Lisa Phin and Prof. Greg Evans

Southern Ontario Centre for Atmospheric Aerosol Research, University of Toronto

Department of Chemical Engineering and Applied Chemistry, University of Toronto

Impactor

An impactor is used to prevent

particles larger than a certain size

from entering the sampling line

while letting the smaller particles

through. The SOCAAR impactors

use a ring of 6mm diameter

nozzles to accelerate the air

towards an impaction plate that

forces the streams into 90o bends.

Larger particles will have too much

inertia to follow the bend in the

streamlines and will hit the plate,

while smaller ones will remain in

the stream [1][2]. An impactor is

typically described by its cut-point

(the particle diameter at which

there is 50% transmission

efficiency).

The chemical and physical properties of atmospheric aerosols are of interest because of

their influence on climate and human health. The Southern Ontario Centre for Atmospheric

Aerosol Research (SOCAAR) studies aerosols using data from a number of sites including a

ground floor laboratory on College Street in the University of Toronto’s Wallberg building.

The lab houses various instruments that collect data on properties of ambient aerosols such

as particle size, number, composition, and black (elemental) carbon content. These

instruments are connected to a sampling system that continuously brings in air from outside

(Figure 1).

• At 4.2µm, the cut-point of the impactor was lower than the predicted value of 4.85µm. However,

this is not a cause for concern as most instruments in the lab analyze particles 2.5µm in diameter

or smaller and are therefore not affected by this difference. In addition, the experimental curve has

a sharper decrease in efficiency with size, which is closer to the ideal impactor efficiency curve.

• The DustTraks are well correlated but a correction factor is required to compare data from the

two instruments. Although it was consistent for over a month of the study the correction factor had

to be changed from 0.7383 to 0.7783 for the August sampling sessions. A possible cause of the

change was routine maintenance performed before the sampling session on 02-Aug.

• The line losses in the sampling duct were higher than predicted but reasonably close considering

the limitations of the equations used to estimate transmission efficiency, including the fact that they

only take into account losses due to deposition by diffusion, inertial impaction, and gravitational

settling.

• The performance of the sampling system is satisfactory and by combining the experimentally

determined efficiency of the impactor with that of the duct one could predict, with reasonable

accuracy, the total particle losses at any probe in the sampling line compared to outside air.

The Sampling System

The two components of the system that were the focus of this study were the impactor and

the sampling duct.

Characterization of Particle Losses in an Aerosol Research Laboratory’s Ambient Air

Sampling System

In order to have accurate, reliable data it is essential that what is analyzed by the

instruments is as representative as possible of the air outside and that differences between

the two are known. Therefore the performance of the sampling system needed to be tested.

Figure 4. Experimental and predicted transmission efficiencies of the impactor

and inlet as a function of aerodynamic diameter. Error bars represent 95%

confidence intervals.

• Instrumentation: Model 3321 Aerodynamic Particle Sizer (APS, TSI Shoreview, MN) to measure number concentrations of

particles for 52 size bins between 0.5 and 20 µm (aerodynamic diameter)

• Sampling: Measurements alternated between analyzing air taken directly from outside a nearby window and air taken from

the first (closest to impactor) probe on the sampling line for 20 minutes at a time (both via the same tubing)

• Data Analysis: The transmission efficiency for each size bin was calculated by taking the average value for each 10

minutes of sampling (i.e. half of each 20 minute sampling period), and taking the ratio of the inside to outside averages for

pairs of consecutive 10 minute periods

Figure 1. View of south wall of SOCAAR’s laboratory on the ground floor of the Wallberg building. Instruments located

in lab are not shown. Air enters through sampling inlets located in each window and flows along the ducts in the

direction indicated by the red arrows.

Figure 2. Impactor a) top view b) front view with plates partially

removed from holderFigure 3. Example set-up for particle loss measurements using two TSI DustTraks (DT#2 and DT#4) on the left sampling duct from Figure 1. The sampling

probes used for sampling are labelled A, B,C,D, and E.

Figure 6. Correlation of DustTrak data from calibration

sessions on 31-May and 07-Jun.

• Instrumentation: Two model 8250 DustTraks (DT#2 and DT#4, TSI, Shoreview, MN) with 2.5µm cut-point impactor inlets to

measure PM mass concentrations

• Calibration: The two DustTraks were connected to the same probe in the sampling line on four different days throughout

the sampling period and the slope of the regression line was used as a correction factor for data from DT#2

• Sampling: The DustTraks were connected to pairs of probes on the sampling line using various combinations of the

positions labelled A,B,C,D, and E in Figure 3

• Data Analysis: The transmission efficiency for each length of the line was determined from the slope of the regression line

when values from the position further along the line were plotted versus those from the position closer to the inlet

Work completed during summer internship funded by the Centre for Global Change Science and

SOCAAR. Technical assistance and guidance provided by Cheol-Heon Jeong. Pictures originally

created by Abby Eldib.

[1] Hinds, W. C. (1999). Aerosol Technology: Properties, Behaviour, and Measurement of Airborne Particles. New York: John Wiley & Sons, Inc.

[2] Marple, V. A., Olson, B. A., & Rubow, K. L. (2001). Inertial, Gravitational, Centrifugal, and Thermal Collection Techniques. In P. A. Baron, & K.

Willeke, Aerosol Measurement - Principles, Techniques, and Applications (2nd Edition). John Wiley & Sons.

[3] Brockmann, J. E. (2001). Sampling and Transport of Aerosols. In P. A. Baron, & K. Willeke, Aerosol Measurement - Principles, Techniques, and

Applications (2nd Edition). John Wiley & Sons.

[4] Lee, K. W., & Gieseke, J. A. (1994). Deposition of particles in turbulent pipe flows. Journal of Aerosol Science , 25 (4), 699-709.

[5] Pui, D. Y., Romay-Novas, F., & Liu, B. Y. (1987). Experimental study of particle deposition in bends of circular cross section. Aerosol Science

and Technology , 7, 301-315.

Sampling Duct

Air flows from the impactor to the 0.1m diameter sampling line along the length of which

several probes are inserted for transport to individual instruments. While travelling through a

sampling line, aerosols can be affected by different processes but deposition should be the

most important here. The deposition rate is affected by factors such as the flow rate, the

size, orientation, and configuration of the tubing, and the size of the particles [3].

Deposition mechanisms include inertial impaction, Brownian diffusion, and gravitational

settling [1][4].

a) b)

Objectives

Methods – Impactor Cut-point

Date Sampling Probes

Distance (m)

Flow velocity* (m/s)

Temperature* (oC)

TransmissionEfficiency

Predicted Efficiency**

31-May B and E 2.45 0.37 30 0.965 0.974

07-Jun B and E 2.45 0.42 25 0.978 0.973

16-Jun B and D 2.25 0.46 26 0.973 0.976

21-Jul A and C 1.08 0.36 37 0.979 0.989

02-Aug A and E 2.60 0.40 31 0.959 0.972

08-Aug A and E 2.60 0.42 29 0.976 0.972

Table 1. Line loss data

0

0.2

0.4

0.6

0.8

1

1.2

0.1 1 10

Tran

mis

sio

n E

ffic

ien

cy

Aerodynamic Diameter (µm)

Experimental

Predicted

For diameters greater than 8.5µm there were

not enough non-zero values to make a proper

determination of efficiency and therefore

values for these larger diameters are not

included in the plot of efficiency versus

aerodynamic diameter (Figure 4).

A cut-point of 4.2µm was determined from the

fitted line and the maximum overall efficiency

(for particles of diameter less than 0.523µm)

was 98 +/- 12%. The predicted efficiencies for

the impactor are also shown in Figure 4 with an

estimated cut-point of 4.85µm.

Results – Line Losses

Discussion and Conclusions

Acknowledgements and References

Background/Motivation

Results – Impactor Cut-point

Methods – Line Losses

The two main objectives of this study were to determine:

1. The cut-point of the impactor

2. The particle losses along the sampling duct

* Approximate average for sampling session

**Estimated for particles 2.5µm in diameter with the same flow velocity and temperature as observed during sampling

y = 0.7383xR² = 0.9959

0

0.02

0.04

0.06

0.08

0 0.02 0.04 0.06 0.08 0.1

Du

stTr

ak #

4 (

mg

/m3)

DustTrak #2 (mg/m3)

The calibration data from the first two

sampling days are shown in Figure 6. As

the same slope of 0.7383 was obtained for

the third calibration session on 21-Jul (R2 =

0.9979) this was the correction factor used

on data from DT#2 for the first four

sampling sessions. However, the slope for

the final session on 08-Aug was 0.7783 (R2

= 0.9928) and this was the correction factor

used for the final two sampling sessions.

Data, including actual and predicted

efficiencies, for each sampling session is

summarized in Table 1.

All the data collected for each

distance was combined to

determine the overall efficiency for

that length of the sampling duct.

These values as well as

estimated efficiencies were

plotted versus distance (Figure 7).

The predicted efficiencies are for

particles 2.5µm in diameter,

assuming a flow rate of 0.46m/s

and a temperature of 25oC. The

equations for the line of predicted

values and for the experimental

trend line are also shown.

y = e-0.011x

y = e-0.014x

R² = 0.6373

0.9

0.91

0.92

0.93

0.94

0.95

0.96

0.97

0.98

0.99

1

0 1 2 3 4

Tra

ns

mis

sio

n E

ffic

ien

cy

Distance (m)

Predicted

Experimental

Figure 7. Transmission efficiency as a function of distance along

sampling line. Predicted line shows efficiency estimated for flow of

0.46m/s at 25oC.

Equations for Estimating Transmission Efficiencies

The transmission efficiency of the impactor can be estimated using equation 1, an empirical equation for calculating the

proportion of particles that will travel around a bend [1][5]. The total transmission efficiency in the duct is the product of the

efficiencies considering deposition by diffusion and inertial impaction (Eq. 2) and by gravitational settling (Eq. 3) [1][3][4]. All

equations are for turbulent flow and efficiency is defined as the ratio of the number of particles entering to the number leaving.

2

πbend angle

n

cpp

ηD

UCdρberStokes numStk

18

2

Efficiencys

dep

inerdiffD

LV-4exp, 32

41

,

*,Re

040

η

Dρ.

U

VV

gdiffdep

diffdep

1

2

,

, 6.50810

15252

-

*f

inerdep

inerdep*τ.fU

VV

Re102 8

,,

-

inerdep*diffdep*dep* VVVUV velocitydepositionV dep*dep

s

tsgrav

πUD

LV-Efficiency

4exp

τglocityettling veterminal sVts

ensityparticle dρp diametererodynamicparticle ad p

locityaverage veU

of airviscosity η

meternozzle diaDn

1522

p

cd

λ.n factor correctioCunninghamC

22

1

mdnpathmean free λ

RT

N Aeculesion of molconcentratn

numberAvogadro'sNAconstantIdeal gas R etemperaturTdiametercollision dm

η

UDρ sgumberReynolds NRe tcoefficiendiffusion D tubelength of L

21Re4

3160.actorfriction ff

ν

τU f

* ion timele relaxatess particdimensionlτ

2

μ

ρd pptimeelaxation particle rτ

18

2

Uelocityfriction vU ff 2

of airviscosity kinematic ν

ensityparticle dρp

gravityon due to acceleratig

)( 882exp (Stk).Efficiencyimp (1)

(2)

(3)