This document compares measurements of particulate matter PM2.5 mass from a SidePak AM510 nephelometer to measurements from a TEOM-FDMS 8500 at multiple sites in Italy. It finds that calibrating the SidePak using a radiative transfer model that accounts for factors like particle composition and humidity provides better agreement with the TEOM-FDMS than using the factory calibration. The calibration is most improved when done using co-located measurements of the two instruments. A sensitivity analysis shows particle density and the nephelometer calibration factor α contribute most to measurement uncertainty.

1. MANCHESTER 9 September 2011 – EAC2011
F. KARAGULIAN, Claudio A. Belis, Friedrich Lagler,
Maurizio Barbiere and, Michel Gerboles
Evaluation of the performances of SidePak AM510
nephelometer compared to the Tapered Element
Oscillating Microbalance (TEOM) method for PM2.5
mass measurement
European Commission – Joint Research Centre, Institute for Environment and
Sustainability, Air-Climate and Human Interactions (AIRCLIM) Ispra, Italy
1

2. Nephelometer vs TEOM-FDMS
MANCHESTER 9 September 2011 EAC 2011
The “Candidate”
SidePak AM510 (TSI)
2
The “Reference”
TEOM-FDMS 8500
(Thermo Scientific)
Portable nephelometer
Applications:
- Personal exposure monitoring/IH studies
- Ambient/work area monitoring
- Trending/screening
- Engineering studies
- Epidemiology health studies
- Environmental sampling

3. Sampling conditions
MANCHESTER 9 September 2011 EAC 2011
3
SidePak AM510:
- Laser beam for light scattering λ TSI ~ 670 nm
- 50% cut off at 2.5 µm
- Flow rate of 1.7 L/min for PM2.5 sampling
- Original factory calibration with the ISO standard 12103-1, A1 Arizona Test Dust
- In situ re-calibration of the flow rate with a Gilan Gilibrator-2 Air Flow Calibrator
(Sensidyne)
TEOM-FDMS 8500:
- sampling head with 50% cut off at 2.5 µm
SidePak AM510 - - - > TSI values
TEOM-FDMS- - - > TEOM values

4. Site Locations
MANCHESTER 9 September 2011 EAC 2011
4
Northern Po Valley
(Italy)
January-February 2010
Rural and kerbsite
Urban site

5. Experimental Setup
MANCHESTER 9 September 2011 EAC 2011
5
Mobile laboratory with
equipment for
meteorological data
TOPAS Gmbh LAP 321
(aerosol size spectrometer)
SidePak in raw for calibration

6. Principle of operation for a nephelometer
MANCHESTER 9 September 2011 EAC 2011
6
Light scattering from particle passing through a monochromatic light:
∞
∫ ( ) (
) ( )
R = C n f d p Pλ d p , λTSI , m d d p = αTSI + β
0
radiance
particle
number
concentration
∞
∫ ( )
cm = Cn ρ p f d p
0
particle mass concentration
of polidisperse aerosol
probability
density function
( )
π 3
dp d dp
6
2
3 m − 1 
4

light flux : Pλ (d p , m) = 4π d p  2
m +2


 1

λ
 TSI




4
π
( ) 6 d d(d )
∫
∫ f ( d ) P ( d )d(d )
3
f dp
cm = R × ρ p
2
p
p
p
p
d
p
λ
d
m: refraction index of
dp: particle size
ρd: particle density

7. Rayleigh approximation
MANCHESTER 9 September 2011 EAC 2011
7
Sampled particle
diameter < 0.5 λ TSI
limit of
Rayleigh scattering
δ(d − d p )
f(d p )
cm =
Rρ p ( λTSI )
4
 m −1 

24π 3  2
m +2


2
2

8. Implications due to relative humidity (RH)
MANCHESTER 9 September 2011 EAC 2011
8
Hygroscopic growth factor GF(RH) for a particle in humid environment:
GF = (1 - RH)-γ
γ parameterizes the light scattering and its hygroscopic dependence
...evaporative (Rev) and dry (Rdry) PM fraction measured by TEOM-FMDS:
Rev = TEOM ev /TEOM
Rdry = TEOM dry /TEOM = 1 - Rev
m=
complex
refraction index
(
)
mdry Rdry + mwet Rev GF 3 − 1
GF 3
mdry and mwet are the partioned dry
and wet refraction index, respectively

9. Rev determination
MANCHESTER 9 September 2011 EAC 2011
Rev measured by TEOM-FDMS ~ 0.25
In order to confirm the evaporative fraction Rev, PM2.5 were sampled
on quartz filters by a PM sampler
Ion chromatography was carried out to determine the
amount of NO3-, SO4-2 and NH4+ on PM2.5
~ 24.8% of the total PM2.5 mass was assigned to ammonium nitrate which
is also the main component of TEOMev (Favez..)
Agreement with TEOM-FDMS measurements
9

10. Calibration through modeling
MANCHESTER 9 September 2011 EAC 2011
10
Total mass concentration measured by TEOM-FDMS:
cm =
( α TSI i + β ) ρ p ( λTSI )
24π 3
4
(
(
)) (
)) (
)
 m ( 1 - R ) + m R ( 1 − RH ) −3γ − 1 2 + 2 ( 1 − RH ) −3γ 2
 dry
ev
wet ev

 mdry ( 1 - Rev ) + mwet Rev ( 1 − RH ) −3γ − 1 2 − ( 1 − RH ) −3γ 2

(
(
)





2
PM2.5 data from
TEOM measurements
fitting of parameters γ, α, β , mdry , mwet
Calibration of SidePak AM510 (“Modeled” TSI data)
TSI Modeled
(
(
))
))
 m ( 1 - R ) + m R GF − 1 + 2( 1 − RH )
ev
wet ev
= a0TSI  dry
2
 mdry ( 1 - Rev ) + mwet Rev GF 3 − 1 − ( 1 − RH ) −3γ

(
3
(
2
− 3γ
2

 + a1


a0 = α ( λTSI ) γ 24π 3
4
a1 = β( λTSI ) γ 24π 3
4

11. The goal
MANCHESTER 9 September 2011 EAC 2011
11
calculation of fitting parameters (γ , α , β , mdry ,mwet ) related to
humidity and site typology
calibration
Calculation of new TSImodeled values for SidePaks where
TEOM-FDMS was located at the same measuring site
Attempt to calibrate the SidePak at different sites without
necessary presence of a TEOM-FDMS

12. Example: fitting parameters for a SidePak
MANCHESTER 9 September 2011 EAC 2011
12
JRC Blg 44(3)
EMEP(4)
Varese(6)
Varese(8)
rural
rural
urban
urban
TEOM (µg/m³)
36 to 129
24 to 110
74 to 124
22 to 47
Temperature (ºC)
-2.5 to 3.9
-6.1 to 1.6
-3 to 4
-2 to 3
RH (%)
65 to 93
64 to 94
58 to 90
62 to 94
Sampling time
25 hrs
85 hrs
24 hrs
9 hrs
α
0.198
0.229
0.194
0.229
β
0
-4.9
-6.9
-4.9
m dry
0.703
0.773
0.735
0.773
m wet
1.110
1.187
1.100
1.187
γ
0.251
0.226
0.143
0.226
Fitting parameters
Calibration factors for the SidePak

13. Modeled TSI vs TEOM
MANCHESTER 9 September 2011 EAC 2011
TSI = a + bTEOM
13
regression
TSI Modeled = a + bTEOM
regression line
 after modeling, new slope b closer to 1* (regression coefficients a and b)
 enhanced coefficient of determination R2 between TSI and TEOM
*Guide for the Demonstration of Equivalence of Ambient Air in Measurement Methods
http://ec.europa.eu/environmental/air/quality/legislation/pdf/equivalence.pdf, 2009

14. Example: results from calibration of a SidePak
MANCHESTER 9 September 2011 EAC 2011
Regression with fitting
parameters calculated
with TEOM-FDMS and
SidePak at the same site:
14
JRC BLG 44(4)
Varese(8)
Rural
urban
urban
0.96±0.01
0.75±0.01
0.72±0.2
1.08±0.06
4.5±0.8
11.2±0.5
28.1±1.8
6.1±2.1
7.4/6.2/0.977
14.0/8.1/0.924
7.2/6.4/0.867
21.8/12.0/0.217
1.14±0.01
0.97±0.01
0.92±0.02
1.47±0.07
-9.9±1.0
1.5±0.6
25.2±1.9
-7.3±2.3
9.0/7.8/0.976
10.9/10.5/0.910
34.8/6.8/0.905
23.0/13.6/0.309
1.24±0.01
0.95±0.01
1.04±0.02
1.07±0.05
-18.6±1.1
-5.6±0.7
-3.5±1.8
-6.3±1.6
10.0/9.4/0.972
….TEOM-FDMS and
SidePak different sites:
Varese(6)
rural
b ± u(b)
a ± u(a)
U/RRSS/R²
EMEP (5)
21.7/11.8/0.881
7.1/6.3/0.935
11.8/8.0/0.556
1.26±0.02
0.93±0.01
1.04±0.02
1.00±0.04
-14.2±1.3
-1.2±0.8
-2.4±1.9
0.3±1.2
18.2/11.0/0.962
18.6/14.1/0.831
8.2/7.1/0.918
6.2/5.5/0.724
Better performance of the TSImodeled vs TEOM only if we use calibration factors
calculated when SidePak were located at the same site of the TEOM-FDMS

15. Improvements in R2
MANCHESTER 9 September 2011 EAC 2011
15
 larger improvements in R2 were measured when larger GF was observed
(usally associated to larger RH variations)

16. Sensitivity analysis
MANCHESTER 9 September 2011 EAC 2011
16
How the variation (uncertainty) in PM2.5 can
be attributed to variations in the inputs of this model?
Mean Relative
Standard Deviation
Relative contribution
of each parameter
to the total uncertainty
Average
|u(Xi)/Xi|
|δcm/δXi/cm.u(Xi)/Xi|
α
0.245
27.%
27%
β
-4.5
-300%
4%
mdry
0.589
49%
13%
mwet
1.281
48%
15%
γ
0.205
25%
0.3%
ρ
1.6
31%
31%
RH
82%
11%
1.9%
Rev
0.278
10%
3.1%
uc(cm)/cm
46%