Sampling of Carbon Fiber Aerosols

Transcription

1 Aerosol Science and Technology ISSN: (Print) (Online) Journal homepage: Sampling of Carbon Fiber Aerosols B. Y. H. Liu, D. Y. H. Pui, X. Q. Wang & C. W. Lewis To cite this article: B. Y. H. Liu, D. Y. H. Pui, X. Q. Wang & C. W. Lewis (1983) Sampling of Carbon Fiber Aerosols, Aerosol Science and Technology, 2:4, , DOI: / To link to this article: Published online: 05 Jun Submit your article to this journal Article views: 214 View related articles Citing articles: 12 View citing articles Full Terms & Conditions of access and use can be found at Download by: [ ] Date: 30 December 2017, At: 02:18

2 Sampling of Carbon Fiber Aerosols B.Y. H. Liu,D.Y. H.Pui,X.Q. Wang,* Particle Technology Laboratory, Mechanical Engineering Department, University of Minnesota, Minneapolis, MN and C. W. Lewis Environmental Sciences Research Laboratory, Environmental Protection Agency, Research Triangle Park, NC This paper describes theoretical and experimental studies undertaken in connection with the problem of sampling carbon fiber aerosols in an ambient environment. Calculations indicate that carbon fiber aerosols with an 8-pm fiber diameter and a density of 1.8 g/cm3 can be sampled by a sampler designed to collect particles in the pm aerodynamic diameter range if fibers up to a few millimeters in length are to be collected. An ap- INTRODUCTION Carbon fibers are high-tensile-strength filaments of approximately circular cross section having typical diameters of about 8 pm. In the early 1960s it was discovered that such filaments could be obtained in a carefully controlled process that combined continuous tensile stress and high-temperature pyrolization of a precursor textilelike fiber. The names "carbon fiber" and "graphite fiber" are often used interchangeably. The distinction arises in the temperature of the pyrolization step, of the order of 1200' and 2400 C for carbon and graphite fibers, respectively. Carbon or graphite fibers are increasingly used in applications where a premium is placed on the combination of hgh strength and light weight, such as in high-performance aircraft components and sporting equipment. The benefits of carbon fiber tech- *Permanent address: East China Petroleum Institute, Dongying, Shandong, China. preach is then described in which carbon fiber aerosols are collected in the impactor of the inlet of a dichotomous sampler. Experimental data are presented showing that the impaction characteristics of the carbon fiber aerosol can be predicted with a reasonable degree of accuracy by approximating the shape of the fibers by a prolate ellipsoid. nology, however, are not realized without some risk. Most concern has focused on the high electrical conductivity of carbon fibers, which, in combination with their airborne buoyancy, cause them to be a potential hazard in electrical environments. Although most applications of carbon fibers involve their consolidation with an epoxy-type binder to form a composite, the binder is less resistant than the fibers to intense heat (in an incinerator or accident), which can lead to some fiber release. A great deal of information has been published recently relating to the potential hazard that would be associated with widespread use of carbon fibers. Much of this has been obtained through the National Carbon Fiber Program (NCFP), which was initiated in 1978 by the President's Office of Science and Technology, Policy to assess this problem. Specific details of the roles of various government agencies in the NCFP are available elsewhere (EPA, 1980) as are a summary of results obtained thus far in the NCFP and Aerosol Science and Technology Q 1983 B. Y. H. Liu et al.

3 500 Liu et al. a comprehensive bibliography (NASA, 1980a). The NCFP has led to the design of a number of instruments and techniques for counting carbon fibers in an ambient environment. These range from simple passive collectors (e.g., sticky paper) to active counters, some with fiber-length measurement capability (eg, a ball counter). These methods and instruments are summarized in considerable detail by Pride et al. (1980) and Newcomb (1980). Nearly all operate on a flux or deposition principle, relying on existing air currents to bring fiber aerosol to the detector rather than on active sampling. Consequently, they are not well suited for determining carbon fiber concentrations (in number per cubic meter) in an open environment affected by variable and uncontrolled wind velocities. The present work is concerned with the problem of active sampling carbon fiber aerosol in an ambient environment. The central concern is the design of an aerosol inlet with high efficiency for carbon fiber aerosol and whose efficiency is relatively unaffected by the environmental variability of the wind speed. The problem has been approached both theoretically and experimentally. Theoretically, the fundamental aerosol physics of the airborne carbon fibers has been developed to a greater degree than was previously available. Aspects of the theoretical results have been experimentally examined. The latter investigation required the development and evaluation of a new generator of carbon fiber aerosol for testing purposes, which is also described. Finally, the design of an integrated active sampler for carbon fiber aerosol is proposed. make the sampler development more difficult, and too narrow a range would result in significant sampling biases and the collection of unrepresentative samples. The aerodynamic properties of a fiberlike particle can be approximated by those of a prolate ellipsoid with appropriate length-towidth ratio. According to Fuchs (1964), a prolate ellipsoid such as that shown in Figure 1 will experience a drag force equal to when moving through a fluid medium. When the motion is along the polar axis, K= K', where When the motion is perpendicular to the polar axis, K = K", where K" = +(p2 - I)/ (3) In Eqs. (1)-(3), 0 = b/a is the aspect ratio, a is the radius of the minor axis of the ellipsoid, b is the radius of the major axis, p is the fluid viscosity (1.83 X lor4 P in air at 20 C), and u is the particle velocity relative FIGURE 1. The prolate ellipsoid. POLAR AXIS I THEORETICAL CONSIDERATIONS To develop a criterion for sampler design, it is necessary to know the aerodynamic properties of the carbon fiber aerosol and the particle size range over whch the sampler must operate. Too wide a size range would

4 Sampling of Carbon Fiber Aerosols to the fluid medium. Under the influence of gravity, the prolate ellipsoid will attain a terminal settling velocity us given by mg = 6apausK, (4) where m is the mass of the particle. Since for the prolate ellipsoid, we have by (4) and (5) where g is the acceleration of gravity and pp is the particle density. The aerodynamic radius a, of a particle is the radius of a sphere of unit density having the same settling speed as the particle in question. For a sphere of unit density, where p, = 1 g/cm3 TABLE 1. Aerodynamic Diameter of Carbon Fiber Aerosolsa Fiber length (pm) Aerodynamic diameter (pm) vertical fall 8h horizontal fall 8h wl app = 1.8 g/cm3; fiber diameter Df = 8.0 pm. b~ sphere. 'An infinitely long cylinder. From comparison of (6) and (7) we see that is the aerodynamic radius of the prolate ellipsoid. In Table 1 the aerodynamic diameter of a carbon fiber particle is calculated by means of Eq. (8) for various fiber lengths. We assume a fiber diameter of 8 pm and a particle density of 1.8 g/cm3, values that are typical of the carbon fibers now in common use. The results are shown in Figure 2. The aerodynamic diameter of the fiber is seen to range from to pm for fiber lengths up to 800 pm for a fiber falling vertically (parallel to the axis). For a fiber falling horizontally (perpendicular to the polar axis), the corresponding range is pm. That there is little dependence of the aerodynamic diameter on the length for carbon fibers is very apparent. A different but equally rigorous approach can be used to check the validity of the prolate ellipsoid model. Consider the limiting case of an infinitely long cylinder having a diameter of 2a and falling horizontally. The drag coefficient C, for this case may be obtained from Lamb's equation (Lamb, 1945): C, = 8m/Re( In Re), (9) where Re is the particle Reynolds number, p, being the fluid density (1.21 X lop3 g/cm3 ). The correct value of the settling velocity us is the one that causes the drag force per unit length, to be equal to.the gravitational force per unit length, The resulting equation may be solved for us,

5 Liu et al. FALLING I I I I I FIBER LENGTH, pm FIGURE 2. Aerodynamic diameter of prolate ellipsoids (fiber diameter, 8 pm; particle density, 1.8 g/cm3). although a trial-and-error approach is required since the dependence on v, is somewhat complicated. Substituting the result for us (2.45 cm/sec) in Eq. (7) gives for the aerodynamic radius of an infinitely long cylinder (a = 4 pm, p, = 1.8 g/cm3) falling horizontally. This result is included in Table 1 and appears to be consistent with the results for the sphere and prolate ellipsoid. This result indicates that representative samples of carbon fiber aerosols can be collected by a sampler designed to collect pm aerodynamic diameter particles if fibers up to a few rmllimeters in length are to be collected. If longer fibers are to be collected, the upper size limit of the sampler must similarly be extended. EXPERIMENT Based on the foregoing considerations, it was decided that a dichotomous sampler (Dzubay and Stevens, 1975) can be adapted for carbon fiber sampling. The reasons for this approach are twofold. First, the dichotomous sampler is already widely used for environmental monitoring. It would be economically advantageous not to deploy a separate set of samplers specifically for carbon fiber monitoring. Second, the sampling inlet being developed for the dichotomous sampler for inhalable particles appeared to have characteristics that are quite suitable for carbon fiber sampling. A separate developmental program can thus be avoided. Figure 3 is a schematic diagram of the inlet that we have developed for the dichotomous sampler for inhalable particles. According to Miller et al. (1979), inhalable particles are particles with aerodynamic diameters of 15 pm or less. It now appears that size-selective particle sampling is likely in the future to be based on a sampling inlet with a 10-pm cutoff for the so-called thoracic par-

6 Sampling of Carbon Fiber Aerosols CIRCULAR COVER SPACER (3) RAIN GUARD DEFLECTION SUPPORT POST IMPACTION CUP ticles, or PM-10 (particulate matter of 10-pm cutoff). However, in 1979, when this work was initiated, 15 pm was considered the appropriate upper cutoff for size-selective particle sampling. Two versions of the inlet shown in Figure 3 have been tested, one identical to the inlet shown and a second with a 30-mesh metal screen around the rain guard. The test was carried out in the wind tunnel facility of the Particle Technology Laboratory of the University of Minnesota. Figure 4 is a schematic diagram of the wind tunnel test facility used. The test was carried out using monodisperse uranine-tagged oleic acid aerosols following a procedure described previously by Liu and REDUCER FOR CONNECTION TO SAMPLER FIGURE 3. A dichotomous sampler inlet for inhalable particles. All dimensions are in centimeters. Pui (1981). Tests were carried out at wind speeds of 1, 8, and 24 km/hr, and the overall sampling efficiency of the inlet was established by comparing the quantity of aerosol collected through the inlet with that collected by an isokinetic sampling probe. The results are shown in Figures 5 and 6. In Figure 5 the points labeled with open symbols were measured in the present work; filled symbols denote measurements done at Lawrence Berkeley Laboratory under static conditions (Loo, private communication, 1981).

7 Liu et al. t VACUUM VACUUM LOW WlND SPEED ( 1-9 kmlhl 1. HIGH WlND SPEED (10-30 kmlh) TEST SECTION TEST SECTION (20"x 20" SQUARE DUCT) (14" CIRCULAR DUCT) FIGURE 4. Schematic diagram of wind tunnel test facility of the Particle Technology Laboratory. To see whether this particular inlet can be used for carbon fiber aerosol sampling, the following experiments were performed using carbon fiber aerosols. Generation of Carbon Fiber Aerosols Before generation of a carbon fiber aerosol, the long continuous carbon fiber filament available from the manufacturer must first be reduced in length and then aerosolized. Loo et al. (1982) has reported on a technique for cutting a single-fiber filament into equallength segments by means of a focused laser beam. Extremely monodisperse carbon fiber aerosols with lengths from 30 to 3000 pm have been generated by this laser-cutting technique. Because only a single filament is cut by the laser at any one time, the quantity of carbon fiber aerosol obtained is quite limited. In the technique reported by Newcomb (1980), carbon fiber filaments are fed continuously by a mechanical feeding device and cut by a rotating knife. The tech- nique is capable of generating a large quantity of carbon fibers of approximately equal length. However, the minimum length obtainable is about 100 pm. In the present study, carbon fibers with lengths ranging from about 10 pm to several hundred microns are needed. Neither of the two techniques we have described are considered suitable. A new method therefore had to be devised. Three separate length reduction techniques were investigated in this study. One involves the use of a ball mill, the second a micromill, and the third a mortar and pestle. The greatest success has been achieved with the micromill followed by the use of mortar and pestle. Stock fiber from Hercules Powder Co. was first cut by hand into 1-3-cm lengths. These fibers were then placed in the micromill, which is a blender-type device with rotary blades in a small cup. The amount of fiber placed in the micromill was important. Too much material would result in fibers being caught on the blades, thus retarding the cutting action of the blades. This occurred with 0.5 g of material in the cutting chamber. With 0.1 g of material the grinding proceeded until the cut fibers rolled into small fiber

8 Sampling of Carbon Fiber Aerosols 80 FIGURE 5. Efficiency of the IP inlet without the screen. (0, W= 1 km/hr, D,, = 15 pm, a, = 1.45; 70 0, W = 5 krn/hr, D,, = 16.5 pm, a, = 1.31; A, W= 24 km/hr, D,, = 14.3 pm, a, = 1.25; where D,,,y is the particle diameter at 50% k efficiency and a, is the geometric + 50 standard deviation. The filled 5 symbols show the impactor trans mission efficiency.) balls 1-2 rnrn in diameter. This took from 4 to 15 mins. The fiber balls when crushed with the mortar and pestle yielded fibers pm in length accompanied by irregular dust that consisted of fiber fragments measuring approximately 1 pm across. By this means sufficient quantities of cut fibers could be produced for experimental purposes. A photomicrograph of the fibers thus produced is shown in Figure 7. The other two devices investigated, the ball mill and the mortar and pestle, both yielded too much dust and were considered to be less desirable than the micromill. To aerosolize the cut fibers, a fluidized bed dust generator was used. The device, described previously by Marple et al. (1978) AERODYNAMIC DIAMETER, pm consisted of a fluidized bed and a chain mechanism for feeding a powder to the bed for aerosolization. In these experiments the feeding mechanism was not used. The cut fibers, which weighed approximately 0.1 g, were placed in the fluidized bed and covered by the bed material, 150-pm-diam brass beads. The fluidizing air was then turned on to aerosolize the fibers. This resulted in an output aerosol concentration that was initially high and that decreased with time. However, this was acceptable for the specific purpose for which the carbon fiber aerosol was to be used. Figure 8 shows the output mass concentration of the carbon fiber aerosol as a function of time following the initial inlet of

9 Liu et al. I 10 AERODYNAMIC DIAMETER, pm FIGUKE 6. Efficiency of the IP inlet with the screen. (0, W = 1 km/hr; 0, W = 5 km/hr; A, W = 24 km/hr.) air to the fluidized bed. The mass concentration was measured by taking 2-min samples of the aerosol with a 47-mm-diam membrane filter in a standard Millipore filter holder and weighing the collected material with an electronic microbalance. In these experiments the fluidizing air was set to 23 l/min. For some of the samples the size distribution of the carbon fiber aerosol was also measured with an optical microscope. A typical length distribution for these fibers is shown in Figure 9. For the particular sample used for Figure 9 the fiber lengths were found to be log-normally distributed with a median length of 19.5 pm and a geometric standard deviation of Only particles that were clearly recognizable as fibers were measured; i.e., fine carbon dusts were ignored in the measurements. The minimum length of the fibers measured was estimated to be approximately pm. Sampling of Carbon Fiber Aerosols The dichotomous sampler inlet shown in Figure 3 allows relatively large particles to enter the inlet opening. The impactor in the inlet then removes the coarse particles above

10 Sampling of Carbon Fiber Aerosols Wpm aerodynamic diameter by impaction. Since the cutoff size of the impactor is 15 pm, carbon fiber aerosols with aerodynamic diameters larger than 15 pm are expected to be collected by the impactor. Table 1 shows that this will include any carbon fibers longer than 32 pm. The surface of the impaction head consisted of a series of concentric V- grooves for the purpose of reducing particle bounce and reentrainment. To test the ability of this V-groove impactor to collect carbon fiber aerosols, the following experiments were performed. Carbon fiber aerosols were generated as previously described and introduced directly into the inlet. A filter was placed immediately downstream of the inlet to allow particles penetrating through the inlet to be collected. Following a length of time sufficient to collect a suitable quantity of carbon fibers, the filter was removed. The V-groove FIGURE 7. Photomicrograph of the carbon fiber aerosol. impactor was disassembled, and the fibers collected in the V-groove were transferred to a filter by inverting the V-groove surface and allowing the fibers to fall onto the filter. The fiber sample thus obtained was placed in an optical microscope, and the number of fibers in each of several length ranges was then counted. The fibers collected on the afterfilter were similarly measured in the microscope. From these data the fraction of fibers that were collected by the impactor within a certain length range could be determined. Since the aerodynamic diameter of a fiber is related to its length, the data can then be compared with the particle penetration characteristics of the impactor as measured by ideal spherical liquid particles. Figure 10 shows the results of ths comparison.

11 Liu et al. FIGURE 8. Output mass concentration of the carbon fiber aerosol generator. TIME, min In Figure 10 the percent of carbon fiber particles penetrating through the impactor as determined for each length category is plotted against the aerodynamic diameter of the fiber of that length. It is assumed that the fibers impact against the impactor surface horizontally (parallel to the impactor surface) or vertically (perpendicular to the impactor surface). It is interesting to note that the data, when plotted in this manner, show good correlation with the impactor penetration data obtained by means of ideal spherical particles. This suggests that the impaction behavior of the carbon fiber aerosol can be predicted with adequate accuracy by means of the prolate ellipsoid approximation. Previous experimental and theoretical work has shown that significant breakage occurs when individual carbon fibers undergo impaction at velocities as low as m/sec (NASA, 1980b). The impaction velocity encountered in the inlet of Figure 3 is more than an order of magnitude smaller, however, so that breakage phenomena should be entirely absent. This is consistent with Figure 10. Some additional experiments were also performed to see what fraction of fibers would impact horizontally on the impaction surface and what fraction would impact vertically. For these experiments the V- groove impaction surface was replaced with a smooth glass cover slip coated with a thin layer of grease. The collected fibers when viewed in an optical microscope could be seen either to lie flat on the surface or to make some nonzero angle with it. The result was that approximately 45% of the fibers were found to lie flat against the surface and thus were presumed to have impacted hori-

12 Sampling of Carbon Fiber Aerosols FIGURE 9. Length distribution of the carbon fiber aerosol sample. (Median length, 19.5 pm; geometric standard deviation, 2.24.) zontally, while the remaining 55% were found to have a nonhorizontal orientation. CONCLUSIONS The results of this study indicate that the aerodynamic behavior of carbon fiber aerosols can be approximated by that of prolate ellipsoids. A dichotomous sampler inlet with a V-groove impaction surface can be used to sample carbon fiber aerosols from the ambient atmosphere. Fibers with aerodynamic diameters of 15 pm or larger are collected on the V-groove impactor surface and can be easily recovered for examination and identification in an optical microscope. The dichotomous sampler inlet appears to be a FIBER LENGTH,pm relative simple and low-cost approach to carbon fiber aerosol sampling from the ambient atmosphere. Considering that the dichotomous sampler is already widely used for environmental monitoring, the addition of carbon fiber sampling capabilities represents a cost-effective approach to the environmental monitoring of carbon fiber aerosols. Although the research described in this article was funded wholly or in part by the U.S. Environmental Protection Agency through cooperative agreement No. CR to the University of Minnesota, it has not been subjected to Agency review and therefore does not necessarily reflect the views of the Agency, and no official endorsement should be inferred.

13 Liu et al. 1.O AERODYNAMIC DIAMETER,pm FIGURE 10. Comparison of impactor transmission characteristics for carbon fiber aerosols and spherical particles. The curve is an empirical representation of the liquid particle data. For the V-groove impaction surface, squares denote points for liquid particles, triangles points for fibers falling horizontally, and circles points for fibers falling vertically. REFERENCES Dzubay, T. G., and Stevens, R. K. (1975). Ambient air analysis with dichotomous sampler and x-ray fluorescence spectrometer, Enuiron. Scz. Technol. 9: EPA (1980). Data base review and assessment of carbon fiber release into the environment, Final Report, contract No , U.S. Environmental Pro- tection Agency, Cincinnati, OH. Fuchs, N. A. (1964). The Mechanics of Aerosols. Pergamon, New York. Lamb, H. (1945). Hydrodynamics, Dover, New York. Liu, B. Y. H., and Pui, D. Y. H. (1981). Aerosol sampling inlets and inhalable particles, Atmos. Enuiron. 15: Loo, B. W., Cork, C. P., and Madden, N. W. (1982). A laser-based monodisperse carbon fiber generator, J. Aerosol Sci. 13: Marple, V. A,, Liu, B. Y. H., and Rubow, K. L. (1978). A dust generator for laboratory use, Am. Znd. Hygiene Assoc. J. 39: Miller, F. J., Gardner, D. E., Graham, J. A., Lee, R. E., Jr., Wilson, W. E., and Bachmann, J. D. (1979). Size considerations for establishing a standard for inhalable particles, J. Air PoNut. Control Assoc. 29: NASA (1980a). Risk to the public from carbon fibers released in civil aircraft accidents, NASA Rep. SP- 448.

14 Sampling of Carbon Fiber Aerosols 511 NASA (1980b). Experimental and analytical studies for Pride, R. A,, McHatton, A. D., and Musselman, K. A. the NASA carbon fiber risk assessment, NASA Con- (1980). Electronic equipment vulnerability to firetractor Rep , NASA contract NAS released carbon fibers, NASA Rep. TM Newcomb, A. A,, Jr. (1980). A carbon fiber exposure test facility and instrumentation, NASA Rep. TM Received 22 September 1982; accepted 13 June 1983