- Email: [email protected]
A microslot impactor for organic aerosol sampling
Pergamon
Vol. 1. Aerosol SC;. ,C’ 1997 Elsevier
PII: SOO21-8502(96)00484-3
A MICROSLOT
IMPACTOR
28, No. 7, pp. 1283S1290, 1997 Science Ltd. All rights reserved Printed in Great Britain mm-8502/97 s17.00 + 0.00
FOR ORGANIC AEROSOL SAMPLING
Susanne Hering,* Lara Gundel+ and Joan M. Daisey+ *AerosolDynamics ‘Indoor
Environment
Program,
Energy
Inc., Berkeley, CA 94710, U.S.A. and Environment Division, Lawrence Berkeley Berkeley, CA 94720, U.S.A.
(Received 4 December
1995; accepted
National
Laboratory,
28 October 1996)
Abstract-A microslot impactor has been designed with the objective of collecting large, concentrated samples suitable for chemical analysis of trace organic constituents. It employs a single slotted orifice per stage, with orifice widths as small as 0.1 mm. Using a relatively small pump (13 kg), five cutpoints from 0.1 to 3.5 pm are obtained at a flow rate of 36 Imin-’ (6 x 10e4 m3s-‘). The microslot impactor operates with incompressible flow and minimal pressure drops for the upper four ~I__._ -..A --..1-a-... .L-_.._I. .L- ,--A ,ClIL\ ^I__^ ~I.^^___^_I _..r__l_r. P^_.L_ :________.11-,_ a_... sragcs, anu some uow lnrvugn lue I~SL(mm, slage. IYI~~LSUI~‘ UUL~O~I~~ U 101 hue IIIGVIII~ITS~~UI~: uvw stages are somewhat higher than predicted, with critical Stokes numbers corresponding to 10 to 30% greater than theory. 01997 Elsevier Science Ltd
INTRODUCTION
The distribution of specific chemical species with respect to particle diameter is an important factor in evaluating their effects, their sources, and their formation mechanisms. Speciated size distributions for submicrometer particles are especially of interest for the study of respirable aerosols that may affect human health. Traditional cascade impactors, which are used for the size-segregated particle collection for species size distributions, do not offer much size resolution below 0.4 pm. For sampling of cigarette smoke or photochemically generated smog aerosolq as much as half of the mass can penetrate the last stage of the impactor to be collected on the backup filter. In order to obtain smaller cutpoints with impactors, two types of impactors have been developed: micro-orifice impactors and low-pressure impactors. Both instruments can provide cutpoints as small as 0.05 ,um. Micro-orifice impactors employ very small orifices, 40 to 200 pm in diameter, and operate more closely to atmospheric pressure than do low-pressure impactors. The MOUDI, a micro-orifice impactor developed at the University of Minnesota, has been used extensively for measurements on ambient aerosols (Marple et al., 1991). Low-pressure impactors resemble ordinary impactors but are operated at reduced pressures, as low as 5 to 40 kPa, which decrease the particle aerodynamic drag. An advantage of this design is that the aerosol sample on each stage can be concentrated in a 1 mm diameter spot that may be analyzed directly by a variety of methods for sulfur ITT~. and Friediander, (~enng i982j or carboil iAiien ei (ll., 1 nnr\ A ~~__ l-.--L_^ :_ cl-r rl1-__. lYY4).
‘4
UlS:iiUViilllq+
IS
llkll
LIIC: IUW
pressure can lead to loss of volatile species during sampling (Biswas et al., 1987). In this work we report the development of a hybrid, a microslot impactor designed to take advantage of the small deposit area of low-pressure impactors as well as the higher pressure of the micro-orifice impactor. The design was chosen to be suitable for use with the particle trap collection surface of Biswas and Flagan (1988), who found that the particle trap eliminates the need for a grease adhesive on the impaction surface and permits larger loading of collected material within the impactor. The microslot design has the additional advantage that the impactor can be operated with a smaller pump than either the MOUDI or Berner lowpressure (Berner et al., 1979) impactors, which have comparable cutpoints and sampling rates. XITrD~PT ,*IILI,vu~v
n-r
I
TA,fDApTnlJ 11*11 KxL
IV1\
nE‘ cTc\l YL.“IUI.
AhTn _I._
rnNc:TPlTr7-TnN UVI.“AI\“UIIVI.
The microslot impactor has five stages, with a single slotted orifice per stage. The impactor follows a rather traditional design, with incompressible flow and minimal pressure 1283
Susanne
1284
Hering
et uf.
drops in the upper four stages, and sonic flow through the last (fifth) stage. The flow rate of 36 lmin-‘(6 x 104m3 s-l ) is controlled by the critical flow through the final stage. Particles bypassing the final stage are collected by an after-filter. Impactor orifice dimensions and stage operating pressures are shown in Table 1. The dimensions for the impactor slots were selected on the basis of the Stokes number for 50% collection efficiency: St
=PD:oUC
(1)
50
9VW
where p is the particle density (g cmp3), D,, is the particle diameter at 50% collection efficiency, C is the Cunningham slip factor, U is the jet velocity, p is the air viscosity, and W is the slot width. Following the guidelines of Marple and Rubow (1986), a value of St50 = 0.59 was assumed in the design calculations. The jet velocity for the first four stages is calculated from the incompressible Aow relation, U = QP,/PLW, where Q is the volumetric flow rate at the inlet pressure PO, P is the pressure immediately above the stage, and L is the orifice length. For the final stage the jet velocity reaches the sonic limit and
Table 1. Dimensions
Stage no.
Orifice width W (mm)
Orifice length L (mm)
1 2 3 4 5
1.38 0.56 0.31 0.18 0.10
33.4 41.3 42.4 31.5 31.5
and operating
parameters
for microslot
Flow Re
Orifice length to width, L/W
Orifice-plate separation to Pressure* width, S/W (kPa)
1144 808 901 1019 2040
24 84 137 211 368
1.8 1.8 1.6 2.1 6.5
100.4 100.3 100.0 98.7 92.0
impactor
Orifice velocity (m s-‘) 13.0 22.7 45.6 89.9 314
Calibration stokes no.’
Cutpoint (flm)
0.76 0.64 0.52
3.5: 1.7: 0.84 0.38 0.10
*Absolute pressure above stage orifice at inlet pressure of 100.4 kPa. +Stokes number at measured 50% efficiency cutpoint. t Estimated based on Stokes number of 0.76 which was derived from Stage 3 calibration.
Stage 1 Stage 2
Stage 3
Stage 4
Stage
0.030” stainless shim welded to form orifice slit of width W.
5
Orifice Plate, Bottom View
Microdot lmpactor Fig. 1. Schematic
of microslot
impactor,
showing
enlargement
Orifice Plate, Cross-sectional View
of orifice plate.
A microslot
impactor
for organic
aerosol
sampling
1285
U = J2y/(y + 1) RTO (where y is the heat capacity ratio, R the ideal gas constant and To the absolute temperature; Biswas and Flagan, 1984; Liepmann and Roshko, 1957). The impactor design also takes into account the orifice Reynolds numbers, defined by Re = PairW U/p where Pair is the density of air. For the microslot impactor 800 < Re -C 2100 (see Table l), which is in the range recommended by Marple and Rubow for optimum impactor performance. The construction of the impactor is shown in Fig. 1. The impactor body is anodized aluminum. The orifice plates are stainless steel and are removable. Pressure taps on each stage permit the measurement of operating pressures. The orifice slots were obtained by spot-welding two pieces of 0.030 in (0.76 mm) stainless steel shim stock to the orifice plate. The two pieces are aligned by means of another piece of shim stock of the thickness of the slot width. The length of the orifice is determined by a slot 0.125 in (3.18 mm) wide, milled into the orifice plate. The width of the orifices obtained by this fabrication process equaled the design specification for the last (fifth) stage but were slightly larger than specifed for the third and fourth stages. The dimensions shown in Table 1 are the actual dimensions, taken from measurements of the impactor orifices. The orifice-to-plate spacing, S, is determined by the thickness of the “feet” mounted at the ends of the orifice slots, and it varies from 1.6 to 2.1 W except for the fifth stage, where S/W = 6.5. The impactor is operated by a 13 kg, 500 W, carbon vane pump.
CALIBRATION
Experimental
PROCEDURES
methods
The impactor was calibrated with aerosol comprised of dioctyl sebacate (DOS), a nonvolatile liquid with a specific gravity of 0.92. The experimental configuration, shown in Fig. 2, follows the approach described by Marple et al. (1991). DOS aerosol was generated
I
Microdot lmpactor
Pinch Clamp Dilution Air1 >c Sheath Air Optical Particle Counter
High-Flow Differential Mobility Analyzer
Vacuum Throttling Valve >(
I
Flow Metering Orifice
Vacuum
Fig. 2. Calibration system showing aerosol generation with atomizer followed by a high-flow differential mobility analyzer and detection for test stage efficiency with an aerosol electrometer. An optical particle counter is used to monitor the calibration aerosol.
Susanne Hering et ~1.
1286
by atomization of a solution of DOS in isopropanol and passed through a high-flow differential mobility analyzer (HF-DMA) in order to select particles of a specified electrical mobility (Stolzenburg et al., 1996). The high-flow differential mobility analyzer is larger but similar in geometry to the mobility analyzer designed by Liu and Pui (1974). Specifically, the geometric factor L,/[ln(R,/R,)] which enters into the mobility equations is larger by a factor of 6.7 (where Ld is the length of the drift tube and R2 and R, are the radii of the annular gap). As a result, it can handle larger flow rates for the same aerosol mobility. For this calibration the sheath air flow was fixed at 15 lmin- ‘, with aerosol input and output providing a mobility window of 20%. flows of 3 lmin-‘, For calibration, the aerosol concentrations immediately upstream and immediately downstream of the impactor test stage were measured with an aerosol electrometer taken from a TSI Model 3030 (St. Paul, Minnesota). The pressures above and below the test stage, and the flow rate through the impactor were monitored throughout the calibration tests. An optical particle counter (Particle Measuring Systems model ,uLPC-HS, Boulder, Colorado), coupled to a pulse amplifier and 1024-channel multichannel analyzer, was used to check the size of generated particles and to count the relative proportion of singly and doubly charged particles for particles above 0.14 pm. Measurements on a blank stage, i.e., one without an orifice, were used to confirm equal transmission of aerosol through the upstream and downstream sampling lines. Although the calibration particles carry unit charge, this is not expected to alter the cutpoint. Ejiciency
calculations
The calibration aerosol from the HF-DMA contains both singly and doubly charged particles of equal electrical mobility, and this was taken into account in the evaluation of the stage collection efficiencies. The collection efficiency for singly charged particles was calculated by the relation: y1
100
=
(
1 _
B _ (%- y$
-f)
4f
>
(2)
where A is the electrometer current measured above the test stage, B is the current measured below the test stage, y2 is the stage collection efficiency for the larger, doubly charged particles of equal electrical mobility, and f is the fraction of the electrometer current A attributable to singly charged particles. For particles above 0.14 pm the ratio of doubly to singly charged particles was measured by the optical particle counter, and f was calculated directly. For smaller particles, the relative number of multiply charged particles was calculated on the basis of the Boltzmann equilibrium charge distribution and the input size distribution, as indicated by the current as a function of HF-DMA voltage. Specifically, the fraction of the current attributed to singly charged particles at HF-DMA voltage I’ is I” - I v+2-
I v+3
f= IV =
1 _
=2,(12v
-
2G4vI4v
IV
-
3H,vIw)
- 3H3vtI3v
- 2GJw)
(3)
IV
where Iv = the total aerosol current measured at voltage I/, Iv+2 = is current attributed to doubly charged particles at voltage I/, Iv+3 = is current attributed to triply charged particles at voltage I’, Izv = the total measured current at voltage 2V, GZV = the equilibrium ratio of doubly to singly charged particles at the particle corresponding to voltage 2V, H,, = the equilibrium ratio of triply to singly charged particles at the particle corresponding to voltage 3V,
size size
A microslot
impactor
for organic
aerosol
sampling
1287
and Z3”, I,, and IsV are defined analogous to Izv, and GdV and HGV are defined analogous to Gzv and H3”. This form of the correction accounts for doubly and triply charged particles at HF-DMA voltages of I/ and 2V but ignores the contribution of multiply charged particles at voltage 4V and 6V. CALIBRATION
RESULTS
Calibration curves for Stages 3-5 are shown as a function of particle aerodynamic diameter in Fig. 3. Measured 50% efficiency cutpoints are 0.84, 0.38, 0.10 ,um. Repeated calibration runs were made with different concentrations of DOS in the atomizer solution,
90 80
20
Stage
3, Run l-48 (50%
stage
4. Run II-41
Stage
4, Run l-43 (6%
Stage
4, Run l-52 (50%
DOS)
(O.i%OOS) DOS) DOS)
Stage
5, Run Ii-49
(0.04%
DOS)
Stage
5, Run II-51
(0.01%
DOS)
Stage
5, Run II-55
(0.01%
DOS)
10 0 0.00
0.10
0.20
0.30
0.40
Particle
0.50
Aerodynamic
0.60 Diameter
0.70
0.80
0.90
1 .oo
(pm)
Fig. 3. Collection efficiency as a function of particle aerodynamic diameter shown for several runs, at differing DOS solution concentrations
for Stages 3-5. Data are in the atomizer.
100
T 90 80 i 70 i
3 g
L” s
.v
2 s
.t
50 40 X Theoretical Corr.
30 60 I
??
0.00
0.10
0.20
0.30
0.40
0.50
0.60
OPC Meas. Corr.
0.70
0.80
0.90
1.00
Aerodynamic Diameter (pm) Fig. 4. Comparison of the efficiency calculations for Stage 4 obtained by correction for multiply charged particles measured by the optical counter to that obtained by correction based on Boltzmann charge equilibrium assumptions.
Susanne
1288
Hering
ef cd
+-
0.6
12
1.4
for Stages 3 -5 as a function
of particle
0.8
1.0
--+--~“--_l
1.6
1.8
2.0
dStokesNumber
Fig. 5. Collection
efficiency
Stokes number.
and thus differing proportion of multiply charged particles in the calibration aerosol, showed consistent results. For the calibration of Stage 5, all corrections for doubly charged particles were done using the formulation of equation (3) because particle sizes were too small to be efficiently detected optically. For Stage 4, stage collection efficiencies evaluated using the theoretical estimation of doubly charged particles of equation (3) are compared to that from the direct measurement off’ from the optical particle counter. These results, shown in Fig. 4, indicate no effect on the collection efficiencies at and above the 50% cutpoint but do show some discrepancies in the shape of the low-efficiency portion of the collection curve. In Fig. 5 the collection efficiencies are shown as a function of the square root of the particle Stokes number, defined in accordance to the relation of equation (1). Although the collection efficiency curves for three stages are similar, they are not exactly the same. The critical Stokes number for collection varies from 0.52 for Stage 5 to 0.76 for Stage 3. The theoretical value for slotted impactors is 0.59 (Marple and Rubow, 1986).
CHAMBER
STUDY
COMPARISON
The microslot impactor was used in a chamber study of environmental tobacco smoke to measure polycyclic aromatic hydrocarbon distributions. It was operated in parallel with the integrated organic vapor-particle sampler of Gundel et al. (1995). The vapor-particle sampler (VPS) uses an annular denuder, a filter, and a second annular denuder operated in series to collect gas-phase, particle-phase, and volatilized particle-phase species, respectively. The impactor was operated with uncoated aluminum foil impaction surfaces, and with a Teflon-coated glass fiber after-filter. Extracts from both the denuders, the impaction substrates, and the filters were analyzed for polycyclic aromatic hydrocarbons. Particle-phase concentrations measured by both methods for pyrene, triphenylene, 1,2 benzofluorene, benzo(a)anthracene, chrysene, benzo(b)fluoranthene, benzo(k)fluoranthene and benzo(a)pyrene are shown in Table 2. The agreement between the two methods was better than 15% for these compounds.
PARTICLE
TRAP
EXPERIMENTS
In the chamber study of environmental tobacco smoke the impactor was operated with uncoated substrates because smoke aerosol is inherently a sticky substance. In general, impactor collection surfaces must be coated with grease or another adhesive in order to
A microslot
impactor
for organic
aerosol
sampling
1289
prevent “bounce”, or rebound of particles from the collection surface. But the carbon content of grease coatings poses a problem for organic analyses. Biswas and Flagan (1988) developed a particle trap impactor that does not require an adhesive coating for the collection of solid particles. It consists of an empty cavity below the impaction jet, with an opening equal to 1.9 jet diameters, and it is designed to provide pocket of still air into which the particles impact. This approach is attractive for organics analysis because of its capacity to collect large samples without contamination from an adhesive coating. In fact, one of the motivations behind the design of the microslot impactor is its greater suitability to a particle trap impaction surface than a multi-orifice impactor. Several types of slotted particle trap collection geometries were tested on Stage 4 of the microslot impactor. Trap # 1 was a simple slotted cavity with an opening equal to 2W and a depth of 16 W where W = width of the impactor orifice. Trap # 2 and Trap # 3 each had a lip with an opening equal to 2 W, an internal width of 6 W, and a depth of 8 W. In all cases the length of the collection slot was equal to the length of the orifice slot. The traps were aligned with the impactor orifice visually, using a bright light on the upstream side of the impactor orifice. A sketch of these traps and the calibration results for liquid aerosols are shown in Fig. 6. The data show that the lip was necessary to achieve high collection efficiencies, yet even with the lip the maximum collection efficiencies are about 90%. With Trap # 3 experiments were also conducted with solid particles, and it was found that the solid particles bounced. The extent of bounce could not be measured because the particles picked up a contact Table 2. Comparison
of particle
phase concentrations of PAH measured and microslot impactor PAH particle
Species
Vapor-particle
1,2-Benzofluorene Benz(a)anthracene Chrysene Benzo(b)fluoranthene Benzo(k)fluoranthene Benzo(a)pyrene
2.6 10.4 30.1 9.4 2.4 5.4
sampler
concentration (VPS)
by vapor-particle
sampler
(ng m-‘)
Microslot
impactor
2.2 11.8 29.6 9.4 2.5 6.2
Impactor/VPS
ratio
0.85 1.13 0.98 1.00 1.05 1.14
. Particle Trap #I A Particle Trap #2 Particle Trap #3
20 i
??
10
3 FlatSurface. t
oI,-+ 0.00
I
I 0.10
0.20
0.30 0.40
0.50 0.60 0.70
0.80 0.90
1.00
AerodynamicDiameter(pm)
Fig. 6. Stage 4 collection efficiency for liquid particles using particle traps of two different Open circles show results for a normal flat surface geometry.
designs.
1290
Susanne
Hermg
et ul.
charge on rebound, presumably from the clear plastic used on the back surface of the trap. (Contact charge refers to the transfer of electric charge from the impaction surface to the particle during contact.) In contrast to the work of Biswas and Flagan, recent work by Tsai and Cheng (1995) found rebound of solid particles from particle traps.
SUMMARY
An impactor employing a single. narrow slotted orifice per stage has been designed to permit the collection of sufficently large and concentrated samples suitable for chemical analysis of trace constituents. Calibration with liquid aerosol yields 50% efficiency cutpoints of 0.10,0.38 and 0.84 /lrn for the lowest three stages. The smallest cutpoint of 0.10 pm is obtained with a pressure above the stage which is equal to 92% of the inlet pressure. Measured cutpoints for the incompressible flow stages are somewhat higher than predicted, with critical Stokes numbers 10 to 30% greater than theory. The microslot impactor was successfully used in a chamber study of environmental tobacco smoke, and measured concentrations of particle-phase polycyclic aromatic hydrocarbon compounds species were comparable to those determined by a denuder-filter-denuder system. Preliminary tests with particle trap collection surfaces showed that with a lipped trap the liquid particle collection efficiency is as high as 90%; however, solid particles were found to bounce out of the trap. Aclino~~lerl~lenle,lrs This work was supported by Grant I ROl HL42490from the National Institutes of Health to the Regents of the University of California and administered by Lawrence Berkeley Laboratory. The views and opinions of the authors expressed herem do not necessarily state or reflect those of the United States Government or any agency thereof. or the Regents of the University of California, and shall not be used for advertising or product endorsement purposes.
REFERENCES Allen. D. T.. Palen. E. J.. Haimov, M. 1.. Hering, S. V. and Young, J. R. (1994) Fourier transform infrared spectroscopy of aerosol collected in a low-pressure impactor: method development and field calibration. Aerosol Sci. Tr~hd. 21, 3255242. Berner. A.. Lurzer. C. H.. Pohl. L.. Preining. 0. and Wagner. P. (1979) The size distribution of the urban aerosol in Vienna. Sci. TotcilEwir. 13, 2455261. Biswas, P., Jones, C. L. and Flagan. R. C. (1987) Distortion of srze distributions by condensation and evaporation in aerosol instruments. Aero,\o/ Sc,i. Tec~hm~l. 17, 231 -246. Biswas. P. and Flagan. R. C. (19X4) High-velocity inertial impactors. Entlir. Sci. Technol. 18, 611-616. Biswas. P. and Flagan, R. C. (198X) A particle trap impactor. J. Aero.d Sci. 19, 113-121. Gundel, L. A.. Lee, V. C.. Mahanama. K. R. R., Stevens. R. K. and Daisey, J. M. (1995) Direct determination of the phase distrtbutions of polycyclic aromatic hydrocarbons using annular denuders. Atmos. Enoiron. 29, 1719-m1733. Hering. S. V. and Friedlander, S. K. (I 982) Origins of aerosol sulfur size distributions in the Los Angeles Basin. Atmos. Elrriron. 16, 2647~ 2656. Liepmann. H. W. and Roshko. A. (1957) Elemrnts of Gus Dgnumics. Wiley, New York. Lui, B. Y. H. and Pui. D. Y. H. (1974) A submicron aerosol standard. J. Colloid Intrrfucr Sci. 47, 155171. Marple. V. A. and Rubow. K. L. (1986) Theory and design guidelines. In Cuscude lmpuctor Sampliny und Datu Amdy~~is(edited by Lodge, J. P. and Chan. T. L.), pp. 79- 102. American Industrial Hygiene Association, Marple. V. A.. Rubow. K. L. and Behm, S. M. (1991) A microoritice uniform deposit impactor (MOUDI): description. calibration and use. Acrosol SC,;. Techol. 14, 436446. Stolzenburr, M. R.. Kreisberg. N. M. and Herinr, S. V. (in press) Atmosoheric size distributions measured bv differential mobility opticalparticle srze spectrometry. ,4r&l SC;. Teclr~rol. (in press). Tsai. C.-J. and Cheng, Y.-H. (1995) Solid particle collection characteristics on impaction surfaces of different designs. Aerosol Sc~i. ‘khnol. 23, 96- 105.
Vol. 1. Aerosol SC;. ,C’ 1997 Elsevier
PII: SOO21-8502(96)00484-3
A MICROSLOT
IMPACTOR
28, No. 7, pp. 1283S1290, 1997 Science Ltd. All rights reserved Printed in Great Britain mm-8502/97 s17.00 + 0.00
FOR ORGANIC AEROSOL SAMPLING
Susanne Hering,* Lara Gundel+ and Joan M. Daisey+ *AerosolDynamics ‘Indoor
Environment
Program,
Energy
Inc., Berkeley, CA 94710, U.S.A. and Environment Division, Lawrence Berkeley Berkeley, CA 94720, U.S.A.
(Received 4 December
1995; accepted
National
Laboratory,
28 October 1996)
Abstract-A microslot impactor has been designed with the objective of collecting large, concentrated samples suitable for chemical analysis of trace organic constituents. It employs a single slotted orifice per stage, with orifice widths as small as 0.1 mm. Using a relatively small pump (13 kg), five cutpoints from 0.1 to 3.5 pm are obtained at a flow rate of 36 Imin-’ (6 x 10e4 m3s-‘). The microslot impactor operates with incompressible flow and minimal pressure drops for the upper four ~I__._ -..A --..1-a-... .L-_.._I. .L- ,--A ,ClIL\ ^I__^ ~I.^^___^_I _..r__l_r. P^_.L_ :________.11-,_ a_... sragcs, anu some uow lnrvugn lue I~SL(mm, slage. IYI~~LSUI~‘ UUL~O~I~~ U 101 hue IIIGVIII~ITS~~UI~: uvw stages are somewhat higher than predicted, with critical Stokes numbers corresponding to 10 to 30% greater than theory. 01997 Elsevier Science Ltd
INTRODUCTION
The distribution of specific chemical species with respect to particle diameter is an important factor in evaluating their effects, their sources, and their formation mechanisms. Speciated size distributions for submicrometer particles are especially of interest for the study of respirable aerosols that may affect human health. Traditional cascade impactors, which are used for the size-segregated particle collection for species size distributions, do not offer much size resolution below 0.4 pm. For sampling of cigarette smoke or photochemically generated smog aerosolq as much as half of the mass can penetrate the last stage of the impactor to be collected on the backup filter. In order to obtain smaller cutpoints with impactors, two types of impactors have been developed: micro-orifice impactors and low-pressure impactors. Both instruments can provide cutpoints as small as 0.05 ,um. Micro-orifice impactors employ very small orifices, 40 to 200 pm in diameter, and operate more closely to atmospheric pressure than do low-pressure impactors. The MOUDI, a micro-orifice impactor developed at the University of Minnesota, has been used extensively for measurements on ambient aerosols (Marple et al., 1991). Low-pressure impactors resemble ordinary impactors but are operated at reduced pressures, as low as 5 to 40 kPa, which decrease the particle aerodynamic drag. An advantage of this design is that the aerosol sample on each stage can be concentrated in a 1 mm diameter spot that may be analyzed directly by a variety of methods for sulfur ITT~. and Friediander, (~enng i982j or carboil iAiien ei (ll., 1 nnr\ A ~~__ l-.--L_^ :_ cl-r rl1-__. lYY4).
‘4
UlS:iiUViilllq+
IS
llkll
LIIC: IUW
pressure can lead to loss of volatile species during sampling (Biswas et al., 1987). In this work we report the development of a hybrid, a microslot impactor designed to take advantage of the small deposit area of low-pressure impactors as well as the higher pressure of the micro-orifice impactor. The design was chosen to be suitable for use with the particle trap collection surface of Biswas and Flagan (1988), who found that the particle trap eliminates the need for a grease adhesive on the impaction surface and permits larger loading of collected material within the impactor. The microslot design has the additional advantage that the impactor can be operated with a smaller pump than either the MOUDI or Berner lowpressure (Berner et al., 1979) impactors, which have comparable cutpoints and sampling rates. XITrD~PT ,*IILI,vu~v
n-r
I
TA,fDApTnlJ 11*11 KxL
IV1\
nE‘ cTc\l YL.“IUI.
AhTn _I._
rnNc:TPlTr7-TnN UVI.“AI\“UIIVI.
The microslot impactor has five stages, with a single slotted orifice per stage. The impactor follows a rather traditional design, with incompressible flow and minimal pressure 1283
Susanne
1284
Hering
et uf.
drops in the upper four stages, and sonic flow through the last (fifth) stage. The flow rate of 36 lmin-‘(6 x 104m3 s-l ) is controlled by the critical flow through the final stage. Particles bypassing the final stage are collected by an after-filter. Impactor orifice dimensions and stage operating pressures are shown in Table 1. The dimensions for the impactor slots were selected on the basis of the Stokes number for 50% collection efficiency: St
=PD:oUC
(1)
50
9VW
where p is the particle density (g cmp3), D,, is the particle diameter at 50% collection efficiency, C is the Cunningham slip factor, U is the jet velocity, p is the air viscosity, and W is the slot width. Following the guidelines of Marple and Rubow (1986), a value of St50 = 0.59 was assumed in the design calculations. The jet velocity for the first four stages is calculated from the incompressible Aow relation, U = QP,/PLW, where Q is the volumetric flow rate at the inlet pressure PO, P is the pressure immediately above the stage, and L is the orifice length. For the final stage the jet velocity reaches the sonic limit and
Table 1. Dimensions
Stage no.
Orifice width W (mm)
Orifice length L (mm)
1 2 3 4 5
1.38 0.56 0.31 0.18 0.10
33.4 41.3 42.4 31.5 31.5
and operating
parameters
for microslot
Flow Re
Orifice length to width, L/W
Orifice-plate separation to Pressure* width, S/W (kPa)
1144 808 901 1019 2040
24 84 137 211 368
1.8 1.8 1.6 2.1 6.5
100.4 100.3 100.0 98.7 92.0
impactor
Orifice velocity (m s-‘) 13.0 22.7 45.6 89.9 314
Calibration stokes no.’
Cutpoint (flm)
0.76 0.64 0.52
3.5: 1.7: 0.84 0.38 0.10
*Absolute pressure above stage orifice at inlet pressure of 100.4 kPa. +Stokes number at measured 50% efficiency cutpoint. t Estimated based on Stokes number of 0.76 which was derived from Stage 3 calibration.
Stage 1 Stage 2
Stage 3
Stage 4
Stage
0.030” stainless shim welded to form orifice slit of width W.
5
Orifice Plate, Bottom View
Microdot lmpactor Fig. 1. Schematic
of microslot
impactor,
showing
enlargement
Orifice Plate, Cross-sectional View
of orifice plate.
A microslot
impactor
for organic
aerosol
sampling
1285
U = J2y/(y + 1) RTO (where y is the heat capacity ratio, R the ideal gas constant and To the absolute temperature; Biswas and Flagan, 1984; Liepmann and Roshko, 1957). The impactor design also takes into account the orifice Reynolds numbers, defined by Re = PairW U/p where Pair is the density of air. For the microslot impactor 800 < Re -C 2100 (see Table l), which is in the range recommended by Marple and Rubow for optimum impactor performance. The construction of the impactor is shown in Fig. 1. The impactor body is anodized aluminum. The orifice plates are stainless steel and are removable. Pressure taps on each stage permit the measurement of operating pressures. The orifice slots were obtained by spot-welding two pieces of 0.030 in (0.76 mm) stainless steel shim stock to the orifice plate. The two pieces are aligned by means of another piece of shim stock of the thickness of the slot width. The length of the orifice is determined by a slot 0.125 in (3.18 mm) wide, milled into the orifice plate. The width of the orifices obtained by this fabrication process equaled the design specification for the last (fifth) stage but were slightly larger than specifed for the third and fourth stages. The dimensions shown in Table 1 are the actual dimensions, taken from measurements of the impactor orifices. The orifice-to-plate spacing, S, is determined by the thickness of the “feet” mounted at the ends of the orifice slots, and it varies from 1.6 to 2.1 W except for the fifth stage, where S/W = 6.5. The impactor is operated by a 13 kg, 500 W, carbon vane pump.
CALIBRATION
Experimental
PROCEDURES
methods
The impactor was calibrated with aerosol comprised of dioctyl sebacate (DOS), a nonvolatile liquid with a specific gravity of 0.92. The experimental configuration, shown in Fig. 2, follows the approach described by Marple et al. (1991). DOS aerosol was generated
I
Microdot lmpactor
Pinch Clamp Dilution Air1 >c Sheath Air Optical Particle Counter
High-Flow Differential Mobility Analyzer
Vacuum Throttling Valve >(
I
Flow Metering Orifice
Vacuum
Fig. 2. Calibration system showing aerosol generation with atomizer followed by a high-flow differential mobility analyzer and detection for test stage efficiency with an aerosol electrometer. An optical particle counter is used to monitor the calibration aerosol.
Susanne Hering et ~1.
1286
by atomization of a solution of DOS in isopropanol and passed through a high-flow differential mobility analyzer (HF-DMA) in order to select particles of a specified electrical mobility (Stolzenburg et al., 1996). The high-flow differential mobility analyzer is larger but similar in geometry to the mobility analyzer designed by Liu and Pui (1974). Specifically, the geometric factor L,/[ln(R,/R,)] which enters into the mobility equations is larger by a factor of 6.7 (where Ld is the length of the drift tube and R2 and R, are the radii of the annular gap). As a result, it can handle larger flow rates for the same aerosol mobility. For this calibration the sheath air flow was fixed at 15 lmin- ‘, with aerosol input and output providing a mobility window of 20%. flows of 3 lmin-‘, For calibration, the aerosol concentrations immediately upstream and immediately downstream of the impactor test stage were measured with an aerosol electrometer taken from a TSI Model 3030 (St. Paul, Minnesota). The pressures above and below the test stage, and the flow rate through the impactor were monitored throughout the calibration tests. An optical particle counter (Particle Measuring Systems model ,uLPC-HS, Boulder, Colorado), coupled to a pulse amplifier and 1024-channel multichannel analyzer, was used to check the size of generated particles and to count the relative proportion of singly and doubly charged particles for particles above 0.14 pm. Measurements on a blank stage, i.e., one without an orifice, were used to confirm equal transmission of aerosol through the upstream and downstream sampling lines. Although the calibration particles carry unit charge, this is not expected to alter the cutpoint. Ejiciency
calculations
The calibration aerosol from the HF-DMA contains both singly and doubly charged particles of equal electrical mobility, and this was taken into account in the evaluation of the stage collection efficiencies. The collection efficiency for singly charged particles was calculated by the relation: y1
100
=
(
1 _
B _ (%- y$
-f)
4f
>
(2)
where A is the electrometer current measured above the test stage, B is the current measured below the test stage, y2 is the stage collection efficiency for the larger, doubly charged particles of equal electrical mobility, and f is the fraction of the electrometer current A attributable to singly charged particles. For particles above 0.14 pm the ratio of doubly to singly charged particles was measured by the optical particle counter, and f was calculated directly. For smaller particles, the relative number of multiply charged particles was calculated on the basis of the Boltzmann equilibrium charge distribution and the input size distribution, as indicated by the current as a function of HF-DMA voltage. Specifically, the fraction of the current attributed to singly charged particles at HF-DMA voltage I’ is I” - I v+2-
I v+3
f= IV =
1 _
=2,(12v
-
2G4vI4v
IV
-
3H,vIw)
- 3H3vtI3v
- 2GJw)
(3)
IV
where Iv = the total aerosol current measured at voltage I/, Iv+2 = is current attributed to doubly charged particles at voltage I/, Iv+3 = is current attributed to triply charged particles at voltage I’, Izv = the total measured current at voltage 2V, GZV = the equilibrium ratio of doubly to singly charged particles at the particle corresponding to voltage 2V, H,, = the equilibrium ratio of triply to singly charged particles at the particle corresponding to voltage 3V,
size size
A microslot
impactor
for organic
aerosol
sampling
1287
and Z3”, I,, and IsV are defined analogous to Izv, and GdV and HGV are defined analogous to Gzv and H3”. This form of the correction accounts for doubly and triply charged particles at HF-DMA voltages of I/ and 2V but ignores the contribution of multiply charged particles at voltage 4V and 6V. CALIBRATION
RESULTS
Calibration curves for Stages 3-5 are shown as a function of particle aerodynamic diameter in Fig. 3. Measured 50% efficiency cutpoints are 0.84, 0.38, 0.10 ,um. Repeated calibration runs were made with different concentrations of DOS in the atomizer solution,
90 80
20
Stage
3, Run l-48 (50%
stage
4. Run II-41
Stage
4, Run l-43 (6%
Stage
4, Run l-52 (50%
DOS)
(O.i%OOS) DOS) DOS)
Stage
5, Run Ii-49
(0.04%
DOS)
Stage
5, Run II-51
(0.01%
DOS)
Stage
5, Run II-55
(0.01%
DOS)
10 0 0.00
0.10
0.20
0.30
0.40
Particle
0.50
Aerodynamic
0.60 Diameter
0.70
0.80
0.90
1 .oo
(pm)
Fig. 3. Collection efficiency as a function of particle aerodynamic diameter shown for several runs, at differing DOS solution concentrations
for Stages 3-5. Data are in the atomizer.
100
T 90 80 i 70 i
3 g
L” s
.v
2 s
.t
50 40 X Theoretical Corr.
30 60 I
??
0.00
0.10
0.20
0.30
0.40
0.50
0.60
OPC Meas. Corr.
0.70
0.80
0.90
1.00
Aerodynamic Diameter (pm) Fig. 4. Comparison of the efficiency calculations for Stage 4 obtained by correction for multiply charged particles measured by the optical counter to that obtained by correction based on Boltzmann charge equilibrium assumptions.
Susanne
1288
Hering
ef cd
+-
0.6
12
1.4
for Stages 3 -5 as a function
of particle
0.8
1.0
--+--~“--_l
1.6
1.8
2.0
dStokesNumber
Fig. 5. Collection
efficiency
Stokes number.
and thus differing proportion of multiply charged particles in the calibration aerosol, showed consistent results. For the calibration of Stage 5, all corrections for doubly charged particles were done using the formulation of equation (3) because particle sizes were too small to be efficiently detected optically. For Stage 4, stage collection efficiencies evaluated using the theoretical estimation of doubly charged particles of equation (3) are compared to that from the direct measurement off’ from the optical particle counter. These results, shown in Fig. 4, indicate no effect on the collection efficiencies at and above the 50% cutpoint but do show some discrepancies in the shape of the low-efficiency portion of the collection curve. In Fig. 5 the collection efficiencies are shown as a function of the square root of the particle Stokes number, defined in accordance to the relation of equation (1). Although the collection efficiency curves for three stages are similar, they are not exactly the same. The critical Stokes number for collection varies from 0.52 for Stage 5 to 0.76 for Stage 3. The theoretical value for slotted impactors is 0.59 (Marple and Rubow, 1986).
CHAMBER
STUDY
COMPARISON
The microslot impactor was used in a chamber study of environmental tobacco smoke to measure polycyclic aromatic hydrocarbon distributions. It was operated in parallel with the integrated organic vapor-particle sampler of Gundel et al. (1995). The vapor-particle sampler (VPS) uses an annular denuder, a filter, and a second annular denuder operated in series to collect gas-phase, particle-phase, and volatilized particle-phase species, respectively. The impactor was operated with uncoated aluminum foil impaction surfaces, and with a Teflon-coated glass fiber after-filter. Extracts from both the denuders, the impaction substrates, and the filters were analyzed for polycyclic aromatic hydrocarbons. Particle-phase concentrations measured by both methods for pyrene, triphenylene, 1,2 benzofluorene, benzo(a)anthracene, chrysene, benzo(b)fluoranthene, benzo(k)fluoranthene and benzo(a)pyrene are shown in Table 2. The agreement between the two methods was better than 15% for these compounds.
PARTICLE
TRAP
EXPERIMENTS
In the chamber study of environmental tobacco smoke the impactor was operated with uncoated substrates because smoke aerosol is inherently a sticky substance. In general, impactor collection surfaces must be coated with grease or another adhesive in order to
A microslot
impactor
for organic
aerosol
sampling
1289
prevent “bounce”, or rebound of particles from the collection surface. But the carbon content of grease coatings poses a problem for organic analyses. Biswas and Flagan (1988) developed a particle trap impactor that does not require an adhesive coating for the collection of solid particles. It consists of an empty cavity below the impaction jet, with an opening equal to 1.9 jet diameters, and it is designed to provide pocket of still air into which the particles impact. This approach is attractive for organics analysis because of its capacity to collect large samples without contamination from an adhesive coating. In fact, one of the motivations behind the design of the microslot impactor is its greater suitability to a particle trap impaction surface than a multi-orifice impactor. Several types of slotted particle trap collection geometries were tested on Stage 4 of the microslot impactor. Trap # 1 was a simple slotted cavity with an opening equal to 2W and a depth of 16 W where W = width of the impactor orifice. Trap # 2 and Trap # 3 each had a lip with an opening equal to 2 W, an internal width of 6 W, and a depth of 8 W. In all cases the length of the collection slot was equal to the length of the orifice slot. The traps were aligned with the impactor orifice visually, using a bright light on the upstream side of the impactor orifice. A sketch of these traps and the calibration results for liquid aerosols are shown in Fig. 6. The data show that the lip was necessary to achieve high collection efficiencies, yet even with the lip the maximum collection efficiencies are about 90%. With Trap # 3 experiments were also conducted with solid particles, and it was found that the solid particles bounced. The extent of bounce could not be measured because the particles picked up a contact Table 2. Comparison
of particle
phase concentrations of PAH measured and microslot impactor PAH particle
Species
Vapor-particle
1,2-Benzofluorene Benz(a)anthracene Chrysene Benzo(b)fluoranthene Benzo(k)fluoranthene Benzo(a)pyrene
2.6 10.4 30.1 9.4 2.4 5.4
sampler
concentration (VPS)
by vapor-particle
sampler
(ng m-‘)
Microslot
impactor
2.2 11.8 29.6 9.4 2.5 6.2
Impactor/VPS
ratio
0.85 1.13 0.98 1.00 1.05 1.14
. Particle Trap #I A Particle Trap #2 Particle Trap #3
20 i
??
10
3 FlatSurface. t
oI,-+ 0.00
I
I 0.10
0.20
0.30 0.40
0.50 0.60 0.70
0.80 0.90
1.00
AerodynamicDiameter(pm)
Fig. 6. Stage 4 collection efficiency for liquid particles using particle traps of two different Open circles show results for a normal flat surface geometry.
designs.
1290
Susanne
Hermg
et ul.
charge on rebound, presumably from the clear plastic used on the back surface of the trap. (Contact charge refers to the transfer of electric charge from the impaction surface to the particle during contact.) In contrast to the work of Biswas and Flagan, recent work by Tsai and Cheng (1995) found rebound of solid particles from particle traps.
SUMMARY
An impactor employing a single. narrow slotted orifice per stage has been designed to permit the collection of sufficently large and concentrated samples suitable for chemical analysis of trace constituents. Calibration with liquid aerosol yields 50% efficiency cutpoints of 0.10,0.38 and 0.84 /lrn for the lowest three stages. The smallest cutpoint of 0.10 pm is obtained with a pressure above the stage which is equal to 92% of the inlet pressure. Measured cutpoints for the incompressible flow stages are somewhat higher than predicted, with critical Stokes numbers 10 to 30% greater than theory. The microslot impactor was successfully used in a chamber study of environmental tobacco smoke, and measured concentrations of particle-phase polycyclic aromatic hydrocarbon compounds species were comparable to those determined by a denuder-filter-denuder system. Preliminary tests with particle trap collection surfaces showed that with a lipped trap the liquid particle collection efficiency is as high as 90%; however, solid particles were found to bounce out of the trap. Aclino~~lerl~lenle,lrs This work was supported by Grant I ROl HL42490from the National Institutes of Health to the Regents of the University of California and administered by Lawrence Berkeley Laboratory. The views and opinions of the authors expressed herem do not necessarily state or reflect those of the United States Government or any agency thereof. or the Regents of the University of California, and shall not be used for advertising or product endorsement purposes.
REFERENCES Allen. D. T.. Palen. E. J.. Haimov, M. 1.. Hering, S. V. and Young, J. R. (1994) Fourier transform infrared spectroscopy of aerosol collected in a low-pressure impactor: method development and field calibration. Aerosol Sci. Tr~hd. 21, 3255242. Berner. A.. Lurzer. C. H.. Pohl. L.. Preining. 0. and Wagner. P. (1979) The size distribution of the urban aerosol in Vienna. Sci. TotcilEwir. 13, 2455261. Biswas, P., Jones, C. L. and Flagan. R. C. (1987) Distortion of srze distributions by condensation and evaporation in aerosol instruments. Aero,\o/ Sc,i. Tec~hm~l. 17, 231 -246. Biswas. P. and Flagan. R. C. (19X4) High-velocity inertial impactors. Entlir. Sci. Technol. 18, 611-616. Biswas. P. and Flagan, R. C. (198X) A particle trap impactor. J. Aero.d Sci. 19, 113-121. Gundel, L. A.. Lee, V. C.. Mahanama. K. R. R., Stevens. R. K. and Daisey, J. M. (1995) Direct determination of the phase distrtbutions of polycyclic aromatic hydrocarbons using annular denuders. Atmos. Enoiron. 29, 1719-m1733. Hering. S. V. and Friedlander, S. K. (I 982) Origins of aerosol sulfur size distributions in the Los Angeles Basin. Atmos. Elrriron. 16, 2647~ 2656. Liepmann. H. W. and Roshko. A. (1957) Elemrnts of Gus Dgnumics. Wiley, New York. Lui, B. Y. H. and Pui. D. Y. H. (1974) A submicron aerosol standard. J. Colloid Intrrfucr Sci. 47, 155171. Marple. V. A. and Rubow. K. L. (1986) Theory and design guidelines. In Cuscude lmpuctor Sampliny und Datu Amdy~~is(edited by Lodge, J. P. and Chan. T. L.), pp. 79- 102. American Industrial Hygiene Association, Marple. V. A.. Rubow. K. L. and Behm, S. M. (1991) A microoritice uniform deposit impactor (MOUDI): description. calibration and use. Acrosol SC,;. Techol. 14, 436446. Stolzenburr, M. R.. Kreisberg. N. M. and Herinr, S. V. (in press) Atmosoheric size distributions measured bv differential mobility opticalparticle srze spectrometry. ,4r&l SC;. Teclr~rol. (in press). Tsai. C.-J. and Cheng, Y.-H. (1995) Solid particle collection characteristics on impaction surfaces of different designs. Aerosol Sc~i. ‘khnol. 23, 96- 105.