Occurrence of Aerosol-Bound Fullerenes in the Mediterranean Sea

Jul 18, 2011 - fullerenes aerosol phase concentrations were 0.06 ng/m3 and 0.48 ng/m3 respectively for the Mediterranean Sea atmosphere. C70 fullerene...
0 downloads 0 Views 3MB Size
ARTICLE pubs.acs.org/est

Occurrence of Aerosol-Bound Fullerenes in the Mediterranean Sea Atmosphere Josep Sanchís,† Naiara Berrojalbiz,† Gemma Caballero,† Jordi Dachs,† Marinella Farre,†,* and Damia Barcelo†,‡,§ †

Department of Environmental Chemistry, Institute of Environmental Assessment and Water Research (IDAEA-CSIC), C/Jordi Girona, 18-26, 08911, Barcelona, Catalonia, Spain ‡ Catalan Institute of Water Research (ICRA), C/Emili Grahit, 101, 17003, Girona, Catalonia, Spain § King Saud University, P.O. Box 2455, 11451, Riyadh, Saudi Arabia

bS Supporting Information ABSTRACT: This work describes the assessment of a selection of fullerenes including C60 and C70 fullerene, N-methylfulleropyrrolidine, C60 pyrrolidine tris-acid ethyl ester, [6,6]-Phenyl-C61 butyric acid butyl ester and [6,6]-Thienyl C61 butyric acid methyl ester, in airborne particulate from the Mediterranean Sea collected during two sampling campaigns from Barcelona to Istanbul and Alexandria, respectively. The analysis of the samples was carried out using a new method based on liquid chromatography coupled to mass spectrometry (LCMS) presenting sensitivities between 5.4 and 20.9 pg/m3. A total number of 43 samples covering the different basins of Mediterranean Sea were analyzed. Fullerenes were detected in all analyzed samples and quantifiable concentrations were found in 28 of the analyzed samples. The median of C60 and C70 fullerenes aerosol phase concentrations were 0.06 ng/m3 and 0.48 ng/m3 respectively for the Mediterranean Sea atmosphere. C70 fullerene was the most frequently detected compound and also it was found in the higher concentrations for most samples, reaching 233.8 ng/m3. The modeled back-trajectories disclose that those samples with higher concentrations of fullerenes were related to air masses which had been circulating over regions with an intense industrial activity, but the variability of the C70/C60 ratio suggests multiple different sources. These results are related to the incidental emissions from urban and industrial development, underpinning the need of studying the possible risks associated to carbon nanoparticles in the environment and the need of evaluating the possible consequences of their ubiquitous occurrence.

’ INTRODUCTION Natural sources of nanoparticles (NPs) in the atmosphere include volcanic eruptions, forest fires, hydrothermal vent systems, physical and chemical weathering of rocks, precipitation reactions, and biological processes. However, the natural background of NPs in the atmosphere is low in comparison to those caused by combustion processes, diesel and gasoline fueled vehicles and stationary combustion sources which have for many years contributed to the particulate material in the atmosphere in a wide size range, including NPs. Carbon based nanomaterials (NMs) of different kinds have also been reported to occur in ordinary hydrocarbon flames 1,2 and, emitted from common heat sources.1,3 It has been assessed that the amount of incidental NPs in the atmosphere due to human activity is more than 36% of the total particulate number concentrations, and the forecast for the next coming years is that there will be a strong increase due to emissions from nanotechnology industry.4 The total global investment in nanotechnologies was around $10 billion in 2005 and it is estimated that the annual value for all nanotechnology-related products will be $1 trillion by 20112015.5 r 2011 American Chemical Society

Among them, fullerenes have attracted considerable interest in many fields of research and have found numerous applications in personal care products,6 textile industry,7 analytical chemistry,8 and microelectronics,9 as well as several potential uses in medicine.10,11 Therefore, it is essential to determine the risk that these materials may pose to human health and the environment.12,13 Most of current environmental research on the impact of carbon based NMs have been directed to elucidate ecotoxicological aspects. However, only few quantitative analytical methods for measuring NPs in natural systems are available, which results in a serious lack of information about their occurrence in the environment. During the last years, some analytical methods for the analysis of fullerenes in aquatic ecosystems were presented,1416 but most of these methods present limits of detection (LOD) that are too high for their use in quantitative analysis Received: March 5, 2011 Accepted: July 8, 2011 Revised: June 20, 2011 Published: July 18, 2011 1335

dx.doi.org/10.1021/es200758m | Environ. Sci. Technol. 2012, 46, 1335–1343

Environmental Science & Technology of real environmental samples. In addition, the presence of carbon based NMs in environmental compartments other than aquatic’s remains completely unexplored. Whereas the emission of carbon NPs have been proved that can occur in combustion processes, to date, no information has been reported to assess their presence in the atmosphere of natural environments. Regarding atmospheric particulate, an early precedent reported the fullerene detection by MALDI-TOF-MS in soot samples of southwest Detroit.17 In this work, the presence of fullerenes was postulated to be related to combustion engines and coal-based power plants. Some recent works have studied the fullerenes potential for occupational exposure in laboratory conditions,18 or during their production processes.19 In order to better understand the whole environmental risk associated to nanosized pollution, their relations with other contaminants, and their fate and behavior it is of high importance to evaluate their presence in open atmosphere. The overall aim of this study was to assess the occurrence of six fullerenes including, C60 fullerene, C70 fullerene, N-methylfulleropyrrolidine C60 (MFP), pyrrolidine tris-acid ethyl ester (CPTAE), [6,6]-phenyl C61 butyric acid methyl ester (PCBM) and [6,6]thienyl C61 butyric acid methyl ester (ThPCBM) associated to aerosols from the open Mediterranean Sea. The selection of the studied compounds was done according to their possible incidental emission, as well as, their industrial application. MFP is used as intermediate of synthesis of substituted fullerenes in industrial processes,14 as well as, an electron acceptor20 while CPTAE is an adduct derived from the N-methylfulleropyrrolidine. PCBM and ThPCBM are derivatives, which are employed as n-type organic semiconductors.20 Therefore, the specific objectives were (a) to develop an analytical method for the analysis of selected fullerenes in airborne particulate, (b) to assess the occurrence of selected compounds in 43 aerosol samples collected during two consecutives sampling campaigns in the open Mediterranean Sea (c) to estimate possible relations between air flux trajectories and concentration of fullerenes.

’ EXPERIMENTAL SECTION Sampling. Samples were collected through the Mediterranean Sea during two sampling campaigns in June 2006 and May 2007 on board of the oceanographic vessel B/O García del Cid (CSIC). Sampling was carried out according to the procedure described elsewhere.21,22 Briefly, forty-three transects were completed comprehensively covering a significant area of the Mediterranean Sea (transect coordenates are detailed in Table S1 and schematically represented in Figure S1, both in the Supporting Information (SI)). High volume air samplers (Echo PUF high volume sampler, TCR Tecora, Milan, Italy) were installed on the upper deck of the research vessel “Garcia del Cid”, above 67 m the sea level. Contamination from boat engines was prevented by installing a directional valet that automatically stopped the sampling process when the ship emissions could potentially interfere in it. The air particulate phase was collected on quartz fiber filters of 102 mm diameter (QM-A type, particle retention: 0.7 μm, Whatman International Ltd., Brentford, Middlesex, UK). The total air volume, the total particulate weight, the organic carbon content and the soot carbon content of the particulate were calculated. The sampled air volumes and the particulate characterization data are also included in SI Table S1. Filters were folded and kept wrapped in aluminum foil in a zip

ARTICLE

plastic bag. The samples preservation was accomplished by freezing the filters at 20 °C until just prior to the beginning of the analytical process. In addition, in order to compare the pattern and levels of fullerenes in the open Mediterranean Sea atmosphere with those of an urban area, aerosol samples were taken at the roof of IDAEA-CSIC in Barcelona. This is the first study that reports fullerenes bound to aerosol. An important issue is weather QMA filters will retain effectively most atmospheric fullerenes. Obviously, fullerenes are much smaller than the cutoff size of the QMA filters. So it is relevant to address the gas-particle partitioning of fullerenes. There are no previous reports on this partitioning for nanoparticles. As discussed elsewhere,16 the applicability of the methods available for hydrophobic compounds partitioning has not been comprehensively assessed for their applicability to predicting the environmental transport and fate of fullerenes. However, due to their low vapor pressures (5  106 tor),25 and using the estimation method of the soot-air partition coefficient suggested by Van Noort et al.,26 for PAHs, we obtain a soot-air partition coefficient of 1014. For the sorption to organic carbon from the gas phase, the partition coefficient may be significantly lower (23 orders of magnitude). Still, with soot-air and organic carbon-air partition coefficients of this order of magnitude, more than 99% of fullerenes should be associated to aerosol particles, similarly to what has been observed for PAHs with more than five aromatic rings.27 Fullerenes will be associated especially to the organic and soot phases. However, because of their very low volatility, this gas particle partitioning will not be very effective for repartitioning to other aerosols after they are associated it a given aerosol particle. The aerosol size distribution of aerosol mass, organic carbon and soot carbon peak for aerosols larger than 1 μm. Therefore, QMA filters retain most aerosol organic carbon and soot carbon mass, and thus will presumably retain most fullerenes. It is unlikely that fullerenes can persist for long time as “free” atmospheric nanoparticles and they will sorb to aerosols after a short atmospheric residence time. In any case, there is a need to determine experimentally the fundamental partitioning properties of NP, and size-distributed aerosol bound concentrationsof fullerenes to gain knowledge on their phase partitioning and transport potential. Chemicals. C60 fullerene (99.9%, reference 572500), C70 fullerene (99%, reference 482994), MFP (highest purity, reference 668184), CPTAE (>97%, reference 709093), ThPCBM (>99%, reference 688215) and PCBM (>99.5%, reference 684449) were purchased from Sigma-Aldrich (Steinheim, Germany). Labeled standard 13C60 (>99%, reference MER613) was used as surrogate and 13C70 (>95%, reference MER713) was used as internal standard. Both labeled standards were purchased from MER Corporation (Tucson, Arizona, USA). LC-MS/MS analysis of the labeled standards did not identify any interference at quantifiable levels. Stock standard solutions of individual compounds were prepared in toluene. A seven points standard calibration curve ranging from1.00 μg/L to 1.0 mg/L of each compound was prepared in toluene/ methanol 1:1 and 13C60 and13C70 (0.1 mg/L). Methanol, toluene and HPLC-grade water were supplied by Merck (Darmstadt, Germany). Nitrogen used as drying gas with 99.995% purity was acquired from Air Liquide (Spain). Sample Pretreatment and Extraction. An eighth part of each QMA filter was carefully cut out for the analysis of the present 1336

dx.doi.org/10.1021/es200758m |Environ. Sci. Technol. 2012, 46, 1335–1343

Environmental Science & Technology work. Then, filters were warmed to 50 °C for 12 h and kept in a drier for 24 h. Each filter portion was placed in a beaker and they were spiked with 10.0 μL of a surrogate standard solution consisting in 10.0 ng/μL of 13C60 in toluene. Surrogate standard addition was used for internal standard calibration. After 30 min of stabilization, 100 mL of toluene were added to each filter portion and the extraction was carried out using the ultrasonic bath during 15 min. Toluene extracts were collected and the process was repeated twice. The extracts were combined and the total extract was evaporated under soft conditions (bath temperature 50 °C and pressure of 76 mbar) until a volume close to 1 mL. Then the residue was quantitatively transferred into a vial, and it was concentrated to near dryness by means of a gently flow of nitrogen. The residue was spiked with the 13C-internal standards to obtain a final concentration of 0.1 mg/L and reconstituted to 1.0 mL with a mixture of toluene/methanol (1:1). In order to minimize matrix interferences in the mass spectrometer, 0.2 mL of each extract was diluted to 1.0 mL with toluene/methanol (1:1). Internal standard addition was used to assess the possible matrix effects of the samples. Instrumental Analysis. LC-tandem MS analysis was performed in a system consisting of a LC chromatograph Alliance waters 2690 coupled to a Micromass Waters LC triple quadrupole, equipped with an electrospray ionization (ESI) interface. The LC analytical column employed to achieve the separation of the analytes was a Luna 2.5u C18 (2)-HST (50  2.0 mm, particle size 2.5 μm) from Phenomenex (Torrance, California, USA). The mobile phase used for elution of the target compounds consisted of toluene (A) and methanol (B) and was delivered at a flow rate of 0.3 mL/min. The chromatographic elution program started at 50% B and rose to 80% B in 0.5 min; this percentage was maintained for 1 min. Finally, the mobile phase was returned to initial conditions in 1.3 min. Initial conditions were maintained for 7 min and equilibrated for further 10 min. The injection volume was set at 20 μL. Mass spectrometric acquisition was performed in single reaction monitoring (SRM) mode, with the ESI operation in negative ionization conditions (NI). As it has been observed in several works,14,15 C60 and C70 cannot be fragmented using the ESI. Thus, molecular peak signals [C60] 3 and [C70] 3 were used for quantification, whereas oxidized ions [C60O] and [C70O], which are also present in a minor proportion, where used for confirmation purposes. The optimal transitions for functionalized fullerenes MFP, CPTAE, ThPCBM, and PCBM were optimized. Because of the high cone voltages required for the ionization of fullerenes, few product ions could be obtained in addition to [C60] 3 . Hence, transition [Mf720] 3 was proved to be the more intense for most of the cases, and [MfM] transitions were used for confirmation purposes. The whole set of transitions for each analyte, and the optimal relevant parameters values, are summarized on SI Table S2. Source and desolvation temperatures were set to 125 and 250 °C, respectively; the cone gas flow was kept at 22 L/hour, and the capillary energy was set at 3.51 kV. Argon was used for the collision cell at 0.5 bar and the electron multiplier was set at 650 V. Quality Assurance. The validation of the proposed method was looking at the linearity range, intraday and interday precision, method limits of detection and quantification decision limit (CCR) and recovery rates. The validation experiments were performed by spiking blank filters at three concentration levels. Blank filters were fortified

ARTICLE

with 10.0, 300 and 600 μL of a mixture of 1.0 mg/L of each compound, and analyzing six replicates of each one. The spikes were processed and analyzed according to the previously described protocol. According to the 2002/657/EC Decision,23 since no certified reference materials were available for the analytes and matrices of interest, the recovery from fortified blank filters was measured as an alternative to trueness. The fortified blank filters were fortified at three levels of concentration in quintuplicate. In all cases, extraction recoveries ranged from 60.0% to 70.6%. Instrumental blanks, extraction blanks, and full procedural blanks were analyzed alongside the spiked extracts. Instrument blanks are composed of a mixture methanol/toluene (1:1) and were analyzed at the beginning of the run. Since no analyte concentration was present in the instrumental blanks no further actions were taken. Extraction blanks consisted on blank filters analyzed along with the samples. No interference or contamination was revealed by the analysis of this blank. Finally, field and procedural blanks were collected during sampling and were analyzed with the samples after instrumental blanks. Procedurals blanks were at the limit of detection levels, ensuring that no crosscontamination was taking place. SI Figure S2 presents an example of a blank chromatogram, a standard solution, and a real sample. The identification of the target compounds was accomplished by comparing the relevant retention time and the MS/MS signals of the analytes in the matrix to those obtained by standard solutions, analyzed under the same conditions. In accordance with the 2002/657/EC Decision, positive identification was achieved when a retention time agreement was within 2% and when the relative abundance of the selected ion transitions was within a margin of (20%.23 An acceptable chromatographic separation was achieved for the majority of the target analytes. Identification and quantification using four identification points for each analyte, as required by the EC, was achieved by monitoring two transition products for every precursor ion corresponding to each target analyte in the SRM mode. The most intense transition served for quantification purposes, and the second transition monitored was used for confirmation of the relevant analyte. The limits of detection (LODs) for fullerenes were calculated as 3 and 10 times the signal-to-noise (S/N) value for the quantifier ion of each analyte. ILDs ranged from 0.1 to 0.8 pg injected while ILQs ranged from 0.3 pg to 2.6 pg injected. Method limits of detection (MLD) were ranging from 5.4 and 20.9 pg/m3 and method limits of quantification (MLQ) were ranging from 17.9 to 69.5 pg/m3. These limits were studied by analyzing real spiked particulate air filters (volumetric concentrations are presented considering a volume of 550 m3). For optimization purposes, and in order to ensure the stability of the instrumental equipment along the different sequences, an eight point calibration curves, based on peak areas, were constructed using least-squares linear regression analysis, from application of the overall method to toluene:methanol (1:1) solutions containing the standard fullerenes mixture at concentrations ranging from 1 μg/L to 1 mg/L. Quantification was performed by isotope dilution using labeled 13C60 as surrogate internal standard. As it has been exposed in the sample pretreatment and extraction section, all the samples analyzed were fortified with surrogate internal standard. 1337

dx.doi.org/10.1021/es200758m |Environ. Sci. Technol. 2012, 46, 1335–1343

Environmental Science & Technology

ARTICLE

Figure 1. Volumetric concentrations of C60 and C70. None of the four technological fullerenes where detected above the limit of detection.

In addition the control of possible matrix interferences in the electrospray interface was performed using 13C70 internal standard added after sample enrichment. CCR was calculated for each analyte, according to the EU Decision 2002/657/EC, as the sample concentration at which the method is able to discriminate with a statistical certainty of 99% if the analyte is present or not. The overall method repeatability was calculated as intraday and interday precision. Intraday precision is expressed as the average of the relative standard deviation (RSD) of the areas obtained for each analyte after the replicate (n = 6) analyses of three filters spiked at the lowest available level. Intraday precision, or within-day reproducibility, was calculated as the percent relative standard deviation (RSD %) of the six replicate samples analyzed during three consecutive days. Interday precision was expressed as the RSD % between six replicate samples extracted and analyzed within the same day. Quality parameters are summarized in SI Table S3. OC and EC aerosol content was determined using the thermaloptical transmittance in a Sunset Laboratory Carbon Analyzer using the NIOSH temperature protocol.24 PCB and HCH concentrations were determined from half of the same filter used for C60 and C70. Briefly, QMA filters were Soxhlet extracted with dichloromethane after addition of surrogates (PCB 30 and PCB 142). Extracts were rotary evaporated for purification on a column filled with 12 g of anhydrous sodium sulfate over 3 g of 15% deactivated neutral alumina (aluminum oxide 90, activated at 400 °C for 12 h). The elution of the column was made with 5 mL of hexane and 12 mL of hexane/ dichloromethane (1:2, v/v) and collected in two separate fractions. The first fraction selected for chlorinated compound analysis was concentrated to 0.5 mL by vacuum rotary evaporation, transferred to a GC vial with isooctane and evaporated to 100 μL under a nitrogen stream. PCB and HCH separation and quantification were performed by gas chromatography coupled to an electron capture detector (Agilent Technologies) and compounds were quantified by the internal standard procedure. The separation was achieved with a 60 m x 0.25 mm i.d.  0.25 μm HP-5MS capillary column (Agilent Technologies) with splitless injection mode. The oven temperature was programmed

from 90 °C (holding time 2 min) to 190 at 15 °C/min (holding time 1 min) to 203 at 3 °C/min (holding time 5 min), then to 290 at 3 °C/min (holding time 1 min), and finally to 310 at 5 °C/ min keeping the final temperature for 10 min

’ RESULTS AND DISCUSSION The analytical method used in the present study was developed for the analysis of six fullerenes in airborne particulate. This analytical method was based in a previous one14 by our group, but here three more fullerenes were included, and the sample pretreatment and the extraction procedures were adapted to the filters employed by the air sampling. The performance of the method was evaluated through determination of the linearity, sensitivity, repeatability and recovery of the method. Results obtained are summed up in SI Table S3. The method was proved to be suitable for the assessment of selected fullerenes bound to aerosols and it was applied to evaluate 43 samples collected during two consecutive sampling campaigns. The concentration of fullerenes in airborne samples from the Mediterranean Sea is presented in Figure 1. C60 and C70 fullerenes were detected over the LOQ in most of the samples. Both compounds were quantified in 28 samples in concentrations ranging from low ng/m3 to pg/m3. The median concentrations of C60 and C70 fullerenes aerosol phase concentrations were 0.06 ng/m3 and 0.48 ng/m3, respectively, for the Mediterranean Sea atmosphere. C70 fullerene was the more frequent compound and it was present in 36 of the 44 samples analyzed. Details of C60 and C70 fullerenes concentration in the samples are summarized in SI Table S4, and volumetric concentrations of fullerenes are presented in Figure 1. Functionalized fullerenes studied were not detected in any of the sample. This can be explained because the sources of buckyballs to the atmosphere are multiple, as have been exposed before, whereas functionalized fullerenes have their only origin in nanotechnology. In addition the degradation of attached chains will revert in C60 and C70 fullerenes or their oxidized form. These results are consistent with the physicochemical properties of fullerenes, especially buckyballs. C60 and C70 fullerenes are highly nonpolar and hydrophobic compounds. As discussed elsewhere,16 the 1338

dx.doi.org/10.1021/es200758m |Environ. Sci. Technol. 2012, 46, 1335–1343

Environmental Science & Technology applicability of the methods available for hydrophobic compounds partitioning has not been comprehensively assessed for their applicability to predicting the environmental transport and fate of fullerenes. However, due to their low vapor pressures (5  106 tor),25 and using the estimation method of the soot-air partition coefficient suggested by Van Noort et al.,26 for PAHs, we obtain a soot-air partition coefficient of 1014. For the sorption to organic carbon from air, the partition coefficient may be significantly lower (23 orders of magnitude). Still, with soot-air and organic carbon-air partition coefficients of this order of magnitude, more than 99% of fullerenes should be associated to aerosol particles, similarly to what has been observed for PAHs with more than six aromatic rings.27 In any case, there is a need to determine experimentally the fundamental partitioning properties of NP. With this limitations in our knowledge of the environmental partitioning of fullerenes, the aerosol-bound concentrations reported here are a lower estimate of their occurrence in the atmosphere since it does not include the occurrence of fullerenes associated to small aerosols (size 0.96, p < 0.05), as presents SI Figure S3. However, it must be noted that sample S2, corresponding to the southwestern Sea of Sardinia, presented an extreme value which has a strong influence the final correlation. When this value was discarded the correlation presented between the volumetric concentrations of C60 and C70 fullerenes explained a lower fraction of variability (R2 > 0.73), but still significant (p < 0.05), as can be seen in SI Figures S3. Even though there is a significant correlation between the C60 and C70 aerosol phase concentrations (slope C70/C60 = 4.7), the ratio C70/C60 shows an important variability ranging from 1.7 to 179, with a median value of 5.6. This fact indicates that even though there is a covariability of C70 with C60, the proportion between them is variable even though C70 is always higher than C60 fullerene in these samples. The proportions of C60 and C70 were different than those reported in previous works, where C60 fullerene was the predominant compound in wastewater14 and in the sediments from CretaceousTertiary boundary.28 A possible explanation can be found in the difference between matrices and the higher polarity of C70 fullerene in comparison with C60 fullerene, which confer a higher stability to C70 in air.29 The production of fullerenes in benzene sooty flames leaded to C70/C60 fullerenes ratio between 0.26 and 5.7 being the higher values obtained when operating at high pressures.30 Whereas, low pressures conditions as car combustion processes will be linked to ratios inferior to 1. This agrees with the results found in the analysis of aerosol concentrations from the urban atmosphere of Barcelona. The samples from Barcelona atmosphere shows that only C60 fullerene was detected with C70 fullerene concentrations lower than the detection limit. This contrasts with the predominance of C70 fullerene in the open Mediterranean Sea atmosphere and suggests that cities contribute only with C60 fullerene to the atmosphere, and other industrial processes have to be considered as major sources of fullerenes to the environment. This work is the first reporting atmospheric concentrations of fullerenes and no comparison can be done with other works. The ratio of C60 and C70, obtained in this work is consistent with these observations, although more empirical data are needed to

ARTICLE

confirm whether the variability observed in the C70/C60 ratio is due to the occurrence of a diverse group of sources, and/or a dominating combustion source with a C70/C60 ration depending on the combustion conditions, and/or a different persistence of C70 versus C60 during atmospheric transport. Comparisons between fullerenes and other organic contaminants. In general, the predominant chemical components of marine air particulate matter are sulfate, nitrate, ammonium, sea salt, mineral dust, organic compounds, and black or elemental carbon, each typically contributing between 5 and 30% of the overall aerosol mass load. At different locations, times, meteorological conditions, and particle size fractions, however, the relative abundance of different chemical components may vary by an order of magnitude or more. In addition, sea salt content can be high in some samples due to formation of primary marine aerosol.31 Therefore, the concentrations of fullerenes are not necessarily in correlation to the particulate weight or to the ratio particulate/volume, as happened in this study. Instead, organic carbon and soot carbon content usually constitute better tools to understand the mechanisms of sorption of hydrophobic persistent organic pollutants in the atmosphere,27 and below we assess the information they provide when assessing the environmental fate of fullerenes. The total organic carbon (OC) and the soot or elemental carbon content (EC) was determined for some selected samples with different values of fullerene concentrations. These results are presented in the SI Table S5 and OC concentrations ranged from 1.4 to 5.3 μg/m3 with a median of 4.1 μg/m3, while EC concentrations ranged from 0.2 to 1.1 μg/m3 with a median of 0.6 μg/m3. It is not clear with the used analytical method to estimate aerosol OC and EC to which extend C60 and C70 concentrations are estimated as part of OC or part of EC. The concentrations of C70 (the more abundant detected fullerene) account between 0.003% and 0.9% of the OC concentrations, and between 0.13% and 6.3% of the measured EC concentrations. There is no significant correlation between the C60, C70 fullerenes, and C70/C60 fullerenes ratio with neither OC or EC concentrations, or the OC/EC ratio. The lack of correlation between buckyballs and soot carbon is surprising at first sight, but this does not necessarily indicate that fullerenes and soot carbon are not coemitted. In fact, the scenario reported here is consistent with fullerenes and soot carbon emitted together, but with a ratio C60/EC and C70/EC depending on the combustion source. Different combustion processes would emit different proportions of EC, C60, and C70 leading to the lack of correlations between soot carbon and fullerenes. In addition, since the vapor pressure of fullerenes is very low, gas-particle partitioning is not an efficient process affecting the environmental fate of fullerenes. Once emitted, C60 and C70 will stick to the coemitted particles without the potential to repartition to other aerosol particles. Indeed, gas phase C60 and C70 concentrations may be too low for repartitioning processes be relevant for the environmental fate of fullerenes. In addition, if any C60 or C70 nanoparticle is emitted alone or as a NP aggregate to the atmosphere, it will stick immediately to aerosol particles, specially its OC and EC fractions, due to its low vapor pressure and high hydrophobicity. Even though high fullerene content is observed for some of the high aerosol OC samples, there is not a significant correlation between fullerenes and aerosol OC. Similarly, C60 and C70 can not repartition once emitted to the atmosphere to other aerosol particles, and they will stay with the same aerosol population 1339

dx.doi.org/10.1021/es200758m |Environ. Sci. Technol. 2012, 46, 1335–1343

Environmental Science & Technology

ARTICLE

Figure 2. Atmospheric concentrations of Fullerenes, PCBs and HCH in the 2006 campaign. For comparison purposes, only samples with data available for Fullerenes, PCBs and HCH have been represented.

during its atmospheric transport. Of course, the composition of these aerosols can vary during transport due to formation of secondary aerosols or aging processes, but this would not modify the content of fullerenes unless there are relevant degradation processes affecting fullerenes during transport. This behavior contrasts with that of legacy POPs such as polychlorinated biphenyls (PCBs) and hexachlorociclohexanes (HCH). Concentrations of PCBs and HCH in aerosol samples

were also determined for the two sampling cruises (Figure 2). Their occurrence in the aerosol phase is dominated by gasparticle partitioning since most PCBs and HCH are found in the gas phase. Therefore, for these compounds, organic matter usually dominates as an absorptive matrix.32 However, there is no correlation between fullerene concentration and the aerosolbound concentrations of PCBs and HCH. This is consistent with the fact that fate and transport processes of fullerenes, PCBs and 1340

dx.doi.org/10.1021/es200758m |Environ. Sci. Technol. 2012, 46, 1335–1343

Environmental Science & Technology

ARTICLE

Figure 3. Hysplit modeled back-trajectories of the sampled air masses, at the initial and final transect coordenades. Model covers the trajectories at two different heights (15 m AGL, in red; 100 m AGL, in blue) and during the last 48 h before the sampling.

HCH may differ, since the vapor pressure of the former is several orders of magnitude lower than for PCBs and HCH. Gas-particle partitioning processes may not be relevant for fullerenes, and thus their occurrence in aerosols may be related to their use profile, and how they are emitted to the atmosphere. Conversely, gas-particle partitioning of PCBs and HCH is an efficient process and legacy POPs can repartition during atmospheric transport accumulating in those phases with higher fugacity capacity. Back Trajectories. The back trajectories of air masses were studied using the Hybrid Single Particle Lagrangian Integrated Trajectory (HYSPLIT) model.33 The backward-trajectories were calculated for the 48 h previous to the sampling at 15 and 100 m above ground level (AGL) and examples of backward-trajectories are presented in Figure 3. The results of this study showed that air masses which had moved over industrialized areas presented the highest concentrations of fullerenes, for example, S1, S2 S4, S6 and S8 were related to air masses moving over France, Catalonia, Sicilian, Ionian and the Athens Basin. High concentrations of chlorinated POPs were also found in some of these sampling sites (Figure 2). Some of these air masses corresponded to air masses coming from Az-Zawiyah, which is a Libyan region with an important oil refinery industry. The samples from the Black Sea, S12 and S13, also presented high concentrations of fullerenes and the corresponding air masses came from the Zonguldak province having an intense port and coal mining activity.

Most of these results suggest that fullerenes, which are present in the samples, are originated by incidental sources, such as industry, car breaks and other combustion processes. Just one exception was found corresponding to the sample S26. In this case low concentrations were found whereas the corresponding back trajectory indicated that the air mass had displaced directly from the Iberian Peninsula, including along its way industrialized and considerably populated areas of Catalonia, however, sources may be very variable, since as shown above, C70 fullerene could not be detected in aerosol samples from the urban atmosphere of Barcelona. Sample S13 presents a unique case, as it is the only significantly polluted sample in which air simulation offers significant differences between the trajectories at 15 m and at 100 m AGL. At 100 m the trajectory moved relatively distant, through the Black Sea, whereas near the Sea level, the air still came from Zonguldak city as it had done in the sample S12. Because fullerene concentrations of S12 and S13 and low-height trajectories are similar, while they differ in their high height trajectory, it can be argued that the lowest atmospheric layers are responsible for the transport of fullerenes. On the other hand, air masses that had moved over wide sea areas, far from the coast (S14, S18, S31, S32, S34, S37), presented the lowest levels of fullerenes. The surprisingly low concentrations obtained in Alexandria, a highly populated and industrially 1341

dx.doi.org/10.1021/es200758m |Environ. Sci. Technol. 2012, 46, 1335–1343

Environmental Science & Technology thriving area, can be also explained by air masses back-trajectories, as they circulated from sea to land. The results showed that concentrations of fullerenes were closely dependent on the backtrajectories of the air masses reflecting different and variable sources, and also help to explain the substantial differences between the results between sampling campaigns carried out during 2006 and 2007. The obtained results suggest that deposition of anthropogenic fullerenes is a quick process, which is likely to affect the coastal marine zones as well as the continental ecosystems. The dry deposition velocity of aerosols (vd) for the Mediterranean is estimated to be average 0.2 cm/s;20 with these deposition velocities it is possible to estimate the dry deposition flux (FDD, in ng/m2.d) by: F DD ¼ 864  vd  CA Where CA is the volumetric concentration of the chemical in the aerosol phase (ng/m3) and 864 is a unit conversion factor. Then the estimated dry deposition fluxes for the C70 fullerene range between 13 and 40 400 ng/m2d averaging 1500 ng/m2d and a median of 154 ng/m2d. The dry deposition fluxes for the C60 fullerene range between 5.7 and 8500 ng/m2d, with a mean of 474 ng/m2d and a median of 21 ng/m2d. There is no data related to the analysis of fullerenes in marine water, although it seems probable that when deposited to the sea surface, fullerenes would be likely to sink to marine sediment due to their high apolarity and their tendency to absorb into organic matter.34 In order to further understand the behavior, transport and fate of fullerenes, other environmental matrices should be taken into account in future analysis.

’ ASSOCIATED CONTENT

bS

Supporting Information. Additional information including five tables and three figures. This material is available free of charge via the Internet at http://pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected].

’ ACKNOWLEDGMENT The Spanish Ministry of Science and Innovation has funded this work through the projects Nano-Trojan (CTM201124051), CEMAGUA (CGL2007-64551/HID), and it has been partly funded by King Saud University grant number (KSU-VPP105). We gratefully acknowledge the National Oceanic and Atmospheric Administration Air Resources Laboratory (NOAA-ARL) for the provision of the HYSPLIT model, which has been used in this publication. ’ REFERENCES (1) Murr, L. E.; Soto, K. F. A TEM study of soot, carbon nanotubes, and related fullerene nanopolyhedra in common fuel-gas combustion sources. Mater. Charact. 2005, 55 (1), 50–65. (2) Murr, L. E.; Garza, K. M. Natural and anthropogenic environmental nanoparticulates: Their microstructural characterization and respiratory health implications. Atmos. Environ. 2009, 43 (17), 2683– 2692.

ARTICLE

(3) Murr, L. E.; Soto, K. F.; Guerrero, P. A.; Lopez, D. A.; Ramirez, D. A. TEM observations of carbon nanotubes and related nanocrystal aggregates collected from domestic and commercial fuel gas combustion sources. Microsc. Microanal. 2004, 10 (SupplementS02), 410–411. (4) Wiesner, M. R.; Lowry, G. V.; Alvarez, P.; Dionysiou, D.; Biswas, P Assessing the risks of manufactured nanomaterials. Environ. Sci. Technol. 2006, 40, 4336–4345. (5) Navarro, E.; Baun, A.; Behra, R.; Hartmann, N. B.; Filser, J.; Miao, A.-J.; Quigg, A.; Santschi, P. H.; Sigg, L. Environmental behavior and ecotoxicity of engineered nanoparticles to algae, plants, and fungi. Ecotoxicology 2008, 17, 372–386. (6) Lens, M. Use of Fullerenes in Cosmetics. Recent Patents on Biotechnology 2009, 3, 118–123. (7) Akkurt, F.; Kaya, N.; Alicilar, A. Phase transitions, order parameters and threshold voltages in liquid crystal systems doped with disperse orange dye and carbon nanoparticles. Fullerenes, Nanotubes, Carbon Nanostruct. 2009, 17 (6), 616–624. (8) Trojanowicz, M. Analytical applications of carbon nanotubes: a review. TrAC, Trends Anal. Chem. 2006, 25 (5), 480–489. (9) Martín, N. G., F.; Prato, M. Fullerene Polymers: Synthesis, Properties and Applications; Wiley: Hoboken, NJ, 2009. (10) Constantin, C.; Neagu, M.; Ion, R.-M.; Gherghiceanu, M.; Stavaru, C. Fullerene-porphyrin nanostructures in photodynamic therapy. Nanomedicine 2010, 5 (2), 307–317. (11) Partha, R.; Conyers, J. L. Biomedical applications of functionalized fullerene-based nanomaterials. Int. J. Nanomed. 2009, 4, 261–275. (12) Environmental, Health, and Safety Research Needs for Engineered Nanoscale Materials; The National Nanotechnology Initiative, 2006. (13) Nanotechnology White Paper; U.S. Environmental Protection Agency: Washington, DC, 2007. (14) Farre, M.; Perez, S.; Gajda-Schrantz, K.; Osorio, V.; Kantiani, L.; Ginebreda, A.; Barcelo, D. First determination of C60 and C70 fullerenes and N-methylfulleropyrrolidine C60 on the suspended material of wastewater effluents by liquid chromatography hybrid quadrupole linear ion trap tandem mass spectrometry. J. Hydrol. 2010, 383 (12), 44–51. (15) Isaacson, C. W.; Usenko, C. Y.; Tanguay, R. L.; Field, J. A. Quantification of fullerenes by LC/ESI-MS and its application to in vivo toxicity assays. Anal. Chem. 2007, 79 (23), 9091–9097. (16) Isaacson, C. W.; Kleber, M.; Field, J. A. Quantitative analysis of fullerene nanomaterials in environmental systems: A critical review. Environ. Sci. Technol. 2009, 43 (17), 6463–6474. (17) Utsunomiya, S.; Jensen, K. A.; Keeler, G. J.; Ewing, R. C. Uraninite and fullerene in atmospheric particulates. Environ. Sci. Technol. 2002, 36 (23), 4943–4947. (18) Johnson, D. R.; Methner, M. M.; Kennedy, A. J.; Steevens, J. A., Potential for occupational exposure to engineered carbon-based nanomaterials in environmental laboratory studies. Environ. Health Perspect. 118, (1), 49-54. (19) Yeganeh, B.; Kull, C. M.; Hull, M. S.; Marr, L. C. Characterization of airborne particles during production of carbonaceous nanomaterials. Environ. Sci. Technol. 2008, 42 (12), 4600–4606. (20) Roncali, J. Linear p-conjugated systems derivatized with C60fullerene as molecular heterojunctions for organic photovoltaics. Chem. Soc. Rev. 2005, 34, 483–495. (21) Castro-Jimenez, J.; Eisenreich, S. J.; Ghiani, M.; Mariani, G.; Skejo, H.; Umlauf, G.; Wollgast, J.; Zaldívar, J. M.; Berrojalbiz, N.; Reuter, H. I.; Dachs, J. Atmospheric occurrence and deposition of polychlorinated dibenzo-p-dioxins and dibenzofurans (PCDD/Fs) in the open Mediterranean Sea. Environ. Sci. Technol. 2010, 44 (14), 5456–5463. (22) Berrojalbiz, N.; Dachs, Del Vento, S.; Ojeda, M. J.; Valle, M. C.; Castro-Jimenez, J.; Wollgast, M. J.; Hanke, G Persistent organic pollutants in Mediterranean seawater and processes affecting their accumulation in plankton. Environ. Sci. Technol. 2011, 44 (14), 5456–5463. (23) Comission Decision 2002/657/EC, 2002. Commission Decision of 12 August 2002 implementing Council Directive 96/23/EC 1342

dx.doi.org/10.1021/es200758m |Environ. Sci. Technol. 2012, 46, 1335–1343

Environmental Science & Technology

ARTICLE

concerning the performance of analytical methods and the interpretation of results . Official Journal L 221, 17/08/2002, p 00080036 (24) Birch, M. E.; Cary, R. A. Elemental carbon-based method for monitoring occupational exposures to particulate diesel exhaust. Aerosol Science & Technology 25 1996, 221–241. (25) WSES Research. http://sesres.com/PhysicalProperties.asp (accessed Feb. 1, 2011). (26) Van Noort, P. C. M. A thermodynamics based estimation model for sorption of low-polarity organic compounds by high surfsce area carbonaceous materials in environmental sorbents. Environ. Toxicol. Chem. 2003, 22, 1179–1188. (27) Dachs, J.; Eisenreich, S. J. Adsorption onto aerosol soot carbon dominates gas-particle partitioning of polycyclic aromatic hydrocarbons. Environ. Sci. Technol. 2000, 34, 3690–3697. (28) Heymann, D.; Wolbach, W. S., Fullerenes in the CretaceousTertiary boundary. In Natural Fullerenes and Related Structures of Elemental Carbon; Rietmeijer, F. J. M., Ed.; Springer: New York, 2006. (29) Juha, L.; Ehrenberg, B.; Couris, S.; Koudoumas, E.; Leach, S.; Hamplova, V.; Pokorna, Z.; M€ullerova, A.; Kubat, P. Single-photon photolysis of C60,C70,C76, and C84 in solutions. Chem. Phys. Lett. 2001, 335 (56), 539–544. (30) Ullmann’s Encyclopedia of Industrial Chemistry, 6th ed.; WileyVCH Verlag GmbH & Co.: Federal Republic of Germany, 2003; Vol. 1, p 372. (31) G€otz, C. W.; Scheringer, M.; Macleod, M.; Roth, C. M.; Hungerb€uhler, K. Alternative approaches for modeling gas-particle partitioning of semivolatile organic chemiclas: model development and comparison. Environ. Sci. Technol. 2007, 41, 1272–1278. (32) Sicre, M. A.; Marty, J. C.; Saliot, A.; Aparicio, X.; Grimalt, J.; Albaiges, J. Aliphatic and aromatic hydrocarbons in different sized aerosols over the Mediterranean Sea: Occurrence and origin. Atmos. Environ. 1987, 21 (10), 2257–2259. (33) READY (Real-time Environmental Applications and and Display sYstems), by Air Resources Laboratory (ARL). http://www.arl. noaa.gov/ready.html (accessed Oct 12, 2010). (34) Lohmann, R.; Breivik, K.; Dachs, J.; Muir, D. Global fate of POPs: Current and future research directions. Environ. Pollut. 2007, 150, 150–165.

’ NOTE ADDED AFTER ASAP PUBLICATION A portion of the SI was missing in the version published on August 3, 2011. The correct SI published on November 16, 2011.

1343

dx.doi.org/10.1021/es200758m |Environ. Sci. Technol. 2012, 46, 1335–1343