Part. Part. Syst. Charact.
19 (2002) 387 ± 390
Technical Note
Design and Experimental Characterization of Low-Volume
PM10/2.5/1.0 Trichotomous Sampler Inlet
Hyun T. Kim*, Ken W. Lee*, Seung J. Lee**
(Received: 14 March 2002; resubmitted: 15 May 2002; accepted: 12 July 2002)
Abstract
This paper presents the development and laboratory
evaluation of a PM10/2.5/1.0 trichotomous sampling
inlet that consists of two main parts: a previously
designed PM10 size-selective inlet part and a PM2.5/1.0
two-stage virtual impactor, which was newly fabricated
Keywords: inlet, trichotomous sampler, virtual impactor
1 Introduction
Particulate matter smaller than 10 mm in aerodynamic
diameter (PM10) is known as thoracic particles that are
capable of reaching the thoracic region of humans. PM10
is further divided into two size ranges, fine particles
(nuclei mode plus accumulation mode) and coarse
particles, based on different sources and chemical
composition [1]. Fine particles can penetrate deep into
the alveolar region of the human lungs, while coarse
particles be deposited in the upper respiratory system [2].
Although there is clearly a separation between the two
size ranges, it is still difficult to select a definite cutoff size
for an air monitoring purposes. Most scientists agreed
that the cutoff size should lie somewhere between 1.0 and
2.5 mm [3, 4]. The ideal cutoff size to separate two size
ranges would be dependent on the conditions of the air
monitoring sites. Therefore, information on PM2.5
(particulate matter smaller than 2.5 mm in aerodynamic
diameter) and PM1.0 (those smaller than 1.0 mm in
aerodynamic diameter) is required. Although there is an
abundance of concentration measurement data for
* H. T. Kim, Prof. K. W. Lee (corresponding author), Depart-
ment of Environmental Science and Engineering, Kwangju
Institute of Science and Technology, 1 Oryong-dong, Puk-gu,
Kwangju 500-712 (South-Korea). E-mail: lee@kjist.ac.kr
** S. J. Lee, Department of Environmental Health, Harvard
School of Public Health, HSPH-Landmark Center, P.O. Box
15 677, Boston, MA 02215 (USA).
¹ 2002 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
387
and attached serially to the PM10 size selective inlet
part. Particles are collected in three locations through
the trichotomous sampling inlet to provide for not only
particle concentration measurements of PM10, PM2.5
and PM1.0, but also those of PM2.5 ±10 and PM1.0 ±2.5.
PM10 and PM2.5, much less is known about PM1.0.
There have been only a few attempts to develop a PM1.0
sampling device [4 ± 6].
The purpose of the trichotomous sampling inlet descri-
bed in this paper is to provide information not only on
PM10, PM2.5, and PM1.0, but also on those in between
those sizes (i.e. PM2.5 ± 10 and PM1.0 ± 2.5). To achieve
this goal, a PM10 size-selective inlet is supposed to
remove all particles larger than 10 mm. Thus, the PM10
size-selective inlet, which was tested in our recent work
[7 ± 8], was attached serially in front of the PM2.5/1.0
two-stage virtual impactor that was newly designed and
tested. The penetrating particles less than 10 mm in size
can be subsequently classified at 2.5 and 1.0 mm by the
PM2.5/1.0 two-stage virtual impactor.
2 Evaluation and Results
Figure 1 is a schematic diagram of PM10/2.5/1.0 trichot-
omous sampling inlet. The PM2.5/1.0 two-stage virtual
impactor was designed based on the research of Loo and
Cork [9]; the ratio of minor flow to major flow ( 10 %)
was chosen as a compromise between higher losses at
lower values and higher contamination of fine particles
in the coarse particle stream at higher values; the
diameter of the acceleration nozzle is critical in
determining the particle velocity, and hence the cutoff
size, as expressed by the critical Stokes number, and a
critical ratio of collection probe to the acceleration
0934-0866/02/0506-0387 $ 17.50+.50/0
388 Part. Part. Syst. Charact. 19 (2002) 387 ± 390

Fig. 1: Schematic diagram of PM10/2.5/1.0 trichotomous sam-

pling inlet.

nozzle in the range 1.3 ± 1.4 must be used to minimize the

loss peak near the cutoff size, and the ratio of the spacing
between the acceleration nozzle and the collection probe
to the acceleration nozzle should be kept within the
range 1.2 ± 1.8.
There are two acceleration nozzles and two particle
collection probes, as shown in Figure 2. Particles larger
than 10 mm are removed by the PM10 size-selective
inlet. Particles passing through the inlet which are
smaller than 10 mm are effectively separated at 2.5 mm
by the PM2.5 virtual impactor (Stage 1). At the first
virtual impactor, particles larger than 2.5 mm, which are
actually PM2.5 ± 10, are collected at position 1 shown
in Figure 2, while particles smaller than 2.5 mm pass
towards the PM1.0 virtual impactor (Stage 2). In the
same way, PM1.0 ± 2.5 and PM1.0 are collected at
positions 2 and 3, respectively, shown in Figure 2.
The information on PM10, PM2.5 ± 10, PM2.5,
PM1.0 ± 2.5 and PM1.0 can be obtained simultaneously.
This two-stage virtual impactor operates at the same flow
rate, 25 l/min, as the PM10 size-selective inlet. The
Fig. 2: Schematic diagram of two-stage virtual impactor.
Reynolds number of the impaction nozzle can influ-
ence the separation characteristics. The study of Lee
et al. [10] showed that there was little effect on the shape
and slope of the separation characteristic curves for
values of Reynolds number from 2 000 to 20 000. Both
Reynolds numbers for the PM2.5/1.0 two-stage virtual
impactor are within this range. Detailed operating
conditions are summarized in Table 1.
The performance of the designed two-stage virtual
impactor was evaluated in the laboratory. Monodisperse
polystyrene latex (PSL, Duke Scientific Corporation)
particles ranging from 0.5 to 4.0 mm were used for the
impactor. For each tested PSL particle size generated by
an atomizer (TSI Inc., Model 9302), the measured
geometric standard deviation ranged from 1.10 to 1.16.
The mass density of PSL is 1.05 g/cm3. The particle-laden
air was drawn by a vacuum pump through a heating tube,

Table 1: Cutoff size and sharpness of tested two-stage virtual impactor.

Inflow Minor Major Flow Cutoff size (mm) Sharpnessb Re
(l/min) flow flow ratio
(l/min) (l/min) Designed Tested Difference

Stage 1 ( PM2.5 virtual impactor) 25 2.2 22.8 1 : 10.4 2.5 2.58 ‡ 3.2% 1.56 9 820
Stage 2 ( PM1.0 virtual impactor) 22.8a 2.2 20.6 1 : 9.4 1.0 0.97 3.0% 1.40 16 110
a
Same as major flow of Stage 1.
b
Calculated using sg ˆ ( D15.9/D84,1)0.5.
Part. Part. Syst. Charact. 19 (2002) 387 ± 390
a dilution chamber, and the two-stage virtual impactor at
the desired flow rate. For each particle size, five
replications of particle concentration were measured
by the Aerosizer/Diluter combination (API Inc., Model
Mach P-LD). Isokinetic sampling probes were used to
sample representative particle concentrations. After the
highest and the lowest data points had been discarded,
the remaining data were averaged to obtain the
separation efficiency. All flows were controlled by a
mass flow controller (Bronkhorst Hi-Tech Inc., EL-Flow
F-112AC) and mass flow meters (SIERA Instrument
Inc., No. 822S-M-3-OV1-PV1-V1). Short sampling tubes
of equal length and diameter were used in the experi-
ments to minimize loss and bias. Throughout the
experiments, it was found that the data were reprodu-
cible to within  5 %.
The laboratory evaluation for the PM2.5/1.0 two-stage
virtual impactor resulted in the cutoff characteristics
shown in Figure 3. The indicated particle sizes shown in
the figure are the aerodynamic particle diameter as
measured by the Aerosizer. The separation efficiency
curves were fitted based on all experimental data using
the four-parameter sigmoidal equation in SIGMAPLOT
version 5.0 software. The 50 % cutoff sizes are 2.58 and
0.97 mm and the sharpnesses are 1.56 and 1.40, respec-
tively, for the PM2.5 virtual impactor (Stage 1) and the
PM1.0 virtual impactor (Stage 2). Detailed experimental
results are also summarized in Table 1. Particle losses of
virtual impactors have a maximum near the cutoff size
[11, 12]. Particle losses were calculated using a mass
balance of particle concentrations. The average particle
loss is 22 % for particle ranging from 0.5 to 4.0 mm with
higher values in the neighborhood of the 50 % cutoff
sizes.
Fig. 3: Performance curve for two-stage virtual impactor.
389
3 Summary and Conclusions
A low-volume trichotomous PM10/2.5/1.0 sampling inlet
has been developed and evaluated in the laboratory. The
sampling inlet operates at a flow rate of 25 l/min. It was
shown experimentally that the cutoff sizes of two-stage
virtual impactor are 2.58 and 0.97 mm. The two-stage
virtual impactor can effectively separate particles into
three particle size ranges. First, particles larger than
10 mm are removed by the pre-selector device. Then, the
remaining particles smaller than 10 mm are separated at
2.5 mm and 1.0 mm in the aerodynamic diameter by the
PM2.5 virtual impactor and the PM1.0 virtual impactor,
respectively. Finally, the information on PM2.5 ± 10,
PM1.0 ± 2.5 and PM1.0 can be obtained directly. By
simple calculation, the information on PM10 and PM2.5
can also be obtained. Thus, PM10, PM2.5 and PM1.0 can
be obtained simultaneously.
This was only a laboratory study. It is therefore
recommended that field tests be conducted to validate
the experimentally determined cutoff characteristic of
the PM2.5/1.0 two-stage virtual impactor. We are
planning on measuring PM10, PM2.5 and PM1.0 using
the trichotomous sampling inlet and MOUDI (Micro
Orifice Uniform Deposit Impactor, MSP Co., Mod-
el 100) in several locations to provide confirm agreement
between two measurements.
4 Acknowledgements
This work was supported in part by the Korea Science
and Engineering Foundation (KOSEF) through the
Advanced Environmental Monitoring Research Center
at Kwangju Institute of Science and Technology.
5 References
[1] K. T. Whitby, G. M. Svendrup, California Aerosols: Their
Physical and Chemical Characteristics. Adv. Environ. Sci.
Technol. 1980, 10, 477.
[2] W. C. Hinds, Aerosol Technology. 2nd Ed. John Wiley and
Sons, New York 1999.
[3] D. A. Lundgren, R. M. Burton, Effect of Particle Size
Distribution on the Cut Point Between Fine and Coarse
Ambient Mass Fractions. Inhal. Toxicol. 1995, 7, 131 ± 148.
[4] D. A. Lundgren, D. N. Hlaing, T. A. Rich, V. A. Marple,
PM10/PM2.5/PM1 Data from a Trichotomous Sampler.
Aerosol Sci. Technol. 1996, 7, 353 ± 357.
[5] V. A. Marple, B. A. Olson, A High Volume PM10/2.5/1.0
Trichotomous Sampler, in Particulate Matter: Health and
Regulatory Issues. Proc. International Speciality Conference,
Air & Waste Management Association, Pittsburgh, PA, 1995,
VIP.49, 237.
390
[6] C. Sioutas, M. C. Chang, S. Kim, P. Koutrakis, S. Ferguson,
Design and Experimental Characterization of a PM1 and a
PM2.5 Personal Sampler. J. Aerosol Sci. 1999, 30, 693 ± 707.
[7] H. T. Kim, Y. J. Kim, K. W. Lee, New PM10 Inlet Design and
Evaluation. Aerosol Sci. Technol. 1998, 29, 350 ± 354.
[8] Y. J. Lee, H. T. Kim, K. W. Lee, Development of Monitoring
Technology for Airborne Particulate Matter. Environ. Monit.
Assess. 2001, 70, 3 ± 20.
[9] B. W. Loo, C. P. Cork, Development of High Efficiency
Virtual Impactors. Aerosol Sci. Technol. 1988, 9, 167 ± 176.
Part. Part. Syst. Charact. 19 (2002) 387 ± 390
[10] K. W. Lee, D. S. Han, J. C. Kim, Particle Collection of
Impactors in Different Gases. J. Colloid Interface Sci. 1990,
137, 183 ± 191.
[11] B. T. Chen, H. C. Yeh, Y. S. Cheng, A Novel Virtual
Impactor: Calibration and Use. J. Aerosol Sci. 1985, 16,
343 ± 354.
[12] C. Sioutas, P. Koutrakis, B. A. Olson, Development and
Evaluation of a Low Cutpoint Virtual Impactor. Aerosol Sci.
Technol. 1994, 21, 223 ± 235.