and

[5b]

[6]

This hydrolysis rate is also shown in Fig. 4 as a function of pH. A point of comparison is available from the solubility experiment described above, in which 8.7 ± 1.3 × 10^-4 sec^-1 was measured, a value that is in good agreement with the previous measurement (39). HNCO hydrolyzes slowly enough (t ≈ 20 min) at physiological pHs that there is ample time for carbamylation chemistry, but just fast enough that tradition analytical methods, e.g. aqueous acid/base extractions etc., will underestimate its concentration. While detailed kinetic parameters are lacking, there are a number of studies that show that carbamylation is rapid relative to hydrolysis (8, 43, 44).

Given the above solubility data and the absence of fast gas-phase loss processes, uptake of HNCO on aerosol and cloud droplets, or natural surfaces, will likely be the relevant loss process. The rate of this process can be estimated from knowledge of the Henry’s Law solubility and liquid-phase reaction rate. The loss of a reactive species to a liquid surface of an aerosol particle or droplet can be thought of as a network of resistances, in series and in parallel, that represent diffusion and reaction processes (45). Often one process is limiting, simplifying the representation. The time scale for gas-phase diffusion to particles is on the order of seconds to a few minutes depending on particle size and number. The time scales for loss of HNCO in liquid aerosol or cloud droplets are much longer and can be estimated by the following equation (46);

[7]

where F_l is the volume fraction of liquid water, which ranges from 10^-12 for an aerosol of total surface area of 200 μm²/cm³, and 0.2 μm mean diameter, and up to 10^-6 for fogs or clouds, and k_hyd is k^I. The estimated atmospheric lifetimes for HNCO then range from > 10⁴ years for reaction on an atmospheric aerosol at pH 3, to approximately 0.5 d for reaction in cloud or fog water of pH 5.5 and the above liquid water content. Clearly, the lifetime of HNCO against uptake on aerosols and clouds is limited by liquid-phase reactions.

Slightly different considerations govern the uptake of HNCO on ground or natural water surfaces (lakes, oceans). In these cases it can be assumed that the solubility of HNCO is the dominant factor. Because those surfaces are essentially at neutral pH or above, HNCO should behave similarly to other highly soluble species, such as nitric acid (HNO₃). Nitric acid has lifetimes against deposition in the planetary boundary layer in the range of an hour to half a day (33). A comparison of the above lifetimes for HNCO can be made to the other biomass-derived CN compounds, HCN, and CH₃CN, which are not very soluble at environmental pHs, and have lifetimes of 5 mo (HCN), and 6.6 mo (CH₃CN) against loss, mainly due to deposition to the ocean (29).

Potential Health Impacts.

The chronic and acute health effects of smoke are well documented (2, 6, 8, 47), however detailed causal biochemical pathways are not completely understood and are the subject of current research. The eyes, respiratory, and cardiovascular systems are the areas of the human body that show chronic effects from smoke exposure (2). The potential for health impacts due to HNCO in smoke can be traced to two features of its chemistry: high solubility at physiologic pH as noted above, and the reaction of HNCO with amine, hydroxyl, and sulfhydryl groups, by addition across the N-C bond, to form a carbamyl group, -H₂NC(O)-, in a process termed carbamylation (43):

[8]

[9]

[10]

Recent work has shown that carbamylation of proteins is a key step in the inflammatory response that links smoking to cardiovascular disease (8) and rheumatoid arthritis (6), and the link between carbamylation and cataracts has been recognized for some time (7). Wang et al. (8) identified isocyanic acid/cyanate ion as a key intermediate in this reaction. However, the source of cyanate in this chemistry was postulated to be the enzymatic oxidation of thiocyanate (NCS^-) by hydrogen peroxide. Our work shows that smoke provides a route to uptake and absorption of HNCO into the blood stream that will drive protein carbamylation directly. The effective Henry’s Law solubility of 10⁵ M/atm at pH 7.4 means that a 1 ppbv (10^-9 atm) mixing ratio in inhaled breath will produce an equilibrium aqueous concentration of 100 μM, a concentration that mimics carbamylation in vitro (8). The transport of HNCO into the blood stream and into the sensitive tissues of the eye depends on membrane transport, the precise aspects depending on its pK_a and oil/water partition coefficient (48), which is apparently not known for HNCO. However, we would expect HNCO to have similar behavior to formic acid, because of its similar size, polarity, and acidity (pK_a = 3.75), which has moderate permeability in lipid bilayer membranes (49). As a result, transport of HNCO should be rapid enough that solution concentrations will be close to those calculated from Henry’s Law equilibrium.

Conclusions.

Our work shows that there is potential for significant exposure of humans to HNCO as a consequence of biomass burning, biofuel usage, cooking, and tobacco usage, and perhaps by the use of new diesel SCR emission control systems. Studies of indoor CO in rural areas of China, where biomass or coal is burned in open indoor fires for cooking and heating, report average concentrations in the range 4–12 ppmv (25). If our data from Fig. 1 are typical of biomass cooking fuels, then HNCO levels up to 10 ppbv or higher could exist in those homes and result in blood NCO^- levels far in excess of those shown to produce protein carbamylation. The tobacco source of HNCO has apparently not been quantified directly, but pyrolysis of urea, (a major tobacco additive) produces HNCO directly (20), implying that this source could be considerable. HNCO mixing ratios in the 1 ppbv range were not measured in urban areas, but the possibility that new diesel SCR sources could increase ambient HNCO is a real concern. Wildfire impacts on populated areas and firefighters could also be significant based on our Firelab measurements (Fig. 1B).

These results suggest several issues that require further study. Accurate HNCO emission factors are needed for the different biofuels and coals used for cooking, and the fuels that burn in wildfires and tropical deforestation. Emission ratios for diesel SCR vehicles need to be determined. Human exposure to HNCO needs to be studied in depth, including lung and membrane transport, and their relation to blood and tissue levels of NCO^-. Increased biomass burning emissions are expected with warmer, drier regional climates (50), and diesel SCR controls are being phased-in by the EU and some states in the United States. Extensive source characterization and targeted toxicological and exposure studies are needed to better understand and mitigate this potentially harmful HNCO exposure.

Materials and Methods

Acid CIMS Instrument.

Our HNCO method is described by Veres et al. (51) and Roberts et al. (12) and involves [1] selective ionization of HNCO by reaction with acetate ion via proton transfer in the gas phase and [2] detection of the NCO^- ion at 42 amu by a quadrupole mass spectrometer. Contributions to mass 42 from other compounds such as HN₃ and other H C N O isomers, were considered and rejected for reasons discussed by Roberts et al. (12). The possibility that mass 42 is produced by electron transfer from can be ruled out by the fact that had at most a 0.1% abundance relative to acetate ions in our Firelab and ambient studies, so even if the reaction occurred at the gas kinetic limit, sensitivity to propene would be about 0.3%, because acetate-acid reactions occur at about 1/3 the gas kinetic limit (16). The detection limit of this measurement is 5 pptv for a 1 sec measurement, and the uncertainties are ± (25%+5 pptv) for ambient measurements. Biomass burning emissions measurements were conducted at the US Department of Agriculture, Fire Science Laboratory in Missoula, MT (4). The LA measurements were made at the CalNex-LA Pasadena ground site (34.140582 N, 118.122455 W) (30), and the Boulder measurements were made right outside the NOAA/ESRL laboratory (39.991431 N 105.261032 W). The inlet time constant was a few seconds in the case of LA and Firelab measurements and was estimated to be 10 to 15 s during the Boulder Fourmile Canyon measurements.

HNCO Calibration.

Calibration of the HNCO signal was accomplished by FTIR measurement of a gas-phase diffusion source of HNCO. A gas stream of HNCO at mixing ratios in the range of 1–2 ppmv in 50 sccm (standard cubic centimeters per minute) zero air was produced by the thermal decomposition (210°–230 °C) of cyanuric acid, the trimer of HNCO. Limiting the HNCO mixing ratio to a few ppmv was found to be essential to avoid polymerization as the gas stream cooled to room temperature down stream of the source. The concentration of HNCO in the diffusion source was measured by FTIR in a room pressure (620 Torr) multipass cell (4.8 m length). The standard HNCO absorption cross-section from the database described by Sharpe et al. (52) and Johnson et al. (53) was used to calculate mixing ratios. The HNCO source was diluted in several stages by larger flows of zero air or nitrogen to the ranges appropriate for the Firelab study and the LA study.

Acetonitrile Measurement.

The gas chromatographic/mass spectrometric method used for the measurement of acetonitrile in Boulder during the Fourmile Canyon fire, was described by Gilman et al. (54). The instrument is a custom built two channel GC coupled to an Agilent 5973 quadrupole mass spectrometer. Air was sampled from the west side of the laboratory building at 7 L/ min through a short (several meter) Teflon PFA sample tube. A small subflow of 70 sccm was sampled off of this flow for 5 min through a series of traps designed to reduce water, carbon dioxide, and O₃ interferences (55). Acetonitrile was separated on a semipolar (Restek-MXT-624) capillary column [20 m × 0.18 mmID (inner diameter)] and detected at m/z 41. The GC separation required 25 min, permitting a 5 min sample to be analyzed every 30 min. Calibrations were accomplished by dynamic dilution of gravimetrically prepared gas-phase standards.

Henry’s Constant Measurement.

Henry’s Law describes the equilibration between a gas-phase chemical species and a liquid-phase, at infinite dilution. At equilibrium, the partial pressure of a species in the gas phase is proportional to its concentration in the liquid phase:

[11]

where the Henry’s Law constant, H is typically given in units of M/atm. Measurement of H is relatively simple if the compound of interest is stable in both phases. However, HNCO hydrolyzes at appreciable rates in a pH-dependent way. Therefore a dynamic method was used in this work, following the description by Kames and Schurath (40), and utilizing the apparatus described by Roberts (56), and the HNCO sources described by Roberts et al. (12). The apparatus consisted of two fritted bubblers (Aldrich 250 mL Gas Reducing Flask) placed in series: the first bubbler contained de-ionized (DI) water to humidify the gas stream, and the second contained the sample of DI water buffered to pH = 3 ± 0.1 with a citric acid/NaOH/NaCl buffer (Fluka Chemicals), through which a flow of zero air, in the flow range 100 to 1,000 sccm, was directed. The HNCO source was connected to the gas stream in between the two bubblers by means of a 3-way valve, so that the source could be directed into the bubbler stream, or to vent, without perturbing the main flow. The outlet of the bubbler stream was teed into the inlet of the NI-PT-CIMS so that the pressure at the outlet remained at room pressure. Equilibration experiments were conducted by placing a sample of pH = 3.0 solution (25 ± 0.25 mL) into the bubbler and measuring the HNCO at the outlet at a series of flow rates, as the HNCO source was switched in-line and equilibrated with the solution, and then switched out of line and observed to decay due to a combination of hydrolysis and loss to the gas phase. The exponential decays were then fit to the relationship given in Eq. 3, as shown in Fig. 3. Measurement of H at physiologic pH is not possible with this method as it is too high to yield decay curves on a meaningful time scale, rather we rely on the well demonstrated relationship in Eq. 4 to calculate H at pH = 7.4.