Article pubs.acs.org/est
Airborne Trifluoroacetic Acid and Its Fraction from the Degradation of HFC-134a in Beijing, China Jing Wu,†,‡ Jonathan W. Martin,§ Zihan Zhai,† Keding Lu,† Li Li,† Xuekun Fang,† Hangbiao Jin,§ Jianxin Hu,† and Jianbo Zhang*,† †
State Key Joint Laboratory of Environmental Simulation and Pollution Control, College of Environmental Sciences and Engineering, Peking University, Beijing 100871, People’s Republic of China ‡ China Waterborne Transport Research Institute, Beijing 100088, People’s Republic of China § Department of Laboratory Medicine and Pathology, University of Alberta, Edmonton, Alberta T6G 2B7, Canada S Supporting Information *
ABSTRACT: Trifluoroacetic acid (TFA) has been attracting increasing attention worldwide because of its increased environmental concentrations and high aquatic toxicity. Atmospheric deposition is the major source of aquatic TFA, but only a few studies have reported either air concentrations or deposition fluxes for TFA. This is the first study to report the atmospheric concentrations of TFA in China, where an annular denuder and filter pack collection system were deployed at a highly urbanized site in Beijing. In total, 144 air samples were collected over the course of 1 year (from May 2012 to April 2013) and analyzed directly using high-performance liquid chromatography−tandem mass spectrometry (HPLC-MS/MS) or following derivatization by gas chromatography−mass spectrometry (GC−MS). The annual mean atmospheric concentration of TFA was 1580 ± 558 pg/m3, higher than the previously reported annual mean levels in Germany and Canada. For the first time, it was demonstrated that maximum concentrations of TFA were frequently observed in the afternoon, following a diurnal cycle and suggesting that a major source of airborne TFA is likely degradation of volatile precursors. Using a deposition model, the annual TFA deposition flux was estimated to be 619 ± 264 μg m−2 year−1. Nevertheless, a box model estimated that the TFA deposition flux from the degradation of HFC-134a contributed only 14% (6−33%) to the total TFA deposition flux in Beijing. Source analysis is quite important for future TFA risk predictions; therefore, future research should focus on identifying additional sources.
■
INTRODUCTION Trifluoroacetic acid (CF3COOH, TFA) has recently attracted worldwide scientific attention because of its aquatic toxicity and increasing environmental concentrations.1−5 Surface waters are the predominant environmental sink of TFA, and no environmentally significant chemical or biological loss pathways for TFA have been reported.4,6,7 Thus, after entering the surface water, TFA is environmentally persistent. When the aquatic concentration of TFA exceeds 100 μg/L, TFA may become toxic to some aquatic life forms.6,8 Previous studies have shown that, in the year 2000, atmospheric degradation of hydrochlorofluorocarbons (HCFCs) and hydrofluorocarbons (HFCs) were the main anthropogenic sources of TFA.9 However, presently 1,1,1,2-tetrafluoroethane (CH2FCF3, HFC-134a) appears to contribute most to TFA formation among the substitutes for ozone-depleting substances (ODS).10 Other sources may include atmospheric oxidation of the fluorinated inhalation anesthetics halothane (CF3CHClBr), desflurane (CF3CHFOCHF2), and isoflurane (CF3CHClOCHF2) or thermolysis of perfluorinated polymers.11,12 At present, the © 2014 American Chemical Society
environmental levels of TFA are still far below the threshold for toxicity;13−19 however, because of the wide usage and emission of its precursors, aquatic concentrations of TFA will continue to increase.20−23 In the future, these increases may lead to aquatic thresholds being exceeded in local surface waters, such as seasonal wetlands.5 This is an important consideration if hydrofluoro-olefins (HFOs), such as 2,3,3,3-tetrafluoropropene (CH2CFCF3, HFO-1234yf) and 1,3,3,3-tetrafluoropropene (CF3CHCHF, HFO-1234ze), are applied as ODS replacements, because their shorter atmospheric lifetimes and higher yields of TFA have the potential to increase aquatic TFA concentrations in source regions.1,3,4,10 Atmospheric deposition of TFA is the main source of increased concentrations in surface waters and marine environments.4,24 However, air concentrations and the associated Received: Revised: Accepted: Published: 3675
November 12, 2013 March 4, 2014 March 5, 2014 March 17, 2014 dx.doi.org/10.1021/es4050264 | Environ. Sci. Technol. 2014, 48, 3675−3681
Environmental Science & Technology
Article
deposition fluxes of TFA have seldom been reported. The gasparticle distribution of TFA is a major factor determining its deposition, but only one study has estimated TFA deposition flux based on the direct measurements of gas- and particlephase TFA.25 Global three-dimensional models have also been used to estimate the annual deposition fluxes of TFA.20,23 In the current study, airborne TFA was measured in China for the first time and two modeling assessments were performed. First, a deposition model was applied to estimate the annual TFA deposition flux based on the atmospheric concentrations of TFA in gas and particle phases over an entire year. Second, a box model was used to estimate annual TFA deposition flux exclusively from the atmospheric degradation of HFC-134a, using regional observational data for HFC-134a (see details in the Supporting Information). The deposited TFA fraction from the degradation of HFC-134a was then determined by combining the two estimates to aid in environmental risk projection and the possible mitigation of TFA in China.
Before sampling, denuders were prerinsed with doubledistilled water, acetone, and hexane. The internal walls of the denuders were coated with a water/methanol solution (50:50, v/v) containing 1% Na2CO3 and 1% glycerol. After drying under a flow of high-purity nitrogen, the denuders were capped before use. Quartz filters were baked at 450 °C for 6 h and wrapped in aluminum foil. From May 2012 to April 2013, a total of 114 samples (each representing a total air intake of 48 m3 collected over a period of 48 h) were collected (5−13 samples per month) at a rate of 16.7 L min−1 (mass flow), to evaluate monthly variation in airborne TFA. Moreover, to evaluate diurnal variation in airborne TFA for the first time, a total of 30 air samples (each representing a total air intake of 4 m3 collected over 4 h) were continuously collected at the same rate from January 9 to 13, 2013. Field blank filters and coated denuders were always placed at the sampling site and then returned to the lab with each set of environmental samples for blank analysis. Sample Preparation and Analysis. For each gas-phase sample, the two denuders were extracted with three consecutive additions of double-distilled water (10, 10, and 5 mL) by transferring each addition from the first denuder to the second, with shaking in each denuder, and a combined extract (25 mL) was obtained. For each particle sample, the 47 mm ringed filters were weighed before and after sampling to determine particle mass (PM2.5). Particles were then entirely extracted using two consecutive additions of double-distilled water (each 10 mL) and subjected to ultrasonication for 30 min, using an additional 5 mL of water for rinsing. The combined particle extract (25 mL) was centrifuged prior to analysis. A field blank was analyzed along with every set of samples. Two analytical methods, gas chromatography−mass spectrometry (GC−MS) and liquid chromatography−tandem mass spectrometry (LC−MS/MS) as developed by Scott et al.18,28 and Taniyasu et al.,29 respectively, were applied to detect airborne TFA. We conducted sample preparation and analysis using GC−MS, as previously described by Hu et al.30 Briefly, TFA and the internal standard [perfluoropropionic acid (PFPA)] were derivatized to the respective acid anilide in the presence of 2,4-DFAn and DCC. Samples were cleaned and reduced to 1 mL for analysis using GC−MS (QP-2010 SE, Shimadzu, Kyoto, Japan) with ions of 225 and 275 amu for TFA and PFPA, respectively. GC separation was performed using a DB-5 ms column (30 m × 0.32 mm × 0.25 μm) with helium as a carrier gas. The initial oven temperature was 50 °C for 2 min, after which it was increased at a rate of 30 °C/min to a maximum temperature of 215 °C and then maintained at that temperature for another 10 min. The injector, transfer line, ion source, and detector temperatures were maintained at 200, 250, 200, and 250 °C, respectively. Using LC−MS/MS, extracts were directly injected into LC (UFLC XR, Shimadzu, Kyoto, Japan) interfaced to a 4000Q TRAP (MDS Sciex, Concord, Ontario, Canada) operated in the negative electrospray ionization mode. LC separation was performed using a Rspak JJ-50 2D column (2.0 mm × 150 mm × 5 μm, Shodex, Showa Denko K.K., Kawasaki, Japan) at a flow rate of 200 μL/min via an isocratic elution. For mobile phase A, we used 20% 50 mM ammonium acetate in water, and for mobile phase B, we used 80% methanol and 20% water. The MS/MS transition used to quantify TFA was m/z 112.9/68.9. Quality Assurance (QA)/Quality Control (QC). The method detection limits (MDLs) for TFA were defined as the mean level of all field blanks plus 3 times the associated
■
MATERIALS AND METHODS Chemicals and Reagents. Chemical standards of TFA (99%) and perfluoropropionic acid (PFPA, 99%) were obtained from Acros Organics (Geel, Belgium). 2,4-Difluoroaniline (2,4-DFAn, 99%) was supplied by J&K Chemical (Greensboro, GA). N,N′-Dicyclohexylcarbodiimide (DCC, >99%) was obtained from Fluka Chemical (Milwaukee, WI). Methanol [high-performance liquid chromatography (HPLC) grade)] and HPLC-grade water were purchased from Fisher Scientific (Ottawa, Ontario, Canada). Ammonium acetate (HPLC grade, 99%) was supplied by Sigma-Aldrich (Oakville, Ontario, Canada). Organic residue analysis grades of ethyl acetate, methanol, toluene, dichloromethane, acetone, and hexane (95% n-hexane) were obtained from J. T. Baker (Phillipsburg, NJ). Optima-grade concentrated hydrochloric acid (36−38%) and reagent-grade glycerol (99.9%) were purchased from Beijing Chemical Works (Beijing, China). Optima-grade anhydrous sodium sulfate (99.5%) was supplied by Tianjin Jinke and baked overnight at 600 °C (Tianjin, China). Reagent-grade sodium bicarbonate (>99.5%) and sodium chloride (>99.5%) were obtained from Xilong Chemical (Beijing, China), immersed in methanol and ethyl acetate for 10 min in sequence, dried overnight, and baked at 80 and 450 °C for 24 h, respectively. Silica gel (60, 0.063−0.100 mm) was purchased from Merck (Rahway, NJ), eluted by dichloromethane, dried overnight, and baked overnight at 550 °C. All glassware was solvent-washed with acetone and n-hexane before use. Air Sampling. To collect TFA in the gas and particle phases separately, an annular denuder/filter pack collection system developed by Martin et al. was deployed on top of a six-story building of the Peking University (PKU) campus in Beijing (40.00° N, 116.31° E; see Figure S1 of the Supporting Information). This site is located in the northwest of Beijing, approximately 10 km from the city center, representing a typical urban environment in Beijing.26,27 The entire apparatus was vertically stationed in a sampling box consisting of a cyclone inlet (2.5 μm cut, URG-2000-30EH, URG, Chapel Hill, NC), two alkalized glass annular denuders (URG-2000-30x2423CSS, URG, Chapel Hill, NC), a one-stage filter pack (URG2000-30FG-2, URG, Chapel Hill, NC) containing a 47 mm ringed quartz fiber filter (Whatman), and a sampling pump equipped with a mass flow sensor (URG-3000-02BA, URG, Chapel Hill, NC), all in series. 3676
dx.doi.org/10.1021/es4050264 | Environ. Sci. Technol. 2014, 48, 3675−3681
Environmental Science & Technology
Article
Hypothesis 1: Atmospheric Concentrations of HFC-134a near the Ground within the Fifth Ring Road Are Higher than Those Outside the Fifth Ring Road. The regional observational results of HFC-134a (see details in the Supporting Information) revealed that the regional mean level of HFC134a observed within the Fifth Ring Road was significantly higher than that outside the Fifth Ring Road (p < 0.01). The atmospheric lifetime of HFC-134a is up to 13.4 years; therefore, we can assume that the airborne HFC-134 within and outside the Fifth Ring Road maintained a dynamic balance and that the emission rate of HFC-134a nearly equaled the net flow rate of HFC-134a out of the box. Hypothesis 2: Within the Fifth Ring Road, the Atmospheric Concentration of HFC-134a above the Annual Mean Boundary Layer Linearly Decreases to the Background Level of Beijing with Increasing Altitude. The spatial distribution of HFC-134a mixing ratios suggested significant HFC-134a emissions within the Fifth Ring Road (see details in the Supporting Information). Therefore, HFC-134a levels at sites near the surface were relatively high, while HFC-134a levels at high altitudes were not affected by HFC-134a emission and approximated the background level in Beijing. We assumed that the atmospheric concentration of HFC-134a above the annual mean boundary layer decreased linearly with increasing altitude to reach the background level in Beijing. The vertical profile of HFC-134a taken by aircraft measurements over the eastern U.S. supports this hypothesis.38 Hypothesis 3: TFA Is in the Quasi-steady State. The atmospheric lifetime of TFA is approximately 9 days,20 which is much shorter than the atmospheric lifetime of HFC-134a (13.4 years) and shorter than the time scale of our estimation (1 year). Therefore, we assume that TFA is in a quasi-steady state, as expressed by the following equation:
standard deviation. For GC analysis, the MDLs were 128 and 30 pg/m3 for 48 h gas and particle samples, respectively. For HPLC−tandem mass spectrometry (MS/MS) analysis, the MDLs were lower, at 50 and 2.6 pg/m3 for 48 h gas and particle samples, respectively. For the 4 h gas-phase samples, which were only analyzed by HPLC−MS/MS, the MDL was 105 pg/m3. Recovery tests were performed 3 times in the field to determine the gaseous recovery efficiency of TFA from alkalized denuders. The mean recovery was 101 ± 3.1%, indicating excellent collection efficiency. To examine the comparability of the two methods, three unknown spiked solutions were prepared and analyzed using both methods. The methods differed by 2−10%, indicating a good agreement.
■
MODEL CALCULATION Deposition Model. The Atkinson deposition model31 was applied to estimate the monthly dry and wet deposition fluxes of TFA. Table 1 lists observed monthly mean concentrations of Table 1. Monthly Meteorological Data at the Weather Station (39.48° N, 116.28° E) in Beijing from May 2012 to April 2013a
a
month
monthly mean pressure (0.1 kPa)
monthly mean temperature (K)
monthly mean relative humidity (%)
monthly total precipitation (0.1 mm)
5/2012 6/2012 7/2012 8/2012 9/2012 10/2012 11/2012 12/2012 1/2013 2/2013 3/2013 4/2013
10048 9993 9986 10045 10107 10144 10171 10243 10241 10226 10146 10102
296 298 300 299 294 288 277 269 268 272 279 286
43 60 71 71 60 54 51 49 61 51 45 39
313 1039 2840 599 812 214 811 76 30 34 107 55
d[TFA] = P − D − (Fout − Fin) dt
where d[TFA]/dt is the change with time in the abundance of TFA inside the box, P is the chemical production rate of TFA, D is the deposition rate of TFA, and Fout and Fin are the flow rates of TFA out of the box and into the box, respectively. Hypothesis 4: Deposition Is the Main Loss Process for Airborne TFA. The spatial distributions of HFC-134a mixing ratios suggested uniform HFC-134a emissions outside the Fifth Ring Road of Beijing and around Beijing (see details in the Supporting Information). Therefore, the TFA generated from HFC-134a should also be evenly distributed in the atmosphere. The outflow or inflow will not significantly change the atmospheric level of TFA in the box. Therefore, the equation in hypothesis 1 could be converted into the equation below.
Data source: China Meteorological Data Sharing Service System.
TFA, monthly mean temperature, and monthly total precipitation. The total dry deposition velocity rate was assumed to be 0.5 cm s−1 based on the value used by Martin et al. Our estimates of deposition are therefore more conservative than other modeling results, which often adopt a higher dry deposition velocity for nitric acid (∼2 m/s) in models of dry TFA deposition.2,3,20,23,32,33 For particle-phase TFA, the wet washout ratio is generally in the range from 105 to 106;34 therefore, we adopted a midpoint value of 5 × 105. For gaseous TFA, deposition is dominated by TFA partitioning between the air and aqueous phases; therefore, the wet washout ratio is a function of KH and temperature.35 Note that KH is not constant but is dominated by the pKa of TFA and the temperature.35,36 Box Model. On the basis of regional atmospheric observational results for HFC-134a (see details in the Supporting Information), an Eulerian box model was set up to estimate annual TFA deposition from the degradation of HFC-134a (DTFA). Beijing was assumed as a simple box: the box bottom was the ground; the surface area was 16 801 km2; and the box top was the annual mean tropopause in Beijing, 16.6 km.37 Four hypotheses were proposed for this box model, as listed below.
P=D
In other words, we assume that the deposition is the main process leading to TFA loss, and thus, the annual deposition of TFA from HFC-134a is nearly equal to the annual production of TFA generated by HFC-134a. On the basis of these assumptions, DTFA can be acquired by calculating the annual TFA production generated by the atmospheric oxidation of HFC-134a.39 On a regional scale, the difficulty with this method is how to estimate the annual mean atmospheric burden of HFC-134 in Beijing (BHFC‑134a). 3677
dx.doi.org/10.1021/es4050264 | Environ. Sci. Technol. 2014, 48, 3675−3681
Environmental Science & Technology
Article
The model was set up as shown below:
By analyzing sequential and short-term air samples (4 h), we were able to determine the diurnal variation in airborne TFA concentrations (see Figure 1). To our knowledge, this is the
estimation for TFA deposition from the degradation of HFC134a (DTFA) DTFA = PTFA =
BHFC‐134a YHFC‐134aM TFA τHFC‐134aMHFC‐134a
(1)
where DTFA is TFA deposition from the degradation of HFC134a in g/year, PTFA is the TFA production generated by the degradation of HFC-134a in g/year, BHFC‑134a is the annual mean atmospheric burden of HFC-134 in Beijing in g, τHFC‑134a is the atmospheric lifetime of HFC-134a in years, YHFC‑134a is the molar yield of TFA from HFC-134a, and MTFA and MHFC‑134a are the molar masses of TFA and HFC-134a, in 114 and 102 g/mol, respectively. estimation of the atmospheric burden of HFC-134a (BHFC‑134a) BHFC‐134a = NC backgroundMHFC‐134a
(αSinter + Souter) S
(2) Figure 1. Gas-phase concentrations of TFA observed at the Peking University site from January 9 to 13, 2013. Different legends indicate different sampling dates. For example, 0109 indicates January 9.
where BHFC‑134a is the atmospheric burden of HFC-134a in Beijing in g, N is the total atmospheric amount above Beijing in moles, Cbackground is the regional background concentration of HFC-134a in pptv, S, Sinter, and Souter are the total area and the area within and outside the Fifth Ring Road in Beijing in km2, and α is the revised factor for the atmospheric burden of HFC134 within the Fifth Ring Road.
first time that such temporal resolution has been achieved. The results show that the atmospheric concentrations of TFA peaked (1190 ± 426 pg/m3) in the afternoon between 12:00 and 16:00 and were at a minimum (420 ± 263 pg/m3) in the morning between 4:00 and 8:00. Figure 2 presents the seasonal
estimation of the total amount of air (N)
N=
S∑ ρi Hi Mair
(3)
where N is the total atmospheric amount above Beijing in moles, ρi is the atmospheric density at various altitudes, Hi is altitude in m, Mair is the molar mass of air, 29 g/mol, and i is the layer of atmosphere, where the ground is the first layer. estimation of the revised factor for the atmospheric burden of HFC-134 within the Fifth Ring Road in Beijing (α) α=
∑ [(Ninter)i (C inter)i ] NinterC background
(4)
where (Ninter)i is the atmospheric total amount in layer i within the Fifth Ring Road in Beijing and (Cinter)i is the regional mean concentration of HFC-134a in layer i within the Fifth Ring Road in Beijing.
■
Figure 2. Monthly mean and standard deviation for atmospheric concentrations of TFA observed at Peking University site from May 2012 to April 2013. Dark error bars represent particle-phase TFA, and thin error bars represent gas-phase TFA.
RESULTS AND DISCUSSION Airborne TFA. TFA was detected in all atmospheric samples. The annual mean concentration of TFA was 1580 ± 558 pg/m3 (mean ± standard deviation), much higher than previously reported annual mean levels in Feiburg, Germany (44 pg/m3) in 1995,11 and Guelph, Canada (760 pg/m3) in 2000.25 The highest value observed in Beijing was 5583 pg/m3, which is slightly higher than the reported maximum value in previous studies (e.g., 5200 pg/m3 in Reno, NV in 1994),16 but average peak levels during the summer in Beijing were similar to those reported in Toronto, Canada in 2000.25 More TFA was present in the gas phase (1330 ± 530 pg/m3) than in the particle phase (245 ± 116 pg/m3) for all months (Figure 2). The annual mean TFA fraction of particulates (ϕ) was 17 ± 11%, lower than reported in Guelph, Canada (29%).25
variation in TFA, with maximum levels observed in spring and summer. Together, these data are consistent and suggest that a major source of TFA is from the secondary transformation of precursors driven by photochemically derived oxidants, such as the hydroxyl radical.40,41 Albeit less likely, another possible explanation is that the atmospheric concentrations of volatile precursors increase midday and in the spring and summer.42 Total Deposition Flux of TFA. The results of deposition modeling revealed that the total deposition flux of TFA peaked in July: July accounted for 20% of the total annual flux, followed by June > November > May (17, 16, and 14%, respectively; 3678
dx.doi.org/10.1021/es4050264 | Environ. Sci. Technol. 2014, 48, 3675−3681
Environmental Science & Technology
Article
Table 2. Monthly Mean and Annual Mean Atmospheric Densities at Different Altitudes in Beijing in January, April, July, and October (kg/m3)a
Figure 3. Monthly deposition flux of TFA at the Peking University site from May 2012 to April 2013. These deposition fluxes were simulated by a deposition model based on atmospheric measurements of TFA.
layer
altitude (m)
January
April
July
1 2 3 4 5 6 7 8 9 10 11 12 13
20 40 79 159 321 653 1040 1534 3242 5393 7840 11404 18208
1.26 1.25 1.25 1.24 1.22 1.19 1.14 1.08 0.96 0.77 0.60 0.41 0.20
1.17 1.17 1.16 1.15 1.14 1.11 1.07 1.03 0.92 0.76 0.59 0.42 0.21
1.11 1.10 1.10 1.09 1.08 1.06 1.02 0.98 0.88 0.73 0.57 0.41 0.23
October
annual mean density
standard deviation
1.17 1.17 1.16 1.15 1.14 1.11 1.07 1.02 0.92 0.75 0.59 0.43 0.22
1.18 1.17 1.17 1.16 1.15 1.12 1.08 1.03 0.92 0.75 0.59 0.42 0.22
0.062 0.061 0.061 0.060 0.058 0.054 0.049 0.043 0.031 0.019 0.011 0.006 0.013
a
Monthly mean values were calculated using the hourly mean atmospheric density simulated by MM5 and MCIP. Annual mean values are averages of the monthly mean values.
see Figure 3). Because of the high airborne TFA concentration (≈1960−4150 pg/m3) or high monthly precipitation (e.g., 104 mm in June and 284 mm in July), both dry and wet depositions were relatively high in May, June, and July, resulting in higher modeled total deposition than in other months. For November, the relatively high deposition flux of TFA was mainly attributable to high levels of wet deposition caused by high monthly precipitation and the high wet washout ratio for gaseous TFA. Dry and wet depositions in other months are both relatively low; therefore, deposition in other months was responsible for
