United States Environmental Protection Agency Atmospheric Research and Exposure Assesment Laboratory Research Triangle Park NC 27711 Research and Development EPA/600/S3-90/048 Sept. 1990 EPA Project Summary Development of a Sampling Procedure for Large Nitrogen- and Sulfur-Bearing Aerosols Stephen J. Randtke, Dennis p. Lane, and Terry E. Baxter The objective of this research project was to develop a procedure that would permit accurate sampling and analysis of large (>2.5 urn) aerosols, especially those containing nitrate, ammonium, and sulfate, without interfering with the sampling and analysis of the gases and fine particle species of interest In dry deposition and acid aerosol sampling. Although other sampling devices were initially considered, the experimental work focused exclusively on a single- stage Impactor modified to utilize a re- movable impaction surface mounted on the end of an annular denuder. The cut point of the modified Impac- tor, used with a polycarbonate (PC) filter coated with 5 |iL of silicone oil, was 2.5 \an, with less than 1% particle bounce occurring for 8-^.m particles. Significant particle bounce occurred when using uncoated or grease-coated impaction surfaces composed of PTFE or porous glass. Saturating a porous glass impac- tion disk with oil reduced particle bounce to about 2%, but the impacted particles could not be easily removed from the porous glass impaction surface. Non-isoaxial sampling resulted in substantial undersampling. Isoaxial sampling of ambient air requires either a horizontally mounted sampling system or a vertically mounted system with an inlet bend. Horizontal configurations performed poorly; greased and lightly oiled surfaces exhibited a high degree of bounce and oil flowed sideways off oil- saturated disks. Vertically mounted systems employing a bent-tube inlet performed reasonably well, but inlet losses were quite high for very large particles and for blunt-tipped Inlets. Inlet losses averaged only 2.7% for 7.9-(im uranine particles sampled using a sharp- edged 90° bent-tube Inlet. Losses of solid particles within the combined impactor/denuder system were greatest for particles close in size to the cut point. For particles having an aerodynamic diameter of 2.0-2.8 |im, losses in the denuder ranged from 4-8% of the mass sampled, while losses on the spacer between the impactor and de- nuder ranged from 10-17% of the mass sampled. Losses for particles of other sizes (larger and smaller) were generally negligible (i.e., <1%). A number of oils and greases were found to be compatible with the analysis of impacted particles by ion chromatog- raphy. PC filters were found to be free of nitrate, nitrite, and sulfate. PC filters, silicone oil, and Vaseline exposed to high concentrations of HNO3 and SO2 were found to adsorb little, if any, nitrate or sulfate. TFE impaction disks may adsorb traces of HNO3, but it is not nec- essary to include them in the analysis if they are overlain by a PC filter. Impacted particles of varying com- position were exposed to high concen- trations of potentially reactive gases (HNO3, SO2, or NH) for four hours at a nominal flowrate of 16.7 Lpm. NH4CI and NH4NO3 impacted on an uncoated PC filter did not volatilize when exposed to purified air. NH Cl and NaCI particles Impacted on a sllicone-greased PC filter did not react with SO2. NaCI and NH4NO, particles impacted on a PC filter coated ------- with flvauL of silicons oil did not react to a significant extent with HNO or SO., respectively. Acidic NaHSO3 particles Impacted on a greased surface reacted to a limited extent with NH, (i.e., they wore 1.7 to 3.5% neutralized). In a 24-hour field test employing four collocated sampling systems, the concentrations of coarse-particle nitrate- N and sulfate were determined to be 0.939 ± 0.032 u,g/m*(± 3.4%) and 0.222 ± 0.007 u.g/m3 (± 3.3%), respectively. However, substantial percentages of the nitrate, sulfate, and ammonium (36%, 44%, and 50%, respectively) were col- lected on the bent-tube Inlets rather than on the Impactlon disks. This Project Summary was devel- oped by EPA's Atmospheric Research and Exposure Assessment Laboratory, Research Triangle Park, NC, to an- nounce key findings of the research project that Is fully documented In a separate report of the same title (see Project Report ordering Information at back). introduction So that reliable estimates of the loads of nitrogen and sulfur associated with dry deposition can be made, scientists at the U.S. EPA and elsewhere have been work- ing to develop quantitative methods to sample and analyze the major nitrogen- and sulfur-containing species present in gaseous and aerosol form in ambient air. These include HNO3, HNO,, NO, NO,, NH,. n'rtrate, nitrite, ammonium, SO,, SO, H,SO., sulfate, and sulfite. Progress in developing sampling methods for dry deposition has been ham- pered by the chemical complexity of the system; several of the chemical species of interest are quite reactive and there are interactions between the solid, liquid, and gaseous components. As a result, a con- siderable number of artifacts have been encountered, including: 1) adsorption of gaseous species (e.g., HNO3 and SO2) on thefittermediaorthe sampling device used to collect aerosols; 2) adsorption of HNO3 and other gaseous species by particles deposited on the filter media; 3) volatiliza- tion of paniculate species such as NH.NO3 and NH4CI; 4) and formation and subse- quent volatilization of HNO3 as a result of metathetfcal or displacement reactions. A number of investigators have been working to develop sampling devices and procedures able to circumvent these problems. One system currently under develop- ment and potentially able to circumvent these problems consists of a filter pack preceded by one or more annular denuders and a Teflon-coated glass impactor (Figure 1). The filter pack is designed to capture fine particles and any gases that may vola- tilize from them, while the denuders capture reactive gases and allow fine particles to pass through unaltered into the filter pack. . The impactor is designed to remove large particles (i.e., those having an aerodynamic diameter greater than 2 u.m), which might otherwise be trapped in the denuder or which, if they were alkaline, might neutralize acidic aerosols captured by the filter pack. Initial attempts to circumvent the prob- lems associated with large particles have focused on the use of either a Teflon-coated glass impactororaTef Ion cycloneto remove large particles ahead of the first denuder tube. However, there are potential problems with each of these devices, including: 1) particle resuspensipn or bounce, especially at high mass loadings; 2) volatilization of constituents of interest from the surfaces of large particles; and 3) reactions of gaseous constituents with impacted particles, with thesampling device itself, orwith any grease oroilusedtotraptheparticles. Furthermore, neither device is inherently well suited to analysis of the large particle fraction, which would be highly desirable in view of its potential significance. Although particulate sulfate is "primarily associated with fine particles, large particles can be very sig- nificant in terms of total particle mass, par- ticulate nitrate, and particulate ammonium. The objective of this investigation was to develop a sampling procedure for large aerosols that would: 1) permit accurate de- termination of the mass and chemical com- position of large particles present in ambi- ent air over a 24-hour period; 2) employ a device that could be used in conjunction with an annular denuder and filter pack Couplers PTFE/FEP-Coated Bent-Tube Inlet TFE Sleeve Coupler PTFE/FEP-Coated Glass-lmpactor Nozzle (4.0 mm) Silicone-Rubber-Filled TFE Spacer TFE Impaction Disk Annular Denuder with Pedestal (Coated with Sodium Carbonate) 2nd Annular Denuder (Coated with Citric Acid) 3-Stage TFE Filter Pack Mass Flow Controller To Vacuum Figure 1. Glass-impactor annular-denuder filter-pack sampling system. 2 ------- without interference in the determination of the gaseous and fine-particle constituents; 3) avoid the formation of chemical artifacts; and 4) be simple enough to deploy in a nationwide monitoring network. Although other sampling devices were initially considered, the experimental work focused exclusively on a single-stage im- pactor modified to utilize a removable im- paction surface mounted on the end of the first annular denuder (Figure 1). This modified design was conceived by re- searchers at the University of Kansas and designed and constructed (in cooperation with University Research Glassware, Carrboro, NC) as part of another project previously funded by the U.S. EPA. Al- though it had not yet been fully character- ized, it was subsequently incorporated into sampling systems being used by several other teams of investigators. Experiments were conducted to exam- ine: the cut point of the modified impactor; impactor inlet losses; particle bounce utiliz- ing various impaction surfaces and impac- tion-surface coatings; horizontal versus vertical mounting; collection efficiency as a function of mass loading; particle deposi- tion patterns; the compatibility of various oils and greases with chemical analysis of impacted particles; the reactivity of various oils, greases, and impaction surfaces with gases (SO,,, HNO3, and NH3); the volatility and reactivity of impacted particles on coated and uncoated surfaces; and the precision of the method underfield sampling conditions. Procedure A. Sampling System Components The sampling system (Figure 1) con- sisted of the modified impactor coupled to a bent-tube inlet (ASTM method D3685-78) and the annular-denuderfilter-pack assem- bly previously described by Vossler et al. (Atmos. Environ., 22,8,1729-1736,1988). Several types of impaction disks were used. The most successful impaction disk (Figure 2) was machined from virgin TFE and featured a small (77-u.L) "well" 14 mm wide and 0.5 mm deep. This well was designed to accept either a small pool of oil (underlain by a 13-mm PC membrane filter) or an oiled or greased 13-mm membrane or glass-fiber -filter. l| was most frequently used with a silicone-oiled PC (Nuclepore) filter, which was found to migrate if placed on a flat impaction surface. Disks fitted with a coarse 10-mm fritted (porous) glass disk were also used. B. Oils and Greases Four greases and three oils were se- lected for study: petroleum jelly (Vaseline); high-vacuum silicone grease (Dow Corn- 16 mm i 14 mm 0.5 mm p 5.75 mm- Figure 2.' TFE impaction disk with well. ing); Halocarbon grease, type 25-5S (Halo- carbon Products Corp.); Fluorolube grease, type GR-290 (Hooker Chemical Co.); sili- .cone diffusion-pump oil (type 704, Dow Corning Corp.); and two mineral oils (Fisherbrand 19 and Duo Seal). C. Analytical Methods Anions were determined by ion chro- matography; ammonium was determined using an Alpkem continuous flow analyzer; and uranine was determined using a flu- orometer. Oiled and greased impaction surfaces were extracted with pentane and then reagent water; and the aqueous phase was subsequently analyzed for the con- stituents of interest. D. Particle Generation and Impac- tion Particles were generated using a Ber- glund-Liu Monodisperse Aerosol Genera- tor equipped with a Krypton-85 charge neu- tralizer. The size, shape, and uniformity of the particles were verified by scanning elec- tron microscopy (SEM). Particles spanning asize range of < 1 u.m to 30 u.m and particles composed of uranine, NHLCI, NH4NO3, (NH4)2SO4, NaCI, and NaHSO4 were used 2.5 mm t i 5 mm 1.5 mm 10 mm over the course of the investigation. Par- ticles generated in the laboratory were im- pacted by inserting the impactor inlet (with or without the bent-tube inlet) into either a wind tunnel or a small-diameter plastic pipe into which the particles had been introduced. The wind tunnel was used for low mass loadings and the pipe was used for high mass loadings. E. Preparation of Impaction Surfaces After scrupulous cleaning, impaction disks were prepared for use in one of the following ways: 1) oiled PC filters were prepared by carefully placing a clean 13- mm, 0.4-u.m-pore filter in the well and then placing the desired amount of oil (usually 5 u.L of silicone oil) on top of the filter; 2) uncoated PC filters were held in place by first adding a very small dab of silicone grease to the top of the impaction disk; 3) greased membrane filters were prepared by placing ten 50-u.L portions of a 1% solu- tion of grease (usually silicone) in pentane on top of the filter, allowing the solvent to evaporate after adding each 50-u.L portion; 4) greased impaction disks were prepared as described in item 3, omitting the filter; and 5) oiled fritted glass disks were pre- ------- Bell Connectors (To Mount Optional Denude~r) Norprene Tubing Air Purifier 47-mm Filter Filter Ambient ="-< Mr Air «Red« Purifier Coupler 1/4" FEP 1/8" TFE TFE Sleeve frnpactor Nozzle TFE Spacer Impaction Disk Annular Denuder Bell Connector 2-Way Valve (SS) Bypass Line (Norprene Tubing) 3-Way Valve (SS) TEE(SS) 1/4" ~FEP~ 2 5-mm Filter Microvalve Reservoir A/2@ 5 PSI Vacuum Figure 3. Impaction-surface exposure system. ------- pared by adding oil dropwiseto saturate the disk or adding a fixed amount of oil in a pentane solution as described in item 3 above. F. Tests for Chemical Artifact Formation Oils, greases, impaction surfaces, and impacted particles were exposed to various gases (SO2, HNO?, NH3, or purified air) by placing an appropriately prepared specimen into an impaction-surface exposure system (Figure 3) or an impacted-particle expo- sure system (similar to the system shown in Figure 3, except that a control sample without particles was exposed simulta- neously with the particles). A side-stream of nitrogen, carrying the reactive gas of inter- est, was introduced through a stainless- steel Swagelock tee. A solution of HNO3 and H2SO4 was used to generate HNO3 vapor; a solution of NH4OH was used to generate NH3 vapor; and a solution of H2SO3 and KH2PO4 was used to generate SO2 vapor. G. Field Test A 24-hr field test was conducted to determine the precision of the method. Four identical sampling systems were mounted on the roof of afour-story building (Learned Hall) on the campus of Kansas University. The bent-tube inlets were turned in the direction of the prevailing wind (north to northwest) on the day of sampling. Mass flow controllers having an accuracy of ±1 % and a precision of ± 0.2%, were used to maintain a flow of 12.5 Lpm (STP: 25°C and 1 atm); prior to use they were cross-cali- brated and found to agree within 0.5%. Results and Discussion A. Physical Performance Characteistics of the Modified Impactor The use of an oil-saturated impaction surface dictates a vertical impactor orien- tation (to prevent oil from dripping off), whereas the use of greaseorathin film of oil allows either a vertical or a horizontal ori- entation to be used. In the latter case, particle bounce may occur at high mass loadings. Since ambient winds are gener- ally horizontal or nearly so, non-isoaxial sampling conditions will exist when the im- pactor is used in the vertical position, and particles may be elutriated. A 90° bend can be used with a vertically positioned impac- tor to sample isoaxially and thus eliminate the elutriation effect, but the inlet bend presents an additional surface where par- ticle deposition and chemical artifact forma- tion may occur. Thus, regardless of which impaction surface coating is used, its influ- ence on the orientation, performance, and operation of the impactor must be consid- ered. To gain insight into certain aspects of the impactor's physical performance char- acteristics, experiments were conducted to examine: 1) the cut-point of the modified impactor; 2) particle collection efficiency under non-isoaxial sampling conditions; 3) inlet losses as a function of particle size; and 4) particle bounce as afunction of mass loading and impaction surface type. These experiments were not intended to provide a comprehensive solution to the age-old large- particle sampling problem. Rather, they were intended to characterize the physical performance (and limitations) of the im- pactor, to aid in the conduct and interpre- tation of subsequent experiments address- ing chemical artifact formation, to reveal problems requiring further investigation, and most of all to determine the size of particles that could be successfully sampled. The cut point of the modified impactor, when used with a PC membranef iltercoated with 5 |o.L of silicone oil, was experimentally determined to be 2.5 u.m, with less than 1% particle bounce for 8-u.m particles. Signifi- cant particle bounce was found to occur when using uncoated or grease-coated im- paction surfaces or when using a porous glass impaction surface (Figure 4). Satu- rating a porous glass impaction disk with oil reduced particle bounce to about 2%, but the impacted particles could not be easily removed from the porous glass impaction surface. Sudden increases in flowrate caused impacted particles to be dislodged from uncoated or greased surfaces. Non-isoaxial sampling resulted in sub- stantial undersampling of large particles, so isoaxial sampling (perhaps using a vane- mounted system) appears to be a necessity. Iso'axial sampling of ambient air requires either a horizontally mounted sampling sys- tem or a vertically mounted system with an inlet bend. The horizontal configurations tested performed poorly; greased and lightly oiled surfaces exhibited a high degree of particle bounce and oil tended to flow side- ways off oil-saturated disks. Vertically mounted systems employing a 90° bent- 100 90 80 70 T T T =T =T Upright, Membrane Filter, 5 \iL Si Oil Inverted, Porous Glass Dish, Mineral Oil (Sat'd) Inverted, Porous Glass Disk, 50 pi. SI Oil Inverted, Porous Glass Disk, 20 \iL Si Oil Upright, Porous Glass Disk, 20 pi. SI Oil WOO 2000 Mass of Uranine Sampled, micrograrns 3000 Figure 4. Impactor collection efficiency as a function of mass loading for solid 7.9-y.m uranine particles. ------- tuba \n\Q\ performed reasonably well, but Inlet losses were quite high for very large particles (e.g., 91 to 96%for26.5-u.muranine particles) and for blunt-tipped inlets (e.g., 7- 14%for8-u,muranineparticles). Inletlosses averaged only 2.7% for 7.9 urn uranine particles sampled using asharp-edged bent- tube Inlet in an upright position; the pattern of the deposited particles indicated the presence of a double vortexing flow condi- tion, which did not appear to adversely affect impactor performance. Losses of solid particles within the combined impactor denuder system (ex- cluding the inlet) were greatest for particles aboutthesizeofthecut point (Figures). For particles having an aerodynamic diameter of 2.0-2.8 urn, losses in the denuder ranged from 4-8% of the mass sampled, while losses on the spacer between the impactor and denuder ranged from 10-17% of the mass sampled. However, losses for particles of othersizes (largerand smaller) weregener- ally negligible (i.e., <1%). B. Tests for Chemical Artifact Formation Selected oils and greases that might be used to prevent particle bounce were examined.and experiments were conducted on the most promising candidates to de- termine: 1) their purity; 2) their compatibility with analysis of trace anions by ion chro- matography; 3) their gravimetric stability; 4) their ability to react with gaseous constitu- ents to form chemical artifacts; and 5) their 100 80 70 60 50 30 20 + Passing Impaction Disk Loss on Spacer A. Loss In Denuder 2345 678 Particle Aerodynamic Diameter, microns to ability to prevent reactions that might occur between reactive gasses and impacted particles. A number of oils and greases were found to be compatible with the analysis of impacted particles using ion chromatogra- phy, i.e., the anions of interest were not leached or sorbed, impacted particles ap- peared to be completely extracted, and no interfering peaks were observed. PC filters were found to be free of nitrate, nitrite, and sulfate. PC filters, siliconeoil, and Vaseline exposed to high concentrations of HNO3 and SO2 were found to adsorb little, if any, nitrate or sulfate. TFE impaction disks may adsorb HNO3; but there is no need to ana- lyze them, since the overlying impaction filters can readily be removed for analysis. NH4CI and NH NO3 particles impacted on an uncoated PC filter did not volatilize when exposed to purified air. Impacted particles of varying composition were im- pacted on uncoated, greased, or oiled sur- faces and subsequently exposed to high concentrations of potentially reactive gases (HNO3, SO2, or NH3) at room temperature and low relative humidity for four hours at a nominal flowrate of 16.7 Lpm. Table 1 summarizes the results. It should be noted that there may have been particles on the rims of the impaction disks exposed prior to August 18; and par- ticles may have been physically dislodged from the disks by the force of the jet prior to August 23, after which a "soft-start" proce- dure was implemented. Considering only those samples exposed after August 23, it can be seen that: 1) NaCI particles impacted on oiled orgreased filters reacted only slightly with HNO3; 2) NH4CI and NaCI particles impacted on uncoated and greased filters, respectively, did not react with SO2> nor did NH4NO3 particles impacted on an uncoated filter; 3) NH Cl particles impacted on greased filters reacted to only a very limited extent with HNO3; and 4) acidic NaHSO4 particles impacted on greased filters were partially (1.7 to 3.5%) converted to NaNH4SO4 by reaction with NH3. C. Field Test In a 24-hour field test employing four collocated sampling systems, the concen- trations of coarse-particle nitrate-N and sul- fate were determined to be 0.939 ± 0.032 u.g/m3 (± 3.4%) and 0.222 ± 0.007 u.g/m3 (± 3.3%), respectively. However, substantial percentages of the nitrate, sulfate, and ammonium (36%, 44%, and 50%, respec- tively) were collected on the bent-tube inlets rather than on the impaction disks. Figure 5. Particle tosses on the spacer and in the denuder as a function of particle size. 6 ------- Conclusions 1. The cut point of the modified im- pactorwas experimentally deter- mined to be 2.5 u.m, identical to its theoretical outpoint. 2. Non-isoaxialsampling resulted in a substantial reduction in sam- pling efficiency, with less than 10% of 8-(4.m particles entering an inlet orthogonal to the air stream. Accurate sampling in the field will require isoaxial sam- pling. When the samplers were mounted horizontally, minus the bent-tube inlets, significant frac- 4. tions (2.0-9.0%) of 8.34-jim uranine particles were collected in the impactor inlets, presum- ably as a result of sedimentation or turbulent diffusion. Extensive particle bounce oc- curred when using uncoated impaction surfaces. Hence, oil or Table 1. Summary of Particle Reactivity Experiments* Date (1988^ 8/01 8/02 8/03 8/04 8/05 8/09 8/09 8/12 8/15 8/16 8/18 8/19 8/22 8/23 8/31 9/07 9/14 9/21 9/28 ft /Ort 9/30 10/03 10/07 10/07 10/12 10/12 10/21 10/28 11/04 11/11 11/18 11/25 12/01 Particle Iwaa NH4CI NH4CI NH4CI NH4CI NH4CI NH4CI NH4CI NaCI NaCI NaCI After this date, NaCI NaCI Particle Oil or Mass. u.g Grease 62.2 None 87.0 None 129.4 106.8 74.0 83.4 32.7 26.8 148.4 70.9 impacted 239.7 389.8 Oil Oil Oil Grease Grease None Oil Grease Test Gas Type ng/m3 HNO3 338 HNO3 318 HN03317 HN03 301 HNO3 343 HNO3 328 HN03114 HN03 341 HNO3 340 HN03 320 particles were transferred to a fresh TFE Oil Oil After this date, the "soft start" procedure was NaCI NACI NACI NaCI 10.9 41.6 58.9 94.1 Oil Oil Oil Oil HNO3 304 HNO3 307 used. HN03 30.1 HN03 60.0 HN03 71.0 HNO3 58.1 Mass Found on Disk, ng 2.57 (N03-N) 3.13(NO3-N) 0.17(N03-N) 0.48 (NO3-N) 0.00 (NO3-N) 0.93 (NO3-N) 2.88 (NO3-N) 1.57(NO3-N) 0.72 (NO3-N) 2.83 (N03-N) impaction disk prior to 3.08 (N03-N) 2.94 (N03-N) 0.03 (NO3-N) 0.10 (NO3-N) 0.11 (NO3-N) 0.26 (N03-N) Displacement to Denuder JPJ1 %. B. Chloride 27.5 1.74 Chloride 24.6 1.79 Chloride Chloride Chloride Chloride Chloride Chloride Chloride Chloride exposure. Chloride Chloride Chloride Chloride Chloride Chloride 1.5 3.2 0.9 63.9 48.5 43.2 2.3 14.7 5.6 2.7 0.00 1.0 0.58 1.3 3.10 1.86 15.07 1.44 1.77 1.14 1.04 1.11 0.88 0.00 1.03 0.93 1.11 After this date, afresh citric-acid-coated denuder was used to scrub the inlet air when N-containing particles were exposed. NH.NO, 4 J NH4NO3 NH4NO3 NH4N03 NH4CI NH4CI NaCI NaCI NaCI NH4CI NH4CI NaHSO4 NaHS04 199.9 20.1 208.6 298.8 78.4 46.8 159.7 207.6 190.6 72.6 163.0 461.9 495.0 Oil None None None None None Grease Grease Grease Grease Grease Grease Grease SO2 371 SO2 337 SO2 486 Air SO2 452 Air SO2 375 HN03101 HN03118 HN03 139 HN03 125 NH3 39.8 NH3 114 0.11(SO4) 0.00 (SO4) 0.00 (SO4) NA 0.00 (SO4) NA 0.00 (S04) 0.33 (NO3-N) 0.73 (NO3-N) 1.48(N03-N) 1.06(N03-N) 1.90{NH4-N) 1.00(NH4-N) Nitrate Nitrate Nitrate Nitrate Chloride Chloride Chloride Chloride Chloride Chloride Chloride 0.00 0.00 0.67 0.11 4.8 3.6 0.50 0.45 1.34 7.43 0.00 (3.5%toNaNH4S04) (1.7%toNaNH4So4) 0.00 NA NA 0.66 0.84 0.96 0.00 Particle mass = initial mass on particles on the impaction disk; oil =silicone oil; grease = high-vacuum silicone grease; R is the control- corrected raiio of moles of ion displaced to moles of contaminant ion found on the impaction filter; NA = not applicable ------- grease must be used to prevent particle bounce. 5. When using a silicone-greased impaction surface, significant particle bounce occurred at a uranine-particle mass loading of only 38 u.g, about 70% of a mono- layer. Hence, use of silicone grease will limit mass loadingsto very low levels unless a rotating surf ace or other means of reduc- ing the mass of particles per unit area is developed. 6. Using a porous glass impaction surface coated with 50 uL of silicone oil, 27-29% of the par- ticle mass bounced off the im- paction surface when sampling 7.9-u.m uranine particles at a mass loading of only 4-5 |ig. Saturating the porous glass sur- face with mineral oil reduced particle bounce to approximately 2% of the particle mass sam- pled. Under severe mass load- ing conditions (126-2,920 u.g), only about 1% bounce was ob- served using a PC filter coated with only 5 \iL of silicone oil. 7. Since excessive bounce oc- curred with uncoated, greased, or lightly oiled impaction sur- faces, and since the oil tended to flow sideways off oil-saturated porous glass impactions disks when the sampling system was horizontally mounted, it was de- termined that isoaxial sampling could best be accomplished us- ing a vertically mounted system with a bent-tube inlet. 8. Inlet losses were observed to be 91 to 96% for 26.5-u.m uranine particles sampled using sharp- edged 90° bent-tube inlets, 48- 52% for 15.2-u.m ammonium- suifate particles sampled using bent-tube inlets constructed from thin-walled glass tubing, 7-14% for 8-u,m uranine particles sampled using blunt 90° bent- tube inlets, 3.7-9.5% for 7.9-u.m uranine particles sampled using sharp-edged 90° bent-tube inlets in an inverted position, and 0.5- 5.3% (averaging 2.7%) for 7.9- ]im uranine particles sampled using sharp-edged 90° bent-tube inlets in an upright position. 9. For 7.9-u.m uranine particles, at least 12-15% of the total particle mass sampled using an oiled porous glass impaction surface became entrapped within the pores and was not readily ex- tractable for subsequent chemi- cal analysis. Uranine particles collected on oiled membrane fil- ters were readily extracted. 10. Deposition of 7.9-u.m uranine particles collected on a silicone- oiled PC filter indicated the pres- ence of a double-vortexing flow condition, presumably resulting from the use of the bent-tube in- let. The particle collection char- acteristics of the impactordonot appear to be adversely affected by this condition. 11. Particle losses in the combined impactor-denuder system (ex- clusive of the inlet losses) were found to occur predominantly with particles about the size of the impactor's cut-point diam- eter. Within the denuder, losses of particles having an aerody- namic diameter of 2.0-2.8 u.m ranged from 4-8% of the total mass sampled. Particles were also lost on the TFE spacer be- tween the impactor and the de- nuder, representing 10-17% of the mass sampled for 2.0-2.8 u.m particles. Losses of particles having other aerodynamic diam- eters were negligible. 12. Analysisoffourgreasesand three oils revealed that they were of suitable purity and were compat- ible with analysis of trace con- centrations of anions by ion chromatography after extraction with pentane and water. They did not generally leach or sorb sig- nificant amounts of the anions of interest. The mineral oils were found to contain detectable amounts of extractable sulfate, but not enough to be significant when a small quantity of oil is used in a 24-hour test. 13. Silicone oil and Vaseline ap- peared to be compatible with gravimetric analysis of impacted particle mass. However, gravi- metric analysis would only be useful for high mass loadings (i.e., those for which weighing would give reasonably accurate results), a dust-free weighing room might be needed, and addi- tional testing is needed to de- termine how long the samples must be equilibrated prior to weighing. 8 14. PC filters were found to be free of nitrate, nitrite, and sulfate; but mixed-fiber (cellulose acetate/ cellulose nitrate) filters contained measurable amounts of leach- able nitrate. 15. Silicone oil, PC filters, and Vaseline exposed to HNO3 and SO2 were found to sorb little if any nitrate or sulfate. The traces of nitrate and sulfate which were often detected in exposed speci- mens may very well have been due to handling rather than to chemical artifact formation; and the amounts detected would generally not be significant for samples collected over a 24- hour period. 16. TFE impaction disks may adsorb traces of HNO ; but it is not nec- essary to analyze the TFE disk when it is overlain by a PC filter. 17. NH4CI and NH4NO particles im- pacted on uncoated PC filters did not volatilize from the collection surface when exposed to purified air. 18. NH4CI, NaCI, and NH4NO3 par- ticles impacted on uncoated or greased PC filters did not react with SO2. 19. NaCI and NH NO3 particles im- pacted on a silicone-oiled PC fil- ter did not react to a significant extent with HNO3 or SO2, re- spectively. 20. Acidic NaHSO4 particles im- pacted on a silicone-greased PC filter reacted to a limited extent (1.7 to 3.5%) with NH3. 21. In a 24-hour field test employing four collocated sampling sys- tems incorporating the modified impactors, the concentrations of coarse-particle nitrate-N and sul- fate were determined to be 0.939 ±0.032 ug/m3 (±3.4%) and 0.222 ± 0.007 ng/m3 (± 3.3%), respec- tively. However, substantial per- centages of the nitrate, sul- fate, and ammonium (36%, 44%, and 50%, respectively) were col- lected on the bent-tube inlets rather than the impaction disks. Recommendations 1. Impaction surfaces used to sample particles (or to prevent their entry into downstream de- nuders and filter packs) should be coated with a suitable oil or ------- grease to prevent bounce. At high mass loadings, excess oil should be used. 2. If an oiled porous glass disk is used to collect particles for sub- sequent chemical analysis, the analyst should make sure that the extraction procedure used to remove the particles from the disk is adequate. 3. The nature of the nitrate and sulfate collected on the impactor inlet during collection of field samples should be investigated, and experiments should be con- ducted to ascertain whether jarge particles deposited in the inlet will react to a significant extent with HNO3, SO.,, NH3, or other gaseous pollutants. 4. A small vane system able to turn a bent-tube inlet mounted on a rotary union and orient it into the wind should be developed and field tested. 5. The efficiency with which both anions and cations associated with various types of particles can be extracted from oils and greases into water should be thoroughly examined and docu- mented. 6. The procedure proposed herein for gravimetric determination of large particles should be perfected and field tested. 7. The reactivity of gases (not only HNO SO and NH,, but also O3, NO, NO,,, PAN, and espe- cially SO3) with oils and greases, with various types of Teflon and Teflon coatings, with various types of particles (especially acidic and basic particles) col- lected on oiled and greased sur- faces, and with denuders and their coatings should be explored in much more detail. The effects of temperature and relative hu- midity on such reactions also need to be thoroughly examined. 8. Additional field tests should be conducted to determine the pre- cision of the procedure under a variety of ambient conditions and to explore ways to improve the precision and to simplify the pro- cedure for use by individuals with minimal experience. 9. Research should continue toward the development of a sampling system capable of quantitatively sampling very large (i.e., 15-20 u.m) particles. 10. The sampling system herein de- scribed should be thoroughly ex- amined vis-a-vis its potential use in sampling acid aerosols. Par- ticular attention should be given to: interactions between gases and impacted particles; proce- dures for quantifying the acidity or basicity of impacted particles; and loss or alteration of submi- cron particles as they pass through the denuders. ------- ------- ------- SJ. Randtke, D.D. Lane, and T. E. Baxter are with the University of Kansas, Lawrence, KS 66045. Russell Wiener Is the EPA Project Officer (see below). The complete-report, entitled "Development of a Sampling Procedure for Large Nitrogen- and Sulfur-Bearing Aerosols," (Order No. PB90-235 789/AS; Cost: $23.00 cost subject to change) will be available only from: National Technical Information Service 5285 Port Royal Road Springfield, VA 22161 Telephone: 703-487-4650 The EPA Project Off her can be contacted at: Atmospheric Research and Exposure Assessment Laboratory U.S. Environmental Protection Agency Research Triangle Park, NC 27711 United States Environmental Protection Agency Center for Environmental Research Information Cincinnati, OH 45268 Official Business Penalty for Private Use $300 EPA/600/S3-90/048 . ------- |