Road Dust Emission Sources and Assessment of Street Washing Effect

Although previous studies report on the effect of street washing on ambient particulate matter levels, there is a lack of studies investigating the results of street washing on the emission strength of road dust. A sampling campaign was conducted in Madrid urban area during July 2009 where road dust samples were collected in two sites, namely Reference site (where the road surface was not washed) and Pelayo site (where street washing was performed daily during night). Following the chemical characterization of the road dust particles the emission sources were resolved by means of Positive Matrix Factorization, PMF (Multilinear Engine scripting) and the mass contribution of each source was calculated for the two sites. Mineral dust, brake wear, tire wear, carbonaceous emissions and construction dust were the main sources of road dust with mineral and construction dust being the major contributors to inhalable road dust load. To evaluate the effectiveness of street washing on the emission sources, the sources mass contributions between the two sites were compared. Although brake wear and tire wear had lower concentrations at the site where street washing was performed, these mass differences were not statistically significant and the temporal variation did not show the expected build-up after dust removal. It was concluded that the washing activities resulted merely in a road dust moistening, without effective removal and that mobilization of particles took place in a few hours between washing and sampling. The results also indicated that it is worth paying attention to the dust dispersed from the construction sites as they affect the emission strength in nearby streets.


INTRODUCTION
Road dust has been acknowledged as an important source of urban air pollution as resuspended particles from paved roads can considerably enhance atmospheric particulate matter (PM) levels (Thorpe and Harrison 2008;Amato et al., 2009a;Karanasiou et al., 2009Karanasiou et al., , 2011)).Particulate emissions from paved roads are a complex mixture derived from different sources.The particles deposited on the road surface originate from direct emissions from vehicles tailpipe and particles emitted by tire and brake wear processes.In addition, particles that settle onto the road from both nearby and distant sources become part of the mixture and can be subsequently resuspended by vehicles or wind.
Several toxic pollutants, embedded in road dust, such as heavy metals and Polycyclic Aromatic Hydrocarbons, PAHs (Rogge et al., 1993) have been associated to increased risk of adverse effects to human health (Nriagu, 1988;Mittal and Grieken, 2001;Lee and Dong, 2011).Therefore, characterising the composition and the sources of road dust particles is important for the development of appropriate mitigation strategies to protect human health.
Previous studies on the chemical composition of road dust particles found that road dust consists mainly of mineral particles but it is also enriched in carbon species, heavy metals and metalloids (Vega et al., 2001;Ho et al., 2003;Zhao et al., 2006;Amato et al., 2009b).Concerning the sources of road dust particles only a few studies have identified and quantified them.In a recent study carried out in three European cities four sources of road dust were identified: road abrasion/mineral dust, vehicles exhaust emissions, brake and tire wear (Amato et al., 2011).In the same study it was demonstrated that road abrasion/mineral dust is the predominant source in Spanish cities (with an average mass contribution of 60%), but represents only 30% of road dust loadings in Zürich (Switzerland) where contributions were more equally distributed among the four main sources of road dust.In the study of Jia et al. (2011) conducted in three Chinese cities, road dust originated mainly from soil dust, coal and cement dust and their mass contributions to road dust were 48%, 24% and 19% respectively.
One of the methods to mitigate road dust resuspension is street sweeping and washing as it might reduce the amount of dust on the road surface and/or reduce their ability to suspend as the increased moisture might capture the particles on the road surface (Amato et al., 2010).Local authorities normally use water flushing in combination with sweeping given that the water jet alone could hardly displace dust till achieving the sewage system unless a large water flow is used.
In the study of Norman and Johansson (2006) in Sweden, street washing with high-pressure water systems resulted in marginal reduction of the ambient PM 10 levels (~6%).For Madrid urban area Karanasiou et al. (2011) found that ambient PM 10 daily levels after street sweeping and washing were 2-15% lower than those present during dry, unwashed conditions.In that study street washing had a positive effect on the resuspension of road dust particles that lasted only for a few hours.In a similar study carried out in Barcelona the reduction in ambient PM 10 due to street washing and sweeping activities was 7-10% (Amato et al., 2009c).However, in the abovementioned studies parameters like background conditions and meteorology are mentioned as factors that could influence the variation of PM 10 concentrations along with street washing.Prior to the present study, Amato et al. (2009c) report on the effectiveness of street sweeping and washing on the inhalable road dust load (fraction below 10 µm).They have found high efficiencies (> 90%) in reducing the mobility of road dust particles by street washing activities.
The principal aim of the present study is to quantify the sources of inhalable road dust and at the same time examine the impact of street washing on the road dust sources.In attempt to identify the possible diurnal patterns of road dust accumulation the samples were collected twice every day in two sites, namely Reference site where the road surface was not washed and Pelayo site where street washing was performed every night.The sources of road dust were resolved by means of receptor modeling and the mass contribution of each source was calculated for the two sites.The novelty of this work lays in the evaluation of the washing effect on the individual road dust sources.

Measurement Campaign
An intensive campaign of road dust sampling was conducted during July 2009 in central Madrid.The city of Madrid is characterized by dense road traffic, high population density and abundant commercial activity.Madrid is one of the few big European cities with low relevant industrial activity nearby, thus atmospheric pollution can be considered to be typically urban (Artiñano et al., 2003).Recently, Karanasiou et al. (2011Karanasiou et al. ( , 2012) ) report that traffic emissions dominate PM concentrations in Madrid urban area with vehicles emissions and road dust contributing to ambient PM 10 concentrations by 60% approximately.Other sources that were identified in the urban area of Madrid were the mineral dust and secondary aerosol.The local authorities of Madrid employ street sweeping and washing daily in a major part of the urban area of the city mainly for aesthetic and health reasons.With the aim of characterizing the effect of street washing in the road dust emission strength, two locations were selected across the city centre and within the area where street washing was applying on a routine basis, Reference and Pelayo sites (Fig. 1).In Reference site the road surface was not washed while at Pelayo site the road surface was washed daily at 23:00 h (local time).The sampling campaign lasted nine consecutive days (9-17 July 2009).The meteorological conditions during the sampling campaign were stable, with no precipitation, and a similar daily temperature variation (21-43°C).Road traffic along Velázquez St. (Reference site) and Pelayo St. (Pelayo site) was dominated by passenger cars (85%), followed by light duty vehicles (7%), motorcycles and buses.Traffic was almost constant between weekdays, typically 32,000 vehicles per day pass through the sampling sites.The highest traffic density occurred around 09:00 h local time (2,500 veh/h).
Street washing was carried out by the urban cleaning agency using a pressured (minimum pressure 10 Kp/cm 2 ) groundwater volume of 4 L/m 2 for each washing.In addition, previously to this wash a mechanical sweeper vehicle was employed with the aim of vacuuming coarser deposited particles.The mechanical sweeper vehicle first moistens sediments, then two rotating brooms (85 cm diameter at a speed of 80 rounds per minute -rpm) sweep and afterward vacuum clean street sediments at a low driving speed of 5 km/h, followed by high pressure washing.The street cleaning vehicles used were Iveco Stralis, (Fig. 2) while the water supplied was recycled groundwater that then was moved along into the urban drainage system.The washed area comprised sidewalks, active lanes and parking lanes.
Sampling areas were selected on the outer lane (on the right lane at the Pelayo site and on the left lane at the Reference site) of the street, excluding the gutter where mass is not directly resuspended.Both sampling sites had similar characteristics concerning the traffic type and volume so they were influenced by the same sources.The road dust sampling protocol included two samplings per day: in the morning (09:00 h, local time) and evening hours (18:00 h, local time) when the traffic peaks.Sampling was performed in a surface area of 0.5 m -2 for 15 min.This procedure focused on examining the diurnal trends of road dust loadings and also to investigate possible build-up of the particles after street washing activities.

Road Dust Sampling Procedure
Road dust was collected using the sampling device described in Amato et al. (2009).Briefly a field resuspension chamber is designed to directly vacuum, at an air flow rate of 25 L/min, the mobile fraction below 10 µm of road dust onto filters at sampling sites.The mobile road dust term is defined as the portion of road dust reservoir below 10 µm that is also capable to suspend, excluding those particle stacked on the road surface by moisture or bond together into aggregates larger than 10 µm (Amato et al., 2012a).The measurement device used does not permit to collect the fixed portion of road dust, which is less likely to suspend in air.Road dust particles were aspired from the pavement of active traffic lanes, using a Becker pump powered by a Honda field generator (located at some distance downwind with respect to the sampling area).Particles were immediately resuspended in a PVC deposition chamber and the particles small and/or light enough entered a Negretti stainless steel elutriation filter designed to allow passage to only PM 10 .The particles able to penetrate this barrier were finally collected on 47 mm diameter quartz fiber filter (Munktell).Sampling was performed in a surface area of 0.5 m -2 for 15 min.The sampling area was gradually moved from one area to another allowing us to sample different areas in each sampling.At both sites each sampling was duplicated, collecting 72 samples in total.

Chemical Analysis
All quartz fiber filters (72 in total) used were first pretreated at 200°C and conditioned at constant temperature and humidity before and after sampling.They were then weighed at least three times (in 24 h) to obtain constant values.Road dust concentrations were determined gravimetrically.Subsequently, all road dust samples were analyzed for major elements (Al, Ca, K, S, Mg, Fe, Na) and 46 trace elements by inductively coupled plasma atomic emission and mass spectrometry respectively following the procedure described by Querol et al. (2001).A section of 1.5 cm 2 of the filter was used for the determination of organic (OC) and elemental carbon (EC) by an ECOC analyzer (Sunset Laboratory thermal optical transmittance method) using the EUSAAR-2 temperature protocol.The road dust mass concentrations and the concentrations of the chemical components were calculated as the arithmetic mean of the two filters collected in each site.The standard deviation of the duplicate measurements was less than 15%.

Source Apportionment
The investigation of sources responsible for road dust build-up was carried out by means of a Positive Matrix Factorization (PMF, Paatero and Tapper, 1994).PMF is a least squares factor analysis based on mass conservation principle to assist in identification of sources and their contributions to observed pollutant loadings.In Eq. ( 1), x ij is the jth species loading measured in the ith sample, g ik is the contribution of the kth source to the ith sample, f kj is the loading of the jth species in the kth source and e ij is the residual associated with the jth species loading measured in the ith sample.The standard PMF model was extended by means of the Multilinear Engine programming language (ME-2; Paatero, 1999) in order to add specific physical constraints (e.g., the sum of a single source profile cannot exceed the unity).The loading of road dust per square meter was introduced as independent variable (mg/m 2 ), while loadings of road dust components (mg/m 2 ) were the dependent variables.Chemical species for PMF analysis were selected based on the signal to noise ratio (Paatero and Hopke, 2003), the percentage of values above detection limit and to the database size requirements.Individual uncertainties of data were estimated following the approach of Amato et al. (2011).Factors are classified taking into account several diagnostic criteria: i) PMF-resolved chemical profile; ii) the variation explained for every species by each factor (specific tracers can be identified); iii) the spatial evolution of contributions.

Loadings of Inhalable Road Dust
The inhalable fraction, PM 10 of the road dust in Madrid urban area comprises small variation, between 1.35-4.40mg/m 2 , Table 1.These mass concentrations are in the range of variability of other studies (Amato et al., 2011(Amato et al., , 2012a) ) in Spain, Switzerland and Netherlands.Comparing the two sites, small differences are observed on the load of mobile road dust.The average concentration in the Reference site was 3.0 ± 0.8 mg/m 2 while in the Pelayo site was somewhat lower, equal to 2.4 ± 0.7 mg/m 2 .The mass difference between the two sites ranged from 0.05 to 1.75 mg/m 2 .This difference in the road dust load could be attributed to several factors such as street washing activities, Table 1.Average mass concentrations of inhalable road dust in mg/m 2 for the Reference and the Pelayo site during the sampling period.The mass difference between the two sites is also given.As the road surface in the Reference site was washed on 8 th of July and then was left untreated one would expect that the deposited particles would accumulate and their mass concentration would increase until the equilibrium state between deposition and resuspension is reached.However, at both sites, we did not observe an increase of the mobile road dust load.This may suggest that the nightly street washing did not provoke a significant decrease in road dust particle mobility measured at 9 am of the following morning.However, we did observe that the difference between the treated and untreated site is higher during the morning sampling, but again this difference is not statistically significant (t-test p-value = 0.95).

Date
The explanation for this could be found in the duration of the effect of street washing.Previous studies have shown that the beneficial effect of sweeping/washing on ambient PM 10 concentrations is short-lived lasting no more than 2-3 h (Chou et al., 2007).Specifically for the studied area we have found that nightly street washing has a beneficial effect on ambient PM 10 concentrations that lasts until 08:00 h, while later on, the prevailing atmospheric conditions and particulate emissions are the main processes that control the ambient PM concentration levels (Karanasiou et al., 2011).It is possible that this beneficial effect has the same duration for the particles deposited on the street surface.Indeed, Amato et al. (2012) showed that in Barcelona the effect of precipitation on the road dust particle mobility is short-lived, with the mobility recovered by 50%, 75%, and 90% within 8, 16 and 27 hours after the rain, respectively.This suggests that performing the measurement 9 hours (or 12 h for the first sampling on 9 th of July) after the washing does not permit to observe a significant (with respect to measurement error) decrease in the road dust load.As the first sampling took place 12 hours after the last street washing in the Reference site it is possible that the equilibrium between mobilization/deposition and moistening/ resuspension processes was already reached.A different street washing protocol (early in the morning) and road dust sampling just before the morning rush hour would allow us to observe larger differences and higher variability between the sampling sites.

Influence of Street Washing on the Chemical Composition of inhalable Road Dust
Table 2 provides the average concentrations of chemical components in inhalable road dust particles collected from both sites which were above their respective limit of detection (LOD) in at least 80% of the samples.The highest mass concentrations are observed for Ca (18 ± 9%), OC (14 ± 8%), Fe (10 ± 3%), S (7 ± 5%) and Al 2 O 3 (7 ± 5%).
To investigate the influence of street washing activities in the load of inhalable road dust is important to express the chemical composition data in terms of µg/m 2 in order to account for the load of each component on the road surface.Thus, it is possible to quantify the micrograms of road dust components at the road surface that were available (mobile) for resuspension.The difference in mass concentrations of elements in the road dust from the two sampling sites could be an indicator of street washing effect.It is possible that different road dust components (or sources) are more depleted by street washing than others: different solubility, hygroscopicity and size distribution of particles play a role in determining the recovery of dust mobility.Street washing activities carried out in the Pelayo site could reduce previously deposited particles like brake, tire and road wear material resulting in road dust samples less enriched in these species.
Table 3 shows the levels of major and trace elements per square meter for the not washed Reference site and for the washed Pelayo site.Comparing the two sites, a mass difference is observed for the OC content, being significantly lower in the Pelayo site around 50% of the average concentration levels.Other major components like Al, Ca, Na and S and trace elements had lower concentration in the Pelayo site but this mass difference was not statistical significant as it was lower of the relative standard deviation of the measurements (t-test p-value > 0.05).For the two metals Fe and Mg only marginal differences were observed in their mass concentration.On the other hand, EC had higher average concentration in the Pelayo site but again this was comparable to the variability of its concentration values.

Source Apportionment
For the source apportionment analysis the samples from the two sites were combined assuming that the emission sources are the same since the distance between the sampling sites is about 1.5 km.The resolved source profiles (Fig. 3) are similar with those resolved by Amato et al. (2011) in three European cities.The first factor is attributed to mineral dust and road abrasion since it consists of crustal species like Al, Ca, Fe, K related to the soil and road pavement.A brake wear source is easily identified by the high loadings of break wear tracers such as EC, Fe, Ba and Cu.This profile reflects the composition of brake linings commonly used in Spain that contain graphite and metallic iron as main components (Amato et al., 2012b).The third source is attributed to tire wear as it comprises high contribution of Zn a typical tire wear tracer and OC from tire rubber (Weckwerth, 2001;Adachi and Tainosho, 2004;Shauer et al., 2006).The low contribution of EC in this source profile has been previously reported by Schauer et al. (2006) where the ratio OC/EC in tire wear was found equal to 140.
A carbonaceous emission source is also resolved whose signature is the high abundance of OC, followed by Al, Cu, Ni and S, explaining most of the variability of Ni and Cu.A vehicle exhaust factor was not resolved since EC a typical tracer for diesel emissions appears in other sources, like mineral dust and break wear indicating that diesel emissions are probably mixed with these two sources.The low variability of pollutants during the campaign did not allow a clear separation of the road dust sources, therefore source factors and contributions have to be taken cautiously.
The factor containing high concentrations of Ca and S suggests the presence of construction works in the area where cement and gypsum was being mixed in the vicinity of the sampling sites.The average source contributions of each source to the mobile road dust loadings are shown in Fig. 4. Mineral dust and construction dust had the highest contributions, 34% and 25% respectively.Carbonaceous emissions contribute 19% to the mass concentrations while tire wear and brake wear have similar contributions of 10% and 12% respectively.To evaluate the effectiveness of street washing on the emission source, we compare the sources mass contributions between the two sites, Table 4. Two sources namely brake and tire wear have lower concentrations (approximately 40%) in the Pelayo site while mineral dust and carbonaceous emissions have higher concentrations.However, these differences as in the case of road dust mass are not statistically significant.The observed mass differences between the two sites lie in the range of the concentrations standard deviation.When the time series of the source contributions between the two sites are compared a clear trend cannot be seen, Fig. 5.The sources contributions do not increase from day to day, or between morning and afternoon sampling.Only in the case of construction dust a clear increase is observed from 11 to 17 July on both sites as a result of the construction activities in the proximity of both sampling sites.For the other sources the relatively small mass differences that are observed between the two sites do not allow us to draw definitive conclusions about the effectiveness of street washing in the specific road dust sources.

CONCLUSIONS
The principal aim of the present study was to examine the impact of street washing on the mobility of road dust particles and on the road dust emission sources.No increase was observed in the mobile road dust load from day to day.The explanation for this could be found in the duration of the effect of street washing.Previous studies have shown that the beneficial effect of sweeping/washing on ambient PM 10 concentrations is short-lived.It is possible that this beneficial effect has the same duration in the particles on the street surface.As the first sampling took place 12 hours after the last street washing in the Reference site it is possible that the equilibrium between mobilization/build-up and moistening/resuspension processes was already reached.Thus, low variability was observed between the two sites and between morning and evening sampling.This observation reveals the need to perform street washing during the early morning hours just before traffic peaks to make the most of its positive effect.The source apportionment results show that it is worth paying attention to the dust dispersed from the construction sites as they clearly affect the emission levels in nearby streets.

Fig. 1 .
Fig. 1.Madrid urban area with the sampling sites marked.

Fig. 2 .
Fig. 2. Street cleaning vehicle (Iveco Stralis) used in the urban area of Madrid.

Fig. 3 .
Fig. 3.Chemical profiles of the road dust emission sources resolved by PMF.

Fig. 4 .
Fig. 4. Average mass contributions of road dust sources determined by PMF.

Fig. 5 .
Fig. 5. Mass contribution of the emission sources for the Reference and the Pelayo site.

Table 2 .
Average mass concentrations and standard deviations (sd) of chemical components in inhalable fraction of road dust in the urban area of Madrid (mg/m 2 for PM 10 , weight percent by mass, % for major components, µg/g for trace elements).

Table 3 .
Mean load of each chemical component at Reference and Pelayo sites, (major components in µg/m 2 and trace elements in ng/m 2 ).

Table 4 .
Average mass contributions (in mg/m 2 ) of the road dust sources between the two sites.