A Multi-Analytical Approach to the Use of Conifer Needles as Passive Samplers of Particulate Matter and Organic Pollutants

The present work regards the analysis of airborne pollutants emitted by an electric arc furnace steel making plant at a test site in Northern Italy and collected by conifer needles. The spatial and temporal trends of accumulation of Mn, Zn, Fe, Cr, Pb, polychlorinated dibenzo-p-dioxins/dibenzofurans and polychlorinated biphenyls dioxin revealed that the contribution from the steel making plant has never been particularly high, whereas traffic emerged as a significant pollution source. The benefits of combining bulk and single particle analysis in air pollution studies from different sources are also discussed.


INTRODUCTION
Conifer needles can be considered passive samplers of particulate matter (Urbat et al., 2004) and a number of semi-volatile organic compound (SVOC) pollutants (Eriksson et al., 1989, Tremolada et al., 1993, Kylin et al., 1994, Tremolada et al., 1996, Di Guardo et al., 1999).For this reason, needles are used for air quality studies based on biomonitoring.The capturing mechanism depends on pollutant state.Gaseous organic pollutants can diffuse through the cuticle, penetrate through the stomata (Urbat et al., 2004) or be trapped in the epicuticular wax (Iozza et al., 2009).Inorganic particles or particle-associated organic pollutants deposit on needles via wet (Eriksson et al., 1989) or dry mechanisms (Turunen et al., 1997) and remain there thanks to the presence of sticky substances like wax or honeydew deposits (Sawidis et al., 2012) and surface roughness (Teper, 2009).
As conifers are evergreen, they have the advantage of accumulating pollutants for several years also during wintertime (Lehndorff and Schwark, 2004).In countries like Finland, needles are routinely employed to evaluate sulphur and heavy metal deposition (Pöykiö et al., 2000).
However, caution is required when using these passive samplers, since monitored concentrations depend on: species, age of the needles, nature and content of wax, aerodynamic factors, temperature and precipitations, location within the forest canopy (Di Guardo et al., 2003).Moreover, concentrations may be underestimated due to particle removal by wind and rain (Urbat et al., 2004).
Bulk analytical techniques, like inductively coupled plasma atomic emission spectroscopy (ICP-AES) and Xray fluorescence (XRF), are widely used to investigate inorganic pollutants, particularly heavy metals.On the other hand, a few studies utilize single particle analysis (SPA) techniques, like electron microscopy (Iwasaki and Tainosho, 2003).This work aims at combining the benefits of the two approaches to identify spatial and temporal trends in the accumulation of particles emitted by a steel making plant located in a test site in Northern Italy, where the concentrations of polychlorinated dibenzo-p-dioxins/ dibenzofurans (PCDD/Fs) and polychlorinated biphenyls dioxin (like PCBs dl) were even measured.
The case of steelmaking plants utilizing electric arc furnaces (EAFs) is of general interest as they emit particulate matter containing hazardous heavy metals (Baiutti et al., 2007), and represent a significant amount of the 60% of plants that in developed countries produce steel from scraps (APAT, 2003;Wichterle et al., 2010).Chemical composition of the dust coming from EAFs depends on the type of scraps and on the type of production plant.Emitted dust would contain several oxide phases, some being characteristic of steel making plants, like zinc, manganese, chromium, lead and iron oxides (Baiutti et al., 2007).Zinc comes from scraps with relicts of anticorrosion coatings.Manganese derives from silico-manganese, a de-oxidising alloying component (Yue et al., 2006).Chromium is from stainless steel scarps; lead, not soluble in steel, may come from several scraps, like automatic steels, but, since it has even other sources, cannot be used as a marker for this industrial activity.The same is true for iron.PCDD/Fs are formed from combustion of scraps containing C, O, H and halogens: paints, plastic and oils (Panizza, 2011).PCBs are present as contaminants of such apparatus parts, like condensers, transformers and containers of mineral oils, that are charged in the EAF in association with vehicle scraps as well as with paints, plastics and oils (Panizza, 2011).
The examined EAF steel making plant has been active for about forty years and is located in the industrial periphery of a Northern Italian town of about 6850 inhabitants located in the floor of a valley 400 m a.s.l.. Surrounding the plant are other industrial and commercial activities, a regional diesel engine railway, some street connecting to the main motorway crossing the valley, a river and rural areas.
Since 1990 the emission suppression system of the plant had two lines: one for the primary emissions and part of the secondary, and one for the secondary emissions only.The two lines are connected with two 40 m high stacks.Emissions were reduced through an Air Pollution Control (APC) system made of a dust bag, a quenching tower, a vertical cyclone and a bag filter.In recent years improvements in the emission collection system were undertaken by the steel making plant to meet the Best Available Technology (BAT) criteria imposed by the local environmental agency.Part of the emission control system is composed of a quenching tower, a horizontal cyclone and a filter at the end of the line.A significant part of the emissions, however, results from those operations of the steelmaking process that must be carried out with the aspiration and off-gas abatement system disconnected.Other diffuse emissions may originate from the movement of the scraps near the plant, the handling of dust collected by the abatement system and the demolition of ladle refractory materials.Moreover, the shop was not perfectly sealed on the sides until the latest improvements, thus fugitive emissions were released at ground level.All dust emitted may be re-suspended by the wind and trucks moving in the plant area (Ciuta et al., 2012).These diffuse contributions may bias pollutant measurements, if not taken into the right account (Schiavon et al., 2013).
In mountainous areas where conifers are available and thermal inversion may hinder the dispersion of pollutants, the approach developed in this study might help estimating time and spatial trends in the distribution of emitted pollutants.The method could also provide data to support air quality modelling, even for those remote areas with no instrumental monitoring systems (Klánová et al., 2009).
The effects of airborne particulate matter (PM) pollution are generally more localized than those of gaseous contaminants (Pöykiö et al., 2010).Therefore, it is expected that airborne metals, which may be considered as tracers of the steel making plant (Sammut et al., 2006;Baiutti et al., 2007), will be mostly found deposited in the proximity of the plant.
The present study proposes a novel approach to study inorganic and organic pollutants sampled using conifer needles.For the characterization of the pollutants a multianalytical approach has been developed based on the combination of bulk and single particle analytical techniques.The data obtained with this approach are particularly suitable as input for modelling the dispersion of pollutants as the concentrations of both organic and inorganic contaminants are obtained with the same sampling strategy.Another interesting aspect emerging from the study is the possibility of investigating temporal evolution in air quality over a time scale of several years.

Sampling
Among conifer species, spruce (Picea abies), was chosen for this investigation as it is widely available in the study location and because in the literature exist many comparison data which were obtained using this species (VDI, 2007).Moreover, the fact that this specie is tolerant to manganese (Wyttenbach et al., 1995) is important in our study as manganese is a tracer for iron and steel industries (Baiutti et al., 2007).
Two sampling campaigns were conducted.In the first, only needles from trees nearby the EAF plant were collected for a preliminary study, not reported herewith, meant to optimize the analytical protocol.The target of the second campaign was to collect samples for two gradient studies: one parallel another perpendicular to the main wind direction, for the characterization of organic and inorganic pollutants (VDI, 2004).The needles were sampled at increasing distances from the EAF plant at relative different position on the branches of the same tree in order to have samples of different ages: 2005 and 2011 and 2008 and 2012 needles were collected (Table 1).The sites, selected on the basis of dispersion models, should represent different levels of pollution and include both high and low concentrations areas (VDI, 2004).
A sample of spruce needles was collected in a pristine location within the Denali National Park, Alaska, USA and used as a reference.

Sample Preparation and Analysis of Inorganic Pollutants
Each sample collected at the sites named F, L, P and M (see Fig. 1 and Table 1) was divided into three subsamples.The first subsample of a few needles was used for scanning electron microscopy (SEM) analysis.Another subsample was used to measure the internal composition after a washing treatment described in the following.These internal element concentrations were subtracted from the total concentrations measured on the third subsample of unwashed needles.The subtraction allows to estimate the atmospheric deposition contribution to total element concentrations.For SEM, needles were mounted on 25 mm diameter aluminium stubs with double-sided adhesive tape.Observations were carried out with a low-vacuum SEM coupled with an energy dispersive X-ray spectrometer (LVSEM-EDXS) system operated at 0.6 Torr at a voltage of 20 kV.
Needles for internal composition were washed in a 1:1 solution of toluene:tetrahydrofuran; they were immersed three times in the solution for 15 s and then rinsed in deionized water.Both the unwashed and the washed  Concentrations were calculated as where conc ICP-AES is the average of three measurements for elements whose concentration are above the instrument detection limit (mg/L); V is the final extraction volume in litres (0.1 l); Blank is the average concentration of that element in the reagent blank sample in mg/L; mass is the total amount of sample weighted and digested (kg).The uncertainty on each concentration measurement was obtained through error propagation.The quality of the chemical analysis was checked from blanks, replicate samples, and measurement of the element concentrations of a certified standard (NIST 1575 pine-needles) with the same protocol.
The precision of the measurements, calculated as relative standard deviation on the concentrations of duplicate samples, is of 0.2 for Mn and Cu, 0.01 for Fe.

Analysis of Organic Pollutants
The determination of PCDD/Fs and PCBs dl was conducted following the United States Environmental Protection Agency (EPA) protocols: EPA 1613 1994 (EPA, 1994) and EPA 1668B 2008(EPA, 2008).These protocols use high resolution capillary column gas chromatography (HRGC)/high resolution mass spectrometry (HRMS).
Several complementary studies were conducted measuring both organic and inorganic pollutants collected by instrumentations (Ragazzi et al., 2012).The obtained results will be compared with those of the present study.

ICP-AES
ICP-AES analyses conducted on unwashed and washed samples allowed the evaluation of the relevant concentrations differences of metals presumably coming from atmospheric deposition.To validate the methodology, the concentrations of interesting elements in the sample from the pristine location in Alaska were measured.While Fe (12.0 ± 3.7 ppm) and Zn (7.0 ± 2.5 ppm) resulted to be present, the other elements (Mn 0 ppm, Cu < limit of detection, C and Pb < instrument detection limit) seem to be present in the sampled needles only due to root uptake.This is in agreement with Wyttenbach et al. (1995), who highlighted that, unless barium and strontium are present to very low extent (≤ 1 ppm), the concentration of Mn due to atmospheric deposition does not exceed 2% of its internal concentration.A similar agreement concerns Cu and Zn, that in polluted environment are mainly exogenous and from root uptake respectively (Nieminen et al., 2004).As concerns Fe deposition on conifer needles, our results seem to parallel those published by Wyttenbach and Tobler (2000).These considerations suggest that, if significant surface concentrations of Mn, Cu, Cr and Pb are detected, these indicate air contamination with that metal.On the other hand, for Fe and Zn the possible presence of an atmospheric contribution ought to be supported by additional information associated with air pollution.
As summarized by Fig. 1, a decreasing trend was observed in the metal concentrations from atmospheric deposition for all elements between 2008 and 2012 at all sites except M (Fe) and L (Cu).Cu increase at site L is not necessarily correlated to the emissions from the steelmaking plant, since Cu has many other sources, like traffic, that is present at that site.Similarly, also Fe at site M may have several sources.The detection of Zn and Fe at certain sites does not necessarily involve that in the period of exposure the atmosphere was contaminated with these metals.To prove that the contamination detected in certain samples is truly due to the presence of the steel making plant it is necessary to exclude that Mn, Cu, Cr and Pb were emitted by other sources, such as other manufacturing activities and non-exhaust traffic emissions.Hence, it is necessary to know in which particles these metals are located.Once the source particle is identified, its morphology is used for source apportionment.

LVSEM-EDXS
LVSEM-EDXS analysis of the needles collected near the steel plant revealed the presence of different kinds of particles.Among the observed morphological categories, two may be related to the EAF steel making plant and they are both chemically characterized by the presence of characteristic heavy metals (Fe, Cu, Zn, Mn, Cr): -non-spherical particles containing heavy metals (diameter ≥ 2.5 µm) Fig. 2(a)) -sub-spherical particles containing heavy metals (Fe, Zn, Cu) (diameter ≥ 1 µm) (Fig. 2(b)) These particles appear brighter in the backscattered electron (BSE) images.
Conifer needles confirmed their capability to accumulate PM in time, as at each site, no matter the level of pollution, thus of deposited materials, there are more particles on 2008 needles than on current year needles (Fig. 3).It is also confirmed the compositional ICP-AES trend among sampling sites, as the needles richest in particles are P, F and L while M needles have practically no deposited materials on the surface.Data for the main wind directions are coherent with this result (Rada et al., 2012).
Needles displayed the typical degradation of the epicuticular wax to different extents (Fig. 4).Although this degradation is accelerated by pollution (Karhu and Huttunen, 1986), it is almost entirely due to the aging of the needles, as a similar morphology is observed on 2008 needles from the most unpolluted site M (Fig. 4).
The particles characteristic of EAF steelmaking plant emissions were detected mainly on the needles from sites P and F, closer to the plant, than in farther sites, like site M.
It must be pointed out that LVSEM observations highlighted the strong contribution of traffic at site P and L where particles characteristic of brake pads wear were detected (Fig. 2(c)).These barite (BaSO 4 ) containing particles, represent the component of filler brake pads (Thorpe and Harrison, 2008;Gietl et al., 2010).Other particles due to non-exhaust vehicle emissions, detected at these sites, were those containing zirconium coming from zirconium silicate used as abrasive in brake lining (Thorpe and Harrison, 2008), Fig. 2. SEM micrographs of particles from different sources present on the needles: A-EAF steelmaking particle containing Fe and Mn; B-EAF fly ash particle containing Fe, Zn and Cu; C-brake wear particle; D-Fe-rich alumino-silicate fly ash.and those containing titanium and potassium, characteristic of semimetallic brakes composed of potassium titanate fibers (Garg et al., 2000).In this respect, it must be highlighted that the trees sampled at site P are close to a cross-road where all cars have to press the brake.
Other particles, such as fly ash (Fig. 2(d)), that may derive from a variety of combustion sources, including, although not exclusively, the steel plant, were also identified.Natural particles coming from rocks have been easily recognized in BSE images so that it was not necessary to analyze them.
Comparing these data with those from needles collected in Alaska, it emerges that, although PM is found also on such needles from a pristine location (Fig. 3), this is mainly of natural origin.A spherical particle containing Mn, Fe, Zn, and Cr was identified on the Alaska needle highlighting again the great screening ability provided by SEM technique and the good correspondence with bulk analysis that detected the presence of Zn and Fe.

Organic Pollutants
As concerns organic pollutants, in addition to the simple concentrations of PCDD/Fs and PCBs congeners, the Toxic Equivalent Concentration (TEQ) can be obtained, using the toxic equivalency factors TEFs.Two types of TEFs are available: the international TEFs (I-TEFs) and the World Health Organization TEFs (WHO-TEFs).The I-TEFs were developed by an international scientific committee convened under the auspices of the North Atlantic Treaty Organization (NATO) in 1989 and they were later extended and updated (NATO/CCMS) (NATO/CCMS, 1988).The WHO-TEFs were revised in 2005 (UNEP/POPS/COP.3/INF/27).Table 2 lists the concentration values obtained from the different samples.The last line of the table shows the value of TEQ of the sample, calculated using the TEF of each congener.The concentrations below the limit of detection (LOD), which were considered to be half of LOD, are marked in blue.
The concentrations obtained are very low, as highlighted by the many values below the LOD: 110 of 136 for PCDD/Fs and 53 of 96 for PCBs dl.This may witness the reduced emissions of these compounds by the steel plant after the 90s and the absence of other sources, which is coherent with the mostly rural land use of the valley.The low concentrations are also demonstrated by a comparison with the measurements on the Alaska sample.
Not always the concentrations of organic pollutants in 2005 samples are higher than those of 2011 (Table 2), as one would expect since the needles of the first are six years older than the second (Iozza et al., 2009).These data contrast the fact that aging should make the cuticular membranes more permeable and thus enhance the capacity of older needles to collect organic pollutants thanks to diffusion through the cuticle.It is also in contrast with the higher concentrations of inorganic pollutants on older needles, previously discussed, that parallel the trend already observed by Turunen et al. (1997).Besides the longer exposure time, aging causes the erosion of the wax (which is produced only during the first growing season (Turunen and Huttunen, 1996)) that in turn favors, alone and in combination with the presence of fungi on the needle surface, the capture of particulate matter (Wang et al., 2011).Since 70% of the canopy of conifers consists of needles produced more than a year ago (Muukkonen, 2005), the higher capture ability of older needles, is a positive characteristic for biomonitoring studies.
In this study the higher concentrations of organic pollutants expressed as TEQ are found in sample number 4 both for 2005 and 2011 needles.This might be related to the fact that spruce number 4 is the only one standing to the west of the plant, in an area characterized by PCDD/F depositions higher than the ones detected eastside (Ragazzi et al., 2012).

CONCLUSIONS AND PERSPECTIVES
This study confirmed that conifer needles are effective proxies of recent pollution history.The temporal trend in the concentrations of pollutants characteristic of steelmaking plant emissions can be used to monitor the efficacy of the improved abatement system.The concentrations of organic pollutants should not be considered as an indicator of pollution as they were comparable to those measured in a sample coming from a pristine location in Alaska, where no industrial plants or other evident sources of pollution were present.This generally positive trend as concerns the reduction of the pollution levels is also confirmed by other data referring to the same environmental context, obtained through conventional sampling methods.
Fe and Cu concentrations were confirmed to be not a good and conclusive marker of pollution from EAF plants, for the interference from traffic emissions.
ICP-AES data, used to identify the concentration trends, were cross checked with particle morphology and To fully assess the efficiency and reliability of conifer needles as passive samplers of PM, a thorough and systematic research of their trapping mechanisms, kinetics and evolution with time is definitely required, especially as we observed a different temporal trend in the concentration of inorganic and organic pollutants.

Fig. 1 .
Fig. 1.Schematic map of the sampling sites with respect to the steelworks and graphs with the concentrations of main elements obtained from ICP-AES.

Table 1 .
Characteristics of the sampling sites.

Table 2 .
PCDD/F and PCB concentrations in needles.An intermediate line and the last line of the table show the values of TEQ of the sample, calculated using the TEF of each congener.Values in Italics are those concentrations below the instrumental detection limit (LOD), which were considered to be half of LOD.2011 values in Bold are the ones higher than the respective ones for 2005.Underlined 2011 values are the ones lower than the respective ones for 2005.