Carbon Nanotube and Nanofiber Exposure Assessments: An Analysis of 14 Site Visits

Abstract

Recent evidence has suggested the potential for wide-ranging health effects that could result from exposure to carbon nanotubes (CNT) and carbon nanofibers (CNF). In response, the National Institute for Occupational Safety and Health (NIOSH) set a recommended exposure limit (REL) for CNT and CNF: 1 µg m⁻³ as an 8-h time weighted average (TWA) of elemental carbon (EC) for the respirable size fraction. The purpose of this study was to conduct an industrywide exposure assessment among US CNT and CNF manufacturers and users. Fourteen total sites were visited to assess exposures to CNT (13 sites) and CNF (1 site). Personal breathing zone (PBZ) and area samples were collected for both the inhalable and respirable mass concentration of EC, using NIOSH Method 5040. Inhalable PBZ samples were collected at nine sites while at the remaining five sites both respirable and inhalable PBZ samples were collected side-by-side. Transmission electron microscopy (TEM) PBZ and area samples were also collected at the inhalable size fraction and analyzed to quantify and size CNT and CNF agglomerate and fibrous exposures. Respirable EC PBZ concentrations ranged from 0.02 to 2.94 µg m⁻³ with a geometric mean (GM) of 0.34 µg m⁻³ and an 8-h TWA of 0.16 µg m⁻³. PBZ samples at the inhalable size fraction for EC ranged from 0.01 to 79.57 µg m⁻³ with a GM of 1.21 µg m⁻³. PBZ samples analyzed by TEM showed concentrations ranging from 0.0001 to 1.613 CNT or CNF-structures per cm³ with a GM of 0.008 and an 8-h TWA concentration of 0.003. The most common CNT structure sizes were found to be larger agglomerates in the 2–5 µm range as well as agglomerates >5 µm. A statistically significant correlation was observed between the inhalable samples for the mass of EC and structure counts by TEM (Spearman ρ = 0.39, P < 0.0001). Overall, EC PBZ and area TWA samples were below the NIOSH REL (96% were <1 μg m⁻³ at the respirable size fraction), while 30% of the inhalable PBZ EC samples were found to be >1 μg m⁻³. Until more information is known about health effects associated with larger agglomerates, it seems prudent to assess worker exposure to airborne CNT and CNF materials by monitoring EC at both the respirable and inhalable size fractions. Concurrent TEM samples should be collected to confirm the presence of CNT and CNF.

INTRODUCTION

Carbon nanotubes (CNT) and carbon nanofibers (CNF) are cylindrical-shaped materials, made of carbon atoms, having a diameter measuring in the nanometer scale with varying structural characteristics including differing lengths and number of walls such as single-walled (SW) and multi-walled (MW) CNT. These characteristics can be altered to modify the materials’ thermal and electrical conductivity, strength, and durability for various applications including advanced composite materials, thermoplastics, batteries, fuel cells, and electronics.

To date, global production of these materials has increased, but quantities have not reached the wide-spread use predicted less than a decade ago (Aitken et al., 2006). It is challenging to enumerate the US workforce handling these materials, but indications point toward steady growth as current advanced material uses shift from pilot plants and development phases to mainstream production and commercialization (Shapira et al., 2011; Schubauer-Berigan et al., 2011; Engeman et al., 2013). With this gradual shift to full scale production and commercialization, the future number of workers in the industry is expected to rise (Invernizzi, 2011).

Concurrently, there have been many animal toxicological studies of certain types of SWCNT, MWCNT, and CNF, suggesting the potential for human health hazards, including pulmonary inflammation, granulomas, interstitial fibrosis, systemic effects, and promotion of tumorigenesis (Lam et al., 2006; Mercer et al., 2008; Mitchell et al., 2009; Porter et al., 2010; Reddy et al., 2010; Erdely et al., 2011, 2013; Kisin et al., 2011; Sargent et al., 2014). In the wake of this evidence, NIOSH published a Current Intelligence Bulletin (CIB) on CNT and CNF with a recommended exposure limit (REL) of 1 µg m⁻³ of elemental carbon (EC), as a respirable particle mass 8-h-time weighted average (TWA) concentration (NIOSH, 2013), designed to protect against pulmonary fibrosis. Due to the lack of epidemiologic data, little is directly known about the potential adverse effects to humans from workplace exposures to CNT and CNF. However, evidence of adverse effects may not yet be apparent due to a long latency period for health endpoints to manifest.

Here, we describe an industrywide exposure assessment study conducted at 14 sites throughout the USA. The sites visited were considered primary and secondary (downstream) manufacturers of CNT or CNF and could be categorized into four distinct industries:

• 1. Primary manufacturers (n = 4; Sites A, B, C, and M)—facilities that produced MWCNT or SWCNT and sold their products to researchers or industry. Most facilities also performed small amounts of development work on their own materials.

• 2. Hybrid producers/users (n = 2; Sites G and L)—facilities that produced MWCNT and subsequently incorporated them into a product.

• 3. Secondary manufacturers (SM) in the electronics industries (n = 4; Sites D, H, K, and N)—facilities that purchased CNT, most often SWCNT, in an aqueous or solvent-based suspension or slurry for use in flexible panel circuits or lighting and other microelectronic devices.

• 4. Secondary manufacturers (SM) in the composites/thermoplastics industries (n = 4; Sites E, F, I, and J)—three facilities solely used MWCNT while one primarily used CNF for the production of various advanced composites for enhanced electro-magnetic shielding, electrical conductivity, strength, and weight reduction.

The main objective in this study was to quantify and describe personal breathing zone (PBZ) exposures to workers producing or handling CNT and CNF within a representative group of industries. Results from the first six sites (A–F) were previously published (Dahm et al., 2012) and are included in the analyses with eight additional sites.

METHODS

Sampling strategy

Site visits occurred between May 2010 and September 2012. Site and process descriptions, which include synthesis methods, volumes handled, and numbers of employees potentially exposed, for sites A–N are provided within the supplemental materials. PBZ samplers were attached to the lapel of the worker while area samplers were positioned within three feet of the specific process, where feasible. Direct reading instruments were co-located to measure particle number, active surface area, and respirable mass. These instruments were used only in area sampling and lacked the specificity required for exposure assessment; therefore, results are not included herein. Application and limitations of direct reading instruments for workplace assessment have been described elsewhere (Evans et al., 2010; Dahm et al., 2013).

PBZ and area samples were collected for analysis of EC and by transmission electron microscopy (TEM). Both sample types were co-located, with sample inlets positioned together, in order to sample the same air space. Over time, the sampling strategy for EC PBZ samples evolved, as methodologies advanced (Table 1). At sites A–K (excluding Sites G and I), only inhalable EC PBZ samples were collected, using methods described previously (Birch et al., 2011; Dahm et al., 2012). For Sites L–N and G, both inhalable and respirable EC PBZ samples were collected, with sample inlets positioned close together. Area and PBZ inhalable TEM samples were collected consistently for all 14 sites.

Table 1.

Sampling strategies used for PBZ and area samples by site.

		EC						TEM
		Sites A–F;H, J, K		Site I		Sites G; L–N		All sites
		Respirable	Inhalable	Respirable	Inhalable	Respirable	Inhalable	Inhalable
PBZ samples	Size fraction collected (Y/N)	N	Y	Y	N	Y	Y	Y
	Cassette size (mm)	—	25	37	—	25	25	25
	Flow rate (lpm)	—	5–6	4.2	—	4.2	5–6	4–5
Area samples	Size fraction collected (Y/N)	Y	Y	Y	Y	Y	Y	Y
	Cassette size (mm)	37	25	37	25	37 & 25	25	25
	Flow rate (lpm)	4.2	5–6	4.2	5–6	4.2	5–6	4–5

EC analysis

The airborne mass concentration of EC was measured using the NIOSH Manual of Analytical Methods (NMAM) 5040 (NIOSH, 2006a). The method is based on a thermal-optical analysis technique for organic and elemental carbon (OC and EC). Twenty-five-mm cassettes and quartz fiber filters (QFFs) were preferentially used over 37-mm cassettes and QFFs, as well as other commercial inhalable samplers, in order to improve the limits of detection and quantitation (LOD/LOQ). Bulk samples of the CNT and CNF materials, where available, were analyzed to obtain their thermal profiles. CNT and CNF aerosols contain micrometer-size agglomerates and the OC–EC split may need adjustment to optimize the CNT and CNF EC speciation, particularly at low loadings. A manual split was therefore assigned, based on results for the bulk and background samples. Manual splits were also used for media blanks to estimate the LOD/LOQ (~0.3 µg cm⁻² filter deposit) (NIOSH, 2013). All materials in this study were fully oxidized during the oxidative mode of the 5040 analysis. Depending on filter load and material type, some materials may require an extended period for oxidation.

Sites A–F, H, J, K

PBZ and area samples were collected using open-faced 25-mm cassettes and QFFs (SKC, Inc., Eighty Four, PA, USA) to estimate the inhalable size fraction. The size-selective characteristics of the open-faced 25-mm cassettes have not yet been fully evaluated, but are anticipated to exhibit a similar sampling efficiency to a 37-mm open-faced cassette. Considering wind-tunnel data described by Bartley (1998), the 37-mm cassette collection efficiency and the inhalable sampling curve are similar for particles with aerodynamic diameters <20 µm. Therefore, samples collected with the 25-mm open-face cassette may be considered to approximate the inhalable size fraction for fine, well dispersed powders encountered in this study. Leland Legacy™ pumps (SKC, Inc.) were used at 6 liters per minute (lpm) to collect inhalable samples, while flow rates for Sites H, J, and K were reduced to 5.5 lpm to decrease pump faults during longer sampling durations. Area samples were collected for the respirable size fraction using 37-mm cassettes and QFFs (SKC, Inc.) with GK 2.69 BGI cyclones (BGI, Inc., Waltham, MA, USA) and AirChek 224-PCXR8 pumps (SKC, Inc.) operating at a flow rate of 4.2 lpm.

Respirable PBZ samples were not collected during site visits conducted in 2010 and early 2011. At that time, a NIOSH REL for CNT and CNF was proposed in a draft CIB, but few NIOSH field studies had found detectable EC concentrations. Therefore, based on earlier surveys, increased air volumes and 25-mm filter cassettes were employed to lower the LOD, as recommended in NIOSH 5040. For respirable aerosol collection, a high-flow respirable cyclone with an adaptor (cyclone base) designed to a fit 25-mm cassette, in order to achieve a lower detection limit, was under development and subsequently used in later site visits [respirable GK 2.69 BGI cyclones using a customized adapter for 25-mm cassettes (BGI, Inc.; catalog number 3503]. The sample collection conditions for these later site visits (Sites G, I, L–N) are described in Table 1.

Electron microscopy analysis

PBZ and area samples were collected at all sites using open-faced 25-mm cassettes to approximate the inhalable size fraction with mixed cellulose ester filters (0.8 µm pore size; SKC, Inc.) and Leland Legacy™ or Airchek XR5000 pumps operating at ~5 lpm. The samples were subsequently analyzed on a JEOL2100F transmission electron microscope (JEOL USA, Inc., Peabody, MA, USA) using a modified NMAM 7402, asbestos by TEM (NIOSH, 2006b). Modifications to NMAM 7402 relate to counting CNT and CNF particles, which were observed mainly as agglomerated structures rather than individual fibers. Modifications consisted of eliminating (i) the steps required for asbestos identification and (ii) the requirement that counted particles have diameters >0.25 µm and meet the traditional definition of a fiber (aspect ratio ≥ 3:1, longer than 5 µm). These samples provided visual evidence of the presence of airborne CNT and CNF as well as the general size, shape, and degree of agglomeration of the particles.

Three 3-mm copper TEM grids prepared from each sample were first examined at low magnification to determine the filter loading and preparation quality. Multiple grid openings from each TEM grid, 40 openings total, were then examined at high magnification and any particles containing CNT or CNF were counted as CNT or CNF ‘structures’, which ranged from single-fiber CNT or CNF (Fig. 1A) to various sized agglomerates comprised of many single CNT or CNF fibers (Fig. 1B,C,D). This differs from how asbestos is counted. All structures, agglomerated or single fibers, were counted as one ‘CNT or CNF structure’. Based on the number of CNT or CNF structures counted, and the collected air volume, CNT or CNF structures per cm³ was calculated.

Figure 1.

TEM images of single fibers and CNT agglomerates from personal breathing zone samples. (A) MWCNT single fiber from a SM-composites/thermoplastics site. (B) SWCNT sgglomerate from an aqueous mixture at a SM-electronics site. (C) MWCNT agglomerate from a SM-composites/thermoplastics site. (D) MWCNT agglomerate from a primary manufacturer.

CNT structures were also categorized by size-bins based on a visual estimate of structure size and degree of agglomeration. Most of the agglomerated CNT particles were roughly spherical in shape (i.e. the longest dimension was <2 times the length of the orthogonal dimension) and therefore the maximum crosswise dimension was used to categorize the structure’s size. The counted CNT structures were placed into six discrete size-bins, based on health relevant size fractions as well as observations from previous work (Dahm et al., 2012). The size-bin ranges included a single fiber bin, CNT structure agglomerates with diameters <1 µm, agglomerates with diameters between 1 and 2 µm, agglomerates between 2 and 5 µm, agglomerates between 5 and 10 µm, and agglomerates with diameters >10 µm. Size categorization was completed for the latter sites (G–N), none of which handled CNF.

Background sampling

Outdoor or indoor background measurements for the respirable and inhalable mass concentration of EC were collected on each day of sampling to assess the potential for interference by anthropogenic sources of EC. These sources could include diesel exhaust, emissions from coal or oil-fired power plants, and the seasonal burning of biomass (Magliano et al., 1999; Christoforou et al., 2000; Streets et al., 2001; Schauer, 2003). Locations for background sampling were selected based on professional judgment and knowledge of the facility.

Statistical analysis

Elemental carbon

Statistical analysis of all outcome variables was performed using SAS version 9.3 (Cary, NC, USA). Concentrations for any samples found below the LOD for NMAM 5040 (n = 39) were calculated by using the LOD of that sample set divided by two. This method was preferred over the maximum likelihood estimate due to the limited number of samples. The concentration was then calculated using the air volume for the specific non-detectable (ND) sample.

The data were background-corrected to adjust for anthropogenic sources of EC by subtracting the site specific background EC concentration, by type of sample (respirable or inhalable). Sites that used clean room or similar atmospheres, where background concentrations were assumed to be zero, were not background-corrected. Sites with missing background EC values were background-corrected by imputing the average value from sites where backgrounds were collected. In instances where a negative background-corrected concentration was found (background concentration > sample concentration), 1/2 the lowest background-adjusted value from the specific sample site and sample type (respirable or inhalable) that was greater than zero was substituted. Each imputed data point was reviewed on a site-by-site basis for validation.

Geometric means (GM) and 8-h TWAs were then calculated using these background-corrected concentrations. When the sampling duration was <480 minutes, zero exposure was assumed for the remaining time in the 8-h TWA value. Generally, these situations included instances where CNT or CNF operations ended after <480min, with no additional work done in the immediate area. In these situations, the employee typically spent the remainder of the working day in an office environment. Short-duration samples (<20min) were excluded in the final analysis. PBZ and area GM concentrations subdivided by industry and material were compared using ANOVA with a Tukey multiple comparison adjustment on the log-transformed concentrations.

TEM

ND TEM samples (n = 34) were those with counts below the LOD, defined as 1 CNT or CNF structure per filter. The concentration per cm³ was then calculated by dividing this LOD by two and then using the air volume for the specific ND sample. Samples that were overloaded (n = 4) with material were visually verified to contain heavy loadings of CNT or CNF and EC by using energy dispersive X-ray spectroscopy. The overloaded samples could not be quantified and, once verified to contain CNT or CNF by microscopy and EC, concentration was estimated by assuming it was 50% higher than the largest CNT or CNF structure concentration found at that specific site. Eight-hour TWAs were calculated in a similar fashion as for the EC samples. The statistical correlations between EC and TEM concentrations in paired samples were examined using the non-parametric Spearman rank correlation coefficient, to accommodate the non-normality of the exposure distributions. Statistical significance was evaluated based on P = 0.05.

RESULTS

Facility demographics

Table 2 presents information on company demographics and the characteristics of CNT and CNF materials used by different industries. Most sites used MWCNT with the general exception of companies in the electronics industries and primary producers who specifically manufactured SWCNT. Only one facility in the composites/thermoplastics industry used both MWCNT and CNF, using larger quantities of CNF for longer durations. The average number of total employees, company-wide, ranged from 13 to 7528, with primary manufacturers having the fewest and hybrid producers/users having the most employees. However, company size did not necessarily reflect the number of employees potentially exposed to CNT and CNF: hybrid producers/users had the highest number of employees potentially exposed while SM-composites/thermoplastics sites had the lowest. Three of the four industry types handled kg quantities in a single day, with the exception of the SM-electronics sites. The maximum quantity observed in use at a SM-electronics site was 300g day⁻¹; however, this was mostly as an aqueous or slurry suspension.

Table 2.

Facility demographics and characteristics by type of industry.

	Primary manufacturer	Hybrid producer/ user	Secondary manufacturer— electronics	Secondary manufacturer— composites/ plastics
No. of facilities	4	2	4	4
Average no. of employees per company	13	7528	166	1180
Average no. of employees potentially exposed	10	32	17	8
Types of material produced/ handled	SWCNT; MWCNT	MWCNT	SWCNT	MWCNT; CNF
Maximum quantities handled per day (kg)	1.5	1	0.03	2.6
Average reported CNT diameter (nm)	1; 15	50	1.3	54; 140
Average reported CNT length (μm)	500; 70	250	250	279; 100

Overall exposures

Overall, 207 samples were collected for EC analysis among the 14 sites: 72 were PBZ samples, 102 were area samples, and 33 were background samples. As seen in Table 3, the background-corrected, respirable EC PBZ exposure concentrations ranged from 0.02 to 2.94 µg m⁻³ with a GM of 0.34 µg m⁻³, an 8-h TWA GM of 0.16 µg m⁻³ and an arithmetic mean (AM) of 0.73 µg m⁻³ [standard deviation (SD), 0.79]. Two TWA PBZ samples were above the NIOSH REL of 1 µg m⁻³. Area respirable samples ranged from 0.02 to 6.94 µg m⁻³ with a single TWA area sample above the NIOSH REL. At the inhalable size fraction, EC concentrations in PBZ samples ranged from 0.01 to 79.57 µg m⁻³, with a GM of 1.21 µg m⁻³ and AM of 6.47 µg m⁻³ (SD, 13.85), and area samples ranged from 0.01 to 210.22 µg m⁻³. The area sample GMs for both the respirable and inhalable size fractions were identical: 0.31 µg m⁻³ while the AMs were 0.92 µg m⁻³ (SD, 1.50) and 4.83 µg m⁻³ (SD, 28.78), respectively. The GM for the EC background samples were 0.36 µg m⁻³ (respirable) and 0.56 µg m⁻³ (inhalable).

Table 3.

Overall EC and TEM exposures by type and size fraction sampled.

	Sample	EC							TEM
		n	GM (μg m−3)	GSD	Min.	Max.	8-h TWA GM (μg m−3)	8-h TWA GSD	n	GM (s cm−3)	GSD	Min.	Max.	8-h TWA GM (s cm−3)	8-h TWA GSD
All sites combined (n = 14)	PBZ Resp.	25	0.34	4.43	0.02	2.94	0.16	3.49	—	—	—	—	—	—	—
	PBZ Inhal.	47	1.21	8.90	0.01	79.57	0.38	9.88	51	0.008	11.59	0.0001	1.613	0.003	13.68
	AS Resp.	49	0.31	4.57	0.02	6.94	0.10	4.02	—	—	—	—	—	—	—
	AS Inhal.	53	0.31	7.52	0.01	210.22	0.10	8.19	52	0.004	7.26	0.0001	0.631	0.001	8.39

All EC values have been background corrected. AS, area sample; Resp., respirable size fraction; Inhal., inhalable size fraction; s cm⁻³, CNT or CNF structures per cubic centimeter.

PBZ samples analyzed by TEM had concentrations ranging from 0.0001 to 1.613 CNT or CNF-structures per cm³, while the area samples ranged from 0.0001 to 0.631 CNT or CNF-structures per cm³. PBZ samples had a GM concentration of 0.008 CNT or CNF-structures per cm³, an 8-h TWA concentration of 0.003 CNT/CNF-structures per cm³, and AM of 0.084 CNT or CNF-structures per cm³ (SD, 0.24). Area samples had a GM concentration of 0.004 CNT or CNF-structures per cm³ and AM of 0.029 CNT or CNF-structures per cm³ (SD, 0.10). Figure 2 displays the distribution of PBZ samples for respirable and inhalable EC and TEM PBZ concentrations.

Figure 2.

EC and TEM structure concentration distributions for PBZ samples.

Exposures by industry

Table 4 presents the data subdivided by industry. The highest respirable PBZ GM exposures were seen in the SM-electronics sites (0.93 µg m⁻³), followed by SM-composites/thermoplastics sites (0.70 µg m⁻³), and hybrid sites (0.68 µg m⁻³), which were found not to differ statistically (Table 5). The lowest respirable PBZ GM was seen in the primary manufacturers (0.05 µg m⁻³). Inhalable PBZ GMs were highest in the hybrid (13.39 µg m⁻³) and SM-composites/thermoplastics sites (5.47 µg m⁻³), but they did not differ statistically. However, a statistically significant difference was seen between inhalable PBZ GMs at SM-electronics (0.52 µg m⁻³) and primary manufacturers (0.19 µg m⁻³). Based on 8-h TWA exposures, the industries with the highest PBZ EC exposures at both the respirable and inhalable size fractions were the hybrid sites (0.41 and 7.93 µg m⁻³), followed by the SM-composites/thermoplastics (0.19 and 0.86 µg m⁻³), SM-electronics (0.18 and 0.12 µg m⁻³), and primary manufacturers (0.04 and 0.11 µg m⁻³), respectively.

Table 4.

EC and TEM exposures by industry.

Industry		Sample		EC										TEM
				n	GM (μg m−3)	GSD	Min.	Max.	8-h TWA GM (μg m−3)	8-h TWA GSD	n	GM (s cm−3)		GSD	Min.	Max.	8-h TWA GM (s cm−3)	8-h TWA GSD
Primary manufacturer (n = 4)		PBZ Resp.		7	0.05	2.29	0.02	0.19	0.04	2.29	—	—		—	—	—	—	—
		PBZ Inhal.		11	0.19	7.78	0.01	1.95	0.11	5.11	11	0.011		2.77	0.003	0.090	0.007	2.34
		AS Resp.		11	0.26	2.77	0.06	1.68	0.16	2.36	—	—		—	—	—	—	—
		AS Inhal.		13	0.06	8.85	0.01	4.29	0.03	5.46	13	0.005		4.61	0.0001	0.034	0.003	4.49
Hybrid producer/ user (n = 2)		PBZ Resp.		9	0.68	2.38	0.15	2.40	0.41	3.05	—	—		—	—	—	—	—
		PBZ Inhal.		9	13.39	3.45	1.59	79.57	7.93	2.34	9	0.042		2.88	0.010	0.169	0.026	2.49
		AS Resp.		12	0.22	3.46	0.05	3.02	0.15	3.32	—	—		—	—	—	—	—
		AS Inhal.		12	2.81	5.18	0.17	210.22	1.96	3.66	12	0.008		8.66	0.0001	0.631	0.005	8.06
Secondary manufacturer— electronics (n = 4)		PBZ Resp.		5	0.93	1.73	0.50	1.94	0.18	1.52	—	—		—	—	—	—	—
		PBZ Inhal.		18	0.52	4.56	0.05	5.21	0.12	6.91	18	0.001		6.30	0.0001	0.214	0.0002	5.65
		AS Resp.		14	0.27	7.49	0.02	6.94	0.04	5.43	—	—		—	—	—	—	—
		AS Inhal.		14	0.19	3.27	0.04	1.45	0.03	4.20	14	0.001		3.29	0.0001	0.015	0.0001	2.12
Secondary manufacturer— composites/ thermoplastics (n = 4)		PBZ Resp.		4	0.70	2.76	0.34	2.94	0.19	2.04	—	—		—	—	—	—	—
		PBZ Inhal.		9	5.47	2.01	1.79	21.23	0.86	3.72	13	0.041		12.26	0.0005	1.613	0.008	13.47
		AS Resp.		12	0.60	4.33	0.05	3.89	0.13	3.52	—	—		—	—	—	—	—
		AS Inhal.		14	0.33	2.46	0.09	1.65	0.07	2.19	13	0.010		7.47	0.0003	0.295	0.002	6.13

All EC values have been background corrected. AS, area sample; Resp., respirable size fraction; Inhal., inhalable size fraction; s cm⁻³, CNT or CNF structures per cubic centimeter.

Table 5.

Differences between EC and TEM GM concentrations by sample type and industry.

Sample	Industry	EC differencesa	TEM differencesa
PBZ Resp.	Primary manufacturer	A	—
	Hybrid producer/user	B	—
	Secondary manufacturer—electronics	B	—
	Secondary manufacturer—composites	B	—
PBZ Inhal.	Primary manufacturer	A	A
	Hybrid producer/user	B	A
	Secondary manufacturer—electronics	A	B
	Secondary manufacturer—composites	B	A
AS Resp.	Primary manufacturer	A	—
	Hybrid producer/user	A	—
	Secondary manufacturer—electronics	A	—
	Secondary manufacturer—composites	A	—
AS Inhal.	Primary manufacturer	A	A
	Hybrid- producer/user	B	A
	Secondary manufacturer—electronics	A,C	B
	Secondary manufacturer—composites	C	A

AS, area sample; Resp., respirable size fraction; Inhal., inhalable size fraction.

^aGM PBZ and area sample concentrations with the same letter are not significantly different in a Tukey multiple comparison ANOVA procedure with a level of significance of 0.05 for each sample type.

The GM TEM structure count data followed a similar trend as the 8-h TWA respirable and inhalable EC data, indicating that the hybrid sites (0.042 CNT-structures per cm³; 8-h TWA 0.026 CNT-structures per cm³) had the highest exposures, followed by the SM-composites/thermoplastics (0.041 CNT or CNF-structures per cm³; 8-h TWA 0.008 CNT or CNF-structures per cm³). However, CNT structure count data indicated that the primary manufacturers had higher exposures (0.011 CNT-structures per cm³; 8-h TWA 0.007 CNT-structures per cm³) compared to the SM-electronics sites (0.001 CNT-structures per cm³; e8-h TWA 0.0002 CNT-structures per cm³), contradicting the 8-h TWA EC data. The 8-h TWA data trends for all PBZ samples by industry group are displayed in Fig. 3.

Figure 3.

Comparison of 8-h PBZ-TWA exposure trends subdivided by industry.

Exposures by material type

Of the 14 sites visited, 5 facilities predominately handled or produced SWCNT, while eight facilities used or produced MWCNT, and one facility predominantly handled CNF. When EC exposures were examined by material type (Table 6), facilities that handled or produced MWCNT were found to have statistically higher GM EC PBZ concentrations, at both the respirable (0.68 µg m⁻³) and inhalable (4.50 µg m⁻³) size fractions, compared to sites that handled or produced SWCNT (0.16 µg m⁻³, respirable; 0.27 µg m⁻³, inhalable). CNT structure count data from the side-by-side PBZ samples were also statistically higher: 0.023 CNT-structures per cm³ for sites that produced or used MWCNT compared to 0.002 CNT-structures per cm³ for those that produced or used SWCNT. The CNF site was not included in the analysis due to a small sample size but had an average inhalable GM EC PBZ of 5.59 µg m⁻³ and a PBZ CNF structure count of 0.325 CNF-structures per cm³.

Table 6.

EC and TEM exposures by material.

Material		Sample		EC								TEM
				n	GM (µg m−3)	GSD	Min.	Max.	8-h TWA GM (µg m−3)	8-h TWA GSD		n	GM (s cm−3)	GSD	Min.	Max.	8-h TWA GM (s cm−3)	8-h TWA GSD
SWCNT (n = 5)		PBZ Resp.		12	0.16	5.40	0.02	1.94	0.08	2.72		—	—	—	—	—	—	—
		PBZ Inhal.		22	0.27	6.91	0.01	5.21	0.09	7.28		22	0.002	5.96	0.0001	0.038	0.001	8.58
		AS Resp.a		19	0.19	5.31	0.02	6.94	0.06	4.96		—	—	—	—	—	—	—
		AS Inhal.		20	0.07	5.71	0.01	1.45	0.02	4.60		20	0.002	4.93	0.0001	0.027	0.001	7.76
MWCNT (n = 8)		PBZ Resp.		13	0.68	2.39	0.15	2.94	0.33	2.84		—	—	—	—	—	—	—
		PBZ Inhal.		23	4.50	4.25	0.53	79.57	1.36	6.13		27	0.023	8.13	0.0003	0.501	0.007	10.52
		AS Resp.a		28	0.37	3.72	0.05	4.46	0.14	3.02		—	—	—	—	—	—	—
		AS Inhal.		31	0.77	5.26	0.08	210.22	0.27	6.73		30	0.006	7.25	0.0001	0.631	0.002	7.59
CNF (n = 1)		PBZ Resp.		—	—	—	—	—	—	—		—	—	—	—	—	—	—
		PBZ Inhal.		2	5.59	1.53	4.15	7.54	0.96	4.55		2	0.325	9.66	0.0653	1.613	0.056	3.24
		AS Resp.		2	2.39	1.22	2.08	2.76	0.31	3.72		—	—	—	—	—	—	—
		AS Inhal.		2	0.53	2.65	0.27	1.06	0.07	1.00		2	0.029	26.95	0.0028	0.295	0.004	10.05

AS, area sample; Resp., respirable size fraction; Inhal., inhalable size fraction; s cm⁻³, CNT or CNF structures per cubic centimeter.

^aOnly the GM EC levels for the area respirable samples were found to not be significantly different using ANOVA with a level of significance of 0.05 for each sample type. The CNF site was excluded from this analysis due to the small sample size. All other EC and TEM values were found to be statistically different. All EC values have been background corrected.

TEM size distributions

TEM-based estimates of the maximum dimension of each CNT structure were grouped into size-bins for 6 of the 14 sites, all of which produced or used CNT (Sites G, I–N; Fig. 1). Figure 4 displays the TEM results as the average number of CNT structures counted per size bin, subdivided by site. As shown in Fig. 4, most site average CNT structure counts were within or above the 2–5 µm size-bin. The most common size bin for CNT structures, when averaged over all sites, was the 2–5 µm (7.7 CNT structures counted per sample), followed by size-bins of 5–10 µm (6.5) and >10 µm (5.4). The three bins with the fewest number of CNT structures averaged over all sites were the <1 µm (1.1), single fiber (1.3), and 1–2 µm (2.7) size bins.

Figure 4.

Average number of CNT structures or fibers per size bin by site. Site M was a primary manufacturer of SWCNT and MWCNT, Sites G and L were hybrid producer/users of MWCNT, Site N was a secondary manufacturer of SWCNT in the electronics industry, and Sites I and J were secondary manufacturers of MWCNT in the composites/thermoplastics industries. Site K was excluded from the figure since all TEM samples found non-detectable results. ‘Single CNT’ size bin includes all non-agglomerated single CNT fibers.

EC-TEM correlations

PBZ and area inhalable EC concentrations showed a positive correlation based on 104 paired TEM samples from Sites A–N (Fig. 5). A significant Spearman correlation of 0.389 was found (P < 0.0001) for the inhalable EC samples and the corresponding TEM data. The correlation was also found to be significant between the 8-h TWA inhalable EC and corresponding TEM values (Spearman ρ = 0.450; P < 0.0001; n pairs = 109). The correlation of the EC mass concentration samples at the respirable size fraction with the corresponding TEM samples was somewhat lower (Spearman ρ = 0.213; P = 0.036; n pairs = 97).

Figure 5.

Scatterplot of inhalable EC versus TEM area and PBZ samples. Samples <20min were excluded from analysis.

DISCUSSION

This study provides detailed information on occupational exposures to CNT and CNF from 14 companies (13 CNT facilities, 1 CNF facility); spanning four distinct industries in which these advanced materials are being produced or used. Schubauer-Berigan et al. (2011) estimated that 44 companies, not including research laboratories or universities, were primary or secondary manufacturers of CNT and CNF in the USA, as of 2008–2009. This group, augmented with an additional six companies, formed the sample base for the present study. The facilities in this study were generally small, with 11 of the 14 sites employing fewer than 50 employees company-wide, and were considered for inclusion in cross-sectional epidemiologic studies, currently underway.

Twelve of the 14 companies typically used non-functionalized CNT and CNF materials while several of the primary manufacturers could functionalize CNT on-site, but only did so at the request of customers. Sites within the electronics industries solely used purified tubes with lower residual metal catalyst, usually as aqueous or solvent suspensions, while all other facilities handled the raw powder form of the CNT and CNF materials, at least initially. Over half of the sites visited incorporated CNT and CNF powders into a less-dispersible (in air) form, such as dispersed within a liquid, post-coated with a resin, or incorporated into a composite or thermoplastic.

Elemental carbon

Overall, nearly all PBZ and area TWA EC samples (96%) were below the NIOSH REL when measured at the respirable size fraction. Although the GMs for respirable samples from all sites were generally well below the REL, inhalable PBZ samples were found to have an overall GM >1 μg m⁻³, with 30% of the 8-h TWA PBZ samples being >1 μg m⁻³. Based on the TEM results, it was not unexpected that the inhalable mass concentrations were significantly higher than the respirable mass, as the majority of the CNT and CNF structures were larger agglomerates, typically >5 μm. These larger agglomerates should contribute more to the inhalable mass as opposed to the smaller respirable fraction. Currently, there are no exposure limits for CNT and CNF at the inhalable size fraction because risk assessments, to date, have focused on the adverse pulmonary effects expected from exposures in the alveolar region. The data collected in this study indicate that occupational exposure is mainly to larger CNT and CNF agglomerates in the thoracic and larger size fractions, though a small respirable sub-fraction also is present, depending on the materials and processes used. Similar findings were reported for a previous study of CNF (Birch et al., 2011). However, it is largely unknown if there are potential human health effects associated with exposures to CNT and CNF particles that are predominately in the inhalable/thoracic size fractions, or whether agglomerates that deposit in the lungs or airways can dissociate into more dispersed fibers, as seen in a recent study (Mercer et al., 2013). The thoracic region is of particular interest for the development of bronchogenic lung cancer. Further toxicological research is needed to answer these questions. Until more information is known about the health effects associated with larger CNT and CNF agglomerates, it seems prudent to assess worker exposure to airborne CNT and CNF materials by monitoring EC at both the respirable and inhalable size fractions. As confirmation, concurrent TEM samples should be collected to confirm the presence of CNT and CNF.

When EC exposures were sub-divided by industry, SM-electronics were found to have the highest respirable GM exposure. Conversely, the 8-h respirable TWA for the SM-electronics sites was among the lowest observed. This can generally be explained by the high number of ND samples, small sample size, and short durations of exposure, typically being less than an hour. Therefore, the higher GM value may be an artifact of these issues, and the 8-h TWA value should be more representative of a typical respirable exposure within the SM-electronics sites. Hybrid producers/users and SM-composites/thermoplastics sites had the next highest respirable GM values, followed by primary manufacturers. The 8-h TWA values for these industries as well as the GM EC and TEM values were in general agreement with this trend (Fig. 3).

When subdivided by material (MWCNT, SWCNT, or CNF) produced or used most often, higher exposures were found at the MWCNT and CNF sites. Six of the eight companies that predominantly handled MWCNT were considered hybrid producers/users or SM-composites/thermoplastics sites and had the highest exposures at both the respirable and inhalable size fractions. A majority of the sites that handled or produced SWCNT were primary producers or SM-electronics sites. These data indicate that companies that handle MWCNT or CNF in secondary manufacturing industries, which generally handle larger quantities of the material in a powder form, may need to implement additional engineering and administrative controls to limit potential exposures.

A large range of GSDs were found for the various groupings reported in Tables 3, 4, and 6 for both the EC and TEM data. The variance was thought to be primarily explained by the differences by company in the quantity of CNT and CNF materials handled, dry powder handling techniques, and the extent to which engineering controls were used.

A recent study on the dustiness of various nanomaterials (Evans et al., 2013) collected data on the respirable-to-total dustiness ratio using a Venturi-based device. Materials tested included a MWCNT, a CNF, and two types of SWCNT, which produced respirable-to-total mass ratios of 0.17, 0.28, 0.41, and 0.84, respectively. In our study, where both the respirable and inhalable samples were collected side-by-side we found ratios of 0.05 and 0.12 at MWCNT sites while SWCNT sites had ratios of 0.22 and 2.49; the latter is likely explained by the small sample size and short durations of exposure at SM-electronics sites. Although only a small number of CNT and CNF materials were tested (Evans et al., 2013), and only a relatively small number of side-by-side respirable and inhalable samples were collected in this study, further research may be warranted to derive a mean ratio between size fractions that could be used to estimate EC concentrations in un-sampled fractions.

Many of the early exposure assessments on CNT and CNF materials focused on collecting samples to estimate CNT and CNF mass gravimetrically, which does not differentiate CNT and CNF from other airborne dusts. Several studies based on total gravimetric mass found concentrations ranging from ND to 331.7 µg m⁻³ (Han et al., 2008) and 7.8 to 320.8 µg m⁻³ (Lee et al., 2010). A study by Maynard et al. (2004) estimated the inhalable mass of CNT based on the concentration of metal catalyst and reported a range of PBZ exposures of 0.7–53 µg m⁻³. More recent studies have collected PBZ samples at the respirable and inhalable size fractions using the chemical-specific mass of EC using NMAM 5040. The EC-based, CNT concentrations found at the inhalable size fraction ranged from ND to 38 µg m⁻³ (Methner et al., 2012). In studies that quantified CNT or CNF as respirable PBZ EC, CNF exposures ranged from 45 to 80 µg m⁻³ (Birch et al., 2011) and CNT exposures ranged from 0.08 to 7.4 µg m⁻³ (Hedmer et al., 2014), while thoracic CNF concentrations ranged from 19.37 to 46.32 µg m⁻³ (Birch et al., 2011). This manuscript reports CNT and CNF PBZ results ranging from 0.02 to 2.94 µg m⁻³ at the respirable size fraction and 0.01–79.57 µg m⁻³ at the inhalable size fraction from 14 sites across the USA, which appear comparable to results reported nationally and internationally.

Electron microscopy

A small number of studies on CNT have used similar collection and analysis methods based on NMAM 7400 and 7402 (Bello et al., 2008, 2009, 2010; Han et al., 2008; Lee et al., 2010; Dahm et al., 2012; Hedmer et al., 2014). Several of these studies counted only fibers with an aspect ratio >3:1, while other studies included both individual fibers, regardless of aspect ratio, as well as agglomerates. To date, specific counting rules and definitions have not been clearly established, complicating comparisons among the different studies (ASTM, 2014). Also, there is little guidance in the form of occupational exposure limits (OEL) for other CNT and CNF exposure metrics, such as number concentration via microscopy.

The OSHA permissible exposure limit (PEL) of 0.1 fibers per cm³ for asbestos could serve as a crude benchmark until a more appropriate, CNT and CNF-specific structure count based OEL is introduced. Although the toxicity of asbestos and some types of CNT and CNF have been compared and suggested to have similar health endpoints, such as mesothelioma (Takagi et al., 2008; Donaldson et al., 2010), there are no data indicating that the asbestos PEL is protective against potential adverse health endpoints that may be specific to CNT and CNF, which are not yet fully known. Similarly, the British Standards Institute (BSI) has also suggested a benchmark OEL for fibrous nanomaterials with aspect ratios >3:1 at 0.01 fibers per cm³ (BSI, 2007). However, as indicated from the TEM data in this manuscript, CNT and CNF agglomerates (Fig. 1B,C,D) do not fit the traditional 3:1 ratio specified by the OSHA asbestos PEL and the BSI benchmark OEL, as most CNT and CNF were large roughly spherical or oblong agglomerates.

Overall, 8-h TWA PBZ exposures, when analyzed by TEM, were 33 times lower than the current OSHA PEL for asbestos and three times lower than the BSI benchmark OEL. Of the 51 PBZ samples, only two were above the OSHA PEL for asbestos while 18 were above the BSI benchmark OEL. When subdivided by industry type, the CNT or CNF structure count data were in relative agreement with the EC data (Fig. 3), indicating that hybrid producer/users had the highest potential exposure, followed by SM-composites/thermoplastics sites, primary manufacturers, and SM-electronics sites. Similar to the EC data, when the TEM data was subdivided by material type, PBZ samples from sites handling MWCNT were seven times higher than sites that handled SWCNT.

In the TEM size-bin analysis, we observed that larger CNT agglomerates of typically >2–5 µm comprised most workplace exposures. These exposures were to a poly-disperse aerosol consisting predominately of CNT agglomerate structures in the thoracic/inhalable size fractions, with few instances of individual fiber CNT or agglomerates <1 µm. In general, these larger CNT agglomerates were comprised of a large number of entangled CNT fibers, often with residual soot. As seen in Fig. 4, the size-bin distributions varied by site, which could relate to the type of CNT (MWCNT or SWCNT), manufacturer, form used (powder or aqueous solution), and the specific processes performed at the site.

As previously reported in Dahm et al. (2012), a statistically significant positive correlation (r = 0.44; P = 0.01) was found between side-by-side PBZ and area samples collected for the mass concentration of EC and CNT or CNF structure counts by TEM analysis from Sites A–F. The additional inhalable samples from sites reported herein showed a similar correlation coefficient with increased statistical significance (P < 0.0001) between EC and TEM samples. This indicates that both metrics reasonably agree on exposure concentrations when sampling the same air space. The correlation between inhalable EC samples and TEM samples increased when the 8-h TWA data were used, which was most likely due to the normalization of outliers with shorter sampling durations from EC and TEM data. However, when the respirable EC data were correlated with the corresponding TEM data, the correlation declined nearly two-fold, but was still statistically significant (P = 0.036). Since the use of a cyclone likely excluded the larger, more common agglomerates ≥4 µm, the EC respirable mass was considerably lower, which probably decreased the correlation with the CNT structure counts collected without a size-selective sampler.

Limitations

Due to evolving sampling methodologies not all sites were sampled identically. When subdivided by industry type, at least one site within each industry group had respirable and inhalable PBZ EC samples at the same site. These samples were considered to be representative exposures for the remaining sites within the industry grouping, where respirable PBZ samples were not collected.

In order to account for the anthropogenic and natural sources of EC, PBZ, and area samples were background-corrected. However, this assumed that a representative background sample was collected at each facility that could be applied to all PBZ and area samples. This, at times, was challenging, since workers can spend varying amounts of time between different buildings with varying types of HVAC systems or access to outdoor air which could influence background EC levels. However, understanding an employee’s job tasks and work flow patterns, along with general knowledge of the facility, will aid in the selection of a representative background sample.

Modification of the thermal program used for thermal-optical analysis of EC was proposed as a means to distinguish CNT and CNF aerosols from amorphous EC, which may be present in the bulk materials to varying extents, and atmospheric (soot) air pollution. The proposed modification targets a higher temperature EC fraction as a measure of CNTs and reportedly improves selectivity against background EC (Ono-Ogasawara and Myojo, 2011, 2013). However, using EC3 as a measure of CNT and CNF materials may underestimate their air concentrations due to negative bias for materials that oxidize at lower temperatures, especially when background is negligible. Instead, the OC-EC splits in this study were based on background samples and the thermal profiles of bulk CNT or CNF materials obtained at the facilities. This approach considers the thermal properties of the actual materials used rather than making assumptions based on materials that may not be representative.

We observed that TEM samples showed varying levels of amorphous soot. It generally seemed that the soot levels were dependent on the type of material produced or used as well as the material’s intended application. In some cases (e.g. raw products), the level of soot not only affected the EC concentration but also demonstrates the difficulty in creating standardized TEM structure count methodologies, since we counted large agglomerates of amorphous soot with a few CNT and CNF fibers embedded the same as a large agglomerate solely comprised of many CNT and CNF fibers.

CONCLUSION

Most exposures to CNT and CNF in this industrywide exposure assessment study of 14 US CNT and CNF manufacturers and users were below the NIOSH REL at the respirable size fraction. However, measured exposures are possibly of health significance since CNT and CNF agglomerates were mainly found to be in the thoracic and inhalable size fractions, though a small respirable sub-fraction was also present. Particular caution should be exercised in workplaces that handle large quantities of CNT and CNF as dry powders, especially in the composites and thermoplastic industries, as well as facilities that produce and incorporate their own materials into various downstream applications on site. However, not all secondary manufacturers had high exposures, such as the sites in the electronics industries, which typically use small amounts of SWCNT materials within aqueous or solvent mixtures.

Overall there was reasonable agreement between exposure trends seen within the TEM and EC data. TEM structure counting was tedious, time consuming, and expensive, but it provided a sensitive, confirmatory means to assess exposure to CNT and CNF and should be considered a potential metric for future OELs. Although an exposure limit based on TEM structure count data would increase the sensitivity and specificity of CNT and CNF exposure monitoring, correlations presented herein indicate that an EC mass-based chemical analysis can provide comparable information. The latter approach is simple, inexpensive, and provides a quantitative indicator of CNT and CNF exposure. However, sampling and analytical challenges remain for assessing dust exposures during the sanding, cutting, or grinding of CNT and CNF composite materials. Such exposures will most likely become common as advanced composites find new applications in different industrial sectors. Exposure scenarios and appropriate monitoring methods for downstream uses of these advanced materials will need to be further addressed.

SUPPLEMENTARY DATA

Supplementary data can be found at http://annhyg.oxfordjournals.org/.

FUNDING

This research was supported in part by an interagency agreement between NIOSH and the National Institute of Environmental Health Sciences (AES12029-001) as a collaborative National Toxicology Program research activity, and the NIOSH Nanotechnology Research Center.