Article pubs.acs.org/est
Formaldehyde Content of Atmospheric Aerosol Kei Toda,*,† Satoru Yunoki,† Akira Yanaga,† Masaki Takeuchi,‡ Shin-Ichi Ohira,† and Purnendu K. Dasgupta*,§ †
Department of Chemistry, Kumamoto University, 2-39-1 Kurokami, Kumamoto 860-8555, Japan Institute of Health Bioscience, The University of Tokushima, 1-78-1 Shomachi, Tokushima 770-8505, Japan § Department of Chemistry and Biochemistry, University of Texas at Arlington, Arlington, Texas 76019-0065, United States ‡
ABSTRACT: Formaldehyde (HCHO) is a highly soluble polar molecule with a large sticking coefficient and thus likely exists in both gaseous and particulate forms. Few studies, however, address particulate HCHO (HCHO(p)). Some report that HCHO(p) concentrations (obtained only with long duration sampling) are very low. The lack of data partly reflects the difficulty of specifically measuring HCHO(p). Long duration filter sampling may not produce meaningful results for a variety of reasons. In this work, gaseous HCHO (HCHO(g)) and (HCHO(p)) were, respectively, collected with a parallel plate wet denuder (PPWD) followed by a mist chamber/hydrophilic filter particle collector (PC). The PPWD quantitatively removed HCHO(g) and the PC then collected the transmitted aerosol. The collected HCHO from either device was alternately analyzed by Hantzsch reaction-based continuous flow fluorometry. Each gas and particle phase measurement took 5 min each, with a 10 min cycle. The limits of detection were 0.048 and 0.0033 μg m−3, respectively, for HCHO(g) and HCHO(p). The instrument was deployed in three separate campaigns in a forest station in western Japan in March, May, and July of 2013. Based on 1296 data pairs, HCHO(p), was on the average, 5% of the total HCHO. Strong diurnal patterns were observed, with the HCHO(p) fraction peaking in the morning. The relative humidity dependence of the partition strongly suggests that it is driven by the liquid water content of the aerosol phase. However, HCHO(p) was 100× greater than that expected from Henry’s law. We propose that the low water activity in the highly saline droplets lead to HCHO oligomerization.
■
INTRODUCTION Indoor formaldehyde (HCHO) is the most important species implicated in sick building syndrome.1,2 Being carcinogenic and genotoxic, HCHO exposure is of considerable concern.3 HCHO is produced from volatile organic compounds (VOCs) via reaction with oxidants, notably by the OH radical, and even in a nonurban area, HCHO may be produced from the (photo)oxidation of biogenic VOCs, notably isoprene.4−8 Atmospheric HCHO undergoes key photochemical reactions;9,10 importantly, some lead to radical amplification.11 Through a series of reactions, it also amplifies tropospheric ozone.12 Atmospheric HCHO is thus of considerable interest. Formaldehyde differs from common VOCs in its high aqueous solubility; Henry’s law constant, KH,HCHO is very high (5020 M atm−1 at 293 K13). Partly this is because HCHO hydrates to CH2(OH)2:
therefore be expected to exist in the particulate/condensed phase. There are a number of studies on HCHO concentrations in fog or cloudwater15−18 (these citations are not comprehensive). Fogwater or cloudwater represents, nevertheless, relatively large amounts of liquid water per unit air volume, far greater than the LWC of ambient aerosols under nonfog conditions. Overall, data on condensed phase HCHO unrelated to fogwater or cloudwater is scant. Sampling for particulate HCHO (HCHO(p)) with an impregnated filter may result in negative errors from destruction by oxidants and by desorption. Positive artifacts can result from oxidation of trapped VOCs. Klippel and Warneck collected aerosols on glass-fiber filters. They reported rural and urban HCHO(p) values of ∼0.04 and ∼0.065 μg m−3, respectively.19 The observed average HCHO(p)/HCHO(g) ratios were ∼10−3 (urban) and 4 × 10−4 (rural). Using filter collection, de Andrade et al. similarly reported that HCHO(p) (100 μg m−3).20 In contrast, Grutter et al.21 first removed ozone and gaseous carbonyl compounds with a denuder, collected particulate material with a downstream filter and followed with a final denuder to collect any HCHO released from the filter. The sum of the HCHO on the filter and the final denuder was taken to be HCHO(p). In Mexico City, HCHO(p) was 20 ± 6% (in a second study22 at same location they reported 15%) of HCHO(g). These results clearly differ greatly from the first two reports. Unfortunately the 4 h time resolution data could not provide diurnal variations, neither was it known whether these results were unique to highly polluted Mexico City. We know of no other reports on HCHO(p). Here we present a simple analytical system that can measure HCHO(g) and HCHO(p) with 10 min resolution. This consists of a parallel plate wet denuder (PPWD),23 and a mist chamber−hydrophobic filter particle collector (PC)24 and a continuous flow analyzer that measures HCHO collected by the two devices. We discuss data collected in a forested area within a semiurban location.
■
aqueous HCHO; for example, when 0.20 L min−1 zero air was bubbled through 0.77 mM HCHO and diluted immediately with 3.8 L min−1 zero air, the resulting stream had a concentration of 12.3 μg m−3 (10 ppbv). The generation solution (prepared from 1 g L−1 stock standard solution, www. wako-chem.co.jp) was in a thermostated bath (NCB-1200, www.tokyo-cci.or.jp) at 20 °C. Analytical System. Referring to Figure 1, air was drawn through the PPWD and the PC serially by an aspiration pump (BA-106F, www.iwakipumps.jp) with flow maintained by a mass flow controller (MFC, SEC-B40, www.horiba.com) at 3 L min−1. The MFC was protected by a mist trap (MT) and air filter (AF, 9900−05-BK, balstonfilters.com) placed upstream. The sample air flowed upward through the PPWD (internal dimensions 415 × 62 × 3 mm, h × w × d).23 Three separate peristaltic pumps (Gilson Minipuls 3, PP1−3) were used; one each serving the PPWD, the PC and the analysis system. The absorber was introduced to each PPWD plate top at 1.0 mL min−1 by PP1 (0.8 × 4.0 mm tubing; all pump tubing were Pharmed). The liquid spread across the width of each plate and flowed down while collecting soluble gases. The hydration of HCHO is acid catalyzed; hence an acidic absorber was used. The solution was aspirated at the PPWD bottom at a higher flow rate by the same pump (1.6 × 4.8 mm tubing) to ensure all effluent is collected. The PC, similar to that described earlier,24 composed of a 76/93 mm i.d./o.d. methacrylate cylinder with a stainless steel needle for absorber injection, was placed immediately above a methacrylate nozzle to nebulize the liquid with the aspirated air. The absorber was pumped at 0.3 mL min−1 by PP2 (0.5 × 3.7 mm tubing). A PTFE filter
EXPERIMENTAL SECTION
Materials and Methods. For both gaseous and aerosol HCHO the absorber was 10 mM H2SO4.25 The Hantzsch reagent was a mixture of 0.01 M 2,4-pentanedione (PD) (www. nacalai.com), 0.25 M acetic acid and 2 M ammonium acetate.26 The reagent solution was prepared daily by mixing equal volumes of 3× concentrated stock solutions of each reagent. All gas volumes hereinafter refer to 298.15 K and 101 kPa. Gaseous HCHO was generated by bubbling pure air through 6637
dx.doi.org/10.1021/es500590e | Environ. Sci. Technol. 2014, 48, 6636−6643
Environmental Science & Technology
Article
(PF020 ϕ47 mm, www.advantec.co.jp) in the air exit path prevented escape of any liquid. The coalesced mist formed drops on the filter, fell to the bottom and was aspirated therefrom by a second PP2 channel. The PPWD and PC effluent were processed by individual air/liquid separators (ALS-1,2, Figure 1); each gas-free stream was alternately selected (for 5 min each) by a three-way valve, 3SV-1. A similar valve 3SV-2 further selected this stream or a blank (2 min sample, 3 min blank) absorber solution for delivery downstream. The solenoid valves (EXAK-3, www. takasago-fluidics.com) were governed by a miniature sequencer (ZEN-10C3DR-D-V2, www.omron.com). The selected sample/blank stream was mixed with the PD reagent (both flows 0.5 mL min−1, PP-3, 0.5 × 3.7 mm tubing). The mixed stream flowed through a reaction coil (0.75/1.5/2200 mm i.d./o.d./ length PTFE tube) wound on a U-shape heater) maintained at 70 °C by a temperature controller (E5CN-RTC, www.omron. com). The product formed, 3,5-diacetyl-1,4-dihydrolutidine, was monitored with a fluorescence detector (FP-2020, www. jascoinc.com, λex/λem 410/510 nm, gain 1000×). A 0.3 × 400 mm i.d./length PTFE restrictor tube at the detector exit inhibited bubble formation.
Figure 2. Collection efficiency of PPWD. Simulated results are shown as the traces and the experimental data are shown as circles.
essentially transmitted without loss. Lack of particle deposition in PPWDs have been reported before.31 Indeed, particle deposition tends to be less than that calculated and less than that in a dry denuder of the same geometry. Diffusiophoresis from Stephan flow of the water evaporating from the wall greatly inhibits particle deposition.32 The slight but perceptibly less than quantitative collection of HCHO(g) at flow rates >3 SLPM may also be due to this reason. We chose therefore a sampling rate of 3.0 SLPM. At this flow rate, the experimentally observed collection efficiencies for 200 ppbv HCHO(g) were respectively 99.91 ± 0.01% for the PPWD and 97.14 ± 1.8% for the PC (obtained with a pair of serial PCs with no denuder upstream) Collection of HCHO(p) was not specifically examined but a similar PC has been shown to have a collection efficiency of 99.5+% for a broad size range of fluorescein-doped (NH4)2SO4 aerosols up to flow rates of 6 L min−1.24 Analytical System Performance. Figure 1 inset shows illustrative raw data: baseline and respective peaks for HCHO(g) and HCHO(p) were obtained every 10 min. The calibration equation used for quantitation was
■
FIELD ANALYSIS OF GASEOUS AND AEROSOL HCHO The instrument was located at the Forestry and Forest Products Research Institute (Tatsuta Natural Park, Kumamoto, N 32°49′ 23″, E 130° 44′ 00″). Although within a city, no vehicular traffic is allowed in the hilly forested park. Outside air was brought to the PPWD via a 1.25 × 250 cm i.d./length PTFE tube. Three different sampling periods, each lasting ≥72 h, during March 30−April 1, May 22−24, and July 10−12, 2013, were used. Multipoint system calibration was conducted just before leaving for the field for each campaign with gas and liquid standards spanning 0−12.3 μg m−3 and 0−600 nM, respectively. The liquid calibration was repeated at the field site before collecting ambient data. Temperature, relative humidity (RH), other meteorological parameters, and NO, NO2, and ozone were monitored simultaneously. Insolation intensity was monitored outside the forest canopy at a location 1000 m away from the sampling station. Data were stored on a datalogger (GL200, www.graphteccorp.com).
HCHO, μgm−3 = 12.8(85.3)μgm−3V −1 × peakHeight, V (2)
where the parenthetical coefficient pertains to the PC signal as the PC has a different liquid influent flow rate. Respective limits of detection (LODs) based on three times the signal-to-noise ratio were 50 ng m−3 (0.04 ppbv) and 3 ng m−3, for HCHO(g) and HCHO(p). The PPWD uses ∼7× greater influent liquid compared to the PC, the LOD difference is mostly because of this. Interferences. The Hantzsch reaction has been widely used for measuring HCHO. In the liquid phase, ethanal, n-propanal, n-butanal, 2-methylpropanal, 3-methylbutanal, acrolein, glyoxalic acid, acetone, and methanol all have responses at least 104× smaller. Relative to HCHO, even glyoxal (CHO−CHO) responds less than 1%.33 The use of an acidic absorber also inhibits uptake of SO2, which may interfere negatively.25 Importantly, Hantzsch-based analyzers have previously been intercompared with different direct spectroscopic instruments with excellent agreement and there is no evidence of any systematic interference, even in relatively polluted atmospheres.34,35 Direct interaction of ozone with the Hantzsch reagent can produce a positive interfrence;9 in the present system 1000 ppbv O3 in the absence of HCHO produced artifact signals of 3.9 and 0.18 μg m−3 for HCHO(g) and HCHO(p), respectively. Given that the maximum O3 concentration encountered during the field measurements was ∼
