Coarse-Particle Passive-Sampler Measurements and Single-Particle Analysis by Transmitted Light Microscopy at Highly Frequented Motorways

20 21 Measuring and characterizing airborne particulate matter (PM) is an important research 22 area because PM can lead to impacts on health and to visibility reduction, material damage 23 and groundwater pollution. In regard to road dust, suspension and re-suspension and the 24 contribution of non-exhaust PM to total traffic emissions are expected to increase as a result 25 of predicted climate scenarios. European environmental regulations have been enforced to 26 reduce exhaust particle emissions from road traffic, but little attention has been paid to 27 reducing non-exhaust coarse particle emissions due to traffic. Therefore, a monitoring 28 program for coarse PM has been initiated in early 2013 to assess the predicted increase in the 29 abundance of non-exhaust particles. Particle particles. The field measurements show that the minimum PM concentration was found in the 39 5 to 12 °C temperature range, whereas the maximum concentration was observed in both the 40 -5 to 5 °C and the 12 to 24 °C ranges, in agreement with previous laboratory measurements. 41 Correlation between super-coarse (d p 10–80 µm, geometric equivalent diameter) PM 42 concentration and precipitation displays a significant increase in concentration with 43 decreasing number of precipitation events (dry weather periods). 44

Measuring and characterizing airborne particulate matter (PM) is an important research 22 area because PM can lead to impacts on health and to visibility reduction, material damage 23 and groundwater pollution. In regard to road dust, suspension and re-suspension and the 24 contribution of non-exhaust PM to total traffic emissions are expected to increase as a result 25 of predicted climate scenarios. European environmental regulations have been enforced to 26 reduce exhaust particle emissions from road traffic, but little attention has been paid to 27 reducing non-exhaust coarse particle emissions due to traffic. Therefore, a monitoring 28 program for coarse PM has been initiated in early 2013 to assess the predicted increase in the 29 abundance of non-exhaust particles. Particle sampling was performed with the passive-30 sampler technique Sigma-2. The subsequent single-particle analysis allows for 31 characterization of individual particles, determination of PM size distribution, and calculation 32 of PM mass concentrations. Two motorways near Cologne (Köln), Germany were selected as 33 sampling sites, and the experimental setup in the field was realized with a so-called twin-site 34 method. The present study reports single-particle analysis data for samples collected between 35 May 31, 2013 and May 30, 2014. Coarse PM, generated through multi-source mechanisms, 36 consists of, e.g., tire-wear, soot aggregates, and mineral dust. The highest mass concentration 37 occurs at both motorways in spring, and the observed PM mainly contains traffic-abrasion 38

50
Particulate matter (PM) in the ambient air is a key environmental problem in most cities 51 around the world (WHO 2005). Policy makers succeeded in reducing exhaust emissions, but 52 did not and still do not address "non-exhaust" traffic emissions from brake wear, tire wear, 53 road wear, and suspension in air of road dust. Road-transport emissions, which are associated 54 with adverse health effects, will become an even more important PM source in the future, 55 especially in urban and congested areas (Denier van der Gon et al., 2013). 56 The literature on the PM 10 particle fraction mainly contains examinations under short-57 term laboratory conditions, and therefore, our objective was to thoroughly examine road-58 transport emissions in a one-year comprehensive field study, the first measurement campaign 59 of this kind. With a continuous year of measurements, all seasons are included. Moreover, 60 our study is the first to include the particle fraction between 10 and 80 µm, the super-coarse 61 fraction, in the single-particle analysis, in addition to the more commonly analyzed coarse 62 particle size range of 2.5-10 µm. Furthermore, meteorological data (precipitation, 63 temperature) and weather classification of the examination period were combined. Two 64 motorways with different traffic flow, traffic load and construction (open vs. bordered by 65 embankments) were chosen. This research, thus, represents a novel approach to coarse PM 66 measurements and study of road-transport emissions at highly frequented motorways. 67 Particulate emissions from traffic do not solely comprise combustion-derived exhaust 68 emissions (Pierson and Brachaczek, 1982). They also contain particles that are collectively 69 known as non-exhaust PM, which is produced through abrasion of brakes, tires, clutches and 70 roadway, as well as mixed-origin dust particles, both suspended and re-suspended from the 71 road surface ( For particle collection, we used the cost-effective and easy-to-handle passive-sampler 116 device Sigma-2 at ground-based sampling sites. This technology ensures a wind-sheltered, 117 low-turbulence air volume inside the sampler. The design of the Sigma-2 device allows for 118 protection of the particles from direct radiation, wind and precipitation (Dietze et al., 2006; 119 VDI 2119, 2013). Particles are collected on a transparent adhesive collection plate, which is 120 A C C E P T E D M A N U S C R I P T exposed for 7 days. The collection plate is designed for subsequent optical single-particle 121 analysis by Transmitted Light Microscopy (TLM) and is also suitable for single-particle 122 analysis via Scanning Electron Microscopy (SEM) combined with Energy-Dispersive X-ray 123 (EDX) spectroscopy (see Sommer et al., 2016). For our study, we selected two different 124 motorways, A 555 and A 4, in North Rhine Westphalia, Germany (Fig.1). The sampling 125 period discussed in this paper covers the period from May 31, 2013 until May 30, 2014 at 126 both motorways, corresponding to a total of 51 weeks of specimen collection (over the 127 Christmas holidays, the sampling duration was two weeks). 128 The A 555 sampling site, located approximately 17 km south of Cologne and 129 approximately two km to the west of the Rhine river ( Fig. 1, symbol A), represents a highly 130 frequented, north-south-directed motorway. The motorway is surrounded by farm fields 131 where the terrain is exposed to a free, largely undisturbed air flow (Fig. 2a). Two passive-132 sampling stations were set up in twin-site mode, i.e. located on either side of the motorway, at 133 about 1.5 m height and 4.6 m horizontal distance from the roadway. 134 One sampling station is situated on the east side (lee side) of the motorway, with 135 driving direction to the north, whereas the second station is on the west side (wind side), with 136 driving direction to the south. Table 1 shows the average traffic counts per day for all  137 vehicles, including HDV, and separately for HDV on the A 555 motorway in the year 2013. 138 The Gremberg, about 6 km east of Cologne and between intersections with exit ramps (Fig. 1,  143 symbol B; see also Fig. 2b). During weekdays, this motorway is characterized by slow-144 A C C E P T E D M A N U S C R I P T 7 moving traffic, with rush hours that regularly cause "stop-and-go" traffic conditions. 145 Engineered road embankments are present along both sides of the motorway, which lead to a 146 disturbed and blocked air flow (Fig. 2b). The traffic lanes are located at an altitude that is 147 approximately five meters lower than the surrounding area, which causes a street-canyon 148 effect. Two sampling sites, also within this street canyon, were set up in twin-site mode, one 149 at the south side, the other at the north side of the motorway, with east-and west-driving 150 directions, respectively. The single-particle analysis was carried out using a computer-controlled light 159 microscope equipped with an adapted automatic scanning stage and a high-resolution CCD 160 digital camera for imaging. We applied automated TLM analysis as a cost-effective, fast and 161 easy to handle method for particle measurement, characterization and differentiation by 162 particle type (opaque vs. transparent). This method provides first information on the origin 163 and composition as well as possible health significance of the particle load. In addition to these repeat determinations, which demonstrate excellent agreement for 182 both size fractions (Fig. 3), the passive-sampler methodology has also been compared to the 183 performance of appropriate active samplers. For such a comparison, the particle size range 184 2.5-10 µm is the most practicable size interval within operational applications of active 185 sampler devices. Therefore, two separate active samplers, one for PM 10 and one for PM 2.5 , 186 have been used to collect data for the calculation of the difference in mass concentration 187 (PM 10 minus PM 2.5 ) in order to compare with the measurement from the passive-sampler for 188 particles in the size range d p 2.5-10 µm (PM 2.5-10 ). The results show that the average PM 2.5-10 189 concentration from the active and passive measurements is 5.1 ± 1.3 µg m -3 and 6.2 ± 2.3 µg 190 m -3 , respectively. The temporal variation in active vs. passive data for a 40-week validation 191 A C C E P T E D M A N U S C R I P T measurement is shown in Figure 4 (no passive-sampler data available for calendar weeks 4 192 and 8 due to contamination of the acceptor plate; Dietze et. al., 2006). 193 In contrast to an active-sampler, the passive-sampler technique allows for deposition of 194 particles in such a way that they are not agglomerated on the sampling medium, which 195 guarantees that TLM analysis can return easily and quickly the necessary properties of 196 individual particles. Therefore, apart from the total particle load and size distribution, the 197 TLM analysis also allows for the differentiation by particle type (opaque vs. transparent), 198 which permits distinction between, and first estimation of, the proportions of anthropogenic 199 and natural particles. The "transparent" particles are mainly of natural origin (e.g., mineral 200 dust, pollen and spores), whereas "opaque" particles are tyically carbon-containing (e.g., 201 coarse combustion-derived particulates) or abrasion particles of anthropogenic origin (e.g., 202 tire wear, metal fragments from brake abrasion). This opacity-based distinction between 203 natural and anthropogenic particles is not clear-cut in all cases, as illustrated in Table 2. In 204 our study, the "transparent" particle fraction mainly represents road wear (dust from the 205 erosion of the roadway aggregate or cement) and specifically excludes water-soluble particles, 206 such as, sulfates (e.g., derived from combustion processes) and halides (e.g., road salt in 207 winter), because of the chosen aqueous immersion medium for optical microscopy. Pollen 208 were stained with Safranin, thus acquiring a typical red color, and were then eliminated for 209 this study by the automated image-processing system. 210 Opaque particles with d p > 10 µm and with a typical cylindrical, roundish or kidney-211 shaped outline ( 4 is remarkable: In contrast to the typical "Gaussian" distribution found at the A 555, we 254 observe a pronounced "shoulder" at the upper right side of the size distribution (d p 20-40 255 µm). In this size interval, the opaque particle fraction is typically dominated by tire abrasion 256 (Fig 5). This can be explained with the stop-and-go traffic mode (low-speed driving 257 conditions) on the A 4 during the daily rush hours. According to Dannis (1974), both particle 258 diameter and particle quantity correlate with the driving speed, whereby the mean particle 259 diameter increases with decreasing velocity. influence on dispersion and transport of PM, especially for the non-exhaust coarse particle 267 fraction, which is mostly generated via abrasion processes. Atmospheric residence time 268 generally decreases with increasing particle size (Jaenicke, 1978). Precipitation events, i.e., 269 the number of days with precipitation in any given particle-sampling period, are important for 270 preventing PM suspension and re-suspension from the road surface. In contrast, dry weather 271 periods enhance suspension and re-suspension. The data obtained during our study allow for 272 quantification of these generally valid statements: Figures 7a and 7b display a significant 273 decrease in total mass concentration of the particle fraction d p 10-80 µm with increasing 274 number of precipitation events (from 0 to 7 days with precipitation per week). With 275 increasing precipitation days per week, the mass concentration was reduced by ~ 66% and ~ 276 60% at the A 555 and the A 4 sites, respectively. This strong correlation (coefficient of 277 determination: R 2 ~ 0.96) confirms that precipitation has a highly significant, quantifyable 278 influence on the super-coarse particle fraction. 279 The influence of precipitation on the PM size fraction d p 2.5-10 µm is relatively low at 280 the A 555 sampling site (Fig. 7c), and not determinable at the A 4 motorway (Fig. 7d). These 281 quantitative results demonstrate that the correlation between precipitation and PM 282 concentration is much weaker for coarse particles than for super-coarse particles (Fig. 7a, 7b). For our study, the weekly average air temperature data from Cologne Airport were 305 chosen to assess the influence of ambient air temperature on the particle load. Three 306 temperature classes have been selected (-5 to 5 °C, 5 to 12 °C, and 12 to 24 °C), and the total 307 particle concentration was measured in the 2.5-10 µm size interval. This size interval was 308 selected because it closely resembles the one measured in the above-mentioned PM 10 309 laboratory studies. The results from the A 555 motorway (Fig. 8b) show that mass 310 concentration decreases from 12.0 µg m -3 to 9.5 µg m -3 with increasing air temperature in the 311 -5 to 12 °C temperature classes, but that it then increases again to 11.4 µg m -3 between 12 and 312 24 °C. Even though the mass concentrations are identical within error, the trend is in 313 surprisingly good agreement with the laboratory measurements described above and shown in 314 Fig. 8a, confirming that the lowest particle concentration appears in the middle temperature 315 range. The A 4 sampling site shows that particle concentration decreases from 17.6 µg m -3 in 316 the -5 to 5 °C class with increasing air temperature to 13.3 µg m -3 in the 5 to 12 °C 317 temperature class (Fig. 8b), which also fits the general trend observed in the indoor road 318 simulator studies, but between 12 and 24 °C, the particle concentration remains constant at 319 13.3 µg m -3 . This result is consistent with the findings of the German Federal Highway 320 Institute, whose laboratory measurements have demonstrated that at an ambient air 321 temperature of 20 °C, low driving speeds, such as those observed in the stop-and-go traffic 322 conditions at the A 4 motorway (i.e., 35-50 km h -1 , temperatures around 20 °C tends in 323 combination with low driving speeds (30-35 km h -1 , e.g. within stop-and-go traffic) lead to 324 reduced particle emissions compared to higher driving speeds (100 km h -1 ) at the same 325 temperature (Fig. 8a). 326 327 weather type (Fig. 9). 336 Figure 9a shows the immission wind rose for the A 555 motorway (driving direction to 337 the south), with the sampling site west of the motorway, and Figure 9b  Similarly, for the SW weather type, the particle mass concentration indicates only small 348 differences between the western and the eastern sampling sites. Typically, SW weather types 349 move marine air masses with moisture-laden clouds from the Atlantic into Germany, thus 350 causing more precipitation days than the E and H weather types. Therefore, mobilization and 351 local transport of PM are considerably reduced compared to the other weather types, as 352 reflected by mass concentrations of 9.6 µg m -3 (western sampling site) and 10.5 µg m -3 353 (eastern sampling site). For the H weather type, with no dominant wind direction, the two 354 sampling locations at the A 555 display, as expected, no major differences (~ 11.5 µg m -3 ). 355 Figure 9c shows the immission wind rose for the A 4 motorway (driving direction to 356 the east), with the sampling site south of the motorway, and Figure 9d for the A 4 motorway 357

A C C E P T E D M A N U S C R I P T
(driving direction to the west), with the sampling site north of the motorway. Similar to the A 358 555, the highest particle concentration at the A 4 occurs during periods of the E weather type, 359 but in this case the airflow is parallel to the motorway. The high-pressure weather type (H), 360 with no dominant wind direction, displays nearly identical concentrations for both sampling 361 sites in the same way we have already seen at the A 555. For the SW weather type, the 362 particle mass concentrations exhibit only small differences between the South and the North 363 sampling sites. 364 In the case of airflow directions perpendicular to the line source (cross-airflow), at the 365 A 4 motorway (west-east aligned) no major differences in PM concentrations are displayed. 366 It is of note that for the S weather type (cross-airflow condition) high PM mass concentration 367 was measured at both the north (16.3 µg m -3 ) and the south (17.7 µg m -3 ) sampling sites of 368 the motorway. This can be explained as due to the street-canyon effect, which is caused by 369 road embankments at either side of the motorway, with driving lanes approximately five 370 meters lower than the surrounding area (see Fig. 2b). In the event of an airflow perpendicular 371 to the motorway, the street-canyon effect leads to a disturbed airflow towards our sampling 372 sites at the A 4 and, additionally, to vortex formation in the skimming flow field as the 373 airstream is able to form recirculation patterns (Fig. 2b). In contrast, the north-south aligned 374 motorway A 555 is surrounded by open terrain, which allows for a largely undisturbed 375 airflow (Fig. 2a), thus resulting in a distinct PM concentration pattern for E and SW weather 376 types (cross-airflow conditions). In general, it can be stated that the traffic-related coarse 377 abrasion particles have a longer residence time in the air within the street canyon at the A 4, 378 which in addition, causes the higher PM concentration at this motorway, as also observed 379 during weather type H. 380

Temporal variation
The seasonal overview of our year-long study (Fig. 10)  In almost all weeks, the PM concentrations measured at the A 4 motorway are higher 388 than those at the A 555. This observation is consistent with higher traffic load, different type 389 and mode of traffic, and different motorway construction of the A4, as described above. 390 There is one exception, however: in summer 2013, during a five-week period, PM values 391 were higher at the A 555. The reason is unknown and, therefore, will not be further explored 392 here. 393 However, for both motorways the seasonal trend reveals the maximum PM values in 394 spring, followed by the summer, with autumn representing the transition to the minimum 395 values observed during winter (turn of the year). To assess a possible influence of climate 396 change (rise in air temperature associated with an increase in the number of heatwaves with 397 droughts), we will examine the summer and spring seasons more in detail. 398 With average maximum temperatures between 22 and 28 °C, summer 2013 represents a 399 very warm to hot period of the year, but the PM concentrations are lower than those of spring 400 2014. The reason for this unexpected finding is that the summer 2013 was dominated by wet 401 weather conditions where the highest temperatures coincide with high precipitation (Fig. 10). 402 Hence, the precipitation was overriding the temperature influence and thus kept the particle 403 load at the second highest level of the year. 404

A C C E P T E D M A N U S C R I P T
In contrast, the six-week spring period in 2014 (highlighted in green in Fig. 10) was 405 characterized by low precipitation. According to DWD Cologne airport weather records, this 406 period in spring was the driest during the last 30 years. With respect to temperature, it was 407 among the warmest three springs in Germany since the year 1881. With average maximum 408 temperatures between 12 and 22 °C, it fit in the temperature class above 12 °C, where both 409 the laboratory experiments and our field measurements (chapter 3.2.2) show an increase in 410 the coarse PM load (see Fig. 8). In summary, the extremely dry weather conditions in spring 411 2014, in combination with high temperatures, led to the highest PM concentrations of our 412 year-long measurement campaign. 413 Based on the temporal variation of the super-coarse PM concentrations shown in Figure  414 10, size distribution diagrams have been calculated for the six-week spring period of 2014. 415 The size distribution for A 555 (Fig. 11a) displays maximum PM concentrations for all 416 particle fractions (transparent, opaque, and total) in the 10-20 µm size interval. This curve 417 shape, similar to a "Gaussian" distribution, is typical for traffic-related coarse abrasion 418 particles (Councell et al., 2004). By contrast, the size distribution for A 4 (Fig. 11b) shows, 419 apart from higher PM concentrations, a significant shift of the maximum PM values to the 420 20-40 µm size interval. The continuous increase of PM concentration between 2.5 and 40 µm 421 for all particle fractions, i.e. a left-skewed size distribution, is an indicator for a traffic mode 422 and type that are different compared to those at the A 555. Apart from the highly congested 423 traffic situation at the A 4, PM concentration is further intensified by the street-canyon effect. 424 In this situation, the generated PM load from the road surface is enclosed between the 425 embankments of the A 4, where the particles are constantly dispersed by turbulences (e.g., 426 caused by HDV), and additionally, due to the lack of ventilation, the air exchange with the 427 environment is limited. 428

A C C E P T E D M A N U S C R I P T
The notable differences in the size distribution of opaque, transparent and total particles 429 between A 555 and A 4 during spring 2014 are also evident in the TLM images shown in 430 Figure 12. These example images demonstrate that for both fractions, transparent and opaque, 431 larger particles occur at the A 4 compared to the A 555. The TLM images further display that 432 most of the opaque particles (d p > 10 µm) feature the cylindrical shape typical for tire-433 abrasion particles, as reported by Baltensperger (1985) and Rauterberg-Wulff et al. (1995) 434 ( Fig. 5a and b). Our quantitative comparison of two distinct motorways allows for a better 453 understanding of the coarse particle distribution and its relationship with traffic load, type, 454 mode and local topography. Higher traffic load with a higher percentage of HDV, lower 455 driving speeds, and enclosing road embankments lead to significantly higher PM 456 concentrations as well as to a shift to larger particle sizes. 457 The overall conclusion for our one-year coarse PM study is that even with a lower 458 temporal resolution and with a lower sampling efficiency in comparison with active sampling 459 methods, the applied passive-sampler technique in combination with automated TLM 460 analysis provides relevant quantitative data for traffic-related coarse particles. 461

462
We will continue our coarse PM measurements at German motorways with TLM 463 particle analysis and, in the near future, expand the characterization of individual particles by 464 including scanning electron microscopy and energy-dispersive X-ray spectroscopy (SEM-465 EDS). This will allow for a more complete classification of the particles according to their 466 chemical composition and provide additional information about the complex traffic-related 467 PM mixture of airborne particles, especially in the size range from 2.5 to 10 µm.  A C C E P T E D M A N U S C R I P T