Aerosol Characterization of Abrasive Blasting Operations

Abrasive blasting in aircraft maintenance generates complex aerosols with significant health implications due to hazardous compounds like hexavalent chromium. Critical knowledge gaps persist regarding particle size distribution and composition for accurate exposure assessment. This study provides a detailed aerosol characterization of these environments using a multipronged approach: real-time particle size distribution analysis, scanning electron microscopy (SEM), energy-dispersive X-ray spectroscopy (EDS), and inductively coupled plasma (ICP) analysis. The results from analyzing bulk dust and inhalable (IOM) samples showed that the bulk dust particles span 20–244 µm, with chromium concentration peaking at 0.29% in the 40–60 µm fraction. EDS confirmed chromium and strontium (indicating aircraft primer) primarily on smaller particles. Metallic contaminants were detected in plastic media operations but absent with steel media. Critically, we challenge the validity of standard correction factors for historical 37-mm closed-face cassette (CFC) data, demonstrating that non-uniform hazardous particle distribution risks over or underestimating exposures. These findings necessitate size-resolved sampling and correction factors between CFC and IOM data. Future work with cascade impactors is recommended to resolve aerosol distribution profiles for evidence-based conversion factors. This research provides a foundation for safeguarding DOD maintenance-shop workers’ health and informs policy transitions to modern sampling.

KEY WORDS: abrasive blasting, aerosol characterization, hexavalent chromium, inhalable sampling, particle size distribution, occupational exposure

It is well established that particulate matter in the workplace poses health hazards to unprotected workers. Like gas and vapor contaminants, the concentration and composition of the particulate matter determine the severity of the hazard; however, 
the distribution of particle sizes also contributes to potential health effects (Mirowsky, 2013; Lippmann, 1999; Kan, 2018). Due to the increasingly narrow branches that make up the lower respiratory tract, only particles up to 4 µm can penetrate the alveolar region of the lungs (Phalen, 1999). Coarse particles ≤10 µm penetrate the thoracic region, and particles larger than 10 µm up to 100 µm penetrate primarily to the upper airways and nose (Phalen, 1999).

Particle size selective (PSS) samplers are designed to mimic the penetration of particles in the respiratory system by abiding a 50% cut point at 4 µm, 10 µm, and 100 µm for respirable, thoracic, and inhalable particles, respectively (Raabe, 1999; Vincent, 1999). It is policy for the American Conference of Governmental Industrial Hygienists (ACGIH) to include PSS guidance for new threshold limit values for aerosol contaminants.

ACGIH recently recommended sampling for hexavalent chromium using a sampler that adheres to the inhalable convention, since large chromium containing particles that deposit in the nose can cause painful ulcers (ACGIH, 2018). In the Department of Defense (DoD), hexavalent chromium exposure is most often found in corrosion control operations during painting and de-painting aircraft and their components (Carlton, 2003; Bennett, 2016; Bennett, 2018). While aerosol size and composition during painting operations have been established, little data are available on the characteristics of the dust generated during abrasive blasting procedures, with estimates of particle size ranging from 1-1,000 µm (Carlton, 1997; Sabty-Daily, 2005; Carlton, 2000). 

Aircraft maintenance and associated processes impact a large workforce and generate a high volume of samples. The bulk of prior exposure data was collected using 37-mm closed-face cassettes (CFCs). These cassettes, designed prior to PSS practice, do not adhere to any health-based size convention but instead have efficiencies that drop precipitously at larger particles sizes, with a reported efficiency of 7% at 38.7 µm compared to the 55% dictated by the inhalable convention (Witschger, 2004; Vincent, 1999). This is in stark contrast with the performance of the best-known inhalable sampler, the Institute of Medicine (IOM) sampler, which closely approximates the inhalable convention. To compare historical data with data obtained using an IOM, it has been proposed that industrial hygienists apply a correction factor to account for the under sampling inherent in the CFC’s design (Vincent, 2007). The issue with this assumption is that it is based on additional assumptions, including consistent composition of the particulate across the entire size distribution. As the Air Force considers adopting inhalable sampling for corrosion control operations, it would be prudent to verify some of the underlying assumptions associated with the standard correction factor so as not to overor under-estimate past exposures.

This study focused on characterizing the dust generated during abrasive blasting procedures using real time particle size distribution data, IOM samples, and a bulk dust grab sample. The samples used a multipronged analytical approach which included microscopic imaging, EDS, elemental composition through ICP, and real-time size distribution data based on optical particle counting.

2.1 Bulk Dust Analysis

A 44.4 g bulk dust sample was taken from an abrasive blasting booth and sieved (Precision e-forming, Cortland, NY) into five fractions: >126 µm, 80-125 µm, 60-79 µm, 40-59 µm, and 20-39 µm. Each section was weighed and transferred to plastic cylinders compatible with acid digestion. The samples were digested and run through an ICP for elemental chromium (i.e., all valence states) at a third-party lab.

Prior to digestion, representative subsamples from each fraction were mounted on SEM stubs with conductive carbon tape. As the blasting media was primarily composed of plastic, a nonconductive material, it was expected that a “charging” effect, where electrons load on the surface and fail to dispel causing areas of 
white, would be observed (Shaffner, 1970). The stubs were sputter-coated with 10 nm of gold to provide a continuous conductive surface to minimize charging while imaging. Imaging of each stub included four non-overlapping sections at a consistent magnification. Due to the polydispersity of the largest size fraction, 
three samples were mounted. From each of these stubs, three images were taken, and the nine images were processed together.

2.2 IOM Filter and Dust Analysis

Five IOM cassettes with polycarbonate filters were pre-weighed, calibrated to 2 LPM, and deployed during abrasive blasting operations (30 min. sampling). The samples were shipped back to the lab and the postweights recorded for the filter and the dust retained on top of the filter. 

The filters were returned with details surrounding the conditions under which they were collected (Table 1). Four of the five samples collected were from booths that used plastic beads as the abrasive material and one 
sample was taken in a booth with steel media.

While the bulk on top of the filters did not appreciably differ between the plastic media samples, the steel media produced a powdery gray solid which was in stark contrast to the plastic media bulk (Fig. 1).  The filters were sputter coated with 10 nm of gold and then imaged using EM-Tec F25 filter sample holder (Rave Scientific, Somerset, NJ) which allowed for processing the entire filter instead of an excised portion.

FIGURE 1 
IOM Filters and Bulk 
Media. A) Bulk Material from Steel Media; B) Bulk Material from Plastic Media

2.3 Image Analysis

All samples were analyzed on a JSM-IT500 (JEOL, Tokyo, Japan) scanning electron microscope. Bulk dust samples were loaded on 10 mm aluminum stubs with conductive carbon tape. Whole filters were mounted on a 25-mm disk holder (Rave Scientific, Somerset, NJ).

2.3.1 Bulk Dust

The researcher processed all SEM images with open-source software ImageJ (NIH) using the pre-packaged application Fiji (v. 1.52q). For images gathered from the bulk dust stubs, automatic particle processing was not possible. The auto-characterization function relies on the ability to turn any image into binary, which itself depends on strong contrast between the particle edge and the background. Despite the layer of gold coating on the stub, charging was an issue during imaging, which resulted in areas of pure white juxtaposed with areas of varying shades of grey. Despite the researcher’s bests efforts to manually set thresholds and the use of auto-thresholding techniques available in the ImageJ package, it was not possible to capture crisp outlines for most particles.

Instead of turning each image to binary, the researcher manually outlined each particle using the freehand selection tool, used ImageJ’s measurement tool, then filled in each particle with white to prevent the accidental characterization of the same particle twice.

2.3.2 Filters

Due to the more homogenous surface of the small particles on the filters, the 10 nm layer of gold greatly reduced charging effects. This reduction in charging allowed for higher contrast images which, in turn, allowed for automated image processing using the ImageJ as described above. Due to the 0.8 µm pores 
of the polycarbonate, an artificial lower size cut off of 0.5 µm2 was used to prevent ImageJ’s automatic particle identification feature from incorrectly classifying pores as particles.

The grey particulate on top of Filter D was dense enough that individually resolving particles was impossible, therefore the filter was excluded from imaging analysis. Filter A was damaged during specimen mounting, leading to a wrinkled surface that proved impossible to image and its exclusion from the size distribution analysis. It was possible to conduct EDS analysis for all five filters.

2.4 EDS Analysis 

All samples were analyzed on a JSM-IT500 (JEOL, Tokyo, Japan) SEM equipped with an X-maxN 80 EDS (Oxford Instruments, Concord, MA). Each imaged stub was also analyzed for elemental composition. Point analysis, which was used as the gold coating, dominated the response for mapping. A minimum of 100 random points spread over the surface of the 
stub were analyzed for each size fraction or IOM sample. While EDS was performed in the same area as the imaging, all imaging locations were randomly selected. The microscope included a feature where previously visited spots were marked to ensure no overlap in images or EDS. The percent of measured points containing an element were determined by 
dividing all points containing the element of interest by the total number of particles measured for the filter.

2.5 Size Distribution

Using the projected area estimated from the ImageJ processed images, the physical diameter of the particles was estimated using the equation below (Eq. 1) (Fan, 1998) where A is the area of the particle.

EQUATION 1
Particle Physical Diameter

Once the physical diameter was calculated, histograms for each sieve section or filter were generated.

For sieved fractions, bins were generated in five micrometer intervals until counts dropped off. For filter samples, bin intervals varied from 0.8-10 µm. Bins were then normalized by their range, then these normalized values were converted to frequency/µm by dividing by the total number of particles counted. The count median diameter (CMD) was calculated by first determining the cumulative fraction of particles by bin. The 50th percentile was interpolated from the nearest fractions. The geometric standard deviation (GSD) for each fraction was determined using Eq. 12 (Hinds, 1999).

EQUATION 2

While charging was an issue with the bulk dust samples, micrographs showed no areas of brightness indicative of metals (Fig. 2A). 

Combined with EDS results, this confirms bulk dust primarily comprised of plastic blasting media rather than metal-containing paint chips. In contrast, the particles on the IOM filters, when viewed with the SEM, did show areas of brightness that when analyzed with EDS showed evidence of metal compounds (Fig. 2B). The brightness of these smaller, metallic particles is due to their greater ability to backscatter electrons as they have a higher atomic number. The distribution for each bulk dust size fraction is shown in Fig. 3. Significant overlap exists between all size fractions, as sieving relied on separation based on physical diameter rather than aerodynamic diameter.

FIGURE 3

Size Distribution for Each Sieve Fraction.

 The CMD and GSD for the bulk dust were determined by fraction (Table 2). All GSDs were below two, indicating a relatively tight distribution, despite the heterogeneous appearance of the particles in each fraction. While the shape of the largest size fraction (>126µm) was unusual, when the log transformed frequency/µm values were evaluated using the Shapiro-Wilkes test, the results indicated a lognormal distribution. This finding validates the calculation method for finding the CMD and GSD.

The sieved fraction of bulk dust sent off for percent chromium content through ICP analysis showed an increase in chromium as particle size decreased (Table 3). Particles were sieved, placed in flat-bottomed tubes, then weighed. 

EDS identified 25 elements in total among the sieved fractions (Fig. 4A). Results from EDS of the primary compound of concern found in aircraft primer—strontium chromate—showed that the highest percent of chromium was associated with the highest percent of strontium (Fig. 4B). The percentages shown correspond to the number of particles sampled that had a positive match for the element normalized by the total number of particles sampled for sieved fraction.

FIGURE 4 
EDS Results of Bulk Dust: A) All Identified Elements; B) Elements Known to Exist in Aircraft Primer.

The distributions for filters C, F, and G are shown in Fig. 5. All distributions were truncated due to the cutoff introduced through image analysis. Due to the incomplete distribution, the CMD and GSD were not calculated.

The CMD and GSD are predicated on knowing the median particle size. Without seeing the full distribution, which should resemble a skewed normal curve, it is not possible to assert where the median lies.

FIGURE 5 
Size Distribution for Filters C, F, and G (Plastic Blast Media Depainting Operations).

As with the bulk samples, EDS identified 25 elements in total among the five filters (Fig. 6A). The percent chromium identified did not correlate with the strontium (Fig. 6B). It is of interest to note no chromium or strontium was found on Filter D, the only IOM sampled from a blasting booth using steel media. It seems likely the part being blasted did not contain a layer of primer.

A key limitation of this study was reliance on a single bulk sample, restricting generalizability. At present, it is not appropriate to extrapolate the composition and distribution profile across all blasting booths or processes; however, the results obtained do indicate that chromium likely adhered to large particles and those large particles have been observed in the IOM capsule. While a correction factor could not be obtained from the single bulk sample, the ICP analysis indicates that elemental chromium concentration is not linear across all particle sizes nor proportional in a predictable way.

The impact of the limitation introduced during image processing for determining the particle size distribution for those particles adhered to the IOM filter was evident. A more appropriate technique would be to separate particles via a cascade impactor to remove the very large particles, then follow the impactor outlet with an electrometer in order to count and deposit the small particles on a TEM filter. The increased resolution and negligible pore size of the TEM filter would allow for image processing of submicron particles without relying on an artificial area cutoff.

Only a single filter was available from a steel blast media process, and the particulate generated was significantly different in size, shape, and composition from the particulate in plastic blast media samples. While the lack of chromium and strontium in the EDS analysis suggests the particular operation sampled did not involve primer, if other steel blasting operations do involve de-painting, the plastic media distributions and composition would not be generalizable.

All methods used to analyze for chromium did not evaluate valence state, only presence of elemental chromium. While in principle it was more conservative to use this approach, future research should include a validation that hexavalent chromium is the only valence present.

Future research that includes a larger study sample size and impactor samples would improve understanding surrounding abrasive blasting processes and build on the findings presented here.

The size distribution of particulate matters when it comes to potential health effects. For hazardous compounds with detrimental impacts to the nose and upper airways, aerosol sampling using an inhalable sampler is appropriate to measure exposure. The findings from this project indicate abrasive blasting aerosols are heterogeneous and complex. A simple correction factor to convert historical CFC values is likely inappropriate and would overestimate prior exposures.

For an organization as large as the Department of Defense, the decision to move to inhalable sampling represents an enormous commitment of time and resources both for samplers and analysts. To avoid losing decades of exposure data when making this change in sampling, efforts should be made to equate prior CFC data with what an IOM would have collected. The impulse to use a convenient conversion factor should be scrutinized, however, as it could unreasonably bias estimates. The voluntary switch to the more conservative TLV represents an admirable intent to protect workers. It would be a disservice to the spirit of the endeavor to overcorrect past samples if a reasonable alternative exists. Using the methods outlined in this study with the addition of a few key pieces of equipment, a larger study could answer the distribution question and provide an evidence-based correction factor.

ACGIH, Chromium and Inorganic Compounds in TLV Documentation, 2018.

Bennett, J.S., Marlow, D.A., Nourian, F., Breay, J., and Hammond, D., Hexavalent Chromium and Isocyanate Exposures During Military Aircraft Painting Under Crossflow Ventilation, J. of Occ. and Environ. Hygiene, vol. 13(5), pp. 365-371, 2016.

Bennett, J.S., Marlow, D., Nourian, F., Breay, J., Feng, A., and Methner, M., Effect of Ventilation Velocity on Hexavalent Chromium and Isocyanate Exposures in Aircraft Paint Spraying, J. of Occ. and Environ. Hygiene, vol. 15(3), pp. 167-181, 2018.

Carlton, G.N. and Flynn, M.R., Influence of Spray Painting Parameters on Breathing Zone Particle Size Distributions, Applied Occ. and Environ. Hygiene, vol. 12(11), pp 744-750, 1997.

Carlton, G.N. and England, E.C., Assessing Worker Exposures During Abrasive Blasting: Industrial Hygiene Field Guidance for Bioenvironmental Engineers, United States Air Force IERA, Report IERA-RS-BRTR-2000-0001, 2000.

Fan, L. and Zhu, C., Size and Properties of Particles, In Principles of Gas-Solid Flows (Cambridge Series in Chemical Engineering), Cambridge: Cambridge University Press, pp. 3-45, 1998. DOI:10.1017/CBO9780511530142.002

Hinds, W.C., Aerosol Technology: Properties, Behavior, and Measurement of Airborne Particles, 2nd Ed., New York, NY: John Wiley & Sons, 1999.

Igathinathane, C., Pordesimo, L.O., Columbus, E.P., Batchelor, W.D., and Sokhansanj, S., Sieveless Particle 
Size Distribution Analysis Of Particulate Materials Through Computer Vision, Comp. and Elect. in Agric., vol. 66, pp. 147–158, 2009.

Phalen, R.F., Airway Anatomy and Physiology, Particle Size-Selective Sampling for Particulate Air Contaminants, Cincinnati, OH: ACGIH, pp. 29-49, 1999.

Raabe, O.G. and Stuart B.O., Sampling Criteria for the Thoracic and Respirable Fractions, Particle SizeSelective Sampling for Particulate Air Contaminant, Cincinnati, OH: ACGIH, pp. 73-95, 1999

Sabty-Daily, R.S., Harris, P.A., Hinds, W.C., and Froines, J.R., Size Distribution and Speciation of Chromium in Paint Spray Aerosol at an Aerospace Facility, Ann. Occup. Hyg., vol. 49(1), pp. pp 47-59, 2005.

Shaffner, T.J. and Ved, R.D.V., ‘Charing’ Effects in the Scanning Electron Microscope, J. of Physics E: Scientific Instruments, vol. 4(9), 1970.

Takashimizu, Y. and Iiyoshi, M., New Parameter of Roundness R: Circularity Corrected by Aspect Ratio, Prog. in Earth and Planetary Sci., vol. 3(2), 2016.

Vincent, JH., Aerosol Samplers in Workplaces in Aerosol Sampling, Hoboken, NJ: John Wiley & Sons Ltd, pp. 544-545, 2007.

Vincent, J.H., Sampling Criteria for the Inhalable Fraction, Particle Size-Selective Sampling for Particulate Air Contaminants, Cincinnati, OH: ACGIH, pp. 51-72, 1999

Witschger, O., Grinshpun, S.A., Fauvel, S., and Basso, G., Performance of Personal Inhalable Aerosol Samplers in Very Slowly Moving Air When Facing the Aerosol Source, Ann. Occup. Hyg., vol. 48(4), pp.351-368, 2004.