exposure assessment of particulate matter from abrasive ... · exposure assessment of particulate...

Aerosol and Air Quality Research, 15: 1906–1916, 2015 Copyright © Taiwan Association for Aerosol Research ISSN: 1680-8584 print / 2071-1409 online doi: 10.4209/aaqr.2015.02.0086

Exposure Assessment of Particulate Matter from Abrasive Treatment of Carbon and Glass Fibre-Reinforced Epoxy-Composites – Two Case Studies Alexander C.Ø. Jensen1*, Marcus Levin1,2, Antti J. Koivisto1, Kirsten I. Kling1, Anne T. Saber1, Ismo K. Koponen1

1 National Research Centre for the Working Environment, Lersø Parkallé 105, Copenhagen DK-2100, Denmark 2 Department of Micro- and Nanotechnology, Technical University of Denmark, DK-2800, Lyngby, Denmark ABSTRACT

The use of composites is ever increasing due to their important structural and chemical features. The composite component production often involves high energy grinding and sanding processes to which emissions workers are potentially exposed. In this study we investigated the machining of carbon and glass fibre-reinforced epoxy composite materials at two facilities. We measured particle number concentrations and size distributions of the released material in near field and far field during sanding of glass- and carbon fibre-reinforced composites. We assessed the means of reducing exposure during the work by means of different working style, local exhaust ventilation, and enclosing the process area. Machining processes released particles primarily in < 100 nm size range. Without enclosure, process particle concentrations were 3.9 × 104 cm–3 in the near field and 1.3 × 104 cm–3 in the far field. Therefore workers in the same area may not be aware of being exposed to the process particles and the need of wearing protective outfits. Comparison of workers working style and effect on near field particle concentrations showed that a careless working style increased particle concentrations of 1.1% and 14.1% when comparing with a careful working style. Investigating the effect of the local exhaust ventilation showed that a maximum flow rate caused removal of close to 100% of particles from the working zone. A 28% reduction in the flow rate reduced particle removal efficiency more than 50%. With the enclosure around the process area, the process particle concentrations were 1.7 × 106 cm–3 in the near field, while the concentration in the far field was negligible. The enclosure used was shown to provide an average protection factor of more than 100. Keywords: Aerosol; Nanoparticle; Occupational hygiene; Emission control; Sanding. INTRODUCTION

Composite materials are combinations of two or more distinct phases or material types which in combination produce structural or functional properties not present in any individual component. Typical to composites is a high structural strength-to-weight ratio and corrosion resistance to a wide range of chemicals. This makes them as one of the most widely used materials. Fibre-reinforced composites are used in increasing quantities in nearly all technological areas such as transportation, marine, wind energy, aerospace, pipes and tanks, construction, electrics and electronics, and consumer goods (Zaman et al., 2013). The global fibre-reinforced composite market was in 2010 up to 1 × 107 metric tonnes with end-product value estimates in range of 40 to 70 billion euros (Gutiérrez and Bono, 2013). The * Corresponding author.

Tel.: +45- 391-652-00 E-mail address: [email protected]

majority of fibre-reinforced composites are made from glass fibres of which the composite industry used 2.57 × 106 metric tonnes which are followed by carbon fibres with a use of 3.9 × 104 metric tonnes (Gutiérrez and Bono, 2013).

Fibre-reinforced composites often require reworking by machining such as sanding, sawing, grinding or drilling to obtain the final product with the process producing dust in the respirable size range (Midtgard and Jelnes, 1991; Bello et al., 2009; Bello et al., 2010; Cena and Peters, 2011). Recently, the development of advanced composites that utilize engineered nanomaterials raised a concern that workers machining nanocomposites are exposed to airborne particulate matter including individual engineered nanoparticles (see e.g., Thostenson et al., 2005; Ging et al., 2014; Kingston et al., 2014). The first laboratory studies found that sanding of fibre-reinforced composites as well as sanding machinery emits high amounts of nanoparticles (Zimmer & Maynard, 2002; Koponen et al., 2009; 2011; Gomez et al., 2014). Bello et al. (2009) also found that dry cutting of alumina and carbon fibre-reinforced composites produced submicron rod-like fibres. Previous studies have shown that reworking of composites with engineered nanomaterials embedded

Jensen et al., Aerosol and Air Quality Research, 15: 1906–1916, 2015 1907

into the matrix do not release free engineered nanoparticles, the released fragments show that the filler is not separated from the polymer matrix (Koponen et al., 2011; Wohlleben et al., 2011; Gomez et al., 2014).

Only a few studies on the toxicity of particles generated during mechanical processing of composite fibre-epoxy dusts have been published: An intratracheal instillation study in rats testing dusts from six different fibre-epoxy composites showed that some of the dusts induced fibrosis in the lungs of rats 1 month after a single intratracheal instillation while others did not (Luchtel et al., 1989). Likewise, the scientific literature on the toxicity of dust from nanocomposites is very limited: To date, we know of four papers that have reported in vivo assessments of degradation fragments from nanocomposites such as paints and lacquer with different nanoadditives (Saber et al., 2012a, b; Smulders et al., 2014), and plastic and cement with carbon nanotubes (Wohlleben et al., 2011). None of these studies found increased toxicity of the sanding dust or other types of degradation fragments from nanocomposites compared to the products without nanomaterials: The studies suggested that toxicity is derived mainly by particles from the composite matrix and not the filler. However, it is important to stress that compared to the above mentioned studies the content of fillers (glass and carbon fibre) used in the present study is much higher than the content of nanomaterial in paints and cement (ranging from 2–15%). Individual glass and carbon fibres released during the reworking process may also have impact on the toxicity. Only a limited database on the toxicity of traditional carbon fibres exists. In one study, rats exposed to carbon fibres (7 µm in diameter and 20–60 µm in length) for six hours a day and five days a week for up to 16 weeks at an average chamber concentration of 20 mg m–3 did not respond with pathological changes in the lungs (Waritz el al., 1998). Recently, a 90-day inhalation study of rats exposed to carbon nanofibres showed minimal inflammation in the terminal bronchiole and alveolar duct areas of the lungs at 2.5 mg m–3 exposures (Delorme et al., 2012). Glass fibres have a fibrous structure that is similar to asbestos and inhalation leads to deposition throughout the respiratory tract (IARC, volume 81, 2002). However, in contrast to asbestos glass fibres have a low biopersistance in the lungs.

This study presents workers process-specific exposures to airborne particles during reworking of polymer-based composites reinforced with carbon fibres and glass fibres. Measurements were performed in two separate facilities where composites were sanded. We studied the effect of working style and emission controls using near field/far field measurements, estimating spatial exposure to process particles in the facilities. Particle samples were collected for gravimetric and microscopy analysis. Regional deposition rates were calculated for the sanding processes by using gravimetrically measured mass distribution and a lung deposition model. MATERIALS AND METHODS Work Processes and Measurements

The measurements on workers in this study were carried

out at two separate workplaces located in rural areas in close vicinity of agricultural activities. During reworking of the composites the workers were clad in full body Tyvek suits (Dupont, Wilmington, Delaware, USA), facial screens (EN 166 3:9:B, 3M, Saint Paul, MN, USA), N95 filtering facepiece respirators (FFR, 3M, Saint Paul, MN, USA ), work gloves (leather), and ear muffs (Peltor, 3M, Saint Paul, MN, USA).

The first workplace (Case 1) consists of a research and development facility shown in schematics in Fig. 1(A). The rework by sanding of the composite material included a pneumatic driven angle grinder (GTG21 8500 RPM, Atlas Copco, Nacka, Sweden) and random orbital sander (LST 31, Atlas Copco, Nacka, Sweden) equipped with on-tool extraction Local Exhaust Ventilations (LEVs). Sanding discs were new and grit sizes used for the angle grinder were 36 and 60 (named as AG36 and AG60) and for the random orbital sander were 40 and 120 (named as ROS40 and ROS120). After each activity settled dust on the workstation was removed with a vacuum cleaner. The effect of working practices on particle emissions was studied by simulating a careless working style as if carried out by an inexperienced worker. In the careful working style, as carried out by an experienced worker, the instrument was tightly fitted onto the substrate with a constant pressure and smooth movements. The inexperienced working style included loose pressure onto the substrate and frequently lifting the instrument during sanding. The LEV efficiency was studied by reducing the LEV flow rate stepwise from maximum (4.3 m3 min–1) to zero. While the angle grinder LEV was turned off (QLEV = 0 m3 min–1) we assumed that particle penetration into the near field was maximum. The LEV efficiency was estimated by normalizing the near field concentrations measured at different QLEV values with the particle concentration measured at QLEV = 0 m3 min–1.

In the second facility (Case 2), composite materials were reworked by sanding using angle grinders without LEVs (Fig. 1(B)). Workers were monitored conducting routine work. The facility was split in three parts lengthwise by plastic sheets hanging from the ceiling. A tent (length 3 m, width 2 m, and height 2 m) was present to enclose the workstation. The tent was a passive shielding without ventilation, a closed floor, or an airlock around the tent entrance. As the reworking of the composite structure progressed the tent was moved in order to follow the working site. The near field Dust Monitor (DM) and filter sampler were placed inside the tent while the near field Fast Mobility Particle Sizer (FMPS) was sampling from outside of the tent with a 60 cm conductive tube with an 8 mm inner diameter. The tent protection factor, i.e., penetration of the process particles from the tent to the rest of the facility, was defined by the difference between the near field particle concentration and particle concentration outside the enclosure using the near field FMPS. The protection factor was calculated as

NF Facility

Facility

C CPF

C

(1)


Fig. 1. The experimental set-up in (A) case 1 and (B) case 2. The instrument location is indicated in relation to the working area and in the near field and far field, respectively.

In both cases the process particle concentrations were calculated by subtracting the averaged pre-activity concentration from the corresponding averaged total particle concentration. During measurements, other work activities were carried out elsewhere in the facilities. Composites

In case 1, the composite component (50 × 50 cm) material was made from a two part layering and reinforced with

glass and carbon fibres (Devold AMT, Langevåg, Norway). The first part of the layer contained 30–40% carbon fibres and 60–70% glass fibres (so-called E-glass) and the second part of the layer was a fibre fabric consisting of 19.5–30% carbon fibre, 39–52.5% glass fibre, and 25–35% polyamide/ polypropylene matrix. The second layer made up to 65–75% of the composite material in mass. In case 2, the substrate was a windmill wing composed of approximately 70% glass fibre and 30% epoxy. The glass fibres have been


classified as Group 2B by IARC as possibly carcinogenic to humans (IARC, volume 81, 2002). Instruments

Online particle number concentrations were monitored using two Condensation Particle Counters (CPC; models 3007 and 3022A TSI Inc., Shoreview, MN, USA) with a detection range from 10 and 7 nm, respectively, to > 1 µm, and a Grimm model 5.403 (GCPC; Grimm Aerosoltechnik, Ainring, Germany) with a detection range of 4.5 nm to > 3 µm. Particle mobility size distributions were measured using two FMPSs (model 3091, TSI Inc., Shoreview, MN, USA) in 32 size channels ranging from 5.6 nm to 560 nm in 1 s intervals. Optical particle size distributions were measured using two DMs (model 1.109, Grimm Aerosoltechnik, Ainring, Germany) in 31 size channels between 250 nm and > 30 µm in 6 s intervals. Aerodynamic particle size distributions were measured using an Electrical Low Pressure Impactor (ELPI; ELPI+, Dekati Ltd., Tampere, Finland) in 14 size channels between 6 nm and 10 µm with 1 s intervals.

Samples for Transmission Electron Microscopy (TEM) analysis were retrieved using a Micro-INertial Impactor (MINI; Kandler et al., 2007), equipped with Ni-TEM grids (Plano, Wetzlar, Germany) on two stages with a nominal cut off diameter of 0.7 and 0.05 µm, respectively (see details from Lieke et al., 2013). Deposited dust from the process was collected passively and dropped on carbon substrate and silicon wafers for further analysis by TEM and EDX. Respirable mass concentrations (see definition CEN, 1993 or ACGIH, 2005) were assessed by collecting particles on 37 mm filters (Teflon with a pore size of 1 µm and cellulose-nitrate with a pore size of 0.8 µm; Millipore, Billerica, MA, USA) using a Triplex Cyclone (BGI Inc., Waltham, MA, USA, Qs = 1.05 L min–1) for the near field and far field, and a BGI Model GK2,69 (Qs = 4.2 L min–1; Stacey et al., 2014) for personal sampling. Mass distribution was measured using a gravimetrical cascade impactor (DGI, model DGI-1571, Dekati Ltd., Tampere, Finland) with a sample flow of 70 L min–1.

Air velocity in the room was measured using a hot-wire probe (435-2, Testo, Lenzkirch, Germany, measurement range 0.03–20 m s–1), and the air flow of the Local Exhaust Ventilation (LEV) was measured using a vane probe (435-2, Testo, Lenzkirch, Germany, measurement range 0.6–40 m s–1). The humidity and temperature was measured using a Gemini tiny tag plus (TGP-1500, Gemini Data Loggers Ltd, West Sussex, UK). Calculation of Regional Deposition Rates

The regional deposition rates (µg min–1) were calculated from the gravimetrical mass size distribution measured by the DGI under the assumption that the respirable mass concentrations had the same mass size distribution as assessed from gravimetrical filter samples. The mass size distributions were multiplied by the respiratory minute volume during light exercise (25 L min–1) and the regional deposition fractions while air is inspired and expired (ICRP, 1994; Hinds, 1999). The method provides a first approximation to estimate magnitude of the deposited dose in the respiratory tract

(see details from e.g., by Koivisto et al., 2012a). RESULTS Exposure Measurements in Case 1

In this facility, the average relative humidity was 57% and the average temperature was 22°C. Without reworking activity, the air velocity inside the facility was below the instrument detection limit of 0.03 m s–1. Figs. 2(A) and 2(B) show the particle number concentrations and Figs. 3(A)–3(D) show the process particle size distribution. During the work day, there were eight processes which were named according to the tool and type of used sanding discs, as AG36, AG60, ROS40, and ROS120. Particles from the processes were released in bursts in the near field while in the far field particle concentration changes were slower. The pre-activity concentrations were defined as an average concentration from 5 minutes prior to each activity (Figs. 2(A) and 2(B)). These concentrations consist of background particles from other processes occurring in the facility, outdoor particles penetrating indoors via ventilation, and residual particles from earlier composite reworking processes. At the end of the work day pre-activity particle concentrations were twice as high as before the start of the experiments and behaving similarly both in the near field and far field as measured by the FMPSs. The increasing trend was attributed to an increase in particles smaller than 40 nm (see Supplementary information Fig. S1) and were not detected by the DMs (Figs. 2(A) and 2(B)).

The angle grinder processes increased the process particle number concentrations in the near field particle from 6.0 × 103 to 7.5 × 104 cm–3 and in the far field from 5.0 × 103 to 1.3 × 104 cm–3 (Table 1). Negative concentration values mean that during process the average concentration was below pre-activity concentration level. The process particle concentrations calculated from the DM measurements were negligible because the emitted particles were mainly below 100 nm in diameter (Figs. 3(A)–3(D)). The ROS40 and ROS120 processes did not increase the near field or far field total particle concentration considerably above the pre-activity particle concentration (Table 1, Figs. 2(A) and 2(B), process particle size distributions are not shown).

Analysis of the TEM samples (Figs. 4(A) and 4(B)) showed that particles originating from the sanding were found in agglomerates, some with parts of fibres protruding from the collected epoxy fragments (Fig. 4(A)). Particles allocated to the sanding process in the sub-micron size range consisted mainly of carbon-containing epoxy material, whereas fibres (glass and carbon) are present in the entire sampled size range. Droplet residues were found to include fibres and particles of 100 nm and smaller (Fig. 4(B)). In the scope of this study, it could not be concluded whether these nano-sized fibres were carbon- or glass fibres.

Process specific mass concentrations were defined from samples collected for 3 times 10 minutes (Table 2). The ROS40 and ROS120 processes yielded higher near field mass concentrations than during the AG processes. During ROS40, the far field mass concentration was 1.058 mg m–3 which is exceptionally high compared to other far field mass


Fig. 2. Case 1; total number concentrations measured from the (A) near field and (B) far field. The events AGx-1 and AGx-2 are done using a careful working style, whereas the events AGx-3 are done using a careless working style. Case 2; total number concentrations measured from the (C) near field (D) and far field. Blue horizontal lines indicate times when the FMPS inlet was moved outside the tent to determine the tent protection factor. Pre-activity concentration averaging interval is indicated with horizontal magenta lines. Dotted and dashed lines indicate start and end of activity, respectively.


Fig. 3. Case 1; averages of the angle grinder process particle size distributions measured from the near field (A) AG36 and (B) AG60, and from the far field (C) AG36 and (D) AG60. (E) Case 2; average process particle size distributions measured from the near field, FMPS data from above 250 nm is not seen as these yielded negative particle concentration values after neglecting pre-activity concentrations from the process particle concentrations. Dashed lines show the ± 1 standard deviation of the concentration, where truncated lines indicate negative values.

Table 1. Case 1 and 2; mean process particle concentrations. WD denotes the average process particles concentration of the whole workday (WD).

Process FMPS Near field, ×103 [cm–3] Far field, ×103 [cm–3]

DM CPC ELPI FMPS DM GCPC Case 1

AG36 6.9 –0.0001 6.0 10 13 0.0096 7.8 AG60 39 0.043 23 75 10 0.0087 5.0 ROS40 –0.043 –0.0008 –0.007 –0.17 –0.02 –0.0008 –0.19 ROS120 0.64 0.0002 0.83 1.2 0.78 –0.0004 0.71

Case 2 Activity 1700 4.6 - - - –0.034 –0.24 Pre-activity 8.2 0.43 - - - 0.300 7.6

concentrations (Table 2). Though the DM did not measure elevated concentrations (Figs. 2(A) and 2(B)) this is most likely due to an instrumental artefact due to the positioning of the instrument (The DM was too far away from the process

and was moved closer after first measurements.). The personal mass concentrations were higher than measured from the near field. The ELPI particle number size distributions were converted to mass size distributions by assuming spherical


Fig. 4. Case 1; TEM micrographs from case 1 showing (A) larger sanding fragments with fibres imbedded into the epoxy matrix, insert shows a close up of the fibres. (B) Fibrous material in droplet residual of unknown material.

Table 2. Case 1 and 2; mean mass concentration calculated from the mass collected on filter samplers in the near field and far field and the deposition rate.

Process Personal, [mg m–3] Near field, [mg m–3] Far field, [mg m–3] Deposition rate, [µg min–1]a

Case 1 AG36 < 0.18 < 0.18 < 0.18 - AG60 < 0.18 < 0.18 < 0.18 - ROS40 0.280 < 0.18 1.058 0.4 ROS120 0.417 < 0.18 < 0.18 0.5

Case 2 0.260 0.240 < 0.031 0.3 aDeposition rates were calculated without taking into account any protection factor from respirators.

particles and unit density (1 g cm–3) which were compared with the mass distribution quantified using the DGI. It was found that the shape of the ELPI and DGI mass distributions were similar up to 1 µm (Fig. 5). The effect of working practices on particle emissions was done during AG36-3 and AG60-3 work activities (Figs. 2(A) and 2(B)). The events AGx-1 and AGx-2 are done using a careful working style, whereas the events AGx-3 are done using a careless working style. According to the near field CPC, the careless work style increased the total particle concentration by 1.1% for AG36-3 and 14.1% for AG60-3 when compared to AG36-1,2 and AG60-1,2, respectively. Local Instrument Ventilation

It was found that overall particle removal efficiency did not depend strongly on particle size (Fig. S2(A)). Half of the particles were removed at QLEV = 3.2 m3 min–1 and nearly all particles were removed at the highest flow rate of 4.3 m3 min–1 (Fig. S2(B)). The near field and far field particle number concentration time series shows how the LEV effects on exposure levels (Figs. S2(C) and S2(D)). Exposure Measurements in Case 2

In the near field, the average relative humidity was 28% and the average temperature was 16°C. Without reworking activity, the near field air velocity measured was below the

instrument detection limit of 0.03 m s–1. Figs. 2(C) and 2(D) show the particle number concentrations and Fig. 3(E) shows the process particle size distribution. Reworking process with the angle grinder increased rapidly the near field particle concentrations measured inside the tent, and at the end of the process, concentrations started to decay slowly (Figs. 2(C) and 2(D)). The near field concentrations measured by the FMPS and DM increased respectively up to 3 and 2 orders of magnitude from the pre-activity concentration levels (Table 1). Process particle size distribution measured form the near field were similar in shape as in Case 1 but concentrations were approximately 2 magnitudes higher (Fig. 3(E)). The personal mass concentration was 0.260 mg m–3. The near field concentration was clearly higher than in Case 1 and the far field concentration was below the detection limit (Table 2).

Measuring the particle concentration outside the tent (blue horizontal lines in Fig. 2(c)) showed an average concentration of 3.9 × 104 cm–3. At times ± 5 min of these measurements the average particle concentration inside the tent was 1.7 × 106 cm–3. Thus, the tent average protection factor was in 117 with a minimum value of 12 and a maximum value of 275. Just outside the tent the concentration was 5 times higher than the average far field particle concentration of 7.4 × 103 cm–3 showing that a fraction of process particles were leaking from the tent to the rest of the facility.


Fig. 5. Case 1; mass distribution measured by the DGI and calculated from the ELPI particle number size distribution averaged over the respective DGI sampling time. Horizontal lines show sampling size range of the DGI and vertical lines show the standard deviation of the mass concentration measured by the DGI. Dashed lines show the ± 1 standard deviation of the ELPI concentration, where truncated lines indicate negative values.

Personal Exposure and Deposited Dose Rates in the Respiratory Tract

During the rework processes with the angle grinder or random orbital sander and considering the size range of 0.2–1 µm, approximately 80% of the mass concentration was in the size range of 0.2 to 0.5 µm while 20% was in the size range of 0.5 to 1 µm as measured by the DGI (Fig. 5). The process specific deposition rates were calculated using the mass concentration listed in Table 2, and by assuming that the mass distribution in the personal exposure is similar as measured by the DGI. This gives mean deposition rates of 0.4, 0.5, and 0.3 µg min–1 for ROS40 and ROS120 in case 1 and for Case 2, respectively. The mass regional deposition was: 50% in head airways, 5% in the tracheobronchial region, and 45% in the alveolar region. The particles were deposited in the same way in the respiratory system because the mass size distribution was assumed to be the same in all processes. For an eight hour working day, the mean deposition rates were 144 to 240 µg day–1. DISCUSSION

This study shows how exposure in the near field and far field are affected by working style, enclosing the near field with a tent, and instruments LEVs. These factors are essential for pollution control and for mathematical exposure modelling (e.g., Fransman et al., 2011) which are needed in increasing amounts to fulfil the Registration, Evaluation, Authorization, and Restriction of Chemicals (REACH; European Union, 2006) regulation.

Composite rework using the random orbital sander

influence on the near field concentration levels was not measureable while the angle grinder process increase was clear (Table 1). In an open workstation, the far field particle concentrations were increased up to 1.3 × 104 cm–3. This means that a person 6 m from the working station is subjected to a large and constant amount of airborne particles originating from the activity which may even lead to a higher process particle exposure than in the near field. The far field exposure is further increased if the LEV efficiency is reduced (Figs. S2(C) and S2(D)).

The LEV efficiency measurements showed that the near field particle number concentration (Fig. S2(C)) is largely unaffected by the LEV flow rate below 3.8 m3 min–1, whereas there is a clear indication (Fig. S2(D)), that with higher LEV flow rate less material reaches the far field measurement position with only trace amount still present at the maximum flow rate of 4.3 m3 min–1. This is consistent with what is reported by Akbar-Khanzadeh et al. (2007) for silica dust during the grinding of concrete. We found that at the maximum LEV flow rate particle emissions were negligible, and during the process, the near field particle concentrations remained at the pre-activity level (Figs. S2(C) and S2(D)).

The tent was found to limit efficiently particles dispersion into the far field (Figs. 2(C) and 2(D), Table 1). However, the near field concentrations were increased up to 2 to 3 magnitudes as compared to the near field exposure in Case 1. This was because the angle grinder was not using LEV and the tent reduced the dilution volume. The tent protection factor was on average 117 which is 3 to 12 times higher than the protection factor for the N95 filtering facepiece respirators which was reported by Reponen et al. (2011).


They measured in laboratory conditions using sub-micrometre particles N95 filtering facepiece respirators protection factors varying between on average 10 to 40, where the most penetrating particle sizes was approximately 70 nm.

The process-specific dose rates can be used to estimate deposited doses in different exposure scenarios (see Koivisto et al., 2012a, b, 2014; Mølgaard et al., 2015). Those can be further used in the risk assessment when particle hazards are assessed with in vivo experiments for specific toxicological endpoints and translated to the human equivalent dose-response (ECETOC, 2010; Ling et al., 2011; ECHA, 2012). In previous studies, considering deposited dose rates, the gravimetrical mass size distribution was not measured. Here, the mass size distribution was measured by using the DGI and then we assumed that the gravimetrical mass samples in Table 2, have a similar mass size distribution. This assumption was most likely valid because particle size distributions were similar in shape between different processes (Fig. 3). Fig. 5 shows that using the calculated mass size distribution from the ELPI number size distribution would here lead to misleading dose characterization. For particles larger than 1 µm the ELPI overestimates mass concentrations. This was most likely because of the poor large particle detection sensitivity of the ELPI (Dekati, 2006). Because the mass concentration level was low and sampling times were short, five out of ten samples were below detection limit. Calculated deposition rates were on average 0.4 µg min–1 which is clearly higher than modelled dose rate of 0.026 µg min–1 of urban air sub-micrometre particles for a male with typical habits living in Helsinki, Finland (Hussein et al., 2013). The use of respirator was not taken into account in this study since the respirator performance was not determined. For example Koivisto et al. (2015) showed that the respirator protection factors may be significantly higher than the assigned protection factor. The considerable decrease in density was most likely due to an artefact from the differences between instruments collection stages size ranges (i.e., different response functions) and inaccuracy of the ELPI particle number to mass conversion because of the particles morphology (Fig. 4). Furthermore, the density of the collected particles is hard to estimate due to their mixing state and unknown composition in the small size ranges (100–500 nm). The droplet residuals found in the TEM samples included particles and fibres. The internal mixing of particles and fibres makes estimations of particle density highly speculative as we cannot evaluate whether the nanoparticles and fibres were mixed and trapped inside the droplets upon collection, and thus were airborne before as single nanoparticles and nanofibers, or if they were trapped before collection. CONCLUSION

In this study we assessed workers exposures in the near field and far field during reworking of glass- and carbon fibre-reinforced composites in two separate facilities and evaluated methods for reducing potential exposure. Sanding processes increased primarily particle number concentrations below 100 nm in diameter while the mass

median diameter was 336 nm. Without enclosure, process particle concentrations were 3.9 × 104 cm–3 in the near field and 1.3 × 104 cm–3 in the far field. Therefore workers in the same area may not be aware of being exposed to the process particles and the need of wearing protective outfits. With the enclosure around the process area, the process particle concentrations were 1.7 × 106 cm–3 in the near field, while the concentration in the far field was negligible. Calculated deposition mass dose rates varied from 0.3 to 0.5 µg min–1. The working style was accountable for an increase of process particle concentration between 1.1 and 14.1% when comparing a careful working style with a careless working style. Efficient local exhaust ventilation was found to be a crucial part of removing particulate matter. The local exhaust ventilation particle removal efficiency was found to be sensitive to the ventilation flow rate as a 28% reduction in flow rate reduced the removal efficiency to less than 50%. Without enclosing the near field, process emissions were measured in the far field and timed with the activities. Introduction of an enclosure around the near field was found to efficiently prevent the process particles from dispersing into the far field where the particle concentration remained at pre-activity particle concentrations. The enclosure added an average protection factor of 117. Reducing the exposure of process particles to workers in the far field can be achieved by using a careful working style, enclosing the work station, and having sufficient local exhaust instrument ventilation. All have been shown to contribute to a reduction in the number of particles being released into the surrounding indoor environment. ACKNOWLEDGEMENTS

[Following acknowledgement waiting for European Commission approval: “The research leading to these results has received funding from the European Community's Seventh Framework Programme (FP7/2007-2013) under NanoValid (grant agreement N°263147)”] and this work was supported by a grant from the Danish Centre for Nanosafety (grant agreement N°20110092173/3) by the Danish Working Environment Research Fund. SUPPLEMENTARY MATERIALS

Supplementary data associated with this article can be found in the online version at http://www.aaqr.org. REFERENCES Akbar-Khanzadeh, F., Milz, S., Ames, A., Susi, P.P., Bisesi,

M., Khuder, S.A. and Akbar-Khanzadeh, M. (2007). Crystalline Silica Dust and Respirable Particulate Matter during Indoor Concrete Grinding - Wet Grinding and Ventilated Grinding Compared with Uncontrolled Conventional Grinding. J. Occup. Environ. Hyg. 4: 770–779.

American Conference of Governmental Industrial Hygienists (ACGIH) (2005). TLVs and BEIs: Based on the Documentation of the Threshold Limit Values for


Chemical Substances and Physical Agents and Biological Exposure Indices. Cincinnati, OH.

Bello, D., Wardle, B.L., Yamamoto, N., Guzman deVilloria, R., Garcia, E.J., Hart, A.J., Ahn, K., Ellenbecker, M.J. and Hallock, M. (2008). Exposure to Nanoscale Particles and Fibers during Machining of Hybrid Advanced Composites Containing Carbon Nanotubes. J. Nanopart. Res. 11: 231–249.

Bello, D., Wardle, B.L., Zhang, J., Yamamoto, N., Santeufemio, C., Hallock, M. and Virji, M.A. (2010). Characterization of Exposures to Nanoscale Particles and Fibers during Solid Core Drilling of Hybrid Carbon Nanotube Advanced Composites. Int. J. Occup. Environ. Health 16: 434–450.

Cena, L.G. and Peters, T.M. (2011). Characterization and Control of Airborne Particles Emitted During Production of Epoxy/Carbon Nanotube Nanocomposites. J. Occup. Environ. Hyg. 8: 86–92.

Dekati (2006). A.1.9 Measurement Accuracy of ELPI. In ELPI User Manual, Ver 4.01, Dekati Ltd., p. A10–A11.

DeLorme, M.P., Muro, Y., Arai, T., Banas, D.A., Frame, S.R., Reed, K.L. and Warheit, D.B. (2012). Ninety-Day Inhalation Toxicity Study with a Vapor Grown Carbon Nanofiber in Rats. Toxicol. Sci. 128: 449–460.

ECETOC (2004) ECETOC Targeted Risk Assessment. Technical Report No. 93. European Centre for Ecotoxicology and Toxicology of Chemicals, Brussels, Belgium.

European Centre for Ecotoxicology and Toxicology of Chemicals (ECETOC) (2010). Guidance on Assessment Factors to Derive a DNEL. Technical Report No. 110, http://members.ecetoc.org/Documents/Document/20110131112906-ECETOC_Technical_Report_ 110.pdf, Last Access: 13 May 2014.

European Chemical Agency (ECHA) (2012). Guidance on Information Requirements and Chemical Safety Assessment, Ch. R.8. http://echa.europa.eu/documents/ 10162/13632/information_requirements_r8_en.pdf, Last Access: on 13 May 2014.

European Committee for Standardization (CEN) (1993). Workplace Atmospheres-size Fraction Definitions for Measurement of Airborne Particles (Report No. BS EN 481:1993). CEN, British Standards Institute, London, England. ISBN 0-580-221407.

European Union (2006). Regulation (EC) No 1907/2006 of the European Parliament and of the Council of 18 December 2006 concerning the Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH). Official Journal of the European Union L136: 3–280. http://www.reach-compliance.eu/english/legislation/docs/launchers/.

Fransman, W., Van Tongeren, M., Cherrie, J.W., Tischer, M., Schneider, T., Schinkel, J., Kromhout, H., Warren, N., Goede, H. and Tielemans, E. (2011). Advanced Reach Tool (Art): Development of the Mechanistic Model. Ann. Occup. Hyg. 55: 957–979.

Ging, J., Tejerina-Anton, R., Ramakrishnan, G., Nielsen, M., Murphy, K., Gorham, J.M., Nguyen, T. and Orlov, A. (2014). Development of a Conceptual Framework for

Evaluation of Nanomaterials Release from Nanocomposites: Environmental and Toxicological Implications. Sci. Total Environ. 473–474: 9–19.

Gomez, V., Levin, M., Saber, A.T., Irusta, S., Dal Maso, M., Hanoi, R., Santamaria, J., Jensen, K.A., Wallin, H. and Koponen, I.K. (2014). Comparison of Dust Release from Epoxy and Paint Nanocomposites and Conventional Products during Sanding and Sawing. Ann. Occup. Hyg. 8: 983–994.

Gutiérrez, E. and Bono, F. (2013). Review of Industrial Manufacturing Capacity for Fibre-reinforced Polymers as Prospective Structural Components in Shipping Containers. JRC Scientific and Policy Reports, JRC77823

Hinds, W.C. (1999). Respiratory Deposition. In Aerosol Technology: Properties, Behavior, and Measurement of Airborne Particles, 2nd ed., John Wiley & Sons Inc., New York, USA, pp. 233–259.

Hussein, T., Londahl, J., Paasonen, P., Koivisto, A.J., Petaja, T., Hameri, K. and Kulmala, M. (2013). Modeling Regional Deposited Dose of Submicron Aerosol Particles. Sci. Total Environ. 458–460: 140–149.

Kandler, K., Benker, N., Bundke, U., Cuevas, E., Ebert, M., Knippertz, P., Rodríguez, S., Schütz, L. and Weinbruch, S. (2007). Chemical Composition and Complex Refractive Index of Saharan Mineral Dust at Izana, Tenerife (Spain) Derived by Electron Microscopy. Atmos. Environ. 41: 8058–8074.

Kingston, C., Zepp, R., Andrady, A., Boverhof, D., Fehir, R., Hawkins, D., Roberts, J., Sayre, P., Shelton, B., Sultan, Y., Vejins, V. and Wohlleben, W. (2014). Release Characteristics of Selected Carbon Nanotube Polymer Composites. Carbon 68: 33–57.

Koivisto, A.J., Aromaa, M., Mäkelä, J.M., Pasanen, P., Hussein, T. and Hämeri, K. (2012a). Concept to Estimate Regional Inhalation Dose of Industrially Synthesized Nanoparticles. ACS Nano 6: 1195–1203.

Koivisto, A.J., Lyyränen, J., Auvinen, A., Vanhala, E., Hämeri, K., Tuomi, T. and Jokiniemi, J. (2012b). Industrial Worker Exposure to Airborne Particles during the Packing of Pigment and Nanoscale Titanium Dioxide. Inhalation Toxicol. 24: 839–849.

Koivisto, A.J., Palomäki, J.E., Viitanen, A.-K., Siivola, K.M., Koponen, I.K., Yu, M., Kanerva, T.S., Norppa, H., Alenius, H.T., Hussein, T., Savolainen, K.M. and Hämeri, K.J. (2014). Range-Finding Risk Assessment of Inhalation Exposure to Nanodiamonds in a Laboratory Environment. Int. J. Environ. Res. Public Health 11: 5382–5402.

Koivisto, A.J., Aromaa, M., Koponen, I.K., Fransman, W., Jensen, K.A., Mäkelä, J.M. and Hämeri, K.J. (2015) Workplace Performance of a Loose-fitting Powered Air Purifying Respirator during Nanoparticle Synthesis. J. Nanopart. Res. 17: 177.

Koponen, I.K., Jensen, K.A. and Schneider, T. (2009). Sanding Dust from Nanoparticle-containing Paints: Physical Characterisation. J. Phys: Conf. Ser. 151: 012048.

Koponen, I.K., Jensen, K.A. and Schneider, T. (2011). Comparison of Dust Released from Reworking Conventional and Nanoparticle-doped Wall and Wood


Coatings. J. Exposure Sci. Environ. Epidemiol. 21: 408–418.

Lieke, K.I., Rosenørn, T., Pedersen, J., Larsson, D., Kling, J., Fuglsang, K. and Bilde, M. (2013). Micro- and Nanostructural Characteristics of Particles before and after an Exhaust Gas Recirculation System Scrubber. Aerosol Sci. Technol. 47: 1038–1046.

Ling, M.P., Chio, C.P., Chou, W.C., Chen, W.Y., Hsieh, N.H., Lin, Y.J. and Liao, CM. (2011). Assessing the Potential Exposure Risk and Control for Airborne Titanium Dioxide and Carbon Black Nanoparticles in the Workplace. Environ. Sci. Pollut. Res. Int. 18: 877–889.

Luchtel, D.L., Martin, T.R. and Boatman, E.S. (1989). Response of the Rat Lung to Respirable Fractions of Composite Fiber-Epoxy Dusts. Environ. Res. 48: 57–69.

Midtgård, U. and Jelnes, J.E. (1991). Toxicology and Occupational Hazards of New Materials and Processes in Metal Surface Treatment, Powder Metallurgy, Technical Ceramics, and Fiber-reinforced Plastics. Scand. J. Work Environ. Health 17: 369–379.

Mølgaard, B., Viitanen, A.K., Kangas, A., Huhtiniemi, M., Larsen, S.T., Vanhala, E., Hussein, T., Boor, B.E., Hämeri, K. and Koivisto, A.J. (2015) Exposure to Airborne Particles and Volatile Organic Compounds from Polyurethane Molding, Spray Painting, Lacquering, and Gluing in a Workshop. Int. J. Environ. Res. Public Health 12: 3756–3773.

Reponen, T., Lee, S.A., Grinshpun, S.A., Johnson, E. and McKay, R. (2011). Effect of Fit Testing on the protection Offered by N95 Filtering Facepiece Respirators Against Fine Particles in a Laboratory Setting. Ann. Occup. Hyg. 55: 264–271.

Saber, A.T., Jacobsen, N.R., Mortensen, A., Szarek, J., Jackson, P., Madsen, A.M., Jensen, K.A., Koponen, I.K., Brunborg, G., Gutzkow, K.B., Vogel, U. and Wallin, H. (2012a). Nanotitanium Dioxide Toxicity in Mouse Lung Is Reduced in Sanding Dust from Paint. Part. Fibre Toxicol. 9: 4, doi: 10.1186/1743-8977-9-4.

Saber, A.T., Koponen, I.K., Jensen, K.A., Jacobsen, N.R., Mikkelsen, L., Møller, P., Loft, S., Vogel, U. and Wallin, H. (2012b). Inflammatory and Genotoxic Effects of

Reworking Dust Generated from Nanoparticle-containing Paints and Lacquers. Nanotoxicology 6: 776–788.

Smulders, S., Luyts, K., Brabants, G., Landuyt, K.V., Kirschhock, C., Smolders, E., Golanski, L., Vanoirbeek, J. and Hoet, P.H. (2014). Toxicity of Nanoparticles Embedded in Paints Compared with Pristine Nanoparticles in mice. Toxicol. Sci. 141: 132–140.

Stacey, P., Lee, T., Thorpe, A., Roberts, P., Frost, G. and Harper, M. (2014). Collection Efficiencies of High Flow Rate Personal Respirable Samplers When Measuring Arizona Road Dust and Analysis of Quartz by X-ray Diffraction. Ann. Occup. Hyg. 58: 512–523.

The International Commission on Radiological Protection (ICRP) (1994). Human Respiratory Tract Model for Radiological Protection. Ann. ICRP 24: 1–482.

Thostenson, E.T., Li, C. and Chou T.W. (2005). Nanocomposites in Context. Compos. Sci. Technol. 65: 491–516.

Waritz, R.S., Ballantyne, B. and Clary, J.J. (1998). Subchronic Inhalation Toxicity of 3.5-micron Diameter Carbon Fibers in Rats. J. Appl. Toxicol. 18: 215–23.

Wohlleben, W., Brill, S., Meier, M.W., Mertler, M., Cox, G., Hirth, S., von Vacano, B., Strauss, V., Treumann, S., Wiench, K., Ma-Hock, L. and Landsiedel, R. (2011). On the Lifecycle of Nanocomposites: Comparing Released Fragments and Their in-Vivo Hazards from Three Release Mechanisms and Four Nanocomposites. Small 7: 2384–2395.

Zaman, A., Gutub, S.A. and Wafa, M.A. (2013). A Review on FRP Composites Applications and Durability Concerns in the Construction Sector. J. Reinf. Plast. Compos. 32: 1966–1988.

Zimmer, A.T. and Maynard, A.D. (2002). Investigation of the Aerosols Produced by a High-speed, Hand-held Grinder Using Various Substrates. Ann. Occup. Hyg. 46: 663–672.

Received for review, February 10, 2015 Revised, May 6, 2015

Accepted, May 6, 2015