About Us | Help Videos | Contact Us | Subscriptions

Journal of Environmental Quality - Article



This article in JEQ

  1. Vol. 42 No. 1, p. 21-29
    Received: Apr 16, 2012
    Published: December 6, 2012

    * Corresponding author(s): ecwitt@usgs.gov


Preliminary Assessment of an Economical Fugitive Road Dust Sampler for the Collection of Bulk Samples for Geochemical Analysis

  1. Emitt C. Witt *ab,
  2. David J. Wronkiewiczbd and
  3. Honglan Shicd
  1. a U.S. Geological Survey, National Geospatial Technical Operations Center, Rolla, MO 65409
    b Missouri Univ. of Science & Technology, Dep. of Geologic Sciences and Engineering, Rolla, MO 65409
    d Missouri Univ. of Science & Technology, Environmental Research Center for Emerging Contaminants, Rolla, MO 65409
    c Missouri Univ. of Science & Technology, Dep. of Chemistry


Fugitive road dust collection for chemical analysis and interpretation has been limited by the quantity and representativeness of samples. Traditional methods of fugitive dust collection generally focus on point-collections that limit data interpretation to a small area or require the investigator to make gross assumptions about the origin of the sample collected. These collection methods often produce a limited quantity of sample that may hinder efforts to characterize the samples by multiple geochemical techniques, preserve a reference archive, and provide a spatially integrated characterization of the road dust health hazard. To achieve a “better sampling” for fugitive road dust studies, a cyclonic fugitive dust (CFD) sampler was constructed and tested. Through repeated and identical sample collection routes at two collection heights (50.8 and 88.9 cm above the road surface), the products of the CFD sampler were characterized using particle size and chemical analysis. The average particle size collected by the cyclone was 17.9 μm, whereas particles collected by a secondary filter were 0.625 μm. No significant difference was observed between the two sample heights tested and duplicates collected at the same height; however, greater sample quantity was achieved at 50.8 cm above the road surface than at 88.9 cm. The cyclone effectively removed 94% of the particles >1 μm, which substantially reduced the loading on the secondary filter used to collect the finer particles; therefore, suction is maintained for longer periods of time, allowing for an average sample collection rate of about 2 g mi−1.


    CFD, cyclonic fugitive dust; EDS, energy dispersive spectroscopy; HDPE, high-density polyethylene; PVC, polyvinyl chloride

Fugitive dust is an airborne contaminant that is common in rural areas where unsurfaced gravel and dirt are the predominant road surfaces and in urban areas close to heavy industry and truck traffic, thus generating higher population exposures and accompanying public health impacts. The health effects of fugitive dust range from aggravation of the respiratory system to systemic addition of trace elements in the blood through inhalation and ingestion. Studies have shown that Pb in dust is responsible for significant increases in blood-Pb levels of persons living in close proximity to Pb-dust sources such as mining, smelting, recycling, and other industrial activities (Murgueytio et al., 1998; Kerin and Lin, 2010; Johnson and Bretsch, 2002; Thornton et al., 1990; Needleman et al., 1990). Trace metals associated with fly ash and bottom ash released from power plants and incorporating onto road surfaces can also have a direct effect on human health (Zeneli et al., 2011). Contaminated unsurfaced roads can be a persistent and unavoidable health hazard for those living beside or regularly traveling these roads. Given the human health concern of fugitive dust, field methods need to be refined to provide a representative, repeatable, and sufficient quantity of sample to better characterize the chemical hazard of this exposure mechanism.

The collection of samples for the chemical characterization of fugitive dust has been largely limited to point sources where swabs are obtained from homes, vehicles, roadside vegetation, deposits on soils, cascading impactors, or collected by installing open pans near unsurfaced roadways in potentially contaminated areas (Erel and Torrent, 2010; Duggan and Inskip, 1985; Kerin and Lin, 2010). These methods have served an important role in defining fugitive dust contamination; however, with respect to fugitive road dust sampling where an 8- to 16-km reach of roadway is to be sampled from a moving vehicle, these methods do not support repeatability and representativeness (Que Hee et al., 1985). Swabs from homes and dashboards of vehicles only represent those particular locations, and contaminated dust collected in these environments is difficult to track back to a road surface. Roadside vegetation only represents a single point along a roadway, and many samples would need to be collected to represent an 8- to 16-km reach. Pan collection is generally unattended and can be compromised by vandals or weather and roadside vegetation; several of these would need to be deployed to represent an entire road reach. Surface soils that are alleged to be “transported dust” do not provide defensible results because the sample can be a mixture from several sources. Impactors are focused on particle size delineation collecting size ranges that will be harmful when inhaled and do not take into consideration that particle ingestion of all suspended sizes is a much larger contributor to elevated blood-Pb levels (Steele et al.,1990; Biggins and Harrison, 1980; Barltrop and Meek, 1979). Also, these samplers use greases, oils, and other compounds that facilitate collection of samples for analysis. This has been recognized as a problem for representative chemical analysis (USEPA, 1983). Furthermore, and most importantly, these methods neither produce the needed quantity of sample for rigorous geochemical characterization and reference that would include the need for several grams of material to perform X-ray diffraction, total and sequential extraction, particle size analysis, gravity separation, and repeat analysis or provide a spatially integrated characterization of the health hazard. To provide the scientific community with another technique for collecting fugitive dust, an inexpensive collection approach was developed to chemically characterize dust that is collected directly from suspension before it interacts with other substrates in the immediate environment and without gross assumptions regarding its origin.

The sampling approach presented herein uses a cyclonic dust collection design that is commonly used by particle-separation industries throughout the world and is inspired by the pioneering aerosol collection research proposed by May (1945), where successive plates were used to impact particles as they pass through a fluid stream. Cyclonic separation effectively removes particles from a fluid by establishing a helical flow pattern inside a conical container. The vortex acts on the entrained particles, creating centrifugal forces that push the particles to the walls of the conical container, allowing them to slow by friction processes and slide down the container sides and into a collection bin. Standard conical cyclones (vertical configurations) have collection efficiencies ranging from 10 to 100% for particle sizes ranging from 0.3 to 10 μm, respectively (Wark et al., 1998). Incorporating a cyclone into the collection air stream removes the large particle component of the fugitive dust before affecting the filter. This prevents filter blockage due to mesh space filling and loss of suction in the sampler, promoting greater collection efficiency and longer collection times and enables a larger volume of sample per unit distance of road surface. There are cyclonic collectors commercially available for the collection of environmental samples, but they are expensive, do not provide the quantity of sample needed to perform rigorous and repeated geochemical analyses, and have not been demonstrated to be useful in fugitive road dust studies where samples are collected from moving vehicles over long distances and from varying road surface heights.

The purpose of this paper is to present the design of a novel and economical cyclonic fugitive dust (CFD) collection system that can be used for fugitive dust studies where trace metal contamination of these suspended solids is a concern. Also presented are data from a preliminary assessment of the CFD sampler when deployed using a moving vehicle collection process. This was accomplished through repeated deployment in an identical sampling environment and through use of scanning electron microscopy (SEM), energy dispersive spectroscopy (EDS), and partial-digestion chemical analysis to characterize the samples collected. The results presented here include a description of the sampling device, methodology used to collect samples, characterization of test samples collected using particle size analysis, and qualitative and quantitative chemical analysis. The results of this study demonstrate the utility and representativeness of the sampling method and address the variables affecting sample collection and quantity.

Materials and Methods

Sampler Design and Deployment

The CFD sampler is relatively simple to construct with inexpensive materials and was specifically built for collecting fugitive dusts generated by a moving vehicle for the purpose of collecting a large amount of sample to facilitate chemical digestion, SEM, X-ray diffraction, and other analyses. However, with minimal modifications this sampler could be used in stationary situations where a representative sample is needed to characterize fugitive dust effects at a single location. The sampler is designed to be quickly dismantled in the field for cleaning between sampling locations. It consists of a wooden frame that can be constructed in any configuration to hold a 10-L Nalgene cyclone collection container, a cyclone (conical tube), intake tubing, and a small 5.5 peak horsepower shop vacuum with an airflow rate of 5.52 m3 min−1 (Fig. 1). The configuration presented here uses clear flexible polyvinyl chloride (PVC) tubing (6.35 cm inside diameter) for the air intake and a high-density polyethylene (HDPE) cyclone (Dust Deputy, Onieda Air Systems). The cyclone is mounted to a piece of wood with a hole cut out for the cyclone opening, and below the opening is a number 415 wide-mouth Nalgene cap that is glued into a recess that was cut out with a router. This recessed area allows for easy replacement of exchangeable sample containers between sampling locations. A 5-μm mesh tortuous-path filter is attached to a plastic intake adaptor that comes with the vacuum. This effectively traps extremely fine particles that do not deposit in the HDPE cyclone collection container. A static pressure sensor was added to the intake portion of the vacuum housing before the filter. This sensor was attached to an analog gauge that measures pressure changes in units of water column and was used to test the airflow changes resulting from filter loading. All parts that come into contact with the sample are either PVC or HDPE, with the exception of the in-line static pressure tube, and are acid washed with 5% HNO3 solution, rinsed with deionized water, and allowed to air dry before deployment. The vacuum is powered by a DC/AC inverter mounted inside the cab of the truck. The 10-L sample container is weighed before and after sampling in the field using a portable electronic scale with measurement capability to the nearest gram.

Fig. 1.

Photographs showing the prototype cyclonic fugitive dust system designed by the authors. Left image: Red arrows show vacuum air flow direction; white arrows show locations of cyclone chamber for collection of >1 μm–sized particles and filter for capture of <1 μm–sized particles. Right image shows prototype device placed in a pickup truck bed for trial collection of dust particles. The sampler has been field tested with up to 2 g mi−1 collected.


The CFD sampler is deployed by positioning the unit in the rear-most area of a pickup truck bed so the intake reaches over the edge behind the rear tires and on the opposite side of the exhaust pipe (Fig. 1). The optimal distance from the back of the truck to the CFD intake was determined to be about 90 cm; however, this may vary depending on the make and model of truck. The CFD sampler intake also should be positioned so it collects suspended dust and not fine sand to granule-size particles thrown by the vehicle tires. This is accomplished by angling the sampler unit at 45 degrees from the longitudinal axis of the tailgate.

After setting the sampler in the proper position and securing it to prevent movement, the track log setting is selected on a hand-held Geospatial Positioning System unit, the vacuum is turned on, and the route of unsurfaced road is driven for a user-defined length at a speed of 40.2 to 56.3 km h−1. At the conclusion of the sampling event, the 10-L collection container is weighed, and the sample is transferred from the 10-L collection container to a certified acid-washed 118 mL storage container and labeled with pertinent information. The secondary filter is removed from the vacuum housing and stored in a labeled polyethylene sampling bag. All field notes are reconciled to the identification information on the sample storage containers, and the containers are placed in a clean dry cooler for transport to the laboratory. Before collecting another sample, the entire unit is dismantled for washing and acid-rinsing of the cyclone, replacement of the PVC intake, and cleaning of the vacuum intake and filter holder. Because of the time it takes to clean and dry the PVC intakes between sample locations, used tubing is replaced with precleaned and acid-rinsed tubing of the same specifications. The PVC tubing was chosen because the material can be bent around a small radius without collapsing, is clear so obstructions can be identified and removed, and resists repeated acid washing without becoming brittle through degradation. No attempt is made to clean the 10-L cyclone sample collection container in the field. Additional pre-cleaned and acid-rinsed 10-L containers were exchanged between each sampling site. It is important that all sampler components be dry before sample collection or a substantial amount of sample may be lost to the walls of each component. Under optimal conditions and with careful pre-field planning, 6 to 10 samples can be collected in a 10-h day.

The secondary filter chosen for this study is a household dust mask with a nominal mesh size of 5 μm. This mask is a tortuous-path fiber filter that would typically be used by the public (National Institute for Occupational Safety and Health–N95) for reducing exposure to dust and other airborne contaminants. This filter was chosen because it represents a filtration product typically used by the general population for limiting dust inhalation. Furthermore, it has low loading potential and is effective at grabbing charged particles that are smaller than the filter mesh size. Figure 2 shows a SEM image of the secondary filter with particles smaller than the mesh size attached to the media by Van der Waals forces. Other filters can be used, but loading is a factor that must be considered if a long road distance and large sample quantity are needed for the study. Incorporating smaller pore size filters in the system requires successive layers from larger to smaller pore and mesh sizes. This would reduce loading and provide the best analysis of submicrometer particles. Optimal designs for adding finer filter sizes in succession are being investigated for this system but have not been tested. The use of commercially available cascading impactors may be an appropriate addition for postcyclone particle separation if the collection plates can be shown to be free from trace element contamination. Furthermore, adding this feature will substantially increase the cost of the sampler.

Fig. 2.

Scanning electron microscope image showing fugitive road dust particles aggregated and attached to the fibers of a standard tortuous-path dust mask (NIOSH-N95) with a nominal mesh size of 5 μm.


Field Testing

The field testing component of this validation effort was done on an unsurfaced gravel/dirt road in rural Ozark County, Missouri (beginning lat. 36.620468, long. −92.26936; ending lat. 36.620511, long. −92.269356). This location was selected because it is extremely rural with minimal vehicular traffic. To minimize variability, all samples needed to be collected in 1 d and sampling conditions needed to be identical; other vehicles on this road at the time of sampling would have compromised the results. After a 1-wk period of dry weather in October 2011, nine separate samples were collected on the same 5.63-km segment of the selected road. The strategy used was to collect five samples with the intake set at 50.8 cm above the road surface and five samples at 88.9 cm above the road surface. The intake hose fell off during collection of sample six resulting in the invalidation of this sample for analysis. The selected sample size was focused on the number of samples that could be collected during a single day. Estimated sample size was calculated using the average number of grams of sample collected per km and substituting this as a population size estimate. Then the equation of Krejcie and Morgan (1970) was used to calculate the sample size of 8 to 10 samples.

The sampler is designed so the user can select any height above the road surface simply by using plastic wire clamps to hold the PVC tubing to the sampler frame. The 50.8-cm height was selected to provide a distance from the road surface that would collect a large quantity of sample without collecting coarse material not normally suspended that could be thrown by the tires. Initial testing of the sampler with the intake set directly behind the tires resulted in gravel-size particles being drawn into the suction air stream. This was eliminated by angling the CFD sampler at 45° to avoid tire spray so the sample that is collected represents only particles that were suspend for 10 or more seconds. The 88.9-cm height was selected to represent the average height to the window opening of a compact to a standard size sedan vehicle. This would be the exposure region of suspension for individuals driving unsurfaced roads with their windows down. Wind shear generated along the sides of the moving sampling vehicle can affect sampling efficiency; however, this variability was not quantified. Sample collection times for this analysis ranged from 15 to 17 min, with no stoppages during the collection period. There was no change in static pressure from the beginning to the end of each sample, so it was assumed there was no loss of suction during collection due to filter loading. Therefore, per vacuum design specifications, the sampler processed about 83.5 m3 of dust-laden air. Airflow during sampling would likely change for longer road reaches; therefore, this technique may be improved by adding a more precise flow-metering device linked to a continuous data logger. The cyclone sample container was weighed before and after each sampling period to determine the quantity collected. Samples collected by the cyclone were labeled C1 through C9, and those collected by the secondary filter were labeled F1 through F9.

Two additional samples (DU1 and DU2) were collected from unsurfaced gravel or dirt roads located in forested rural settings of the Viburnum Trend mining subdistrict of Missouri’s Pb-mining and metals processing region. Soils in the region have been shown to contain persistent Pb contamination, making this an ideal place to test the CFD sampler’s ability to collect contaminated dust (Bolter et al., 1975; Bornstein, 1989; Rucker, 2001). Sample DU1 represents an 8-km reach of unsurfaced road (beginning lat. 37.657778, long. −91.178611; ending lat. 37.683333, long. −91.166667) that was traveled twice to accumulate a large quantity of sample. This sample was collected in September 2011, and the weather conditions were extremely dry with an air temperature of 36.7°C (98°F). The road surface was primarily composed of quarry stone layers on top of native soil with an overall well maintained appearance. The road terminated at a gated fire-access road that was not passable by the air-sampling vehicle. There was no other traffic on this road during the 30-min sampling period.

Sample DU2 represents a 20.4-km reach of unsurfaced road that follows a ridge top in a heavily forested area (beginning lat. 37.533333, long. −91.050; ending lat. 37.566667, long. −91.133333). The road is maintained by the U.S. Forest Service and is primarily used for logging operations but also receives some traffic from local residents. Quarry stone additions to this road surface were minimal, so the sample primarily represents dust generated from exposed native soils. This sample was collected in October 2011, and the weather was a mild 15.5°C (60°F) with a clear sky. There was no other traffic on this road during the 35-min sampling period. An odor assumed to be derived from the nearby secondary Pb processing plant toward the end of the road reach was noted.

Particle Size Analysis

Particle sizes were determined from images collected using a Hitachi S-4700 SEM located at the Missouri University of Science & Technology (MS&T), Advanced Materials Characterization Laboratory. Samples collected by the cyclone were prepared by suspending small aliquots of dust in ethanol and pipetting drops onto SEM stubs covered in double-sided carbon adhesive tape. To reduce any potential particle size bias, additional samples were prepared by dipping stubs covered with the carbon adhesive tape into an aliquot of sample followed by rinsing with compressed air. The ethanol suspension method alone may be preferential to the smaller particles because the larger particles tend to drop out of the solution more quickly. To improve the randomness of the analysis, images from both preparation methods were used. All cyclone samples were sputter coated using a gold-palladium (Au-Pd) target, with a sputter time of 120 s being sufficient to reduce charging effects.

Dust that was collected on the secondary filter was rinsed from the filter into a collection dish using ethanol. To improve image quality and encourage particle separation, the samples were mounted on a silicon wafer that was mounted on a SEM stub with the carbon adhesive. To increase the number of particles per image, some samples were prepared by simply dipping the carbon-taped stub onto the surface of the filter. All filter dust samples were sputter coated for 240 s with Au-Pd.

Imaging conditions for the cyclone-collected samples included a 10 kV accelerating voltage, 11 nA emission current, 12.3 mm working distance, and zero tilt. The lower secondary electron detector within the microscope produced the best quality image with the least amount of charging effect. All images were acquired at 300× magnification. Instrument imaging conditions for the filter-collected samples included a 10 kV accelerating voltage, 9 nA emission current, 12 mm working distance, and zero tilt. The best image quality was obtained at 10,000× magnification. Particle size was measured manually using the ImageJ software (National Institute of Health, version 1.44p) by measuring the longest axis of a given particle (maximum Feret diameter) identified in each of 10 randomly selected horizontal tracks across the image. Cubic particles were measured on an axis diagonal to the corners on a single plane. Between five and eight images were acquired for each sample. A total of 118 images were collected; of these, 100 images were used for particle size analysis. Graphical and inferential statistics were performed using STATA 9 software (StatCorp).

Chemical Analysis

Cyclone-collected particle samples were analyzed at the MS&T Environmental Research Center Laboratory for the trace elements Pb, Zn, Co, Ni, Cu, Cd, and As. For each sample and standard reference material, approximately 0.5 g of material was weighed accurately and added to 50-mL, acid-washed disposable digestion vessels and digested using U.S. Environmental Protection Agency method 200.2 (USEPA, 2003). The chosen digestion method was designed to partially digest dry soil and sediment samples for the determination of trace element concentration with the exclusion of metals bound into digestion-resistant silicate minerals. In general, digestion was accomplished by adding 2 mL of diluted HNO3 mixture (500 mL concentrated trace metal–grade HNO3 to 500 mL ASTM Type I water) and 5 mL of diluted HCl mixture (200 mL concentrated trace metal–grade HCl to 400 mL of ASTM Type 1 water) to the digestion vessels containing the samples. The vessels were digested at approximately 95°C for 45 min. Ribbed polypropylene watch glasses were placed on the open digestion vessel for the duration of the digestion process. After the hot block digestion time, the samples were allowed to cool, the watch glass residue was rinsed into the vessel using ASTM Type 1 water, and the sample volume was brought to 50 mL using the same water. Samples were capped and shaken before allowing to settle for 24 h. Extracts were decanted from the digestion vessels into precleaned, 200-mL polypropylene storage bottles.

Immediately after digestion, samples were analyzed by inductively coupled plasma–mass spectrometry after appropriate dilution. A model Elan DRCe inductively coupled plasma–mass spectrometry instrument (PerkinElmer SCIEX) equipped with a cyclonic spray chamber with a Meinhard nebulizer and platinum cones was used for analysis. The samples were delivered at 1 mL min−1 by a peristaltic pump. The RF power was set at 1500 W. Argon flow rates for the plasma and auxiliary gas were 15.0 and 1.2 L min−1, respectively. Quantitation was performed using an internal standard method. A multi-element internal standard mixture purchased from PerkinElmer (PerkinElmer SCIEX) was added continuously online. Arsenic as arsenate was detected by DRC mode to eliminate the chloride interference from the HCl addition during digestion. Oxygen was used as the DRC reaction gas.

To ensure good quality data, USEPA-recommended QA/QC methods were followed. Laboratory quality control included measurement of blanks, standards, and duplicates. All digestion vessels, lids, watch glasses, and extract containers were soaked in 5% HNO3 for 24 h, rinsed with trace element–grade deionized water, and allowed to air dry before using. All laboratory consumables were used once and discarded. National Institute of Standards and Testing standard reference materials SRM 2711 Montana II Soils and SRM 2587 Trace Elements in Soil (contains lead from paint) were analyzed in triplicate to assure method recovery and to measure the variability of the analytical process. Instrument calibration was performed at the element concentration of 0.02 to 50 μg L−1 linear range. Good linearity (R2= 0.9999–1.0) was obtained. Energy Dispersive Spectrums were acquired using the Hitachi S-4700 and EDAX system. These were used to estimate dust particle compositions and to assist with estimating dilutions for the sample digestions before analysis. The EDS analyses are conducted without standards and thus are semiquantitative.

Results and Discussion

Size distribution analyses for 2657 particles were performed from cyclone samples C1 through C9. Sample C9 had the fewest measurements with 150 particles measured, and C2 had the most with 450 particles measured. Distribution of particle size is nearly the same for each sample with some variation in outliers (Fig. 3). Group 1 (samples C1–C5) represents samples collected at 50.8 cm above the road surface. The quantity of sample collected at this level ranged from 3 to 11 g (average, 7 g), resulting in a collection average of 2 g mi−1. Group 2 (samples C6–C9) represents suspended particles collected at 88.9 cm above the road surface. The quantity of these samples ranged from 2 to 4 g (average, 2.7 g), resulting in a collection average of 0.77 g mi−1. For group 1, particle sizes ranged from 1.97 to 92.6 μm; for group 2, particle sizes ranged from 3.31 to 96.1 μm. An uneven paired t test of the means suggests there is no significant difference in particle size within and between the two groups. In general, larger particles were found more often in samples collected at a height of 88.9 cm; however, the presence of these particles was not sufficient to alter the median particle size relative to those collected at 50.8 cm. The mean particle sizes for groups 1 and 2 were 17.3 and 18.6 μm, respectively. A paired t test of these means indicates there is no significant difference. Given this, it was determined that the expected average particle size to be collected by the cyclone is 17.9 μm. This effort does show that large particles ranging from 30 to 90 μm can be suspended to heights of 88.9 cm and possibly higher above the road surface. From a human health perspective, these larger particles could be problematic in contaminated areas where suspension would likely be a driving mechanism for ingestion or inhalation, especially if these large particles are associated with a trace metal contaminant such as Pb.

Fig. 3.

Boxplots showing the distribution of particle size for cyclone and filter mesh samples of fugitive dust collected at 50.8 and 88.9 cm above the road surface while traveling at speeds ranging from 40.2 to 56.3 km h−1.


Particle analysis was not performed on samples DU1 and DU2; however, one image was acquired of cyclone-collected DU2 dusts, and 55 particles were measured. Although the results of this count are insufficient in quantity to assure a random sampling, the range of particles is consistent with those observed in Fig. 3. The minimum and maximum particle sizes were 4.0 and 86.1 μm, respectively. Average DU2 particle size was computed at 20 μm (SD, 13.9 μm).

Scanning electron microscopy images show the range of particles size and shape (Fig. 4). Energy-dispersive spectroscopy mapping combined with phase morphology suggests that most of the particles are quartz, calcite, and Al-Si clays. Ambient soils in the study area are derived primarily from carbonate rock, but most of the particles observed in the imagery are likely from the periodic additions of quarry stone by road maintenance crews. Except for the large difference in scale, the particles in the images do not look that much different from the aggregate on the road surface.

Fig. 4.

Scanning electron microscope images showing particles collected by the cyclone at 50.8 cm (left) and 88.9 cm (right) above the road surface. There is no statistically significant difference between the two sample populations.


More than 800 particles were measured from images of the mesh filter media. Particles captured by F1 through F5 ranged from 0.183 to 3.19 μm, and F7 and F9 ranged from 0.187 to 1.63 μm. The distribution of particle size for the seven samples is shown in Fig. 3. In general, there was less variability in the filtered particles than in the cyclone particles, and fewer large outliers were observed. The average particle sizes captured for the filtered samples at 50.8 and 88.9 cm above road surface were 0.625 and 0.561 μm, respectively. A paired t test between all sample means indicates there is no difference. Furthermore, less than 6% of the particles on the filters were larger than 1 μm, indicating that the cyclone efficiently removes about 94% of the larger particles.

Figure 5 shows the particle distribution from a sample that was prepared by dipping the SEM stub directly onto the collection surface of the filter. Particles collected by the dust filter tend to be more rounded and grouped relative to those captured by the cyclone. Energy dispersive spectroscopy was used to provide a semiquantitative analysis to assist in the determination of particle chemistry and mineralogy. Results suggest that filter-collected particles are primarily quartz, CaCO3, Fe oxide minerals, and Al-Si clays.

Fig. 5.

Scanning electron microscope images showing filter collected particles magnified 10,000× (left) and 15,300 times (right).


Dust samples were further characterized by performing the previously described partial-digestion chemical analysis. Analytical results of samples C1, C2, C3, C5, C7, C8, C9, DU1, DU2, and the standard reference materials are presented in Table 1. Samples C4 and C6 were saved to be used as control samples when the CFD sampler is deployed during field studies. Because trace elements will be the focus of collection when this sampler is deployed for environmental contamination studies, Pb, Zn, Co, Ni, Cu, Cd, and As were measured. Cyclone-collected samples were singled out for this study because this is where the bulk of sample will be collected. Particles in this part of the sampler represent the size ranges that have been shown to be associated with large trace metal concentrations (Wixson et al., 1975).

View Full Table | Close Full ViewTable 1.

Results of partial-digestion analysis of cyclone particle samples, digestion blank, standard reference materials digested in triplicate, and test samples from two lead-contaminated roads.

Trace element
Sample designation Pb Zn Co Ni Cu Cd As Sample quantity Intake height
mg kg−1 g cm
C1 14.4 23.2 3.1 7.2 4.7 0.2 10.7 3 50.8
C2 14.2 24.8 3.4 7.8 4.7 0.3 12.6 6 50.8
C3 11.1 22.8 2.9 7.4 3.5 0.2 11.7 6 50.8
C5 10.8 23.9 3.0 7.6 3.8 0.2 11.7 11 50.8
C7 11.4 25.2 3.0 7.4 4.7 0.2 10.2 4 88.9
C8 11.9 25.4 3.1 7.7 4.3 0.2 11.8 3 88.9
C9 8.5 21.7 2.5 8.0 3.2 0.2 17.4 2 88.9
Mean (C1–C9) 11.8 23.9 3.0 7.59 4.13 0.21 12.3
Standard deviation (C1–C9) 2.04 1.37 0.27 0.27 0.63 0.04 2.38
DU1 4588 433 13.4 21.9 284 6.3 135 27 50.8
DU2 58.5 31.0 4.2 4.7 11.7 0.4 5.6 13 50.8
Digestion vessel blank 0.025 <0.02† <0.02† <0.02† <0.02† <0.02† <0.02†
SRM 2711–1 1419 354 7.9 17.7 122 51.9 116
SRM 2711–2 1431 358 7.9 17.7 125 52.9 116
SRM 2711–3 1422 358 7.9 17.7 125 53.3 117
SRM 2587–1 3367 304 8.0 22.2 146 2.0 17.4
SRM 2587–2 3491 317 8.6 23.7 153 2.3 17.9
SRM 2587–3 3399 316 8.2 21.9 161 2.0 17.0
Less than method determination level.

Variability between samples within the same group height and between samples of different group height was small and within analytical error. A paired t test was performed on groups 1 and 2 for each element and resulted in a determination of no significant difference. The greatest degree of variability between samples was for Pb. Three samples (C1, C2, and C9) fell slightly outside the 95% confidence interval (9.867–13.647) calculated for the mean Pb concentration of the seven samples. Analysis of standard reference materials in triplicate validates the repeatability of the analytical process at larger trace metal concentration—Pb concentration varied less than 0.5% from the mean concentration of the three replicates.

Samples DU1 and DU2 were collected from roadways in regions where Pb contamination has been identified, and it was expected that Pb and Zn concentrations would reflect the concentrations documented by others for soils in the area (Bolter et al., 1975; Bornstein, 1989; Rucker, 2001). The unsurfaced road where sample DU1 was collected is about 4.82 km north of an active secondary recycling smelter and within close proximity to former Pb-ore hauling roads. The road where DU2 was collected also is near the active smelter but to the south and east about 3.22 km. This road was not used for Pb-ore hauling. Background concentration of Pb in Missouri ranges from 20 to 24 mg kg−1 (Tidball, 1984). Samples DU1 and DU2 exceeded the background concentration for Pb by 191 and 2.4 times, respectively. It was expected that Pb would be elevated in these samples with respect to the background concentration but not to the degree of variability observed. These results demonstrate the value of the CFD sampler for collecting contaminated dusts. Furthermore, this sampler and the methods tested demonstrate the capability for collecting substantial quantities of fugitive road dusts in Missouri’s Pb-producing region. Approximately 15 to 20 g of fugitive dusts were collected from each road surface, with some retention of dust in the PVC intake tube and on the walls of the 10-L cyclone collection container.

Many factors will affect the results of data collection and will show deviations from the results of this analysis. For example, ambient soil moisture and relative humidity at the time of sampling will substantially affect the suspension of dust from a road surface. Therefore, the prevailing weather conditions are an important consideration in planning a field study. Sampling during the dry summer months should provide the best overall outcome, but winter months could be equally productive as long as there is a substantial period of dry weather before sampling.

The type of road material also will affect suspension. During several test runs on various unsurfaced roads throughout Missouri, it was observed that roads with a quarry stone base produced substantially more dust than roads composed solely of the local soil. These roads tended to produce a white suspension that persisted in the atmosphere for more than 10 s and in some cases longer than 1 min. Local soil roads that appear to be predominantly clay, and sand seemed to produce less dust than soil roads mixed with natural stone. The stone component of the road surface mixture tends to support fine particle suspension by increasing surface area friction associated with the moving vehicle tires. Road surfaces composed primarily of local soil produced a brown to reddish suspension. Samples DU1 and DU2 were characteristic of this suspension because both samples are brown to red in color.

Rural unsurfaced roads have reaches that are open to sunlight and covered by forest canopy. During the testing phase of this sampler, it was observed that traveling on shaded reaches of road produced less dust than reaches in full sunlight. Although there has been no quantification of sample volumes surrounding these conditions, it may be appropriate to increase the collection distance where shaded road surfaces predominate. This can usually be accommodated by doubling back on an already sampled reach if a shorter reach is within your study design. Dust suspension also is affected by road surface grading. This activity occurs frequently in rural areas as the road surface becomes rough or in rare cases impassible. Roads that have been recently graded produced the most dust if grading was done after an extended period of dry weather; otherwise, road grading can expose moist layers causing a reduction in dust generation. In many cases, field planning for these conditions can be adjusted by contacting the local county road maintenance office. They generally keep records on grading dates, quarry stone additions, and general road repairs. All is valuable information when interpreting the results of data collection.

The speed of the collection vehicle also can result in variable quantities of dust collected. The optimal collection speed was determined to be between 40 and 56 km h−1 in this study. Traveling speeds less than 40 km h−1 generally do not produce suspensions high enough off the road surface, and speeds in excess of 56 km h−1 cause shear velocities at the CDF intake nozzle, causing preferential selection of larger particle sizes. It was observed that local residents using unsurfaced roads on a daily basis generally do not exceed speeds of 64 km h−1 because speeds faster than this can be unsafe, damage vehicle finishes, and erode tires. This shows that the optimum sample collection speeds are well aligned with the representative activity in rural areas where these roads predominate.

Wind speed may also be a contributing factor to collection and should be taken into account. High wind speeds likely carry dust away from the CFD sampler before it passes in front of the intake. As a rule, wind speeds should not exceed 16 km h−1 on a given sampling day. Exceeding this reduces the quantity of sample collected and may affect a representative collection of particles sizes.

It is not an uncommon occurrence in rural areas for one or more vehicles to be following each other on unsurfaced roads; when this occurs there is substantially more dust put into suspension. In fact, during periods of dry weather exceeding more than a week, a following vehicle on an unsurfaced road may have visibility reduced to 3.05 or 4.57 m, with the dust suspension plume reaching heights exceeding 6.10 m above the road surface. Although not performed in this study, sample collection during these conditions should increase the sample quantity and provide a more representative sample of fugitive dust exposure to the local population. However, following another vehicle may introduce contamination not specifically associated with dust generated from the road surface. This must be taken into account when interpreting results, and it is suggested that, in areas of expected trace metal contamination where this scenario occurs, a second sampling of the road reach should be performed to confirm results.

The components of the sampler can play a role in the quantity and quality of the sample collected. The filter chosen for this analysis represents one that would be used by the local population to remove airborne contaminants, but future work will include adding filters of progressively smaller mesh size to better quantify submicrometer particle sizes. Also adding a series of commercially available impactors may help to maintain airflow while capturing submicrometer particles. The decision to add smaller mesh–size filters should be done with caution because this could reduce the airflow through the cyclone, rendering it less efficient. Chemical analysis of the dust captured on the filter used for this study is currently being investigated and was not a part of this reporting.


The development of the CFD sampler provides a new tool with improved capabilities for those conducting fugitive road dust studies worldwide. Under the controlled sampling conditions presented herein, sample collection volume was best at collection heights closer to the road surface; however, both collection scenarios will produce a substantial amount of sample in an 8- to 16-km reach of unsurfaced road. Adequate sample volume will support an extensive array of physical and chemical analyses while allowing for the preservation of reference samples for future investigations.

The SEM images, particle size, and chemical analysis performed during this study provide assurance that the CFD produces relatively consistent results. Measurement of more than 3400 particles from both the cyclone and filters indicates that particle sizes did not differ significantly between samples and sampling heights. Furthermore, chemical analysis for trace elements shows small variability in results. These results, while limited in robustness, provide a basic assurance that the CFD sampler will perform under conditions that minimize variability. The CFD sampler is a fugitive dust collection technology that is ready for use by the scientific community, and it is encouraged that investigators perform their own testing and validation as a quality assurance measure.


The authors thank MS&T ceramic engineering graduate students Derek King, Ben Lei, and Timothy Strack for assistance in preparing samples for imaging, particle size analysis, and EDS and the rural residents of Ozark County road 319 for not traveling the road while our samples were being collected. Their patience ensured that variable field conditions would remain at a minimum during the sample collection effort. This article is Geology & Geophysics Contribution 44 of Missouri University of Science and Technology.




Be the first to comment.

Please log in to post a comment.
*Society members, certified professionals, and authors are permitted to comment.

Facebook   Twitter