RTI International

RTI International is a trade name of Research Triangle Institute. www.rti.org

An Evaluation of Robotic Optical Carbon Analysis on Teflon Filter Media

Paige Presler-Jur, Prakash Doraiswamy, Lisa C. Greene, and Oki Hammond RTI International

Neil Frank and Joann Rice US EPA / OAQPS

RTI International

Outline Background and Objectives Discussion of the Robotic Weighing System (RWS) Quality Assurance

– Comparison of the RWS OT21 to Benchtop SootScanTM

OT21 – Sensitivity of filter blank used for light attenuation calculation

Relationship of Light Attenuation and Thermal Elemental Carbon (EC) Analysis

Future Work

2

RTI International

Background

Black carbon (BC) has been related to adverse human health and environmental effects.

Predominant methods of analysis are labor intensive and/or expensive such as Thermal EC analysis of quartz filters or continuous instruments such as Aethalometers.

Development of an inexpensive, nondestructive, fast, optical carbon analysis method to estimate BC would benefit air quality research.

Ability to measure BC at a low cost on both archived and current filters would provide modelers and data end users with data for health and climate change effect research.

3

RTI International

Objectives

Evaluate the ability of the robotic weighing system to perform optical carbon analysis. – Comparison to the established benchtop optical carbon method – Evaluation of the relationship between optical carbon and

thermal EC analyses – Preliminary investigation by season and region

Verify robotic optical carbon analysis to illustrate its validity as a cost-effective means to: – Obtain a vast amount of BC data from archived filters – Maximize data collection for future samples with an efficient

determination of both net gravimetric mass and BC on a single Teflon® filter

4

RTI International

Instrumentation and Data Sets Instrumentation MTL Corporation AH-225

Precision Weighing System – Inline Magee

SootScan™ Model OT21 Transmissometer

Benchtop Magee SootScan™ Model OT21 Transmissometer

Data Sets Earlier measurements of CSN

Teflon® filters performed in 2010 (sampled in 2007)

Repetition of 2010 measurements on Benchtop and RWS OT21s in 2014

New measurements of filters sampled in 2014 on both OT21s

Corresponding quartz filters from the CSN analyzed by Thermal EC STN Method (2007) IMPROVE_A Method (2007 and

2014) Thermal/optical reflectance (TOR)

and transmittance (TOT) method

5

RTI International

MTL Precision Weighing System AH-225 a. b.

c. d.

MTL AH-225 Precision Weighing System

Figures b-d provided by MTL Corporation

6

RTI International

0.0000

0.0010

0.0020

0.0030

10/4/13 10/6/13 10/7/13

99.970

99.985

100.000

100.015

100.030

Repeatability for Continual Weighing St

anda

rd W

eigh

t (m

g)

Weight after Set of 10 Filters

Date

Plots show variation in weight for groups of 10 filter weights for a single filter weighing continuously for a. 100 mg standard weight, b. 400 mg standard weight, c. standard deviation.

Stability of standard weights indicate slight drift in standard deviation is due to filter media not instrument.

400.040

400.050

400.060

400.070

400.080a. b.

c.

Weight after Set of 10 Filters

Stan

dard

Dev

iatio

n (m

g)

7

RTI International

Magee OT21 SootScan™ Transmissometer

Analyzes filter media for BC content collected from both ambient and source applications

Measures the IR (880 nm) and UV (370 nm) transmission of sample media

Non-destructive and applicable for archived filters

OT21 Distinction Benchtop RWS Light Diffusion

Media T60 Filter Medium GP47 Permanent Backing Filter

Analysis Manual

(1.5 min/filter required labor)

Automated

Neutral Density Filters Available In Development

8

RTI International

RWS OT21 SootScan™ Transmissometer The RWS Transmissometer is

equipped with a GP47 glass filter with Teflon® coating as a permanent backing filter. – Borosilicate microfibers reinforced with

woven glass cloth and bonded with PTFE

backing filter installed in the

Magee filter holder

top view of the OT21 cavity without permanent backing filter

top view of the OT21 cavity with permanent backing filter

Images provided by MTL Corporation

9

RTI International

RWS OT21 SootScan™ Transmissometer

Image provided by MTL Corporation

Back of the RWS Transmissometer illustrating the location of the diffusing filter inside the OT21 cavity and fork entrance

10

RTI International

Benchtop OT21 Analysis Comparison

Comparison between the Benchtop OT21 measurements analyzed in 2010 and in 2014

IR ATN Unit y = 0.9843x + 0.3139

R² = 0.9936

UV ATN Unit y = 1.0201x - 1.6639

R² = 0.989

-5

15

35

55

75

95

115

135

155

175

-5 45 95 145 195

2014

Ben

chto

p AT

N U

nit

2010 Benchtop ATN Unit

UV ATN Unit IR ATN Unit

11

n = 92

RTI International

Comparison between OT21 SootScan Transmissometers

Comparison between the RWS OT21 analyzed in 2014 and Benchtop OT21 measurements analyzed in 2010 and in 2014

IR ATN Unit y = 0.9693x - 1.5692

R² = 0.9607

UV ATN Unit y = 0.959x - 3.3316

R² = 0.9668

-20

0

20

40

60

80

100

120

140

-5 15 35 55 75 95 115 135 155

RW

S AT

N U

nit

Benchtop ATN Unit

UV ATN Unit IR ATN Unit

12

n = 557 unique filters = 447

RTI International

Comparison to Thermal EC Analyses

Light absorption is approximated by the light attenuation (batt) from PM deposited on the filter surface. batt = Area (m2)/Volume (m3) * LN(I0/I)

I0 = transmission through a blank filter I = transmission through an exposed filter

The BC concentration (µg/m3) of the PM is calculated with the mass absorption efficiency (σatt[λ] (m2/g)). [BC] = batt (Mm-1)/σatt[λ] (m2/g)

A linear regression scatter plot of the EC by thermal analysis and optical carbon batt can empirically derive the σatt[λ] (m2/g). σatt[λ] (m2/g) is the slope of the least-squares regression analysis.

13

RTI International

Variation in Blank Filter Batches

150,000

200,000

250,000

Individual Filters

Nov-2009

Oct-2012

Jun-2013

Jul-2013 Nov-2013 Dec-2013

Month-Year Received at RTI

Blank filters are used to measure initial transmittance, I0. IR and UV transmission readings of blank filters are compared by receipt

date at RTI. Results indicate that transmission values can vary by lot or over time.

When possible, blank filters from the same time period should be used.

UV Transmission IR Transmission

14

Raw

IR a

nd U

V Tr

ansm

issi

on R

eadi

ngs

RTI International

Elemental Carbon by IMPROVE_A Method Compared to the RWS Light Attenuation

-5

5

15

25

35

45

55

0 2 4 6

EC by IMPROVE_A TOR and TOT was compared to the RWS light attenuation. – A.: RWS I0 was the blank filter analyzed with

the exposed filters. – B.: RWS I0 was the average of blank filters

from batch blank test.

EC by IMPROVE_A

Type

Slope σatt[λ] (m2/g)

Intercept r2 Count (n)

TOR – A 10.3 0.51 0.67 370

TOR – B 9.42 0.27 0.67 370

TOT – A 11.1 2.24 0.70 370

TOT - B 11.6 1.27 0.58 370 -5

5

15

25

35

45

55

0 1 2 3 4 5

TOR TOT

A.

B.

Teflo

n-m

embr

ane

filte

r bat

t (M

m-1

)

CSN Reported EC by IMPROVE_A (µg/m3)

15

RTI International

-10

0

10

20

30

40

0 1 2 3 4 5

-10

0

10

20

30

40

0 2 4 6 8

Elemental Carbon by IMPROVE_A Method Compared to OT21 Light Attenuation

EC by IMPROVE_A TOR and TOT was compared to the OT21 light attenuation. – TOR.: EC by IMPROVE_A TOR – TOT.: EC by IMPROVE_A TOT

OT21 – TOR or TOT

Slope σatt[λ] (m2/g)

Intercept r2 Count (n)

RWS – TOR 8.71 1.87 0.73 95

Benchtop – TOR 9.34 0.30 0.87 95

RWS – TOT 7.27 3.09 0.60 95

Benchtop – TOT 7.58 5.20 0.68 95

Benchtop OT21 RWS OT21

TOR

TOT

Teflo

n-m

embr

ane

filte

r bat

t (M

m-1

)

CSN Reported EC by IMPROVE_A (µg/m3) 16

RTI International

EC by STN Method compared to OT21 SootScan Transmissometers

RWS OA batt = 10.939*EC - 2.2685

R² = 0.5922

Benchtop OA batt = 11.457*EC + 0.1074

R² = 0.6323

-20

0

20

40

60

80

100

120

0 1 2 3 4 5 6 7

Tef

lon-

mem

bran

e fil

ter

b att

(Mm

-1)

EC by STN method was compared to light attenuation measured on the RWS and Benchtop.

batt was comparable to filters measured by IMPROVE_A method.

17

CSN Reported EC by IMPROVE_A (µg/m3)

n = 305

RTI International

Future Data Analysis

Understanding the relationship of the net mass loading on the optical carbon analysis

Comparing the batt and thermal carbon analysis as a function of – seasonal variation – EPA regions – Regional Planning Organizations

Images from epa.gov

18

RTI International

Light attenuation as a function of the Mass Concentration of the Teflon filter shows a possible plateau at higher concentrations.

Further investigation is needed to evaluate the limit of the capability of the optical absorption method at high net mass loadings.

RWS Light Attenuation as a Function of Mass Concentration

-20

0

20

40

60

80

100

120

-5 45 95 145

Teflo

n-m

embr

ane

filte

r bat

t (M

m-1

)

CSN Reported Mass Concentration (µg/m3)

19

RTI International

A focused look at the low mass concentrations illustrates increased possibility of negative light attenuation.

This further illustrates the importance of blank filter at low net mass loadings.

RWS provides the capability of looking at a large number of filters with minimal labor time to further understand this variance.

RWS Light Attenuation as a Function of Mass Concentration

-20

-10

0

10

20

30

40

0 5 10 15 20 25

Teflo

n-m

embr

ane

filte

r bat

t (M

m-1

)

CSN Reported Mass Concentration (µg/m3)

20

RTI International

-10

0

10

20

30

40

50

0 1 2 3 4 5

Teflo

n-m

embr

ane

filte

r bat

t (M

m-1

)

CSN Reported EC by IMPROVE_A (µg/m3)

Seasonal Variation of Light Attenuation as a Function of EC by IMPROVE_A Method

Seasonal variation of the Teflon filter light attenuation as a function of EC by IMPROVE_A method for filters sampled in 2014

Season Slope (σatt[λ], m2/g)

Intercept r2 Count (n)

Fall – TOR 8.61 1.14 0.60 252

December – TOR 12.0 3.41 0.91 37

Fall – TOT 10.8 0.26 0.47 252

December – TOT 13.6 1.44 0.88 37

Fall TOR December TOR Fall TOT December TOT

21

RTI International

Seasonal Variation of Light Attenuation as a Function of EC by IMPROVE_A Method

Seasonal variation of batt as a function of EC by IMPROVE_A method for filters sampled in 2007

Season Slope σatt[λ] (m2/g)

Intercept r2 Count (n)

Summer – TOR 8.91 1.97 0.73 29

Fall – TOR 9.02 2.12 0.74 43

December – TOR 8.51 1.48 0.88 10

Summer – TOT 9.95 0.90 0.65 29

Fall – TOT 6.13 4.14 0.59 43

December – TOT 12.68 0.24 0.89 10

-10

-5

0

5

10

15

20

25

30

35

40

0 2 4 6 8

Teflo

n-m

embr

ane

filte

r bat

t (M

m-1

)

CSN Reported EC by IMPROVE_A (µg/m3)

Summer TOR Fall TOR December TOR Summer TOT Fall TOT December TOT

22

RTI International

EPA Regional Variation of batt as a Function of EC by IMPROVE_A TOR

Variation between EPA regions of batt as a function of EC by IMPROVE_A method for filters sampled in 2014

EPA Regions

Slope σatt[λ] (m2/g)

Intercept r2 Count (n)

1 & 2 9.53 2.27 0.75 49

3 & 4 8.20 0.90 0.48 80

5 & 6 8.60 0.94 0.56 65

7 & 8 11.14 1.94 0.67 47

9 & 10 10.57 2.78 0.84 48

-10

-5

0

5

10

15

20

25

30

0 0.5 1 1.5 2 2.5 3

Teflo

n-m

embr

ane

filte

r bat

t (M

m-1

)

EC by IMPROVE_A TOR (µg/m3)

Region 1 & 2 Region 3 & 4 Region 5 & 6 Region 7 & 8 Region 9 & 10

23

RTI International

Regional Planning Organization Variation of batt as a Function of EC by IMPROVE_A TOR

Variation between Regional Planning Organizations (RPO) of batt as a function of EC by IMPROVE_A method for filters sampled in 2014

RPO Slope σatt[λ] (m2/g)

Intercept r2 Count (n)

MANE-VU 9.38 2.25 0.74 70

VISTAS 8.29 0.68 0.46 42

Midwest 9.05 1.62 0.64 46

CENRAP 8.96 0.53 0.47 38

WRAP 10.70 2.81 0.83 59

MANEVU VISTAS Midwest CENRAP WRAP

-10

0

10

20

30

40

50

0 1 2 3 4

Teflo

n-m

embr

ane

filte

r bat

t (M

m-1

)

EC by IMPROVE_A TOR (µg/m3)

24

RTI International

Conclusions and Next Steps Conclusions Robotic system shows excellent

agreement with the Benchtop OT21. Relationship between thermal EC

and RWS batt provides σatt comparable to RTI and EPA Benchtop OT21s and ranges observed in literature.

Ability of the RWS to measure optical carbon at a high throughput makes it a valuable, cost-effective tool for – black carbon and gravimetric analysis

as part of routine network sampling. – black carbon analysis of filters archived

as part of the CSN, FRM, and other network activities.

Next Steps Further investigation of the

relationship of optical carbon analysis to – mass concentration – seasonal variation – regional variability

Development of neutral density filters for RWS

Analysis of reference materials produced by RTI

25

RTI International

Acknowledgements

Linda Andrews Ed Rickman Wayne Winstead Eric Poitras Cindi Salmons

Project was funded by

RTI International.

MTL Corporation – Joel Englund

Magee Scientific – Tony Hansen

26

RTI International

More Information

Paige Presler-Jur Research Environmental Scientist 919.541.6813 pjur@rti.org

Prakash Doraiswamy Research Environmental Scientist 919.990.8648 pdoraiswamy@rti.org

27

RTI International

28

RTI International

RTI International is a trade name of Research Triangle Institute. www.rti.org

Extra Slides

29

RTI International

Repeatability for Continual Weighing Analysis of a single filter weighed over a three-day period

showed collection of filter weights by AH-225 to have a high level of precision.

Standard Deviation for groups of 10 filter weights for a single filter weighing continuously. Sets of 10 were divided by the weighing of standard weights. St

anda

rd D

evia

tion

(mg)

Sets of 10 Filters over Time

0.0000

0.0005

0.0010

0.0015

0.0020

0.0025

0.0030

10/4/13 10/6/13 10/7/13

30

RTI International

99.990

99.995

100.000

100.005

100.010

Repeatability for Continual Weighing St

anda

rd W

eigh

t (m

g)

Filte

r Wei

ght (

mg)

Weight after Set of 10 Filters

Date / Time

Variation in weight of a. 100 mg Standard Weight, b. 400 mg Standard Weight, c. Filter Weight for single filter repetitions.

Stability of standard weights indicate drift is due to filter media not instrument.

400.040

400.050

400.060

400.070

147.200

147.210

147.220

147.230

147.240

147.250

10/4/13 10/6/13 10/8/13

a. b.

c.

Weight after Set of 10 Filters

31

RTI International

-20

0

20

40

60

80

100

0 2 4 6 8

Teflo

n-m

embr

ane

filte

r bat

t (M

m-1

)

CSN Reported EC by STN (µg/m3)

Seasonal Variation of Light Attenuation as a function of EC by STN Method

Seasonal variation of batt as a function of EC by STN method for filters sampled in 2007

Season Slope (σatt[λ], m2/g)

Intercept r2 Count (n)

Spring 8.53 2.89 0.66 92

Summer 10.03 2.12 0.81 67

Fall 8.24 1.03 0.76 42

December 16.80 4.21 0.59 70

Spring Summer Fall December

32

RTI International

EPA Regional Variation of batt as a function of EC by IMPROVE_A TOT

Variation between EPA Regions of batt as a function of EC by IMPROVE_A method. Filters sampled in 2014.

EPA Regions

Slope (σatt[λ], m2/g)

Intercept r2 Count (n)

1 & 2 10.67 0.56 0.65 49

3 & 4 12.68 0.19 0.41 80

5 & 6 9.20 1.23 0.38 65

7 & 8 16.12 1.37 0.61 47

9 & 10 13.05 2.17 0.84 48

Region 1 & 2 Region 3 & 4 Region 5 & 6 Region 7 & 8 Region 9 & 10

-10

-5

0

5

10

15

20

25

30

0 1 2 3Teflo

n-m

embr

ane

filte

r bat

t (M

m-1

)

EC by IMPROVE_A TOT (µg/m3)

33