Future Emissions and Atmospheric Fate of HFC-1234yf from Mobile

Article pubs.acs.org/est

Future Emissions and Atmospheric Fate of HFC-1234yf from Mobile Air Conditioners in Europe Stephan Henne,†,* Dudley E. Shallcross,‡ Stefan Reimann,† Ping Xiao,‡ Dominik Brunner,† Simon O’Doherty,‡ and Brigitte Buchmann† †

Laboratory for Air Pollution/Environmental Technology, Empa, Swiss Federal Laboratories for Materials Science and Technology, Ü berlandstrasse 129, 8600 Dübendorf, Switzerland ‡ School of Chemistry, University of Bristol, Cantocks Close, Bristol BS8 1TS, U.K. S Supporting Information *

ABSTRACT: HFC-1234yf (2,3,3,3-tetrafluoropropene) is under discussion for replacing HFC-134a (1,1,1,2-tetrafluoroethane) as a cooling agent in mobile air conditioners (MACs) in the European vehicle fleet. Some HFC-1234yf will be released into the atmosphere, where it is almost completely transformed to the persistent trifluoroacetic acid (TFA). Future emissions of HFC-1234yf after a complete conversion of the European vehicle fleet were assessed. Taking current day leakage rates and predicted vehicle numbers for the year 2020 into account, European total HFC-1234yf emissions from MACs were predicted to range between 11.0 and 19.2 Gg yr−1. Resulting TFA deposition rates and rainwater concentrations over Europe were assessed with two Lagrangian chemistry transport models. Mean European summer-time TFA mixing ratios of about 0.15 ppt (high emission scenario) will surpass previously measured levels in background air in Germany and Switzerland by more than a factor of 10. Mean deposition rates (wet + dry) of TFA were estimated to be 0.65−0.76 kg km−2 yr−1, with a maxium of ∼2.0 kg km−2 yr−1 occurring in Northern Italy. About 30−40% of the European HFC-1234yf emissions were deposited as TFA within Europe, while the remaining fraction was exported toward the Atlantic Ocean, Central Asia, Northern, and Tropical Africa. Largest annual mean TFA concentrations in rainwater were simulated over the Mediterranean and Northern Africa, reaching up to 2500 ng L−1, while maxima over the continent of about 2000 ng L−1 occurred in the Czech Republic and Southern Germany. These highest annual mean concentrations are at least 60 times lower than previously determined to be a safe level for the most sensitive aquatic life-forms. Rainwater concentrations during individual rain events would still be 1 order of magnitude lower than the no effect level. To verify these results future occasional sampling of TFA in the atmospheric environment should be considered. If future HFC-1234yf emissions surpass amounts used here studies of TFA accumulation in endorheic basins and other sensitive areas should be aspired.

■

INTRODUCTION In Europe the F-gas Regulation No 842/2006 and the MAC Directive 2006/40/EC were issued in 2006 and became effective in 2011. These regulations forbid the use of cooling agents with a 100 year global warming potential (GWP) greater than 150 for mobile applications. In essence, this leads to the phase-out of the commonly used HFC-134a (1,1,1,2-tetrafluoroethane) as a cooling agent in MACs. With a GWP of 41 the newly proposed substitute HFC-1234yf (2,3,3,3-tetrafluoropropene) fulfills the regulations. When released to the atmosphere HFC-1234yf undergoes two major destruction pathways: (1) the path initiated by OH radicals with a 100% molar yield of trifluoroacetyl fluoride (TFF) and (2) the path initiated by chlorine radicals (Cl) with a 92% molar yield of TFF.2 Further reaction pathways initiated by reaction with O3 or NO3 are only of minor importance.2,3 Once formed in the atmosphere TFF hydrolyses readily in water forming TFA, which is finally removed from the atmosphere by wet and dry deposition. Several currently used hydrofluorocarbons (HFCs) also form TFA in the atmosphere. © 2012 American Chemical Society

However, compared with HFC-1234yf degradation (∼12 day lifetime with respect to OH), these HFC’s atmospheric lifetimes are larger and TFA molar yields are smaller (e.g., 14.6 years and 7−20% for HFC-134a 4,5), suggesting that TFA deposition from HFC-1234yf will be enhanced and more localized. TFA is mildly phytotoxic but accumulates in the hydrosphere,6,7 since thermal degradation of TFA is thought to be slow, with a lifetime of tens of thousands of years.8 While natural sources of TFA seem to exist in the oceans,6,7,9−11 there is consensus that the major present day contribution to TFA in the atmosphere, precipitation and surface fresh waters is anthropogenic, originating from the atmospheric oxidation of HFCs.12 Despite its persistence, current day levels and Received: Revised: Accepted: Published: 1650

September 30, 2011 January 3, 2012 January 6, 2012 January 6, 2012 dx.doi.org/10.1021/es2034608 | Environ. Sci. Technol. 2012, 46, 1650−1658

Environmental Science & Technology

Article

Table 1. Fraction of Cars, f, and Emission Factors, e in (g yr−1 MAC−1), for Different HFC-1234yf Loss Processes from MACs for the Low and High Emission Scenarioa low high

f rl

erl

f il

eil

ff

ef

f rs

ers

f rn

ern

fes

ees

fen

een

0.9 0.9

17.9 35.8

0.017 0.017

550 550

0.9 0.9

0.42 0.42

0.9 0.81

2.92 2.92

0 0.09

29.2 29.2

0.45 0.225

8.33 8.33

0 0.225

33.3 33.3

a

rl: Regular diffusive losses. il: Irregular losses. f: Initial filling losses. rs: Refilling losses (skilled personnel). rn: Refilling losses (non skilled personnel). es: End of life losses (skilled personnel). en: End of life losses (non skilled personnel).

2020, we choose this approach to give an estimate of the final TFA deposition after the total conversion. The average filling of a MAC was presumed to be 550 g, the same amount used by Papasavva et al.14 Regular losses from MACs are due to leakages, occurring during normal usage through hoses and junctions. Based on test bench measurements it is commonly assumed that MACs have a higher leakage rate in warmer climates.15 This was considered by Papasavva et al.14 to adjust emission factors in different regions in the U.S. For Europe, Schwarz and Harnisch16 estimated that emissions were indeed higher in Portugal compared with Germany. However, emissions in Sweden were as high as in Portugal, possibly due to a more corrosive environment. Considering the encountered variability, we did not include a climate adjustment factor for regular emissions. In our highemission scenario the average regular annual loss rate was 6.5% of the initial filling of 550 g, erl = 35.8 g yr−1 MAC−1. This number relies on emissions derived from in-use cars in 2002/ 200316 and assumes no improvement in loss rates by 2020. For the low-emission scenario it was assumed that leakages will reduce due to technical improvements and will be only 3.25% of 550 g, erl = 17.9 g yr−1 MAC−1. Our lower estimate is only slightly higher than the average yearly loss rate of 13.6 g yr−1 MAC−1 for the US calculated by Papasavva et al.,14 who set their loss rates to two times the one estimated for new MACs tested in laboratory studies.17 In an update to this study Yu and Clodic18 estimated yearly emissions of new MACs in Europe to be around 10 g yr−1 MAC−1. Values were identical after running the MACs for 9 months on the road. Anyhow, it is expected that the average 12-year usage of European cars leads to higher emissions, because vibrations and adverse environmental conditions have cumulative effects on hoses and connections, as supported by the real-world tests by Schwarz and Harnisch.16 Hence, it was assumed that values from Yu and Clodic18 represent minimum leakage rates. Additional evidence for higher real-world emissions comes from Souza19 who tested emissions of MACs which had been artificially aged in the lab. Emissions from regular leaks were between 20 and 56 g yr−1 MAC−1, similar to our rates of 18−36 g yr−1 MAC−1. Irregular losses occur when MACs are completely emptied due to sudden leaks from accidents, stone impact and component defects. Assuming the initial filling of 550 g an emission factor of eil = 550 g yr−1 MAC−1 results. Schwarz and Harnisch16 estimate that 1.9% of MACs suffer complete leakage per year, giving f il = 0.017 (1.9% of MAC multiplied by 90% of cars using MACs) for both scenarios. Losses from the initial filling were assumed to be very low (5 g MAC−1), which has to be divided by the average MAC lifetime of 12 years, resulting in ef = 0.42 g yr−1 MAC−1. With an annual loss rate of 6.5%, MACs have to be refilled at least once in their lifetime. For the low-emission scenario it was assumed that 100% of the MACs will be refilled by skilled personnel with an average loss of 35 g per refill (regular refill), which corresponds to an average yearly loss of ers = 2.92 g yr−1

deposition rates of TFA are not thought to threaten the most sensitive aquatic life forms.4,12,13 However, additional TFA deposition from newly introduced substances like HFC-1234yf will need to be evaluated before their global market launch. While this was done for the U.S.,2,14 no such study was performed for Europe as yet. HFC-1234yf emissions within Europe will be more concentrated, situated further to the North and, considering the specific meteorological conditions, might be dispersed slower compared with the U.S. Thus, the purpose of this study is to assess the impact of a complete transition of European vehicles to HFC-1234yf as a cooling agent in their MACs. We first estimate country wide HFC-1234yf emissions from MACs and then use these as input to two different atmospheric chemistry transport models that predict transport, conversion to TFA, deposition and rainwater concentrations. Possible environmental risks are discussed.

■

MATERIALS AND METHODS Emissions. European HFC-1234yf emissions from MACs in 2020 were estimated using two scenarios with a lower and upper limit for emission factors, ej, of HFC-1234yf for different activities/processes; that is, filling/refilling, usage (regular/ irregular) and disposal. These scenarios should be understood as the 95% confidence band that we put on our estimate. The part of Europe considered in this study is the 27 European Union countries plus Croatia, Norway, Switzerland and Turkey, hereafter called EU-27+. Multiplying emission factors by the fraction f j of cars involved in each activity and by the predicted number Ni of passenger cars per country in 2020 yields country total emissions Ei of HFC-1234yf,

E i = Ni ∑ e j f j j

(1)

Passenger car numbers as predicted for the year 2020 were taken from the TREMOVE 2.7b database (available online at http://www.tremove.org/). Emissions from additional mobile applications (e.g., trucks) were presumed to account for an additional 10% of the emissions. Emission factors and passenger car fractions for the different processes are described below and summarized in Table 1. The following assumptions were made: In 2020 all countries of the EU-27+ share the same legislation. Therefore, no intercountry differences exist in the quality of filling and refilling of MACs. However, for the disposal (end-of-life losses) it was assumed that differences still exist in handling the remaining cooling agent. Using the age-distribution of cars within the TREMOVE database, the average lifetime of a passenger car in 2020 was calculated to be 12 years. It was presumed that 90% of the passenger cars will be equipped with a MAC in 2020. The changeover from HFC-134a to HFC1234yf was assumed to be completed. Although a considerable percentage of cars might still be equipped with HFC-134a in 1651

dx.doi.org/10.1021/es2034608 | Environ. Sci. Technol. 2012, 46, 1650−1658


Article

MAC−1. For the high-emission scenario it was assumed that only 90% of the MACs will be refilled by skilled personnel. For the remaining 10% of the MACs it was supposed that the residual HFC-1234yf will be completely vented before refilling. Taking an average content of 350 g when the refilling is due, this corresponds to an average yearly loss of ern = 29.2 g yr−1 MAC−1. Considering that 50% of decommissioned EU27+ passenger cars will be exported (as today), that is, mostly to Russia and West Africa, and only the remainder will be dismantled within EU-27+ itself, only end-of-life emission from the latter were taken into account. It was assumed that before disposal an average of 400 g MAC−1 would still be present in the system. For the low-emission scenario it was assumed that HFC-1234yf will be recycled from 100% of the cars, which are dismantled in the EU-27+. Also with the best available practice about 100 g of cooling agent remains in the MACs16 and will finally be vented to the atmosphere. This results in yearly emissions of ees = 8.33 g yr−1 MAC−1 considering a 12-year lifetime. For the highemission scenario it was assumed that only 50% of the cars undergo regular recycling and for the other 50% the remaining HFC-1234yf (400 g) is completely vented to the atmosphere, giving een = 33.3 g yr−1 MAC−1. For use by the transport models, country total emissions were disaggregated onto a regular grid. The nondriving mode emissions (assumed to 70%) were distributed relative to population densities for the reference year 2000.20 Emissions during driving (remaining 30%) were set proportional to EDGAR4.0 on-road CO emissions (traffic category 1A3b_c_e) for the reference year 2005.21 FLEXPART Simulations. The Lagrangian particle dispersion model FLEXPART (Version 8.1) 22,23 was used to simulate atmospheric transport and degradation of European HFC-1234yf and deposition of TFA. FLEXPART was driven by three-dimensional meteorological fields obtained from the European Centre for Medium-Range Weather Forecasts (ECMWF) Integrated Forecasting System (IFS) with a resolution of 1° ×1° on 91 levels. Fields were available every 3 h (alternating between analysis fields at 00, 06, 12, and 18 UTC and 3 h forecast fields at 03, 09, 15, and 21 UTC). FLEXPART was extended to incorporate more highly resolved OH fields from a multimodel ensemble climatology24 and prescribed Cl fields. Furthermore, the model’s description of incloud scavenging was improved by including cloudwater content from ECMWF IFS instead diagnosing it from relative humidity and precipitation rates. HFC-1234yf reactions with OH and Cl radicals were incorporated in FLEXPART using the temperature-dependent rate constant kOH = 1.26 × 10−12 exp(−35/T) cm3 molec−1 s−125 and kCl = 7.03 × 10−11 cm3 molec−1 s−1.3 Cl concentrations in the marine boundary layer (MBL) were set to an average concentration of 1.8 × 104 molecules cm−3 with a latitudinal and seasonal variability as suggested by Allan et al.26,27 In the continental planetary boundary layer (PBL) and the free troposphere, Cl concentrations were set 1 order of magnitude smaller following the same seasonal cycle. Since hydrolysis of TFF in cloudwater is fast (hydrolysis rate of 150 s−128) compared with the FLEXPART time step (600 s), hydrolysis was considered as an instantaneous complete conversion of TFF to TFA as soon as cloudwater was present. Dry and wet (in cloud and below cloud scavenging) deposition follow the approaches of refs 29 and 30, respectively. TFA dry deposition was assumed to be identical to that of nitric acid.2

Simulations were carried out with the meteorology of the year 2010. The model was allowed to spin up for 1 month (December 2009) and was integrated throughout 2010. Model particles were released proportionally to HFC-1234yf emissions on a 0.1° × 0.1° grid and were followed for 40 days. After this time, any HFC-1234yf, TFF and TFA remaining on terminated particles was considered to contribute to a hemispheric background TFA deposition and was removed from the simulation. About 2 million model particles were present at any time of the simulation. Model particle masses and deposition fluxes were referenced to a Eulerian grid of 0.5° x 0.5° and 500 m thickness at the surface. CRI-STOCHEM Simulations. FLEXPART does not include a full tropospheric chemistry scheme and therefore does not consider any feedbacks of HFC-1234yf onto the prescribed OH and Cl field. Such feedbacks were shown to be minor given the low predicted atmospheric levels of HFC-1234yf.2 Nevertheless, they were considered by the second model (CRISTOCHEM), allowing an influence of HFC-1234yf on OH concentrations and possible nonlinear responses to HFC1234yf emissions. CRI-STOCHEM31−34 is a global 3-dimensional CTM which uses a Lagrangian approach to advect 50 000 air parcels. The transport and radiation models are driven by meteorological data for the year 1998, generated by the UK Met Office numerical weather prediction model with a resolution of 1.25° × 0.83° on 12 vertical levels extending to 100 hPa. The common representative intermediates mechanism (CRIv2-R5),35−37 which represents the chemistry of methane and 22 nonmethane hydrocarbons was employed in the model. Each parcel contains the concentrations of 219 species involved in 618 photolytic, gas-phase and heterogeneous chemical reactions, with a 5 min time step. The scheme was extended to incorporate OH degradation of HFC-1234yf. Dry deposition rates in the model differ over land and oceans and are described by differing species-dependent deposition velocities. Species dependent scavenging coefficients for convective and dynamic precipitation are taken from Penner et al.,38 with values for TFA assumed to be the same as those for nitric acid. Model results were referenced to an Eulerian grid of 5° × 5° and ∼1000 m thickness at the surface. Additional sensitivity runs with 40 Gg yr−1 and 80 Gg yr−1 of HFC-1234yf emitted in Europe, a comparative run with 11 Gg yr−1 of HFC-1234yf released in the U.S., and runs with varied rate constants for the HFC-1234yf reaction with OH were conducted.

■

RESULTS AND DISCUSSION Emissions. Total predicted emissions of HFC-1234yf for the EU27 countries for the year 2020 were estimated to be 9.8 and 17.2 Gg yr−1 for the low and high emission scenario, respectively. Additional emissions from Croatia, Norway, Switzerland and Turkey totalled 1.1 and 2.0 Gg yr−1, resulting in total emissions of 11.0 and 19.2 Gg yr−1 for EU27+ that were used for the model simulations (Supporting Information (SI) Table S1 and Figure S1). HFC-1234yf emissions from MACs in the U.S. were estimated for the year 2017 assuming complete changeover.14 Considering similarly sized car fleets in Europe and the U.S. (290 millions EU27+ vs 280 millions U.S.), emissions in the low scenario agreed very well (U.S.: 11.4 Gg yr−1), while slightly higher per vehicle emissions for the high scenario were determined for the U.S. (24.7 Gg yr−1). Predicted Atmospheric Mixing Ratios. In the following, model results are presented at different levels of temporal and 1652



Article

Table 2. Summary of Model Resultsa emission scenario

Europe (annual) FLEXPART

Europe (annual) STOCHEM

Europe (summer) FLEXPART

mean

max

mean

max

mean

max

U.S.b (summer) Luecken et al.2 mean

max

HFO-1234yf (ppt)

low high

1.5 2.6

10 18

0.9 1.5

2.1 3.7

0.83 1.5

8.4 15

5 7.4

na 300

TFA (ppt)

low high

0.06 0.1

0.3 0.6

0.07 0.12

0.20 0.35

0.09 0.15

0.54 0.94

na na

na >0.8

TFA conc. in rainwater (ng L−1)

low high

330 580

990 1700

465 800

1260 2160

690 1200

7000c 12 000c

na 500

na 1264

combined TFA deposition (kg km−2 yr−1)

low high

0.37 0.65

1.4 2.5

0.44 0.76

1.1 1.9

0.56d 1.0d

2.6d 4.4d

0.32b 0.48b

na 2.34b

a

Mean and maximum are spatial aggregates deduced from annual (seasonal) mean fields. Where available results for a low and high emission scenario are given, providing the likely range of future concentrations. For comparison the results of the summer-time study for the U.S.2 are given as well. na: information not available bOnly summer time average available (mid June to end of August). Deposition values were up-scaled under the assumption that during this period ∼50% of the emission occurred. cExtreme values were reached in low precipitation areas over the eastern Mediterranean. Maximum concentrations over central Europe were 3000−5000 ng L−1. dSeasonal result up-scaled to annual flux.

Figure 1. Annual mean surface mixing ratios of TFA simulated with (a) FLEXPART and (b) CRI-STOCHEM for the high emission scenario. The numbers at the top of each panel give the minimum, mean and maximum mixing ratio within the indicated box over Europe.

emission scenario. Our main results are compared to those of previous studies in Table 2. Annual mean mixing ratios of HFC-1234yf in Europe as predicted by FLEXPART were between 0.5 and 18 ppt (average 2.6 ppt) and mostly below 1 ppt outside Europe (see SI Figure S3a), while CRI-STOCHEM predicted considerably lower mean (1.5 ppt) and maxima (3.7 ppt, see SI Figure S3b). These differences can partly be explained by different output grid resolutions (see discussion in the SI) but also by different OH levels in the models. Due to its coarse resolution (5° × 5°) CRI-STOCHEM is known to produce greater OH levels in areas of high nitrogen dioxide emissions like Central Europe.39 In contrast, FLEXPART uses an OH climatology taken from a multimodel ensemble of global CTMs with a grid scale of the order of 2° × 2°. Consequently, annual average OH surface concentrations in CRI-STOCHEM were ∼1.7 × 106 and 2.6 × 106 molecules cm−3 for northern and southern Europe, respectively, whereas in FLEXPART the respective values were between 1.25 × 106 and 2.0 × 106 molecules cm−3. Therefore, HFC-1234yf was more quickly lost in CRI-

spatial aggregation. We refer to the temporal mean over the entire simulation period as annual mean (mixing ratios, rainwater concentrations) and to the temporal sum as annual total (deposition). Furthermore, seasonal mean or total values are given following the standard definition of meteorological seasons (spring: MAM; summer: JJA; autumn: SON; winter: DJF) and are referenced as seasonal mean or total. In addition, spatial aggregates of the annual or seasonal means/totals are given either for the whole globe or for a box covering Europe (−10° to 30° East and 35° to 65° North). These are referred to as global or European (mean, maximum, minimum, total) and seasonal global or European. Here the maximum and minimum refer to the spatial extremes, that is, to a local maximum or minimum in the distribution of a seasonally or annually averaged field. Only results from the high-emission scenario with emissions of 19.2 Gg yr−1 are presented here. Since concentrations and deposition fluxes in both model simulations scaled almost perfectly with emissions (see Figure S2 in the SI), the values derived from the high emission scenario can be divided by a factor of 1.7 to derive the results for the low 1653



Article

deposition (64%) took place outside Europe but was much more dispersed. This “domestic” deposition fraction decreased to 16% in winter and was above 50% in summer. In CRISTOCHEM, annual global total deposition of TFA were 7.5 Gg yr−1 and 10.9 Gg yr−1 for dry and wet deposition, respectively. Of this deposition, 42% was calculated to be deposited in Europe (3.2 Gg yr−1 and 4.6 Gg yr−1 for dry and wet deposition). This greater fraction, as compared with FLEXPART, is likely due to greater OH concentrations and resulting faster degradation of HFC-1234yf as discussed. In CRI-STOCHEM about 0.6 Gg yr−1 of TFA was not lost through dry or wet deposition but was subject to OH degradation. In accordance with the study in the U.S.,2 wet deposition of TFA was simulated to account for about 77% of total deposition in FLEXPART with little variation during the seasons, while in CRI-STOCHEM wet deposition accounted for 60% of total deposition. CRI-STOCHEM does not capture intense wet deposition events due to the model formulation and this is thought to be the main reason for these differences. Combined (dry plus wet) annual European total deposition rates of TFA in Europe were estimated to be 0.65 kg km−2 yr−1 on average (European maximum 2.5 kg km−2 yr−1, FLEXPART). In close agreement, CRI-STOCHEM simulated mean European total deposition rates of 0.76 kg km−2 yr−1 and European maximum values of 1.9 kg km−2 yr−1. Annual total dry deposition in FLEXPART was largest in Northern Italy, reaching up to 0.7 kg km−2 yr−1 (Figure 2a). In contrast, wet deposition mostly occurred in central Western Europe with a maximum around the Alpine area of 1.7 kg km−2 yr−1 (Figure 2b). The combined annual total deposition was largest in central Western Europe, with a maximum over the Po Valley (Figure 2c). Mean summer-time European deposition rates were predicted to be 0.25 kg km−2 (see SI Figure S6), which is very similar to the mean rate of 0.24 kg km−2 for the U.S.2 However, considering that the U.S. study assumed 61% of the total annual emission to occur in summer, European summertime deposition seems to be about twice as concentrated as over the U.S. This might be explained by more concentrated emissions in Europe but also by the fact that the European PBL undergoes slower ventilation compared with North America.40 Previously reported TFA deposition rates in Germany and Switzerland were between 0.05 and 0.2 kg km−2 yr−1 6,7 and of the order of 0.02 kg km−2 yr−1 in rural California and remote Canada.41,42 Predicted average deposition rates over Europe are likely to be up to 1 order of magnitude larger. A considerable fraction of the European emissions of HFC1234yf was subject to long-range transport toward Asia and across the Sahara desert and was deposited as TFA in Central Asia and in Tropical Africa, respectively (see SI Figure S7). However, annual total deposition rates did not surpass 0.2 kg km−2 in these areas. In additional CRI-STOCHEM simulations, emissions of 11 Gg yr−1 of HFC-1234yf from the U.S. increased TFA mixing ratios over the Iberian Peninsula and Northern Africa by ∼0.02 ppt (see SI Figure S8) compared with ∼0.1 ppt from European HFC-1234yf emissions (low scenario, both models). No significant influence of North American TFA on the rest of Europe was simulated. However, together with the seasonal analysis of HFC-1234yf and TFA transport from Europe, this suggests that long-range transport of HFC-1234yf from an upwind continent should not be neglected altogether and might be especially important in winter. This is in contrast to Luecken et al,2 who inferred little long-range influence for

STOCHEM than in FLEXPART despite the neglect of Clchemistry in CRI-STOCHEM. The largest annual mean HFC-1234yf mixing ratios (up to 18 ppt) were simulated at emission hot spots in Northern Italy, the Ruhr area and other metropolitan areas such as Paris (see SI Figure S3a). Seasonal mean HFC-1234yf mixing ratios varied considerably throughout the year, with largest values reached in winter and lowest in summer (see Figure S4 in the SI). Slower winter-time degradation of HFC-1234yf lead to a considerably greater area that is influenced by HFC-1234yf levels above 1 ppt. Our summer-time mean and maximum HFC-1234yf mixing ratios of 1.5 and 15 ppt, respectively, can be compared with the results of the summer-time simulation for the U.S.2 Their summer-time mean mixing ratio of 7.4 ppt is higher than the European value of 1.5 ppt but this difference can be explained by the fact that Luecken et al.2 assumed that 61% of annual total emissions of HFC-1234yf are released during the summer months, whereas a lower fraction of 25% was used in our study. Assuming that this factor of 2.4 in the emissions can be applied directly, summer-time European mean mixing ratios would be about 3.7 ppt, still slightly smaller than the value given for the U.S. European maxima were about 1 order of magnitude smaller than those predicted for the U.S., most likely owing to different model grid designs (see discussion of Figure S3 in the SI). In FLEXPART, the major fraction of emitted HFC-1234yf (84%) was oxidized via OH to TFF, whereas the rest was oxidized by Cl. Annual mean mixing ratios of TFA in Europe were between 0.006 and 0.6 ppt (European mean 0.1 ppt) in FLEXPART (Figure 1a) and the European mean and maximum were 0.12 and 0.35 ppt, respectively, in CRI-STOCHEM (Figure 1b). Highest levels were reached over Northern (FLEXPART) and Central (CRI-STOCHEM) Italy. In general TFA mixing ratios were relatively large in the Mediterranean, which can be related to higher abundances of OH and Cl radicals and slower wash-out due to smaller precipitation in this region (see SI Figures S10 and S11). Opposite to HFC-1234yf, TFA levels were lowest during winter and largest during summer (see Figure S5 in the SI). Especially mean TFA mixing ratios over Northern Italy (0.94 and 0.75 ppt in FLEXPART and STOCHEM, respectively) and the Mediterranean adjacent to Italy (∼0.4 ppt) were increased in summer. These values compare with maximum TFA levels of ∼0.8 ppt that were predicted off the North American East coast.2 Again a factor of 2.4 in the expected emissions should be applied to compare the studies, yielding comparable results for the two outflow areas. Mean European summer-time mixing ratios of 0.15 ppt (FLEXPART, SI Figure S5) and 0.22 ppt (CRI-STOCHEM) would enhance the formerly measured concentrations in background air in Germany and Switzerland of around 0.01 ppt (45−60 pg m−3)7 by more than a factor of 10.

■

TFA DEPOSITION AND RAINWATER CONCENTRATIONS Annual global total TFA deposition resulting from HFC-1234yf emissions of European MACs was estimated by FLEXPART to be 18.6 Gg yr−1 (sum of wet and dry deposition and end of air parcel life deposition). The remaining part of HFC-1234yf emitted (0.4 Gg) was either lost through degradation of TFA by OH (0.2 Gg) or accumulated during the simulation period. The annual total deposition of TFA over the European area was 6.4 Gg yr−1 (or 34% of the emissions). The remaining 1654



Article

smaller (up to 0.1 kg km−2 yr−1) in the major source regions of the Northern Hemisphere.43,44 Considering a lifetime of 14 year of HFC-134a, this TFA deposition will continue for several years even when HFC-134a emissions ceased. Replacing HFCs with shorter lived HFC-1234yf will flip the deposition pattern and likely lead to slight reductions in TFA deposition rates in the tropics and strong increases close to the major emitters. European mean TFA concentrations in rainwater were estimated by FLEXPART to be 580 ng L−1 (Figure 3a).

Figure 3. FLEXPART simulated TFA in rainwater concentrations; (a) annual mean and (b) 90th percentile of daily estimates.

Within EU27+, annual mean concentrations were predicted to be highest in the Czech Republic and Southern Germany (1700 ng L−1). CRI-STOCHEM results were slightly higher (800 ng L−1 for European mean rainwater concentrations and 2160 ng L−1 for European maxima), again related to the faster HFC1234yf degradation in CRI-STOCHEM. Highest TFA annual mean concentrations of 2000−2500 ng L−1 were simulated over Northern Africa (Figure 3a), where precipitation volumes are small but considerable amounts of TFA exported from Europe were present. However, TFA total deposition in these areas will be smaller than in Europe itself. Expected summer-time mean TFA in rainwater concentrations (see SI Figure S9) over the continent were mostly below 2000 ng L−1, larger than those predicted for the U.S.2 (∼1300 ng L−1). In addition, even larger values (up to 12 000 ng L−1) were simulated for some hot spots over the Mediterranean. These differences reflect the poorer

Figure 2. FLEXPART simulated annual total deposition rates of TFA from (a) dry, (b) wet deposition and (c) both processes.

U.S. summer-time HFC-1234yf emissions. Ultimately the range within which TFA might be transported from a source continent will depend on the seasonal cycle of the emissions. If, as assumed by Luecken et al.,2 emissions mainly occur during summer, long-range transport might well be neglected. However, if HFC-1234yf is released at constant rates, wintertime long-range transport will become important. Previous model studies analyzed TFA deposition from current HFC degradation in the atmosphere and predicted largest deposition rates for the tropics of about 0.2 kg km−2 yr−1, while rates were 1655



Article

European rainwater will be 6−8 times higher than the average of 110 ng L−1 measured in the 1990s. The no-effect level of TFA in water for the most sensitive algae was determined to be 120 000 ng L−1.13 Thus, even the largest simulated daily rainwater concentrations will be a factor of 10 smaller than this and European maxima of the annual mean rainwater concentrations will stay a factor of 60 below this critical level. Nevertheless, it would be advisible to probe TFA deposition and TFA concentrations in rainwater in the future in order to verify deposition rates predicted by this analysis and Luecken et al.2 Furthermore, the accumulation of TFA in endorheic basins should be investigated. While there are no major endorheic basins in Europe, TFA of European origin will still influence the endorheic basins in central Asia (e.g., Lake Aral, Caspian Sea) and Northern Africa (e.g., Lake Chad).

ventilation conditions of the European PBL and the HFC1234yf and TFA export into arid regions, respectively. To evaluate the maximal expected TFA concentrations in rainwater for individual rain events, FLEXPART derived daily wet deposition was divided by daily accumulated precipitation for each grid cell. It needs to be emphasized that these daily mean concentrations might be associated with large uncertainties introduced through the uncertainty in simulations of precipitation location and amount. Thus, as a more robust estimate of the highest expected daily mean TFA concentrations in European rainwater we considered the annual 90% percentiles of daily mean concentrations. These can reach up to 13 000 ng L−1 over northern Italy (Figure 3b). Previously measured concentrations of TFA in rainwater in Europe were in the range of 10 to 1500 ng L−1.6,7,45,46 Thus, in the future formerly infrequent high rainwater concentrations will become the average rainwater concentrations and future extreme concentrations might be 2 orders of magnitude larger than previously observed.

■

ASSOCIATED CONTENT

S Supporting Information *

■

Additional overview tables on country-wide emissions, sensitivity runs, seasonal model results, and long-range transport are provided. This material is available free of charge via the Internet at http://pubs.acs.org.

ENVIRONMENTAL IMPLICATIONS While there is no evidence of natural occurring TFA in ancient (older than 2000 years) groundwater, in ice cores or in ancient spring water,6,7,11 TFA was found in recent rainwater (110 ng L−1 in 1995/96 in Southern Germany7), in recent spring water (16−123 ng L−1 in Switzerland6) and in surface water (81 ng L−1 (rivers) and 119 ng L−1 (lakes)) average concentrations in Switzerland in 1996/1997.6 Concentrations in surface waters found in Europe were comparable with those in the U.S. with median values of 21 ng L−1 in remote regions and 144 ng L−1 in the San Francisco Bay Area.47 Discussions about the recent sources of TFA in rain and surface fresh water could not be resolved definitely. However, different HFCs and HCFCs partly degrade to TFA48 and it was recently concluded that “this source is predominantly anthropogenic”.12 TFA is naturally present in oceans.9,10 However, conflicting results exist in the literature about the consistency of a stable natural background. Whereas Frank et al.9 observed a stable background of 200 ng L−1 in both the Atlantic and the Southern Ocean (near the Antarctic continent), Scott et al.10 observed comparable concentrations of ∼150 ng L−1 in most oceans down to the deepest depths, but found TFA concentrations lower than 10 ng L−1 in the South Pacific ocean. Vertical profiles taken near hydrothermal vents suggest that these might be a natural source of TFA.10 For future deposition and accumulation of TFA in oceans it was estimated by a recent review that constant global deposition of 50−100 Gg yr−1 of TFA for the next 100 years would lead to an average increase of 3.7−7.4 ng L−1 in the oceans.12 However, this was done under the assumption of complete oceanic mixing, which seems to be a rather simplistic approach. To compare the TFA levels in rainwater and surface water observed in the 1990s with those expected in future rainwater over Europe, it has to be emphasized that highest concentrations are only expected in individual rain events after a period of TFA build up in the atmosphere. Even if highest daily concentrations in precipitation may reach up to 13 000 ng L−1 in northern Italy and similarly large summer-time average concentrations were observed over the Eastern Mediterranean (Table 2), annual mean rainwater concentrations in the same areas are expected to be below 1’500 ng L−1 (Figure 3) and European maxima will be around 2000 ng L−1. Expected average values of 600−800 ng L−1 of TFA in

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected].

■

ACKNOWLEDGMENTS The study was financially support by the European Fluorocarbon Technical Committee (EFCTC). The Swiss National Science Foundation is acknowledged for partly financing the IPAZIA computational cluster (project 206021_128754) on which FLEXPART calculations were performed.

■

REFERENCES

(1) Orkin, V. L.; Martynova, L. E.; Ilichev, A. N. High-accuracy measurements of OH reaction rate constants and IR absorption spectra: CH(2)CF-CF(3) and trans-CHFCH-CF(3). J. Phys. Chem. A 2010, 114 (19), 5967−5979, DOI: 10.1021/jp9092817. (2) Luecken, D. J.; Waterland, R. L.; Papasavva, S.; Taddonio, K. N.; Hutzell, W. T.; Rugh, J. P.; Andersen, S. O. Ozone and TFA impacts in North America from degradation of 2,3,3,3-tetrafluoropropene (HFO1234yf), A potential greenhouse gas replacement. Environ. Sci. Technol. 2010, 44 (1), 343−348. (3) Nielsen, O. J.; Javadi, M. S.; Andersen, M. P. S; Hurley, M. D.; Wallington, T. J.; Singh, R. Atmospheric chemistry of CF3CFCH2: Kinetics and mechanisms of gas-phase reactions with Cl atoms, OH radicals, and O-3. Chem. Phys. Lett. 2007, 439 (1−3), 18−22, DOI: 10.1016/j.cplett.2007.03.053. (4) Tang, X.; Madronich, S.; Wallington, T.; Calamari, D. Changes in tropospheric composition and air quality. J. Photochem. Photobiol., B 1998, 46 (1−3), 83−95. (5) Wallington, T. J.; Hurley, M. D.; Fracheboud, J. M.; Orlando, J. J.; Tyndall, G. S.; Sehested, J.; Mogelberg, T. E.; Nielsen, O. J. Role of excited CF3CFHO radicals in the atmospheric chemistry of HFC134a. J. Phys. Chem. 1996, 100 (46), 18116−18122. (6) Berg, M.; Muller, S. R.; Muhlemann, J.; Wiedmer, A.; Schwarzenbach, R. P. Concentrations and mass fluxes of chloroacetic acids and trifluoroacetic acid in rain and natural waters in Switzerland. Environ. Sci. Technol. 2000, 34 (13), 2675−2683. (7) Jordan, A.; Frank, H. Trifluoroacetate in the environment. Evidence for sources other than HFC/HCFCs. Environ. Sci. Technol. 1999, 33 (4), 522−527.

1656



Article

(8) Lifongo, L. L.; Bowden, D. J.; Brimblecombe, P. Thermal degradation of haloacetic acids in water. Int. J. Phys. Sci. 2010, 5 (6), 738−747. (9) Frank, H.; Christoph, E. H.; Holm-Hansen, O.; Bullister, J. L. Trifluoroacetate in ocean waters. Environ. Sci. Technol. 2002, 36 (1), 12−15. (10) Scott, B. F.; Macdonald, R. W.; Kannan, K.; Fisk, A.; Witter, A.; Yamashita, N.; Durham, L.; Spencer, C.; Muir, D. C. G. Trifluoroacetate profiles in the Arctic, Atlantic, and Pacific Oceans. Environ. Sci. Technol. 2005, 39 (17), 6555−6560. (11) Nielsen, O. J.; Scott, B. F.; Spencer, C.; Wallington, T. J.; Ball, J. C. Trifluoroacetic acid in ancient freshwater. Atmos. Environ. 2001, 35 (16), 2799−2801. (12) Tang, X.; Wilson, S. R.; Solomon, K. R.; Shao, M.; Madronich, S. Changes in air quality and tropospheric composition due to depletion of stratospheric ozone and interactions with climate. Photochem. Photobiol. Sci. 2011, 10 (2), 280−291. (13) Boutonnet, J. C.; Bingham, P.; Calamari, D.; de Rooij, C.; Franklin, J.; Kawano, T.; Libre, J. M.; McCulloch, A.; Malinverno, G.; Odom, J. M.; Rusch, G. M.; Smythe, K.; Sobolev, I.; Thompson, R.; Tiedje, J. M. Environmental risk assessment of trifluoroacetic acid. Human Ecol. Risk Assess. 1999, 5 (1), 59−124. (14) Papasavva, S.; Luecken, D. J.; Waterland, R. L.; Taddonio, K. N.; Andersen, S. O. Estimated 2017 refrigerant emissions of 2,3,3,3tetrafluoropropene (HFC-1234yf) in the United States resulting from automobile air conditioning. Environ. Sci. Technol. 2009, 43 (24), 9252−9259, DOI: 10.1021/es902124u. (15) Clodic, D.; Yu, Y.; Tremoulet, A.; Palandre, L.; Fayolle, F.; Castel, J.; Lallemand, J. A.; Dujardin, B. Research study on the definition of the implementation of a method of measurement of annual leakflowrates (LFRs) of MAC systems. ACEA/ARMINES 2006, in press. (16) Schwarz, W.; Harnisch, J. Establishing the leakage rates of mobile air conditioners, European Commission (DG Environment), 2003; http://digital.library.unt.edu/ark:/67531/metadc29372/ (accessed Aug 15, 2011). (17) Yang, X.; Pyle, J. A.; Cox, R. A. Sea salt aerosol production and bromine release: Role of snow on sea ice. Geophys. Res. Lett. 2008, 35 (16), -. (18) Yu, Y. Z.; Clodic, D. Leak flow rate of MAC systems and components 1-Laboratory tests, fleet tests and correlation factor. Int. J. Refrig.-Rev. Int. Froid. 2010, 33 (7), 1465−1477, DOI: 10.1016/ j.ijrefrig.2010.05.017. (19) Sousa, D. Study of refrigerant emissions from mobile airconditioning compressor shaft seals. Ph.D. Dissertation, Ecole des Mines Paris, 2008. (20) Center for International Earth Science Information Network (CIESIN) - Columbia University; Centro Internacional de Agricultura Tropical (CIAT); Socioeconomic Data and Applications Center (SEDAC), Columbia University: Palisades, NY, 2005, available from http://sedac.ciesin.columbia.edu/gpw (accessed March 1, 2011). (21) European Commission Joint Research Centre (JRC)/Netherlands Environmental Assessment Agency (PBL) Emission Database for Global Atmospheric Research (EDGAR), release version 4.0, 2009; http://edgar.jrc.ec.europa.eu (accessed March 1, 2011). (22) Stohl, A.; Hittenberger, M.; Wotawa, G. Validation of the Lagrangian particle dispersion model FLEXPART against large-scale tracer experiment data. Atmos. Environ. 1998, 32 (24), 4245−4264. (23) Stohl, A.; Forster, C.; Frank, A.; Seibert, P.; Wotawa, G. Technical note: The Lagrangian particle dispersion model FLEXPART version 6.2. Atmos. Chem. Phys. 2005, 5, 2461-2474. (24) Fiore, A. M.; Dentener, F. J.; Wild, O.; Cuvelier, C.; Schultz, M. G.; Hess, P.; Textor, C.; Schulz, M.; Doherty, R. M.; Horowitz, L. W.; MacKenzie, I. A.; Sanderson, M. G.; Shindell, D. T.; Stevenson, D. S.; Szopa, S.; Van Dingenen, R.; Zeng, G.; Atherton, C.; Bergmann, D.; Bey, I.; Carmichael, G.; Collins, W. J.; Duncan, B. N.; Faluvegi, G.; Folberth, G.; Gauss, M.; Gong, S.; Hauglustaine, D.; Holloway, T.; Isaksen, I. S. A.; Jacob, D. J.; Jonson, J. E.; Kaminski, J. W.; Keating, T. J.; Lupu, A.; Marmer, E.; Montanaro, V.; Park, R. J.; Pitari, G.; Pringle,

K. J.; Pyle, J. A.; Schroeder, S.; Vivanco, M. G.; Wind, P.; Wojcik, G.; Wu, S.; Zuber, A. Multimodel estimates of intercontinental sourcereceptor relationships for ozone pollution. J. Geophys. Res. 2009, 114, DOI: 10.1029/2008JD010816. (25) Papadimitriou, V. C.; Talukdar, R. K.; Portmann, R. W.; Ravishankara, A. R.; Burkholder, J. B. CF3CFCH2 and (Z)CF3CFCHF: temperature dependent OH rate coefficients and global warming potentials. Phys. Chem. Chem. Phys. 2008, 10 (6), 808− 820, DOI: 10.1039/b714382f. (26) Allan, W.; Struthers, H.; Lowe, D. C. Methane carbon isotope effects caused by atomic chlorine in the marine boundary layer: Global model results compared with Southern Hemisphere measurements. J. Geophys. Res. 2007, 112 (D4), DOI: 10.1029/2006JD007369. (27) Allan, W.; Struthers, H.; Lowe, D. C.; Fletcher, S. E. M. Modeling the effects of methane source changes on the seasonal cycles of methane mixing ratio and delta C-13 in Southern Hemisphere midlatitudes. J. Geophys. Res. 2010, 115, DOI: 10.1029/ 2009JD012924. (28) George, C.; Saison, J. Y.; Ponche, J. L.; Mirabel, P. Kinetics of mass-transfer of carbonyl fluoride, trifluoroacetyl fluoride, and trifluoroacetyl chloride at the air/water interface. J. Phys. Chem. 1994, 98 (42), 10857−10862. (29) Wesely, M. L. Parameterization of surface resistances to gaseous dry deposition in regional-scale numerical-models. Atmos. Environ. 1989, 23 (6), 1293−1304. (30) Hertel, O.; Christensen, J.; Runge, E. H.; Asman, W. A. H.; Berkowicz, R.; Hovmand, M. F.; Hov, O. Development and testing of a new variable scale air-pollution model-Acdep. Atmos. Environ. 1995, 29 (11), 1267−1290. (31) Utembe, S. R.; Cooke, M. C.; Archibald, A. T.; Jenkin, M. E.; Derwent, R. G.; Shallcross, D. E. Using a reduced common representative intermediates (CRIv2-R5) mechanism to simulate tropospheric ozone in a 3-D Lagrangian chemistry transport model. Atmos. Environ. 2010, 44 (13), 1609−1622. (32) Archibald, A. T.; Cooke, M. C.; Utembe, S. R.; Shallcross, D. E.; Derwent, R. G.; Jenkin, M. E. Impacts of mechanistic changes on HOx formation and recycling in the oxidation of isoprene. Atmos. Chem. Phys. 2010, 10 (17), 8097−8118. (33) Derwent, R. G.; Stevenson, D. S.; Doherty, R. M.; Collins, W. J.; Sanderson, M. G.; Johnson, C. E. Radiative forcing from surface NOx emissions: spatial and seasonal variations. Clim. Change 2008, 88 (3− 4), 385−401, DOI: 10.1007/s10584-007-9383-8. (34) Collins, W. J.; Stevenson, D. S.; Johnson, C. E.; Derwent, R. G. Tropospheric ozone in a global-scale three-dimensional Lagrangian model and its response to NOx emission controls. J. Atmos. Chem. 1997, 26 (3), 223−274. (35) Jenkin, M.; Watson, L.; Utembe, S.; Shallcross, D. A common representative intermediates (CRI) mechanism for VOC degradation. Part 1: Gas phase mechanism development. Atmos. Environ. 2008, 42, 7185−7195. (36) Utembe, S. R.; Watson, L. A.; Shallcross, D. E.; Jenkin, M. E. A common representative intermediates (CRI) mechanism for VOC degradation. Part 3: Development of a secondary organic aerosol module. Atmos. Environ. 2009, 43 (12), 1982−1990. (37) Watson, L.; Shallcross, D.; Utembe, S.; Jenkin, M. A common representative intermediates (CRI) mechanism for VOC degradation. Part 2: Gas phase mechanism reduction. Atmos. Environ. 2008, 42 (31), 7196−7204. (38) Penner, J. E.; Atherton, C. S.; Graedel, T. E. Global emissions and models of photochemically active compounds. Global Atmos.Biospheric Chem. 1994, 48, 223−247 261.. (39) Archibald, A. T. Numerical modelling of the Lower Atmosphere. Ph.D., University of Bristol, Bristol, UK, 2009. (40) Stohl, A. Intercontinental Transport of Air Pollution; Springer: Berlin, 2004. (41) Scott, B. F.; Mactavish, D.; Spencer, C.; Strachan, W. M. J.; Muir, D. C. G. Haloacetic acids in Canadian lake waters and precipitation. Environ. Sci. Technol. 2000, 34 (20), 4266−4272. 1657



Article

(42) Cahill, T. M.; Thomas, C. M.; Schwarzbach, S. E.; Seiber, J. N. Accumulation of trifluoroacetate in seasonal wetlands in California. Environ. Sci. Technol. 2001, 35 (5), 820−825. (43) Kanakidou, M.; Dentener, F. J.; Crutzen, P. J. A global threedimensional study of the fate of HCFCs and HFC-134a in the troposphere. J. Geophys. Res. 1995, 100 (D9), 18781−18801. (44) Kotamarthi, V. R.; Rodriguez, J. M.; Ko, M. K. W.; Tromp, T. K.; Sze, N. D.; Prather, M. J. Trifluoroacetic acid from degradation of HCFCs and HFCs: A three-dimensional modeling study. J. Geophys. Res. 1998, 103 (D5), 5747−5758. (45) Rompp, A.; Klemm, O.; Fricke, W.; Frank, H. Haloacetates in fog and rain. Environ. Sci. Technol. 2001, 35 (7), 1294−1298, DOI: 10.1021/es0012220. (46) Von Sydow, L. M.; Grimvall, A. B.; Boren, H. B.; Laniewski, K.; Nielsen, A. T. Natural background levels of trifluoroacetate in rain and snow. Environ. Sci. Technol. 2000, 34 (15), 3115−3118. (47) Cahill, T. M.; Seiber, J. N. Regional distribution of trifluoroacetate in surface waters downwind of urban areas in Northern California. USA. Environ. Sci. Technol. 2000, 34 (14), 2909−2912. (48) Montzka, S. A.; Reimann, S. C. L. A.; Engel, A.; Krüger, K.; O’Doherty, S.; Sturges, W. T.; Blake, D.; Dorf, M.; Fraser, P.; Froidevaux, L.; Jucks, K.; Kreher, K.; Kurylo, M. J.; Mellouki, A.; Miller, J.; Nielsen, O.-J.; Orkin, V. L.; Prinn, R. G.; Rhew, R.; Santee, M. L.; Stohl, A.; Verdonik, D. Ozone-depleting substances (ODSs) and related chemicals. In Scientific Assessment in Ozone Depletion: 2010, Chapter 1, Global Ozone Research and Monitoring Project-Report No. 52; World Meteorological Organization: Geneva, Switzerland, 2011

1658


Future Emissions and Atmospheric Fate of HFC-1234yf from Mobile

Recommend Documents