Analysis of Carbonyl Compounds in Ambient Air by a Microreactor Approach.

Aldehydes including formaldehyde, acetaldehyde, and acrolein are toxic organic components of air pollution that cause lung cancer and cardiovascular disease with chronic exposure. The commonly used method for determining the levels of carbonyl compounds based on the derivatizing agent 2,4-dinitrophenylhydrazine is of limited use for ketones and unsaturated aldehydes because of issues such as low capture efficiencies, unstable derivatives, and long sample collection times. This work details the analysis of carbonyls in ambient air by a microreactor approach. The microreactor is fabricated on a silicon wafer and has thousands of micropillars in a microfluidic channel for uniformly distributing the air flow through the channel. The surfaces of the micropillars are coated with a quaternary ammonium aminooxy reagent, 2-(aminooxy)ethyl-N,N,N-trimethylammonium iodide (ATM), for chemoselective capture of carbonyl compounds by means of oximation reactions. ATM-carbonyl adducts are eluted from the microreactor and directly analyzed by Fourier transform ion cyclotron resonance mass spectrometry and ultrahigh-performance liquid chromatography-mass spectrometry. More than 20 carbonyls were detected in ambient air samples. Acetone, 2-butanone, acetaldehyde, and formaldehyde were the most abundant carbonyls in ambient air of the studied urban areas.

Introduction

Carbonyl compounds are common incomplete combustion products of direct emissions from motor vehicles[1−4] and chemical industrial plants and other combustion sources,[5,6] produced by the oxidation of virtually all hydrocarbons, and formed by photochemical reactions of precursor molecules with atmospheric ozone.[7−10] Carbonyls in ambient air in urban and rural areas and in emission gases from vehicles have been frequently monitored and studied.[11−16] Some carbonyls, notably formaldehyde, acetaldehyde, and acrolein, cause cardiovascular disease after chronic exposure.[17] Consequently, the concentrations of these aldehydes in the atmosphere are continuously monitored by the United States Environmental Protection Agency (EPA) under the National Air Toxics Assessment (NATA) program.[18]

The commonly used method for the analysis of carbonyl compounds in air is the EPA method TO-11A, which requires 2,4-dinitrophenylhydrazine (DNPH)-coated solid sorbent cartridges.[19] Carbonyls are collected as DNPH derivatives and the resultant DNPH–carbonyl adducts are subsequently eluted and analyzed by high-performance liquid chromatography (HPLC).[20−24] However, only a few studies have reported the carbonyl capture efficiency of DNPH.[25−27] In recent years, the TO-11A method has been reported to have issues involving low capture efficiencies of ketones and problems related to the formation of dimers and trimers with acrolein.[28−31] Although collection of air samples in a mist chamber containing an aqueous bisulfite solution followed by derivatization using pentafluorobenzylhydroxylamine (PFBHA) is an efficient method for acrolein determination, this approach is not suitable for measuring formaldehyde and acetaldehyde because of the low capture efficiencies of the mist chambers for these aldehydes.[31−33] Solid-phase microextraction (SPME), a popular preconcentration method introduced by Pawliszyn in the early 1990s,[34] is a rapid extraction technique for the analysis of volatile organic compounds from a variety of matrices. SPME, however, is only a semiquantitative method because of competitive adsorption on the SPME fiber. Microfabricated gas preconcentrators and gas chromatography columns have demonstrated the potential for real-time gas detection.[35−37] However, microfabricated preconcentrators have low physical adsorption efficiencies and require thermal desorption for the analyses of captured analytes.

The objective of this work is to analyze carbonyls in environmental air using silicon microreactors to capture trace carbonyls. The microreactors have the advantage of overall small microliter volume and high surface area because of thousands of micropillars coated with 2-(aminooxy)ethyl-N,N,N-trimethylammonium iodide (ATM), a quaternary ammonium aminooxy salt in comparison with DNPH-based packed cartridges.[38−41] ATM has fast reaction kinetics with both aldehydes and ketones via oximation reactions and thus provides high carbonyl capture efficiencies for both classes. Furthermore, a much smaller volume of gaseous sample (∼1 L) is required for the analysis. ATM imparts a permanent positive charge to all adducts and thus ensures soft efficient ionization of the carbonyl analytes. In this work, Fourier transform ion cyclotron resonance mass spectrometry (FT-ICR-MS) was initially used for rapid identification of volatile carbonyl adducts because of its highest mass accuracy and resolution. Ultra HPLC–MS (UHPLC–MS) was subsequently used for the separation and analysis of isomeric ketones and aldehydes. The microreactors were used to capture carbonyls from indoor and outdoor air samples collected during sunny and rainy days in urban Louisville, Kentucky. More than 20 carbonyls were detected, and their concentrations varied with weather and location.

Results and Discussion

Characterizations of the microreactors and measurements of capture efficiencies of trace carbonyls in high-purity helium by the microreactors have been reported.[39] The capture efficiencies of three aldehydes (propanal, butanal, and hexanal) and three ketones (acetone, 2-butanone, and 2-pentanone) all reached above 92% using the microreactors with ATM loaded in the microreactor at an ATM to carbonyl molar ratio larger than 100:1 and at a sample flow rate of 5 mL/min through the microreactor. Increasing the sample flow rate decreases the capture efficiencies of carbonyls. Thus, a fixed air flow rate of 5 mL/min was used in this work. The microreactor was also tested for the capture of acetone in a TO-15 standard with a capture efficiency of 97%.[39] Furthermore, deuterated acetaldehyde, pentanal, acetone, and 2-butanone were spiked into air and exhaled breath samples for the measurement of capture efficiencies, and the results indicated that capture efficiencies were not affected by relative humidity (RH) because of high excess amount of ATM loaded in the microreactor and its fast reaction kinetics.

After the microreactors were characterized for the capture of carbonyl compounds in high-purity helium and synthetic air, we began to use the microreactors for the capture of carbonyls in ambient air. A typical FT-ICR-MS spectrum of an air sample collected outdoors in urban Louisville on a sunny day is shown in Figure S1. More than 20 ATM adducts of carbonyls were detected in both outdoor and indoor air samples. The ultrahigh mass resolution of FT-ICR-MS provides a quick way to identify ATM–carbonyl adducts. Constitutionally isomeric ketones and aldehydes are indistinguishable by direct infusion FT-ICR-MS, but the measured m/z (molecular weight) with the best mass resolution provides accurate molecular formulas. Table shows the calculated mean concentrations of carbonyl compounds in air samples. All concentrations in Table were blank-corrected by subtraction of the mean field blank values of the same carbonyls from the measurements, with resulting negative values set to the nondetected (ND) level. All values were included in the analysis, even if the concentration fell below the method detection limit (MDL). A few other compounds including glyoxal and tolualdehyde were detected in some outdoor samples but were not included in the table. The air samples collected in 2017 were analyzed by UHPLC–MS, and 13 compounds were identified. The concentrations of acetone, propanal, 2-butanone, butanal, pentanone, and pentanal were obtained from the air samples collected in 2017, while the concentrations reported for all other carbonyls were measured mean values obtained in 2012 and 2017. Hydroxyacetaldehyde, hydroxyacetone, and 3-hydroxy-2-butanone were detected. Five formulae (C4H6O, C6H12O to C9H18O) in Table were mixtures of isomeric aldehydes and ketones that were not separated by UHPLC.

Table 1

Average Concentrations and Standard Deviation (n = 3) of Carbonyl Compounds in Ambient Air and MDLs of Carbonyl Compounds

average carbonyl concentrationsa (μg/m3)	gas station (sunny)	roadside (sunny)	roadside (raining)	Chem Lab	MDL (μg/m3)
formaldehyde	6.59 ± 1.08	5.96 ± 0.15	1.72 ± 0.11	7.02 ± 1.10	4.41
acetaldehyde	9.02 ± 0.88	9.29 ± 1.79	4.19 ± 0.52	9.47 ± 1.01	6.77
acrolein	0.07 ± 0.01	0.07 ± 0.01	0.01 ± 0.01	0.09 ± 0.01	0.07
acetone	16.43 ± 1.00	16.26 ± 3.35	8.72 ± 0.82	16.94 ± 1.95	8.61
propanal	0.23 ± 0.03	0.21 ± 0.05	NDb	0.11 ± 0.01	0.005
C4H6O	0.09 ± 0.01	0.12 ± 0.01	ND	0.69 ± 0.05	0.008
2-butanone	9.66 ± 1.50	9.97 ± 2.16	0.97 ± 0.17	8.54 ± 1.47	9.17
butanal	4.17 ± 0.44	4.76 ± 0.55	ND	3.91 ± 0.27	0.10
pentanone	1.46 ± 0.30	1.16 ± 0.18	0.29 ± 0.03	1.67 ± 0.24	1.64
pentanal	0.41 ± 0.19	0.28 ± 0.18	ND	0.47 ± 0.29	0.50
benzaldehyde	1.02 ± 0.08	1.24 ± 0.08	1.29 ± 0.13	0.97 ± 0.03	0.01
hydroxyacetaldehyde	0.77 ± 0.17	0.62 ± 0.02	0.16 ± 0.01	0.75 ± 0.19	0.70
hydroxyacetone	0.09 ± 0.01	0.15 ± 0.03	0.12 ± 0.03	0.15 ± 0.03	0.06
3-hydroxy-2-butanone	0.14 ± 0.02	0.11 ± 0.03	0.13 ± 0.02	0.14 ± 0.02	0.09
C6H12O	0.55 ± 0.03	0.54 ± 0.05	0.14 ± 0.01	0.91 ± 0.05	0.01
C7H14O	0.40 ± 0.04	0.41 ± 0.06	0.25 ± 0.02	1.12 ± 0.01	0.21
C8H16O	0.44 ± 0.03	0.45 ± 0.06	0.36 ± 0.02	1.07 ± 0.01	0.25
C9H18O	0.44 ± 0.05	0.44 ± 0.05	0.51 ± 0.03	1.31 ± 0.07	0.27

All values of carbonyl concentrations were blank-corrected by subtraction of the mean blank values from the measurements.

ND—nondetected level.

The concentrations of carbonyl compounds obtained near a gas station and at a roadside area about 2 miles away from the gas station (shown in Figure S2) during sunny days were very close with only slight differences for some compounds (Table ). The total concentration of all carbonyls near the gas station had a mean level of 51.6 μg/m3. The total concentration of all carbonyls at the roadside area was 52.0 μg/m3 during sunny days and 16.9 μg/m3 during rainy days. Rain washes away carbonyl compounds in air. Table indicates that acetone, 2-butanone acetaldehyde, and formaldehyde were the most abundant carbonyl compounds in air. The total concentration of formaldehyde and acetaldehyde counts for about 30% of all carbonyl compounds in air at the roadside.

The total concentration of all carbonyls in a chemistry laboratory had a mean level of 55.3 μg/m3, slightly higher than the total concentration outdoor during sunny days (Table ). The concentrations of acetone, 2-butanone, formaldehyde, and acetaldehyde in outdoor air are similar to that in the chemistry laboratory. A comparison of the measurements of these carbonyls in outdoor air during sunny days with an average RH of 71% and rainy days with an average RH of 97% and in the chemistry laboratory in Table indicates that the environmental air in the area was polluted by the carbonyls and that the weather affects carbonyl compound levels in environmental air.

Table lists the concentrations of formaldehyde, acetaldehyde, acrolein, acetone, and 2-butanone in environmental air determined from this work and the data reported in the literature. Formaldehyde in this work was similar to the prior reports, but acetaldehyde was even higher than that measured in New York[23] and California.[12] The level of acetone was near the highest level reported in the literature (measured in Guangzhou, China[21]) and the level of 2-butanone was the highest among all the reported data. One likely reason is air pollution from the ever-expanding chemical industry in the city. Another possible reason is that the acetone and 2-butanone levels, according to the literature, were determined using DNPH-based methods, and these had issues of low capture efficiencies.[28] The principal chemical industrial site in Louisville, known as Rubbertown (shown in Figure S2), is in the southwest end of the city. The area has a history of air pollution, and the air quality has been monitored by the EPA. The average concentration of acrolein was considerably lower than most of the published data but was in the range of the data reported by Cahill.[31] The capture efficiency of acrolein by the microreactor was within the 50–60% range under the experimental conditions. Therefore, the acrolein concentration measured by the microreactor approach as reported in Table is roughly half of the actual concentration.

Table 2

Comparison of Concentrations of Five Carbonyl Compounds in Ambient Air in Recent Studies (μg/m3)

locations	study seasons	method	formaldehyde	acetaldehyde	acrolein	acetone	butanone	references
Rio de Janeiro, Brazil	May–Nov 2000	DNPH	10.84	10.43	0.82	4.14	1.42	Grosjean, 2002[20]
New York, US	Summer 1999	DNPH	5.3	4.2	NA	NA	NA	Sax et al., 2004[23]
CA, NJ, and TX, US	1999–2001	DNSH	6.42	5.44	0.46	4.19	NA	Liu et al., 2006[12]
Guangzhou, China	June–Sept 2003	DNPH	13.68	8.33	NA	17.76	5.02	Feng et al., 2005[21]
Austin, TX, US	April–July 2007	DNPH	0.641	2.70	0.48	NA	0.857	Clements et al., 2009[22]
California, US	Sept–Oct 2013	PFBHA	NA	NA	0.17 (0.046–0.41)	NA	NA	Cahill, 2014[31]
Beijing, China	June–Aug 2008	DNSH	29.3	27.1	2.3	NA	NA	Altemose et al., 2015[11]
Louisville, KY, US	July 2012, July 2017	ATM	6.29	9.78	0.069	15.01	9.82	this study

The hourly average concentrations of formaldehyde, acetaldehyde, and acetone measured during daytime in July 2012 are shown in Figure . A slight change in the concentrations, especially the concentrations of formaldehyde and acetaldehyde, were observed at the roadside area over the 1 day sampling period. During the sunny day (Figure a), the concentrations of formaldehyde and acetaldehyde slightly increased from 8:00 to 10:00 because of emission gases from vehicles during the morning traffic peak. The concentrations then decreased from 10:00 to 12:00, followed by a slight increase from 12:00 to 18:00, reaching a maximum at 18:00 once again because of the traffic. Photochemical reactions also contribute to the total formaldehyde and acetaldehyde concentrations observed during the mid-day. The carbonyl concentrations on rainy days were much lower than those observed during sunny days (Figure b). The concentrations of formaldehyde and acetaldehyde in both sunny and rainy days slightly increased from 8:00 to 10:00 because of traffic emission.

Figure 1

Average carbonyl concentrations (n = 3) as a function of time: (a) roadside (sunny days) and (b) roadside (rainy days).

This work demonstrated that the microreactor approach is suitable for the analysis of carbonyls in ambient air. The present approach has several advantages over current methods[11,12,20−24,31−33] for the collection and analysis of carbonyls in air including small air sample size (<1 L), microfluidic solvent volume (∼7 μL) for adduct elution from the microreactor, and the permanent positive charge imparted by ATM for increasing the sensitivity of MS analysis. Deuterated carbonyl compounds can be added into air samples for monitoring the capture efficiencies and measurement accuracy. The approach is convenient for monitoring carbonyls in environmental air at specific times as shown in Figure . High capture efficiencies of both saturated aldehydes and ketones make the microreactor approach suitable for the analysis of these carbonyls (Table ). The limitation of the microreactors is a relatively low sample flow rate (<10 mL/min) in order to achieve high capture efficiencies. Finally, depending on column elution conditions, UHPLC may not cleanly separate isomeric ketones and aldehydes, particularly for carbonyls with more than five carbon atoms. To further improve this microreactor approach, we aim to develop new derivatizing agents for the microreactor to capture unsaturated aldehydes, particularly acrolein, with higher capture efficiencies for accurate analyses of such aldehydes.

Materials and Methods

Materials

All reagents and solvents, including formaldehyde, acetaldehyde, acetone, acrolein, methanol, deuterated acetaldehyde, and acetone (acetone-d6) (99.9%), were purchased from Sigma-Aldrich (Milwaukee, WI). The quaternary ammonium aminooxy compound ATM was synthesized according to a published method.[42] Tedlar bags were purchased from Supelco (Bellefonte, PA). Syringes (0.5 L, clear acrylic, purchased from SGE, Orlando, FL) and 1 L size Tedlar bags were cleaned with high-purity nitrogen and tested free of carbonyl compound contamination. Stainless steel canisters (6 L size) for the collection of air samples were purchased from Restek Corporation (Bellefonte, PA).

Microreactor Fabrication

The microreactors were fabricated using the techniques common for the fabrication of microelectromechanical system devices. The fabrication procedures are standard processes that take about 1 week to complete.[38−40]Figure A shows a microreactor chip having inlet and outlet ports connected to 360 μm O.D., 250 μm I.D. deactivated fused silica tubes. Figure B shows an optical micrograph of the micropillar array within the microreactor, and Figure C shows a scanning electronic microscopy (SEM) micrograph of the micropillars. The surface coating of the micropillars using ATM was done by flowing a known amount (3.65 × 10–7 mol) of ATM–iodide salt in methanol (15 μL) into the microreactor channel through one connection port, followed by evaporation of methanol using a vacuum oven. The slightly negative charge of the silicon dioxide surface of the micropillars enhances association between ATM cations and the surfaces of the micropillars. ATM on the surface of the micropillars reacts with carbonyls via oximation reactions.

Figure 2

(A) Size comparison of the microreactor with a U.S. coin, (B) micrograph of the microreactor, and (C) SEM micrograph of the micropillars.

Air Sample Collection and Concentration

Ambient air samples from indoor areas (university facilities) and outdoor areas (roadside, gas station) in urban Louisville, Kentucky, were collected using a 0.5 L size syringe to fill Tedlar bags. Initially, the measurements of carbonyl compounds in air samples collected in Tedlar bags were compared with measurements of air samples collected in stainless steel canisters. The Tedlar bags were free of carbonyl compound contamination after cleaning three times with high-purity nitrogen, followed by evacuation each time. There was no statistically significant difference between measured concentrations of carbonyls for the samples collected in Tedlar bags versus canisters. Therefore, all air samples were collected in Tedlar bags. After sample collection, a 1 L air sample in a Tedlar bag was connected to the microreactor in the laboratory at an ambient temperature of 20 °C as shown in Figure S3. The air sample was passed through the microreactor at a flow rate of 5 mL/min. The ATM–carbonyl adducts and unreacted ATM were then eluted from the microreactor by flowing methanol (150 μL) from one slightly pressurized vial through the microreactor and into an empty collection vial. The eluted solutions were directly used for either FT-ICR-MS or UHPLC–MS analysis without further processing. To determine the concentrations of the carbonyl compounds, a 5 μL solution containing 1.14 × 10–8 mol of acetone-d6 completely reacted with ATM in methanol was added to each eluted methanol solution as an internal reference (IR) for quantitative analysis by FT-ICR-MS and UHPLC–MS. The amounts of captured carbonyl compounds were determined by using calibration curves of the known concentrations of ATM–carbonyl adduct solutions versus the intensity ratios of the ATM–carbonyl to the IR ATM–acetone-d6 adduct for FT-ICR-MS analyses. The calibration curves of UHPLC–MS were built by the known concentrations of ATM–carbonyl adduct in methanol versus the ratios of the peak area of the ATM–carbonyl to the peak area of the IR ATM-acetone-d6. The samples collected in July 2012 were analyzed only by FT-ICR-MS because no UHPLC–MS was available in-house at the time. Although high-resolution FT-ICR-MS was used for quick identification of carbonyls, other lower resolution mass spectrometers can be used for the analysis. The samples collected in July 2017 were analyzed by UHPLC–MS for the separation of isomeric compounds. ATM–carbonyl adducts stored as methanol solutions in a refrigerator were stable for at least 1 month.

Quality Assurance

The capture efficiencies of ketones and aldehydes by the ATM-coated microreactors have been characterized and published.[38−40] Saturated ketones and aldehydes were captured with efficiencies above 95%. Deuterated acetaldehyde, 2-butanone, and pentanal were also added to air samples for verifying the capture efficiencies of these compounds. The capture efficiencies of aldehydes were slightly higher than those of isomeric ketones because of faster reaction kinetics. Field blanks were defined as the average of measurements of individual carbonyls from three microreactors loaded with ATM without any active flow of air but eluted with methanol. These field blanks were then used as background contamination of carbonyls. The MDLs of ATM–carbonyl adducts were defined as the mean field blanks plus three standard deviations of the field blanks. At least three duplicate air samples were collected and analyzed for each location, and the average concentrations were reported.

FT-ICR-MS and UHPLC–MS Analyses

The eluted methanol solutions containing the ATM–carbonyl adducts were analyzed on a hybrid linear ion trap FT-ICR-MS instrument (Finnigan LTQ-FT, Thermo Electron, Bremen, Germany) equipped with a TriVersa NanoMate ion source (Advion BioSciences, Ithaca, NY) with an electrospray chip (nozzle inner diameter 5.5 μm). The TriVersa NanoMate was operated in a positive ion mode by applying 2.0 kV with no head pressure. The LTQ-FT was tuned and calibrated according to the manufacturer’s default standard recommendations, which achieved better than 1 ppm mass accuracy and a 400 000 resolution capability at m/z of 400. ATM and ATM derivative species were assigned on the basis of their accurate mass by first applying a small (typically <0.0005) linear correction based on the observed mass of the internal standard.

For the separation of isomeric ATM–carbonyl adducts, the eluted solutions were injected into a Vanquish UHPLC system fitted with a Syncronis HILIC column (2.1 mm × 100 mm, 1.7 μm) (Thermo Fisher, USA). Elution was performed using a 95/5% mixture of acetonitrile/100 mM aqueous ammonium formate solution (pH = 3). The column was thermostatized at 30 °C, and the chromatographic flow rate was set to 0.4 mL/min. The eluent from the column was introduced into an electrospray ionization source (3.5 kV). Ions were introduced into a Q Exactive Focus Orbitrap (Thermo Fisher Scientific, CA, USA) mass spectrometer working at a resolution of 35 000 at m/z of 400.