Transverse approach between tunnel environment and corrosion: Particulate matter in the Grand Mare tunnel

Abstract

A tunnel-type semi-enclosed atmosphere is characterized by a higher particulate pollution than urban zones and highlights the particulate species having an impact on material degradation. Therefore, a transverse approach between air composition and its consequences upon longevity of materials is necessary, requiring a better knowledge of tunnel atmosphere and a better understanding of material degradation inside a tunnel for operating administration. The characterization of particulate matter collected inside a road tunnel in Rouen (France) allows us to set up the features of the particle characteristics of the real conditions of field exposure. Two sampling campaigns include analyses of organic and water-soluble ionic fractions. The current work shows that organic species, grouped into two sets derived primarily from engine exhaust and debris with wear particles resuspended by the traffic, are divided into two groups: a majority comprising n-alkanes, alkanoic acids, phthalates, ketones, and benzothiazole and a minority one composed of BTEX (benzene, toluene, ethylbenzene, and xylenes), polycyclic aromatic hydrocarbons (PAHs), fatty acid methyl esters (FAMEs), furans, phenols, and alkenes. As regards the water-soluble ionic fraction, the ionic species such as Cl⁻, SO₄ ²⁻, CH₃COO⁻, HCOO⁻, NO₃ ⁻, NH₄ ⁺, and Na⁺ are involved in the degradation process. The inorganic particles (insoluble and slightly soluble), debris and wear particles, organic acids, and relative humidity play a key role and are important factors to consider in the degradation process.

Implications:

Particles, together with relative humidity, can play an important role in the degradation of materials inside a road tunnel. Depth knowledge of the environment prevailing inside a tunnel improves the criteria for selection of materials of equipments contained therein. In the present study, it appears that the particulate mass can be divided into two groups with distinct effects: a water-soluble particle group is the source of ionic species involved in the degradation process, and a slightly water-soluble or insoluble particle group that acts as reservoir of condensed water when covering the surface of the material.

Introduction

Road traffic is one of the largest sources of air pollutants in the urban environment. It should be noted that vehicle exhausts are the main source of fine particles in the atmosphere. This particulate pollution can cause cardiopulmonary troubles and this particulate matter also contains toxic compounds that affect human health. Intense research is consequently focused on particle emission. There are two methods to determine vehicular emissions: the first consists of carrying out dynamometric tests on individual engines and the second consists of outdoor studies inside road tunnel or near heavy road traffic.

The particulate matter from vehicle exhausts collected inside tunnel has been studied, including emission factors (Weingartner et al., 1997; Kristensson et al., 2004; Cheng et al., 2006; Handler et al., 2008), and the size distribution being determined in laboratory (Kittelson et al., 2006) or in tunnel (Gouriou et al., 2004; Kristensson et al., 2004; Abu-Allaban et al., 2004). The chemical characterization of particular emissions has prompted many laboratory studies to find suitable compound markers characteristic of exhaust emissions in the atmosphere or to study the chemical profiles of these particles emitted by vehicle engines as function of technological evolution of engine or engine type (Rogge et al., 1993a; Cadle et al., 1999; Liu et al., 2010; Chiang et al., 2012). The PM₁₀ (particulate matter with an aerodynamic diameter <10 μm) emissions tend to increase with vehicle age due to both changes in vehicle technology and the increased likelihood of poor maintenance. Emissions are also tending to be higher in the winter (Caddle et al., 1999a). From model year 2004 and 2007 engine studies, evolution such as combustion strategies, aftertreatment components of heavy-duty diesel engines, and fuel composition contribute to reduce the portion of elemental carbon and organic matter, resulting in larger concentration from other sources such as inorganics (Liu et al., 2010). Rogge et al. (1993b) conclude that the particulate species in the atmosphere are composed not only of combustion products from fuel and oil but also of wear products from tires, break linings, and road construction and resuspension of road and soil dust. However, these tests, carried out in laboratory conditions, may not provide a good representation of the real-world traffic because of a large variation of engine types. In addition, these tests do not account for the nonexhaust emissions from wear products and the resuspension of particulate matter. The road tunnel studies are used to define an average vehicle exhaust profile representative of overall flow in circulation, with little interference with other sources such as urban heating or industrial activities. The chemical composition of primary particles from tailpipe emissions (PM_2.5 and PM₁₀) has prompted many studies that focused on different topics such as inorganic and (or) trace metallic speciation (Kristensson et al., 2004; Grieshop et al., 2006; Landis et al., 2007; Handler et al., 2008), water-soluble species (Laschober et al., 2004; Chiang and Yao-Sheng, 2009), metallic emissions (Sternbeck et al., 2002; Chellam et al., 2005), and organic speciation (Fraser et al., 1999; Song et al., 2005; Chellam et al., 2005; He et al., 2006). Very few studies concerning the chemical characterization of particles or aerosols under tunnel (Haddad et al., 2009) or in urban area (Roth et al., 2008) were conducted in France.

Nevertheless, unlike the studies of air particle effect on human health and the environment, no study has shown the relationship between particulate pollution and the degradation of metallic materials exposed to stuffy environments. The studies concerning the corrosion of materials such as copper, zinc, or nickel and achieved during outdoor exposure campaigns are especially focused on the effects of gaseous pollutants (Graedel et al., 1987; Graedel, 1987; Jouen, 2000; Lefez et al., 2001; Krätschmer et al., 2002; Jouen et al., 2004). Particle impact has never really been considered in outdoor field exposition campaigns. The atmospheric particles contain a large number of different species, including both organics and inorganic species, and can complicate the corrosion process. The inorganic compounds of particular mass have been given special attention through laboratory studies. The main source of chlorides in the atmosphere comes from marine aerosol that has been transported inland by coastal wind in the form of solid particle or liquid droplet. Calcium or sodium chloride is also used to treat roads in winter. Now, it is known that NaCl plays an important role in the atmospheric corrosion of materials and a substantial amount of work has been done in this field. In a humid pure air environment, the corrosion rate of zinc is proportional to the amount of NaCl added (Svensson and Johansson, 1993; Lindström et al., 2000). The degradation of copper or zinc substrates in the presence of NaCl accelerates according to the relative humidity (RH) and becomes very important at 95% RH. A single addition of NaCl (70 μg/cm²) before exposure resulted in higher corrosion rates, compared with the cases when the same total amount of NaCl was deposited in five successive additions (14 μg/cm²) before and during exposure (Svensson and Johansson, 1993). Furthermore, the NaCl particles induce a degradation of copper or zinc coupons at RH below the deliquescence relative humidity, 75.3% RH at 25 °C (Tang and Munkelwitz, 1993) (Svensson and Johansson, 1993; Chen, 2005). The combined effects of gaseous pollutants and NaCl particles have also been studied in laboratory tests. The atmospheric corrosion of zinc in the presence of NaCl and CO₂ is approximatively independent of temperature. The presence of CO₂ slows down NaCl-induced corrosion at 22 and 38 °C (Lindström et al., 2000). In the presence of NaCl particles, similar corrosion rate was found after exposure to pure humid air or humid air containing gaseous pollutants SO₂, NO₂, O₃, and NO₂+O₃ (Chen, 2005). The corrosion rate was not dependant on the concentration of pollutant except for the combination SO₂+O₃, which leads to a significantly higher corrosion rate at higher concentrations (Chen, 2005). About zinc, Svensson and Johansson (1993) point out, for small amount of NaCl added, that a synergetic effect was observed for SO₂+NaCl and NO₂+NaCl combination at 75% RH. In the presence of NaCl, SO₂+NO₂ environment did not result in any synergetic effect on the corrosion rate of zinc at 70% RH (Svensson and Johansson, 1993). The degradation caused by (NH₄)₂SO₄ particles present in airborne has been reported in the literature. Copper corrosion in air induced by submicrometric (NH₄)₂SO₄ particles (0.05–1.5 μm) depends most sensitively on the relative humidity; it is only possible if RH ≥ deliquescence relative humidity of (NH₄)₂SO₄ (75% RH at 373 K; 81% RH at 300 K) (Lobnig et al., 1993, 1994). According to the authors, particle size, in the size range studied, plays a minor role in the corrosion of copper. Nevertheless, even at 70% RH, corrosion attacks were observed on (NH₄)₂SO₄ deposited copper surface. The size of particles affects the corrosion rate; small particles (<10 μm) result in higher corrosion rate than large particles (<100 μm) at equal amount of deposited particles (Chen, 2005). The copper mass loss increases both with exposure time and with the amount of (NH₄)₂SO₄ particles deposited on the surface (Chen, 2005). On the other hand, zinc reacts with (NH₄)₂SO₄ particles as soon as RH ≥65%, value below the deliquescence relative humidity (DRH) (Lobnig et al., 1996). The influence of chloride or sulfate salt particles on zinc substrate has been mentioned. The chlorides NaCl and NH₄Cl accelerate degradation of zinc substrate; the results obtained underline the effect in order NaCl > NaCl + NH₄Cl > NH₄Cl (Qu et al., 2005). Similarly, Na₂SO₄ and (NH₄)₂SO₄ affect zinc; the mass gain decreases in the order as Na₂SO₄ > Na₂SO₄ + (NH₄)₂SO₄ > (NH₄)₂SO₄ (Qu et al., 2006). Lindström et al. (2002) studied the effect of four chlorides and sulfates on zinc substrate exposed during 4 weeks at 95% RH and 22 °C. NaCl and Na₂SO₄ are respectively the most corrosive among the chloride and sulfate investigated. The corrosion of zinc is directly correlated with the amount of sodium ion (NaCl or Na₂SO₄). According to the authors, there is a clear influence of the cation nature on the solubility of the salt and on its ability to attract water vapor to form an aqueous layer on the substrate surface. According to Askey et al. (1993), particulate species in atmosphere can accelerate the corrosion of metal in one or more of three ways: (1) differential aeration effects that occurs with chemically inert particles; (2) increasing the conductivity of the surface moisture layer after dissolution of soluble ionic species contained within the particles; and (3) promoting the oxidation of SO₂ to sulfuric acid by the catalytic activity of transition metal oxides associated with particulate matter. The aerosols play a critical role in corrosion (Lau et al., 2008), with size and composition effects.

This work is a part of a project dealing with the degradation of materials in a semiconfined tunnel-type atmosphere and with the equipment reliability and lifetime. The understanding of corrosion process involves knowledge of the real conditions of material exposure. In a previous study, it has been shown that among the gaseous pollutants identified, the SO₂-NO₂ combination can be considered as an important factor during the corrosion process (Ameur-Boudabbous et al., 2012). The five dominant aldehydes emitted by automotive traffic—formaldehyde, acetaldehyde, butanal, propanal, and acrolein—representing 90–95% of total aldehyde emissions, will also have an accelerating effect on the corrosion process. This work completes the characterization of exposure atmosphere composition with a special focus on particulate matter. Detailed knowledge on the composition of PM is required for each exposure location inside the tunnel, before any attempt of correlation with the observed material degradation. Characterization of the particles has been conducted over a long and adequate period corresponding to the different exposure times of materials in the tunnel. The composition of the atmosphere inside the Grand Mare tunnel (France), namely, organ specificity and water-soluble ionic species, has been followed, taking into account possible differences between the two tubes. No distinction of vehicle type or engine has been made owing the fact that counting during a long period is more tedious.

Methodology

The Grand Mare tunnel

The Grand Mare tunnel is an urban tunnel parted in two separate tubes (north bore and south bore) with two traffic lanes per bore and currently used by around 41,500 vehicles per day (2008 traffic count). Vehicles are only distinguished by their length: 12% of heavy-duty vehicle (>5 m) and 88% of low-duty vehicles. The geometrical characteristics of the urban tunnel have been indicated in a previous study (Ameur-Boudabbous et al., 2012). The tunnel length is around 1532 m with a slope of ˜3%. According to France Road Union Association, the proportion of passenger cars equipped with diesel engine is increasing continuously over the years (2007–2010: 52–58%), even for light commercial vehicles (2007–2010: 85–90%).

Sampling

A sampling campaign of metallic substrates has been conducted over a period of 5 yr. Only the results concerning the analysis of particulate matter are presented in this paper. Three locations were chosen for both campaigns, namely, UA, UC, and DA, as shown in Figure 1. Thermohygrometric conditions were simultaneously registered using a Vaisala transductor HUMICAP HMT 330 (Vaisala Corp., France). From a large number of data on the relative humidity, a ratio between the exposure time to a minimum relative humidity of 70% during the month and the monthly exposure time has been calculated for each exposure month. Figures 2 and 3 show the evolution of this ratio and of monthly average temperature over time.

Figure 1. Geometrical characteristics of Grand Mare urban tunnel and sampling locations in the upward and downward tubes.

Figure 2. Ratio evolution N _RH>70%/N_T as function of time. N _RH>70% = hour number per month at relative humidity ≥70%; N_T = total hour number of exposure per month.

Figure 3. Average temperature evolution inside the tunnel as function of exposure time. Error bars indicate ±standard deviation.

Particles dispersed in the atmosphere within the tunnel are depositing on the sample holder structure made of poly(methylmethacrylate) (PMMA), chemically inert. Figure 4 illustrates the specific exposition stand that was adopted for this study of materials aging and the different sample holder. The exposition stands of materials are fixed on the tunnel wall at a height of 1.5 m and at a length of 0.5 m from the traffic left line. The mass of the particulate matter collected on sample holder is a function of the tube and time, namely, 0.5 g (3 months) to 9 g (24 months) in the upward tube and 0.3 g (3 months) to 4 g (24 months) in the downward tube. These particles are representative of the one deposited continuously on materials exposed up to form a full coverage. The collect procedure adopted in this study does not permit collecting ultrafine particles from the sample holder. Two sampling campaigns were conducted: the first took place in summer from July 2007 to July 2009 and the second took place in winter from January 2009 to January 2010.

Figure 4. Photographs of front and side views of the exposition stand installed in the downward tube.

Chemical analyses

The chemicals used were of analytical reagent grade. All standard solutions were prepared using deionized Milli-Q water (18 MΩ). All the salts were in the form of sodium except acetate that was in ammonium form. Standard stock solutions containing anions such as CH₃COO⁻, HCOO⁻, Cl⁻, NO₂ ⁻, NO₃ ⁻, CO₃ ²⁻, and SO₄ ²⁻ with anion concentration of 1000 mg/L were prepared and stored at 4 °C. For the cations, all the salts were in the form of chloride with the exception of ammonium (nitrate form) and potassium (iodide form). A second set of standard stock solutions containing Na⁺, NH₄ ⁺, K⁺, Mg²⁺, and Ca²⁺ (cation concentration of 1000 mg/L) was prepared and stored at 4 °C.

Instrumentation

Ion chromatography was carried out with a Dionex model ICS-900 (Dionex Corp., France) ion chromatograph, 10 μL loop, Dionex DS5 conductivity detector, and a mobile-phase flow rate 1 mL/min.

For anions separation and identification, a Dionex Ion Pac AS-11HC analytical column and Dionex Ion Pac AG-11HC and Dionex Ion Pac CG3 guard columns were used (Dionex Corp., France). The eluent conductivity was chemically suppressed by an ASRS-300 4-mm self-regenerating suppressor (Dionex Corp., France). Isocratic analysis required 18 mM KOH as mobile phase. Conditions used for the gradient analysis are an initial concentration of 1 mM KOH, a gradient time of 25 min, a final concentration of 25 mM, and a final hold time of 5 min. The eluent KOH solution was prepared with the Dionex Reagent Free Controller RFC30 and a Dionex eluent generator cartridge EGC II KOH.

Cations were analyzed using a Dionex Ion Pac SCS1 analytical column and a Dionex Ion Pac SCG1 guard column. The eluent used—3 mM methanesulfonic acid (isocratic condition)—was prepared by diluting concentrated acid (99.5%).

Calibration was obtained from freshly prepared standard solutions in the range from 1.0 to 10.0 mg/L. The ions are identified from their elution/retention times and are quantified from the conductivity peak area.

GC-MS was performed on an Agilent 6890N GC system equipped with an Agilent 5973N mass-selective detector (Agilent Technologies, Agilent Corp., France). A cooled injection system (Gertsel CIS4; Gerstel GmbH, R.I.C. SAS, France) was coupled with an external thermal desorption system (Gerstel TDS2; Gerstel GmbH). Sample (≈20 mg) were directly introduced into extraction tube plugged with quartz wool then loaded in the thermal desorption unit. The thermal desorption is carried at 280 °C for 5 min. The column flow was 1.0 mL/min and the split vent flow was 20 mL/min. The GC column was 30 mm length × 0.25 mm i.d. × 0.25 μm film thickness optima 5 Accent (Machery-Nagel EURL, France). The oven temperature was kept at 50 °C for 10 min and then it was raised to 300 °C at 10 °C/min. Final temperature was kept for 40 min. The mass scan range was 36–500 m/z amu. The identification of compounds was obtained by comparison with the National Institute of Standards and Technology (NIST) spectral reference library (NIST98).

Identification of crystalline products was performed by X-ray powder diffraction (Cu K_α radiation) operating at a voltage of 35 kV, a current of 40 mA, and a scanning rate of 5 sec from 2θ = 5° to 70° with angular pitch of 0.1°. Morphological and elemental studies were performed with scanning electron microscopy equipped with energy dispersive spectroscopy (SEM-EDS).

Treatment

The major peaks observed in the thermodesorption total ion current chromatogram (TIC) were quantified using the peak area and a model for the prediction of response factors. The model according to Fitch and Sauter (1983) presents a scheme for the calculation of electron impact total ionisation ratio for organics molecules (for electron impact measured at 70 eV). This scheme could be used in the prediction of GC-MS response factor in the case of electron impact ionization and was used by Kaal et al. (2007) to calculate the relative ionization cross-section of compound i, the contribution of compound i, and the mass percentage of compound i on a TIC GC-MS chromatogram.

The relative ionization cross-section of compound i, Q_i , is calculated according to Fitch and Sauter (1983):

(1)

In this equation, nX _i is the number of X atoms in compound i (X = C, H, O, N, S, Cl).

The contribution of compound i was calculated according to Kaal et al. (2007):

(2)

in which A_i is the peak area of compound i, MW_i the molecular weight of compound i, and Q_i is the calculated relative ionization cross-section of compound i.

The thermodesorption products were then grouped into the following classes: alkanes, alkenes and alkynes, n-alkanoid acids, aromatic acids, phthalates, ketones and aldehydes, alcohols and phenols, esters, polycyclic aromatic hydrocarbons (PAHs), hopanes, furans, BTEX (benzene, toluene, ethylbenzene, and xylenes), and N- and/or S-containing compounds.

Results and Discussion

Physical characteristics of particulate matter

The exposition site used in this study was previously selected for a study focused on particulate concentration and size distribution in a size range from 30 nm to 10 μm (Gouriou et al., 2004). The mean profile of particle concentration as function of distance from the entrance (upward or downward tube), extracted from the study achieved by Gouriou et al. (2004) is shown in Figure 5. The particulate concentration represents the number of particles per volume unit expressed as number of particles per cm³, all the particles being collected independently of their size. The number of particles per cm³ in tunnel atmosphere extracted from this graph is respectively 2.4 × 10⁵, 4.8 × 10⁵, and 1.4 × 10⁵ at UA, UC, and DA locations selected in this study. Beyond 400 m from the entry, the average level is increasing as a function of distance in both tubes, with average level at the exit of the tunnel 3 times higher in the upward tube than in the downward tube. Gouriou et al. (2004) have shown from board measurements that the particle size does not change between entry and exit of the tunnel. Only the total particulate concentration evolves. The Grand Mare tunnel presents an average particle concentration 3 times higher in the upward tube compared with the downward tube (Figure 5; Gouriou et al., 2004). Figure 6 shows SEM micrograph with accumulation of particles on PMMA substrate after a 3-month exposure (Figure 6a) and micrograph of particles collected from sample holder and then deposited on polished copper substrate (Figure 6b). Surfaces are covered with porous agglomerates from the accumulation over time of resuspension of road dust caused by the movements of vehicle and the particulate emissions from road traffic itself. The X-ray diffraction analysis of dust collected on the sample holder structure after 3 and 6 months of exposure reveals the presence of chemically inert crystalline products: Na₂Si₂O₅·5H₂O, NaAlSi₃O₈ or/and CaAlSi₃O₈, SiO₂, slightly soluble CaCO₃ and CaSO₄·2H₂O, and NaCl. These products are the same in both tubes. Notice that NaCl has been identified after 3 months of exposure (July–October) inside the tunnel. During this period, no preventive treatment against ice has been made on roads. The tunnel is never treated; only exterior pavements are treated in winter. Road traffic carries salt particles inside the tunnel; they will remain there for a more or less period. All the particles analyzed by X-ray diffraction and present on the sample holder as well as the exposed substrates come from a resuspension of particles in the atmosphere. The elemental analysis by means of SEM-EDS reveals the predominant presence of elements such as O, Si, Ca, Cl, S, Na, and Al corroborating the results obtained by X-ray diffraction. Fe, Mo, K, Ti, Mg, Ba, Rb, and Br are also detected. Elements such as Si, Ca, Al, Fe, and Mg have been identified as well by Kristensson et al. (2004), Grieshop et al. (2006), and Handler et al. (2008). Kristensson et al. (2004) have also identified the presence of Cl, S, and K in particulate matter, but no information about Na was mentioned by these authors, whereas the PM_2.5 particulate sampling was conducted in winter. Bromine has been detected in PM_2.5 particulate matter by Kristensson et al. (2004) and Lough et al. (2005). From a literature survey, it is possible to summarize the different sources of elements detected in this study. Ca and Mg identified in coarse particle fraction point out the resuspension of road dust and soil as the main source (Handler et al., 2008), whereas the same elements, Ca and Mg, in fine particles are also present in lubricating oils as detergent additives (Cadle et al., 1999). The source of Ba is linked to brake wear (Sternbeck et al., 2002; Garg et al., 2000). Ba in fine particulate particles can be linked quantitatively to diesel engine emission and considered as an elemental tracer for heavy-duty trucks (Chellam et al., 2005). Ti is used in the paint for road markings (Handler et al., 2008) or connected to the brake wear (Garg et al., 2000). Mo is used as a component of automotive catalyst and as an antifriction additive in lubricating oil (Grieshop et al., 2006). Rb source can be associated with the degradation of catalytic converters (Lough et al., 2005).

Figure 5. Particle concentration evolution along the upward and downward tubes at the three sampling locations UA, UC, and DA, respectively middle position, exit position in the upward tube, and middle position in the downward tube. The profile of particle concentration as function of distance from the entrance is extracted from Gouriou et al. (2004). © Elsevier. Reproduced by permission of Elsevier. Permission to reuse must be obtained from the rightsholder.

Figure 6. SEM micrographs. (a) Exposure of PMMA sample during 3 months at UA location. (b) Particle collected on sample holder and deposited on copper substrate.

Chemical characteristics of identified organic compounds

Regarding the organic compounds, only the fraction of identified components can provides useful information with respect to the degradation of materials. Indeed, according to literature data, this fraction represents about 4% of particulate organic matter (Fraser et al., 1999; Haddad et al., 2009) and the results discussed below will focus on this fraction. Only data from the first campaign will be discussed here. The pie chart (Figure 7) shows the different species present in the Grand Mare tunnel. Values mentioned in the figure represent the average values calculated from the results collected after 6, 9, 12, 18, and 24 months of exposure. Two groups of organic compounds have been identified: the first with contribution ≥6% and the second <6%.

Figure 7. Pie chart of organic species detected in particulate matter during the first campaign at UA, UC, and DA. The three sampling locations are respectively middle position, exit position in the upward tube, and middle position in the downward tube. The average values are calculated from the results collected after 6–24 months of exposure.

Major group of organic compounds (≥6%)

The n-alkanes belong to an important class of organic components in atmospheric aerosols inside the tunnel (Figure 7). The distribution of n-alkanes (C₁₂–C₃₂) in the three targeted zones UA, UC (upward tube), and DA (downward tube) is illustrated by Figure 8a. These profiles are dominated by C₂₀–C₂₁ alkanes in the upward tube and C₂₁–C₂₂ in the downward tube, results similar to those obtained by Haddad et al. (2009) for PM_2.5 (aerodynamic diameter ≤2.5 μm) matter. The main difference lies in the more important relative proportion of C₂₇–C₃₀ alkanes in this study compared with Haddad et al.’s results; the sampling protocol (methodology and duration) is quite different. On the other hand, the results obtained in the Grand Mare tunnel are quite different from those obtained by Fraser et al. (1998) and He et al. (2006) with a C number maximum located at 24. The hydrocarbons of lower molecular weight (C₁₀–C₂₅) are most commonly derived from diesel fuel, whereas lubricating oils contain primarily high-molecular-weight hydrocarbons (Schauer et al., 1999; Voss et al., 1997), although significant overlap exists (Zielinska et al., 2004). During dynamic test carried out on individual engine, Rogge et al. (1993a) have quantified n-alkanes that range from C₁₉ to C₃₂ in fine particle vehicle exhausts. The profiles obtained during tests on heavy-duty diesel truck engines show a predominance of low-molecular-weight n-alkanes (C ≤ 22), with a maximum at C₂₀ (Figure 8b, inset). According to the authors, this result is linked to the relatively high lubricating oil consumption of diesel engines and results also from unburned diesel fuel itself. Although the conditions are different, the profiles obtained remain similar. The characterization of n-alkane emissions on gasoline engines equipped with catalytic converters reveals a lower emission rate (≤20 μg/km) with a C number maximum located at C₂₅ (emission rate: 18 μg/km), whereas the emission is about 13 μg/km for C₂₀ alkane (Rogge et al., 1993a). Considering the results obtained by the authors, the ratio calculated from the sum of n-alkanes respectively emitted by a diesel engine and a gasoline engine equipped with catalytic converters reaches a value close to 34 and shows that the contribution of n-alkanes emitted by gasoline engine equipped with converter is negligible. Knowing that the current fleet comprises approximately 60% of passenger vehicles equipped with a diesel engine and that the totality of high commercial vehicles and heavy trucks is diesel engine, we can reasonably assume that the distribution of lower-molecular-weight n-alkanes (C ≤ 25) determined in this study is associated with diesel exhaust whatever the observation site. For the C number >26, many sources can exist out of exhausts. Regarding the tire wear particles, the major fraction of elutable and identified matter is composed of n-alkanes, the C₃₀–C₄₁ range is almost all n-alkanes, with C number maximum located at 37 (Rogge et al., 1993b). From the collected data, it is not possible to demonstrate a source of tire wear particles from n-alkanes due to upper limit range of CG/MS setup. Rogge et al. (1993b) noted that small amounts of n-alkanes (range: C₁₉–C₃₆) were identified in the extractable fraction of brake wear particles. It appears rather as a minor source. Biogenic materials and garden soil organics also contribute to the street dust. The distribution of n-alkanes on these vegetative detritus shows a predominance C₂₉–C₃₁–C₃₃ carbon numbers (Rogge et al., 1993b). Concerning n-alkanes >26 of this study, the direct contributors may be residues of lubricating oil, street surface weathered particles, and to a lesser extend the brake wear particles constituting particulate matter resuspended by car movements.

Figure 8. Distribution of n-alkanes. (a) At different locations inside the Grand Mare tunnel (UA, UC, and DA represent respectively the three sampling locations: middle position, exit position in the upward tube, and middle position in the downward tube). (b) Laboratory tests on heavy-duty truck engine from Rogge et al. (1993a). © American Chemical Society. Reproduced by permission of the American Chemical Society. Permission to reuse must be obtained from the rightsholder.

About the second major groups, for n-alkanoic acids, a predominance of n-hexadecanoic and octadecanoic acids is observed, a result in agreement with those obtained by Fraser et al. (1998), He et al. (2006), and Haddad et al. (2009). Determination of n-alkanoic acids present in exhaust during dynamic test carried on heavy-duty truck engine or gasoline engine equipped with converter reveals the presence of hexadecanoic or octadecanoic acid; only hexadecanoic acid, identified also in this study, appears in significant amount during heavy-duty engine test (Rogge et al., 1993a). The wear particles from tires and brakes contain also n-alkanoic acids; the contribution of brake wear particles remains weak and hexadecanoic and octadecanoic acids are not dominant (Rogge et al., 1993b). However, the n-alkanoic acid such as octadecanoic, hexadecanoic, then tetradecanoic, ranked in decreasing order, are predominant in tire wear particles (Rogge et al., 1993b). We can consider that the n-alkanoic acids found in different samples come from resuspended particles consisting partly of tire wear particles. The n-alkanoic acids identified in particulate matter can also form during the combustion process. Yet, considering sampling conditions, the automobile exhaust from diesel engine is also a source for hexadecanoic and octadecanoic acids. This study highlighted the significant presence of acetic acid in the particulate matter collected on a sample holder structure. The automobile emission data reported by Kawamura et al. (1985) suggest that alkanoic acids are directly emitted by motor vehicles as incomplete products of fuel and that the emissions are enriched in acetic acid compared with formic acid. Talbot et al. (1988) measured formic and acetic acids in a tunnel near Hampton, Virginia, USA, and observed that more acetic acid than formic acid was emitted by mobile source. The principal loss process for gas-phase formic and acetic acid in the atmosphere appears to be dry and wet deposition (Hartmann et al., 1991). Khwaja (1995) indicates that acetic acid was the most abundant alkanoic acid, the concentration of which varies from 87 to 328 ng/m³ in particulate phase. The presence of acetic acid detected in this study may be explained by a dry or wet deposition process of both gas-phase and particulate-phase acetic acid. Khwaja (1995) indicates in his study possible sources of alkanoic acids, among which aqueous-phase reaction of aldehydes with OH radical. The aqueous oxidation of aldehydes into organic acids in the gas phase was suggested to be an important generation process for organic acids (Chameides and Davis, 1983). Pena et al. (2002) and Kawamura et al. (2000) have established the importance of the oxidation of aldehydes as a relevant source of organic acids. In a previous study inside the same tunnel, five dominant aldehydes emitted by automotive traffic have been identified (Ameur-Boudabbous et al., 2012). Taking into account the humidity inside the tunnel, an aqueous oxidation of aldehydes can also be considered and contributes to generate organic acids. The other important reactions—such as photochemical reaction process of organic precursors or gas-phase reaction of olefins (alkenes) and O₃ mentioned for the formation of alkanoic acids (Khwaja, 1995)—are not possible inside the tunnel due to the absence of solar radiation and the very low O₃ concentration.

The relative proportion of phthalates remains constant in the middle of the tunnel (upward and downward tubes) and is higher at the exit of the upward tube (UC). Among the phthalates detected in the Grand Mare tunnel, bis(2-ethylhexyl)phthalate (DEHP) is present, in agreement with Haddad et al. (2009); phthalic anhydride and dibutyl phthalate (DBP) are also detected in lower amounts. The presence of phthalates in the particulate matter is rather linked to other source than vehicle exhaust emission. Phthalate esters are widely used as plasticizers in polymeric materials, paint pigments, caulk, adhesives, and lubricants (Staples et al., 1997). Phthalate anhydride is used in the manufacture of polyester resins, resins in paints, and rubber scorch inhibitor. The presence of these phthalates in the collected particulate matter may originate from construction materials. Coating materials containing resins and polymers can appear as potential sources. The results concerning the proportion of phthalate detected in the Grand Mare tunnel are higher than those determined by Haddad et al. (2009) inside the Marseille tunnel; this difference can be explained from the sampling conditions, namely, an accumulation over several months (6–24 months).

From compounds containing both N and S, the amount of which is relatively constant regardless of the observation site; the results show that benzothiazole is predominant. Kim et al. (1990) showed that benzothiazole is a suitable tracer to determine the contribution of tire wear in ambient aerosols. Rogge et al. (1993b) have also detected the presence of benzothiaziole in the analysis of tire wear; it originates from the pendant group of vulcanization accelerators classified as thiazoles and sulfomanides. The identification of benzothiazole indicates the presence of tire wear on sample holder and exposed materials inside the tunnel, presence that cannot be detected with n-alkanes due to the limitation of equipment.

Finally, among relatively large amount of identified species (≈8%), this study found that the aldehydes are in minor amount (benzaldehyde and substituted benzaldehyde) and ketones in major proportion. The main identified ketones are ethyl methyl ketone (MEK), acetophenone, p-methyl acetophenone, and methyl isobutyl ketone (MIBK). They are used as solvents for paints, and are involved in the manufacture of lacquers, varnishes, glues, adhesives resins, or stabilizers of plastics against ultraviolet light. Zielinska et al. (1996) have detected the presence of MEK inside Fort McHenry Tunnel in the analysis of volatile organic compounds (VOCs) emitted by motor vehicles, principally heavy-duty vehicles. It seems difficult to identify a single source of ketone emission, but due to the significant use of ketones in building materials, the effect of particle redeposition would be preponderant. The benzaldehyde concentration inside the tunnel is lower, representing 0.4% of the total aldehyde emission (Ameur-Boudabbous et al., 2012). It is the same for the amount adsorbed on the particles.

Minor group of organic compounds (<6%)

In the second class of organic compounds, the contribution of BTEX remains constant, independently of the observation spot place. In a previous study, the results about total BTEX concentration showed a higher value at the exit position compared with middle position in the upward tube and a higher value in the upward tube with regard to the downward tube (Ameur-Boudabbous et al., 2012). No identical trend appears about the adsorption of BTEX on particles.

Focusing on the second group, the fatty acid methyl esters (FAMEs) have a contribution substantially constant—about 3–5%—independently of the observation site. Among the esters, the methyl ester of linoleic acid is preponderant. The presence of FAMEs is due to the incorporation of the 7% DIESTER^® (biodiesel) in the diesel fuel supplied in stations since 2008 and 30% DIESTER^® for buses and public and professional utility vehicles. These FAMEs arise directly from exhaust emission of diesel engine equipping the passenger cars and light commercial vehicles. Considering the actual diesel fuel, the FAMEs, especially the methyl ester of linoleic acid, can be appearing as suitable tracer, typical of diesel engine exhausts for the ambient aerosols. This result has never been mentioned in the literature hitherto.

The presence of PAHs has been detected and their relative contribution remains constant in the tunnel. These results are in agreement with those obtained by He et al. (2006) and Haddad et al. (2009). Many factors can influence PAH emission rate, including the fuel composition, PAH concentration of lubricating oil, and engine operating mode (load, cold start) (Cadle et al., 1999). The PAHs have been introduced into tire tread, either directly from organic additives or as contaminants in the carbon black addition (Rogge et al., 1993b). Among the analyzed species, this study reveals the presence of furan class, including a large number of compounds with a relative global contribution of 1–3% depending of exposure sites. Furfural, 2-methylfuran (alkyl furan derivates), 4,7-dimentylfuran, and 2,3-dihydrobenzofuran appear to be predominant. Furan and its derivates 2-methylfuran and 3-methylfuran are heterocyclic compounds dispersed into the atmosphere from the combustion of fossil fuels, waste biomass, and degradation products of biogenics (Graedel et al., 1986). The 2-methylfuran, 2-furan carboxaldehyde and 2-furan methanol, detected in particles collected inside the tunnel, are known as in pyrolysis products from fresh and composted waste and derived from polysaccharides (Dignac et al., 2005). Furfurals are a family of aromatic compounds produced from dehydration of sugar and are a major component of bio-oil (He et al., 2009; Elliott and Hart, 2009). Among switchgrass pyrolysis oil species cited in literature, the presence of furfural, 4,7-dimethylbenzofuran, and 3-metholbenzofuran was found (Imann and Capareda, 2010). The first two species are clearly identified in the particulate matter samples collected in the Grand Mare tunnel. Gandini and Belgacem (1997) published a review on macromolecular structures bearing furan ring or moieties arising from them. Furan derivatives were used as initiating, transfer, or terminating agent. Conventional polymers were modified by furanic reagents. This review shows the importance of exploiting the specificities of furan chemistry to elaborate original structures and thus obtain interesting properties. Furan resins are a likely alternative to phenolic resin because they can be obtained by replacing the toxic formaldehyde by furfural obtained from a wide range of agricultural residues containing pentoses (Rivero et al., 2011). According to Simoneit (2002), furan derivatives, described as components of smoke and attributed to cellulose combustion/decomposition products, are generally minor constituents of smoke from biomass burning. The waste incinerator located near the road tunnel is no longer in service for over 10 yr; therefore, the biomass burning source is unlikely. The increasing use of biowaste or solid waste compost in agriculture and private gardens is source of furans. Runoff of rainwater from the environment up tunnel and the wind transport of the particles can be suggested to explain the presence of furans inside the downward tube. It seems unlikely in the upward tube. A second explanation of their presence inside the tunnel is linked to compost conveying activities. Furans and its derivatives used in thermoplastic resins for paints and varnishes, polymer, or plastic container are also a potential source, which has been previously reported above. A single source of furans cannot be highlighted in this study. However, due to the extensive use of furans in polymer chemistry and presence in the atmosphere of components of polymers, resins, etc., through the detection of various organic species, this chemical source may be the main source of furans. Several reports have mentioned that vehicle exhausts are also a source of polychlorinated dibenzofurans (PCDFs), particularly in tunnel air (Miyabara et al., 1999; Chang et al., 2004; Deng et al., 2011); in our study, no feature indicates the presence of these species.

The analysis of particles after long-time exposition up to 24 months allows highlighting alcohols and phenols. The phenols are used in the manufacture of synthetic resin, coloring agents, lubricant, and solvent. It is also naturally occurring in some varieties of wood. The components such as phenol, phenol, 2-methyl and, phenol,4-methyl together with phenol dimethyl are the major organic components of lubricating bio-oil (Gupta and Demirbas, 2010). The alcohols are, in turn, present both in the resin and in plasticizers or additives in fuel.

Alkenes are also detected inside the Grand Mare tunnel, with a constant relative contribution. In this alkene group, the squalene but also 1-heptadecene, 1-hexacosene, 1-octadecene, 10-heneicosene, and 1-tridecene are major compounds. The alkenes analyzed in this study have a carbon number ranging from 13 to 30. Squalene, a biogenic hydrocarbon—which is an intermediate in the biosynthesis of steroids and triterpenoids—is detected within Los Angeles urban area during a severe photochemical smog episode (Fraser et al., 1997). During the period considered, no photochemical smog episodes have been mentioned. No information about the presence of squalene in the particulate matter taking from the tunnel atmosphere has been reported in the literature before this study. The combustion of biomass is an important primary source of soot and organic particulate matter in the atmosphere. Studies have reported the chemical composition of organic species in smoke particulate matter emitted by flaming and smoldering combustion of fuel from conifers (Oros and Simoneit, 2001a), deciduous trees (Oros and Simoneit, 2001b), and grasses (Oros et al., 2006). Among the organic compounds, the authors emphasize the presence of alkene groups, including heptadecene, octadecene, heneicosene, hexcosene, and tricosene, compounds also identified in the particulate matter collected in the tunnel. Nevertheless, the relative abundance of alkenes is respectively 3.1% and 4.6% in conifer and deciduous tree smokes (Oros and Simoneit, 2001a, 2001b). Simoneit (2002) points out that the major source of C₁₅–C₃₇ alkenes comes from coal and biomass during a combustion process. The n-alkenes are formed primarily by the thermal dehydration of n-alkanols (Mazurek and Simoneit, 1984) and to a minor degree from the n-alkanes by oxidation during incomplete combustion (Abas et al., 1995). The presence of significant amounts of alkenes (up to 20%) is reinforced by the fact that alkenes are formed during the early stages of reaction of alkanes (Vanhove, 2004). It appears that a single source of alkenes seems unlikely but the engine exhaust is still the main source.

About aromatic acids, the results (˜0.3%) are in agreement with those obtained by Haddad et al. (2009) inside the Marseille road tunnel. Only benzoic acid has been detected and is used as common retarder additive in the manufacture of tire (Rogge et al., 1993b) or present in vehicle exhaust (Kawamura et al., 1985). Fraser et al. (1998) have also pointed out the presence of this acid in the Van Nuys tunnel. Concerning the organic acids in motor exhausts collected under laboratory car tests, Kawamura et al. (2000) pointed out that benzoic acid is among the most abundant acids in gasoline exhaust but is also present in diesel exhaust with significant concentration. However, the engine tests were achieved with engine of 1970s–1980s.

Determination of water-soluble species

The average mass concentration of water-soluble ionic species and standard deviation are summarized in Table 1. It has been calculated for each species to execute t test with the standard confidence level of 95%. The mass concentration values cannot be compared with the results published in the literature that indicate the particle concentrations in atmosphere. In the present study, a passive accumulation over time was used, so the experimental conditions were totally different. Whatever exposure campaigns, the mass concentration, expressed in μg/mg, remains constant for almost all of chemical species analyzed (Figure 9). Two species groups appear: the majority one—Cl⁻, Na⁺, CO₃ ⁻, SO₄ ²⁻, Ca²⁺—and the minority one—NO₃ ⁻, NH₄ ⁺, K⁺, Mg²⁺, Zn²⁺, CH₃COO⁻, HCOO⁻, F⁻—independently of sampling period. As shown in Figure 9, the Na⁺ average concentrations are not significantly different regardless of the slope of the tunnel and sampling period (P = 94%). As regards chlorides, the average concentration is significantly higher than those obtained for Na⁺ for both two tubes (P _UA = 0.14%; P _DA = 0.57%). The profiles of mass concentration of chloride and sodium species evolve similarly over time independently of the tube slope and sampling period (Figure 10); the gap between Cl⁻ and Na⁺ concentrations remains constant over time. These results show that one or more sources of chlorides, apart from that attributed to road salts, exist. Various sources of chlorides have been reported in literature. First of all, the chlorides present in lubricating oils come mainly from dispersants, which are used to retain dirt in suspension and thereby protect the engine (Dyke et al., 2007). “Conventional dispersants” are manufactured using a process that results in a residual chloride concentration in the oil attached to long-chain organic molecules; typical levels are now in the range of 100–150 mg/km (few ppm) (Dyke et al., 2007). In addition, chlorine is not routinely specified in engine fuels. Dyke et al. (2007) have also analyzed the chloride content in both gasoline and diesel fuels sold by service stations. For nominally similar fuel sources from forecourts in Europe, there is a wide variation in measured chlorine level (0.1–6.9 mg/km; diesel average value: 1.23 mg/km and gasoline average value: 3.6 mg/km). When the automobile is run for a short period of time, the exhaust gas temperature does not rise so high, resulting in the formation of exhaust gas condensate in the low temperature of the muffler; therefore, the condensate water contains NH₄ ⁺, CO₃ ²⁻, SO₄ ²⁻, Cl⁻, and organic acids (Sato and Tanoue, 1995; Doche et al., 2006; Lee et al., 2006). Chlorides in condensates lead to assuming the presence of chlorine or hydrochloric acid in gaseous form from engine exhausts. Other sources of hydrochloric acid—potential source of chlorine—present in air atmosphere come from the coal or waste combustion (Lightowler and Cape, 1988). The waste incinerator, located near the road tunnel, is no longer in service for decade; only coal combustion may prove to be a source of hydrochloric acid. The excess of chlorine obtained in the soluble fraction can be derivatized as a soluble compound. Among the crystalline products identified in X-ray diffraction, only CaSO₄·2H₂O and/or CaCO₃ may react with hydrochloric acid to form CaCl₂, soluble in water. No specific features of CaCl₂ have been identified by X-ray diffraction. Laschober et al. (2004) and Chiang and Yao-Sheng (2009) have also detected the combined presence of Cl⁻ and Na⁺ species in particulate mass collected in tunnel. The results obtained at the center of Chung-Lio Tunnel indicate a Na⁺ concentration systematically higher than the chloride level for PM_2.5 and PM_2.5–10 particulate matter (Chiang and Yao-Sheng, 2009). However, the results obtained inside Kaisermühlen Tunnel show a chloride concentration higher than the one obtained for Na⁺ (Laschober et al., 2004), results similar to those obtained in the present study. Possible sources of chloride, with the exception of road salt, do not seem very well identified.

Table 1. Average mass concentrations of ionic species with standard deviations during two sampling periods at the center of the two tubes (unit: μg/mg)

	Upward Tube				Downward Tube
	Summer		Winter		Summer		Winter
Species	Average	Standard Deviation	Average	Standard Deviation	Average	Standard Deviation	Average	Standard Deviation
F^−	0.24	0.09	0.09	0.02	0.07	0.03	0.03	0.01
CH3COO^−	0.30	0.11	0.21	0.03	0.37	0.12	0.18	0.05
HCOO^−	0.20	0.09	0.11	0.03	0.20	0.12	0.12	0.02
Cl^−	25.85	4.68	29.83	9.37	28.27	8.50	31.63	13.36
CO3 ^2−	9.27	1.93	9.53	0.65	11.79	2.31	9.78	0.65
SO4 ^2−	22.11	2.33	11.60	2.38	11.27	2.55	6.39	2.44
NO3 ^−	2.57	1.83	2.15	1.39	1.73	1.05	1.78	0.99
Na^+	13.69	1.71	16.30	5.67	14.50	4.12	16.01	5.46
NH4 ^+	1.09	0.54	1.07	0.70	0.84	0.42	1.38	0.43
K^+	0.62	0.18	0.46	0.17	0.39	0.13	0.30	0.16
Mg^2+	0.34	0.12	0.24	0.06	0.18	0.10	0.12	0.04
Ca^2+	12.68	0.94	6.08	3.46	6.79	1.44	4.50	1.72
Zn^2+	0.77	0.08	0.69	0.18	0.62	0.23	0.54	0.20
HCOO^−/CH3COO^−	0.66		0.52		0.54		0.66

Figure 9. Average mass concentration (μg/mg) for each water-soluble specie detected in particulate matter. (a) Summer sampling period (July 2007–July 2009). (b) Winter sampling period (January 2009–January 2010). Error bars indicate ±standard deviation.

Figure 10. Average mass concentrations (μg/mg) of Cl⁻ and Na⁺ as function of time during the first campaign at center of tunnel. Error bars indicate ±standard deviation. Road salt: 50% Na⁺, 50% Cl⁻.

Analysis of results shows significant amount variations of SO₄ ²⁻ and Ca²⁺ species (Figure 9). The average mass concentration of SO₄ ²⁻ is significantly higher in the upward tube (UA) compared with the downward tube (DA), regardless of sampling period (P _summer = 0.055%; P _winter = 1.11%). In a previous study, the results have shown that the average concentration of SO₂ was significantly higher in the upward tube than that observed in the downward tube during two campaigns (May–July 2007 and January–April 2009) and correlated to the slope of the tunnel (Ameur-Boudabbous et al., 2012). We reach similar conclusions in terms of mass concentration of SO₄ ²⁻. The presence of sulfates can be explained by the adsorption of SO₂ or sulfuric acid vapor on particles collected on sample holders. About Ca²⁺, the average concentration is significantly higher in the upward tube compared with the value obtained in the downward tube (P = 0.003%) during the first campaign (summer). The results collected during the second campaign (winter) show no significant difference between the two tubes. Ca is present in the form of CaCO₃ and CaSO₄·2H₂O, as evidenced by X -ray diffraction analysis. The expected solubilities in pure water are respectively 0.015–0.020 g/L for CaCO₃ and 2 g/L for CaSO₄·2H₂O. These values can change depending on the composition of the aqueous layer. Calcite can be converted into CaNO₃ following reaction with gaseous nitric acid (Harrison and Kitto, 1990). The solubility of calcium sulfate in water has been extensively studied and it is known that the solubility increases with the temperature up to ˜50 °C. In water, gypsum (CaSO₄·2H₂O) is a saturating solid phase for temperatures at least as high as 100 °C. The addition of sulfuric acid to the system causes a substantial increase of the solubility of gypsum; this one increases with increasing acid concentration up to ˜10% wt H₂SO₄ (Dutrizac, 2002). The comparison of the mass concentration of Ca²⁺ and CO₃ ²⁻ showed no consistent trend between the two campaigns achieved inside the upward tube. Nevertheless, inside the downward tube, the CO₃ ²⁻ content is significantly higher than that observed for Ca²⁺, independently of the sampling period. On the other hand, the sulfate concentrations are significantly higher than those obtained for Ca²⁺, regardless of the slope of the tunnel and sampling period. Knowing that a proportion of sulfate comes from the SO₂, the combined presence of NaCl and/or sulfuric or hydrochloric acid promotes rather the dissolution of gypsum at the expense of calcite. Ca²⁺ cations would result from the dissolution of gypsum.

Among minor species, the average mass concentrations of anionic species and some cations such as NH₄ ⁺, Zn²⁺, and K⁺ show no difference between the two tubes regardless of sampling period. Only the Mg²⁺ mass concentration is significantly higher in the upward tube compared with the downward tube (P _summer = 2.31%; P _winter = 1.08%); the evolution of the concentration over time remains constant. It should be noted that the concentration of NO₃ ⁻ and NH₄ ⁺ are not significantly different with the slope of the tunnel or sampling period. This leads to considering the presence of ammonium nitrate (NH₄NO₃), aerosol-specific vehicle exhaust.

Among the anionic species, the identification of CH₃COO⁻ and HCOO⁻ anions reveals the presence of formic and acetic acids generated by engine exhausts and aqueous-phase oxidation of aldehydes. Only the alkanoic acids (C₁–C₄) are readily soluble in water. In addition, the results obtained by CG/MS show that (C₅–C₁₀) alkanoic acids are negligible. However, in aqueous layer, the butanoic acid can be oxidized by sulfuric acid and form CO₂ and acetic acid. Alkanoic acids have been the subject of particular attention, with focus on their presence in rainwater collected in urban or semiurban area (Kawamura et al., 2001; Pena et al., 2002; Willey et al., 2011) or in the urban or semiurban atmosphere itself (Khwaja, 1995; Kuwamura et al., 2000). From theses studies, it appears that formic and acetic acids are the most abundant species. This study shows the same result for the particles collected in the Grand Mare tunnel atmosphere. The ratio of formic to acetic acid is an indicator of the relative importance of direct emission (low ratio) and in situ formation by photochemical processes (high ratio) (Khwaja, 1995). The changes in this ratio can be used to infer influence of anthropogenic source (particularly vehicle) because automotive emissions contain relatively more acetic acid than formic acid (Avery et al., 2001). From our results, the HCOO⁻/CH₃COO⁻ average mass concentration ratio has been calculated and summarized in Table 1. The ratio values vary within the range from 0.5 to 0.66; this ratio remains constant regardless of the period of observation and the location point selected within the tunnel. It is attributed to automotive traffic. The results are consistent with those obtained by Khwaja (1995) on alkanoic acids present in a semiurban atmosphere, especially in the morning period associated with automotive traffic.

Interpretation of the results

This study in the Grand Mare tunnel sets out the real conditions of material exposure and highlights the presence of specific particulate pollutants. All the species identified are summarized in Figure 11. Table 2 detailed the water-soluble species present and organic compounds detected in particles. It also indicates the likely sources that we have selected, based on a literature survey. Due to climatic conditions inside the tunnel, the impact of such species on the degradation of materials is difficult to undertake without preliminary study of their individual action. Thus, laboratory studies under carefully controlled conditions have been made to identify the effect of each potentially aggressive particle and to investigate the potential synergetic effect of particle–gaseous effluent combinations (Table 3). Studies have been conducted on copper or zinc in different environments with different kinds of particle and gas effluents, at different relative humidity values. Selected particles studied are often soluble in water. Table 3 summarizes the experimental conditions listed from the literature data as well as the impact on the corrosion rate. From this survey, some interesting pieces of information can be extracted.

Figure 11. Summary of identified species.

Table 2. Organic and ionic soluble species detected in the particulate matter and the possible sources from the literature

		Sources
		Combustion/Unburned Species				Nonexhaust Emission
Organic Species	Ionic Soluble Species, S	Exhaust			Lubricating Oil	Brake Tire Road Wear	Resins, Paint, Plasticizers	Plant	Road Treatment	Construction Materials	Aerosol
Alkanes	Cl^−	P		S	P	P			S
Alkanoic acid	Na^+		P			P			S
N+S products	CO2 ^−		S			P				S
Phthalate	Ca^+				(P)		P			S
Ketone	SO4 ^2−		S				P
Aldehyde	NO3 ^−		P								S
FAMEs	NH4 ^+		P								S
BTEX	K^+		P
Phenol	Mg^2+				(P)		P			(S)
Alcohol	Zn^2+		P				P			(S)
PAH	F^−		P		P	P
Furan	CH3C OO^−	P		S	P		P	P
Alkenes	HCO O^−	P		S				P
Hopane			P
Halogene					P
Aromatic acid			P			P
Notes: Organic species—present: P; may be present: (P); water-soluble ionic species—present: S; may be present:(S)

Table 3. Experimental conditions of laboratory studies mentioned in literature

Parameter	Löbnig et al. (1994)	Löbnig et al. (1996)	Svensson and Johanssson (1993)	Askey et al. (1993)	Chen (2005)	Qu et al. (2005)	Lindström et al. (2000, 2002)
Particles	(NH4)2SO4	(NH4)2SO4	NaCl	Fly ash (oil, coal)	NaCl	NaCl/NH4Cl	NaCl, NH4Cl, ZnCl2, MgCl2, Na2SO4, MgSO4, ZnSO4, (NH4)2SO4
RH (%)	65, 75, 88	43, 60, 65, 71, 79	70, 95	65, 80, 90	55, 75, 95	80	95
θ (K)	373	300, 373			298	298	298
Materials	Cu	Zn	Zn	Zn—steel	Cu	Zn	Zn
			SO2: 0.78 ppm	HCl	SO2, O3, NO2		CO2: 350 ppm
Environment	Air–H2O		NO2: 1.06 ppm	SO2	SO2+NO2
					Time: 10 days		Time: 4 weeks
Impact on degradation of materials exposed	Effect with RH value at RH > DRH	Effect at RH ≥65% Effect at RH < DRH	NaCl: Important corrosion rate Corrosion rate function of NaCl amount Corrosion rate linked to RH value Degradation at RH < DRH Moderate or high amount of NaCl: SO2 small inhibitive action RH = 75% NaCl + SO2+NO2: No synergetic effect RH = 95% NaCl + SO2+NO2: Synergetic effect Small amount of NaCl: Synergic effect of (NaCl + NO2) on corrosion rate Effect of single addition of NaCl as opposed to repeated addition	Unpolluted atmosphere: Corrosion rate important with RH value and type of fly ashes HCl: Corrosion rate as function of HCl concentration at RH = 80% SO2: No effect with fly ashes SO2 + HCl: No additional effect with contamination by fly ashes	Corrosion attack at RH = 55% < DRH Corrosion acceleration with RH NaCl particles added: No effect of SO2, O3, NO2, and SO2+NO2 Synergetic effect: NaCl with high concentration of SO2+O3	NaCl, NH4Cl: Effect on corrosion rate: Effect NaCl + NH4Cl > Effect NH4Cl	NaCl: Most corrosive of studied salts NaSO4: Higher corrosion rate than others sulfate salts Corrosion rate function of amount of sodium on surface substrate CO2: Corrosion rate independent of temperature

The effect of NaCl particles on Zn substrate is important and significant on copper; the corrosion rate is linked to NaCl amount and rises with increasing RH. NaCl appears as the most corrosive salt studied. It also appears that NaCl induces the degradation at RH below the deliquescence value of the salt. It is well known that SO₂ associated with O₃ and or NO₂ has a synergetic effect on the corrosion of metals without particulate pollution. This effect becomes minor compared with impact of NaCl particles alone on copper. Indeed, the corrosion rate of copper during NaCl deposits is similar in both humid environment without pollutant and SO₂, NO₂, O₃, and SO₂+NO₂ polluted atmosphere; therefore, there is no synergistic effect between NaCl and gaseous pollutants. The conclusions regarding zinc are little different and depend on the pollutants. A slight synergism can be observed when the Zn substrate is exposed in environment containing small amount of NaCl and traces of SO₂. In the presence of moderate to high amount of NaCl, SO₂ has rather an inhibitive action. In combination with small amount of NaCl, NO₂ stimulates the corrosion. There is a synergistic effect of the combination NaCl+SO₂+NO₂ that is highly dependant on the relative humidity; no effect at 75% RH but significant at 95% RH. It was highlighted that the corrosion of zinc is highly linked to Na⁺ concentration; this result focused on the important role of Na⁺ cation. Many studies have also considered that poly anions present in the electrolyte had an influence on the degradation mechanism. The action of ammonium sulfate depends on the nature of substrate. Copper degradation appears only when the relative humidity reaches the deliquescence relative humidity (DRH) of this salt; about zinc substrate, the degradation occurs at a minimum value of 65% RH, which is below the DRH value. Note that in an atmosphere free of pollutants, the corrosion rate increases as a function of RH and the type of particle (inert, coal, or fly ashes). SO₂ has no impact on zinc substrate contaminated by fly ashes. The increase in atmospheric corrosion rate due to particulate contamination is predominantly due to the quantity of species that can be leached from the particulates accumulated on the substrate. If the leachate has a high pH or contains a significant quantity of inhibitive cations such as Ca²⁺, then the corrosion rate may not increase as significantly (Askey et al., 1993). The inhibitive effect is linked to the formation of insoluble corrosion products. Amount of Ca²⁺ cation is therefore a parameter. The Ca²⁺ concentration detected inside the Grand Mare tunnel does not allow asserting an inhibitive effect. In a tunnel-type environment, pollutant sources that may cause material degradation are numerous and can have an impact amplified by a synergistic effect. The concentration ratio NO₂/SO₂, calculated from the data collected inside the tunnel, has a value greater than 12 (Ameur-Boudabbous et al., 2012), accordingly, the mixed atmosphere SO₂+NO₂ is expected to have a clear synergic effect, especially above critical (RH, T) couple values, with 95–99% in winter and around 40% in summer. But adding NaCl can modify the impact of these effluents. It is therefore essential to quantify theses species (Na⁺, Cl⁻) in order to predict their potential impact and potential synergism in accordance with their respective concentration. Particulate matter rather contains a small amount of NaCl, about 3.5% of particulate mass. A synergetic effect of NaCl+NO₂ can exist in our experimental conditions. Taking into account the measured RH value (Figure 12), the synergetic effect of the combination NaCl+SO₂+NO₂ may be possible. The presence of hydroxysulfates among corrosion products during outdoor exposure campaigns (Jouen et al., 2000, 2004; Lefez et al., 2001) makes it necessary to quantify sulfates that may be present in the aqueous film in order to discuss the possible formation of such corrosion products in a semienclosed environment. Theses sulfates are connected, on the one hand to the presence of SO₂ and sulfuric acids vapor, and the other hand to the partial dissolution of CaSO₄·2H₂O. Chen (2005) shows that the oxidation of SO₂ to sulfate is much faster in the presence than in the absence of NaCl particles and the fast oxidation of SO₂ to sulfate occurs in the electrolyte droplets. With the amount of NaCl particles on copper surface lower than 15 μg/cm², higher corrosion rate was found with <1 ppm CO₂ than with 350 ppm CO₂ (ambient level); but with higher amount of NaCl (>15 μg/cm²), the corrosion was higher with 350 ppm CO₂ (Chen, 2005). It has been shown that CO₂ strongly influenced NaCl-induced atmospheric corrosion of zinc, aluminium, and magnesium, with a decreasing corrosion rate in the presence of ambient levels of CO₂ ( Lindström et al., 2000, 2002). CO₂ concentration inside the tunnel is higher than the ambient value. However, in the presence of SO₂, CO₂ is not expected to affect the corrosion rate (Chen, 2005). As a result, CO₂ has more limited impact in our experimental conditions due to the SO₂ concentration. The presence of NH₄NO₃ has been highlighted; it exists predominantly in the form of fine particles (<2 μm) from the reactions involving HNO₃ and NH₃ (Wall et al., 1988; Pierson and Brachaczek, 1988). The experimental data obtained in this study do not reveal the presence of (NH₄)₂SO₄. To our knowledge, no laboratory study has been conducted on the impact of NH₄NO₃ particles on materials such as copper and zinc. Because of its volatibility, NH₄NO₃ is very difficult to study and thus its hydration and dehydration properties have been the subject of rather limited studies (Lighstone et al., 2000). However, as (NH₄)₂SO₄ has an impact on the degradation of materials depending on the RH, a possible effect of NH₄NO₃ cannot be excluded. The relative humidity prevailing in the tunnel plays a key role in the degradation process. Table 4 summarizes the data extracted from the literature on relative humidity related to deliquescence and recrystallization of salts. From this table, (NH₄)₂SO₄, NH₄NO₃, and NaCl liquefy respectively at 80%, 62%, and 75.3% at ambient temperature. These salts remain in liquid phase, forming a metastable liquid state until the relative humidity linked to crystallization point is reached, at which point the highly concentrated particulate aerosol nucleates to form a solid. Regarding NH₄NO₃, it will deliquesce gradually and eventually become droplet until RH reaches its DRH value (Park et al., 2009). But laboratory experiments show that they never reach crystallization point, even at 8% RH (Dougle et al., 1998). Impurities can initiate crystallization in bulk salts, inducing the crystallization at RH higher than the crystallization point of the pure salt aerosol (Ianniello et al., 2011). Dougle et al. (1998) found that soot is highly unlikely to promote crystallization in ambient aerosols. Ambient aerosol contains more “organic” carbon than soot; the authors suggested that the crystallization of ambient aerosol has been induced by the more abundant organic carbon. Considering the relative humidity prevailing in the tunnel (Figure 12) and water retention in the particle layer leading to a RH value higher than that measured in the tunnel atmosphere, the crystallization point of these salts will never be reached. These salts remain in ionic form in the liquid phase and participate actively in the degradation process.

Table 4. Deliquescence (DRH) and recrystallization (RH) relative humidity of soluble salts

			Deliquescence	Recrystallization
Salt	DRH (%)	Temperature (K)	References	RH (%)	Temperature (K)	References
NaCl	75.3	298	Tang and Munkelwitz (1993)	44	293	Cohen et al. (1987)
	75	298	Xu et al. (2003)
NH4Cl	78	298	Seinfeld and Pandis (2006)
NH4NO3	61.8	298	Tang and Munkelwitz (1993)	<8	Undefined	Dougle et al. (1998)
	61.8± 0.3	298	Lightstone et al. (2000)
(NH4)2SO4	80.3	298	Tang and Munkelwitz (1993)	42	Undefined	Dougle et al. (1998)
	80 ±1	298	Hu et al. (2011)	33	298	Cizczo et al. (1997)
	80	Undefined	Dougle et al. (1998)

Figure 12. Evolution of maximum and minimum RH inside the tunnel as function of time.

CH₃COO⁻ and HCOO⁻ anions reveal the presence of the formic and acetic acids in the particulate matter. Notice the significant presence of acetic acid, detected by CG/MS, in the particulate matter collected on a sample holder structure. In a previous study, it has been shown that the relative contribution of formaldehyde and acetaldehyde is respectively 34.3% and 18% (Ameur-Boudabbous et al., 2012). The aqueous oxidation of aldehydes to organic acids (especially formic and acetic acids) could be a process in the formation of these last compounds. Due to their higher water solubility, organic acids such as formic and acetic acids are more effectively scavenged from the atmosphere into raindrops than aldehydes (Kawamura et al., 1996b). Taking into account the climatic conditions in the tunnel, these acids are present in the aqueous layer and contribute to the degradation of exposed materials. Corrosion products such as formate and acetate are found on sensitive items displayed in museum enclosures contaminated with acetic or formic acid vapors on lead substrate (Niklasson et al., 2008) or formic acid vapors on brass and lead (Gibson and Watt, 2010). It was therefore necessary to determine the concentration of these anions in order to take into account their effect on the degradation.

The influence of the nature of the particles on the corrosion of zinc has been observed through laboratory testing. It appears that the metal loss caused by the presence of ZnSO₄ or MgSO₄ particles is 8–10 times lower than that caused by the presence of Na₂SO₄. Similarly, the metal loss caused by the presence of chlorides such as MgCl₂ or ZnCl₂ is 4–6 times lower than that observed with NaCl (Lindström et al., 2002). In this study, the concentration of Zn²⁺ or Mg²⁺ in aqueous film is small; these cations are not expected to the degradation of materials.

The interaction of water with atmospheric particulate matter has been described in many research papers, focusing mainly on the study of relationship between relative humidity and water adsorbed on PM. A quantitative determination of adsorbed water was attempted only in a few studies. Water adsorption resulted to be a relevant parameter for fine particles and for particles containing water-soluble inorganic salt, mainly ammonium sulfate and hygroscopic organic species such as dicarboxylic acids (Canepari et al., 2013, and references cited). The water-absorbing capability of atmospheric aerosol particles depends in a complicated way on the hygroscopic properties of primary particles and on numerous atmospheric processes modifying the chemical composition of these particles (Kerminen, 1997). Kerminen (1997) points out that among primary anthropogenic particles, those resulting from fossil fuel combustion usually are hydrophobic or slightly hygroscopic, and that secondary particles formed via multicomponent nucleation routes in the atmosphere are expected to be hygroscopic. Vartiainen et al. (1994, 1996) showed that weights of diesel soot samples increased by 2% as relative humidity increased from 40% to 90% and these particles became more hygroscopic after 9–10 hr of exposure in smog chamber. Weingartner et al. (1995) came to the same conclusion for carbon particles and freshly combustion particles. The particulate matter collected inside the Grand Mare tunnel contains both hydrophobic and hydrophilic compounds. As regards the organic fraction, two sets appear: one comprising the species emitted by the exhausts (lubricating oil and fuel) and the second related to debris such as resins, paints, plasticizers, polymers, as well as wear particles of brakes, tires, and roads. The latter set is insoluble, inert, and hydrophobic in our experimental conditions, namely, a temperature ranging from 0 to 20 °C and a RH of 30–99%. This set has not a priori a direct impact on the deterioration of materials. However, this mixture of organic particles plus inorganic insoluble and very slightly soluble particles play an indirect role in the degradation of materials exposed. These agglomerates formed as a result of accumulation of particles over time are constituted by a wide range of capillaries, as shown in the SEM micrographs (Figure 6). Canepari et al. (2013) conducted a quantification of water in particulate matter such as soot from vehicle exhausts and road dust. They show a water contribution due to adsorbed moisture on the particles. They also found that moisture was adsorbed on both the molecular sieves but also inside the pores of the sieve. In this present study, the porous particulate layer behaves like an irregular sieve. Considering the climatic conditions prevailing in the tunnel (Figures 2 and 12), water adsorption takes place in both the porous particulate layer and inside the pores and reaches even the substrate-particle interface. Due to the presence of capillaries, water retention is giving birth to a permanent layer loaded with aggressive species at the particle-substrate interface. These agglomerates have therefore an indirect key impact on the degradation of materials; this particulate layer is similar to a sponge absorbing water vapor and gas-phase species.

Among species from engine exhaust, only organic acids and aldehydes may be involved in the degradation. As regards aldehydes, identified by CG/MS, their relative contribution is low; they will not cause any significant impact in terms of degradation. The agglomerates identified in this study may behave quite differently from the pure particles (NH₄NO₃, (NH₄)₂SO₄, NaCl) under varying humidity conditions. The organic fraction is expected to have a strong impact, especially due to the hygroscopicity of these multiconponent aerosols (Lighstone et al., 2000). These authors observed slight decrease of the DRH value for particulate mixture of ammonium nitrate/succinic acid, the latter being slightly soluble water. In addition, Xu et al. (2003) showed that formic acid vapors reduce the DRH value of NaCl significantly from 75% in the absence of acid to 68.5% at a pH of 2.7. The decrease is monotonic and continues to increase over the entire concentration range studied. According to these authors, the deliquescence occurs only in the presence of water layer; the DRH lowering is expected to be largest for those species that are both abundant and highly soluble. In tunnel exposure conditions, acetic and formic acids detected by the presence of CH₃COO⁻ and HCOO⁻ anions can significantly reduce DRH. Generally, the organic acids will also play an indirect role by reducing the DRH value. As the result, the degradation starts at lower relative humidity. The sulfuric acid condenses irreversibly on the NaCl aerosol (dry or wet) (ten Brink, 1998), which may occur on NaCl particles during the exposure conditions inside the Grand Mare tunnel. Nitric and sulfuric acids, which occur in much higher abundances than formic or acetic acid, can also contribute to a significant reduction of the DRH value.

Conclusion

This work presents the atmospheric particulate generated by traffic in a road tunnel in upper Normandy, France. Characterization of particulate matter by physical methods shows crystalline phases such as SiO₂, CaCO₃, and NaCl but also reveals the presence of CaSO₄·2H₂O and silicates. The elemental analysis confirms the results obtained by X-ray diffraction but also indicates the presence of elements such as Fe, Mo, Ti, K, Mg, Ba, Rb, and Br. Concerning the recognizable organic fraction, two groups appear: a majority consisting of n-alkanes, alkanoic acids, phthalates, ketones, and N+S compounds, especially benzothiazole, and the second, minority group is composed of BTEX, PAHs, FAMEs, furans, phenols, and alkenes. The results for alkanes, alkanoic acids, and PAHs are similar to those obtained in similar studies in tunnel. In this study, phthalates, ketones, and benzothiazole have been clearly identified. Besides the minority fraction, we have demonstrated the presence of FAMEs—having in the source the fuel DIESTER™, alkenes—associated with the exhaust, and phenols, but also furans linked to the presence of plastic debris. Organic species can be grouped into sets associated with two main sources, namely, engine exhausts and debris with wear particles, resuspended by the traffic. The analysis of water-soluble fraction helped to identify the ionic species involved in the degradation process, such as Cl⁻, SO₄ ²⁻, CH₃COO⁻, HCOO⁻, NH₄ ⁺, Na⁺ ,NO₃ ⁻. SO₄ ²⁻, with mass concentration significantly higher in the upward tube than the value measured in the downward tube, varies similarly to results obtained with SO₂. Other sources of chloride outside the identified source (road salt NaCl) exist and may include lubricating oil and fuel itself.

About degradation, the inorganic particles (insoluble and poorly soluble), debris, wear particles contribute to form porous agglomerates having a multiplicity of capillaries, promoting water absorption and leading to the formation of an almost constant aqueous layer. This set of chemically inert insoluble particles under our experimental conditions play a key role. Together, the organic compounds and especially organic acids play also a key role, contributing to a reduction of deliquescence relative humidity. Soluble salts such as NaCl and NH₄NO₃ have an impact on the degradation process. The particles inside the tunnel and the relative humidity are important factors to consider in the process of degradation of materials.

Acknowledgment

The authors thank the Interdepartmental Directorate of North West Roads and especially F. Harel and P. Langlois for their participation in the project.

Funding

This work was supported by the Carnot ESP Institute.