*2.1. Sampling*

Road dust samples were collected in September and October 2018 on representative streets of Viana do Castelo (latitude: 41◦4135.63 N; longitude: −8◦49 58.33 W), the most northern Atlantic city in Portugal with a population of about 40,000 in the most central urbanised area, although the municipality as a whole reaches 90,000 inhabitants. The city is located between Santa Luzia hill and the mouth of the river Lima. Its major industries are related to naval construction and repair, with one of the few large shipyards still in operation. It is also home to a large cluster of wind green electricity and car-parts industries. The old downtown streets are closed to tra ffic.

Three streets were selected for road dust sampling (Figure S1):

(S1) Suburban context—Rua Alto Xisto is a street with cobbled pavement made of granite cubes in a residential area on the outskirts of the city. One side of the street consists of terraced houses. The other side is flanked by an agricultural farm with some animals, and vineyard, corn and potato cultivation;

(S2) Urban context—Local road within the campus of the Higher School of Technology and Management. It is composed of stone mastic asphalt pavement and located a few meters from the beach and the shipyards;

(S3) Urban context—Avenida Combatentes da Grande Guerra is a central artery connecting to the train station. It is an avenue with shops, services and an elementary school. Its cobbled pavement is made of granite cubes.

The selection of streets was carried out in collaboration with the city council. It was considered that the urban area could be subdivided into three sectors: (i) a residential area with a lower population density and some agricultural influence, (ii) a central area with a higher volume of tra ffic where the main commercial activities and public services take place, (iii) and another area on the coast line, also with busy streets, but with industrial influence. For each sector, a street representing the dominant pavements and tra ffic was chosen. The various samplings took place on 3 weekends in September and October 2018 and implied the tra ffic cut by the police authorities. Road dust was collected on delimited lane sections using a vacuum cleaner, following the procedure described by the United Stated Environmental Protection Agency (USEPA) in its AP-42 document [17]. In each street, several subsamples were obtained by vacuuming segments of 20 m in length and width corresponding to the lanes. The collection of the first subsample started at a distance of 50 m from the intersection with another street. Distances of approximately 50 m were maintained between sampled segments. The weight of subsamples ranged from about 200 to 600 g. For each road, a composite sample from a minimum of 3 incremental subsamples was created.

#### *2.2. Geochemical, Mineralogical and Morphological Characterisation*

After sampling, the vacuum cleaner bags were stored in polyethylene bags and brought to the laboratory, where the USEPA methodology for analysis of surface/bulk dust loading samples was followed [18]. Samples were wet sieved with addition of distilled water through standard Taylor screens, dried at ~40 ◦C, and weighed in a microbalance with 1 μg sensitivity. While particles >1 mm were rejected, the fraction <0.074 mm (F1) and that between 0.0074 and 1 mm (F2) were retained for further analyses.

The chemical composition of the road dust fractions was determined by X-ray fluorescence (XRF) using an Axios PW4400/40 X-ray (Marvel Panalytical, Almelo, The Netherlands) fluorescence wavelength dispersive spectrometer. The limit of detection (LOD) (i.e., the concentration at which it is possible to distinguish the signal from the background) is provided in Table 1. Mineralogy was determined by X-ray di ffraction (XRD) using a X'Pert-Pro MPD di ffractometer (Marvel Panalytical, Almelo, The Netherlands) with a Cu-K α radiation at 45 kV, 40 mA, and with a step size of 0,02◦2θ, in 2◦–90◦ 2θ angle range. A Hitachi S-4100 scanning electron microscope (SEM) equipped with a Quantax 400 Energy Dispersive Spectrometer (EDS) (Bonsai Advanced, Madrid, Spain) was used for point and area chemical analyses. Particle size distributions of road dusts were determined by X-ray sedimentation technique with a Sedigraph III Plus grain size analyser (Micromeritics Instrument Corp., Norcross, GA, USA). This technique for determining the relative mass distribution of a sample by particle size is based on two physical principles: sedimentation theory (Stokes' law) and the absorption of X-radiation (Beer-Lambert law). Precision and accuracy of analyses and digestion procedures were monitored using internal standards, certified reference material and quality control blanks.

#### *2.3. Estimation of Emission Factors*

Road dust emission factors (*EFs*) were estimated using the equation outlined in the AP-42A document of USEPA [19] for paved roads:

$$EF = k \text{ (sL/2)}^{0.65} \times (\text{W/3})^{1.5} - \text{C},\tag{1}$$

where:

> *EF* = PM10 emission factor (g veh−<sup>1</sup> km−1),

*k* = particle size multiplier for particle size range and units of interest (0.46 g veh−<sup>1</sup> km−<sup>1</sup> for PM10),

*s* = surface material silt content, defined as particles that pass through a 200-mesh screen, which corresponds to 74 μm (%),

*L* = surface material loading, defined as mass of particles per unit area of the travel surface (g m<sup>−</sup>2), *W* = average weight (tons) of the vehicles travelling the road (a value of 2 tons was assumed),

*C* = emission factor for 1980's vehicle fleet exhaust, brake wear and tyre wear (0.1317 g veh−<sup>1</sup> km−<sup>1</sup> for PM10).

#### *2.4. Enrichment Index*

The enrichment index (*EI*) of an element is based on its concentration and the concentration of a reference element. The latter is chosen based on its low occurrence variability in order to identify geogenic and anthropogenic contributions. Due to the abundant natural occurrence on Earth's crust, Al was selected for this study. *EI* was calculated as follows:

$$EI\_x = \left(X / \mathbb{C}\_{ref}\right) \text{sample} / \left(X / \mathbb{C}\_{ref}\right) \text{crust}\_\prime \tag{2}$$

with *EIx* is the enrichment index of the element *X*, *X* the concentration of the element and *Cref* the concentration of the reference element (Al). The Earth's crust individual elemental composition was taken from Riemann and Caritat [20]. *EI* < 1 indicates no enrichment (natural contribution), 1 ≤ *EI* <3 minor anthropogenic enrichment, 3 ≤ *EI* < 5 moderate anthropogenic enrichment, 5 ≤ *EI* < 10 moderately severe anthropogenic enrichment, 10 ≤ *EI* < 25 severe anthropogenic enrichment, 25 ≤ *EI* < 50 very severe anthropogenic enrichment, and *EI* ≥ 50 extremely severe anthropogenic enrichment [21].

#### *2.5. Human Health Risk Assessment of Exposure to Potential Toxic Elements in Road Dust*

A human health risk assessment assumes that residents, both children and adults, are directly exposed to potential toxic elements through three different routes: ingestion, dermal absorption and inhalation of particles [22–24]. For road dust, it was assumed that intake rates and particle emission are similar to those established for soils. Equations (3)–(5) were used to estimate the chronic daily intake (*CDI*route, ingestion and dermal in mg kg−1·d−1; inhalation in mg m<sup>−</sup><sup>3</sup> for non-carcinogenic effects, and μg m<sup>−</sup><sup>3</sup> for carcinogenic effects) of each exposure route and for separated elements:

$$\text{CDI}\_{\text{ing}-\text{nc}} = \frac{\mathbb{C} \times \text{IngR} \times EF \times ED}{BW \times AT\_{\text{nc}}} \times 10^{-6} \text{ } \tag{3}$$

$$\text{CDI}\_{\text{drm}-\text{nc}} = \frac{\text{C} \times \text{SA} \times \text{SAF} \times \text{DA} \times \text{EF} \times \text{ED}}{\text{BW} \times \text{AT}\_{\text{nc}}} \times 10^{-6} \text{ } \tag{4}$$

$$\text{CDI}\_{\text{inh-nc}} = \frac{\text{C} \times \text{InhR} \times EF \times ED}{PEF \times BW \times AT\_{\text{nc}}},\tag{5}$$

where, *C* (mg kg−1) is the concentration of the element in road dust, *IngR* is the ingestion rate (200 and 100 mg d−<sup>1</sup> for children and adults, respectively), *InhR* is the inhalation rate (7.6 and 20 m<sup>3</sup> d−<sup>1</sup> for children and adults, respectively), *EF* is the exposure frequency (180 d yr<sup>−</sup>1), *ED* is the exposure

duration (6 and 24 years for non-carcinogenic effects in children and adults, respectively, and 70 years is the lifetime for carcinogenic effects), *BW* is the average body weight (15 and 70 kg for children and adults, respectively), *ATnc* is the averaging time for non-carcinogenic effects *(AT* days = *ED* × 365), *SA* is the exposed skin area (2800 and 5700 cm<sup>2</sup> for children and adults, respectively), *SAF* is the skin adherence factor (0.2 and 0.07 mg cm<sup>−</sup><sup>2</sup> for children and adults, respectively), *DA* is the dermal absorption factor (0.03 for As and 0.001 for other elements), and *PEF* is the particulate emission factor (1.36 × 10<sup>9</sup> m<sup>3</sup> kg−1) [22–26].

For each element and route, the non-cancer toxic risk was estimated by computing the chronic hazard quotient (*HQ*, Equation (6)) for systemic toxicity [24]. A *HQ* less than or equal to 1 indicates that adverse effects are not likely to occur, and thus can be considered to have negligible hazard, while *HQ* > 1 suggests that there is a high probability of non-carcinogenic effects to occur. To estimate the overall developing hazard of non-carcinogenic effects, it is assumed that toxic risks have additive effects. Therefore, it is possible to calculate the cumulative non-carcinogenic hazard index (*HI*), which corresponds to the sum of *HQ* for each pathway (Equation (7)) [27].

$$HQ\_{\text{route}} = \frac{CDI\_{\text{route}}}{R\_f D\_{\text{route}}},\tag{6}$$

$$dH = \sum HQ = HQ\_{\rm ing} + HQ\_{\rm drum} + HQ\_{\rm inh'} \tag{7}$$

with *RfD* being the chronic reference dose (values estimated as given in RAIS) [24].

The probability of an individual to develop any type of cancer over his lifetime (Risk), as a result of exposure to the carcinogenic hazards, was computed for each route according to Equation (8). The carcinogenic Total Risk is the sum of risk for each route (Equation (9)). A cancer risk below 1 × 10−<sup>6</sup> is considered insignificant, being this value the carcinogenic target risk. A risk above 1 × 10−<sup>4</sup> (a probability of 1 in 10,000 of an individual developing cancer) is considered unacceptable [22–24,27]:

$$\text{Risk}\_{\text{route}} = \mathbb{C}DI\_{\text{route}} \times \mathbb{C}SF\_{\text{route}} \tag{8}$$

$$\text{Total Risk} = \sum \text{Risk} = \text{Risk}\_{\text{ing}} + \text{Risk}\_{\text{drum}} + \text{Risk}\_{\text{lium}} \tag{10}$$

$$\text{=} = \text{CDI}\_{\text{img-ca}} \times \text{CFS}\_{\text{img}} + \text{CDI}\_{\text{inhr-ca}} \times \text{ILIR} + \frac{\text{CDI}\_{\text{lbm-ca}} \times \text{CSF}\_{\text{ing}}}{\text{ABS}\_{\text{gl}^i}} \tag{9}$$

$$\text{CDI}\_{\text{imh-ca}} = \frac{c \times \text{Img}\_{\text{adj}} \times EF}{AT\_{\text{cr}}} \times 10^{-6} \,\text{s} \tag{10}$$

$$\log R\_{\text{adj}} = \frac{ED\_{\text{child}} \times \log R\_{\text{child}}}{BW\_{\text{child}}} + \frac{(ED\_{\text{residual}} - ED\_{\text{child}}) \times \log R\_{\text{adult}}}{BW\_{\text{adult}}},\tag{11}$$

$$\text{CDI}\_{\text{inh-ca}} = \frac{\mathbb{C} \times EF \times ET \times ED}{\text{PEF} \times 24 \times AT\_{ca}} \, ^\circ \tag{12}$$

$$\text{CDI}\_{drm-ca} = \frac{c \times DA\_d \times EF \times DSF\_{adj}}{AT\_{ca}} \times 10^{-6} \,\text{s} \tag{13}$$

$$\text{DSF}\_{\text{adj}} = \frac{\text{ED}\_{\text{child}} \times \text{SA}\_{\text{child}} \times \text{SAF}\_{\text{child}}}{\text{BW}\_{\text{child}}} + \frac{(\text{ED}\_{\text{residual}} - \text{ED}\_{\text{child}}) \times \text{SA}\_{\text{adult}} \times \text{SAF}\_{\text{adult}}}{\text{BW}\_{\text{adult}}} \tag{14}$$

where *CSF* is the chronic slope factor (ingestion, (mg kg−<sup>1</sup> <sup>d</sup>−1)−1; dermal*, CSFing*/*ABSgi*), *ABSgi* is the gastrointestinal absorption factor, *IUR* is the chronic inhalation unit risk ((μg m<sup>−</sup>3)−1), *DFSadj* is the soil dermal contact factor-age-adjusted (362 mg yr kg−<sup>1</sup> <sup>d</sup>−1), *ATca* is the averaging time carcinogenic effects (*AT* days = *LT* × 365) and *ET* is the exposure time (8 h <sup>d</sup>−1) [27].

#### **3. Results and Discussion**

#### *3.1. PM10 Emission Factors*

For granite cube paved streets, very close PM10 emission factors were found: 277 mg veh−<sup>1</sup> km−<sup>1</sup> (central avenue, S3) and 330 mg veh−<sup>1</sup> km−<sup>1</sup> (street on the outskirts of the city, S1). On the asphalt paved street (S2), a much lower PM10 emission factor was obtained (49 mg veh−<sup>1</sup> km−1). Using an in-situ resuspension chamber, PM10 emission factors in the range 12.0–29.4 mg veh−<sup>1</sup> km−1, averaging 18.6 mg veh−<sup>1</sup> km−1, were determined for asphalt roads in the city of Oporto [1]. However, a much higher value of 1082 mg veh−<sup>1</sup> km−<sup>1</sup> was estimated for a cobbled pavement in the same city. According to the researchers, this grea<sup>t</sup> emission factor was due to the very high roughness of cobblestones, which may have promoted the build-up of road sediments. Moreover, the joints between granite cubes filled with soil may have constituted an additional source of resuspendable dust. Following the USEPA methodology, PM10 emission factors for typical urban roads in Milan within 13–32 mg veh−<sup>1</sup> km−<sup>1</sup> were recently reported, which were found to be in the central range of the literature values in Europe [28]. Based on micro-scale vertical profiles of the deposited mass, PM10 emission factors from 5.4 to 9.0 mg veh−<sup>1</sup> km−<sup>1</sup> were obtained at inner-roads of Paris, whilst a higher value was estimated for the ring road (17 mg<sup>−</sup><sup>1</sup> veh−<sup>1</sup> km−1) [16]. Based on a literature review, Boulter et al. [29] summarised the available information on emissions from road dust resuspension for the United Kingdom, USA, central and northern Europe. Differences in emission factors may be associated with local factors, such as road pavement type, regularity of street washing and precipitation events. On the other hand, while in Southern Europe high solar radiation is usually recorded, favouring the mobility of deposited particles, winter sanding or salting of roads and the use of studded tyres in Scandinavian and Alpine regions may promote the enhancement of road dust loadings. Road PM10 emission factors documented for Europe are, in general, lower than those reported for the USA. However, it must not be forgotten that American databases are older than European ones.

#### *3.2. Geochemical Characterization of Dust and Enrichment Index*

Road dusts are known for their heterogeneous mix with diverse natural and anthropogenic origins. The composition can vary depending on geographical location, resuspended soil, atmospheric deposition and anthropogenic sources, which include traffic related particles such as metallic components, eroded road pavement, but also building construction and demolition, and power generation [14,30].

Cluster analysis of the XRF results (Figure S2) of the studied samples (S1 suburban environment influenced by agricultural activities; S2 and S3 urban streets) confirmed the chemical difference between the two analysed fractions (F1 with samples <0.074 mm; F2 with sizes >0.074 mm and <1 mm). Elemental concentrations (Table 1) revealed Si as the most representative constituent, followed by Al > K > Fe > Ca > Na. The highest Si content was found in F2 (fraction >0.074 mm and <1 mm). The enrichment index (Figure S3) suggests a low influence of anthropogenic activities in both size fractions (*EI* < 1.5). Crustal erosion and parent materials (e.g., Variscan granite parent rock) may also influence the concentrations of other elements, such as Fe, Al, Mg, Mn, Rb, Na, Ti, Mo, V and Zr [31]. Previous studies sugges<sup>t</sup> that heavy metals such as Mn, V, Cu, Fe, Ni, Pb and Zn are linked to traffic emissions [32,33]. Low enrichment indices (<3) were also obtained for manganese, indicating a major natural source. However, this element is used in fuel additives, aluminium based alloys offering a high corrosion resistance and formability, vehicle applications, such as inner panels, and heater and radiator tubes.


**Table 1.** Summary descriptive statistics of elements analysed in two size fractions of road dusts from three selected locations. Concentrations in mg kg−1.

F1—fraction <0.074 mm; F2—fraction >0.074 mm and <1 mm; min—minimum; max—maximum; med—median; SD—standard deviation; nd—not detected.

Iron, with higher concentrations in fraction F1 (max = 39,895 mg kg−1), showed comparable results to those obtained for road dusts in Xi'an [34], Shanghai [35], Hong Kong [35], Beijing [35], Delhi [36] and Madrid [37], suggesting a significant geogenic contribution, which is confirmed by *EI* < 3. Elements such as Al, Fe, Ti, Zr and Na have a potential source in soil resuspended dusts and marine spray (*EI* < 3). The anthropogenic contribution may be linked to the application of these elements in the production of metal alloys commonly used in vehicles (e.g., Fe as a major component of steel and associated rust); Al alloys associated with other elements such as Mg, Mn and Cu to reduce vehicle weight; alloys with Mn to avoid corrosion and deformation; sulphides (such as Sb2S3, MoS2 and SnS) and sulphates such as barite (BaSO4) applied in brakes; titanium oxide (TiO2) used as a pigment in white paints; and aluminium oxide and iron oxides also employed in brakes [33,38]. The detection of Zr may be, in part, related to vehicle exhaust emissions, since ceria/zirconia (CeO2/ZrO2) mixed oxides have become an essential component of three-way catalysts [39]. Asphalt materials are usually rich in Al, Si, K, and Ca, with smaller amounts of Fe, S, Mg, Zn, and Ti [40]. Elements such as P and K are commonly used in agriculture activities (e.g., phosphate, potassium nitrate). Higher concentrations of these elements were found in samples from the suburban location, indicating the contribution of the surrounding agricultural environment to road dust. The median concentrations of lead, chromium and copper in samples of urban streets, particularly in F1 fraction, were similar to those of other studies (e.g., Thessaloniki [41], Shiraz [42], Urumqi and Zhuzhou [43]). Lead, chromium and copper concentrations in the ranges 48–375, 2.0–498 and 47–995 mg kg−1, respectively, have been reported for street dusts of different cities on various continents [44], and references therein.

The enrichment index of Zn in the finest fraction (F1) ranged from 10.6 to 43.5, with a moderate (suburban location) to very severe anthropogenic enrichment (urban streets). Zn has been reported to be present as a minor constituent of silicate minerals and linked to fly-ash particles and to traffic related materials. Tyre rubber (ZnO and Cu/Zn layers formed during vulcanisation) and break wear resuspended particles are major sources of Zn, together with Cu, in urban areas [45]. Zn is also used as engine oil additive. Sternbeck et al. [46] suggested that fuel combustion may be a significant source of Zn. Nickel is a ubiquitous natural metal, used in the production of stainless steel and other nickel alloys with high corrosion and temperature resistance. It is considered a tracer of oil combustion [47]. The high concentration of Ba is likely related to brake wear since BaSO4 (barite) is used as filler for brake lining materials [48].

Antinomy concentrations revealed a very high enrichment index (F1 = 29.3; F2 = 158) in road dust from the central avenue (S3). This is a busy street with frequent braking, especially in front of the elementary school. Since there is no dedicated parking, parents stop at the lane to drop off or pick up their children. Antimony increases the hardness and mechanical strength of lead and is a significant metallic component in brake linings. It is also used in batteries and antifriction alloys, as additive in some types of oils, and applied in semiconductors and Sb2O3 in rubber vulcanisation flame retardants [40,49]. Cooper in the coarser fraction (F2) of samples from urban streets showed an *EI* of 73.6 and 18.6, while the corresponding value for the suburban location was 5.3. This element is commonly associated with traffic related activities, e.g., a component of brake pads wear. A mean Cu/Sb concentration ratio of 6.3 was obtained, in line with the 4.6 ± 2.3 diagnostic criteria proposed by Sternbeck et al. [46] for brake wear particles.

The maximum lead concentration in the coarse fraction did not exceed the minimum value recorded in the finest fraction. The highest concentration was found in the city centre (310 mg kg−1), while the minimum was recorded in the suburban area with rural influence (81 mg kg−1). According to Ferreira [50], the concentration of Pb in natural soils in this region is in the range 30–45 mg kg−1. Thus, the higher values observed in the present study sugges<sup>t</sup> anthropogenic sources. This element presented an *EI* of 22.1 in fraction F2 from urban locations, whilst a value of 4 was determined for the suburban area. Lead and lead compounds are used in batteries and as pigments in paints. It has been documented that lead weights, which are used to balance motor vehicle wheels, are lost and deposited in urban streets, that they accumulate along the outer curb, and that they are rapidly abraded and ground into tiny pieces by vehicle traffic [51]. Lead oxide is a component of brake friction materials [15]. Its elevated concentrations in urban dust could be a consequence of common use of PbO4 as a gasoline additive in Portugal until the year 2000. Most of Pb is bound to stable fractions and only a negligible percentage is mobile, which contributes to its long persistence in the environment [15,45]. Tin revealed high *EI*, ranging from 11.6 in road dust samples from the suburban location to 36.3 in samples from urban streets. Tin was found to be one of the major elements in bulk brake pad samples, as well in airborne nano/micro-sized wear particles released from low-metallic automotive brakes [52]. Urban samples revealed a high W anthropogenic enrichment (*EI* up to 20.9). Tungsten is linked to break pads and tyre wear [53]. The abundant use of Br as flame retardant in several types of materials may explain the high *EI* of this element (23.3 and 47 in F1 of urban samples) [54].

An arsenic concentration of 180 mg kg−<sup>1</sup> (*EI* = 72.7), six times higher than in urban samples, was observed in fraction F1 of the suburban sample. Arsenic rich particles in this road dust sample may originate from resuspended agricultural soil. It should be borne in mind that the street where the sampling took place is flanked by a farm. In addition to the abundant natural origin, there are many agricultural sources of arsenic to the soil, from pesticide application, livestock dips, organic manure to phosphate fertilisers [55]. Elemental As is used in the manufacture of alloys, particularly with lead (e.g., in lead acid batteries) and copper. Arsenic compounds are also extensively used in the semiconductor and electronics industries, in catalysts, pyrotechnics, antifouling agents in paints, among others. Fossil fuel combustion is considered a major contributor of anthropogenic As emissions to the air (mainly AsIII). Arsenic is present in the air of suburban and urban areas mainly in the inorganic As<sup>V</sup> form. Background concentrations in soil range from 1 to 40 mg kg−1, with a mean value of 5 mg kg−<sup>1</sup> [56], much lower than the levels found in this study.

#### *3.3. Mineralogical Composition*

The mineralogy of the two particle size fractions of the road dust samples was compared. X-ray diffraction results showed that road dust samples consist primarily of mineral matter accounting for a minimum of ~60%. The most abundant mineral was quartz [SiO2], especially in the coarse fraction. Quartz has higher structural hardness than other minerals preventing physical weathering and abrasion of road surfaces. Other minerals also present were muscovite [KAl3Si3O10(OH)2], albite [NaAlSi3O8], kaolinite [Al2Si2O5(OH)4], microcline [KAlSi3O8], Fe-enstatite [(Fe,Mg)2Si2O6] and graphite [C]. Minor proportions of calcite [CaCO3] and rutile [TiO2] were also observed, mostly in the coarse fraction of road dust collected in the suburban location. Clay forming minerals increase with the decrease of particle size. A significant proportion of amorphous content was detected in all samples, particularly in the finest fraction. This content may originate from weathered minerals, clays with low detection limit, organic matter and anthropogenic sources [57].

The bulk composition suggests that the SiO2 content of all samples reflects the abundance of leucocratic phases (quartz and feldspar) in the coarser fractions of road dust, while Fe, Ti and Mg indicate the higher abundance of melanocratic phases (biotite, garne<sup>t</sup> and hornblende) in the finer fractions. The ternary diagram Al2O3/CaO+Na2O+K2O/FeO+MgO (Figure 1) [58] of the studied samples defines three compositional trends, suggesting slightly different sources for each location. The composition of road dust from the suburban street (S1) reveals a feldspars/muscovite trend, S2 (urban) plots in the feldspars/biotite, and S3 (central avenue) in the feldspars/biotite with more compositional Ca. Feldspars are the most abundant mineral in all fractions. The ternary diagram Al2O3/CaO+Na2O/K2O suggests that all samples are closer to K-feldspar composition than plagioclase, and present also high content of micas (lever rule). This graphical analysis is in line with the XRD analysis, being quartz the most abundant mineral that is not plotted in these ternary diagrams.

The chemical index of alteration (CIA = (Al2O3/(Al2O3+CaO\*+Na2O+K2O)) × 100, where CaO\* is CaO if CaO < Na2O, but if CaO > Na2O, CaO = Na2O) [58,59] indicates: (a) low weathering if 50–60; (b) intermediate weathering if 60–80; and (c) intense weathering when >80 [60]. The coarser fraction of road dust from urban streets (S2 and S3) reveal a low weathering, with CIA = 57.3 and 58.4, respectively. The same road dust fraction of the sample collected at the suburban location (F2 of S1) indicate a low intermediate weathering with CIA = 61.6. The finest fractions (F1) of all samples present intermediate weathering, ranging from 61.3 to 67.6.

To assess the alumina abundance compared to other major cations, the index of compositional variation (ICV = ((Fe2O3+K2O+Na2O+CaO+MgO+MnO+TiO2)/Al2O3) was calculated [61,62]. Minerals such as kaolinite, illite and muscovite present ICV < 1, whereas minerals such as plagioclase, K-feldspar, biotite, amphiboles and pyroxenes have ICV > 1. The road dust sample from the suburban location influenced by agricultural activities, for both size fractions, presented an ICV < 1 (0.84 and 0.92), while the urban samples S2 and S3 revealed an ICV > 1 for both fractions (1.06 to 1.73), indicating an enrichment in rock forming minerals. The limitation of using the elemental composition to identify road dust sources should be noted, as not only the natural but also the anthropogenic contribution must be considered.

**Figure 1.** Ternary diagrams (**a**) Al2O3—CaO+Na2O+K2O—FeO+MgO, and (**b**) Al2O3—CaO+Na2O +K2O—FeO+MgO of the studied samples (adapted from [58]). Stars plotted indicate the average composition of the Earth's crust and granite [20]. S1—suburban street, cobbled pavement; S2—urban street, asphalt pavement; S3—central avenue, cobbled pavement.
