TY - JOUR
AU - Mamolini,, Giuseppe
AB - Abstract The aim of this work was to offer a state-of-the-art critical survey for characterizing airborne nano- and microparticles by means of electron microscopy (EM) techniques and to highlight advantages and limits of different possible operation modes. Procedures of collection and sample preparation are revisited and improved to analyse airborne particles deposited on filtering membranes by using various sampling methods. Three kinds of electron microscopes are used to this end: scanning electron microscope (SEM), field emission scanning electron microscope (FE-SEM) and transmission electron microscope (TEM). Following and extending previous studies, we optimized procedures by varying both the sample collection/preparation and the operational parameters of the microscopes. In particular, we diversified the sampling methods applied, using ad hoc filters as well as common filters for standard gravimetric measures. This approach enabled us to achieve a simple and clean procedure allowing direct SEM or TEM observation of the collected particulate matter. PM, nanoparticles, FE-SEM, TEM, PC filters, INCA-Feature software Introduction Air pollution due to nano- and microparticles is of increasing interest to both the scientific community and the general public. In this context, the acronym PM (particulate matter) is used to denote solid and liquid particles (organic or inorganic) suspended in air (WHO). European Community (EC) directives dictate European Countries the introduction of laws setting air quality standards for PM10 and PM2.5, while the PM1 concentrations are at a preliminary stage of regulation. The acronyms PM10, PM2.5 and PM1 refer to particles of less than 10, 2.5 and 1 μm, respectively, in aerodynamic diameter, which is defined as the diameter of a sphere of unit density (1 g cm−3) which has the same terminal falling speed in air as the particle of interest, at the same conditions of temperature, pressure and relative humidity. The quality standard for PM10 is referred to the maximum annual mean concentration, that is, 40 µg m−3, linked with the limit of 35 days in a year exceeding a daily mean concentration of 50 µg m−3 (European Standard EN 12341; for details on the reference method for sampling and measuring PM10, see [1]). Morphology and chemical composition are not referred to; however, it is becoming clear that these parameters are of great importance to the human health, particularly with regard to ultrafine particles or nanoparticles (<0.1 µm). Epidemiological studies of Qian et al. [2] point out that both morbidity and mortality increase markedly in the presence of micro- and nanoparticles. The nexus between exposure of some categories of workers to aerosol containing fine and ultrafine dusts and the onset of respiratory diseases is well established [3]. Moreover, Lee et al. [4] have reported that many cases of cancer have been detected in workers exposed to prolonged inhalation of specific kinds of nanoparticles dispersed in the working environment. On the whole, the diseases involved are asthma, emphysema, bronchitis, pneumoconiosis and lung cancer as well as cardiovascular diseases. In particular, PM10 can reach bronchial tubes and alveoli, PM2.5 can reach arteries, and nanoparticles are able to pass through alveolar cells and reach the whole body, and for this reason they are more harmful [5]. Particle composition and diameter could therefore play a major role in the pathogenesis: the need to physically and chemically quantify and characterize the particles responsible for environment pollution in heavily anthropized areas is by now quite obvious. Recent studies focused on airborne nanoparticles have been made using various analytical methods such as inductively coupled plasma atomic emission spectroscopy [6,7], real-time single-particle mass spectrometer [8] and aerosol time-of-flight mass spectrometer [9]. A large number of samples collected in various environments is currently available in Italy at Italian Public Organizations, e.g. Istituto Superiore di Sanità [10] and Agenzie Regionali per la Protezione dell'Ambiente. The standard massive analyses commonly performed, as imposed by law, however, only provide information on the PM weight content per volume unit and possibly on the mean chemical composition of the collected particulate. On the other hand, the determination of the numerical distribution by particle size and the morphological and chemical characterization of single particles, although extremely important, in particular, for nanoparticles, are so far lacking. In a recent extensive review, Kumar et al. [11] demonstrate that particle number concentration is a fundamental indicator for dealing with the impact of the finest atmospheric particulate on the human health: submicrometric particles constitute indeed the major contribution to the total number concentration of PM. In this context, electron microscopy (EM), coupled with energy-dispersive X-ray spectrometry (EDS), can play a fundamental role and begins to be recognized as a powerful tool for counting particles in dimensional classes and for characterizing individual particles. Compared with bulk analysis [12], single-particle analysis [13,14] offers distinctive advantages, as studying shape and chemical structure, describing variation of composition with size, detailing clusters and surface layers. Many recent studies [15–17] supplement mass and bulk chemical measurements with scanning electron microscope (SEM)–EDS analysis to obtain information on the morphology, composition, number and volume-size distribution of atmospheric aerosols. More detail on the crystalline structure of individual particles can be obtained by transmission electron microscope (TEM) analysis [18]. Most of the studies employing EM techniques are until now focussed on coarse (<PM10) or fine (<PM1) PM, but SEM and TEM analyses are becoming more and more important for nanoparticulate speciation studies too and have been recently coupled to in vitro biological tests to assess nanoparticle cytotoxicity in indoor and outdoor environments [19]. Meanwhile, a new discipline is developing, ‘Nanoparticle aerosol science and technology (NAST)’, about formation, properties and behavior of nanoparticles in gases [20]. This work deals with the development of procedures for PM collection, sample preparation and data acquisition, to characterize airborne micro- and nanoparticles by means of SEM and TEM. The state of the art of these methods is reviewed and new ideas are suggested; at the same time, opportunities and limits connected with each different operational choices are pointed out. Materials and methods Sample collection In order to collect airborne particles, we used two samplers: TCR-TECORA and Gilian AIRCON-2, both certified according to EN12341 regulation. The instruments can be equipped with a preselector which, if necessary, can trap larger particles, allowing the sampling filter to collect only a particular PMx fraction. For PM10 sampling (with TCR-TECORA), we used two standard kinds of filters: quartz fibre filters (Whatman Ø 47 mm – pore size 2.0 µm); polytetrafluoroethylene (PTFE) filters (Albet® Ø 47 mm – pore size 2.0 µm). However, for total suspended particulate (TSP) sampling (with Gilian AIRCON-2), we tested ad hoc polycarbonate (PC) filters: Millipore ISOPORE Ø 23 mm – pore size 0.8 µm; Millipore ISOPORE Ø 23 mm – pore size 0.2 µm. PM10 sampling complies with EN 12341 regulation, while TSP sampling allowed us to determine filter-type choice, air volume to be sampled and suitable procedures to quantify and identify airborne matter using SEM analysis directly on PC filters. Some sampling was run by first sticking a copper mesh on the surface of the PC filter, in order to examine the mesh directly with TEM, following an idea already implemented in the past by Dye et al. [21]. Sample preparation Depending on the particular method and filter type used during the sample collection, different preparation procedures are needed prior to any EM observation. The sample preparation is a very critical step, since the final specimen must retain the characteristics of the particle population originally present in the collected air. The ideal filter for SEM–EDS analysis, according to EPA guidelines [22], should have these features: flat surface providing uniform contrast; minimal noise with regard to the EDS spectra of the specimen; thermal stability (insensitivity to localized heating due to the electron beam) and compatibility with high vacuum; minimal electrostatic charging; mechanical strength. Procedure A: filters from standard sampling For standard sampling of PM10 (following European Standard EN 12341), a considerable amount of deposited particulate is required to minimize weighting errors. The loading of the filters is therefore high and direct SEM observation may not have enough resolution to discriminate single particles. However, local and international regulations impose massive sampling and analysis of PM10 and PM2.5 for environmental monitoring, therefore these kinds of samples are easily available. In order to reutilize them also for FE-SEM and TEM examinations, we developed a suitable operational procedure. The steps we adopted are analogous to the well-established procedure employed for catalyst nanoparticles and asbestos fibers, that is, we transferred an aliquot of a particulate on a PC filter, where particle identification is easier. Step 1 The samples were collected according to the European Standard EN 12341, using TCR-Tecora samplers with sampling heads PM10/PM2.5 and PTFE or quartz fiber filters. Step 2 In order to extract the PM from the filter: a little part of the filter was carefully cut paying attention to avoid contamination and to choose a representative sub-sample from the original; the part was immersed in a known volume of dispersing liquid. We checked the performance of water, isopropyl alcohol, ethyl alcohol and hexane and found that water causes dissolution of the soluble components and so their loss; conversely, hexane does not detach most of the particulate. Eventually the solvent of choice was 95% ethanol [16]; the system was sonicated (100 W, 50 kHz) to obtain particulate detachment until a homogeneous suspension is obtained; 15 min is a proper time interval. Step 3 An aliquot of the suspension obtained in step 2 is percolated through a PC filter (0.2 µm pore size) in such a way that the whole particulate is evenly dispersed on the polycarbonate surface. However, if the above-mentioned suspension is still too concentrated, previous dilution is needed. The amount of filtered suspension must provide a light loading on the PC filter (the filter takes on a pale grey colour): this way, the particles will be easily identifiable and can be chemically and morphologically characterized by EM. Step 4 The PC filters prepared in step 3 have to be treated in different ways according to the instrument we want to employ (SEM and FE-SEM or TEM). PC filters are cut in half with a scalpel: one half is attached to a sample stub with double-sided tape, coated with carbon and analyzed with SEM or FE-SEM; the other half is dissolved on a mesh which retains the particles for TEM observation. Filter dissolution is done on a Cu lacey-carbon mesh in a chloroform thermostatic bath at 55°C for 12 h, according to the well-known EPA procedure [23] and other guidelines of Spurny [24]: the procedure involves a mesh placed on a suitable framework coated with a filtering paper; the PC filter is then put over the mesh. The framework is a steel mesh or a folded Teflon foil. It is put in a weighing bottle containing CHCl3 (the mesh must not be covered by the solvent) and inserted into the thermostatic bath. The filtering paper imbues with chloroform and the high temperature speeds up the CHCl3 evaporation and the polycarbonate dissolution, causing deposition of the particles on the mesh. After a proper time interval (12 h are more than enough), PC is dissolved and it is possible to withdraw the mesh which is ready to be analyzed with TEM. On the basis of our tests, we can conclude that PC filters used in Step 3 are ideal for SEM analysis of the airborne particulate owing to their smooth surface on which particles are well distinguished. Moreover, their very good solubility in chloroform makes them the best answer for particulate transfer on the TEM mesh. Procedure B: filters from innovative sampling In order to minimize the number of steps required in the preparation procedure, we also examined other possible options based on a new sampling method. In this case, the sample can be directly observed; one can mount it on a sample stub and lightly coat the specimen with Au or more commonly carbon (carbon is mandatory for SEM–EDS analysis) for SEM or FE-SEM examination. We screened various kinds of filters (of different material and pore size) and sampling procedures (sampled air volume, type of particulate subfraction, sampling time) to obtain samples as representative as possible of airborne micro- and nanoparticle distribution. The samplers employed were Gilian AIRCON-2. The results so obtained are discussed in the following section. Instrumentation The samples were analyzed with the following microscopes: Transmission Electron Analytic Microscope Jeol 2010 200 kV equipped with X-EDS analytical system: Oxford Link with a Pentafet Si-Li. Scanning Electron Microscope Leo Stereoscan440, equipped with X-EDS Oxford Link System, with a Si–Li detector and back-scattering detector. Field Emission Scanning Electron Microscope Zeiss Supra 40 VP equipped with back-scattering detector, X-EDS analytical system: Oxford Link with a Pentafet Si-Li and INCA-Feature software. During the preparation of the samples, the following instruments were used: Vacuum oven (Bicasa Vuotomatic 50). High vacuum evaporator (Polaron Range High Vacuum Evaporator). Thermostatic bath (P.M. Tamsonn V). Results and discussion Evaluation of the collection efficiency of various sampling modes During the sampling procedure we can choose to collect TSP or a particulate fraction (PM10, PM2,5). The sampling mode could affect the dimensional distribution of the submicrometric particles collected on the filter: in order to investigate this possibility we analyzed some PC filters (0.2 µm pore size) coming from TSP, PM10 and PM2,5 samplings in the same collection site (the harbor area of Genoa, Italy). We used the INCA-Feature software installed on the FE-SEM in SE signal mode at high magnification (50 000×): in the automated mode the software counts the collected particles relying on a grayscale threshold that allows it to discriminate between a particle and the filter surface. The results are shown in Fig. 1, where each curve refers to a specific sampling mode. The number percentage of particles per liter of sampled air is plotted as a function of their mean Feret diameter, so the graph allows us to compare the relative abundance of the various dimensional classes for the three sampling modes. Fig. 1. Open in new tabDownload slide Comparison of the dimensional distributions (% of particle number per liter) in particulate matter collected by different sampling modes (TSP, PM10, PM2,5). Results obtained by FE-SEM analysis, SE mode signal, magnification 50 000×, INCA-Feature software elaboration. Fig. 1. Open in new tabDownload slide Comparison of the dimensional distributions (% of particle number per liter) in particulate matter collected by different sampling modes (TSP, PM10, PM2,5). Results obtained by FE-SEM analysis, SE mode signal, magnification 50 000×, INCA-Feature software elaboration. It can be seen that the sampling mode (TSP, PM10 or PM2,5) does not affect seriously the resulting numerical distribution of submicrometric particles; some fluctuation can be due to possible variations depending on the specific sample examined and on the particular sampling moment. Moreover, the graph shows clearly that the relative abundance (percentage of the total number) of the coarse particulate fraction is in any case almost negligible; this fraction is however overestimated by the massive analyses, that give prominence to the heaviest particles. So gravimetric measurements, performed according to standardized procedures set by law, may prove to be unreliable as an indicator of micro- and nanoparticle pollution: however, studies on the connection between exposure and the onset of adverse health effects hypothesize that particle number rather than particle mass is a fundamental factor to take into account [5]. Tests on blank filters We checked whether possible contaminations of the samples with extraneous particles can take place during the preparation procedures that precede EM observation. Various tests have been applied to find potential artifacts or particle identification mistakes. We tested in particular the coating step: for this purpose we carried out carbon nanofilm deposition, which is the standard preparation for FE-SEM and SEM observation, on one half of some blank filters of PC (0.2 µm pore size) and left the other half untreated. We performed observations at increasing magnification with FE-SEM in SE signal, hand-operated modality, both on filters coated with a carbon nanofilm and on untreated blank filters: we did not detect any artifact. Calibration was done for EDS analysis too. These tests were run at higher energy (30 kV), making use of variable pressure conditions in order not to damage the specimen and achieve a good signal during observations. Again the carbon peak shows the same intensity in both EDS analyses performed on the above-mentioned coated and uncoated PC filters, as it can be seen in Fig. 2 and in Table 1, reporting a statistical analysis on a set of measurements. Table 1. Statistical results after 31 EDS analyses on untreated and treated blank PC filters . Sample without carbon coating . Sample with carbon coating . Mean (C: Weight %) 79.76 79.94 Variance 1.25 1.35 Standard deviation 1.12 1.16 Degree of freedom 30 30 Test F 1.08 Critical value F 1.84 Test z 0.56 Critical value z 1.96 . Sample without carbon coating . Sample with carbon coating . Mean (C: Weight %) 79.76 79.94 Variance 1.25 1.35 Standard deviation 1.12 1.16 Degree of freedom 30 30 Test F 1.08 Critical value F 1.84 Test z 0.56 Critical value z 1.96 The weight percentage of carbon (C) was considered. Performing F and z tests, critical values F and z (significance level: 0.05) turn out to be bigger than calculated values, so there are not significant differences between the untreated and treated samples. Open in new tab Table 1. Statistical results after 31 EDS analyses on untreated and treated blank PC filters . Sample without carbon coating . Sample with carbon coating . Mean (C: Weight %) 79.76 79.94 Variance 1.25 1.35 Standard deviation 1.12 1.16 Degree of freedom 30 30 Test F 1.08 Critical value F 1.84 Test z 0.56 Critical value z 1.96 . Sample without carbon coating . Sample with carbon coating . Mean (C: Weight %) 79.76 79.94 Variance 1.25 1.35 Standard deviation 1.12 1.16 Degree of freedom 30 30 Test F 1.08 Critical value F 1.84 Test z 0.56 Critical value z 1.96 The weight percentage of carbon (C) was considered. Performing F and z tests, critical values F and z (significance level: 0.05) turn out to be bigger than calculated values, so there are not significant differences between the untreated and treated samples. Open in new tab Fig. 2. Open in new tabDownload slide FE-SEM-EDS analysis of a PC blank filter (a) uncoated and (b) coated with a carbon nanofilm: the carbon signals are very similar. Fig. 2. Open in new tabDownload slide FE-SEM-EDS analysis of a PC blank filter (a) uncoated and (b) coated with a carbon nanofilm: the carbon signals are very similar. Similarly for TEM observations, to check possible contaminations due to foreign particles coming from PC dissolution on TEM meshes, we followed this procedure: preparation of a suspension of polystyrene particles in ethanol; the reference sample, obtained from Agar Scientific, is composed of spherical particles with a known certified diameter (86 nm, standard deviation 6.4 nm); filtration on a PC filter (50 µm pore size); dissolution of the PC filter on a Cu mesh with chloroform as described in step 4 of the previous section; analysis with TEM. We observed polystyrene particles only; the specimen was free from iron, steel or other inorganic substances. Effect of the type of sampling filter Following the EN 12341 directive, quartz fiber or PTFE filters are commonly used for PM10 massive sampling, but they may not be suitable for EM observations: in fact, sampling is carried on, complying with the directive, for a day at a flow of 2.3 m3 h−1, and this produces high loadings. Following the ‘procedure A’ described in the subsection ‘Sample preparation’ we can transfer the PM on PC filters: however, if the extracted material forms a multilayer on the PC surface, as we can see in Fig. 3a, single-particle characterization becomes impossible. Fig. 3. Open in new tabDownload slide (a) SEM image of a PC filter which collected PM sampled with a PTFE filter. The material forms a multilayer on the PC surface, thus preventing particle characterization. (b) SEM image of an identical PC filter, which collected PM from a quartz fiber filter, after dilution of the original suspension. There is a great deal of quartz fibres mixed with the particles. (c) FE-SEM image of a PC filter, which collected PM from a PTFE filter, after dilution of the original suspension. The particles are well separated and evenly distributed. Fig. 3. Open in new tabDownload slide (a) SEM image of a PC filter which collected PM sampled with a PTFE filter. The material forms a multilayer on the PC surface, thus preventing particle characterization. (b) SEM image of an identical PC filter, which collected PM from a quartz fiber filter, after dilution of the original suspension. There is a great deal of quartz fibres mixed with the particles. (c) FE-SEM image of a PC filter, which collected PM from a PTFE filter, after dilution of the original suspension. The particles are well separated and evenly distributed. If the PC filter is prepared diluting the extracted particulate suspension (step 3 of the above mentioned procedure) the particles are well separated and evenly distributed. However it can be noted that, starting from a quartz fiber filter, during the extraction many fibers are released and settle on the PC surface, invalidating particulate analysis and characterization (Fig. 3b). Hence PTFE or PC sampling filters (Fig. 3c) are to be preferred to perform at the same time massive analysis and EM characterization: the sample preparation procedure is clean, the extracted particles are well observable and they are in good correlation with those deposited on the starting filter. Anyway we must remark once more that the use of samples collected for subsequent gravimetric measurements is not the optimal choice to perform any analysis through EM techniques. Moreover, the unavoidable extraction procedure (steps 2 and 3 in ‘procedure A’) can cause a loss of soluble components. We then suggest a specific method allowing direct EM observation of the sampling filter and avoiding the laborious procedure previously described. The guidelines we propose are: sampling filters of polycarbonate, with pore size ≤0.2 µm to capture ultrafine PM fraction; volume of sampled air between 2 and 3 Nm3 calibrated for a filter diameter of 23 mm, corresponding to about 450–700 l cm-2, to obtain both a light loading and a representative collected sample. This procedure is applicable for sampling both TSP and subfractions of the PM (PMx); in the latter case suitable preselectors are obviously necessary. In our tests we found that the particles remain well separated on the filter surface and it is possible to perform directly by SEM a chemical, morphological and dimensional analysis of the collected PM. To reproduce the dimensional distribution of the airborne PM, it is important to select an appropriate pore size for the PC filters. The plot in Fig. 4 shows a raw assessment of the theoretical overall collection efficiency of two types of PC filters we have used, performed according to the model proposed by Yamamoto and coworkers [25]. Filters with 0.8 µm pore size appear to have a collection efficiency less than 50% for particles smaller than 200 nm, while filters with 0.2 µm pore size, the value lastly selected for our experiments, seem rather more efficient for nanoparticles also. Fig. 4. Open in new tabDownload slide Theoretical collection efficiency of two PC filters with different porosity (0.2 and 0.8 µm pore size, respectively), calculated following Yamamoto et al. [25]. Fig. 4. Open in new tabDownload slide Theoretical collection efficiency of two PC filters with different porosity (0.2 and 0.8 µm pore size, respectively), calculated following Yamamoto et al. [25]. In a recent work of Cyrs et al. [26], the surface and overall collection efficiencies of PC filters were experimentally measured and theoretically estimated from a sophisticated model, which considers different particle collection mechanisms: diffusion to filter surface, interception and impaction and diffusion to pore walls. The authors studied the performance of PC filters with nominal pore sizes of 0.4 and 0.8 µm at two different flow rates, 0.06 and 0.3 m3 h−1. The results demonstrate that the overall collection efficiencies are higher for the smaller pore size filter; furthermore they indicate that, for nanoparticles, surface collection efficiency can be lower than the overall one: the smallest particles can be trapped inside the filter pores and they are not detected during subsequent SEM observations. Additionally, it is found that surface collection efficiency, equal to 100% for particles larger than about 240 nm, drops for particles smaller than 100 nm, while it returns appreciably higher for the finest particles. This happens because the smallest particles deposit by diffusion; as the particle size increases, deposition by diffusion decreases while collection by impaction is poor, so surface collection efficiency reaches a minimum. As the particle size further increases interception/impaction mechanism becomes more important and collection efficiency increases again. These effects cannot be detected by the simpler model we have used; for a detailed discussion, we refer to the clarifying article of Cyrs and coworkers. They conclude proposing the creation of size-specific correction factors to estimate correctly the particle distribution of the original population. Another finding of these authors is that surface collection efficiency for nanoparticles is higher if the air flow rate is lower, since this condition favors deposition by diffusion. The flow rates they selected are considerably lower than 2.3 m3 h−1, the value established by the EN 12341 directive, so during standardized samplings an important fraction of nanoparticle population can be missed. If particles are to be directly observed with TEM, it could be convenient, before collecting the samples, to mount a copper grid (lacey-carbon kind) directly on the PC filter. The collection of particles from air for direct TEM observation was been already accomplished in the past by Farrants et al. [27], although on non-porous supports and for micrometric particles. Guidelines for PM characterization by means of EM techniques Three kinds of electron microscopes can be used to evaluate number, morphology and chemical characteristics of individual airborne particles deposited on PC filters: SEM, FE-SEM and TEM. Fig. 5 shows the overall information acquired for particle population using each different technique. Fig. 5. Open in new tabDownload slide Information provided by the various EM techniques depending on the explored dimensional range. Fig. 5. Open in new tabDownload slide Information provided by the various EM techniques depending on the explored dimensional range. In the course of our tests we tried to set the optimum value for the EM run parameters. We propose the following guidelines. SEM: Amperage: 500 pA. Voltage: 30 kV. Focal distance: 25 mm. FE-SEM, equipped with INCA-Feature software: Amperage: 400 pA. Voltage: 30 kV. Focal distance: 8.5 mm. TEM: Amperage: 105 µA. Voltage: 200 kV. Taking advantage of INCA-Feature analytical software installed on FE-SEM (also used by Xie et al. [13]), automated mode analysis of suitably prepared filters is possible after manual calibration. In the following the critical operating steps are analyzed. Selection of the examination field A representative field is chosen after verifying sample uniformity. A preliminary examination is performed [22] scanning the sample at a given magnification, usually a tenth of the final one (50 000× for Secondary Electrons (SE) analysis, detecting the whole PM, and 5000× for Back-Scattered Electrons (BSE) analysis, to detect the metallic particles only). Fig. 6 shows an example of two sets of photos taken at two different fields of the same sample (TSP collected in the urban area of Genoa, Italy) with increasing magnification up to 50 000×. Fig. 6. Open in new tabDownload slide Series of micrographies of the same sample (TSP collected in the urban area of Genoa – Italy) at various magnifications, showing how the operator assesses the sample uniformity. Fig. 6. Open in new tabDownload slide Series of micrographies of the same sample (TSP collected in the urban area of Genoa – Italy) at various magnifications, showing how the operator assesses the sample uniformity. We can observe a uniform distribution of the deposited particles over the two filter fields, taking into account an unavoidable degree of variability which becomes more evident at a nanometric scale. This variability of course converges if we increase the number of analyzed fields: for example, scanning 10 different fields on the same sample we counted a mean number of particles equal to 51.5 (standard deviation 3.3) at 500×, and equal to 292.5 (standard deviation 13.9) at 50 000×. Subsequently, the operator chooses a representative field of the filter and enters the coordinates of two fiducial points that define the rectangle to be scanned by the instrument: the standard area is about 0.003 mm2. The rectangle is partitioned by the software in separately scanned fields, in order to achieve the count and the chemical analysis of every particle in automated mode. Setup of the parameters A particle or a cluster of particles is chosen to calibrate the grayscale and the intended parameters are inserted in the proper working window (signal type, field magnification, maximum number of particles per field, scanning modality, etc.). The morphological parameter we have chosen in order to describe the particle dimensions is the mean Feret diameter. During the automated analysis the software relies on a grayscale threshold, pre-selected by the operator, to discriminate between a particle and the filter background. A correct choice of this parameter is then critical and not always simple if the particles under consideration are of many different types; in particular, particles <50 nm could be not detected by the software if the contrast was too low. In order to compare the results from the automated analysis with the results obtained by a direct observation, a manual count was made on some filters for particles smaller than 50 nm, and then for particles between 50 and 500 nm, using Image-Pro Plus software at the same magnification. For a PC filter coming from a TSP sampling in the urban area of Genoa we counted particles in 20 fields of 34 µm2 at 50 000× using SE signal. The results (Table 2), normalized for the standard scanned area (3050 µm2) and compared with the automatic counting, show that the ratio between automated and manual analysis is about 1:3 for particles smaller than 50 nm, while in the dimensional range 50−500 nm this ratio is close to 1:1. Table 2. Comparison between manual counting and automatic counting (× 50 000), on a standard area of 3050 µm2, for particles smaller than 0.050 µm and for particles in the range 0.050–0.5 µm Field (area = 34 µm2) . Number of particles ≤0.050 µm . Calculated number of particles on the standard area . Ratio between automateda and manual count . Number of particles in the range 0.050–0.5 µm . Calculated number of particles on the standard area . Ratio between automatedb and manual count . a 83 7387 1.92 36 3204 0.96 b 118 10 501 2.73 45 4005 1.20 c 184 16 375 4.25 47 4183 1.26 d 138 12 281 3.19 39 3471 1.04 e 154 13 705 3.56 46 4094 1.23 f 93 8277 2.15 35 3115 0.94 g 115 10 235 2.66 41 3649 1.10 h 133 11 836 3.07 44 3916 1.18 i 122 10 857 2.82 39 3471 1.04 j 141 12 548 3.26 40 3560 1.07 k 155 13 794 3.58 43 3827 1.15 l 104 9256 2.40 38 3382 1.02 m 126 11 213 2.91 37 3293 0.99 n 127 11 302 2.93 30 2670 0.80 o 102 9078 2.36 31 2759 0.83 p 117 10 412 2.70 34 3026 0.91 q 99 8811 2.29 32 2848 0.86 r 123 10 946 2.84 37 3293 0.99 s 133 11 836 3.07 40 3560 1.07 t 119 10 590 2.75 38 3382 1.02 Mean 124 11 062 2.87 39 3435 1.03 Standard deviation 22.88 2035.86 0.53 4.73 420.64 0.13 Mean and fiducial intervals (Probability 95%) 2.87 ± 0.25 Mean and fiducial intervals (Probability 95%) 1.03 ± 0.06 Field (area = 34 µm2) . Number of particles ≤0.050 µm . Calculated number of particles on the standard area . Ratio between automateda and manual count . Number of particles in the range 0.050–0.5 µm . Calculated number of particles on the standard area . Ratio between automatedb and manual count . a 83 7387 1.92 36 3204 0.96 b 118 10 501 2.73 45 4005 1.20 c 184 16 375 4.25 47 4183 1.26 d 138 12 281 3.19 39 3471 1.04 e 154 13 705 3.56 46 4094 1.23 f 93 8277 2.15 35 3115 0.94 g 115 10 235 2.66 41 3649 1.10 h 133 11 836 3.07 44 3916 1.18 i 122 10 857 2.82 39 3471 1.04 j 141 12 548 3.26 40 3560 1.07 k 155 13 794 3.58 43 3827 1.15 l 104 9256 2.40 38 3382 1.02 m 126 11 213 2.91 37 3293 0.99 n 127 11 302 2.93 30 2670 0.80 o 102 9078 2.36 31 2759 0.83 p 117 10 412 2.70 34 3026 0.91 q 99 8811 2.29 32 2848 0.86 r 123 10 946 2.84 37 3293 0.99 s 133 11 836 3.07 40 3560 1.07 t 119 10 590 2.75 38 3382 1.02 Mean 124 11 062 2.87 39 3435 1.03 Standard deviation 22.88 2035.86 0.53 4.73 420.64 0.13 Mean and fiducial intervals (Probability 95%) 2.87 ± 0.25 Mean and fiducial intervals (Probability 95%) 1.03 ± 0.06 Data deriving from a representative sample collected on a PC filter in the urban area of Genoa (Italy), TSP sampling mode. aINCA-Feature software detected 3851 particles on the same area (3050 µm2). bINCA-Feature software detected 3325 particles on the same area (3050 µm2). Open in new tab Table 2. Comparison between manual counting and automatic counting (× 50 000), on a standard area of 3050 µm2, for particles smaller than 0.050 µm and for particles in the range 0.050–0.5 µm Field (area = 34 µm2) . Number of particles ≤0.050 µm . Calculated number of particles on the standard area . Ratio between automateda and manual count . Number of particles in the range 0.050–0.5 µm . Calculated number of particles on the standard area . Ratio between automatedb and manual count . a 83 7387 1.92 36 3204 0.96 b 118 10 501 2.73 45 4005 1.20 c 184 16 375 4.25 47 4183 1.26 d 138 12 281 3.19 39 3471 1.04 e 154 13 705 3.56 46 4094 1.23 f 93 8277 2.15 35 3115 0.94 g 115 10 235 2.66 41 3649 1.10 h 133 11 836 3.07 44 3916 1.18 i 122 10 857 2.82 39 3471 1.04 j 141 12 548 3.26 40 3560 1.07 k 155 13 794 3.58 43 3827 1.15 l 104 9256 2.40 38 3382 1.02 m 126 11 213 2.91 37 3293 0.99 n 127 11 302 2.93 30 2670 0.80 o 102 9078 2.36 31 2759 0.83 p 117 10 412 2.70 34 3026 0.91 q 99 8811 2.29 32 2848 0.86 r 123 10 946 2.84 37 3293 0.99 s 133 11 836 3.07 40 3560 1.07 t 119 10 590 2.75 38 3382 1.02 Mean 124 11 062 2.87 39 3435 1.03 Standard deviation 22.88 2035.86 0.53 4.73 420.64 0.13 Mean and fiducial intervals (Probability 95%) 2.87 ± 0.25 Mean and fiducial intervals (Probability 95%) 1.03 ± 0.06 Field (area = 34 µm2) . Number of particles ≤0.050 µm . Calculated number of particles on the standard area . Ratio between automateda and manual count . Number of particles in the range 0.050–0.5 µm . Calculated number of particles on the standard area . Ratio between automatedb and manual count . a 83 7387 1.92 36 3204 0.96 b 118 10 501 2.73 45 4005 1.20 c 184 16 375 4.25 47 4183 1.26 d 138 12 281 3.19 39 3471 1.04 e 154 13 705 3.56 46 4094 1.23 f 93 8277 2.15 35 3115 0.94 g 115 10 235 2.66 41 3649 1.10 h 133 11 836 3.07 44 3916 1.18 i 122 10 857 2.82 39 3471 1.04 j 141 12 548 3.26 40 3560 1.07 k 155 13 794 3.58 43 3827 1.15 l 104 9256 2.40 38 3382 1.02 m 126 11 213 2.91 37 3293 0.99 n 127 11 302 2.93 30 2670 0.80 o 102 9078 2.36 31 2759 0.83 p 117 10 412 2.70 34 3026 0.91 q 99 8811 2.29 32 2848 0.86 r 123 10 946 2.84 37 3293 0.99 s 133 11 836 3.07 40 3560 1.07 t 119 10 590 2.75 38 3382 1.02 Mean 124 11 062 2.87 39 3435 1.03 Standard deviation 22.88 2035.86 0.53 4.73 420.64 0.13 Mean and fiducial intervals (Probability 95%) 2.87 ± 0.25 Mean and fiducial intervals (Probability 95%) 1.03 ± 0.06 Data deriving from a representative sample collected on a PC filter in the urban area of Genoa (Italy), TSP sampling mode. aINCA-Feature software detected 3851 particles on the same area (3050 µm2). bINCA-Feature software detected 3325 particles on the same area (3050 µm2). Open in new tab Of course a manual count performed on photos taken at 50 000× gives more accurate information about nanoparticle identification, because visual observation allows higher resolution than the automated scanning simply based on the selected grayscale; moreover, touching particles could be not distinguished as individual entities in the computer controlled analysis. However we must keep in mind that also the manual count is error-prone because it analyses a smaller area (34 µm2) than the automated mode (hence the variability of the data in column 2 of Table 2), so the mean value reported in column 3 of Table 2 has to be interpreted as an assessment of the number of particles that could be statistically found on the filter area usually analyzed in standard conditions. On the basis of our findings, the number of nanoparticles <50 nm is undoubtedly underestimated by the automated scanning with FE-SEM: we are currently studying if it is possible to set suitable correction factors to be applied to these types of particles. Actually, the potential loss of nanoparticle fractions is a hard and so far unsolved problem at any level (sampling, pretreating, analyzing the PM), as pointed out also in the recent work of Cyrs et al. [26] mentioned above. The quantification of the nanoparticle number percentage potentially lost, for a variety of reasons, when a filter-based apparatus is used deserves undoubtedly major research efforts. Analysis execution The software scans fields one at a time, detecting the single particles and recording morphology and chemical composition data. The instrument memorizes the spectrum of each single particle and its digitalized image. The number of detected and analyzed particles is closely related to the magnification factor used, as one can notice observing the representative results reported in Table 3 (from a TSP sampling in the urban area of Genoa). In order to count particles >1 µm, 5000× is a sufficient magnification, while the use of high magnifying factors (50 000× in SE signal) is mandatory for nanoparticle characterization. Table 3. Automated count and partition of particles into dimensional classes as a function of the magnification factor (FE-SEM, INCA-Feature software) . . Class (µm) . Mag . Total count (n per liter) . ≤0.05 . (0.05–0.1] . (0.1–0.2] . (0.2–0.5] . (0.5–0.75] . (0.75–1.0] . (1.0–2.5] . (2.5–5.0] . (5.0–10.0] . >10.0 . × 200 2.37E + 02 0 0 0 0 0 0 0 2.40E + 01 9.20E + 01 1.20E + 02 × 2000 1.79E + 05 0 0 0 4.13E + 04 6.76E + 04 3.14E + 04 2.85E + 04 1.96E + 03 5.51E + 02 2.26E + 02 × 10 000 5.79E + 06 0 5.08E + 05 2.02E + 06 2.20E + 06 5.68E + 05 1.48E + 05 9.99E + 04 1.73E + 03 1.16E + 03 0 × 50 000 3.63E + 07 1.40E + 07 1.00E + 07 6.35E + 06 3.51E + 06 6.75E + 05 2.23E + 05 5.87E + 04 0 0 0 . . Class (µm) . Mag . Total count (n per liter) . ≤0.05 . (0.05–0.1] . (0.1–0.2] . (0.2–0.5] . (0.5–0.75] . (0.75–1.0] . (1.0–2.5] . (2.5–5.0] . (5.0–10.0] . >10.0 . × 200 2.37E + 02 0 0 0 0 0 0 0 2.40E + 01 9.20E + 01 1.20E + 02 × 2000 1.79E + 05 0 0 0 4.13E + 04 6.76E + 04 3.14E + 04 2.85E + 04 1.96E + 03 5.51E + 02 2.26E + 02 × 10 000 5.79E + 06 0 5.08E + 05 2.02E + 06 2.20E + 06 5.68E + 05 1.48E + 05 9.99E + 04 1.73E + 03 1.16E + 03 0 × 50 000 3.63E + 07 1.40E + 07 1.00E + 07 6.35E + 06 3.51E + 06 6.75E + 05 2.23E + 05 5.87E + 04 0 0 0 Data derived from a campaign of characterization of airborne particles in the city of Genoa. TSP sampling mode, PC filter with pore size of 0.2 µm and diameter of 23 mm, volume of sampled air 2 Nm3. Open in new tab Table 3. Automated count and partition of particles into dimensional classes as a function of the magnification factor (FE-SEM, INCA-Feature software) . . Class (µm) . Mag . Total count (n per liter) . ≤0.05 . (0.05–0.1] . (0.1–0.2] . (0.2–0.5] . (0.5–0.75] . (0.75–1.0] . (1.0–2.5] . (2.5–5.0] . (5.0–10.0] . >10.0 . × 200 2.37E + 02 0 0 0 0 0 0 0 2.40E + 01 9.20E + 01 1.20E + 02 × 2000 1.79E + 05 0 0 0 4.13E + 04 6.76E + 04 3.14E + 04 2.85E + 04 1.96E + 03 5.51E + 02 2.26E + 02 × 10 000 5.79E + 06 0 5.08E + 05 2.02E + 06 2.20E + 06 5.68E + 05 1.48E + 05 9.99E + 04 1.73E + 03 1.16E + 03 0 × 50 000 3.63E + 07 1.40E + 07 1.00E + 07 6.35E + 06 3.51E + 06 6.75E + 05 2.23E + 05 5.87E + 04 0 0 0 . . Class (µm) . Mag . Total count (n per liter) . ≤0.05 . (0.05–0.1] . (0.1–0.2] . (0.2–0.5] . (0.5–0.75] . (0.75–1.0] . (1.0–2.5] . (2.5–5.0] . (5.0–10.0] . >10.0 . × 200 2.37E + 02 0 0 0 0 0 0 0 2.40E + 01 9.20E + 01 1.20E + 02 × 2000 1.79E + 05 0 0 0 4.13E + 04 6.76E + 04 3.14E + 04 2.85E + 04 1.96E + 03 5.51E + 02 2.26E + 02 × 10 000 5.79E + 06 0 5.08E + 05 2.02E + 06 2.20E + 06 5.68E + 05 1.48E + 05 9.99E + 04 1.73E + 03 1.16E + 03 0 × 50 000 3.63E + 07 1.40E + 07 1.00E + 07 6.35E + 06 3.51E + 06 6.75E + 05 2.23E + 05 5.87E + 04 0 0 0 Data derived from a campaign of characterization of airborne particles in the city of Genoa. TSP sampling mode, PC filter with pore size of 0.2 µm and diameter of 23 mm, volume of sampled air 2 Nm3. Open in new tab To express the data we used the number concentration, that is the number of particles in a specific dimensional interval, per unit of volume aspirated: C = (AN)/(naV), where C is the number of particles per liter, A the total exposed area of sampling filter, N the number of total counted particles in the whole area explored, n the number of scanned fields, a the area of each explored field and V the sampled air volume. The EM analysis highlights that the number of the largest particles is by far lower than the number of the finest ones; we remember that these latter, more harmful for the human health, although making the main numerical contribution to the PM population, come out to be insignificant if the analysis is performed according to conventional gravimetric methods. In addition to counting particles in the various dimensional classes, the program can measure, for each single particle, several morphological parameters, e.g. length, breadth, area, aspect ratio, perimeter, etc. Figure 7a shows the shape details observable with an FE-SEM for a siliceous nanoparticle at high magnification (250 000×). Fig. 7b exemplifies the much more sophisticated capabilities peculiar to TEM analysis. In this case (magnification 400 000×), a nanoparticle aggregate is showed where crystal lattices differently oriented are distinguishable. TEM indeed is one of the most powerful tools to reveal extremely detailed structural information; moreover, it is worth remembering that the associated X-EDS analytical system is able to detect also trace elements. Fig. 7. Open in new tabDownload slide (a) FE-SEM image of a siliceous particle at high magnification (250 000×); (b) TEM image of a nanoparticle aggregate containing Fe and Mn oxides (400 000×). Fig. 7. Open in new tabDownload slide (a) FE-SEM image of a siliceous particle at high magnification (250 000×); (b) TEM image of a nanoparticle aggregate containing Fe and Mn oxides (400 000×). It is also possible to obtain, by means of an FE-SEM analysis, INCA-Feature software, the numerical and dimensional distribution of the particles for every identified chemical element; the minimal detectable concentration of an element in a particle is 0.5% by weight. The data reported as an example in Table 4 refer to a sampling of PM10 performed in the harbor area of Genoa using a PC filter with pore size of 0.2 µm and diameter of 23 mm; the volume of sampled air was 2 Nm3. Table 4. Numerical and dimensional distribution of the particles for every identified chemical element (FE-SEM, INCA-Feature software) from a campaign of characterization of airborne particle in the city of Genoa . . Class (µm) . El. . Count (n per liter) . ≤0.05 . (0.05–0.1] . (0.1–0.2] . (0.2–0.5] . (0.5–0.75] . (0.75–1.0] . (1.0–2.5] . (2.5–5.0] . (5.0–10.0] . >10.0 . C 1.28E + 06 6.93E + 05 1.54E + 05 1.55E + 05 2.03E + 05 3.93E + 04 1.61E + 04 1.02E + 04 5.67E + 02 0 0 Al 7.56E + 02 3.78E + 02 0 0 0 0 0 1.89E + 02 0 0 0 As 5.67E + 02 5.67E + 02 0 0 0 0 0 0 0 0 0 Ca 1.02E + 04 4.73E + 03 2.46E + 03 1.13E + 03 9.46E + 02 1.89E + 02 0 5.67E + 02 0 0 0 Cu 2.65E + 03 1.70E + 03 1.89E + 02 0 5.67E + 02 1.89E + 02 0 0 0 0 0 Fe 9.46E + 02 3.78E + 02 0 0 0 1.89E + 02 1.89E + 02 0 1.89E + 02 0 0 K 5.11E + 03 3.03E + 03 3.78E + 02 3.78E + 02 9.46E + 02 3.78E + 02 0 0 0 0 0 Mg 1.32E + 03 1.13E + 03 0 0 0 0 0 1.89E + 02 0 0 0 Na 1.12E + 04 7.00E + 03 9.46E + 02 7.56E + 02 1.32E + 03 3.78E + 02 0 3.78E + 02 0 0 0 P 7.56E + 02 1.89E + 02 0 0 1.89E + 02 1.89E + 02 0 1.89E + 02 0 0 0 Pb 1.89E + 02 0 0 0 0 1.89E + 02 0 0 0 0 0 S 1.10E + 05 6.20E + 04 1.13E + 04 9.08E + 03 1.91E + 04 3.03E + 03 2.46E + 03 3.78E + 02 0 0 0 Si 1.11E + 06 6.31E + 05 1.12E + 05 1.09E + 05 1.73E + 05 3.48E + 04 1.57E + 04 9.83E + 03 5.67E + 02 0 0 Zn 3.78E + 02 1.89E + 02 0 0 0 1.89E + 02 0 0 0 0 0 . . Class (µm) . El. . Count (n per liter) . ≤0.05 . (0.05–0.1] . (0.1–0.2] . (0.2–0.5] . (0.5–0.75] . (0.75–1.0] . (1.0–2.5] . (2.5–5.0] . (5.0–10.0] . >10.0 . C 1.28E + 06 6.93E + 05 1.54E + 05 1.55E + 05 2.03E + 05 3.93E + 04 1.61E + 04 1.02E + 04 5.67E + 02 0 0 Al 7.56E + 02 3.78E + 02 0 0 0 0 0 1.89E + 02 0 0 0 As 5.67E + 02 5.67E + 02 0 0 0 0 0 0 0 0 0 Ca 1.02E + 04 4.73E + 03 2.46E + 03 1.13E + 03 9.46E + 02 1.89E + 02 0 5.67E + 02 0 0 0 Cu 2.65E + 03 1.70E + 03 1.89E + 02 0 5.67E + 02 1.89E + 02 0 0 0 0 0 Fe 9.46E + 02 3.78E + 02 0 0 0 1.89E + 02 1.89E + 02 0 1.89E + 02 0 0 K 5.11E + 03 3.03E + 03 3.78E + 02 3.78E + 02 9.46E + 02 3.78E + 02 0 0 0 0 0 Mg 1.32E + 03 1.13E + 03 0 0 0 0 0 1.89E + 02 0 0 0 Na 1.12E + 04 7.00E + 03 9.46E + 02 7.56E + 02 1.32E + 03 3.78E + 02 0 3.78E + 02 0 0 0 P 7.56E + 02 1.89E + 02 0 0 1.89E + 02 1.89E + 02 0 1.89E + 02 0 0 0 Pb 1.89E + 02 0 0 0 0 1.89E + 02 0 0 0 0 0 S 1.10E + 05 6.20E + 04 1.13E + 04 9.08E + 03 1.91E + 04 3.03E + 03 2.46E + 03 3.78E + 02 0 0 0 Si 1.11E + 06 6.31E + 05 1.12E + 05 1.09E + 05 1.73E + 05 3.48E + 04 1.57E + 04 9.83E + 03 5.67E + 02 0 0 Zn 3.78E + 02 1.89E + 02 0 0 0 1.89E + 02 0 0 0 0 0 PM10 sampling mode, PC filter with pore size of 0.2 µm and diameter of 23 mm, volume of sampled air 2 Nm3. Open in new tab Table 4. Numerical and dimensional distribution of the particles for every identified chemical element (FE-SEM, INCA-Feature software) from a campaign of characterization of airborne particle in the city of Genoa . . Class (µm) . El. . Count (n per liter) . ≤0.05 . (0.05–0.1] . (0.1–0.2] . (0.2–0.5] . (0.5–0.75] . (0.75–1.0] . (1.0–2.5] . (2.5–5.0] . (5.0–10.0] . >10.0 . C 1.28E + 06 6.93E + 05 1.54E + 05 1.55E + 05 2.03E + 05 3.93E + 04 1.61E + 04 1.02E + 04 5.67E + 02 0 0 Al 7.56E + 02 3.78E + 02 0 0 0 0 0 1.89E + 02 0 0 0 As 5.67E + 02 5.67E + 02 0 0 0 0 0 0 0 0 0 Ca 1.02E + 04 4.73E + 03 2.46E + 03 1.13E + 03 9.46E + 02 1.89E + 02 0 5.67E + 02 0 0 0 Cu 2.65E + 03 1.70E + 03 1.89E + 02 0 5.67E + 02 1.89E + 02 0 0 0 0 0 Fe 9.46E + 02 3.78E + 02 0 0 0 1.89E + 02 1.89E + 02 0 1.89E + 02 0 0 K 5.11E + 03 3.03E + 03 3.78E + 02 3.78E + 02 9.46E + 02 3.78E + 02 0 0 0 0 0 Mg 1.32E + 03 1.13E + 03 0 0 0 0 0 1.89E + 02 0 0 0 Na 1.12E + 04 7.00E + 03 9.46E + 02 7.56E + 02 1.32E + 03 3.78E + 02 0 3.78E + 02 0 0 0 P 7.56E + 02 1.89E + 02 0 0 1.89E + 02 1.89E + 02 0 1.89E + 02 0 0 0 Pb 1.89E + 02 0 0 0 0 1.89E + 02 0 0 0 0 0 S 1.10E + 05 6.20E + 04 1.13E + 04 9.08E + 03 1.91E + 04 3.03E + 03 2.46E + 03 3.78E + 02 0 0 0 Si 1.11E + 06 6.31E + 05 1.12E + 05 1.09E + 05 1.73E + 05 3.48E + 04 1.57E + 04 9.83E + 03 5.67E + 02 0 0 Zn 3.78E + 02 1.89E + 02 0 0 0 1.89E + 02 0 0 0 0 0 . . Class (µm) . El. . Count (n per liter) . ≤0.05 . (0.05–0.1] . (0.1–0.2] . (0.2–0.5] . (0.5–0.75] . (0.75–1.0] . (1.0–2.5] . (2.5–5.0] . (5.0–10.0] . >10.0 . C 1.28E + 06 6.93E + 05 1.54E + 05 1.55E + 05 2.03E + 05 3.93E + 04 1.61E + 04 1.02E + 04 5.67E + 02 0 0 Al 7.56E + 02 3.78E + 02 0 0 0 0 0 1.89E + 02 0 0 0 As 5.67E + 02 5.67E + 02 0 0 0 0 0 0 0 0 0 Ca 1.02E + 04 4.73E + 03 2.46E + 03 1.13E + 03 9.46E + 02 1.89E + 02 0 5.67E + 02 0 0 0 Cu 2.65E + 03 1.70E + 03 1.89E + 02 0 5.67E + 02 1.89E + 02 0 0 0 0 0 Fe 9.46E + 02 3.78E + 02 0 0 0 1.89E + 02 1.89E + 02 0 1.89E + 02 0 0 K 5.11E + 03 3.03E + 03 3.78E + 02 3.78E + 02 9.46E + 02 3.78E + 02 0 0 0 0 0 Mg 1.32E + 03 1.13E + 03 0 0 0 0 0 1.89E + 02 0 0 0 Na 1.12E + 04 7.00E + 03 9.46E + 02 7.56E + 02 1.32E + 03 3.78E + 02 0 3.78E + 02 0 0 0 P 7.56E + 02 1.89E + 02 0 0 1.89E + 02 1.89E + 02 0 1.89E + 02 0 0 0 Pb 1.89E + 02 0 0 0 0 1.89E + 02 0 0 0 0 0 S 1.10E + 05 6.20E + 04 1.13E + 04 9.08E + 03 1.91E + 04 3.03E + 03 2.46E + 03 3.78E + 02 0 0 0 Si 1.11E + 06 6.31E + 05 1.12E + 05 1.09E + 05 1.73E + 05 3.48E + 04 1.57E + 04 9.83E + 03 5.67E + 02 0 0 Zn 3.78E + 02 1.89E + 02 0 0 0 1.89E + 02 0 0 0 0 0 PM10 sampling mode, PC filter with pore size of 0.2 µm and diameter of 23 mm, volume of sampled air 2 Nm3. Open in new tab The table shows that by choosing appropriate parameters, such as number of particles in the sampled volume unit and dimensional distribution for each single chemical elements detected, it is possible to acquire important information about the possible origin of the collected PM. These data, not obtainable from massive analyses, are essential to achieve a better control of the atmospheric pollution sources. A detailed morpho-chemical analysis of the PM collected in the urban area of Genoa will be the subject of a following work. Conclusions As we have shown, it is possible to characterize airborne micro and nanoparticulate using EM techniques. The filters employed during the sampling process display different performances with relation to their chemical constitution and porosity. Our results show that quartz fiber filters are not suitable; among sampling filters complying with EN 12341 European Standard, only PTFE filters are viable for SEM and TEM analysis, and only after appropriate treatment. Ethanol allows the complete detachment of the particles from the massive sampling filters during the extraction procedure and transfer on PC filters. However, some chemical class of particles can be underestimated or lost due to solubilization in the course this treatment. On the basis of our tests, a proper sampling procedure for EM analysis of the airborne particulate requires the use of PC collecting filters together with low flow and reduced volumes during the sampler operation. This operating method eliminates any pretreatment of the sample, which could introduce errors and artifacts, and makes single particles deposited on the filter directly observable by SEM. For TEM analysis the particle loading must be transferred on a Cu lacey-carbon grid, dissolving the PC filter in a chloroform thermostatic bath. In fact we have verified that if the grid is previously mounted on the sampling filter also TEM observation can be directly performed. The procedures we suggest are adjustable for sampling both TSP and sub fractions of the PM (PMx). With INCA-Feature equipped FE-SEM analysis, it is possible to achieve quantitative data, especially important in the submicrometric range, for example particle distribution according to dimensional classes and chemical classes. We emphasize the capability to carry out a morphological and chemical characterization of each single particle, which is of utmost importance owing to the correlation between particle composition/diameter and its uptake by the respiratory system. With regard to our TEM analysis, it does not provide quantitative results, but it gives us very important information about nanoparticle morphology and structure, carbon-black presence and degree of aggregation of the particulate. At the present time, a comprehensive view and a detailed assessment of the kinds of airborne particles are not obtainable using a single conventional analyzing apparatus. The gravimetric measurements only provide information on the weight content of total PMx collected per volume unit, but do not take into account the number concentration and the dimensional distribution of the particle population; moreover they underestimate the presence of fine and ultrafine particles, that give a predominant numerical contribution but a negligible massive contribution. The spectroscopic techniques usually associated with massive analyses can provide information on the mean chemical composition of the collected particulate, but do not allow to connect the single particle to its composition. On the other hand, the automated particle counters nowadays can measure number and size distributions of particles down to about 5 nm in real time; such devices however do not provide any information on the chemical composition of the examined particulate. The EM techniques, coupled with Energy Dispersive X-ray Spectrometry, are ideally placed to combine numerical, morphological and chemical information, although with some limitations, particularly in the dimensional range below 50 nm. The possibility of analyzing individual particles imparts to these techniques unique features allowing to identify the different particle types present in the sample: in this way possible pollution sources can be deduced and potential adverse effects can be predicted. Acknowledgements We would like to state our appreciation to Dr Mauro Michetti, Mr Claudio Uliana and Ms Laura Negretti (DCCI) for technical assistance, Dr Enrico Daminelli, Mr Federico Manni (Provincia di Genova) and Dr Eleonora Cuccia (DIFI) for supplying materials. References 1 Williams M , Bruckmann P . A Report on Guidance to Member States on PM10 Monitoring and Intercomparisons with the Reference Method , 2001 Brussels European Commission Working Group on Particulate Matter Draft Final Report 2 Qian Z , He Q , Lin H-M , Kong L , Liao D , Dan J , Bentley Christy M , Wang B . Association of daily cause-specific mortality with ambient particle air pollution in Wuhan, China , Environ. Res. , 2007 , vol. 105 (pg. 380 - 389 ) 10.1016/j.envres.2007.05.007 Google Scholar Crossref Search ADS PubMed WorldCat Crossref 3 Gardiner K , Calvert I A , Van Tongeren M J A , Harrington J M . Occupational exposure to carbon black in its manufacture: data from 1987 to 1992 , Ann. Occup. Hyg. , 1996 , vol. 40 (pg. 65 - 77 ) Google Scholar Crossref Search ADS PubMed WorldCat 4 Lee B-K , Smith T J , Garshick E , Natkin J , Reaser P , Lane K , Lee H K . Exposure of trucking company workers to particulate matter during the winter , Chemosphere , 2005 , vol. 61 (pg. 1677 - 1690 ) 10.1016/j.chemosphere.2005.03.091 Google Scholar Crossref Search ADS PubMed WorldCat Crossref 5 Peters A , Wichmann H E , Tuch T , Heinrich J , Heyder J . Respiratory effects are associated with the number of ultrafine particles , Am. J. Respiratory Crit. Care Med. , 1997 , vol. 155 (pg. 1376 - 1383 ) Google Scholar Crossref Search ADS WorldCat 6 Querol X , Alastuey A , Rodriguez S , Plana F , Mantilla E , Ruiz C R . Monitoring of PM10 and PM2.5 around primary particulate anthropogenic emission sources , Atmos. Environ. , 2001 , vol. 35 (pg. 845 - 858 ) 10.1016/S1352-2310(00)00387-3 Google Scholar Crossref Search ADS WorldCat Crossref 7 Pindado O , Perez R M , Garcia S , Sanchez M , Galan P , Fernandez M . Characterization and sources assignation of PM2.5 organic aerosol in a rural area of Spain , Atmos. Environ. , 2009 , vol. 43 (pg. 2796 - 2803 ) 10.1016/j.atmosenv.2009.02.046 Google Scholar Crossref Search ADS WorldCat Crossref 8 Reinard M S , Adou K , Martini J M , Johnston M V . Source characterization and identification by real-time single particle mass spectrometry , Atmos. Environ. , 2007 , vol. 41 (pg. 9397 - 9409 ) 10.1016/j.atmosenv.2007.09.001 Google Scholar Crossref Search ADS WorldCat Crossref 9 Healy R M , O'Connor I P , Hellebust S , Allanic A , Sodeau J R , Wenger J C . Characterisation of single particles from in-port ship emissions , Atmos. Environ. , 2009 , vol. 43 (pg. 6408 - 6414 ) 10.1016/j.atmosenv.2009.07.039 Google Scholar Crossref Search ADS WorldCat Crossref 10 Paoletti L , De Berardis B , Arrizza L . Inquinamento da polveri e da particolato fino in siti con differenti caratteristiche ambientali , Ann. Ist. Super Sanità , 2003 , vol. 39 (pg. 381 - 385 ) Google Scholar PubMed OpenURL Placeholder Text WorldCat 11 Kumar P , Robins A , Vardoulakis S , Britter R . A review of the characteristics of nanoparticles in the urban atmosphere and the prospects for developing regulatory controls , Atmos. Environ. , 2010 , vol. 44 (pg. 5035 - 5052 ) 10.1016/j.atmosenv.2010.08.016 Google Scholar Crossref Search ADS WorldCat Crossref 12 Lin J J , Lee L-C . Characterization of the concentration and distribution of urban submicron (PM1) aerosol particles , Atmos. Environ. , 2004 , vol. 38 (pg. 469 - 475 ) 10.1016/j.atmosenv.2003.09.056 Google Scholar Crossref Search ADS WorldCat Crossref 13 Xie R K , Seip H M , Leinum J R , Winje T , Xiao J S . Chemical characterization of individual particles (PM10) from ambient air in Guiyang City, China , Sci. Total Environ. , 2005 , vol. 343 (pg. 261 - 272 ) 10.1016/j.scitotenv.2004.10.012 Google Scholar Crossref Search ADS PubMed WorldCat Crossref 14 Liu X , Zhu J , Van Espen P , Adams F , Xiao R , Dong S , Li Y . Single particle characterization of spring and summer aerosols in Beijing: formation of composite sulfate of calcium and potassium , Atmos. Environ. , 2005 , vol. 39 (pg. 6909 - 6918 ) 10.1016/j.atmosenv.2005.08.007 Google Scholar Crossref Search ADS WorldCat Crossref 15 Jones T , Moreno T , Bèrubé K , Richards R . The physicochemical characterisation of microscopic airborne particles in south Wales: a review of the locations and methodologies , Sci. Total Environ. , 2006 , vol. 360 (pg. 43 - 59 ) 10.1016/j.scitotenv.2005.08.055 Google Scholar Crossref Search ADS PubMed WorldCat Crossref 16 Aragon Piña A , Torres Villasenor G , Monroy FernaHndez M , Luszczewski Kudra A , Leyva Ramos R . Scanning electron microscope and statistical analysis of suspended heavy metal particles in San Luis Potosi, Mexico , Atmos. Environ. , 2000 , vol. 34 (pg. 4103 - 4112 ) 10.1016/S1352-2310(99)00526-9 Google Scholar Crossref Search ADS WorldCat Crossref 17 Shi Z , Shao L , Jones T P , Whittaker A G , Lu S , Berube K A , He T , Richards R J . Characterization of airborne individual particles collected in an urban area, a satellite city and a clean air area in Beijing, 2001 , Atmos. Environ. , 2003 , vol. 37 (pg. 4097 - 4108 ) 10.1016/S1352-2310(03)00531-4 Google Scholar Crossref Search ADS WorldCat Crossref 18 Aragon Piňa A , Torres Villaseňor G , Santiago Jacinto P , Monroy Fernandez M . Scanning and transmission electron microscope of suspended lead-rich particles in the air of San Luis Potosi, Mexico , Atmos. Environ. , 2002 , vol. 36 (pg. 5235 - 5243 ) 10.1016/S1352-2310(02)00588-5 Google Scholar Crossref Search ADS WorldCat Crossref 19 Murr L E , Garza K M . Natural and anthropogenic environmental nanoparticulates: their microsctructural characterization and respiratory health implications , Atmos. Environ. , 2009 , vol. 43 (pg. 2683 - 2692 ) 10.1016/j.atmosenv.2009.03.002 Google Scholar Crossref Search ADS WorldCat Crossref 20 Wang C , Friedlander S , Madler L . Nanoparticle aerosol science and technology: an overview , China Particuol. , 2005 , vol. 3 (pg. 24 - 254 ) OpenURL Placeholder Text WorldCat 21 Dye A L , Rhead M M , Trier C J . A porous carbon film for the collection of atmospheric aerosol for transmission electron microscopy , J. Microsc. , 1997 , vol. 187 (pg. 134 - 138 ) 10.1046/j.1365-2818.1997.2260781.x Google Scholar Crossref Search ADS WorldCat Crossref 22 EPA Guidelines for the Application of SEM/EDX Analytical Techniques to Particulate Matter Samples , 2002 600/R-02/070 23 EPA federal register , Rules and Regulations , 1987 , vol. 52 210 (pg. 41880 - 41881 ) OpenURL Placeholder Text WorldCat 24 Spurny K R . Sampling analysis, identification and monitoring of fibrous dust and aerosols , Analyst , 1994 , vol. 119 (pg. 41 - 51 ) 10.1039/an9941900041 Google Scholar Crossref Search ADS WorldCat Crossref 25 Yamamoto N , Kumagai K , Fujii M , Shendell Derek G , Endo O , Yanagisawa Y . Size-dependent collection of micrometer-sized particles using nylon mesh , Atmos. Environ. , 2005 , vol. 39 (pg. 3675 - 3685 ) 10.1016/j.atmosenv.2005.03.001 Google Scholar Crossref Search ADS WorldCat Crossref 26 Cyrs W D , Boysen D A , Casuccio G , Lersch T , Peters T M . Nanoparticle collection efficiency of capillary pore membrane filters , J. Aerosol Sci. , 2010 , vol. 41 (pg. 655 - 664 ) 10.1016/j.jaerosci.2010.04.007 Google Scholar Crossref Search ADS WorldCat Crossref 27 Farrants G , Rieth A , Schuler B , Feren K . A simple direct method for the collection of particles from air for transmission electron microscopy and digital image analysis , J. Microsc. , 1988 , vol. 149 (pg. 159 - 164 ) Google Scholar Crossref Search ADS WorldCat © The Author 2011. Published by Oxford University Press [on behalf of Japanese Society of Microscopy]. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com
TI - Electron microscopy characterization of airborne micro- and nanoparticulate matter
JF - Journal of Electron Microscopy
DO - 10.1093/jmicro/dfr001
DA - 2011-04-01
UR - https://www.deepdyve.com/lp/oxford-university-press/electron-microscopy-characterization-of-airborne-micro-and-F30ToPRFNy
SP - 117
EP - 131
VL - 60
IS - 2
DP - DeepDyve
ER -