The persistence of pesticides in atmospheric particulate phase: An emerging air quality issue

The persistent organic pollutants (POPs) due to their physicochemical properties can be widely spread all over the globe; as such they represent a serious threat to both humans and wildlife. According to Stockholm convention out of 24 officially recognized POPs, 16 are pesticides. The atmospheric life times of pesticides, up to now were estimated based on their gas-phase reactivity. It has been only speculated that sorption to aerosol particles may increase significantly the half‐lives of pesticides in the atmosphere. The results presented here challenge the current view of the half-lives of pesticides in the lower boundary layer of the atmosphere and their impact on air quality and human health. We demonstrate that semivolatile pesticides which are mostly adsorbed on atmospheric aerosol particles are very persistent with respect to the highly reactive hydroxyl radicals (OH) that is the self-cleaning agent of the atmosphere. The half-lives in particulate phase of difenoconazole, tetraconazole, fipronil, oxadiazon, deltamethrin, cyprodinil, permethrin, and pendimethalin are in order of several days and even higher than one month, implying that these pesticides can be transported over long distances, reaching the remote regions all over the world; hence these pesticides shall be further evaluated prior to be confirmed as POPs.

are calculated from the gas-phase reactivity with respect to the OH radicals, using structure-activity relationships (SAR) used by U.S. Environmental Protection Agency, software AOPWIN (Atmospheric Oxidation Program) 11 .
However, the heterogeneous reactions of pesticides which occur on the surface of atmospheric aerosols may proceed at different rates than the gas-phase reactions. Indeed, Socorro et al. 12 have shown that half-lives of 8 commonly used pesticides span from 9 to > 24 days for an atmospheric ozone level of 9.8 · 10 11 cm −3 , demonstrating that these species are very persistent regarding the ozone (O 3 ) reactivity on atmospheric particulate phase 13 . In this context, it is essential to investigate the fate of pesticides adsorbed on aerosol particles to determine their persistence in the atmosphere with respect to the highly reactive hydroxyl radicals. Indeed, OH radicals are considered as the dominant oxidizing and cleansing agent determining the oxidative capacity of the atmosphere.
In the present study we investigated the heterogeneous reactivity of eight pesticides, difenoconazole, tetraconazole, fipronil, oxadiazon, deltamethrin, cyprodinil, permethrin, and pendimethalin enriched in atmospheric particulate phase toward gas phase OH radicals. The emerged results strongly indicate that these commonly used pesticides once adsorbed on the atmospheric aerosols can be transported thousands of kilometres far away from the place where have been applied.
From the environmental pollution point of view these results are extremely important and it should be considered while developing appropriate environmental strategies which in turn will contribute to better describe and understand the atmospheric behavior of pesticides and their persistence in the environment.

Results and Discussion
Silica is commonly a major constituent of mineral dust 14 on which surface the organic compounds are adsorbed and (photo)oxidized during their transport 15 . To study the OH heterogeneous kinetic reactions the pesticides were coated on silica particles (AEROSIL R812). Silica particles are characterised by a high surface to volume ratio which makes them an ideal proxy for investigation of heterogeneous reactions. The pesticides adsorbed on silica particles were simultaneously oxidized by gas-phase OH radicals and gas-phase O 3 which is necessary for the production of OH radicals as explained in the supplementary information.
A detailed procedure describing the calculations of the second order rate constants k OH II part ( ) for the heterogeneous reactions of OH radical with cyprodinil, deltamethrin, permethrin, and pendimethalin is given in the SI Appendix.
In order to determine the second order rate constant k OH II part ( ) , the experimental pseudo first order reaction rate constants k OH I part ( ) were plotted as a function of the OH radical concentrations (Fig. 2). For the heterogeneous reactions of O 3 with of cyprodinil, deltamethrin, permethrin, and pendimethalin, the following second order rate constants (k O II part 3 ( ) ), (8.8 ± 0.2) · 10 −19 , (8.0 ± 0.2) · 10 −19 , (6.0 ± 0.2) · 10 −19 and (3.4 ± 0.1)·10 −19 cm 3 molecule −1 s −1 12 , were used in Eq. S12. The intercept of the plot depicted in Fig. 2 ) which was then subtracted, following the Eq. S12, to finally obtain the k OH . The linearity of the kinetic data depicted in Fig. 2 corresponds to a Langmuir-Rideal mechanism [17][18][19] . In Fig. 2 we assume that the linearity will continue down to ambient OH concentrations. Considering the extremely short life time of OH radicals comprised between 0.01 s and 1 s 16 , it is logical to expect direct bimolecular collision between the gas-phase OH radicals and the particulate pesticides without the adsorption step of OH radical on the surface of particles which would correspond to Langmuir-Hinshelwood mechanism 12 .
These results are in contrast with recent studies 20-23 related to the heterogeneous OH reactions with liquid surfaces which proceed through Langmuir-Hinshelwood mechanism. The different behavior of the kinetic data could be ascribed to the different surfaces i.e. solid versus liquid film.
A study on the dependence of the first order rate coefficients of the deposited sample mass on the particles surface would be important to distinguish between the both mechanisms. Also, the experimental set-up is unable to generate low OH concentrations; hence, the parameter space of OH concentrations is not sufficient. However, even if the slope changes in Fig. 2 and kinetics behave more like Langmuir-Hinshelwood, the rates at ambient OH concentrations will be lower implying even longer half-lives of the pesticides.
According to the Stockholm convention 9 , the classification of POPs is made by use of AOPWIN software. Basically, the program calculates the first order rate constants (k OH I gas ( ) ) by multiplying the second order rate constants (k OH II gas ( ) ) for the reaction of OH radicals with the organic compounds in the gas phase by an average OH radical concentration of 1.5 · 10 6 cm −3 and for an exposure of 12 h by day which corresponds to 7.5 · 10 5 cm −3 day −1 24,25 . It follows from Eq. 1 the half-life ( OH II gas ( ) However, most of the pesticides represent the compounds with low vapor pressure (below 1 Pa) and thus a significant fraction of pesticides is enriched into atmospheric particulate phase. It was already speculated by Scheringer 26 that the adsorption of semivolatile compounds on the atmospheric particles may significantly increase their half-lives. This implies that in reality the half-lives of the pesticides can be much longer than the current estimates by AOPWIN. In addition, the estimations made by AOPWIN gave unrealistically high rate constants for complex and highly chlorinated chemicals 27,28 .
Since the oxidation by OH radicals was considered as the most effective removal pathway of the pesticides, Scheringer et al. 29 suggested a model to describe the reactivity of pesticides with respect to the OH radicals in both the gas phase and the particulate phase (Eq. 2).   is the rate constant describing the reactions between OH radical and the pesticides adsorbed on the surface of atmospheric particles.
The fraction of pesticides adsorbed on the surface of particles, Φ pesticide , can be determined by AEROWIN software based on the Junge-Pankow partitioning model 30 : where c is a factor that depends on the excess heat of desorption from the particle surface (17.2 Pa cm), Θ is the surface of particles per unit of air volume (6.3 · 10 −6 cm 2 m −3 ) and P l s (Pa) is the vapor pressure of considered pesticide at 25 °C.
The estimated Φ pesticide values, the effective rate constants and the calculated half-lives of the eight studied pesticides with respect to the OH reactivity are summarized in Table 1.
The partitioning between the gas phase and particulate phase is largely variable from 0.01 for pendimethalin up to 0.99 for difenoconazole. Therefore, Φ pesticide should be taken into account while estimating the half-life of the pesticides. Table 1 shows that pendimethalin and cyprodinil are predominantly found in the gas phase; thus the gas phase reactivity towards OH radicals contributes essentially to the half-lives of those pesticides.
Tetraconazole and oxadiazon partition more or less equally between the gas phase and the particulate phase. Considering the OH reactivity in both phases, these pesticides exhibit half-lives much higher than 1 day which implies long-range transport.
On the other hand, the estimated Φ pesticide values of deltamethrin and permethrin indicate that these two pesticides are mostly enriched in the particulate phase. The experimentally measured rate constants k OH II part ( ) of these two pesticides are very close to the + k OH II gas part ( ) suggesting that the atmospheric half-lives are determined by the OH reactivity in the particulate phase. The half-lives of deltamethrin and permethrin are higher than 2 days strongly indicating that these two pesticides can be potentially considered as POPs according to the Stockholm convention.
The example of difenoconazole that is completely enriched in the particulate phase is especially striking with half-life much higher than 16 days with respect to the OH reactivity. Such long lived pesticides will be transported far from their place of release impacting the regional and global air quality, human health and wildlife. Figure 3 graphically shows the difference between the OH reactivity towards pesticides dispersed in the gas phase and pesticides adsorbed on the atmospheric particles.

Concluding remarks
Heterogeneous reactivity of OH radical with 8 commonly used pesticides was investigated.
In the past, it has been assumed that these pesticides are predominantly degraded by the OH radical in the gas phase. However, the majority of these pesticides are adsorbed on the atmospheric particle surface; thus their reactivity in particulate phase should be considered prior to take any conclusion about their half-lives and possible hazards.
in the gas phase. This implies that the heterogeneous OH oxidation of pesticides adsorbed on atmospheric particles is a very slow process suggesting that the pesticides can persist long time in the atmosphere prior to be removed and transferred to terrestrial and aquatic ecosystems. Therefore, the estimated half-lives of these pesticides based on AOPWIN estimations are not valid and should not be used during the preparation of adequate legislation.
The emerged kinetic data from this study can be of great help for further validation of AOPWIN program with more complex organic molecules containing more functional groups in order to increase the confidence in the accuracy of the half-lives estimations for pesticides.
Silica particles coating. Commercial hydrophobic silica particles (AEROSIL R812, Degussa, purity SiO 2 content ≥ 99.8%, average primary particle size of 7 nm and specific surface area (BET) of 260 ± 30 m 2 g −1 ) were used as proxy of atmospheric mineral aerosols. Silica particles were coated with the pesticides according to a liquid/solid adsorption. 5 mL of pesticides solution at concentration 20 mg L −1 in dichloromethane (for HPLC, ≥ 99.8%, Sigma-Aldrich) was mixed with 500 mg of silica particles in a Pyrex bulb wrapped with aluminum foil. This bulb was ultrasonicated for 15 min. Then, dichloromethane was evaporated by a rotary evaporator (Rotavapor R-114, Büchi) at 40 °C and 850 ± 85 mbar. The load of pesticides on silica particles was about 0.02% by weight and the percentage of the coated aerosol surface was between 0.2 and 0.4%, less than a monolayer, assuming a uniform particle surface coverage as was detailed by Socorro et al. 12 .

Production and measurements of OH radical. A HS-PTR-MS (High Sensitivity -Proton Transfer
Reaction -Mass Spectrometer, Ionicon Analytik) was used to follow the concentrations of m-xylene and 2,3-dimethyl-2-butene (DMB). DMB is an alkene which produces OH radicals with a yield near unity through the reaction with ozone. m-xylene was used as a tracer to determine the OH radical concentrations.
The HS-PTR-MS allows on-line and continuous monitoring of organic compounds with detection limit of only few part par trillion (ppt). The ionization process is a soft process, meaning the energy transferred during the ionization is small (as compared to electron impact ionization) which limits the fragmentation of the initial compounds.
OH radical reactivity. The coated powders of inert AEROSIL particles with pesticides were exposed in a rotating quartz bulb to six different OH radical concentrations (3 · 10 7 ; 6.1 · 10 7 ; 8 · 10 7 ; 9.3 · 10 7 ; 1.4 · 10 8 and 1.5 · 10 8 cm −3 ). This simplified method is a useful means to expose the pesticides at sub-monolayer thickness to reactive species from the gas phase surrounding the particles (such as ozone and OH radicals).
Detailed descriptions of materials and methods, silica particles coating, description of OH radical production and kinetic experiments, extraction and pesticide quantification, measurements of OH radicals and other procedures used are given in SI Appendix.