Use of Chemical Signature to Trace the Impact of Emission Sources

The composition of organic fraction associated to particulate emissions depends on their nature as well as on contour conditions. Therefore, many Authors have investigated the chemical signature of airborne particulate matter and dusts with the goal of identifying the pollution sources and assessing their impact on the environment and health. Usually, Authors use three complementary tools for this goal; they are specic source markers, concentration ratios of pairs of congeners, and percent distributions of homologues within a group. After the presentation of the state-of-the-art about non-polar aliphatic (alkanes and alkenes), aromatic (PAHs, Nitro-PAHs) and polar (fatty acids, organic halides, polysaccharides) compounds associated to emissions, this paper provides new information with regard to chemical signature non-polar fraction, suitable to trace the impact of sources on airborne particulate matter and settled dust. Non-polar organic fraction comprises short/medium-chain alkenes and alkanes (with carbon numbers ranging from 12 to 23), which display distinct relative abundances in petrol-derived exhausts, microorganism residues and high vegetation leaf debris. Meanwhile, long-chain alkanes associated to tobacco smoke show a peculiar iso/anteiso/normal homologues ngerprint as well as n-hentriacontane percentages higher than other emissions. Based on this particular alkane distribution, two indexes (ATSR and AICR) have tentatively identied and tested though comparing their rates in some sets of particulate samples. Until now, the study of molecular signature has overall limited to qualitative purposes and seldom exploited to achieve quantitative estimates of contributions of sources to air pollution. Future investigations will reach this goal through further clarifying the nature and behavior of organic contaminants associated to airborne and settled particulate matters.


Introduction
The release of toxicants in the atmosphere has recognized to induce harmful effects on humans and injure the environment. Therefore, the knowledge of nature, amount and land spread of emissions appears mandatory if strategies aimed at mitigating the toxicants impact must be implemented [Albaiges et al 1984, Bascom et al 1996 At this regard, organic substances associated to particulate matter (both aerosols and dust settled on surfaces and soil) are important tools of investigation and at the same time subjects of scienti c challenge, due to their complexity and variety. Organic substances are classi ed in a number of groups displaying very different properties like acidity/alkalinity/neutrality, grade of polarity, water and n-octanol solubility, vapor pressure, inertness vs. oxidants and light-They are linear and cyclic aliphatic compounds, polycyclic aromatic hydrocarbons, acids, amines, carbonyls, halides, organic sulfates and phosphates, etc.. Many of them, as peculiarly associated to sources, have identi ed as tracers of living organisms, natural phenomena and man's activities.
Unfortunately, only in a handful of happy cases one substance is unequivocally typical of one emission, so that it allows revealing and even valuing the impact of that source on the environment. More frequently, many chemicals are associated to many emissions, and the compound peculiarity is missed, nevertheless, in this case the composition per groups and the distribution pattern of congeners within organic groups can aid in the goal of identifying the pollution sources. For instance, biofuels are usually richer of esters than fossil fuels, and the opposite occurs for polycyclic aromatic hydrocarbons (PAHs) [ Instead, in the case of dusts the impact of atmospheric reactivity is by far more important, because ne particles need time to settle on surfaces and the collection period of samples usually lasts ≥15 days.
The rst approaches to identify the emission sources of organic toxicants by means of molecular signatures have made in late twentieth century [Daisey et al., 1986, Harrison et al., 1996. Focus was addressed overall to alkanes, PAHs and nitrated derivatives (NPAHs), but some investigations regarded fatty acids, halides (e.g. dioxins and furans PCDD/Fs, polychlorobiphenyls PCBs, polybromodiphenylethers PBDEs), polysaccharides (levoglucosan, mannosan), sterols (cholesterol, stigmasterol, sitosterol) and triterpenols (amyrins). Due to intrinsic limits of the simple molecular signature use, other tools have adopted preferably to trace the pollution sources impact, e.g. PCA, nevertheless, this the study of chemical ngerprints remains proper for the preliminary analysis of i.e.: i) chemicals overall released outdoors and moved into the interiors through building openings and ventilation devices (e.g., hydrocarbons present in motor vehicle exhausts), ii) substances released indoors and outdoors at broadly analogous levels (e.g., nitrogen oxides, psychotropic substances), and iii) compounds predominantly released by typical indoor sources (e.g., deodorants, plasticizers). Besides, chemicals released indoors meet up environmental contours heavily different from open air, e.g., temperatures are much less varying, ozone is much less and surfaces are much larger. All of them change the chemistry of interiors, the lifetime of substances and the gas/condensed phase equilibria. Although human exposure to toxicants occurs predominantly in interiors, legislation dealing with indoor pollution is still insu cient aside of workplaces, while extensive investigations concerning the air quality of interiors deal overall with tobacco smoking.
This paper provides an overview of investigations carried out by our team, aimed at identifying the molecular ngerprints of organic fraction of airborne particulate matter and deposition dust, potentially suitable to draw information about the sources of pollution and the respective impact on the environment. Three major categories of ngerprints will discussed, namely: i) individual tracers, ii) diagnostic concentration ratios, and iii) homologue percent distributions within groups.
2. The State-of-the-art Of Research About Molecular Signatures Of Pollution Sources 2.1. General features of emission pro les. Apart from environmental contour, three key factors in uence the composition of emissions resulting from chemical analysis. They are: i) the operating conditions of emission source, including the kind of fuel, temperature of exhausts, type of abatement devices, ii) the procedure of collection (which includes vapors, particles or both), and iii) the methodology adopted to process samples and determine chemical composition. These factors make hard to assign thorough emission factors to chemicals released from the sources, and to assess precise values to concentration ratios or percentages of substances considered in the molecular ngerprints [Tobiszewski andNamiesnik 2012, Cecinato et al 2014]. Investigations undertaken with different methodological approaches can lead to not comparable results. For instance, the n-alkane pro le detected in organic particulate matter depends on temperature occurring during sampling, which in uences the loss rates of the most volatile compounds. Therefore, the study of n-alkanes percent distribution is mostly restricted to high molecular weight homologues, namely to hydrocarbons with carbon number ≥25 (C 25 ). As for PAHs, compounds with similar vapor pressures and/or high molecular weights are preferred to choose the ratios of concentration to select as diagnostic of sources. Anyway, in our work the organic matter composition was determined by applying similar methodologies for all of emission exhausts, suspended particulates and dusts. The samples have rst analyzed according to state-of-the-art of literature, and then chemical characterization has improved to draw further insights about the nature of sources.
2.2. n-Alkanes. Non-polar fraction of particulate organic matter POM includes numerous groups. They are alkanes, alkylated mono-aromatics and biphenyls, alkenes, branched and cyclic-aliphatic hydrocarbons. Among them, attention has paid overall to n-alkanes (linear homologues), alicyclic compounds have investigated as tracers of petrol products, and methyl substituted alkanes as markers of tobacco smoking.
The saw-tooth distribution of high-molecular-weight n-alkanes has been associated to high trees [Simoneit and Mazurek 1982, Alves et al 2001, Rabhi et al 2018]. Biogenic synthesis leads to preferential generation of even C-numbered fatty acids, then, acids tend to loose CO 2 through the natural process of decarboxylation, and form odd-C numbered n-alkanes (or alkenes, in the case of unsaturated precursors) as nal products. Since this phenomenon is more evident for long-chain homologues, the most used parameter to rate the impact of (high) vegetation is Carbon Preference Index (CPI 25  For instance, we calculated CPI 25 equal to ~1 and exceeding 10, respectively, nearby a highway in the Algiers metropolitan area (Fig. 1A), and in a forest area of Biskra province, Algeria (Fig. 1B), instead, usually a combination of the two distributions has observed in a city garden of Rome, Italy (Fig. 1C).
Other indexes have investigated to put in evidence the impact of vegetation. They are: i) the homologue (C max ) corresponding to the maximum concentration within the n-alkane percent distribution, and ii) the cumulative percentage attributable to natural waxes (NW%) [36,63]. As for C max , motor vehicle exhausts show the prevalence of short and medium chain hydrocarbons (<C 24 ), whilst the leaf debris of high trees is characterized by the predominance of n-C 29 /n-C 31 . The two distinct behaviors have pictured in Fig. 1A and 1B, where the maximums correspond to tricosane (C 23 ) and nonacosane (C 29 ), respectively.
The NW% value is provided by the formula: Where each term at numerator is set equal to zero whereas the actual rate results <0.
Besides, carbon preference indexes analogous to CPI 25 have formulated considering larger n-alkane ranges (e.g., C 11 ÷C 36 ) or the only light homologues (<C 25 ), e.g. C 15 -C 24 . Although CPIs based on light hydrocarbon sequences are biased due to compound volatility, these indexes allow investigating the possible impact of marine biota (algae, plankton) and microorganisms when they are combined with the presence of isoprenoids and with C max corresponding to C 15  composition pro le of organic particulates shows one or two humps of "unresolved mixture" which accompany the n-alkanes sequence, the rst one, comprised of light hydrocarbons, has associated with gasoline and diesel oil residues, and the second, of heavier components, to lubricating oils (see a and b humps in Fig. 2). Noteworthy, the vehicle emission pro le depends on the engine working conditions, the relative importance of hump raises at unregulated driving regimens, e.g. during cold starts.
Both tobacco plant leaves and tobacco smoke fumes exhibit a particular percent pro le of the non-polar organic fraction. In fact, long-chain odd iso-alkanes and even anteiso-alkanes are much richer than in other emissions [Kavouras et al., 1998]. Besides, the normal hentriacontane (nC 31 ) is predominant if compared to nC 29 and nC 33 homologues. This molecular signature has observed in tobacco smoke chambers and in interiors heavily contaminated by smoke.  Figure 3B refers to Leonessa, a mountain town of Central Italy, the dimethyl/ethyl phenanthrenes ngerprint is analogous to that of Rome but holds some differences. Another molecular signature of dimethyl/ethyl phenanthrenes, reported in Fig. C, belonged to a second sample of Leonessa ( Fig. 3C), although collected in the same year period, it was almost identical to that characterizing the wood burning. Noticeably, wood combustion is plenty exploited in that region to heat houses, cook food and dispose brushwood. The percent distribution in Fig. 3C is similar to that found in Leonessa during the summer season (Fig. 3D). For completeness of information, Figure 3E reports the GC-MS trace of retene recorded in the second winter sample of Leonessa. Attention has paid overall to a list of NPAHs whose contribution to toxicity of air particulates has even estimated, they are nitrated derivatives of naphthalene, uorene, anthracene, uoranthene, pyrene, benz[a]anthracene and chrysene.
However, NO 2 -position isomers associated to PM do not correspond to those mainly affecting emissions.
In particular, 2-nitro uoranthene and 2-nitropyrene are commonly absent in exhausts, but exist as products of in-situ reactions developing in the atmosphere (see Figure 4A The molecular signature nitro uoranthenes and nitropyrenes has used to parameterize the relative importance of direct emission and action of oxidants. Taking in account the nitration rate of precursors reacting with OH radicals and NO 2 , it has suggested that photochemical reactivity is more important than vs. direct emission when the 2-NFA/1-NPY ratio exceeds the value of 5 [Pitts et al, 1985].

Materials And Methods
To improve the knowledge of molecular signature of particulate organic matters, we investigated the chemical composition of groups of samples collected in Italy and abroad. The samples were collected from air both indoors and outdoors (airborne particulates and settled dusts). All of them have analyzed in the frame of cooperative research programs with Regional Agencies for Air Pollution Control, Italy Organic fraction has recovered from the substrate through extraction in ultrasonic bath, using a dichloromethane/acetone mixture. After reduction close to dryness, the extracts have fractionated through chromatography on neutral alumina or silica gel column; three classes of polarity have separated through eluting the column with isooctane, isooctane/dichloromethane and dichloromethane/acetone, in sequence. Instrumental analysis was performed by applying GC-MS methods based on (DB5-MS type) capillary column separation, electron-impact ionization of analyte molecules and total ion current (TIC) or selected-ion (SIM) recording. Injection has operated in split-less mode, chromatographic runs has conducted in gradient of temperature up to 290°C, and peak recognition achieved by means of characteristic molecule and fragment ion current signals, coupled with peak retention times. The internal standard method, based on the use of perdeuterated homologues as reference of native compounds, has adopted for quantitative purposes.

Results And Discussion
Until now, scarce attention has paid to unsaturated hydrocarbons. Nevertheless, in the light range of nonpolar fraction of POM (corresponding to hydrocarbons with 12 up to 20 carbon atoms in the molecule, namely C 12 ÷C 20 ) we could distinguish three distinct ngerprints. They were: i) the predominant occurrence of n-alkanes (n-alkenes negligible); ii) the prevalence of n-alkenes (with high dodecene/dodecane and tetradecene/tetradecane ratios, and low octadecene/octadecane and eicosene/eicosane ratios); and iii) a mixed behavior. The three ngerprints have shown in Figure 5. At our knowledge, no exhaustive explanation of the three distinct patterns has provided. Even n-alkene homologues presumably are directly released by unknown sources or originate from the twofold decarboxylation of unsaturated even a,w-diacids, but further investigations are needed to link these ngerprints with anthropogenic and biogenic emissions.
As for tobacco smoking, both in airborne particulate and dust collected at a frequent smoker's home in Rome, Italy, anteiso C 30 , C 32 and C 34 alkanes were more abundant than the respective normal homologues. Besides, the concentration of normal C 31 exceeded the average of nC 29 and nC 33 . Both molecular signatures are similar to those found in smoke chambers ( Figure 6).
Instead of identifying one only ratio among the anteiso/iso/normal alkane sequence, we computed all concentration ratios of ATSR and AICR formulas over the C 29 -C 34 range in order to reduce uncertainty associated to small changes in emission pro les of other possible sources. To assess the correctness of this approach, both ATSR and AICR rates have evaluated in airborne particulates and dusts collected both at urban and rural sites. The results have shown in Table 3.
ATSR values ranged from 0.05 to 1.30, with the minimums corresponding to rural sites and the maximums found in locations heavily affected by tobacco smoke. ATSR and AICR values calculated in the case of outdoor samples put in the evidence the impact of tobacco smoking, which was in agreement with the ubiquitous occurrence of nicotine in the city air [Rabhi et al 2018 and references herein]. In general, dusts showed ATSR and AICR values lower than airborne particulates; that is in accordance with the nature of dust, which includes coarse grains of various origin, whilst tobacco smoke is comprised of ne and ultra-ne particles accounting for only a fraction of total particulate mass. Noticeably, ATSR and AICR values in the dust and at the balcony of smoker house (see the sixth column of Table 3) were quite high, con rming the importance of tobacco smoke in interiors and even at open air. Though still insu cient to draw semi-quantitative information about the contribution of tobacco smoke to environmental pollution, the molecular signature of long-chain alkanes, seems more attractive than nicotine and its derivatives, which are more volatile and degradable. Presumably, it can suitably combine with the monitoring of nicotelline.

Conclusion
Molecular signature of organic contaminants affecting airborne particulates and dusts comprise individual markers and, more often, distribution patterns within the homologue groups. Both tools provide preliminary but suitable information about nature of emissions and with regard to their health impact on environment. Further investigations are necessary to elucidate the sources of chemical ngerprints newly observed. In particular, the progress in knowledge of molecular ngerprints will help investigators in applying more sophisticated approaches (e.g., principal component analysis, source factorization modelling) to assess the relative importance of emissions. That will help to optimize the strategies aimed at controlling air pollution and mitigating the toxicants impact on humans and environment.

Declarations
Author Contributions. This paper originates from speci c contributions of the Authors. Data availability. The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request. Anyway, those not directly produced by personal investigations of Authors are available in the bibliography cited in the paper.
Con icts of Interest. The authors declare they are not under any con ict of interest condition. A (1997