Airborne Nanoparticle Exposures while Using Constant-Flow, Constant-Velocity, and Air-Curtain- Isolated Fume Hoods

Transcription

1 Ann. Occup. Hyg., Vol. 54, No. 1, pp , 2010 Ó The Author Published by Oxford University Press on behalf of the British Occupational Hygiene Society doi: /annhyg/mep074 Airborne Nanoparticle Exposures while Using Constant-Flow, Constant-Velocity, and Air-Curtain- Isolated Fume Hoods SU-JUNG (CANDACE) TSAI 1 *, RONG FUNG HUANG 2 and MICHAEL J. ELLENBECKER 1 1 Center for High-rate Nanomanufacturing, University of Massachusetts Lowell, One University Avenue, Lowell, MA 01854, USA; 2 Department of Mechanical Engineering, National Taiwan University of Science and Technology, Taipei, Taiwan, Republic of China Received 10 April 2009; in final form 24 September 2009; published online 20 November 2009 Tsai et al. (Airborne nanoparticle exposures associated with the manual handling of nanoalumina and nanosilver in fume hoods. J Nanopart Res 2009; 11: ) found that the handling of dry nanoalumina and nanosilver inside laboratory fume hoods can cause a significant release of airborne nanoparticles from the hood. Hood design affects the magnitude of release. With traditionally designed fume hoods, the airflow moves horizontally toward the hood cupboard; the turbulent airflow formed in the worker wake region interacts with the vortex in the constant-flow fume hood and this can cause nanoparticles to be carried out with the circulating airflow. Airborne particle concentrations were measured for three hood designs (constant-flow, constant-velocity, and air-curtain hoods) using manual handling of nanoalumina particles. The hood operator s airborne nanoparticle breathing zone exposure was measured over the size range from 5 nm to 20 mm. Experiments showed that the exposure magnitude for a constant-flow hood had high variability. The results for the constant-velocity hood varied by operating conditions, but were usually very low. The performance of the air-curtain hood, a new design with significantly different airflow pattern from traditional hoods, was consistent under all operating conditions and release was barely detected. Fog tests showed more intense turbulent airflow in traditional hoods and that the downward airflow from the double-layered sash to the suction slot of the air-curtain hood did not cause turbulence seen in other hoods. Keywords: airborne nanoparticle; exposure; fume hood; handling method; nanoalumina INTRODUCTION The use of nanoparticles, both in the laboratory and in the manufacturing, is growing rapidly. However, the proper use of local exhaust ventilation when processing and handling nanoparticles has not been studied widely. Current best practice guidelines usually recommend the use of laboratory fume hoods when handling dry nanopowders (BSI, 2007; University of California Berkeley, 2007). These guidelines assume *Author to whom correspondence should be addressed. Tel: þ ; fax: þ ; SuJung_Tsai@uml.edu that properly designed and operated fume hoods will offer the users adequate protection against nanoparticle exposure. It is likely that all commonly available hood designs will be used for handling nanoparticles. Most laboratories now use some variation on the constant-flow hood, the constant-velocity hood, or the bypass hood (DiBerardinis et al., 1993). DiBerardinis et al. (1993) and Saunders (1993) describe basic elements that all fume hoods should incorporate. The hood face velocity, defined as the average velocity through the hood sash opening, is likely the single design factor that will most strongly affect hood performance. The most effective hood face velocity has been the subject of research for at 78

2 Airborne nanoparticle exposures while using fume hoods 79 least 50 years (Schulte et al., 1954; ACGIH, 1966). The current consensus on proper hood face velocity, as given by scientists (DiBerardinis et al., 1993; Saunders, 1993; Altemose et al., 1998; Burgess et al., 2004), government agencies (OSHA, 1992), and professional organizations (ANSI/AIHA, 2003; AC- GIH, 2007), is that the average face velocity should be in the range of m s 1. In addition, procedures to evaluate the performance of fume hood have been developed using standardized tracer gas tests (BSI, 1994; ANSI/ ASHRAE, 1995; ANSI/AIHA, 2003; CEN, 2003), and leakage under various conditions has been documented (Kim and Flynn, 1991a,b, 1992; Flynn and Ljungqvist, 1995; Johnson and Fletcher, 1996). Hoods that are tested and pass a standard tracer gas evaluation may still not offer adequate protection when nanoparticles are released inside the hood. The ASHRAE 110 (ANSI/ASHRAE, 1995) acceptance criterion for hood performance is a breathing zone (BZ) SF 6 concentration of 0.1 p.p.m. when SF 6 is released inside the hood at a rate of 25 g min 1. Although there are no current standards for nanoparticle exposure, acceptable BZ concentrations may be as low as the asbestos threshold limit value (0.1 particles cm 3 ) for carbon nanotubes, which have been found to possibly cause mesothelioma in mice (Poland et al., 2008), to as high as perhaps 1000 particles cm 3 for a relatively non-toxic material such as nanoalumina. In order to compare a gas to particles, the concentrations must be converted to mass. The SF 6 acceptance criterion of 0.1 p.p.m. corresponds to a mass concentration of 0.6 mg m 3. For simplicity, assume that the nanoparticles being released inside a fume hood are nanoalumina with a diameter of 100 nm and a density of 3600 kg m 3. A number concentration of 0.1 particles cm 3 has a mass concentration of mg m 3, and a number concentration of 1000 particles cm 3 has a mass concentration of mg m 3. If we assume that source release rate is the same for both SF 6 and nanoparticles (i.e. 25 g min 1 ), the maximum BZ concentrations can be compared directly. The maximum allowable mass concentration for the lower particle limit ( mg m 3 ) is a factor of lower than the maximum SF 6 concentration (0.6 mg m 3 ) when a hood passes the ASHRAE acceptance test, and the maximum allowable concentration for the higher particle limit ( mg m 3 ) is a factor of 310 lower than the maximum SF 6 concentration. Thus, fume hoods would have to limit nanoparticle releases by a factor of 310 to times lower than allowed by the SF 6 test to meet the range of possible maximum allowable nanoparticle BZ concentrations. Current evidence suggests that particle number and/or surface area concentration may be more important nanoparticle exposure metrics than mass (Maynard and Kuempel, 2005), although there may be circumstances where mass measurement is appropriate (Maynard and Aitken, 2007). Nonetheless, Maynard and Aitken also describe the serious shortcomings of currently available instrumentation for nanoparticle mass measurement, leaving number and surface area as the preferred exposure metrics for the near future. Techniques such as those used in this research are available for determining airborne number concentrations as a function of particle size for airborne nanoparticles released from fume hoods. At this time, however, there are no standards for acceptable nanoparticle exposure based on number or surface area concentration; given this, any measurable nanoparticle release from a fume hood must be viewed with concern. The research reported here evaluated the magnitude of exposure for several fume hood designs under various conditions to provide guidance to minimize nanoparticle release. Tsai et al. (2009) reported that the handling of dry nanoalumina and nanosilver consisting of nanosized particles inside laboratory fume hoods can result in a significant release of airborne nanoparticles from the fume hood into the laboratory environment and the researcher s BZ. Three types of commonly used fume hoods were studied. When nanoparticles were manually handled in fume hoods, the particle release to the worker s BZ was greatest when using a constant-flow hood, as compared to a bypass and a constant-velocity hood. A recent innovation in hood design is the aircurtain hood by Huang et al. (2007). This hood utilizes an air curtain, which is generated as a narrow planar jet emanating from the double-pane sash and collected by a suction slot located in the hood work surface just behind the sash doorsill. The push jet and the suction flow are designed to create an air curtain (Huang et al., 2005) on the sash plane to aerodynamically separate the interior of the cabinet from the outside atmosphere. This design does not have several features of traditional hoods. The airflow inside the hood goes downward from an open grille at the top of hood toward the floor of hood cupboard, where it enters the suction slot right behind the doorsill, along with the air jet from the doublepane sash. This hood is designed to have almost no face velocity through the sash opening, instead the air curtain remains at a constant velocity for all sash positions and serves to isolate the inside and outside of the hood. Huang et al. (2007) evaluated hood

3 80 S. J. Tsai, R. F. Huang and M. J. Ellenbecker leakage in great detail and found that this new hood design was extremely effective. Since our initial evaluation of constant-flow hoods found many shortcomings when handling nanoalumina and nanosilver (Tsai et al., 2009), the research reported here was undertaken to determine whether this new hood design gave improved performance. The objective of this research was to evaluate three different hood types in their ability to contain aluminum oxide nanomaterials during two different nanomaterial handling activities. Hood performance was established, in this regard, in two ways: (i) by comparing worker BZ exposure levels to source levels inside the hoods during two different handling activities and (ii) by visualizing airflow patterns. In addition to actual manual handling tasks, where the researcher s exposures were monitored, airflow visualization tests were performed utilizing a mannequin at the hood face. MATERIALS AND METHODS Fume hoods Particle handling was studied using a constantflow hood, a constant-velocity hood (ACGIH, 2007), and an air-curtain hood (Huang et al., 2007) (photographs of each hood and manufacturer and model specifications are provided in the on-line supplementary material Figure S01, available at Annals of Occupational Hygiene online): With no special features to help control the hood face velocity, a constant-flow hood has a constant airflow, and the hood face velocity varies inversely with the height of the sash opening. The studied constant-flow hood has a full-opensash-face dimension of 62 cm height (H) 130 cm width (W). A constant-velocity hood, also called a variable air volume hood, typically uses a motor controller to vary the fan speed as the sash is moved in order to maintain a constant hood face velocity of 0.5 m s 1 over most sash positions. The sash of the studied constant-velocity hood has dimensions of 69 cm (H) 163 cm (W). An air-curtain hood induces a push pull air curtain at the face of the hood and downward flow from a top grille, as described above. The sash of the studied air-curtain hood has dimensions of 69 cm (H) 112 cm (W). Materials Aluminum oxide (Al 2 O 3 ) nanoparticles, also called nanoalumina, manufactured using physical vapor synthesis, were used for this study (grade Al ; Nanophase Technologies Corporation, Romeoville, IL, USA) [Materials were retrieved from manufacturer s website, com/technology/capabilities.asp (2007)]. They appear roughly spherical in shape and have a reported density of 3600 kg m 3 and primary particle size ranging from 27 to 56 nm; when dried, these particles formed agglomerates in the bulk material with a nominal size of 200 nm. Nanoalumina particles were dried overnight at a temperature.150 C to remove moisture before use. For each handling, a 400-ml beaker (100 g) of nanoalumina was used. Real-time particle measurement Two instruments were operated simultaneously to record airborne particle concentrations from 5.6 nm to 20 lm. The concentration of airborne nanoparticles for diameters from 5.6 to 560 nm was measured using the Fast Mobility Particle Sizer (FMPSÒ) spectrometer (Model 3091, TSI), with 32 channels of resolution (16 channels per decade). The FMPS performs particle size classification based on differential electrical mobility classification. Particle concentration and size distribution were recorded every second. Two-meter long conductive tubing was connected to the inlet of FMPS to reach the measuring locations. The concentration of airborne particles for diameters from 0.5 to 20 lm was measured using the Aerodynamic Particle Sizer (APSÒ) spectrometer (Model 3321, TSI). The APS provides high-resolution, real-time aerodynamic measurements of particle size. Particle concentration and size distribution were recorded every second. Two-meter long conductive tubing was connected to the air inlet of APS to reach measuring locations. Particle loss in the conductive tubing is a potential source of measurement error. The study by Kumar et al. (2008) indicated that the majority of the decrease in particle number concentration was for particles,20 nm diameter, where the maximum losses are also expected due to the higher diffusivity of smaller particles (Hinds, 1999). The particle number concentration, which were obtained by integrating the particle number distribution, decreased 3, 7, 28, and 32% for tubes 5.47, 5.55, 8.90, and m long, respectively, using 1-m tubing as the baseline. The particle loss.20 nm was negligible for all lengths of tubing tested. Thus, the particle losses for the 2-m conductive tubing used in this study can be neglected due to the short tube length and since the nanoalumina particles of interest were all.20 nm diameter.

4 Airborne nanoparticle exposures while using fume hoods 81 A fog generator and a green light laser producing a laser light sheet with 500 mw at peak power and wavelength of nm (Laserglow) were used for visual tests of airflow patterns. A 163-cm height female mannequin with chest width of 25 cm was placed in front of each hood, and the two arms of the mannequin were positioned inside the hood at the workers manual handling position. Process Measurements were taken at a background location, source location and the researcher s BZ for three sash locations as described below. Experimental design and measuring locations are illustrated in Fig. 1a. Background particle concentration was measured 1 m in front of the hood. Breathing zone concentration was measured near the researcher s nose. Nanopowder handling tasks were performed in the hood on the work surface 15 cm behind the sill, and the source location was measured 8 cm above the beaker being handled. Background, source, and BZ measurements were taken before, during, and after particle handling in the hood. The concentration before handlingtaskswasusedasthebaselinetobesubtracted from the measurements during and after handling tasks to calculate the quantity of particle release caused by particle handling. Particle handling methods developed by Tsai et al. (2009) were performed by manually transferring and pouring nanoparticles between 400-ml beakers as shown in Fig. 1b,c. Experiments were performed using 100 g nanoalumina. The transferring task (Fig. 1a) was performed by using a spatula to transfer nanoparticles from one beaker to another beaker; g of nanoalumina were loaded at the open top of beaker for each spatula transfer. For the pouring task (Fig. 1b), nanoparticles were poured directly from one beaker into a second beaker at the center of the open top, so that the feeding and receiving beakers were adjacent to each other at the open edge. Pouring 100 g nanoalumina took 1 min and transferring took 4.5 min. For the constant-flow hood, sash locations for measurement were at full open, half open, and low chest height, i.e. 62-, 44-, and 16.5-cm open sash respectively. The sash locations of the constant-velocity hood were at full open, half open and low chest height of 64, 38, and 17 cm open, respectively. The sash locations of the air-curtain hood were at the height of 63, 37, and 20 cm. Aerosol nanoparticles released during handling tasks were collected at the BZ location to characterize their morphology. Sampling was performed using a new method developed by Tsai et al. (2008, 2009). Transmission electron microscope (TEM)- copper grids (400 mesh with a formvar/carbon film) were attached to 47-mm diameter polycarbonate membrane filters (0.2-lm pore size) and fiber backing filters were used to support the polycarbonate filters. The filters were placed in a cassette and samples Fig. 1. (a) Illustration of experimental design and measuring locations, (b) transferring task, and (c) pouring task.

5 82 S. J. Tsai, R. F. Huang and M. J. Ellenbecker were collected at airflow of 0.3 l min 1. Sampled particles were characterized using TEM; images of the samples were taken using a Philips EM400 TEM (Eindhoven, The Netherlands) operated at 100 kv. RESULTS Magnitude of release The changes in particle concentration at the worker s BZ measured for the three hoods by using the FMPS are presented in Fig. 2 for measurements at the three hoods; results for both the transferring and the pouring methods are shown. Each data curve is an average of three experiments performed on three different dates with variables of worker s body width, motion of different worker, and room humidity. The data for a single experiment on a certain date are the average particle concentration measured during the total time period of a single transferring or pouring task. The standard deviations shown on each curve of Fig. 2 represent the variability while handling nanoparticles for various conditions at that particular hood. Fig. 2. Particle number concentration increase (5 560 nm) at worker s BZ during handling nanoalumina particles, average of various operations. (a) Constant-flow hood, transfer; (b) constant-flow hood, pour; (c) constant-velocity hood, transfer; (d) constant-velocity hood, pour; (e) air-curtain hood, transfer; and (f) air-curtain hood, pour.

6 Airborne nanoparticle exposures while using fume hoods 83 The results while using the constant-flow hood, as seen in Fig. 2a,b, showed the highest variability in BZ concentration for all three sash locations and for both the transferring (Fig. 2a) and the pouring (Fig. 2b) tasks. The pouring task showed slightly higher exposure and variability than the transfer task. For the constant-velocity hood, particle release was barely detected while transferring nanoparticles and the variability was low as seen in Fig. 2c. However, the pouring method resulted in a more intense nanoparticle exposure as seen in Fig. 2d, with a higher variability than for the transfer task; the concentration peak was found at 200 nm particle diameter. Exposure while using the air-curtain hood was barely detected for both the transfer and the pouring tasks as seen in Fig. 2e,f; in addition, the exposure variability for both tasks was lower than for the other hood designs. The pouring task produces higher exposure variability than the transfer task in the air-curtain hood, which is consistent with the performance of the constant-flow and constant-velocity hoods. Effect of sash location The sash location affects the face velocity of the constant-flow hood and also the constant-velocity hood when it is fully open. Nanoparticle release was affected by changes in face velocity, as shown in Fig. 2. For the constant-flow hood, the highest exposure occurred at the low sash position for both the transfer and the pouring tasks. The released particles were predominantly.100 nm diameter (Fig. 2a,b). The BZ concentration decreased when the face velocity was reduced; the lowest exposure was measured at the high sash location, which had a face velocity of 0.4 m s 1. The middle sash location, which is the normal operating position, provides a face velocity of 0.6 m s 1 and results in a higher concentration for particles,50 nm diameter. The size distribution pattern of the particles from the constant volume hood varied with face velocity, i.e. very few particles of any size were released at a face velocity of 0.4 m s 1, particles of 20 nm diameter predominated at 0.6 m s 1, and particles of 200 nm diameter were released at 1.0 m s 1. The causes of this variation, which was found consistently in the repeated tests, are not known. Perhaps the higher diffusion coefficients of the smaller particles played a role; a 20-nm particle s diffusion coefficient is almost two orders of magnitude higher than that of a 200-nm particle (Hinds, 1999), and thus will diffuse more readily than larger particles. In addition, hood airflow patterns, such as turbulence at the hood face, may vary in unknown ways with changing sash location (face velocity) and affect particle release in different size ranges. Although these tests seem to indicate that a face velocity of 0.4 m s 1 is optimal for a constant volume hood, caution should be used since external airflows such as crossdrafts that were not present for the tested hood may be present in other settings and disrupt performance at this low face velocity. This hood is currently being modeled using computational fluid dynamics in an attempt to clarify the underlying mechanisms. For the constant-velocity hood, an increase in nanoparticle BZ concentration was barely detected when operating at the desired face velocity of 0.5 m s 1 while at the normal operating position (middle sash) or low sash for both the transfer and the pouring tasks (Fig. 2c,d). The exposure became unstable at the high sash, where the face velocity dropped to 0.3 m s 1. More particles were released during the pouring task, which is more energetic than transfer, and the variability of exposure level was the highest as seen in Fig. 2d. The results of air-curtain hood showed the highest stability and the lowest particle release for all the operating conditions tested, and the particle release was barely affected by sash location. The highest exposure measured with the air-curtain hood was at the low sash position using the pouring task. Since the nanometer-sized nanoalumina particles may agglomerate in the bulk powder material, limited experiments were also performed using the APS. Particle release for diameters from 5 nm to 20 lm, using the single operation of pouring for each hood and the same operator, is shown in Fig. 3. Using the more energetic tasks of pouring, the release of particles.500 nm diameter was barely detectable; in other words, particles escaping from the hoods during the handling of nanoalumina particles are almost all,500 nm. The results of the tests in Fig. 3 are consistent with the averaged performance over various conditions shown in Fig. 2. The highest particle concentration escaping from the hoods still occurred in the constant-flow hood. The release was better controlled by using the constant-velocity and air-curtain hoods. The set of operations shown in Fig. 3 was performed by an operator with narrow width 34 cm at chest and 174 cm height, and very slow and gentle motions were applied through all the handling tasks. The high concentration peaking at 200 nm diameter shown for the constant-flow and constant-velocity hoods in Fig. 2 is not seen in the Fig. 3. For a lower release from this set of operations, the optimal handling condition with lowest particle release occurred at the face velocity of 0.4 m s 1 (high sash) for the constant-flow hood

7 84 S. J. Tsai, R. F. Huang and M. J. Ellenbecker Fig. 3. Particle number concentration increase ( nm) during pouring nanoparticles at during one pouring operation. (a) Constant-flow hood; (b) constant-velocity hood; and (c) aircurtain hood. (Fig. 3a) and at the face velocity of 0.5 m s 1 (middle low sash) for the constant-velocity hood (Fig. 3b). Handling nanoparticles in the constant-flow hood at face velocities between 0.4 and 0.5 m s 1 resulted in the lowest exposure under most operating conditions studied. For the newly designed air-curtain hood, particles escaping from this hood were always detected at a very low level no matter the locations of sash or the task performed. However, the low sash position with the shortest air curtain resulted in a slightly higher release than other sash locations (Fig. 3c). Airflow patterns The results of the flow visualization tests of airflow patterns for the three hoods are shown in Fig. 4. Photographs were taken at the middle sash locations for all hoods. For the constant-flow and constant-velocity hoods, which generate an airflow past the operator, fog was generated behind the mannequin; the airflow carried the fog flow around the mannequin and into the hoods. For the constantflow hood, horizontal and vertical views are shown in Fig. 4a,b, respectively. Circulating airflows in the horizontal plane were intense and airflow eddies are readily observed between the mannequin s body and arms (Fig. 4a). The vertical view of airflow at a plane 15 cm behind the doorsill with arms inside the hood duplicating the actual position of handling tasks described above shows very intense airflows circulating individually around each arm of the mannequin (Fig. 4b). These vertical circulations around the arms were seen only in the constant-flow hood. For the constant-velocity hood, the airflow seen in the vertical plane (Fig. 4d) did not show the individual circulating airflow around the two arms, instead a downward airflow was seen to evenly carry smoke around the arms. Circulating airflows were observed in the horizontal plane in the constant-velocity hood, but those eddies formed between the mannequin s body and arms were smaller than for the constantflow hood and the circulation was less intense (Fig. 4c). For the air-curtain hood, the visual tests were taken while generating fog both inside and outside the hood; both showed similar patterns around the doorsill. A photograph taken when generating fog inside the hood is shown in Fig. 4e. This hood design avoids the inward horizontal airflow design of traditional hoods and instead induces a vertical downward flow inside the hood. Consequently, the airflow around the doorsill was almost vertical and did not have the circulation between mannequin s body and arms on the horizontal surface as observed in the other hoods. Air flows directly into the suction slot located just behind the doorsill, as seen in Fig. 4e. The strong suction force exerted on the bottom of air curtain combined with the air jet issued from the sash produce a smooth laminar airflow going downward across the hood face. In addition, circulation in the horizontal plane inside the hood was minimized due to the downward airflow motion from the top grille. The airflow in the vertical plane around arms showed downward airflow and no circulation around arms was seen (Fig. 4f). DISCUSSION The magnitude of particles escaping from the two traditional hoods tested in these experiments was

8 Airborne nanoparticle exposures while using fume hoods 85 Fig. 4. Features of airflow pattern of three hoods at middle sash. (a) Constant-flow hood, horizontal; (b) constant-flow hood, vertical; (c) constant-velocity hood, horizontal; (d) constant-velocity hood, vertical; (e) air-curtain hood, horizontal; and (f) air-curtain hood, vertical. substantial. The resulting BZ concentration depended on various factors (Tsai et al., 2009), including the task performed, the workers physical size and hood design, and operating conditions such as sash position and face velocity. The particle release was minimized by performing tasks in the optimal face velocity range, which for this study was m s 1. This is the same range as that currently recommended by ACGIH (2007). However, workers exposure while handling nanoparticles in traditional hoods within the optimal face velocity still is variable and can be substantial depending on the actual worker activity. All the variables that were found to affect the magnitude of exposure in traditional hoods, such as workers motion, body size, and sash location, did not affect the exposure when using the air-curtain hood. The airflow patterns within and around an operating fume hood play an important role in affecting particle escape and worker exposure. The constant-velocity hood showed very good performance when operating at the recommended 0.5 m s 1 face velocity, with limited particle release during the more energetic task of pouring. The better performance of the constantvelocity hood compared to the constant-flow hood was related to the smaller vortex created and restrained above the sash edge inside the constant-velocity hood which eliminated the vertical circulation around the arms. The vortex in the constant-flow hood was larger and more intense than in the constant-velocity hood and caused the strong circulating vertical airflow around the arms. The particle release from the constant-flow hood was much higher at its optimal face velocity of 0.4 m s 1 than the release from the constantvelocity hood for most tested operating conditions.

9 86 S. J. Tsai, R. F. Huang and M. J. Ellenbecker The air-curtain hood was found to have a stable and very low particle release for all tested operating conditions, with a very small but measurable release occurring at the low sash position. The cause of this small release most likely was related to a slight imbalance between the air jet flow and the suction flow when the air curtain was very short. Since the air-curtain hood used for these tests was a first-generation hood from the pilot manufacturing, for laboratory testing only, the flow control system was not optimized. A comprehensive logistic control system is used in the later generation hoods now being sold for commercial use. When the crossdraft or sash movement is applied, the containment leakages of the traditionally designed fume hoods (either the constant flow or the constant velocity) usually become serious, particularly at a sash velocity,0.5 m s 1 (Tseng et al., 2007). Although operating the traditional hood within the face velocity range of m s 1 under the static environment situation is recommended, the containment leakage may still be noticeable when the hood is operated under the environment with crossdraft or disturbance of human walk-by. The air-curtain hood generally has a much higher robustness against the influence of environment draft (Huang et al., 2007) when compared with that of the traditional fume hoods. Considering the practical situation that most fume hoods are inevitably operated in an environment with more or less drafts and/ or disturbances such as high traffic levels, the air-curtain hood may be the best choice when handling dry nanoparticles. CONCLUSIONS The use of local ventilation systems for nanoparticle handling must be approached with caution since particle escape from traditional hoods can be substantial and the magnitude of the release depends on various conditions of use. Workers exposure can be minimized by operating traditional fume hoods within the optimal face velocity range ( m s 1 ) with very careful attention paid to the worker s motions and tasks performed. Circulation of airflow in the region between the workers body and arms and the hood face was found for constant-flow hood designs. A worker s arm motions, if energetic, can induce strong turbulence that will work with the complex vortex inside the hood to carry nanoparticles to the BZ. When the face velocity is lower than the optimal range, turbulent body and arm wakes are minimized, but more rapid arm motion can still cause higher nanoparticle release during manual handling tasks since the inward velocity across the hood face is lower. In conclusion, it is crucial that handling nanoparticles in traditional fume hoods be done with a hood face velocity within m s 1 range and with very careful motions; exposure could be substantial when these cautions are not followed. The performance of the newly designed air-curtain hood during nanoparticle use was outstanding for the various conditions tested here and avoids the difficulties found when using traditional hoods. FUNDING Nanoscale Science and Engineering Centers for High-rate Nanomanufacturing funded by the National Science Foundation (Award No. NSF ). SUPPLEMENTARY MATERIAL Supplementary Figure S01 can be found at annhyg.oxfordjournals.org/ Acknowledgement We acknowledge the supporter of Institute of Occupational Safety and Health in Taiwan to the aircurtain hood. REFERENCES ACGIH. (1966) Industrial ventilation a manual of recommended practice. 9th edn. Lansing, MI: American Conference of Governmental Industrial Hygienists; pp ACGIH. (2007) Industrial ventilation a manual of recommended practice. 26th edn. Cincinnati, OH: American Conference of Governmental Industrial Hygienists; pp Altemose BA, Flynn MR, Sprankle J. (1998) Application of a tracer gas challenge with a human subject to investigate factors affecting the performance of laboratory fume hoods. Am Ind Hyg Assoc J; 59: ANSI/AIHA. (2003) ANSI/AIHA Laboratory ventilation standard Z9.5. Atlanta, GA: American Industrial Hygiene Association, American National Standards Institute, Inc. ANSI/ASHRAE. (1995) Method of testing performance of laboratory fume hoods (ANSI/ASHRAE Standard ). Atlanta, GA: American National Standards Institute, American Society of Heating, Refrigeration and Air Conditioning Engineers. BSI. (1994) Laboratory fume cupboards. Part 4. Method for determination of the containment value of a laboratory fume cupboard (BS 7258: Part 4: 1994). London: British Standards Institution. BSI. (2007) Nanotechnologies part 2: guide to safe handling and disposal of manufactured nanomaterials (No. PD :2007). London: British Standards Institution. Burgess WA, Ellenbecker MJ, Treitman RD. (2004) Ventilation for control of the work environment. 2nd edn. New York: Wiley Interscience; pp

10 Airborne nanoparticle exposures while using fume hoods 87 CEN. (2003) Fume cupboards part 3: type test methods (pren ). Brussels, Belgium: European Committee for Standardization. DiBerardinis LJ, Baum JS, First MW et al. (1993) Guidelines for laboratory design: health and safety considerations. 2nd edn. New York: Wiley Interscience; pp Flynn MR, Ljungqvist B. (1995) A review of wake effects on worker exposure. Ann Occup Hyg; 39: Hinds WC. (1999) Aerosol technology: properties, behavior, and measurement of airborne particles. 2nd edn. New York: Wiley Interscience; pp Huang RF, Lin SY, Jan SY et al. (2005) Aerodynamic characteristics and design guidelines of push pull ventilation systems. Ann Occup Hyg; 49: Huang RF, Wu YD, Chen HD et al. (2007) Development and evaluation of air curtain fume cabinet with its considerations of aerodynamics. Ann Occup Hyg; 51: Johnson AE, Fletcher B. (1996) The effect of operating conditions on fume cupboard containment. Safety Sci; 24: Kim TH, Flynn MR. (1991a) Airflow pattern around a worker in a uniform freestream. Am Ind Hyg Assoc J; 52: Kim TH, Flynn MR. (1991b) Modeling a worker s exposure from a hand-held source in a uniform freestream. Am Ind Hyg Assoc J; 52: Kim TH, Flynn MR. (1992) The effect of contaminant source momentum on a worker s breathing zone concentration in a uniform freestream. Am Ind Hyg Assoc J; 53: Kumar P, Fennell P, Symonds J et al. (2008) Treatment of losses of ultrafine aerosol particles in long sampling tubes during ambient measurements. Atmos Env; 42: Maynard AD, Aitken RJ. (2007) Assessing exposure to airborne nanomaterials: current abilities and future requirements. Nanotox; 1: Maynard AD, Kuempel ED. (2005) Airborne nanostructured particles and occupational health. J Nanopart Res; 7: OSHA. (1992) General industry, OSHA safety and health standards, CFR Washington, DC: Occupational Safety and Health Administration. Poland CA, Duffin R, Kinloch I et al. (2008) Carbon nanotubes introduced into the abdominal cavity of mice show asbestoslike pathogenicity in a pilot study. Nat Nanotechnol; 3: Saunders GT. (1993) Laboratory fume hoods: a user s manual. New York: Wiley Interscience; pp Schulte HF, Hyatt EC, Jordan HS et al. (1954) Evaluation of laboratory fume hoods. Am Ind Hyg Assoc J; 15: Tsai S, Ada E, Isaacs J et al. (2009) Airborne nanoparticle exposures associated with the manual handling of nanoalumina and nanosilver in fume hoods. J Nanopart Res; 11: Tsai S, Ashter A, Ada E et al. (2008) Control of airborne nanoparticle release during compounding of polymer nanocomposites. Nano; 3: Tseng LC, Huang RF, Chen CC et al. (2007) Effect of sash movement and walk-bys on aerodynamics and contaminant leakage of laboratory fume cupboards. Ind Health; 45: University of California. (2007) Nanotechnology: guidelines for safe research practices. Berkeley, CA: University of California.