Characterization of Urban Particulate Matter by Diffusive Gradients in Thin Film Technique

A diffusive gradient in thin films (DGT) technique was employed in characterization of the particulate matter related to the urban area suffering from heavy traffic. Kinetics of mobilization metals fluxes from the metal-contaminated particulate matter was investigated. To monitor responses of the particulate matter sample, DGT probes of various thickness of diffusion layer were deployed in aqueous model suspensions of the particulate matter for different time periods. Particulate matter samples and exposed DGT resin gels were decomposed in a mixture of nitric and hydrochloric acid in a microwave pressurized PTFE-lined system. Total content of some traffic-related elements (Cd, Co, Cu, Mo, Ni, Pb, Pd, Pt, Rh, Sb, and V) was determined by inductively coupled plasma mass spectrometry. DGT measurements revealed that two metals pools associated with particles could be recognized, which can be characterized as high soluble fraction and almost insoluble fraction. DGT-measured metal fluxes from the labile pool showed significant difference in mobilization and resupply fluxes of individual selected elements, which might reflect the origin of selected metals and their speciation in particulate matter. The DGT technique can be applied as a useful tool for characterization of metals mobilization from the particulate matter.


Introduction
e particulate matter (PM) has received growing attention since many epidemiological studies demonstrated an association between exposure to particles and adverse human health e ects [1,2]. Particulate pollution encompasses emissions from both natural and man-made sources. e main source of PM in urban areas is road transport in addition to the burning of fossil fuels in power stations and factories.
e particulate matter emitted from on-road motor vehicles includes complex mixtures of metals from tires, brakes, parts wear, engine emissions, and dust from road surfaces. Fine particulate material containing platinum group elements (PGE) is also introduced into the environment due to utilization of three-way catalytic converters [3]. e PM comprises a mixture of numerous components, including various organic substances, elemental carbon, and inorganic substances. e metallic fraction is associated with several adverse health e ects. e content of transition metals in PM is signi cantly a ected by vehicle volume and speed, type of engine and its operation conditions, road type, rush time periods, neighboring environment and meteorological conditions, etc. [4,5]. As urban tra c is continuously increasing, tra c-generated atmospheric pollutant loads will impose an even greater impact on human and ecosystem health. Epidemiological studies pointed out a number of adverse health e ects associated with atmospheric particles originating from transportation [6][7][8].
Conventional analytical metal measurements in atmospheric particles usually entail the determination of total metal concentrations. Nevertheless, it is often necessary to quantify speci c metallic forms, species, since bioavailability, solubility, geochemical transport, and metal cycles largely depend on physical-chemical speciation [9]. Consequently, toxicity of transition metals associated with atmospheric PM depends on solubility [10], speciation, and kinetics of metal release from the solid phase. e use of water as a leaching agent was recommended [9] because these conditions are more similar to environmental conditions of extraction by rain water or to conditions in the lung environment, when PM is inhaled.
In recent decades, batchwise equilibrium-based single [11] or sequential extraction schemes [9] have been consolidated as analytical tools for fractionation analysis to assess the ecotoxicological signi cance of metal ions in solid environmental samples. However, they cannot simply provide information about the kinetics of the leaching process. Moreover, the ratio of the solid phase mass to volume of leaching agent, particularly water, can considerably in uence the yield of extraction. Taking into account that naturally occurring processes take predominantly place under dynamic conditions, recent trends have been focused on the development of alternative methods aimed at mimicking environmental events more correctly than their classical batchwise static operational and sequential extraction counterparts [12].
A comprehensive survey of these developments in assessment of bioaccessible trace metal fractions in airborne particulate matter is recently reported in a review [13]. Dynamic extractions procedures provide a continuous ow of the leaching agent through the sample; thereby, soluble species are dissolved and subsequently detected o ine or online providing information about the kinetics of the leaching process. e risk of sample losses as well as contamination is reduced since troublesome and time-consuming sample handling steps of batchwise procedures such as sample shaking, ltration, or centrifugation can be avoided. In literature, dynamic extraction procedures based on the concepts of stirring chambers, rotating coiled columns, packed microcartridges, and ow-through microdialysis are used mostly for the analysis of samples like y ash, rock, soil, and sediments [12,14]. Application of these procedures to fractionation studies (e.g., [15]) is limited because airborne particulate matter, unlike soil, sediments, and y ash, is only available with very limited sample amounts collected on lters, typically at submilligram to milligram level [11]. To overcome problems associated with this special kind of sample, various approaches were introduced [16][17][18]. Instead of whole sample only aliquots of the sample in the form of either circular parts or punches of the sampling lter substrate with homogeneously distributed particulate matter were used, which enables loading of sample into the extraction unit.
Another group of methods for determination of the water-soluble fraction of metals in atmospheric aerosols can be based on the continuous sampling of atmospheric aerosols into deionized water [10,19]. By using the aerosol growth technology, saturated water vapor condenses on the PM forming droplets, which can be collected in various impactors. e sampled particles are leached in interaction with condensed water in time, so that the soluble fraction of elements can be subsequently determined after separation of insoluble part of particles by suitable high-sensitive detection technique.
A new approach, the di usive gradients in thin lm (DGT) technique [20], has been developed to assess the bioavailability of metals in environmental systems such as sediments and soils [21]. DGT technique can measure the uxes of labile species in water, soil, and sediment compartments in situ under undisturbed conditions and without pretreatment. In the uptake process, metals di use from pore water through the di usion layer to the binding phase, typically Chelex 100 resin gel [20,22], which is selective to transition metals. As in an organism uptake, the DGT uptake reduces metal concentration in the aqueous phase in the vicinity of the DGT probe, so that it induces resupply of metals associated with the solid phase.
us, DGT can measure the metal fraction that is accessible from the complete metal pool in the solid phase system [21,23].
Interpretation of DGT-induced uxes indicates the degree of depletion of the solution metal concentration in the vicinity of the DGT probe (as determined by the thickness of the di usive layer), the kinetics of desorption of metals, and the extent of the pool(s) of labile metal species in the particulate phase [24,25]. e scope of this study was to investigate the kinetics of metals resupplied from an urban particulate matter into the aqueous phase by employing the DGT technique. is approach represents a useful tool in estimation of the watersoluble, mobile fraction of metals in particulate matter. Until now, to the best of our knowledge, no results have been reported concerning utilization of the DGT technique for characterization of atmospheric particulate matter in direct contact of the probe with the moisturized PM, providing information on the metal contamination and on the kinetics of metal release.

Particulate Matter Samples.
Within a period of years 2006-2016, the total suspended particulate matter (TSPM) was sampled in annual service intervals from lters of an airconditioning unit in the building of the Institute of Analytical Chemistry of the CAS, v.v.i., in Brno, which is situated at the heavy tra c area on the Veveri street, Brno, Czech Republic, GPS location: 49°12′27.958″N, 16°35′27.845″E. In this study, the sample collected within the period of March 2008 to February 2009 was characterized in more detail. Concentration of the PM 2.5 fraction was found [26] within the range of 20-32 µg·m −3 for this sampling site.

Preparation of DGT Probes.
e polyacrylamide hydrogels used in this study were prepared according to the procedure developed by Zhang and Davison [20]. Gel solution contained 15% by volume of acrylamide (Sigma-Aldrich, Germany) and 0.3% by volume of patented agarose crosslinker (DGT Research, Ltd., Lancaster, United Kingdom). N,N, N′,N′-Tetramethylethylenediamine (TEMED; Sigma-Aldrich) was used as catalyst. A freshly prepared solution of ammonium peroxydisulfate (10%, Sigma-Aldrich) was used as an initiator for polymerization. For preparation of polyacrylamide di usive gels (APA), 10 μL TEMED, and 28 μL ammonium peroxydisulfate solutions were added to 4 mL gel solution. is solution was immediately cast between two glass plates separated alternatively by a plastic spacer of 0.3, 0.5, 1.0, 1.5, and 2.0 mm thickness, respectively, and allowed to set in an oven at 42-46°C for 60 min. e gel was then removed from glass plates and hydrated in ultrapure water at least for 24 h before use. Ultrapure water was exchanged three times during hydration period. Finally, the gels exhibited a stable thickness of 0.45, 0.80, 1.6, 2.35, and 3.2 mm (tolerance ± 0.01 mm), respectively, according to spacers used. Gel discs of 25 mm in diameter were cut from hydrated gel strips using a round plastic knife and then stored in 0.01 mol·L −1 sodium nitrate solutions at room temperature.
e resin gel consisted of 0.5 g of Chelex 100 ionexchange resin (Na form, 100-200 wet mesh; Bio-Rad Laboratories, USA) in 2.5 mL of the gel solution. After one-day swelling of the resin in the gel solution, 3.75 μL TEMED and 15 μL ammonium peroxydisulfate solutions were added. Contrary to preparation of the di usive gel, a half amount of the initiator was added to the mixture to prolong the setting time and consequently to allow settling of the resin beads on one side of the gel. Discs of resin gels were cut by the same way as di usive gels and then stored in ultrapure water at 4°C. ey reached a thickness of 0.44 ± 0.01 mm.
e DGT sampling probes were assembled by inserting the resin gel inside the piston-type DGT probe (R-SDU, DGT Research Ltd., Lancaster, UK) with resin beads oriented towards the sampler exposition window. Afterwards, the resin gel was covered with the di usive gel and a wet protective membrane lter (0.45 μm in pore size, 25 mm in diameter, and 0.13 mm in thickness (Supor ® -450, Pall Corporation, Michigan, USA)), both components representing the overall di usive layer thickness (∆g). Finally, the DGT probes were carefully closed. Blanks of resin gels were treated in the same way including probe preparation.

Deployment Experiments.
e sample of PM taken from the textile lter of the air conditioner was gently stirred in order to prepare su cient amount of homogeneous representative model sample for comparative experiments. A mixture of 20 g PM and 25 g water was equilibrated at laboratory temperature of 23°C for 24 h. e set of three DGT probes packed with polyacrylamide di usive and Chelex 100 resin gels were gently immersed into the surface of the PM suspensions, making sure that there were no air bubbles between the PM slurry and the exposition window of the DGT probe. e probes with di usive gel thicknesses of 1.6 mm were deployed at constant room temperature of 23°C for various time periods (short term up to 24 hours, long term up to 60 days). DGT devices were then retrieved from the PM suspension and rinsed with ultrapure water to remove all the particles adhered on the lter membrane. Analogously, the probes with di usive gels of di erent thicknesses (0.45, 0.8, 1.6, 2.35, and 3.2 mm, resp.), that is, DGT probes with speci c decreasing demand, were also used in a series of experiments in duration of 8 h of deployment to measure responses of PM.

Apparatus and Sample Analysis.
In the analysis of the PM sample, pore water solution was collected from experimental suspensions by centrifugation in plastic tubes for 30 min. e supernatant was ltered through a 0.45 µm membrane lter Supor-450 to remove colloidal components, then stabilized by addition of nitric acid, and stored at 4°C before analysis.
Prior ICP MS analysis, resin gels (0.24 g) and samples of PM (0.1 g) were digested in a mixture of concentrated subboil grade acids consisting of 2 mL·HNO 3 and 0.5 mL·HCl. e samples were treated in precleaned PTFE (110 mL) reaction vessels by using focused microwave (MW) energy in one-stand closed high-pressure autoclave unit (UniClever, Plazmatronika, Wroclaw, Poland). In the rst predigestion step, the sample aliquot was left to react at laboratory temperature for 5 min. Subsequently, it was heated in three steps, each in duration of 5 min, applying 70, 90, and 100 percent of microwave power (150 W) controlling working pressure within limits of 3.5/3.2, 4.0/3.5, and 4.5/4.2 MPa, respectively. After cooling down period of 10 min, the digests were diluted with ultrapure water to the nal mass of 10 g. To achieve complete decomposition of PM samples inclusive silicates, 0.5 mL HF (Suprapur grade, Merck, Darmstadt, Germany) was also added to the mixture of subboil grade nitric and hydrochloric acids. Blank samples of acids (n � 4) and resin gels (n � 6) were processed analogously.
Ultrapure water, prepared by Ultra Clear system (SB Barsbüttel, Germany), was exclusively used throughout the work. e subboil grade HNO 3 and HCl were obtained by purifying concentrated analytical reagent grade acids (Penta, Prague, Czech Republic) in the subboil quartz distillation system model MSBQ 2 (Maasen, Eningen, Germany).
Inductively coupled plasma mass spectrometry was applied as a multielement, high-sensitive technique in the determination of metals in sample solutions. An Agilent 7700 Series ICP MS with MicroMist Nebulizer was employed under operating conditions summarized in Table 1. e Octopole Reaction System (ORS) of the 7700 ICP MS was operated in helium collision mode (He mode) to exclude potential isobaric interferences according to the US EPA Method 6020A, validated by multilaboratory studies [27]. Possible spectral overlaps were checked by matrix interference and recovery tests [28]. For these tests, a series of synthetic-matrix samples and standards were prepared from the certi ed single-element standard stock solution Astasol ® (Analytika, Prague, Czech Republic) to obtain background equivalent concentrations of elements, especially for noble metals. e analysis of real PM sample solutions revealed that the matrix elements were present in solutions below the critical interfering concentrations to cause spectral overlaps when using the He mode of the collision cell. Potential overlaps on 103 Rh + by 87 Sr 16  Hf 16 O + on 195 Pt + were not found. When applicable, ame and electrothermal atomic absorption spectrometry was also applied to verify the results. In these comparative measurements, the bias of results was below 10% of relative standard deviation of values found by ICP MS. e multielement calibration standards for multipoint (level) calibration were prepared by mixing of certi ed Astasol standards. For fast screening of the digested PM, the semiquantitative analysis mode of ICP MS was applied. In Journal of Analytical Methods in Chemistry this instance, a one-point calibration was used, in which a single multielement calibration standard (1 µg·L −1 Li, Mg, Co, Ce, Y, Tl + 10 µg·L −1 As) was introduced before sample analysis to update the response factors. e level of overall blanks related to analysis of PM samples and to analysis of DGT resin disks was negligible. Except platinum group metals, the blank level was 4 to 5 orders of magnitude lower than that of sample solutions, so that it could be omitted in evaluation of results.

Characterization of the PM Sample.
Particulate air pollution constitutes a complex mixture of particles, present in the atmosphere as solids or liquids that vary in mass, number, size, shape, surface area, chemical composition as well as reactivity, acidity, solubility, and origin [2].
Content of the inorganic fraction in the PM sample was assessed by charring approximately 1 g aliquots of the dry PM in ambient air atmosphere in three steps at 200°C for 1 h, 500°C for 3 h, and 900°C for 5 h. In this instance, residual ash represented 57.9 ± 0.5% (n � 3) of the original sample mass. e organic fraction of the PM contained probably some parts of plants, pollen species, common organic compounds as saccharides, organic acids, PAH, and high -molecularweight alkanes [29].
Preliminary examination of the PM sample was performed by using a semiquantitative mode of ICP MS in which content of up to 80 elements could be estimated. e main components of the PM sample were as follows: Al (7.5%), Fe (4.2%), K (1.4%), Mg (0,75%), Na (0.59%), Zn (0.40%), Ti (0.25%), Ba (0.08%), and Mn (0.06%). e metallic elements Ca, Mg, Al, Si, Fe, and Mn are typical markers for soil and resuspended dust sources in the atmosphere [30]. Crustal and other elements were obviously the major contributors to composition of the PM. e results of conventional ICP MS determination of the total content of selected minor, tra c-related elements (Cd, Co, Cu, Mo, Ni, Pb, Pd, Pt, Rh, Sb, and V) are summarized in Table 2. Platinum group elements were found at similar concentration ratio as in German urban areas [3].
In natural media, metal contaminants undergo reactions with water, ligands dissolved in it, and with surface sites of the solid material. e metal partitioning is usually characterized by distribution coe cient, which is the ratio of adsorbed metal concentration (mg·kg −1 ) to the dissolved metal concentration (mg·L −1 ) at equilibrium [31]. In this work, the mobilization release of metals into water under static conditions is described by the reciprocal value, that is, by elution distribution coe cient K d , which re ects the metal distribution between the aqueous phase and the PM during leaching of samples with water. e K d values were evaluated (Table 2) from the pore water metal concentration (mixture of 20 g PM and 25 g water in equilibrium, phase ratio r∼1.25) and the content of metal in the PM (solid phase). e values di er among individual tra c-related metals, decreasing exactly in the following order: ese di erences might re ect the origin and the speciation of individual selected metals.

DGT Experiments.
e basic theory of DGT technique assumes that transport in both the di usion layer of the thickness ∆g (the thickness of polyacrylamide gel and lter membrane) and in the pore water of the slurry of PM was solely driven by molecular di usion and that all labile metal  Journal of Analytical Methods in Chemistry species in the pore water had a single self-di usion coe cient D that is related to free metal ions. If the binding of a metal to the resin is strong and fast in relation to the transport rate of metal species to the resin layer by diffusion, the average ux F, related to the unit area of the resin during the deployment, is given by the following equation [20,22]: where M is the mass of the metal trapped (metal uptake) in the resin (mol), A is the area of the DGT exposition window (cm −2 ) and t is the deployment time (s), D is the di usion coe cient (cm 2 ·s −1 ), ∆g is the thickness of di usion layer (cm), and c (mol·L −1 ) is the metal concentration in the pore water solution at the sampling window of the probe. is equation can be applied in evaluation and interpretation of DGT-measured data, in assessment of kinetic phenomena of metal release from the PM. e plots of DGT-measured uptake and ux of lead versus deployment time (Figures 1(a)-1(c)) are presented as examples of results of DGT experiments. Similar results were obtained also for other monitored metals listed in Table 2.
e DGT uptake sharply decreases during a short deployment period of time. After few hours of deployment, the resupply rate from the solid phase to the solution is evidently too slow to sustain metal concentration in the aqueous phase, that is, to follow the demand of the DGT probe. Consequently, the metal ux into the DGT probe asymptotically approaches zero (Figure 1(c)). Based on the overall concentration of a metal in the aqueous phase, uptake yields below and close to only one percent of the nominal value were recorded by the DGTprobes. is observation documents that di usive processes from the PM slurry are predominantly responsible for the resupply of metals into the solution adjacent the DGT probe [32]. is phenomenon was also re ected by the e ect of the demand of the DGT probe, controlled by the di usive gel thickness ∆g (Figure 1(d)), even for a very thick di usive layer thickness of 3.33 mm. Initial DGT-measured resupply uxes of Cd, Co, Cu, Mo, Ni, Pb, Pd, Pt, Rh, Sb, and V reached values of 3.7, 8, 160, 2.7, 27, 4, 1.1 ×10 −2 , 1 ×10 −5 , 1.1 ×10 −3 , 8.4, and 1.9 nmol·cm −2 ·day −1 , respectively.
Generally, two metal pools associated with particles could be recognized, that is, a labile pool characterized by weak interactions (high solubility fraction) and an immobile, inert pool (insoluble fraction) incorporated strongly in the solid phase of the PM. Similar dissolution behavior was also observed for some elements in natural and anthropogenic particulate matter samples [15]. e general consensus is that ionic, water-soluble metal forms are the most bioavailable species, having the potential to access cells and organs of biota. Solubility and mobility (mobile metal fraction) of individual tra c-related metals are presented in Table 2. As documented by these data and Note. Solubility is represented by the water-soluble fraction (n � 5) of the total metal content after 1-day extraction (r � 1.25 mL·g −1 ), K d is the average (n � 10) distribution coe cient for aqueous extraction, the mobile metal fraction is the percentage of the C DGT -measured concentration from the overall metal concentration in the aqueous phase for the lowest demand of the probe (∆g � 3.33 mm, 8 h, n � 3) applied in the DGT experiment, and change of DGTmeasured resupply rate is the percentage decrease of the metal ux per day within one-day short-term experiment (∆g �1.73 mm, n � 5). n: number of replicates in evaluation of data from series of experiments; a limits of detection (LOD) of 11 ng·L −1 Pd, 0.14 ng·L −1 Pt, and 0.05 ng·L −1 Rh for determination of these metals in aqueous media (supernatant, digest) and 0.56 ng·Pd, 0.002 ng·Pt, and 0.001 ng·Rh in DGT resin gel disks were achieved, respectively; b n.d.: not determined.
results of DGT experiments (Figure 1), the soluble fraction of the metal pool is released within a very short time period, that is, within preparation of aqueous PM suspension and its equilibration period of one day. It is interesting to note that overall solubility of metals di ers among elements of interest. It exceeded 30% of the total metal content of Cd and Sb, whereas solubility of other elements did not reach 10%. e residual fraction of the metal pool was almost insoluble. e kinetics of the subsequent release of metals was very slow even within the period of several months.
Particulate matter is a complex, heterogeneous mixture. It encompasses many di erent chemical components and physical characteristics. Many of them have been cited as potential contributors to toxicity [33]. Each component might have multiple sources. erefore, identi cation and quanti cation of speci c components or source-related mixtures represent one of the most challenging areas of environmental health research. e main source of PM in urban areas is the road transport [2]. Consequently, the metals of interest are divided into the source categories. It is evident from the comparison of the data in Table 2 that low content and minor soluble fraction do not represent low mobility, as documented for elements Co, Pd, and Rh. In this instance, mobility of elements can be expressed by the DGTmeasured mobile fraction of the leachate, presented in Table 2, within the rst stage of deployment experiments, for example, by the slope of a tangent in Figure 1(d). is mobile fraction reaches approximately 31, 26, and 54% of the total soluble fraction of Co, Pd, and Rh in aqueous leachate, respectively.
Resupply uxes can be estimated from the short-term DGT experiments, from relation of calculated metal ux into the DGT probe and duration of deployment period. Calculated metal uxes should be normalized [20] by the metal pore water concentration, as shown in Figure 2, in order to make comparison of these relations possible. Quasilinear relationship was found between the ux and the deployment time for most of the metals investigated. is decrease of the normalized ux (Table 2), expressed by the slope of the linear function, re ects depletion of a pool capacity of the mobile metal fraction with duration of deployment. e capability of resupplying metal is low when the slope is high. e slope of these lines decreased in the following order: Co, Ni, V, Rh, Cd, Cu, Mo, Pb, Pd, and Sb. In addition, the level of the normalized ux in Figure 2 represents the mobile fraction of the metal pool. In this graph, it is re ected by the virtual intercept of the line with ordinate axis. ese plots can provide very interesting characteristics of PM with respect to behavior of individual elements in interaction with aqueous phase. Signi cant di erences can be found among the elements investigated. For example, when comparing results presented in Figure 2 with the data in Table 2, they show for Sb total content of 104 µg·g −1 Sb, relatively high solubility (32.8%), very low mobility of soluble species (1.8%), and relatively wellbalanced resupply kinetics (bu ering capacity) both latter assessed by DGT experiments. On the other hand, results for Pb exhibit higher total content of 411 µg·g −1 ·Pb, low solubility (2.3%), higher mobility (5.7%), and slower resupply kinetics.
Similar results were also obtained in series of experiments for other PM samples taken within a period of years 2006-2016 on the same sampling site. ey are not presented in this communication, because no signi cant trends or changes in these data have been observed.

Conclusion
e results of aqueous leaching and simultaneous DGT experiments showed that DGT technique can be successfully applied for characterization of total suspended particulate matter taken in urban area, for characterization of transition metal soluble fractions and their dissolution kinetics. DGT probes deployed in the moisturized PM disturb metal concentration close to the sampling window of the probe, so that the DGT-measured data can monitor for the response of the PM, that is, for the kinetics of metals release from the solid phase. e study of mobilization of important metals related to tra c, to various car components and fuel, showed considerable di erences in metal fractionation, that is, in their content and solubility, in mobility of soluble species. DGT resupply experiments revealed that most metals in the particulate matter from Brno city were present in two fractions, in a very soluble mobile one and in an almost insoluble immobile one. is observation is in contrast to the typical behavior of soils, which exhibit more balanced resupply uxes, that is, higher bu ering metal pools for deployment time periods of several days [34,35].
In this work, homogenized PM, representing annual accumulation on large textile ber lters in air-conditioning system, were taken for pilot model DGT experiments on particulate slurries. Nevertheless, based on these results, the DGT technique can be successfully applied in future in a modi ed arrangement for measurements with routinely used lters, for example, nitrate cellulose disks of the relevant diameter, for monitoring purposes, for assessment of soluble, mobile fractions of toxic elements in aerosols of interest.

Conflicts of Interest
e authors declare that there are no con icts of interest regarding the publication of this paper.