Quantifying particulate matter reduction and their deposition on the leaves of green infrastructure

7 The green infrastructure (GI) is identified as a pa ssive exposure control measure of air 8 pollution. This work examines particulate matter (P M) reduction by a roadside hedge and its 9 deposition on leaves. The objectives of this study are to (i) quantify the relative difference in 10 PM concentration in the presence of GI and at an ad jacent clear area; (ii) estimate the total 11 mass and number density of PM deposited on leaves o f a hedge; (iii) ascertain variations in 12 PM deposition at adult (1.5m) and child (0.6 m) bre athing levels on either side of a hedge; 13 (iv) illustrate the relationship between PM deposi ti n to leaves and ambient PM 14 concentration reductions; and (v) quantify the elem ental composition of collected particles of 15 the leaves on different heights and sides of hedge. PM reduction of 2-9% was observed 16 behind hedge compared to a clear area and followed a trend of ΔPM1 >ΔPM10 >ΔPM2.5. 17 Counting of particles was found to be an effective m thod to quantify deposition than 18 weighting methods. Sub-micron particles (PM 1) dominated particle deposition on leaves at all 19 sampling points on both sides of the hedge. PM mass deposition and number concentration to 20 the leaves on traffic-facing side was up to 36% and 58% higher at 0.6m compared with 1.5m 21 Corresponding author. Address as above. Email: p.ku mar@surrey.ac.uk; Prashant.Kumar@cantab.net (P. Kumar) height, respectively. Such a difference was absent on the backside of the hedge. The SEM22 EDS analysis showed up to 12% higher traffic-origin ated particles deposited to leaves on the 23 traffic-facing side compared to the backside. The n aturally occurring particles dominated in 24 identified particles on leaf samples from all colle ction points on the hedge. These new 25 evidence expand our understanding of PM reduction o f GI in the near-road environment and 26 its variations in particle deposition, depending on height and sides of GI, which could allow a 27 better parameterisation of dispersion-deposition mo dels for GI assessment at micro-scale. 28


Introduction 31
Air pollution is a major cause of premature death in Europe, responsible for more than 32 400,000 premature deaths (EEA, 2019; HEI, 2018). In the UK, traffic-generated pollution has 33 been identified as a major contributor to particulate matter (DEFRA, 2017). Considerable 34 research findings have indicated a variety of adverse health impacts for those who spend 35 significant amounts of time near major roads (HEI, 2010). Such impacts are attributed to 36 relatively high concentrations of air pollutants within a few hundred metres of the road 37 influence of weather, the sampling day was sunny, dry, and without high wind speeds

Quantification of PM deposition 169
Samples for SEM imaging and elemental analysis were prepared from archived 170 leaves, following preparation methods described by Ottelé  energy dispersive X-ray spectrometer, using Backscattered Electrons (BSE) at 15 kV, in high 178 vacuum and with carbon coating (SI Figure S2). Four randomly selected points on each leaf 179 sample were micrographed in series, with a constant centre of view at two different 180 magnifications: 500× and 1200×, targeting coarse (2.5-10µm) and fine (1-2.5µm, and 0.27-181 1µm) PM size ranges, respectively. All micrographs were taken at the same working 182 distances, and with consistent contrast and brightness levels, to ensure uniform image 183 procuration procedures for all samples. 184 Pathfinder 2.0 X-ray Microanalysis Software (Thermo Fisher) was employed to quantify the 185 PM number density for different fractions (PM 10-2.5 , PM 2.5-1 , PM 1-0.27 ) on each micrograph, by 186 using binary BSE micrographs and counting the pixels composing each particle. The software 187 calculated the area and circularity of each identified particle, visualised as white against the 188 black leaf surface. The software was initialised with a setting to identify adjoining particles enabled the software to capture particles effectively and subtract leaf surfaces. PMi is PM (PM 10-2.5 , PM 2.5 or PM 1 ) deposition (μg cm -2 ). LAI is the total leaf surface area per 205 unit ground area (m 2 m -2 ). The density of particles was taken as 2 g cm -3 (Lin et al., 2018; 206 Zhang and He, 2014). 207

2.5
Comparisons of deposition and total pollution reduction by the hedge 208 Ambient measurements provided instantaneous concentration change data for both 209 scenarios (i.e. in the absence and presence of the hedge). This time series was converted to a 210 daily average for comparison between the two monitoring locations as well as with leaf 211 deposition. Total PM deposition, quantified by image analysis, was applied as part of a ' ( = $%& ) * + , $% ) * + ! 2 Where LAI is leaf area index, LAD is leaf area density, ) * is deposition velocity, C is average 214 PM concentration at the hedge site, and t is the time available for particle deposition to leaves 215 (12 days). The relationship between PM deposition and ambient PM concentration was 216 assessed by comparing time series plots of average ambient PM concentration (with and 217 without the presence of the hedge, from field measurements) with average PM deposition 218 according to equation (3). 219 3 220 Where LAD is leaf area density, C is the ambient concentration of air pollutants, and ) * is 221 deposition velocity, calculated from leaf deposition quantification as described by equation 222 (2). 223

2.6
Elemental analysis of particles collected on leaves 224 The elemental composition of bulk particles collected on leaf samples was determined 225 via SEM-EDS analysis. Particles deposited on leaf samples of the hedge at a breathing height 226 of children (0.6m) and adults (1.5-1.7m) were collected for characterising their chemical 227 composition. Leaf samples were prepared for the quantification of PM elemental 228 composition using a Scanning Electron Microscope (JEOL SEM, model JSM-7100F, Japan) 229 equipped with an energy dispersive X-ray spectrometer at the Micro-Structural Studies Unit 230 of the University of Surrey, UK. The SEM has a spatial resolution of 1.2nm at 30 kV and 231 3.0nm at 1 kV. The SEM was employed at an acceleration voltage of 10kv, with a working 232 distance of 10mm under vacuum conditions. Automated EDS analyses were performed on 233 contrasting Backscattered Electron (BSE) images with bright white collected particles against 234 the black background of the filter paper using Pathfinder 2.0 software (Thermo-Fisher), 235 following procedures described by Abhijith and Kumar (2019). A total of 2 micrographs were

3.
Results and discussion 239

Differences in PM concentrations between GI scenarios and adjacent clear areas 240
A summary of PM concentration data is provided in terms of descriptive statistics 241 with percentage differences of the mean (SI Table S2) and as box plots ( locations are presented in Figure 5, and summary statistics of PM densities are provided in SI 283 Table S3. The fine (PM 1 ) size fraction accounted for the highest PM density across all leaf 284 sampling locations and it was approximately 10-times higher than the lowest PM density PM 2.5-1 > PM 10-2.5 , accounting for 66%, 29% and 5% of total deposited particles, respectively, 288 as shown in SI Figure S3b  surfaces was less significant than on adaxial surfaces across leaf sampling locations. As seen 307 in SI Figure S3a, 18-25% of all particles were accumulated on abaxial surfaces, which is 308 consistent with reported ranges of 24% and 17% by Shi et al. (2017b) and Wang et al. (2006), at the back of hedge than on the traffic-facing front side of the hedge, and adaxial leaf from higher traffic-induced turbulence along the traffic-facing side, leading to less 313 deposition. Conversely, the relatively less turbulent conditions at the backside of the hedge 314 may have allowed increased PM deposition to leaves. Moreover, visual inspection of hedge 315 during the field campaign showed smaller and less healthy leaves at traffic facing side 316 compared to large and heathier leaves on the back of the hedge, as showed in SI Figure S4. 317 Exposure to fresh traffic fumes may be the reason for visual differences in the health of 318 leaves from both sides. This could presumably affect the PM retention on leaves. Further 319 studies investigating air pollution tolerance index and air pollution stress on GI could provide 320 their impact on air pollution reduction and PM removal by GI. These results also suggest that 321 it would be beneficial to plant high air pollution tolerant variants close to traffic facing side 322 while designing GI barrier with multiple rows of various species. 323 The average mass of deposited particles per unit area of the hedge (across sampling points) 324 was estimated by incorporating the mean aerodynamic diameter of particles (SI Table S4), 325 mean PM density (SI Table S3) and LAI of the hedge, and was tabulated in SI Table S5. The 326 mean deposited mass followed an opposite trend to that of PM density (PM 1 < PM 2.5 < PM 2-5- . The highest mass deposited was of the coarse size range (PM 2-5-10 ) and the least was 329 fine (PM 1 ), accounting for 73% and 3% of mean particle deposition to the hedge, respectively 330 (SI Figure S3c). The deposited amount of PM 2.5-10 , PM 1-2.5 and PM 1 ranged from 13-17 µg 331 cm -2 , 4-5 µg cm -2 and 0.5-0.6 µg cm -2 , respectively ( Figure 6). Our finding of 20±2 µg cm -2 332 mean PM deposition to the Fagus sylvatica hedge is generally consistent with the ~17±3 µg 333 cm -2 described by Saebø et al. (2012), who collected leaves during the same seasonal period at 1.5m height on the traffic-facing front side of the hedge reported least than leaves from remaining sampling locations as observed in Figure 6. Coarse (PM 10-2.5 ) particle deposition 337 on the traffic-side was lower than that on the backside of the hedge (Figure 6). The highest 338 PM deposition was observed at 0.6m height on the front, traffic-facing side of the hedge, and 339 mean PM 1 deposition to leaves displayed less variation between sampling points (Figure 6). 340 Table 1 provides further insight into the impacts of different sampling heights and sides of 341 the hedge. Significant differences in PM deposition and, particularly, PM density were 342 observed between leaf sampling heights for the traffic-facing front side of the hedge, whereas 343 no such trend was observed for the backside of the hedge (Table 1). Leaves closer to the 344 ground (0.6m) were closer to the traffic emission source and subject to less traffic-generated 345 turbulence than leaves at 1.5m, resulting in greater PM deposition. In addition, the barrier 346 effect of the hedge may lead to higher concentrations at ground level than at the upper canopy 347 of the hedge (2.2m), further contributing to increased PM deposition at the F 0.6 sampling 348 point (Table 1) Figure S5). These higher rates of a deposition provided only data during entire traffic peak hours (07:00-10:00h and 16:00-19:00h at this site). Higher 388 concentrations and lower dispersion during morning peak hours may have resulted in higher 389 PM deposition and thus led to a larger difference between INF and BHD monitoring points at 390 GI compared to clear-area. As with the relative change (discussed in Sections 3.1), the 391 absolute difference between INF and BHD monitoring points at the GI site was consistently 392 higher than that of the clear area, positively indicating a combination of dispersion and 393 deposition in the reduction of PM by GI (SI Figure S5). 394

Elemental analysis of deposited particles 395
Elemental analysis was carried out on a total of 2741, 2731, 2471, and 2674 particles 396 deposited on leaves collected from sampling points F 1.5m , F 0.6m , B 1.5m , and B 0.6m , respectively.  Table S6. 412 At all four sampling points, most (>64%) of the deposited particles were found to belong to 413 the 'natural' classification. This fraction was as high as 80% (+15% difference) for leaves 414 sampled from the backside the of the hedge (B 1.5m and B 0.6m) rather than the traffic-facing 415 front side (F 1.5m and F 0.6m ). The comparison of sampling heights on both sides of the hedge 416 showed a negligible difference in the proportion of 'natural' particles. Particles of the 417 'vehicle' and 'agglomerates' classifications were -8% to -12% and -5% to -8%, 418 respectively, lower at the backside of the hedge than at the traffic-facing side. This may 419 indicate fewer harmful particles in the ambient air behind the hedge, adding weight to the 420 findings of Abhijith and Kumar (2019). They also reported higher quantities of 'natural' 421 particles in the ambient air behind the hedge than in front, resulting in higher percentages of 422 'natural' particle deposition on the backside of the hedge. While comparing sampling heights, 423 the percentages of the agglomerates and vehicle portions were closer in the backside of hedge 424 .2.5 > PM 2-5-10, comprising 66%, 29% and 5% of total deposited particles, respectively. In 436 addition, adaxial leaf surfaces captured up to three times more particles than abaxial leaf 437 surfaces. The mean deposited PM mass on leaves followed an opposite trend to that of PM 438 density; i.e. PM 1 <PM 2.5 <PM 2.5-10 . Interestingly, leaf sampling height significantly influenced 439 PM deposition. Significant variation in PM density was observed on leaf samples from the