Exposure to Ultrafine Particles in Asphalt Work

ABSTRACT

An epidemiologic study has demonstrated that asphalt workers show increased loss of lung function and an increase of biomarkers of inflammation over the asphalt paving season. The aim of this study was to investigate which possible agent(s)causes the inflammatory reaction, with emphasis on ultrafine particles. The workers’ exposure to total dust, polycyclicaromatic hydrocarbons, and NO2 was determined by personal sampling. Exposure to ultrafine particles was measured by means of particle counters and scanning mobility particles izer mounted on a van following the paving machine. The fractions of organic and elemental carbon were determined.Asphalt paving workers were exposed to ultrafine particles with medium concentration of about 3.4×104/cm3. Ultrafine particles at the paving site originated mainly from asphalt paving activities and traffic exhaust; most seemed to originate from asphalt fumes. Oil mist exceeded occupational limits on some occasions. Diesel particulate matter was measured as elemental carbon, which was low, around 3μg/m3. NO2 and total dust did not exceed limits. Asphalt pavers were exposed to relatively high concentrations of ultrafine particles throughout their working day, with possible adverse health effects.

METHODS
Exposure Assessment
Sampling Strategy

Measurements were carried out between April and October 2005 and 2006 to assess exposure in asphalt production and asphalt paving. Measurements were carried out on one bitumen plant, three asphalt production plants, two asphalt stripper sites, and 14 asphalt paving sites. A total of 51 asphalt workers carried personal samplers in the exposure study, and most (90%) were monitored on more than one occasion. Participation was voluntary.

Exposure to total dust and gases was determined by means of personal sampling, and two or more agents were measured simultaneously for each person for at least 2 days. The sampling duration was 7–8 hr. A typical work shift was 8–10 hr.

Exposure to ultrafine particles was measured during asphalt paving on a four-lane motorway (E20, speed limit 110 kph) in a rural area in Sweden and, to some extent, also during asphalt production. The ultrafine particle monitors are heavy and need to be driven by electricity, so there was no possibility of doing personal sampling of ultrafine particles. However, by using a van it was possible to carry out mobile measurements of particles by following the paving machine at the same speed as the road was paved. The van was equipped with truck batteries to run the instruments. Paving was done in the right lane of the motorway. The research team drove the van on the hard shoulder of the road, while traffic ran at 70 kph in the left lane. 

The van, whenever possible, was driven beside the two screedmen. To get as close as possible to the screedmen, the van driver positioned the front of the van alongside the back of the paving machine, while the side of the van (where the air was sampled) was next to the screedmen. When there was a crash barrier beside the road and the space between the paving machine and the crash barrier was too narrow for the van, sampling was done a few meters in front of or behind the two
screedmen.
 
Air was sampled at mouth level. Stainless steel tubing brought air directly to the instruments inside the van. Particle size distribution was measured by using a Scanning Mobility Particle Sizer (SMPS, model 3071A; TSI GmbH, Aachen, Germany), the particle number by using a P-Trak Ultrafine Particle Counter (model 8525; TSI), and a Grimm Particle Dust Monitor (PDM, model 1109; Grimm Technologies, Inc., Douglasville, Ga.), and the mass of particles by using Dust- rak Aerosol Monitor (model 8520; TSI) during two working days. At the same time, measurement of particulate carbon as elemental and organic carbon was performed via aerosol sampling.

Ultrafine particles were also measured at an asphalt plant. Air was sampled at a position where portions of newly produced asphalt were emptied repeatedly into a carriage on rail for transportation to a large container. Particle measurements were performed for 1 hr using SMPS and P-Trak.

Sampling Methods and Analyses

Total dust and PAHs were collected on glass fiber filters (Whatman International Ltd., Maidstone, U.K.), fitted in 37-mm, closed-faced aerosol filter cassettes (Millipore Corporation, Bedford, Mass.). Gaseous PAHs were collected on tubes filled with the adsorbent XAD-2 (SKC, Blandford Forum, U.K.). Filters and tubes for collection of particulate and gaseous PAHs were mounted in series during sampling. The sampling flow rate was 2 L/min. The particle mass was measured by microbalance (model AT261; Mettler Toledo, Inc., Columbus, Ohio), with a detection limit of 0.031 mg/m3 based on 8 hr of sampling.

The filter and adsorbent were extracted by addition of dichloromethane mixed with internal standards (naphthalen- D8, phenanthren-D10, fluoranthen-D10, benzo(a)pyrene- D12). PAHs were measured by using gas chromatography mass spectrometry. The detection limits of PAHs were 0.005–0.010 μg/m3 depending on the specific component, based on 8 hr of sampling at a flow rate of 2 L/min. Total PAH were calculated as the sum of all measured components on filter and adsorbent. When the concentrations were less than the detection limit, these results were substituted with a concentration level of half the detection limit.

Oil mist was collected on glass fiber filters (Whatman) and oil vapor was collected on tubes containing XAD-2 (SKC). Filters and tubes for collection of oil mist and oil vapor were mounted in series during sampling. The sampling flow rate was 2 L/min. Oil mist was measured by using a Fourier transform infrared (FTIR) spectrophotometer (PE 1600 FTIR; PerkinElmer, Waltham, Mass.); oil vapor was measured by chromatography (CG) with a flame ionization detector (FID). The detection limit of oil mist and oil vapor was 10 μg/m3 and 5 μg/m3, respectively, assuming that the sampling volume is 1 m3.

Respirable dust was collected on 37-mm, cellulose acetate filters with a pore size of 0.8 μm using a cyclone separator (SKC) at a sampling flow rate of 2.2 L/min. The particle mass was measured gravimetrically (with a detection limit of 0.031 mg/m3 based on 8 hr of sampling). The alfa-quartz content in the respirable dust sample was measured by using the FT-IR PE-1600 spectrophotometer. The detection limit was 1.7 μg/m3 assuming that the sampling volume is 1 m3.

Amines (dietanolamine, diethylenetriamine, triehytlenetetramine) were collected on tubes filled with the adsorbent XAD-2. The sampling flow rate was 2 L/min. The adsorbents are extracted by addition of acetonitril. Amines were measured by using high-performance liquid chromatography with UV- detection. The detection limit was 0.0006 mg/m3 assuming that the sampling volume is 1 m3.

The carbon monoxide and nitrogen dioxide concentrations were measured with direct-reading electrochemical sensors with a data-logging facility built into the instrument (type PAC III; Drager Aktiengesellscaft, Lubeck, Germany). An averaging period of one reading every 2 min was selected. The detection limits of carbon monoxide and nitrogen dioxide measurements were 2 ppm and 0.2 ppm, respectively.

Particle size distribution was measured by using an SMPS (model 3071A, TSI). Particle sizes were scanned from 14–764 nm. A condensation particle counter (CPC, model 3022; TSI) was included in the system. The particle size as measured by the SMPS was calibrated by using monodisperse test aerosols, and the concentration of the CPC was calibrated by TSI prior to performing the measurements. One scan of the particle sizes took 5 min to complete. In outdoor conditions, the aerosol properties can vary quickly with time, especially if the wind direction is fluctuating. Therefore, the aerosol particle size distribution presented here is the mean distribution of several scans.

Total number of particles was measured by using a P-Trak (TSI) in the size range 20 nm–1 μm. The P-Trak is a condensation particle counter that monitored the number of particles every second during 8 hr. The Grimm PDM is an optical particle counter that was used to measure the particle concentration in 31 size fractions between 250 nm–32 μm. The monitor was bought from TSI in 2005 and was not calibrated before the measurements that were carried out in 2006.
 
Dust-Trak (model 8520; TSI) measured PM10 mass concentration of particles once every second during the measurements at the paving site.

The sampling of particulate carbon was done at the same time as sampling of UFPs during asphalt paving at the rural paving site. The respirable fraction of the particles was collected onto 37-mm, quartz filters in sampling cassettes with cyclones with a flow rate of 10 L/min. The quartz filters were pre-cleaned in a furnace at 800◦C for3 hr.

A thermo-optical transmittance carbon analyzer (Sunset Laboratory Inc., Tigard, Ore.) using the National Institute for Occupational Safety and Health (NIOSH) 5040 protocol quantified the carbonaceous fraction of collected particles and specified the fractions as organic carbon (OC) or elemental carbon (EC). This is accomplished through stepwise heating, first by an oxygen-free mode where carbon that volatilizes is analyzed as OC. After this step, the temperature drops, oxygen is introduced, and stepwise heating starts again. The rest of the carbon from this second step, correcting for charring events, is termed as EC. The sum of the OC fraction and the EC fraction is termed total carbon (TC).

Data Analysis

With the use of cumulative probability plots, the exposure data were found to be best described by log-normal distributions and were ln-transformed before further statistical analyses. The measured exposure values were used without further adjustment as they were regarded as representative of the whole work shift. Standard measures of central tendency and distributions (arithmetic [AM] and geometric means [GM] and geometric standard deviations [GSD]) were calculated. Kruskal-Wallis tests were used to evaluate the differences in the exposure levels among the job groups. Mann-Whitney tests were used for two-group comparisons for total dust, oil vapor, and PAHs.


Results and discussion

Tables II and III give a summary of exposure levels by agent. Only a few analyses of volatile organic compounds (VOC), respirable dust, and quartz were done due to previous analyses determining low concentrations.

A Kruskal-Wallis test between job groups showed statistical difference for total dust, oil vapor, and total PAHs. The mean exposure levels across job groups showed a moderate variability (Table IV).

The asphalt strippers had a significantly higher exposure to total dust than the other asphalt workers (2.3 mg/m3 vs. 0.5 mg/m3, p < 0.001). The members of the paving team (paver operators, screedmen, roller drivers, and drivers of the binding agent trucks) had a significantly higher exposure to total PAHs compared with other asphalt workers (transport truck drivers, asphalt strippers, and plant operators) (1.8 μg/m3 vs. 0.43 μg/m3,p 0.001).

Although the geometric mean exposure to nitrogen dioxide was less than 1 ppm for all groups, the paving group may be exposed to higher levels when paving asphalt in tunnels (max 3.4 ppm measured). Although the geometric mean exposure of oil mist was less than 0.3 mg/m3, exposure to oil mist also can be significant. On one occasion, oil mist was measured to 2.5 mg/m3 in an asphalt paving machine without cabin (stationary sampling). Exposure to oil mist was correlated to exposure to total dust (Spearman’s rho = 0.82, p < 0.001).
 
Paver operators seated on paving machines without a cabin showed significantly higher levels of total dust (GM0.40 mg/m3 [GSD 2.0] vs. GM 0.23 mg/m3 [GSD 2.0], p 0.047) and oil mist (GM 0.25 mg/m3 [GSD 2.2] vs. GM 0.094 mg/m3 [GSD 2.3], p 0.003) than paving operators on paving machines with a closed cabin.

Paving in tunnels showed significantly higher levels of total dust (GM 0.60 mg/m3 [GSD 2.0] vs. GM 0.23 mg/m3 [GSD 2.3], p < 0.001), total PAHs (GM 3.5 μg/m3[GSD 1.5] vs. GM 0.88 μg/m3 [GSD 2.2], p < 0.001), oil vapor mg/m3 (GM 1.5 [GSD 1.4] vs. GM 0.96 mg/m3 [GSD 1.5], p 0.03), and a nonsignificant higher level of oil mist mg/m3 (GM 0.32 [GSD 1.5] vs. 0.11 mg/m3 [GSD 2.2], p 0.09) than paving outdoor. Only two measurements of oil mist from paving in tunnels were performed.

Although amines had been added to the asphalt mass on the occasions that sampling was carried out, three stationary measurements of amines sampled 20 cm above the screed were under the detection limit of 0.0006 mg/m3.

Particle number concentration as measured by P-Trak at a rural road paving site (about 60 km from Stockholm) on a four- lane motorway is shown in Figure 1. No significant difference in particle concentrations was observed when driving at different positions beside the screedmen. Ultrafine particles at the paving site originated mainly from asphalt paving activities and traffic exhaust. Background number of particles was measured for 1.5 hr before the asphalt paving started but at a time when the road was already busy.

Traffic was passing the paving site at a reduced speed of 70 kph. On the opposite side of the road as compared with the paving site, traffic was running as normal (speed limit 110 kph). The background median number of particles was 7000 cm−3at the paving site (prior to paving), originating mainly from the traffic exhaust. During paving, a higher concentration was observed. A median particle concentration of 3.4x104 cm−3 (interquartile range 1.3x 104–7.5x 104 cm−3) was measured close to the screedmen walking behind the paving machine.

However, frequent peaks often reached particle concentrations of 1x105–2.2x105 cm−3. For outdoor measurements, the number of ultrafine particles may be influenced by atmospheric conditions, e.g., ambient temperature(14) and wind speed and direction.(15,16) On a calm day one could expect a higher particle concentration at the paving site, especially since peak concentrations up to 2.2x105 cm−3 were frequently observed during paving activities.

Compared with larger particles, the number of UFPs (<100 nm) is generally large. Measurements at a busy roadside in Birmingham showed that more than 90% of the particles at that site were ultrafine particles. About 80% were smaller than 50 nm, and more than 50% of the particles were smaller than 30 nm.(17) Therefore, it is important to consider the lower cutoff diameter of the particle monitor when comparing results with other measurements. P-Trak has a lower cutoff diameter at 20 nm, i.e., the smallest particles in the ultrafine range will be too small to be detected. The background as measured in this study using P-Trak was 7000 cm−3 and represents the
particle number contribution from the traffic in a rural area (Sweden).

Molnar et al.(15) reported particle levels of 2000 cm−3 at a suburban roadside in Gothenburg, Sweden, and 5000–7000 cm−3 at traffic intense periods as measured between 10 to 368 nm. Gidhagen et al.(14) measured particle levels of 3x104 cm−3 (cutoff diameter 3 nm) 10 m from a heavily trafficked highway in an open rural area (Sweden). The same authors also reported particle concentrations from urban sites, for example, where a street canyon (Hornsgatan) in Stockholm had a monthly average of 6x 104–7x104 cm−3 (cutoff diameter 10 nm).

Hot processes like welding and combustion (including engine combustion forming exhausts) generate high concentrations of UFPs. Asphalt paving was performed at around 150◦C and was also found to generate UFPs. Most UFPs seemed to originate from the asphalt fumes, since the particle concentration directly dropped to background concentration when paving temporarily stopped and no more hot asphalt exited the paving machine (Figure 1) . There were several brief stops during the day when, for example, the paving machine stood still while being filled with new glue, or when one truck had emptied its asphalt (which was made continuously during paving) into the paving machine, and a new, full asphalt truck was moved into position. All these stops can be seen in Figure 1 as dips in the particle concentration. However, apart from the asphalt fumes, traffic exhaust also contributes to the total exposure of UFPs of the screedmen.

There was a clear difference between the number of ultrafine particles when paving and prior to paving (and when the paving temporarily stopped during the day). The contribution from only the paving activities was 2.7x104 cm−3 after subtracting the background level. To confirm that the asphalt mix generated ultrafine particles, measurements of UFPs were also taken at an asphalt plant where the presence of traffic exhaust could be excluded. Particles were sampled in close vicinity to newly produced asphalt. Particle peak concentrations (when a carriage was filled with asphalt as earlier described) were about the same level as the peak concentrations at the paving site, and it was concluded that hot asphalt itself can generate high concentrations of UFPs. Peak concentrations between 1.5x105 and 2.5x105 cm−3 were observed during the measuring time of 1 hr at the asphalt plant. Observed background concentration at the asphalt plant was 4500 cm−3 (cutoff diameter 20 nm), which can be compared with a particle concentration at a rural site outside Stockholm that was reported to be 3000 cm−3 (cutoff diameter 10 nm).(14) 

Particle number size distribution of ultrafine particles as measured by an SMPS at a road paving site is shown in Figure 2. The size distribution has an irregular shape, since it is the average distribution of several scans during one day. If the wind was fluctuating during a scan or if the paving machine stopped temporarily, the particle concentration could suddenly drop to background levels. Such distributions were excluded from the data set, while all complete scans (where no obvious change in wind direction was observed nor any temporary paving stops occurred) were added, and the result was divided by the number of scans to get an average size distribution. This size distribution gave a picture of what particle sizes the screedmen were mainly exposed to regarding number concentration of particles. The particle size distribution was fitted to a lognormal distribution. The geometric mean diameter of the number size distribution of particles (Dpg) at the paving site was 70 nm. 

The contribution from traffic generates smaller particles, as reported from measurements in street canyons. Gidhagen et al.(14) report a peak of the size distribution close to 20 nm in diameter for fresh exhaust particles, and according to Wehner,(18) the peak was at 15 nm, as measured in a street canyon during rush hour.(9) Molna´r et al.(15) found a number concentration peak between 15–50 nm for traffic- related particles.(15) A small peak at 23 nm is present in the particle size distribution as presented here (Figure 2). This peak is most probably due to traffic exhaust. The particles at the asphalt plant were larger as compared with the particles at the paving site. The geometric mean diameter was 180 nm, possibly due to the higher temperature of the asphalt (160◦C), but it was not further examined.

The number of particles larger than 764 nm (maximum monitored particle size at the operating flows of the SMPS) is usually low compared with the number of UFPs. This was also the case in asphalt paving, confirmed by the results of the Grimm PDM, which showed that the concentration of particles above 700 nm was about 0.1 cm−3. This means that the percentage of ultrafine particles (particles smaller than 100 nm) can be determined from the particle size distribution as obtained from the SMPS. At the paving site, 74% of the measured particles were ultrafine.

The results from the Grimm PDM show that hardly any particles were larger than 4 μm as measured at the level of the breathing zone of the screedmen. Particles larger than 4 μm may have been formed during paving, but many of them probably sediment before reaching the aerosol inlet of the particle measurement instruments at the conditions during the measurements. According to Herrick et al.(19) the mass median diameter of particles in a paving environment was about 1 μm, and the great majority of the mass was contained in particles less than 3.5 μm in aerodynamic diameter, which seems to agree with the observations from hot mix paving in the present study.

Dust-Trak was also used at the paving site to measure PM10 mass concentration. Median background mass concentration (as measured before paving started but when the road was already trafficked) was 21 μg/m3, while median mass concentration during paving was 29 μg/m3. The number of particles does not necessarily correlate with the mass of particles, since ultrafine particles often are numerous, but does not contribute much to the particle mass. The mass concentration as monitored by Dust-Trak, however, followed the same trend as the number concentration as monitored by P-Trak during the measuring day; i.e., an increase in number concentration of particles also gave an increase in mass concentration of particles at the paving site. In this case there seems to be a simultaneous increase in the generation of micron and nano- sized particles at the paving site.

Measurement of particulate carbon was performed during asphalt paving for 8 hr. The arithmetic mean ( SE ) of three samples was for elemental carbon 3.0 ( 0.2) μg EC/m3 and for organic carbon 42.1 ( 4.2) μg OC/m3. This gives a total carbon of 45.2 ( 4.1) TC/m3. The ratio of EC/TC was 6.7%.

Elemental carbon is commonly used as a marker for diesel particulate matter in occupational exposure measurements. Some potential confounders to EC formation (except from diesel fumes) are other sources of combustion, such as cigarette smoke, wood/biomass burning, and gasoline exhaust. No workers were smoking during the working hours of this study. A possible contribution to EC from gasoline-driven engines in passing cars is probably negligible, as gasoline exhaust contains much smaller amounts of soot. No known sources of wood or biomass burning or other combustion were near the site.

Background levels in the air of EC in the United States are about 2–3 μg/m3 according to Birch and Cary.(12) Outdoor occupational exposure levels of EC in Sweden found by Lewne et al.(20) were reported to be 7.8 μg EC/m3 for construction machine operators and 4.1 μg EC/m3 for other outdoor workers exposed to diesel exhaust (calculated as geometric mean). The corresponding values for total carbon in the study by Lewne et al. were 19.7 μg TC/m3 and 9.5 μg TC/m3. In their study, a different method for analyzing carbon was used—the German method BGI 505-44. Good agreement for TC has been shown between for German method BGI 505-44 and the NIOSH 5040 method used in this article, but for EC as analyzed on different diesel samples, the NIOSH 5040 method showed 30–60% lower values.(21) The EC results presented for asphalt paving in this article agree well with the occupational exposure levels of EC reported by Lewne et al. However, the levels of exposure of TC and, therefore, the exposure of OC, in this article were higher than compared with the study by Lewne. This indicates an organic contribution to TC during asphalt paving that was not present in the Lewne´ study where workers were exposed to motor exhaust only.

NIOSH recommends using EC as a marker for DPM, but other agencies (e.g., Mine Safety and Health Administration) regulate exposure of DPM to be measured as TC. If TC is used as a marker for DPM, there are many organic substances that can add on to the OC fraction and therefore interfere with the measurement. Sirianni and colleagues(22) found that cutting oil mists and airborne creosote made OC levels very high in an engine factory and a wood treatment plant where diesel equipment was used.

The EC/TC ratio for diesel exhaust varies with fuel com- position, engine type, maintenance, lube oil consumption, and ambient conditions. The EC/TC ratio in diesel exhaust has been suggested to be around 0.30–0.80 viewing the whole sample as DPM.(23) Much lower EC/TC ratios, as found in this study (0.067), suggest interferences in the OC fraction from the asphalt fumes. Asphalt fumes consist of organic substances with particle sizes in the respirable fraction. This probably explains the high OC levels and the low EC/TC ratio in this study. The results from these measurements of carbon indicate that part of the carbon originates from diesel exhaust. However, the main part of the OC fraction originates from another source, most likely the asphalt fumes.
