masclet1995

Prévia do material em texto

<p>Journal of Atmospheric Chemistry 22: 41-54, 1995. 41</p><p>© 1995 Kluwer Academic Publishers. Printed in the Netherlands.</p><p>Emissions of Polycyclic Aromatic Hydrocarbons</p><p>by Savanna Fires</p><p>P I E R R E M A S C L E T 1, HI~LI~NE C A C H I E R 2, C A T H E R I N E L I O U S S E 2, and</p><p>H E N R I W O R T H A M 3</p><p>1Laboratoire d'l~tudes des Syst~mes Atmosphdriques Multiphasiques, ESIGEC, Universit6 de Savoie,</p><p>BP 1104, 73 000 Chamb6ry, France</p><p>2Centre des Faibles Radioactivit~s, CNRS-CEA, rue de la Terrasse, 91198 Gif-sur-Yvette, France</p><p>3Laboratoire de Physico-Chimie de I'Atmosph&e, Universitd Paris VII, 2 Place Jussieu, 75251, Paris</p><p>Cedex 05, France</p><p>(Received: 15 June 1992)</p><p>Abstract. Although Polycyclic aromatic hydrocarbons (PAH) are known as anthropogenic compounds</p><p>arising from the combustion or the pyrolysis of fossil fuels, they may be also emitted by the combustion</p><p>of vegetation. A field study was caixied out in January 1991 at Lamto (Ivory Coast) as part of the FOS</p><p>DECAFE experiment (Fire Of Savanna). Some ground samplings were devoted to the qualitative and</p><p>quantitative characterization of atmospheric emissions by savanna fires during prescribed burns and</p><p>under background conditions. Specific collections for gaseous and particulate PAHs have shown that</p><p>the African practice of burning the savanna biomass during the winter months is an important source</p><p>of PAHs. These compounds are emitted mainly in gaseous form but a significant fraction, essentially</p><p>heavy PAHs, is associated with fine carbonaceous particles and can therefore represent a hazard for</p><p>human health, since some of these compounds are mutagenic and carcinogenic. Twelve compounds</p><p>were identified during the fire episodes and in the atmospheric background. The total concentration in</p><p>the fires is of the order of 10 ng m -~ for the gas phase and from 0.1 to 1 ng m -3 in the aerosols. In the</p><p>atmospheric background the mean concentrations are regular, 0.15 ng m -3 and 2 pg m -3, respectively.</p><p>These concentrations are comparable with what is observed in European rural zones. The particulate</p><p>emissions of PAHs by the savanna fires are distinguished by the abundance of some compounds which</p><p>can be considered as tracers, although they are also slightly emitted by fossil fuel sources. These com-</p><p>pounds are essentially pyrene, chrysene and coronene. In the gas phase, although no individual PAH</p><p>may be considered as specific of the biomass combustion emissions, the relative abundances of the</p><p>main PAHs are characteristic of the biomass burning. The concentrations of pyrene and fluorene are</p><p>always predominant; these compounds could be considered as characteristic emission products of</p><p>smoldering and flaming episodes, respectively. In the background the PAH composition shows that in a</p><p>tropical region the air consists of a mixture coming from the various sources, but the biomass combus-</p><p>tion is by far the most important source.</p><p>The fluxes of total PAH emitted by savanna biomass burning in Africa were estimated to be of the</p><p>order of 17 and 600 ton yr -I, respectively, for the particulate PAHs and the gaseous PAHs, respec-</p><p>tively.</p><p>Key words: biomass burning, African region, gas phase, aerosols, soot carbon, total carbon, tracers,</p><p>atmosphere.</p><p>1. Introduction</p><p>In urban areas, PAHs are compounds which arise essentially from the combustion</p><p>of fossil fuels (fuel oil, coal, petrol and diesel fuel) (Nikolaou et al., 1984; Masclet</p><p>42 PIERRE MASCLET ET AL.</p><p>et al., 1986) or the pyrolysis of wood (Ramdalll, 1983). These anthropogenic</p><p>sources have been carefully studied because of the carcinogenic and mutagenic</p><p>properties of many PAHs, particularly those with five or more rings (Finlayson-</p><p>Pitts and Pitts, 1985). These compounds are found in both the gas and the particu-</p><p>late phases; the latter is less abundant but constitutes a real hazard since PAHs are</p><p>mainly adsorbed on free particles (submicron) which constitutes the major breath-</p><p>able fraction of the aerosol. PAHs are present in all media: air, water and soil and</p><p>occur worldwide, particularly in the northern hemisphere. Recently seven particu-</p><p>late PAHs were detected in samples collected in the Arctic region atmosphere</p><p>(Jaffrezo et al., 1992). These compounds when injected into the atmosphere,</p><p>undergo a long-range transport (Bjorseth and Olufsen, 1983; Masclet et aL, 1988).</p><p>Physicochemical transformations may occur during transport, modifying the PAH</p><p>profile between the source and the receptors site. Two phenomena are involved in</p><p>these modifications: the gas-particle equilibrium, which depends primarily on the</p><p>temperature, and the photochemical reactions with atmospheric oxidizing agents,</p><p>particularly the OH radicals (Biermann et al., 1985; Arey et al., 1986). While the</p><p>anthropogenic sources are well documented this is not the case for sources due to</p><p>the combustion of the biomass. Also, the domestic habits in tropical countries con-</p><p>sisting of cooking food over open fires and the cooking on barbecue in developed</p><p>countries, result in a significant production of carbonaceous pollutants including</p><p>PAHs. However, to our knowledge, there is no study investigating the production of</p><p>these hydrocarbons by savanna fires.</p><p>As part of the FOS DECAFE programme, a field study was carried out at</p><p>Lamto in the Ivory Coast (60 km north of Abidjan) during January 1991. More</p><p>details on the experiment and the collection strategy are given in this issue by</p><p>Cachier et al. (1995). The aim of this study was to determine qualitatively and</p><p>quantitatively the emission of PAHs by the combustion of the savanna vegetation</p><p>and find tracers able to distinguish this type of combustion from the fossil fuel</p><p>sources. The characteristics of savanna fires allow this type of combustion to be an</p><p>original source of organic and inorganic matter. The PAH load was determined in</p><p>both the aerosols and the gas phase. We also tried to estimate the PAH flux in rela-</p><p>tion to other gases or elements. Normalization with CO and CO2, and subsequent</p><p>emission factor (El=) calculations were done taking into account all the carbon</p><p>emissions during the savanna fires (Crutzen and Andreae, 1990; Kadowaki, 1990).</p><p>2. Experimental</p><p>The samples corresponding to savanna fires were taken under two different condi-</p><p>tions:</p><p>on plots of limited area (about 1000 m 2) with different savanna grass types.</p><p>Fires were lit in the direction of the slope or against it. These conditions cor-</p><p>respond to samples 2, 4, and 6,</p><p>EMISSIONS OF PAH BY SAVANNA FIRES 43</p><p>during large fires affecting about 10 km 2 in the savanna consisting of a mix-</p><p>ture of grasses and small shrubs. These fires propagated naturally over vast</p><p>areas and are called 'big fires' corresponding to samples 8 and 10. Sample 3</p><p>was taken during a violent down slope plot fire.</p><p>In all cases, the sampling equipment, including the HIVOL sampler was close to</p><p>the firefront (about 10 m), and the sample duration of the order of 30 min.</p><p>Background samples were taken during 18-24 h periods before the prescribed</p><p>fires (sample 1) or one or two days after the extinction of the local fires and disper-</p><p>sion of the pollutants produced by these fires (samples 7 and 9). Nevertheless,</p><p>during the collection of these samples, regional fires were ignited over a part of the</p><p>western African region, as shown by aerial photographs.</p><p>We used a HIVOL sampler (SIERRA) working with a flow rate of about 30 m 3</p><p>h -1. For the background atmosphere about 600 m 3 of air were sampled, whereas</p><p>15 m 3 were enough for the fire plots samples. For the purpose of analysis of both</p><p>gaseous and particulate PAH, this sampler was fitted with a 8"x 10" Whatman</p><p>GF/A glass fibre filter previously washed with organic solvents (mixture of cyclo-</p><p>hexane and dichloromethane) and with a metal cartridge filled with Amberlite</p><p>XAD2 adsorbant previously cleaned under the same conditions. As previously</p><p>shown by Bresson et al. (1984), this procedure satisfactorily minimizes collection</p><p>artefacts for both gaseous and particulate PAHs.</p><p>The samples were stored in the dark, wrapped in aluminiu m foil, then sealed in</p><p>plastic bags and stored in a refrigerator since their arrival. They were analysed</p><p>within 6 weeks following sampling. The procedure has been published elsewhere</p><p>(Masclet et al., 1987). Basically, it may be described as follows: the samples are</p><p>extracted for 3 h with 300 mL of a mixture of dichloromethane and cyctohexane</p><p>(2/1), then evaporated to dryness; the dry extract is taken up in 1000 mL of</p><p>methanol (fluorescence quality) for HPLC analysis on a Spectra-Physics 4270</p><p>apparatus fitted with a Vydac C18 column (25 cm long - 5 gm). A ternary gradient</p><p>of acetonitrile-methanot-water is used to elute 16 of the major PAH in 22 min.</p><p>Three wavelengths are used for fluorimetric detection, providing a good compro-</p><p>mise between selectivity and sensitMty. The compounds are identified by com-</p><p>paring their retention times w4th those of standards and in some cases by the</p><p>stopped flow fluorescence spectrum. The concentrations are calculated by an inter-</p><p>nal standard method, using methyl-9-anthracene, a compound which is not present</p><p>in the emissions.</p><p>Particulate PAHs concentrations were compared to the bulk amount of particu-</p><p>late carbon (total carbon) and soot carbon (also referred as black carbon). Soot and</p><p>total carbon analyses were performed on aliquotes of the glass-fiber filters. Details</p><p>on the analytical procedure which is based on the thermal separation of black and</p><p>organic carbon and the subsequent coulometric titration may be found in others</p><p>papers (Cachier et al., 1993).</p><p>44 PIERRE MASCLET ET AL.</p><p>3. Results and Discussion</p><p>Tables Ia and Ib give the total PAIl concentrations found in both the gas and the</p><p>particulate phases, for the plot fires, the big fires and the atmospheric background.</p><p>In this tables, data on the most abundant PAHs only are given, this fraction forming</p><p>about 95% of the total nonsubstituted PAHs in mass. As expected, for the pre-</p><p>scribed fire samples (plots fires and big fires) the concentrations observed are</p><p>scattered but, the concentrations appear to be very homogeneous in the back-</p><p>ground atmosphere. The PAH concentrations in the fire plume samples varied</p><p>approximately from 400 to 40 000 pg m -3 for the gas phase and from 70 to 700 pg</p><p>m -3 for the particulate PAH. For the prescribed fires, sampfing conditions may</p><p>Table I. Gaseous (a) and particulate (b) PAH concentrations for the fires and in</p><p>the atmospheric background (Lamto, Ivory Coast)</p><p>Samples</p><p>P A H 1 2 3 4 6 7 8 9 10</p><p>( ~ Gaeous P A H</p><p>Concen~a t ions (pg m ~ )</p><p>NAP 60 3050 14000 159 68 80 4570 49 1380</p><p>FLU 30 1320 10800 114 37 46 4260 31 1300</p><p>PHE 110 2130 5910 82 88 81 1795 37 320</p><p>ANT 2 92 635 6 - 3 185 2 65</p><p>PYR 15 307 1920 63 214 33 440 25 151</p><p>FLA 4 132 1000 16 - 6 265 5 73</p><p>Z P A H 221 7030 34300 440 407 249 11510 149 3290</p><p>(b) Particulate P A H</p><p>Concentrations (pg m -3)</p><p>FLA 0.36 - 175 - - 0.51 38 0.45 15,5</p><p>PYR 0.92 - 121 - - 0.84 25 0,73 11</p><p>CHR 0.23 - 75 - - 0.19 14 0.24 11</p><p>BaP 0.02 - 12 - - 0.03 2.2 0.02 4</p><p>BeP 0.10 - 32 - - 0.14 9 0.11 14</p><p>BbF 0.07 - 29 - - 0.06 8 0,04 2.7</p><p>BkF 0.04 - 19 - - 0.04 5,8 0.02 2.4</p><p>INP 0.17 - 29 - - 0.16 8 0.15 4</p><p>BghiP 0.20 - 65 - - 0.17 10.7 0.19 5</p><p>COR 0.29 - 92 - - 0.28 16 0.25 8</p><p>Z PAH 2.40 - 649 - - 2.42 137 2.20 78</p><p>NAP: Naphthalene; FLU: Fluorene; PHE: Phenanthrene; PYR: Pyrene; FLA:</p><p>Fluoranthene; ANT: Anthracene; CHR: Chrysene; BaP: Benzo a Pyrene; BeP:</p><p>Benzo e Pyrene; BbF: Benzo b Fluoranthene; BkF: Benzo k Fluoranthene; INP:</p><p>IndenoPyrene; BghiP: Benzo ghi Perylene; COR: Coronene.</p><p>Samples 1, 7 and 9: atmospheric background.</p><p>Samples 2, 4 and 6: plot fires.</p><p>Samples 3, 8 and 10: intense fires.</p><p>EMISSIONS OF PAH BY SAVANNA FIRES 45</p><p>differ and for this reason the absolute concentrations vary widely; this is mainly</p><p>due to variations of the position of the sampler with respect to the fire (sampler</p><p>placed downwind or upwind, distance to the fire front, etc.). Furthermore, the vege-</p><p>tation and combustion variability, from one experiment to another, are likely to be</p><p>important factors. The combination of these parameters implies to be very careful</p><p>about the crude concentration results and requires normalization of the PAH emis-</p><p>sions with a reference gas emitted simultaneously. Some sampling difficulties in the</p><p>prescribed fire plumes did not permit the use of our whole set of samples for com-</p><p>parison purposes. Also, the samples displaying gaseous PAH content of the same</p><p>order as in the background atmosphere were rejected; only samples 2, 3, 8, and 10</p><p>were kept for discussion.</p><p>The reproductibility of the background PAH concentrations for both the gases</p><p>and the particles allows to characterize the local ambient air in tropical savanna</p><p>regions. These atmospheric concentrations are about 150-200 pg m -3 and 2.0 pg</p><p>m -3 for the gas and the particles respectively. These values are to be compared with</p><p>those found during similar studies performed in a rural area (forests in the centre of</p><p>France) and in Corsica (Western Mediterranean sea) (Masclet et al., 1988; Pistiki-</p><p>poulos et al., 1990a). Table II gives values for the three situations. It can be seen</p><p>that the African savanna site of Lamto is as polluted as continental temperate</p><p>regions and more than a marine region severely affected by anthropogenic inputs.</p><p>3.1. Gas Particle Distribution</p><p>The gas to particle ratio is about 35 for the big fires. This value is much smaller</p><p>than that generally observed at emission (of the order of 100 for coal-fired power</p><p>stations, for example). The emissions due to savanna biomass burning appear</p><p>therefore rich in particles and organic compounds attached on particles. In the</p><p>background atmosphere the mean gas-particle ratio is 100, which is greater than</p><p>that observed in a rural area in Europe or in a remote zone (Mediterranean sea for</p><p>example) where the ratio is close to 30 for an average temperature of about 20 °C.</p><p>However, for an ambient temperature equivalent to that of the tropical region</p><p>(Lamto, 28 °C), calculations show that the gas particle ratio would be close to 165</p><p>(Pistikipoulos et aL, 1991a). Although the thermodynamical trend in the gas to par-</p><p>Table II. Comparison of PAH concentrations in an urban area, in a</p><p>remote marine atmosphere and in the atmosphere of savanna region</p><p>(sample 9)</p><p>PAH (pg m -3) Rural Marine Savanna background</p><p>area atmosphere atmosphere (sample 9)</p><p>gaseous PAH 184 52 149</p><p>particulate PAH 6.1 2,1 2.2</p><p>46 P I E ~ MASCLET ET AL.</p><p>ticle PAH ratio values is observed in the background atmosphere, the atmospheric</p><p>load of particles is slightly higher at the Lamto site. In all cases the temperature</p><p>determines the gas-particle distribution. Consequently the relative abundance of</p><p>the gas phase PAH in the atmosphere is due to a shift in the gas-particle equi-</p><p>librium in relation with ambient temperature. An extreme opposite situation is</p><p>observed in polar regions (Jaffrezo et al., 1992) where the gas phase is practically</p><p>nonexistent.</p><p>3.2. Gas Emission Profiles</p><p>Figures 1 and 2 show the gas emission profiles for the plot fire and the background</p><p>samples. The plot fire samples display some significant differences in their profiles</p><p>which may be inferred to changes in the combustion process. In sample 4 which has</p><p>integrated a long period of smoldering, naphthalene and fluorene are predominant</p><p>(36 and 26% respectively). Conversely sample 6 is enriched in pyrene with a rela-</p><p>tive abundance of 53%. The gaseous PAH pattern therefore provides a signature</p><p>for the two types of fire; the very light PAH such as naphthalene and fluorene could</p><p>be characteristic of a smoldering phase, whereas pyrene (a heavy PAH with four</p><p>rings) could correspond to flaming episodes. For the 'big fires' samples the PAH</p><p>behavior is very similar, as shown by samples 3, 8, and 10. The distributions are the</p><p>same with naphthalene (40%) and fluorene (36%) predominant; pyrene</p><p>is not</p><p>absent but less important (6%).</p><p>Therefore, as the study of the gaseous PAHs shows that the distinct types of fire</p><p>may be traced by different PAHs, we have compared the PAH profiles in the fires</p><p>with those found in an urban area (Paris) or a marine area (Mediterranean) (Figure</p><p>3). Naphthalene is abundant in all the samples and thus cannot be used as a specific</p><p>tracer for savanna fires. On the other hand, fluorene and pyrene are particularly</p><p>abundant in the gaseous PAH mixture of the savana fires and may be considered as</p><p>specific tracers, although they are not totally absent from others source emissions.</p><p>So, fluorene could be a tracer of the smoldering and pyrene a tracer of flaming</p><p>phase of the savanna combustion processes.</p><p>3.3. Particulate Emission Profiles</p><p>We do not have the data for the plot fires. As obtained for the gaseous PAH, the</p><p>distribution is rigorously constant for the three large scale fire samples, with the</p><p>same six major compounds: anthracene, fluoranthene, pyrene, chrysene, benzo (e)</p><p>pyrene and coronene. Some of these compounds cannot be considered as specific</p><p>tracers for the biomass combustion since they are found in any type of combustion</p><p>(industry', power stations, petrol or diesel engines). This applies to anthracene,</p><p>fluoranthene and coronene which are abundant in the emissions of all anthropic</p><p>sources. On the other hand, chrysene and pyrene which are found to be particularly</p><p>abundant in the biomass burning smokes could trace the aerosols produced by this</p><p>EMISSIONS OF PAH BY SAVANNA FIRES 47</p><p>O/o 6%</p><p>. . . o i I Z Z 2 ~ _ \ X I ~ t t l I I t t \</p><p>n t / o / j ] a . ' _ L ' . ' ~ _ \ R I I I I I i ! ! ! I\</p><p>/~.~'._'~'~'.~\\I I I I I I I II I\</p><p>I~'._'~'~'~'~I I I I I I I I I I~</p><p>I'~',.'~'~'~_'~III i I I I I I I I I</p><p>I±'~' ~ I I I I I I I I I I</p><p>\ \ x i I I I t t I i /</p><p>\ \ \ \ \ \1 I I I I I I I / \ \ \</p><p>\\\</p><p>3 1 7 0</p><p>/ ' ~ ' ± ' ~ ' _ , _ ' \ \ I I I I i I t I I X</p><p>/ ~ . ~ T Z Z ~ \ \ m I j i I I I i I i \</p><p>/_C..,.".,.'.,.L,? . ,_ \ \ ! ~ t t i i I I I I~.</p><p>L f_ ~ ' _ C S . , _ _ ~ t I 1 I I I 1 i I ' ,</p><p>\ \ " k l I 11 t t I I'14o o ,</p><p>\ \ \ \ \ \ t I l i t I I / /0</p><p>3 7 7 . ~</p><p>I ~ N a p h</p><p>F l u</p><p>P h e</p><p>f - - 1 An,</p><p>F l a</p><p>I P y r</p><p>° 2' 5 ~ .</p><p>L + + + ' ~ ~ m , I , , ~ , I 4 0 / ° 2</p><p>/ ' - , , ~ ÷ + X I I I L i I I I I ~\</p><p>4 i I l l l / ~ ;\111, , , , , 1 \</p><p>I ' , "-,.3 I i I I I I I I I</p><p>\ \ l i t E I I l l j</p><p>~ , \ \ N t I t I t I I I</p><p>\ \ \ \ \ l ~ t l t l I /</p><p>\\\ \\\I J I I I I/</p><p>\ \ \ \ ~ I t l 11</p><p>Fig. 1. Gaseous PAH emissions profiles for the big fires (Lamto, Ivory Coast) (samples 3, 8 and 10);</p><p>evidence for a mean savanna fire pattern.</p><p>48 PIERRE MASCLET ET AL.</p><p>27o 77o</p><p>50°,/o</p><p>~ 1 3 7 o</p><p>. 3 2 ° / o</p><p>19°/o</p><p>| Naph</p><p>Flu</p><p>Phe</p><p>Ant</p><p>Fla</p><p>Pyr</p><p>17"/</p><p>t |</p><p>25</p><p>Fig. 2. Gaseous PAH profiles for the atmospheric background in savanna areas (Lamto, Ivory Coast)</p><p>(samples 1, 7 and 9).</p><p>EMISSIONS OF PAH BY SAVANNA FIRES 49</p><p>pg m -3</p><p>300</p><p>250</p><p>100</p><p>Naph FI.</p><p>pg m -3</p><p>30</p><p>!1!!</p><p>Naph Flu</p><p>pg m'3</p><p>J</p><p>4 5</p><p>3sl</p><p>251</p><p>1 5 '</p><p>5</p><p>Phe Ant Fla Pyr</p><p>Phe Ant Fla Pyr</p><p>Naph Flu Phe Ant Fla Pyr</p><p>Fig. 3. Gaseous PAH profiles for three local situations: urban atmosphere (up); marine atmosphere</p><p>(middle) and savanna background atmosphere (bottom).</p><p>source. Coronene is best known as a tracer of vehicles exhaust and is also abundant</p><p>in barbecue wood fires as it has been shown recently (Freeman and Cattell, 1990).</p><p>Therefore its tracer character is doubtful and contrarily to what has been stated</p><p>recently (Pistikipoulos et al., 1990b), coronene should not be used in models as the</p><p>chemical element balance (Duval and Friedlander, 1981).</p><p>The background savanna burning particulate PAHs profile is given in Figure 4</p><p>and compared to profiles obtained for ambient air in an urban atmosphere (Paris,</p><p>1986) and [or a marine atmosphere. These latter profiles are averages calculated</p><p>50 PIERRE. MASCLET ET AL.</p><p>pg.m-3</p><p>120</p><p>80 ,</p><p>40.</p><p>011</p><p>Ant Fia Pyr Chr BaP BeP BbF BkF inPyr Bghi Cor</p><p>Per</p><p>pg.m-3</p><p>1 . 2 !</p><p>0.8</p><p>0.4¢</p><p>0!t I</p><p>Ant</p><p>'1 J</p><p>!</p><p>I</p><p>I</p><p>Fla Pyr Chr BaP BeP BbF BkF InPyrBglzi Cot</p><p>Per</p><p>i</p><p>pg.m-3</p><p>3</p><p>1</p><p>0</p><p>Ant Fla Pyr Chr BaP BeP BbF BkF lnPyrBghi Cot</p><p>Per</p><p>Fig. 4. Particulate PAH profiles for three local situations: urban atmosphere (up); marine atmos-</p><p>phere (middle) and savanna background atmosphere (bottom).</p><p>for a large number of samples (29 for Paris and 14 for Western Mediterranean</p><p>Sea). The three profiles differ significantly. In all, fluoranthene is abundant. In the</p><p>urban atmosphere the compounds which are considered to be typical of vehicle</p><p>exhausts, such as indenopyrene and benzo (ghi) perylene are also very abundant.</p><p>At Lamto the abundance of the pyrene is straightforward; it comes exclusively from</p><p>biomass burning. Coronene, indenopyrene and benzo (ghi) perylene are also</p><p>present. Their origin is mainly biomass burning but it is not possible to exclude a</p><p>EMISSIONS OF PAH BY SAVANNA FIRES 51</p><p>partial contribution from vehicles emissions as the tracers of both types of combus-</p><p>tions are the same. Anthracene and chrysene are also abundant; these two com-</p><p>pounds are found in fire emissions. Finally, another striking point of our biomass</p><p>burning data could be the particularly high benzo (e) pyrene to benzo (a) pyrene</p><p>ratio if compared to the same ratio in urban areas. Nevertheless, it is impossible to</p><p>use this ratio as a tracer of biomass burning because Benzo (e) pyrene is a low reac-</p><p>tive compound and the Benzo (a) pyrene, a high reactive PAH; and thus this ratio</p><p>reflects the distance from the source and the receptor site better than the nature of</p><p>the combustion source.</p><p>The Mediterranean profile is more complex, governed by two main determi-</p><p>nants: the sources and the PAH reactivits: The most abundant compounds are</p><p>clearly of anthropogenic origin: benzo (ghi) perylene predominates; the abundance</p><p>of pyrene could mean that a significant part of the air sampled in the marine atmos-</p><p>phere comes from regions influenced by biomass burning. As the study of the</p><p>Western Mediterranean Sea atmosphere was carried out in February, the intense</p><p>savanna fires prevailing in tropical Africa could have affected air masses reaching</p><p>the sampling site. This hypothesis should be checked carefully by study of the</p><p>meteorological and satellite data regarding the air mass trajectories. Comparison of</p><p>the three profiles shows that the tropical biomass burning may be trace by a typical</p><p>pattern. At this period of the year, the dry season, the background atmosphere in</p><p>Africa is mainly influenced by savanna fires and the profile is close to the profile</p><p>obtained in the fire plumes. Since sources affecting the atmosphere are very near</p><p>the sampling site, the emission pattern is not modified by the reactivity.</p><p>3,4. PAH Emission Factors</p><p>An indirect estimate of the PAH emission factors was calculated through the rela-</p><p>tionship existing between the carbonaceous particle and the particulate PAIl con-</p><p>centrations (PAHp). Table III gives the values of the ratio of the total particulate</p><p>PAH to the soot or total carbon. A variability of both the PAHp to soot carbon</p><p>(PAHp/SC) and PAHp to total carbon (PAH/TC) ratios may be observed for the</p><p>different samples (from 1.2 to 4.7 ~g g-1 for the first ratio and from 0.24 to 0.74 ~tg</p><p>g-1 for the second ratio). These results compare with those obtained at emission</p><p>from a coal fired power station sample: 2 and 0.6 ~tg gq, respectively (Bresson et</p><p>al., 1984).</p><p>In spite of the variability of the ratios, there is still a correlation between particu-</p><p>late PAHs and carbon concentrations. This correlation is better underlined when</p><p>examining the covariation of the PAHp/SC and TC/SC or PAHp/TC and SC/TC</p><p>(Figure 5). The regression coefficients found for the correlations show that the</p><p>PAHp/TC correlation is more satisfactory, which would indicate that the origin of</p><p>PAHs</p><p>and particulate organic carbon is the sanle. However, the relative good corre-</p><p>lation between PAH and soot carbon could also indicate that PAHs are intermedi-</p><p>ate species between organic matter and soot carbon as expected previously. This</p><p>52 PIERRE MASCLET ET AL.</p><p>Table III. Particulate PAH/soot carbon ratio and particulate</p><p>PAH/total carbon ratio for fires and atmospheric background</p><p>(ratios expressed in ~g/g-1)</p><p>Sample 1 3 7 8 9 10</p><p>SC (~.g m 3) 1.56 97.3 3.13 60.2 1.9 15.7</p><p>TC (~tg m -3) 7.28 884 13.4 461 9.00 133</p><p>PAH 2.40 649 2.42 137 2.20 78</p><p>PAH/SC 1.54 6.67 0,77 2.28 1.16 4.97</p><p>PAH/TC 0.33 0.74 0.18 0,30 0.24 0.59</p><p>2-</p><p>* background</p><p>• fire</p><p>f</p><p>/</p><p>I</p><p>f</p><p>I</p><p>0-</p><p>f</p><p>f</p><p>J</p><p>.,..,Iv"</p><p>J</p><p>f</p><p>f</p><p>%</p><p>0</p><p>~ , 0,4"</p><p>* background</p><p>• fire</p><p>• t</p><p>,,%,</p><p>0~0</p><p>8 I0 0,10 0.15 0.20 0.25</p><p>TC I SC SC I TC</p><p>Fig. 5. Variation of the (a) PAtg/soot carbon and (b) PAH/total carbon ratios vs. soot carbon/total</p><p>carbon ratio (carbon data from Cachier et aL (1993, this issue),</p><p>result is not yet accurate enough but could confirm the hypothesis that PAHs are</p><p>systematically associated with soot or fly ash during combustion processes (Niko-</p><p>laou et al., 1984).</p><p>The correlation between PAHp to total carbon appears in the Figure 5a:</p><p>PAHp/SC = 1.004 TC/SC - 3.61 (r 2 = 0.83) (I)</p><p>The correlation between PAHp to soot carbon appears in the Figure 5b:</p><p>PAHp/TC = -3.225 SC/TC + 0.944 (r 2 = 0.67) (II)</p><p>Emissions factors may be estimated through the correlation existing between par-</p><p>ticulate carbon and particulate PAH concentrations, and recalling the emission fac-</p><p>tor found for carbonaceous aerosols emitted during the same bums. Equation (I)</p><p>may be transformed into:</p><p>PAHp/TC = 1.004 - 3.61 SC/TC; the mean SC/TC value for these six</p><p>samples is 16.97%</p><p>PAHp/TC = 0.392" -6 -1 10 g(PAH) g(c~rbon)</p><p>EMISSIONS OF" PAH BY SAVANNA HRES</p><p>The particulate carbon emission factor is</p><p>EF carbon = 17.2 _+ -I 7.9g(carbon)kg(dry plant)</p><p>The PAH emissions factors are then</p><p>for the particulate PAHs:</p><p>EFpAHp 6.7 + 3.1 x 10-6g(PAHp) • kg(dryplant)</p><p>and for the gaseous PAHs:</p><p>EFpAHg = 2 .4 + 1.1 X 10-4g(PAHg) • kg(dry plant)</p><p>as PAHg/PAHp = 35.</p><p>53</p><p>4. Conclusion</p><p>This study provides the first set of data on biomass burning in a tropical medium, a</p><p>region which is virtually free of anthropogenic emissions. It shows that PAHs are</p><p>produced abundantly, mainly in the gaseous form. The concentrations in ambient</p><p>air are almost the same as found in European countries and in the USA. This result</p><p>must be taken into account in the global PAH estimation since these social prac-</p><p>tices are frequent in nonindustrial countries. The PAH emissions, by biomass</p><p>burning, constitute therefore a hazard to human health.</p><p>This study has also pointed out that some PAHs are excellent tracers for this</p><p>type of combustion, especially fluorene and pyrene for gaseous emissions; the first</p><p>could be characteristic of smoldering episodes and the second of flaming ones. For</p><p>particulate emissions pyrene is also the best tracer although chrysene and coronene</p><p>emissions are significant.</p><p>The evaluation of the emissions factors of gaseous and particulate PAH allows</p><p>to estimate the total flux of PAH emitted in tropical Africa during the biomass</p><p>burning season. According to Delmas et aL (1991) the total biomass burnt in the</p><p>various savanna ecosystems of the African continent could be of the order of</p><p>2.52 x 1015g yr -1. Thus, the annual fluxes of PAHs due to the African savanna fires</p><p>only, are estimated to be 17 + 8 tons yr -1 for the particulate PAHs and 605 + 275</p><p>tons yr -1 for the gaseous PAHs.</p><p>Although the savanna fires occur during few months in the year, for tropical</p><p>Africa only the contribution of this source to the global budget of atmospheric</p><p>PAH is significant compared to the anthropogenic inputs from all the industrial</p><p>combustions.</p><p>References</p><p>Arey, J., Zielinska, B., Atkinson, R., Winer, A. M., Ramdahl, T., and Pitts, J. N. Jr., 1986, The forma-</p><p>tion of nitro PAH from the gas phase reactions of fluoranthene and pyrene with the OH radical in</p><p>the presence of NOx, Atrnos. Environ. 20, 2339.</p><p>54 PIERRE MASCLET ET AL.</p><p>Biermarm, H. W., McLeod, H., Alkinson, R., Winer, A. M., and Pitts, J. N. Jr., 1985, Kinetics of the gas</p><p>phase reaction of OH radical with naphthalene, phenanthrene and anthracene, Environ. Sci. Tech-</p><p>nol. 19, 244.</p><p>Bjorseth, A. and Ohifsen, B. S., 1983, Long-range transfer of polycyclic aromatic hydrocarbons, in A.</p><p>Bjorseth (ed.), Handbook ofPAH, Dekker, New York, p. 517.</p><p>Bresson, M. A., Beyne, S., Masclet, R, and Mouvier, G., 1984, Optimisation des m6thodes de pr61~ve-</p><p>ment et d'analyse des HAP et de leur d6rivgs azotds; d&ermination de leur stabilit6 dans t'atmos-</p><p>ph~re, in B. Versino and G. Angeletti (eds.), Physico-chemical Behaviour of Atmospheric Pollutants,</p><p>D. Reidel, Dordrecht, p. 125.</p><p>Cachier, H., Liousse, C., Buat Menard, R, and Gaudichet, A., 1995, Particulate content of savanna fire</p><p>emissions, J. Atmos. Cherm 22, 123-148 (this issue).</p><p>Crutzen, R J. and Andreae, M.O., 1990, Biomass burning in the tropics: Impact on atmospheric</p><p>chemistry and biogeochemical cycles, Science 250, 1669.</p><p>Delmas, R. A., Loudjani, R, Podaire, A., and Menaut, J.C., 1991, Biomass burning in Africa: an</p><p>assessment of annually burned biomass in J. S. Levine (ed.), Global Biomass Burning, MIT Press,</p><p>Cambridge, Mass. pp. 126-132.</p><p>Duvat, M. M. and Friedlander, S. K., 1981, Source resolution of PAH in the Los Angeles atmosphere.</p><p>Application of a chemical species balance method with first order chemical deca~; Environ. Sci.</p><p>Research Lab., Research Triangle Park, NC 27711, EPA 600/2-81; 161.</p><p>Finlayson-Pitts, B. J. and Pitts, J. N. Jr., 1985, Chemistry and mutagenic activity of airborne polycyclic</p><p>aromatic hydrocarbons and their derivatives, in Atmospheric Chemistry, Fundamentals and Experi-</p><p>mental Techniques, Wiley, New York, p. 870.</p><p>Freeman, D. J. and Cattell, F. C. R., 1990, Woodburning as a source of atmospheric polycyclic aro-</p><p>matic hydrocarbons, Environ. Sci. Technol. 24, 1581.</p><p>Jaffrezo, J. L., Masclet, R, Wortham, H., and Mouvier, G., 1992, Polycyclic aromatic hydrocarbons in</p><p>arctic regions (Greenland), in Proceedings of the 16th Symposium on PoIynuclear Aromatic Hydro-</p><p>carbons, Bordeaux Spet. 1991, in press.</p><p>Kadowaki, S., 1990, Characterization of carbonaceous aerosols in the Nagoya urban area 1: elemental</p><p>and organic carbon concentrations and the origin of organic aerosols, Environ. Sci. Technol. 24 (5),</p><p>741.</p><p>Masclet, R, Mouvier, G., and Nikolaou, K., 1986, Relative decay index and sources of polycyclic aro-</p><p>matic hydrocarbons, Atmos. Environ. 20, 439.</p><p>Masclet, R, Bresson, M. A., and Mouvier, G., 1987, Polycyclic aromatic hydrocarbons emitted by</p><p>power stations and influence of conditions parameters, Fuel 66, 556.</p><p>Masclet, R, Pistikopoulos, R, Beyne, S., and Mouvier, G., 1988, Long range transport and gas/particle</p><p>distribution of polycyclic aromatic hydrocarbons at a remote site in the Mediterranean sea, Atmos.</p><p>Environ. 22, 639.</p><p>Nikolaou, K., Masclet, R, and Mouvier, G., 1984, Sources and Chemical reactivity of PAH in the</p><p>atmosphere - a critical review, Sci. Tot. Environ. 32, 103.</p><p>Pistikopoulos, E, Masclet, R, and Mouvier, G., 1990a, A receptor model adapted to reactive species:</p><p>polycyclic aromatic hydrocarbons; evaluation of source contributions in an open urban site, Atmos.</p><p>Environ. 24A (5), 1189.</p><p>Pistikopoulos, P., Wortham, H., Gomes, L., Maselet, S., Masclet, E, and Mouvier, G., 1990b, Mecha-</p><p>nisms of formation of particulate polyeyclic aromatic hydrocarbons in relation to the particle size</p><p>distribution: effects on meso scale transport, Atmos. Environ. 24A (10), 2573.</p><p>Ramdaht, T., 1983, Retene - a molecular marker of wood combustion in ambient air, Nature 306, 580.</p>