Engineered nanomaterials exposure in the production of graphene

ABSTRACT

The objective of this study was to obtain the multi-metric occupational exposure assessment to graphene family nanomaterials (GFNs) particles of workers engaged in the large-scale production of graphene. The study design consisted of the combination of (i) direct-reading instruments, used to evaluate the total particle number concentrations relative to the background concentration (time series with spatial approach) and the mean size-dependent characteristics of particles (mean diameter and surface-area concentration) and (ii) filter-based air sampling for the determination of size-resolved particle mass concentrations. The data obtained from direct reading measurement were then used to estimate the 8-h time weighted average (8-h TWA) exposure to GFNs particles for workers involved in different working tasks. Workers were generally exposed to 8-h TWA GFNs particle levels lower than the proposed reference value (40,000 particle/cm³). Furthermore, despite high short-term exposure conditions were present during specific operations of the production process, the possibility of significant exposure peaks is not likely to be expected. The estimated 8-h TWA concentration showed differences between the unexposed (<100 particle/cm³; <0.05 µg/m³) and exposed subjects (mean concentration ranging from 909 to 6438 particle/cm³ and from 0.38 to 3.86 µg/m³). The research outcomes can be of particular interest because the exposure of workers in real working conditions was assessed with a multi-metric approach; in this regard, the study suggests that workers who are directly involved in some specific working task (material sampling for quality control) have higher potential for occupational exposure than operators who are in charge of routine production work.

© 2016 American Association for Aerosol Research

EDITOR:

• Tiina Reponen

Introduction

Background

The graphene family of nanomaterials (GFNs) has been recently introduced into many fields of science and technology (Jang and Zhamu 2008; Rafiee et al. 2009). Given the potential occupational and public exposure to graphene due to its versatile applications, scientists are directing more attention toward investigating the safety aspects of these nanomaterials (Hu and Zhou 2013). What is emerging from the available results is a variety of effects that are strictly related to the nature of the graphene used: the size, layer number, chemical groups, and surface of graphene may have a strong impact on the biological and toxicological responses (Hu and Zhou 2013). Nevertheless, to date, no occupational or environmental exposure limits for GFNs have been set by any regulatory agency (Lo et al. 2011), and limited data are available regarding occupational exposure assessment in the GFNs production industry. In summary, further research is still required in this emerging field to draw conclusions regarding the potential hazards and risk characterization by way of the link between these exposure assessments (Allen et al. 2010; Guo et al. 2010; De et al. 2011; Bianco et al. 2013).

The preliminary hypothesis of this study was that the production process may cause occupational exposure arising from the emission and dispersion of GFNs particles, mainly in the form of airborne nanoparticles (“NPs,” i.e., particle with diameter <100 nm). The main objective of the study is to assess the occupational exposures to GFNs of workers engaged in the large-scale production of graphene. The exposure assessment was performed via environmental monitoring and aimed at the multi-metric characterization (i.e., particle number, mass and surface-area concentrations, particle mean diameter) of the NPs exposure concentration associated with each working task and working place.

The production process

The production process in the study company is a continuous manufacturing process based on the chemical intercalation of natural graphite, followed by thermal plasma expansion. The production unit (capacity of 30 tonnes per year) was designed with a modular pattern and to be fully automated with minimal human intervention; only quality-control material sampling, cleaning, and maintenance operations required manual interventions. The production was organized as follows: (i) acceptance of raw materials (graphite) and storage; (ii) plasma expansion; (iii) post-plasma treatment/exfoliation (performed in liquid media); (iv) drying; (v) finishing operations (e.g., packaging), and (vi) storage of final products. Quality controls of raw materials, by-products, and final products were performed through the entire production process and required material sampling from different points of the production process. All of these tasks were performed within a single building, organized to accommodate separated areas for each specific activity: warehouses for raw materials and finished products were placed in a separate area (downstairs); the production site (upstairs) was divided into a productive area designed with a modular pattern, situated in three containment chambers (“graphite expansion,” “exfoliation,” and “drying” room) and an open-space area including both R&D laboratory (materials characterization, quality control and testing) and “engineers” workstations (production process monitoring). Finally, the administration department (offices) was located in a separate, adjacent building.

Risk management strategy

It is globally recognized that a general hierarchical approach in risk management must be implemented to eliminate the hazard when possible, substitute it with a less hazardous material or, if not feasible, control the hazard at or as close to the source as possible. The effectiveness of the exposure controls and measurement methods, especially for engineered nanomaterials (ENMs), remains a key research need (Kuempel et al. 2012). In the absence of regulatory occupational exposure limits (OELs) for most of the existing ENMs, a strategy is required to determine the appropriate levels of exposure controls to protect workers' health (Pietroiusti and Magrini 2014). A typical hierarchical risk management approach (Table 1) was implemented at the study company coherently with a precautionary approach, to control the exposure to NPs associated with the production process (Mirer et al. 2008).

Table 1. Risk management options used for different work tasks.

	Production (expansion, exfoliation, drying)	Maintenance	Cleaning	Laboratory test
Organization, Procedures	Warnings, Training, Specific Operating Procedures (S.O.P.)
Engineering Control	Enclosure of high-exposure potential activity: fixed containment booths: HEPA-filtered, double locked-holding area			Chemical Hoods (EN 14175-compliant)
	Closed-process, fully automated production line; General Ventilation		+ HEPA-filtered Vacuum Cleaners
Personal Protective Equipment	Safety Boots and Gloves; Filtering half-facemask (FFP2- or FFP3-class filter); Nitrile gloves and disposable safety gloves

Methods

The exposure assessment was performed during a 12-month period (preliminary observational survey and six monitoring sessions). First, a preliminary assessment involved identifying the potential source(s) of NPs emissions by reviewing the type of process and work practices, as suggested by Methner et al. when conducting nanoparticles exposure assessment studies with tiered approach (Methner et al. 2009). Measurements were then performed in accordance with the sampling strategy described hereafter. The measurement design consists of the combination of: (i) direct-reading instruments used to evaluate the total particle number concentrations (PNC), including that of the background and the mean size-dependent characteristics of particles (mean diameter, surface-area concentration); (ii) filter-based PM sampling used for the determination of size-resolved particle mass concentrations.

Monitoring strategy

The measuring strategy for the exposure assessment consisted of: (i) determination of the “background” particle number concentration (natural and anthropogenic nanoparticles in the workplace air); (ii) determination of the total particle number concentration during the production process by means of micro-environmental and personal measurement; (iii) distinction of process-related GFNs particles from background aerosols; (iv) estimation of 8-h Time Weighted Average (8-h TWA) exposures for GFNs particles; and (v) comparison of these exposure values with the available occupational exposure thresholds.

Table 2 summarizes the monitoring design and strategy: average airborne PNC was measured in each location (workplace micro-environments) before and after the production or handling of nanomaterial to obtain an average background number concentration, which is then subtracted from the measurements made during processing, manufacturing, or the handling of ENMs (assuming that the emissions during process are stable during the measurements) (Brouwer et al. 2009; Methner et al. 2009, Koivisto et al. 2012a,b; 2014; 2015; Koponen et al. 2015; Jensen et al. 2015). This approach basically assumes that the concentration determined in each location, during no work activity is representative for the background concentration and any increased concentrations during the work activity can be attributed to the process, the nanomaterial or both (Kuhlbusch et al. 2011).

Table 2. Environmental monitoring strategy and instruments contextually used for different work tasks and environments.

	Production workers			Office workers (controls)
	Graphite expansion	Drying	R&D Laboratory	Office
Background Concentrations (site-specific)	CPC: Particle number concentration (time-resolved)
	DSC: Particle number concentration, mean diameter; LDSA (time-resolved)
Micro-environmental fixed-site monitoring	CPC, OPC, DSC, and DLPI: mass concentration, size-resolved (TWA)	CPC: Particle number concentration (time-resolved)
		OPC: Particle number concentration, size-resolved (time-resolved)
		DSC: Particle number concentration, mean diameter; LDSA (time-resolved); PCIS: mass concentration, size-resolved (TWA)
Personal monitoring	DSC: Particle number concentration, mean diameter; LDSA (time-resolved)			n.a.

CPC = Condensation particle counter; DSC = diffusion size classifier; OPC = optical particle counter; DLPI: dekati low pressure impactor; PCIS = Personal Cascade Impactor Sampler; LDSA: lung deposited surface area concentration.

Once the background particle number concentrations have been determined, specific measurements are made with all of the available instruments simultaneously at different locations: the results from this type of measurement should be interpreted as an indicator of workplace's micro-environment exposure concentration. Exposure to GFNs particles was estimated by comparing the workplace concentrations with the background concentration, following different approaches for background distinction (Kuhlbusch et al. 2011; Berges et al. 2013). However, note that these approaches are considered as a proxy for assessing NPs exposures, despite there being several possible errors associated with use of the count-difference methods for background distinction. In this study, a precautionary and conservative approach was adopted, attributing the whole differential particle concentrations as GFNs particles (without any differentiation between incidental and process-related engineered nanoparticles).

The sampling time generally matched the length of time necessary to complete an entire work shift or a half work shift. Direct reading instruments were placed at both the process/task location (where operations involved engineered nanomaterial production or application, i.e., graphite expansion, drying, R&D laboratory) and at the process/task location (where process nanomaterials were not directly involved [office]). The exfoliation room was excluded from environmental monitoring because workers were not required to work for extended periods in this location and, the exfoliation phase was conducted in closed-process conditions, for which GFNs particles emissions in the workplace were excluded. A time-activity diary was also used to separate the continuous data as a function of the different monitored environments and working tasks. Filter-based, size-selective air sampling (“DLPI”: Dekati® Low Pressure Impactor) was also performed in a selected location (graphite expansion, drying) following a worst-case exposure scenario approach (highest expected exposure concentrations). Finally, personal sampling was performed using the miniature diffusion size classifier. Personal measurements were collected in the breathing zone of workers for whom exposure can be expected (e.g., during graphite expansion and drying phases). All of the instruments were used simultaneously for the entire length of the monitoring period, except for the DSC (which was alternatively used for personal sampling and fixed-site monitoring) and for the personal cascade impactor sampler (PCIS), which were used for extemporary measurements in one monitoring session.

Direct-reading measurements

Numerical concentrations of airborne particles were measured using a miniature diffusion size classifier (DSC) and condensation particle counters (CPC). The DSC used for this study (DiSCmini, Matter Aerosol AG, Wohlen AG, Swiss) measures the particle number concentrations (range: 10³–10⁶ particle/cm³) and particles' average diameter in the size range of approximately 10–700 nm (Fierz et al. 2011). DSC also estimates with reasonable accuracy (Fierz et al. 2011) the lung deposited surface area concentration (LDSA), defined as the particle surface area concentration per unit volume of air, weighted by the deposition probability in the lung and calculated according to ICRP report 66 (ICRP 1994). Portable CPCs were also used in this study (P-Trak Ultrafine Particle Counter model 8525; TSI Inc., Shoreview, MN, USA) to quantify the airborne particle number concentration(size range: 0.02 to 1 μm). Finally, numeric concentrations of airborne particles were also measured using optical particle counters (“OPC,” mod. Handheld 3016, Lighthouse Worldwide Solutions, Fremont, CA, USA), which are able to provide real-time measurement of particles with optical diameters in the 0.3–30 µm range (six different dimensional fractions).

Filter-based PM sampling and gravimetric analysis

The gravimetric determination of the airborne size-fractionated airborne particulate matter (PM) was conducted to characterize the mean exposure to size-segregated PM in terms of mass concentrations (µg/m³). A DLPI Low Pressure Impactor was used: DLPI classifies airborne particles into 13 size fractions (from 30 nm to 10 μm) at a sampling flow rate of 30 l/min. Sampling were conducted by means of greased filters (aluminium membranes, 25-mm diameter, greased with Apiezon-L); a continuous control of impactor's pressure (100 ± 5 mbar) was performed during the sampling period to ensure the exact diameter cut point over the measurement period, other than to ensure the accuracy of the sampling volume estimates. A PCIS, which was developed for the analysis of size-segregated particulate matter, was also used for extemporary measurements in one monitoring session. PCIS is a miniaturized cascade impactor, which operates at a flow rate of 9 l/min and consists of four impaction stages with cut-off diameters of 2.5, 1.0, 0.5, and 0.25 µm (PTFE s/PTFE filters; diameter: 25 mm, porosity: 0.8 mm), which are followed by an after-filter for particles <0.25 µm (PTFE w/PMP ring; diameter: 37 mm; porosity: 2 µm) (Misra et al. 2002). Size-resolved mass concentrations (µg/m³) were then determined by gravimetric analysis in accordance with reference methods and with the accepted standard practice (UNI EN 12341:1999; UNI EN 14907:2005). The net PM mass on the filters was measured by weighing the conditioned filters before and after sampling with a microbalance having a resolution of 1 µg in a temperature- and relative humidity-controlled (20 ± 1°C; 50 ± 5%) environment (Activa Climatic; Aquaria, Lacchiarella, MI, Italy). PM masses were corrected by subtracting the mean blank weights (two field blank and two laboratory blank) from the sample weights. The quality of the weighing procedure was assessed using the American Society of Testing and Materials (ASTM) D 6552 method (ASTM 2000).

Data treatment and statistical analysis

The GFNs particles' numeric concentrations (obtained after background concentration subtraction), mean diameter, and LDSA were used to estimate the 8-h TWA exposure to graphene NPs for workers involved in different working tasks, by means of a microenvironmental model. The basic concept of this type of model is that time-weighted average exposure is defined as the sum of partial microenvironmental exposures, which are determined by the product of the GFNs particles concentration and the time spent in each microenvironment/work task. For this purpose, time-activity patterns were derived from time-activity diaries and a simple questionnaire submitted to workers. Workers were a priori classified as “non-exposed” (N = 5; office workers) or “exposed,” including two groups of production workers: “operators” (N = 3; assumed to be exposed and directly involved in the production process) and “engineers” (N = 6; assumed to be exposed but not directly involved in the production process). A further estimation was performed to obtain the 8-h TWA exposure values expressed as mass concentrations (µg/m³). The calculation was performed on the basis of a simple algorithm already applied in previous studies on ultrafine particles (Wittmaack et al. 2002; Spinazzè et al. 2015), based on the relationship between the mean mass density factors of NPs (ρ = 0.05 g/cm³) derived from the gravimetric determination of PM_0.1 in the expansion room, mean particle volume (derived from mean particle diameter), and 8-h TWA exposure expressed as PNC. Note that this technique presents some limitations (Spinazzè et al. 2015) thus, the following results are not expected to provide extremely accurate exposure values but are indicative of the magnitude of the exposure.

The use of non-parametric methods was considered the most appropriate because the data results are skewed. Differences in the median concentrations as a function of working conditions/tasks were tested with the Kruskal-Wallis, one-way ANOVA, and the “MW”: Mann-Whitney U-test performed using an IBM SPSS 20.0 (IBM, Armonk, NY, USA). Statistical results were regarded as a significant when p < 0.05. The results are presented in the text as the geometric mean (GM) ± geometric standard deviation (GSD).

Results

Overall, six monitoring campaigns were performed, after an initial assessment survey, for a total of over 50 of field survey. During the initial survey, direct-reading instruments were used to obtain a semiquantitative indication of the magnitude of potential exposures for each process or work task.

The variation of airborne size-fractionated particle concentrations from background concentrations (Table S1, online supplemental information [SI]) is reported in Figure 1.

Figure 1. Percent variation of airborne size fractionated particles number concentrations from background concentrations (i.e., particle number concentration measured at a specific location, minus the average background concentration—assumed to be constant—in that specific location; this difference was then divided by the background concentration to obtain the percentual variation from the background concentration) at different sampling locations. Histograms and error bars denote the geometric mean ± geometric standard deviation of the percentual variation. Size fractionated particles number concentrations were measured by means of miniature diffusion size classifier (particle < 0.1 µm) and optical particle counter (particle 0.3–0.5 µm and > 1 µm).

The results reported in Figure 1 showed that the airborne particle concentrations during the production phase were consistently higher than the corresponding background levels for the sampling locations in the productive area. A high particle number concentration for particles <0.1 µm, combined with a high particle count in the small size range (0.3–0.5 µm), indicates the possible presence of nanoscale particles (expansion room and drying room); conversely, a low count for particles <0.1 µm, in combination with a high count in the larger size range (>1.0 µm) would indicate the presence of large particles and/or agglomerates (Methner et al. 2009). The massive presence of nanometric and sub-micrometer particles during the production activity (graphite expansion room) was also observed in the expansion area through the gravimetric determination of size-fractionated PM performed by the DLPI sampler (Figure 2). In particular, the mean concentration (GM ± GSD; % cumulative) of PM₁ (97.2 ± 13.0 µg/m³; 85.8%), PM_0.5 (88.8 ± 9.8 µg/m³; 77.6%), PM_0.25 (16.5 ± 4.4 µg/m³; 12.0%), and PM_0.1 (1.4 ± 0.3 µg/m³; 1.0%) clearly reflect the high amount of sub-micrometer particle in the work environment, especially compared with spot measurement of size-fractionated particles (PM_0,25 = 4.3 µg/m³; PM_0.5 = 6.7 µg/m³; PM₁ = 9.3 µg/m³) performed in other non-productive areas using a PCIS.

Figure 2. Mass concentrations (µg/m³) of size-resolved airborne particles (sampling with Dekati® Low Pression Impactor, and gravimetrical analysis) in the graphite expansion room, during the production activity. Marker with error bars denotes the mean ± standard deviation (concentrations were not corrected for background concentrations).

Table 3 reports the multi-metric characterization of exposure concentrations to GFNs particles (personal and microenvironmental monitoring). The background number concentrations defined for each location were subtracted from the measurements made during processing, manufacturing, or handling of ENMs; thus, results from this type of measurement were interpreted as the workplace concentrations of Graphene NPs.

Table 3. Environmental monitoring results as a function of sampling location, after background concentration subtraction.

Position		GM	GSD	p5	p95
Metric
R&D Laboratory
PNC [particle/cm^3]	FS	3779	212	<100	13,521
PNC [particle/cm^3]	P	5882	198	<100	11,927
Particle diameter [nm]	P	91.2	17.8	79.5	104.8
LDSA [µm^2/cm^3]	P	68.02	7.55	65.49	70.51
Graphite expansion room
PNC [particle/cm^3]	FS	6578	3970	3030	52,200
PNC [particle/cm^3]	P	6050	3676	1862	48,885
Particle diameter [nm]	P	96.8	45.3	37.2	199.4
LDSA [µm^2/cm^3]	P	28.49	25.23	10.57	127.80
Drying room
PNC [particle/cm^3]	FS	758	553	620	1860
PNC [particle/cm^3]	P	198	230	192	1061
Particle diameter [nm]	P	203.4	68.8	73.2	300.0
Office
LDSA [µm^2/cm^3]	P	4.41	2.02	3.20	5.91
PNC [particle/cm^3]	FS	<100	73.6	<300	6000
Particle diameter [nm]	FS	136.0	22.9	121.7	150.3
LDSA [µm^2/cm^3]	FS	27.56	5.62	24.58	30.29

FS, fixed-site microenvironmental sampling; P, personal sampling; SD, standard deviation; p(N) = Nth Percentile. PNC: NPs number concentrations [particle/cm³]; LDSA: lung deposited surface area concentrations [µm²/cm³].

The highest mean exposure concentrations were observed in the graphite expansion room, which were one order of magnitude higher than those found in the R&D laboratory and in the drying room. As expected, the mean particle diameters had an inverse relationship with the PNC, while the LDSA had a direct relationship with the PNC. Therefore, larger particles and lower LDSA concentrations were primarily found in the drying room, while the smaller particles, with a higher LDSA were measured at graphite expansion room. Statistical analyses were performed to identify significant differences in the multi-metric OEEC characterization as a function of the different working task reported for the graphite expansion and drying activities. All of these work environments were characterized by statistically significant differences (p_KW < 0.001) in the particle number concentrations (both for fixed-site and personal monitoring), the diameter and the LDSA levels as a function of different work-tasks. In particular, the highest mean concentration and data variability (GM ± GSD) for the graphite expansion process were observed during material-sampling activities (FS: 9677 ± 4719 particle/cm³; P: 9844 ± 7704 particle/cm/cm³) and during the following 30-min period (FS: 15243 ± 5297 particle/cm³; P: 12601 ± 5388 pt/cm³), rather than during the routine production process (FS: 5643 ± 3987 particle/cm³; P: 5687 ± 4091 particle/cm³) or after the end of the production (FS: 4196 ± 323 particle/cm³; P: 3611 ± 619 particle/cm³). The same differences were observed for the average particle diameters (which followed an opposite trend compared to the PNC) and LDSA (positive correlation with PNC), showing large average particle sizes and smaller LDSA after the end of the production process (123.3 ± 29.4 nm; 25.93 ± 4.62 µm²/cm³) rather than during the routine production process (99.9 ± 40.6 nm; 29.58 ± 19.75 µm²/cm³) or during material sampling (53.6 ± 24.5 nm; 30.28 ± 39.43 µm²/cm³), and in the 30-min period after that operation (86.6 ± 28.3 nm; 60.18 ± 45.20 µm²/cm³). It must be noted that the monitoring of all these operations resulted in the determination of GFNs particles showing smaller average particle sizes and larger LDSA, compared to those determined as background concentrations in the same location (209.3 ± 43.3 nm; 18.42 ± 6.26 µm²/cm³). In contrast, smaller but statistically significant differences (p_KW < 0.001) were observed among the concentrations of exposure concentrations measured during the different work tasks monitored in the drying room, such as cleaning (FS: 942 ± 97 particle/cm³; P: 554 ± 244 particle/cm³), graphene-powder packing (FS: 571 ± 98 particle/cm³; P: 414 ± 190 particle/cm³), and drying (FS: 765 ± 597 particle/cm³; P: 110 ± 272 particle/cm³). Statistically significant differences were also observed for the average particle diameters and LDSA, with larger average particle sizes and smaller LDSA during the drying process (248.8 ± 68.8 nm; 4.48 ± .2.26 µm²/cm³), with respect to the background characterization (155 ± 56 nm; 5.8 ± 3.2 µm²/cm³), GFNs packing (146.3 ± 65.6 nm; 3.63 ± 2.19 µm²/cm³), or cleaning activities (170 ± 63 nm; 5.49 ± 2.71 µm²/cm³). In both the graphite expansion room and drying room, each activity-specific multi-metric characterization was significantly different (p_MW < 0.001) from the others.

Finally, the microenvironmental particle PNC, mean diameter, and LDSA concentrations determined for different work environments (Table 3) were used to estimate the 8-h weighted exposure to GFNs particles for workers involved in different tasks (Table 4). The 8-h TWA exposure estimation showed differences between the unexposed and exposed subjects; further, exposed operators involved in graphite expansion process are attended to experience the highest exposure level (for to particle with small diameter and wide LDSA), which are one order of magnitude higher than the exposure levels estimated for drying process operator and engineers. Finally, the obtained 8-h TWA exposure levels expressed as mass concentrations showed differences between the unexposed (<0.05 ± 0.001 µg/m³) and exposed subjects: for these latter group, in particular, further differences were explored: exposed operators involved in graphite expansion process (3.86 ± 0.085 µg/m³) and drying process (3.69 ± 0.078 µg/m³) are expected to be exposed at GFNs particle levels one order of magnitude higher than engineers (0.38 ± 0.001 µg/m³).

Table 4. Time weighted average exposure (TWA 8-h) to graphene NPs for typical workers profile.

Exposure^* profile	GM	GSD	p5	p95
Metric
Not exposed - Office workers (N = 5)
[Time use: Office = 95.8 ± 5.3%; R&D Laboratory = 4.3 ±5.3%]
PNC [particle/cm^3]	<100	80	<100	6071
Particle diameter [nm]	134.1	22.7	134.1	150.2
LDSA [µm^2/cm^3]	29.28	5.70	26.42	32.49
Exposed - Engineer (N = 6)
[Time use: Office = 12.5 ± 30.6%; R&D Laboratory = 87.5 ± 30.6%]
PNC [particle/cm^3]	3319	195	<100	12396
Particle diameter [nm]	96.8	18.4	96.8	111.5
LDSA [µm^2/cm^3]	62.96	7.31	60.48	65.54
Exposed - Operator (Graphite Expansion) (N = 2)
[Time use: R&D Laboratory = 5%; Graphite Expansion room = 95%]
PNC [particle/cm^3]	6438	3782	10659	20113
Particle diameter [nm]	96.6	44.0	46.5	206.4
LDSA [µm^2/cm^3]	30.47	24.35	9.33	125.30
Exposed - Operator (Drying) (N = 1)
[Time use: R&D Laboratory = 5%; Drying room = 95%]
PNC [particle/cm^3]	909	536	854	3113
Particle diameter [nm]	197.8	66.3	108.6	364.0
LDSA [µm^2/cm^3]	7.6	2.3	6.43	9.18

* Occupational Exposure to NPs (obtained after background concentration subtraction).

GM, geometric mean; GSD, geometric standard deviation; p(N) = Nth Percentile; PNC: GFNs particle number concentrations [particle/cm³]; LDSA: lung deposited surface area concentrations [µm²/cm³].

Discussion

The scientific literature about occupational exposure in the production of GFNs is limited: no studies concerning the occupational exposure to nanoparticles reports data on the GFNs production industry (Brouwer et al. 2009; Kaluza et al. 2009; Kuhlbusch et al. 2011; Pietroiusti and Magrini 2014). Making direct comparison with other case studies is quite difficult. Nevertheless, this study shows strong similarities in the exposure characterization to other types of NPs as previous studies showed that (i) under certain circumstances (e.g., maintenance activity, open handling of nanopowders) a release of NPs may occur and that (ii) the pattern of exposure is generally characterized by transient high peaks, linked to specific operations (Magrini and Pietroiusti 2014). In fact, in this study mean particle concentrations (Table 3) were on average of the same order of magnitude as the background level (Table S1, in the SI): by applying the methodology based on the ratio between workplace air concentrations and three times the standard deviation of the background concentration (Asbach et al. 2012a), it may be concluded that nanoparticle emissions during the study process were not significant. Nevertheless, in a specific location (R&D laboratory, graphite expansion room), the peak particle concentrations were one or two orders of magnitude higher than the background levels, and both the standard deviation and 95th percentile were considerably higher than those of the background concentration (Table 3). These are indications of a process in which significantly high short-term exposures may occur. Similarly, the mean exposure concentrations (Table 4) were also in the same order of magnitude (Office workers, Engineers, Drying operators) or one order of magnitude higher (operators working in the graphite expansion room) than background levels (Table S1, in the SI). Further, as mentioned before, the study work environments were characterized by statistically significant differences (p_KW < 0.001) in the particle number concentrations (both for fixed-site and personal monitoring), the diameter, and the LDSA levels as a function of different work-tasks, confirming that high short-term exposure conditions were present during the production process.

In this regard, a recent technical report performed an evaluation of engineering controls for manufacturing and handling graphene nanoplatelets in the workplace and concluded that some specific tasks in the production process were identified as the sources of release of nanoparticles into the workplace (Lo et al. 2011); the same report also stated that appropriate engineering controls could help mitigate exposure to nanomaterials in production areas. Similarly, in the present study, specific work tasks showed the potential to cause serious contamination within the workplace; however, the development of properly guided procedures for workplace and worker surveillance, together with the improvement of removal and containment systems as well as education and training workers, assisted by periodic exposure assessment activities, were all helpful in preventing workplace contamination and to contain workers' exposure. The definition of specific standard operating procedure was part of a comprehensive risk-management program (Table 1) that also addressed other issues, such as the abatement of air contamination in isolated work areas, the protection of non-productive areas from possible contamination events, and the protection of workers engaged in the production process with adequate personal protective equipment. The implementation of up-to-date control strategies (e.g., improvement of the existing local exhaust ventilation systems with mobile HEPA-filtered inlets to be placed in correspondence with the potential sources of NPs) is recommended and is now under consideration: this should contribute to a lowering of occupational exposures, workplace contamination levels, and high-level transient exposure peaks.

It also must be considered that, to date, no legal binding framework concerning ENMs-specific OELs exist, but several unofficial OELs for ENMs are being proposed by national organizations. In particular, no specific regulatory OEL for graphene NPs were defined (Lo et al. 2011; Pietroiusti and Magrini 2014). In the absence of regulatory OELs, alternative strategies are required for risk assessment purposes, e.g., the determination of technical exposure thresholds for the protection of workers' health. In this regard, exposure levels can be compared with reference values (RVs), developed to provide provisional limit values in situations where recognized OELs and DNELs are not available. RVs represent a warning level; when they are exceeded, exposure control measures should be taken. In this case, the nanoparticles were bio-persistent, granular, and fiber-form in the size range of 1–100 nm with a density of <6.0 g/cm³. Therefore, an 8-h TWA RV of 40,000 particles/cm³ may be adopted as RV (Cornelissen et al., 2012; Van Broekhuizen et al., 2012; Pietroiusti and Magrini, 2014). Note that the adopted RV, as the vast majority of the proposed OELs, is based on the number concentration as reference metric (although some of the proposed OELs are also expressed as mass concentration). Until reliable, easy-to-use and inexpensive devices are developed, this metric should remain the reference to allow for data comparability. Other metrics, such as surface-area concentration, still has limited use, and, overall, no proposed OELs have been developed in relation to this metric (Pietroiusti and Magrini 2014). According to the performed measurements and estimations, workers were exposed to GFNs particle levels (8-h TWA) substantially lower than adopted RV; it should be considered, however, that the most probable exposure in the workplaces is represented by transient spikes occurring during some workplace procedures (Pietroiusti and Magrini 2014). Taking this characteristic into account, the American Conference of Governmental Industrial Hygienists established that a nanotechnology process could be considered to require further assessment if (i) short-term exposures exceed three times the RV for more than a total of 30 min per 8-h working day or (ii) a single short-term value exceeds the RV by five times (ACGIH 2010). None of these two conditions apply here because (i) concentrations above three times the adopted RV (120,000 particle/cm³) were achieved only for a single short-term episode (<1 min) (ii) no single value exceeded by five times the RV (200,000 particle/cm³).

Some limitations in the study design and methods could have had an impact on the exposure assessment, including possible errors associated with use of the count-difference method to estimate the particle number concentrations of GFNs particles and direct-reading instruments sensitivity. In particular, the distinction of GFNs particles from background aerosols was performed with a conservative approach, by the quantification of the pre-and post-activity concentrations (background) and assuming that the emissions during process are stable during the measurements (Koivisto et al. 2012a,b, 2014, 2015; Koponen et al. 2015; Jensen et al. 2015). This approach attributes the whole background-differential concentrations to GFNs particles without any differentiation between incidental and process-related nanoparticles; thus, it is plausible that sources of incidental nanomaterials, and the large fluctuations that may result from their presence, may make it impossible to identify actual releases of ENMs and, in conclusion, this approach was likely to result in an overestimation of graphene NPs exposures, mainly for non-productive areas where the presence of graphene NPs is not expected.

Further, at present, the available measurement techniques generally present some limitations in terms of specificity and selectivity, and therefore every attempt to assess the occupational exposure to NPs requires the use of multiple sampling and monitoring techniques. However, these techniques are not yet fully evaluated in terms of accuracy and reliability (in particular those measuring number concentration and surface-area) and are not portable (i.e., multistage impactors for the determination of mass concentrations) so that exposure characterization is deeply limited by the lack of availability of instrumentation for collecting high-quality data, especially for personal monitoring. In this regard nano-specific monitors and samplers are subject to large studies (Kaluza et al. 2009; Asbach et al. 2012b; Kaminski et al. 2013; Price et al. 2014; Zimmerman et al. 2014), which are expected to improve their sensitivity and reliability. To date, however, the presented screening approach represents one of the best strategies for the exposure assessment, and followed the main available technical recommendations on the harmonization of the strategies for measuring exposure to nanoparticles (Methner et al. 2009; Brouwer et al. 2012; Cornelissen et al. 2012; Meier et al. 2013). Further research will focus on the morphological characterization of NPs by SEM-FEG for a complete graphene NPs characterization.

Conclusions

This study was designed to determine the potential graphene particle contamination deriving from the production of GFNs in a specific industrial setting. The obtained information on workplace contamination was then used to estimate the potential workers' exposure to GFNs particles. The research outcomes can be of particular interest because the exposure of workers in real working conditions was assessed with a multi-metric approach (i.e., particle number and mass concentration, particle size distribution, particles diameter, and surface-area concentrations). The multi-metric characterization of occupational exposure to graphene NPs resulted in significantly different results, both for different work environments and for each specific work-task. The study showed that workers are exposed to graphene NPs mean levels lower than the proposed reference values (8-h TWA = 40,000 particle/cm³) and that the possibility of significant exposure by transient high GFNs particles peaks is not likely. However, the results suggest that workers who are directly involved in some specific work task (material sampling for quality control) have higher potential for occupational exposure than operators who are in charge of routine production work.

Conflict of interest

The authors declare that they have no conflict of interest.

Acknowledgments

The authors gratefully acknowledge Directa Plus S.p.A, and in particular Giulio Cesareo (CEO) and Valerio Giugliano (Project Manager and Project Coordinator), for participating in this research project and for the permission to publish; the authors also acknowledge Directa Plus' administrative and operation personnel for their kind and helpful collaboration. The authors acknowledge Marina Limonta, Francesca Borghi, and Giacomo Fanti for their contributions to the environmental monitoring activities.

Funding

This study was supported by a grant in the framework of “The MULAN program” (MULtilevel Approach to the study of Nanomaterials health and safety), a project founded by Fondazione Cariplo (Grant number: 2011-2096).