United States
Environmental Protection
Agency
Office of
Research and
Development
Environmental Sciences
Research Laboratory
Research Triangle Park, N.C. 27711
EPA-600/7-77-128
November 1977
COMPARISON OF WET
CHEMICAL AND INSTRUMENTAL
METHODS FOR MEASURING
AIRBORNE SULFATE
Final Report
Interagency
Energy-Environment
Research and Development
Program Report
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into nine series. These nine'broad cate-
gories were established to facilitate further development and application of en-
vironmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
8. "Special" Reports
9. Miscellaneous Reports
This report has been assigned to the INTERAGENCY ENERGY-ENVIRONMENT
RESEARCH AND DEVELOPMENT series. Reports in this series result from the
effort funded under the 17-agency Federal Energy/Environment Research and
Development Program. These studies relate to EPA's mission to protect the public
health and welfare from adverse effects of pollutants associated with energy sys-
tems. The goal of the Program is to assure the rapid development of domestic
energy supplies in an environmentally-compatible manner by providing the nec-
essary environmental data and control technology. Investigations include analy-
ses of the transport of energy-related pollutants and their health and ecological
effects; assessments of, and development of, control technologies for energy
systems; and integrated assessments of a wide range of energy-related environ-
mental issues.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/7-77-128
November 1977
COMPARISON OF WET CHEMICAL AND
INSTRUMENTAL METHODS FOR MEASURING
AIRBORNE SULFATE
Final Report
by
B.R. Appel, E.L. Kothny, E.M. Hoffer and J.J. Wesolowski
Air and Industrial Hygiene Laboratory
California Department of Health
2151 Berkeley Way
Berkeley, California 94704
Contract No. EPA 68-02-2273
Project Officers
Carole R. Sawicki
Atmospheric Chemistry and Physics Division
Environmental Sciences Research Laboratory
Research Triangle Park, North Carolina 27711
and
Michael E. Beard
Quality Assurance Branch
Environmental Monitoring and Support Laboratory
Research Triangle Park, North Carolina 27711
ENVIRONMENTAL SCIENCES RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NORTH CAROLINA 27711
-------
DISCLAIMER
This report has been reviewed by the Environmental Sciences Research Laboratory,
and by the Environmental Monitoring and Support Laboratory, U.S. Environmental
Protection Agency, and approved for publication. Approval does not signify
that the contents necessarily reflect the views and policies of the U.S.
Environmental Protection Agency, nor does mention of trade names or commercial
products constitute endorsement or recommendation for use.
ii
-------
ABSTRACT
The methyl-thymol blue (MTB), modified Brosset and barium chloranilate sulfate
methods were evaluated for precision, accuracy, working range, interference
effects and degree of agreement with x-ray fluorescence analysis (XRF) using
atmospheric particulate samples. The samples used were collected simul-
taneously with glass fiber, quartz fiber and Fluoropore filters, the latter
being in a dichotomous sampler. The sampling design also permitted an
evaluation of artifact sulfate formation and other filter media-specific
effects. Studies of interference effects were based upon measured concen-
trations of potential interferents extractable from the particulate matter
as well as the filter media. Interferents were evaluated singly, in pairs
and quartets seeking evidence of interactions yielding non-additivity of
effects.
The results demonstrated agreement within 16$ for determining atmospheric
sulfate concentrations by the three wet chemical procedures with all the
filter media. XRF results on the "fine" Fluoropore samples agreed within
1O/0 of those obtained by wet chemical procedures on the same samples and
were, on average and within experimental error, equivalent to results obtained
by the MTB method on 8 x 10" glass fiber high volume samples. Small differences
in results obtained with different filter media in the present study are more
consistent with the effects of analytical interferents rather than artifact
sulfate formation as the cause.
ill
-------
CONTENTS
Abstract ill
Figures vi
Tables vii
Acknowledgments i
I. Introduction 1
II. Summary and Conclusions 3
III. Evaluation of the Automated Methylthymol Blue Method 8
IV. Evaluation of the Semi-Automated Modified Brosset Method.. 21
V. Evaluation of a Manual Barium Chloranilate Method 32
VI. The Analysis of Atmospheric Samples 41
VII. Effects of Interferents 75
VIII. Evaluation of Artifact Sulfate Formation with Atmospheric
Samples 110
IX. Comparison of Sulfate Results on Different Filter Media...117
References 126
Appendices
A. Differences in EPA and AIHL Laboratory Procedures for the
Auto Technicon II MTB Sulfate Method 127
B. Protocol for Determining Working Range of the MTB Method..132
C. The Semi-Automated Modified Brosset Method for Sulfate
Analysis 134
D. Barium Chloranilate Method for Determination of Sulfates
in the Atmosphere 143
E. AIHL Procedure for the EPA-MRI Barium Chloranilate Method.153
F. Determination of Reactive Silicate 154
G. Determination of Phosphates 158
H. Determination of Sulfite 162
I. X-Ray Fluorescence Analysis of St. Louis Aerosol Collected
on Fluoropore Filters 165
-v-
-------
FIGURES
Number Page
1 The Working Curve for the MTB Method Using Linear Regression Fit for 10
6-60 yg/ml Standards (Error "bars +_ 2 a)
2 Variance of Standard Sulfate Solution Analyses vs Concentration for the 11
MTB Method
3 Variance of Atmospheric Sample Extract Sulfate Analyses vs Concentration 12
for MTB Method (Linear Regression 5-60 yg/ml fit)
k Accuracy of MTB Method vs Sulfate Concentration from a Single Atmospheric 15
Sample Extract (Linear Regression Fit 5-60 yg/ml)
5 Working Curve for the Modified Brosset Method (Trial A) 2h
6 Relative Accuracy by the Modified Brosset Method with Atmospheric Samples 28
7 The Variance of Sulfate Determinations on an Atmospheric Sample with the 30
Modified Brosset Method
8 Barium Chloranilate Calibrations with and without Ion Exchange, May 25, 197^ 33
9 Chloranilate Method Calibration 37
10 Accuracy of Barium Chloranilate vs Sulfate Concentration from a Single 39
Atmospheric Sample Extract
11 Variance of Atmospheric Sample Extract Sulfate Analyses vs Concentration kO
for the Barium Chloranilate Method
12 Sampling Sites in the St. Louis Area (1975 Sampling Sites = 12k and 106) k2
13 Analytical Scheme for Atmospheric Samples ^5
lh Comparison of MTB and Modified Brosset Results for St. Louis Samples 53
15 Comparison of Brosset and MTB Results using Linear and Third Order 5^
Regression for MTB Data
16 Comparison of XRF, MTB (3rd Order Regression) and Brosset Results 55
17 Comparison of Barium Chloranilate (BC) and MTB Sulfate Values for 56
St. Louis Samples
18 Comparison of Barium Chloranilate (BC) and MTB Sulfate Values for 57
St. Louis Samples
19 Chloride Calibration Curve 6k
20 Scatter Diagram of Zinc vs Sulfate Concentrations 70
21 Comparison of UV-Visible Scans of Atmospheric Sample Aqueous Extracts 7^
and Candidate Model Chromophores
22 Artifact Sulfate in Filter Sampling 116
vi
-------
TABLES
Number page
1 Determination of Relative Accuracy with an Atmospheric Extract — MTB Ik
Method
2 Comparison of Third Order Polynomial, Power Function and Linear Calibration 16
Curve Fits on Standard Sulfate Solutions
3 Comparison of Third Order Polynomial and Linear Calibration Curve Fits on 17
Atmospheric Sample Extracts
k Comparison of MTB and EPA Results for SRM K^SOi,. Doped Strips (yg/strip) 20
5 Sulfate Determination of SRM Filter Strips with the Brosset Method (yg/strip) 23
6 Standard Error of the Estimate (Sy.x) for Working Curves by the Modified 26
Brosset Method
7 Determination of Relative Accuracy with an Atmospheric Extract Modified 27
Brosset Method
8 Accuracy of the Barium Chloranilate (BC) Method Using EPA Audit Sulfate 35
Strips (yg/strip)
9 Determination of Relative Accuracy and Precision with an Atmospheric Extract 38
10 Samplers and Filter Media Employed 1+3
11 Analysis of Field Samples , Numbers of Determinations kk
12 Sulfate Analysis of Gelman AE 8 x 10" Hi-Vol Filter Samples (yg/m3) 1+7
13 Sulfate Analysis of 126 mm Gelman AE Glass Fiber Filter Samples (yg/m3) k8
lU Sulfate Analysis of 126 mm Pallflex 2500 QAO Quartz Filter Samples (yg/m3) 1^9
15 Sulfate Analysis of "Fine" Fluoropore Falp Filter Samples (yg/m3) 50
16 Precision and Method Comparisons with Atmospheric Samples (Relative to the 51
MTB Method)
17 Comparison of Relative Sulfate Results in the First and Second Years of 58
this Study with St. Louis Samples
18 Silicate Analyses of Aqueous Extracts from 126 mm Glass and Quartz Fiber 6l
Total Filter Samples
19 Phosphate Analyses of 126 mm Glass and Quartz Fiber Filter Samples 62
20 Halide (as Chloride) Analysis of 126 mm Glass Fiber Filter Samples 65
21 Halide (as Chloride) Analysis of 126 mm Quartz Fiber Filter Samples 66
vii
-------
22 Sulfite Analyses of 126 mm Glass Fiber Filter Samples 68
23 Turbidity of Aqueous Extracts (as yg/ml colloidal clay) of 8 x 10 Glass 72
Fiber Hi-Vol Samples from St. Louis
2k Calculated Maximum Concentrations of Potential Interferents Under 76
Conditions Simulating EPA Procedures
25 Calculated Concentrations for Single Interferent Studies 79
26 Calculated Concentrations for Studies of Interferent Pairs (yg/ml) 80
27 Calculated Concentrations for Studies of Interferent Quartets (yg/ml) 8l
28 Sulfate Concentrations for Study of Interferent Pairs and Quartets (yg/ml) 82
29 Interference Effect vith the Methyltnymol Blue Method (yg/ml Observed 83
Sulfate)
30 Interference Effect with Modified Brosset Method (yg/ml Observed Sulfate) 8^
31 Interference Effect with the Barium Chloranilate Method (yg/ml Observed 85
Sulfate)
32 Interference Effect with the MTB Method Using Interferent Pairs 88
33 Interference Effect with Modified Brosset Method Using Interferent Pairs 89
3U Interference Effect with Barium Chloranilate Method Using Interferent Pairs 90
35 Comparison of Results of Single and Paired Interferents with the MTB Method 92
36 Comparison of Results of Single and Paired Interferents with the Brosset 93
Method
37 Comparison of Results of Single and Paired Interferents with the Barium 9^
Chloranilate Method
38 Interference Effect with MTB Method Using Interferent Quartets 96
39 Interference Effect with Modified Brosset Method Using Interferent Quartets 97
UO Interference Effect with Barium Chloranilate Method Using Interferent 98
Quartets
kl Comparison of Results of Single and Interferent Quartets with the MTB 99
Method
k2 Comparison of Results of Single Interferent Quartets with the Brosset 101
Method at 10-11 yg/ml Sulfate
vi ii
-------
U3 Comparison of Results of Single and Interferent Quartets with the Barium 10
Chloranilate Method at 25 and UO ug/ml Sulfate
UU Summary of Single Interferent Results 104
1;5 Maximum Calculated Interferent Concentrations under Conditions used for 10?
MTB, Brosset and BC Analyses (yg/ml)
h6 Estimated Maximum Error for Atmospheric Samples at Mid-Range Sulfate 108
Concentrations
U7 Observed Sulfate Concentrations in 2U-hour Sampling in Columbus, Ohio 111
U8 The pH of the Filter Media used in the Two Year EPA-AIHL Study 113
k$ Correction to "be applied to MTB Sulfate Results for S02 Conversion on 115
Gelman AE 8 x 10" Hi-Vol Filters
50 Comparison of MTB Sulfate Values for St. Louis Samples on Different 120
Filter Media (ug/m3)
51 Comparison of Modified Brosset Sulfate Values for St. Louis Samples on 121
Different Filter Media (ug/m3)
52 Comparison of BC Sulfate Values for St. Louis Samples on Different Filter 122
Media (yg/m3)
53 Statistical Evaluation of Mean Differences in Sulfate Results on Different 123
Filter Media by the MTB Method
5^ Statistical Evaluation of Mean Differences in Sulfate Results on Different 12U
Filter Media by the Modified Brosset Method
55 Statistical Evaluation of Mean Differences in Sulfate Results on Different 125
Filter Media by the BC Method
ix
-------
ACKNOWLEDGMENTS
Other participants in this study included G. Buell and S. Wall, who carried
out the modified Brosset and taethylthymol blue analyses, respectively;
S. Twiss, who supervised data reduction and display; and M. Haik, who super-
vised sample handling and logistics.
In addition, we wish to acknowledge T. Dzubay of the Environmental Protection
Agency, who provided the x-ray fluorescence analyses discussed in this report,
and R. Coutant, Battelle Columbus Laboratory, for his data on artifact sulfate
formation, and J. Stikeleather, Northrup Services, Inc., for helpful dis-
cussions on the Brosset method.
Mrs. Carole Sawicki and Mr. Michael Beard served as Co-Project Officers for
this program. Their helpfulness throughout this work has been sincerely
appreciated.
-------
I. INTRODUCTION
In the first year of this two-year study1, four wet chemical sulfate
methods were evaluated and compared to each other and to total sulfur
determinations by x-ray fluorescence analysis (XRF). The wet chemical
methods studied included:
A- The Bads turbidimetric procedure
B. The methylthymol blue procedure as automated for the Technicon
Autoanalyzer II (MTB)
C. The AIHL microchemical procedure
D. A modified Brosset procedure
The specific parameters evaluated included precision, accuracy, agree-
ment between methods, and the effect of a number of potential inter-
ferents.
The second year of this study had as its objectives:
A. Evaluation of the working ranges of three wet chemical sulfate
methods, including assessment of precision and accuracy.
B. Evaluation of the influence of interferents singly, in pairs, and
quartets on determination of sulfate by the three methods.
C. Analysis of approximately 100 St. Louis air samples for sulfate by
the three methods with and without correction for the observed in-
fluence of interferents.
D. Intercomparison of wet chemical and x-ray fluorescence (XRF) total
sulfur analyses with ambient air samples.
- 1 -
-------
The wet chemical methods were the automated methyl thymol blue (MCB) pro-
cedure, a manual barium chloranilate method (EC) and a semi-automated
modified Brosset (Brosset) procedure. The latter was another version
from that evaluated in the first year of the study, differing in the
solvent and the use of semi-automated equipment.
Since the atmospheric samples were collected with side-by-side sampling
of both glass fiber and more inert filter media, an evaluation of the
extent of artifact sulfate formation has been made for these samples.
This report concentrates on the efforts of the second year. The first
year's study is summarized in EPA Report No. 600/2-76-059.
- 2 -
-------
II. SUMMARY AND CONCLUSIONS
The working ranges of the automated methyl thymol blue (MTB), semiauto-
mated, modified Brosset (Brosset) and manual barium chloranilate (BC)
methods were evaluated using as criteria constancy of variance and
accuracy with standards and atmospheric samples . With atmospheric
samples, accuracy was assessed relative to the sulfate determined for
analyses at the mid-range of each method. While data reduction employed
primarily linear regression with all methods, the evaluation of the MTB
method compared linear regression and a third order polynomial fit to the
working curve. Using linear regression, the working range of the MTB
method was established to be 7-75 tig/til. Using a third order fit, the
working range could be extended down to ca. k jUg/ml. The working range
for the Brosset method was shown to be 3-13
Following the criteria stated, the working range for the BC method would
be from 10 to Uo jug/ml since the variance increased markedly at higher
levels. However, if a variance corresponding to a coefficient of varia-
tion of 6% is acceptable, the working range can be considered 10-50
jug/ml .
Using EPA audit samples (glass fiber filter strips spiked with known
quantities of sulfate), the accuracy of the MTB, Brosset and BC methods
was, on average, 93$, lOU/o and 98$, respectively.
About one hundred atmospheric samples were collected by EPA personnel in
St. Louis with side-by-side sampling using a hi-vol (8 x 10" glass fiber),
two low- volume 126 mm filter samplers using glass fiber and quartz fiber,
- 3 -
-------
and a dichotomous sampler with 37 mm Fluoropore filters. These samples
permitted evaluation of precision, agreement between methods, comparison
with XRF for total sulfur, determination of analytical interferents
present in aqueous extracts of the atmospheric samples, and an evaluation
of filter media effects in sulfate collection.
With atmospheric samples, the precision of the three methods differed
significantly. The MTB method exhibited a coefficient of variation of
1-2$ compared to 5-9$ for the modified Brosset and EC methods. Average
results by the three wet chemical methods differed by up to 1.6% with
the degree of agreement varying with filter type. Apparent sulfate
concentrations with the glass fiber hi-vol and quartz samples agreed
within 3$> while Fluoropore and glass fiber low-vol showed greater
differences relative to the hi-vol samples. Fluoropore filters were
analyzed by XRF in addition to the three wet chemical methods. The XRF
results agreed, on average, within 10$ with those obtained by the MTB
and modified Brosset methods.
The regression equations relating results between methods are as follows:
Brosset = 0.93 [MTB]-0.61 r = 0.75 (126 mm glass fiber)
Brosset = 1.03 [MTB]-0.27 r = 0.997 (126 mm quartz fiber)
Brosset =0.97 [MTB]-O.U9 r = 0.996 (Fluoropore filters)
(3rd order regression for MTB method)
BC = 0.92 [MTB]-K).89 r = 0.999 (Hi-vol (glass))
BC = 0.95 [MTB]-1.^9 r = 0.86 (Fluoropore)
BC = 0.93 [MTB]-1.56 r = 0.99 (126 mm glass fiber)
BC = 0.91 [MTB]+0.68 r = 0.98 (126 mm quartz fiber)
-------
XBF = 0.71 [MTB]+3-37 r = 0.98 (Fluoropore)
(3rd order regression for MTB method)
XRF =0.73 [Bros set 3+3.714. r = 0.98 (Fluoropore)
The average agreement between methods, expressed as a ratio of means,
for the Fluoropore samples was as follows:
Brosset/MTB = 0.93
BC/MTB =0.87
XEF/MTB =0.91
Aqueous extracts from atmospheric samples were analyzed for a number of
potential interferents in sulfate determination including silicate,
phosphate, chloride, turbidity (as colloidal clay) and "yellowness"
(as p-benzoquinone). In addition, filter sections were analyzed for sulfite.
Based on the maximum observed concentrations, a study of interference
effects was designed including evaluations with interferents singly, in
pairs and quartets. The multiple interferents were studied seeking evidence
of possible non-additivity.
The Brosset method proved to be sensitive to the fewest interferents and
the MTB-Jaethod, to the greatest number. The MTB method was subject to
significant (i.e. > 5$) positive interference by silicate, phosphate and
colloidal clay, while colloidal clay and p-benzoquinone gave significant
positive and negative interference, respectively, with the modified Brosset
technique. The BC method was subject to positive interference by phosphate,
chloride, clay and p-benzoquinone. Studies of interferents in quartets
revealed evidence of some interaction between interferents with the MTB and
5 -
-------
BC methods; observed interferences were about half of the sum of the effects
observed singly.
Observed interferents in atmospheric samples resulted from both the filter
media and the particulate matter. Based on measured interferent concen-
trations in atmospheric samples and the interference effects study, an
estimate of the maximum likely error in sulfate concentration has been
made for each filter type and analytical method. The 126 mm glass fiber
filters appear to cause the greatest interference with a maximal error of
1.6 MS/m3 by the MTB method. The error for the glass fiber hi-vol filter
is only 0.5 MS/m3 primarily because of the larger air volumes sampled.
An estimate of artifact sulfate formation was made based on reported
concentrations for 5 of the particulate sampling days. Using these data
and the results of R. Coutant of the Battelle Laboratory (Columbus),
artifact sulfate levels of ca. 1 jug/m3 are estimated for the 2^-hour
8 x 10" glass fiber filters used in the current program. However, the
conductimetric SOg method used, is known to be subject to substantial
positive interferences making the estimate of artifact sulfate probably
too high.
Results on the different filter media depended on the analytical method
used. With the MTB method, the 126 mm glass fiber results were 10-15$
higher than those on the quartz, Fluoropore and 8 x 10" glass fiber
filters, while by the Brosset method, the 126 mm glass and quartz results
agree well but the Fluoropore results (on < 2 /um particles) were
- 6 -
-------
lower. As discussed in Section VIII, these findings are consistent with
the effects of analytical interferents rather than artifact sulfate
formation.
Based upon the present study, the Brosset technique is a rapid and re-
liable procedure for samples in the 3-13 jug/10! sulfate range. Similarly,
the MTB method proved to be a rapid and reliable method for samples in
the 7-75 MS/™! range. The barium chloranilate procedure, while providing
good accuracy with standards, exhibited relatively poor precision and
cannot be recommended for use without further modification.
The key conclusions from this study are the following:
1. The automated methylthymol blue, Brosset and barium chloranilate
procedures yield results agreeing within 16$ when applied to ambient
air samples collected on three different filter media.
2. X-ray fluorescence analysis of samples collected on Fluoropore filters
yields mean sulfur concentrations (as sulfate) within 10$ of those
obtained by the wet chemical procedures. The XRF results on Fluoropore
and those by MTB analysis on 8 x 10" glass fiber hi-volume samples were,
on average, equivalent.
3- Analytical interference with the MTB method can yield sJ,gnifjLcantly
Mghersuifate results with glass fiber compared to other filter
samples.
U. The interferences shown in the wet chemical sulfate methods point
out the need to make appropriate corrections.
- 7 -
-------
III. EVALUATION OF THE AUTOMATED METHYLTHYMOL BLUE (MTB) METHOD
A. Analytical Protocol
The MTB method was set up in the Fall of 197^, and operated follow-
ing the protocol given by Technicon for the Autoanalyzer II (AA II) as
modified by AIHL1. The technique used differs somewhat from that
described in the EMSL/RTP procedure for the Auto Technicon II. These
differences are enumerated in detail in Appendix A. The principal
differences relate to the use of a linearizer and the technique for
data reduction. The EMSL/RTP procedure for the AA II requires a
linearizer. Since such equipment was not available, AIHL analyses
used linear regression for fitting the working curve, analyzed
samples in the linear region of the working curve and compared linear
and non-linear curve fitting techniques, including a power function
and a third order polynomial fit.
B. Working Range
The definition of the "working range" as provided by EPA is the region
of constant variance. Our evaluation employed both this concept, using
standard solutions and particulate extracts, and an evaluation of
accuracy as a function of concentration. The latter is relevant be-
cause the MTB (and other) working curves are non-linear at their
extremes.
To evaluate accuracy with an atmospheric sample, a hi-vol sample
extract was analyzed after dilution to the optimal range of the MTB
- 8-
-------
method. The sulfate concentration calculated for the undiluted
solution was considered "correct". By analyzing aliquots of this
solution diluted to varying degrees, the accuracy of the method was
evaluated relative to the optimal value obtained with the method.
We henceforth refer to this as "relative accuracy". A detailed
protocol is included in Appendix B.
The relative accuracy calculated for the MTB method with sulfate con-
centrations at the extremes of the working curve is expected to depend
on the regression equation employed for fitting the working curve.
Therefore, in addition to the work conducted under the protocol in
Appendix B (which used linear regression), linear regression was com-
pared to other techniques using both standards and a hi-vol extract.
Figure 1 plots the working curve between h and 75 MS/ml- The error
bars are + 2 cr. The line shown is the best linear least squares
fit between 6 and 60 jLtg/ml. This range was the maximum that provided
minimal scatter in the standard error of the estimate (Sy.x) about
the regression line. The working curve is seen to deviate from
linearity by more than 2 a at concentrations <6/jg/ml and > 60 jug/ml.
Considering the variance obtained as a function of concentration,
Figure 2 plots the results for standard solutions. Precision is seen to
remain reasonable constant between 8 and 70 jug/ml but increases below 8.
With an atmospheric sample extract and using linear regression for the
working curve (Figure 3), the variance in observed sulfate increased
&The square root of the abscissa values are the standard deviations in
chart units, where one chart unit corresponds to ca. .8 Mg/ml sulfate.
- 9 -
-------
I
o
100
80
60
Chart
Units
40
20
0
J I
10 20 30 40 50
Concentration (jig/ml)
60
70
80
Figure 1. The working curve for the MTB method using linear regression fit for 6-60 jug/ml standards
(error bars ± 2a).
-------
0.25
0.20
3
-P
0.15
csi
0.10
CD
o
•H
&
>0.05
10
20
30 40 50
Standard SOj ()ig/ml)
60
70
80
Figure 2. Variance of standard sulfate solution analyses versus concentration for the MTB method.
-------
§
5 2.0
SH
•*->
d
o>
o
o
«H
CM
H
(D
a
•H
1.5
CO
*g 1.0
e
o
CO
0.5
• Im
10
20 30 40 50
Observed Sulfate (jig/ml)
60
70
80
figure 3. Variance of atmospheric sample extract sulfate analyses versus concentration for MTB method
(linear regression 5-60 /ng/ml fit).
-------
with sulfate concentration, but never exceeds about 1.0$ of the
observed sulfate level from U to 75 Mg/ml.
Table 1 and Figure 4 displays relative accuracy as a function of
concentration with a single atmospheric extract diluted to varying
degrees (using linear regression). The MTB method is seen to be
accurate within 5$ over the range 8.8 - 75 Mg/ml relative to the
accuracy at the mid-range concentration of the method.
The degree to which three regression equations fit the working curve
can be evaluated by comparing actual sulfate concentrations for
standards used in constructing the working curve to those obtained
from a given regression line. Table 2 compares linear regression,
a power function and a third order polynomial fit of the standards
working curve. Results are shown separately for cases in which the
k jug/ml standard was included or excluded from the regression line.
The greatest deviation from the linear regression fit is at h jug/ml.
Between 6 and 60 jLtg/ml, results are not greatly different for the
three methods but the third order polynomial is, overall, somewhat
better.
A comparison of relative accuracy between linear regression and the
third order polynomial regresssion with atmospheric sample extracts
is given in Table 3» As expected from the results with standards,
below 6 jug/ml the results by linear regression are substantially
- 13 -
-------
Table 1
DETERMINATION OF RELATIVE ACCURACY WITH AN
ATMOSPHERIC EXTRACTa - MTB METHOD
Calculated
Expected
Concentration, Observed
$ of Undiluted jug/ml*1
75
70
65
60
50
ko
30
20
10
8
.6
h
79-2
75.^
71.5
66.6
55.7
lOf.2
33.0
21.2
11.0
8.79
7.08
5.61
Undiluted Cone.
wg/mlc
105.6
107.7
110.1
111.0
m.Ud
110.5
110.2
106.1
110.0
109.8
118.0
1^0.3
Diluted Cone.
ug/mle
83-9
78.3
72.7
67.1
55.9
hk.l
33.6
22.^
11.2
8.95
6.71
k.bf
Observed Cone.
Expected Cone.
.9^5
.963
—\
•985i
.993\
• 9971
.988
.985
.9^8
.983^
.982
1.055
1.255
a. St. Louis sample 618GH
"b. Mean of three determinations.
c. Equals [(Observed)/(Concentration, % of undil.)] x (.01)
d. This value taken as the correct undiluted concentration.
e. Equals (correct undil cone.) x (cone., % of undil.) x (.01).
-------
1.30,-
1.20
Observed
Expected
1.10
sulfate
l.<
0.90
0.80
10
20. 30 40 50
Observed Sulfate (pg/ml)
60
70
80
Figure 4. Accuracy of MTB method versus sulfate concentration from a single atmospheric sample extract
(linear regression fit 5-60 M9/ml).
-------
Table z
COMPARISON OF THIRD ORDER POLYNOMIAL, POWER FUNCT-L* " AND
LINEAR CALIBRATION CURVE FITS ON STANDARD SULFATE SOLJTIONS
MTB METHOD
Actual
Ug/ml
Found, jug/nx
y = a + bx
L x
A %&
Found, ug/i
y = ax
Fitting if - 60
if
6
8
10
20
30
ifO
50
60
if .66
6.10
7.66
9-53
19.6
30.0
ifO.if
50.5
59.5
16.6
1.7
- if.2
- if.7
- 1.8
- .Oif
1.01
l.Olf
- .85
if. 19
5-89
7-67
9-75
20. if
30.7
If0.7
50.2
58.5
Fitting 6-60
if
6
8
10
20
30
1(0
50
60
n^ A $,
if. 87
6.30
7.86
9.71
19.8
30.1
IfO.if
50.5
59-^
. = i no v ^S/ml
21.8
5.1
- 1.8
- 2.9
- 1.2
.20
1.05
•96
- .99
(found) -
If. 31
6.03
7.82
9.91
20.5
30.7
If0.6
lf9-9
ol Found, L
A % y = a +
jug/ml standards
if. 8
- 1.8
- if.l
- 2.5
2.0
2.3
1.8
.1(2
- 2.6
jug/ml standards
7.8
.50
- 2.2
- -93
2.6
2.1f
l.lf
- .2lf
58.0 - 3. if
jug/ml (actual)
ig/ml
bx + ex2 + dx3
if .28
5-98
7-76
9.82
20.2
30.1
39.9
50.0
60.0
l|.55
6.17
7.88
9.88
20.1
30.0
39-9
50.0
60.0
A J
7.0
- .37
- 3.0
- 1.8
• 91
.20
- .21
- .050
- .050
13.6
2.8
- 1.5
- 1.2
.52
.013
- .16
.077
- .Ollf
(actual)
- 16 -
-------
Table 3
COMPARISON OF THIRD ORDER POLYNOMIAL AND LINEAR CALIBRATION
CURVE FITS ON ATMOSPHERIC SAMPLE EXTRACTS8-
MTB METHOD
Expected
ug/ml
U.l*7
6.71
8.95
11.2
22. b
33.6
Wf.7
55-9
67.1
72.7
78.3
83-9
Found, Linear
Regression, jug/mT3 A
-------
high. Between 6 and 80 /ig/inlj linear regression is accurate within
6%, whereas the third order polynomial is somewhat 2°re accurate
over a wider range of concentrations.
The conclusions from the above are as follows:
1. The best linear regression fit (the fit which minimized Sy.x)
of the working curve for the MTB method encompasses the range
6 to 60 jug/ml.
2. The range of constant variance for the MTB method using standard
solutions has a lower limit betwen 6-8 nS/nH and an upper
limit between JO - 75 jug/ml. The variance observed for atmos-
pheric extracts was considered constant throughout the range
investigated, 5-80 jug/ml.
3. The concentration range of acceptable accuracy for the MTB method
depends to some degree upon the regression technique used to fit
the working curve. Based on atmospheric extracts as well as
standard solutions, the working range employing linear regression
(5-6 to 60 jug/ml fit) can be considered to be 7 - 75 Mg/ml if
< 5«5$ error is acceptable. The lower end of the working range
can be extended down to approximately ^ /ug/ml with < J% error by
using a third order polynomial fit.
C. Accuracy Using EPA Audit Strips
EPA audit sulfate samples prepared by Columbia Scientific, Inc. (CSl)
as glass fiber filter strips spiked with potassium sulfate were provided
by the EPA/RTP laboratory. These were extracted in distilled H20 (final
- 18 -
-------
volume 50 ml) by refluxing for 90 minutes. Based on the reported
range for the samples, 1000-6000 jug/strip, a concentration range from
20 to 120 Mg/ml was expected, except for "blanks. Accordingly, all
extracts were diluted by a factor of two before analysis. Table k
compares the MTB results to those obtained at Research Triangle Park.
using the .BaClgturbidimetric method (which were reportedly in ex-
cellent agreement with CSI's theoretical values). The MDB results
averaged 7% low.
- 19
-------
Table
COMPARISON OF MTB AMD EPA RESULTS FOR
SRM KaSOu DOPED STRIPS (yg/strip)
Sample MTBa EPA (BaClg Turbidimetric)*1 MTB/EPA
3000 series 2258 + 5 2^00 +_• 8l 0.9^1
1*000 series 5757 +_12k 61*00 + 128 0.900
5000 series (blank) 205 ±2 30 to 110
6000 series 1*55 +_ 5 U80 +_ 80 0.9^8
9000 series 1381* + 11 1500 +_ 85 0.923
Mean Ratio MTB/EPA (turbidimetric) = 0.93°
Lfean of k strips +^ standard deviation of the mean.
J. Puzak, private communication, 1976. Excepting the blanks, the values
cited vere reported to be in excellent agreement with theoretical values
provided by Columbia Scientific.
£
Excludes blanks.
- 20 -
-------
IV. EVALUATION OF THE SEMI-AUTOMATED MODIFIED BROSSET METHOD
A. Analytical Protocol
In the first year of this study, a manual, modified Brosset method
which used the solvents acetone or dioxane was evaluated. For the
current program, equipment similar to that in Brosset's laboratory
was acquired. The solvent was changed to isopropyl alcohol based
on favorable results observed with this solvent in studies performed
at REP. A detailed protocol for the method is included as Appendix C.
B. Preliminary Evaluation and Accuracy Using EPA Audit Strips
The method requires removal of cationic interferents by ion exchange.
Initial trials were made using Rexyn 101 (H) ion exchange resin.
The precision of the method proved to be excellent (better than 1$ C.V.)
with standard solutions and the working curve reproducible. However,
initial comparisons with the MTB method using extracts from atmos-
pheric samples as well as extracts from the K2S04 doped strips demon-
strated results which were 15-20$ higher. The problem was traced to
the ion exchange resin which liberates a material analyzed as sulfate
in spite of extensive prewashings. For example, 50 cc of resin mixed
with 50 cc water released what corresponded to 150 yg sulfate after
prolonged soaking.
Standard sulfate solutions showed < 1% change in concentration by
passing through resin columns which had been used 6 times previously.
- 21 -
-------
While sufficient washing may eliminate or minimize +-.h. problem, on
the advice of EPA personnel, the resin was changed to AG 50-W-X8,
50 - 100 mesh and the system re-evaluated.
Using the new resin, the extracts from 5 of the K2S04 loaded strips,
previously analyzed by the MTB method, were reanalyzed. The system
was calibrated using sodium sulfate run in duplicate before and
after the measurement of these EPA audit samples. The mean re-
gression line was then used as calculated from four values obtained
for each concentration of the standard solutions. Results are
shown in Table 5 for the means of two determinations of each sample.
In general, the Brosset results agree more closely with the EPA
findings than did the MTB. Excluding the blank value, the ratio
of means differed by only ty$> with Bros set higher.
C. Working Range
The protocol followed in evaluating the working range of the modified
Brosset method is similar to that detailed for the MTB method
(Appendix B).
1. Linearity of Working Curve and Reproducibility
Three trials were run on separate days, and the three working
curves plotted separately. Figure 5 shows the working curve
obtained with the first of three trials. The curve is distinctly
non-linear above lU Mg/ml and expected result because of the
limited capacity of the reagent. It is also somewhat non-linear
- 22 -
-------
Table 5
SULFATE DETERMINATION OF SRM FILTER STRIPS
WITH THE BROSSET METHOD (yg/strip)
Sample ID Brosset EPA Value8* Brosset/EPA
3105 2U39 ± 8 2UOO £ 8l 1.02
UlU5 6280 i 66 6UOO £ 128 0.981
533T (blank) 36 +. l.U 30 to 110
6112 516 +.6 U80 +_ 80 1.08
9200 1620 + 26 1500 +85 1.08
Mean ratio Brosset/EPA (turbidimetric) =
the BaCla turbidimetric method
Excludes blank
°Duplicate analyses of single strip extract
- J83 -
-------
l.O-i
.9-
.8-
.7-
.6-
.3-
.4-
.3-
.2-
.1-
i I I i i I i i i I
123456789 10
12 14
Sulfate (pg/nl)
16
i
18
20
Figure 5. Working curve for the modified Brosset method (Trial A).
- 24 -
-------
below 1 j^g/ml. Ihe goodness of fit to a straight line is tested
for the three trials in Table 6. As suggested both by this table
and by Figure 5, the value of Sy.x increased markedly above lU
/ng/ml. The three working curves determined on separate days
and in differing sequences were not significantly different, by
analysis of covariance, implying reasonable precision with standards.
2. Precision and Accuracy with Atmospheric Samples
Extract from three hi-vol samples, one each from Durham, NC.,.
St. Louis, MO, and Pasadena, CA, were analyzed. Each extract
was diluted to provide 1^ concentrations between approximately
0.5 and 20 jug/ml. Relative accuracy was established from one
trial with each of these three extracts. Precision as a function
of concentration employed three determinations (done on separate
days) with the St. Louis sample extract.
Relative accuracy was defined in paragraph III, B. The expected
concentration was based upon analyses done with solutions in the
6-9 jug/ml range. Table 7 shows, for one of the hi-vol sample extracts.
the observed concentrations and the concentration of the undiluted
solution calculated from each analysis. In this case, the value
calculated from the 6.7 and 9.1 jug/ml solutions were averaged, and
the resulting 22.56 MS/™1 was accepted as the undiluted extract
concentration. Expected values were then calculated from this
and the percent dilution. Results are shown for the extracts from
the three atmospheric samples in Figure 6. With the Durham and
- 25 -
-------
TABLE 6
STANDARD ERROR OF THE ESTIMATE (Sy.x) FOR
WORKING CURVES BY THE MODIFIED BROSSET METHOD
Range of Standards (yg/ml)
0-8
0-9
0-10
0-12
0-14
A
.00302
.00336
.00330
. 00455
.00484
Triald
b
B
.00516
.00519
. 00518
.00571
-,C
C
.0106
.0105
.0103
.0103
.0102
0-16 .0104 .0118 .0131
0-18 .0163 .0183 .0203
0-20 .0234 .0261 .0289
a. Standards run in succession, 0, 1, 2, 3 20 yg/ml
b. Standards run in succession 20, 19, 18 0 yg/ml
c. Standards run in random sequence
d. By analysis of covariance the curves from trials A, B and C are not
significantly different at the 95% level (F * 0.16).
-26-
-------
TABLE 7
DETERMINATION OF RELATIVE ACCURACY WITH AN ATMOSPHERIC EXTRACT^'
MODIFIED BROSSET METHOD
Calculated Expected Observed Cone.
Diluted Cone. Expected Cone.
yg/ml
Concentration,
% of Undiluted
100
80
70
60
50
40
30
25
20
15
10
7.5
5
2.5
Observed
yg/ml
13.4
13.5
13.6
13.6
11.9
9.09
6.72
5.53
4.5i
3.33
2.26
1.74
1.07
0.47
Undiluted,Conc .
yg/ml
13.4
16.9
19.4
22.7
23.8
22.7°
22. 4C
22.1
22.6
22.2
22.6
23.2
21.3
18.9
22.6 0.594
18.1 0.747
15.8 0.861
13.5 1.00
11.3 1.05
9.02 1.01
6.77 0.992
5.64 0.980
4.51 1.00
3.38 0.985
2.26 1.00
1.69 1.03_
1.13 0.946
0.56 0.84
Durham, N.C. sample N 23076.
Equals (Observed) x (Concentration, % of undil.) x (0.01).
°The mean of these values accepted as the correct undiluted concentration.
Equals (correct undil. cone.) x (cone., % of undil.) x (0.01).
-27-
-------
1.2-
Vr 1.0 -J
3
•2 .8-1
!
• •»
• .2 -
Pasadena, Ca. Sample (1974)
C 23063
G. ® © 0 ©-
5.
I I
10
Observed Sulfate (ug/ml)
15
1.2
Vrl.O
"2 .8
8
.6 -
1
St. Louis, Mo. Sample (1974)
M 23067
I I I
5 10
Observed Sulfate (ug/ml)
15
*
fr .6 .
^:;
s -21
Durham, N.C. Sample (1974)
N 23076
i i
5 10
Observed Sulfate (ug/ml)
i i i
15
Figure 6. Relative accuracy by the modified Brosset method with atmospheric samples.
- 28 -
-------
Pasadena samples, accuracy appears acceptable in the range
3-13 jug/iol. With the St. Louis sample (from the first year of
this study) -which was notable for its high level of calcium,
relative accuracy -within 5$ is observed in the range 7-llf jug/ml.
Finally, Figure 7 plots the variance from three determinations
at each concentration level for the St. Louis sample diluted to
varying degrees. Below 2.6 jug/ml, the variance increased
markedly. In the range 2.6-114- jug/ml, the variance is both small
and relatively constant excepting the sample at about 7 ng/ml.
The variance in this case corresponds to a coefficient of varia-
tion of about 3%. There is no apparent explanation for this outlier.
A principal source of the variance observed appears to be the
changes in absorbance readings with time. To minimize this effect,
the spectrophotometer was reset to read 0.800 with a reagent blank
following every third determination. Nevertheless, Just prior to
resetting, the reagent blank typically changed about O.CXA absorb-
ance units. A change of this magnitude corresponds to about 0.2
jug/ml. The data point at 7 fig/ml, had the following absorbance
values and corresponding concentrations in three trials:
Absorbance ug/ml S04~
.616 6.90
.602 7.08
.620 6.70
- 29 -
-------
.056-
.052,
.048.
.044-
.040-
.036-
CM
b
I .032-
| .028-
| .024-
.020-
.016-
.012-
.008-
.004-
0-
St. Louis Sample (1974)
M 23067
Ci)
©
©
©
© ® ©
1 2 3 4 5 6 7 8 9 10 11 12 13 I/
Concentration in ug/nl
Figure 7. The variance of sulfate determinations on an atmospheric sample with the
modified Brosset method.
-30-
-------
These findings are consistent with the + 0.2 jug/ml value claimed
for precision by C. Brosset.
Conclusions from the above are the following:
a. The working curve of the modified Brosset method can be
treated by linear regression in the range 0-ll+ wg/ml with
acceptable goodness of fit.
b. The working curve is reproducible from day to day.
c. The relative accuracy of the modified Brosset method with
atmospheric samples was constant within 5% over the range
3-13 Wg/Ml with two of three samples. A more restricted
range for a St. Louis sample 7-1^ us/nd- suggests that the
presence of other ions can influence the relative accuracy
of the method.
d. The variance for analysis in triplicate for the St. Louis
samples remained approximately constant in the range 2.6-lU
yg/ml with one exception.
e. Based on both accuracy and precision, the working range of
the modified Brosset procedure is calculated to be 3-13 wg/ml.
f. The coefficient of variation in the 2.6-1^ ug/ml. range was,
in all cases, < 3%-
-31 -
-------
V. EVALUATION OF THE MANUAL BARIUM CHLORANIIATE (BC) MF'-TQr
A. Protocol and Preliminary Evaluation
A protocol for conducting manual BC determinations was prepared,
under EPA sponsorship, by Midwest Research Institute (MRl) and is
included as Appendix D. Initial efforts at AIHL to follow this
protocol revealed a number of shortcomings as follows:
1. Plastic containers are leached by isopropyl alcohol. Blanks and
standards made with isopropyl alcohol stored for several days in
these containers had unacceptable scatter.
2. The standard curve without ion exchange was different than that
obtained when ion exchange was included (see Figure 8), possibly
an effect caused by some organics bleeding from the resin. Con-
trol of contact time between standards and resin was very im-
portant to obtain reproducible results.
3- Filtration with washed fritted funnels gave highly scattered
results. This could be avoided by not washing the frit. Sample
carry-over was eliminated by rinsing the frit with about 7-8 ml
of the next sample solution,discarding this portion and then
collecting the remainder of the new sample. This procedure gave
"better results than using fine pore filter paper (Whatman k2)
and filtering by gravity. Chloranilate accumulated on the frit
was removed by scraping and an IPA rinse. An alternative tech-
nique involving use of a disposable glass fiber filter above the
frit proved to yield less precise results than by the scraping
procedure.
- 32 -
-------
,75 r
o
•8
o
i
I
LO
10
Without Ion Exchange
ABS « .0119 [Cone.] - .0122
T - .998
Sy-x - .0140
20 30
Concentration, jig/ml
With Ion Exchange
ABS - .0146 [Cone.] - .0115
r - .997
Sy.x - .0207
40
50
Figure 8. Barium chloranilate calibrations with and without ion exchange, May 25,1976.
-------
k. Use of glass pipets gave better precision than ut;ng plastic
repetitive pipets as called for in the MRI proceaore.
5. A potential source of error is the exchange of the cations in
the samples with the acid form of an ion exchange resin. This
will decrease the solution pH and, therefore, increase the solu-
bility of Ba-chloranilate.
Based on these observations, the MRI procedure was modified, with
the revised protocol included as Appendix E.
Following the revised protocol, the accuracy and precision of the BC
method was evaluated with the EPA audit sulfate strips. Three filter
strips at each of five sulfate levels were extracted by refluxing
for 9° minutes with final volume 100 ml. Each extract was analyzed
with a single trial with results as shown in Table 8. Comparing the
AIHL results to those by EPA indicates that, on average, the AIHL
results were 1$ lower. The coefficient of variation for the BC
determinations was < k% except for the blank strips.
B. Determination of the Working Range of the BC Method
Following a protocol similar to that used with the MTB and Brosset
methods, the working range of the method was evaluated. Again, the
criteria were constancy of variance and accuracy relative to that in
the mid-range of the method.
-------
TABLE 8
ACCURACY OF THE BARIUM CHLORANILATE (BC) METHOD USING
EPA AUDIT SULFATE STRIPS (yg/STRIP)
EPA Value by
Sample ID
2058
2059
2060
3040
3041
3042
5048
5049
5050
7035
7036
7037
8060
8061
8062
Observed
6364
6486
6265
129
58
58
2773
2602
2745
4120
4141
4184
1754
1654
1775
Mean a Modified AIHL(BC)/
± 1 o Theoretical3 BC Method EPA(BC)
6372 ± 111 6549 6585 0.973
81 ± 41 0.0 20
2707 ± 92 2670 2653 1.01
4149 ± 33 4240 4252 0.979
1728 ± 65 1737 1766 0.995
Mean Ratio AIHL/EPA - 0.99°
a. Glass fiber filter strip spiked with known quantities of K2SO,
by Columbia Scientific.
b. Extracted by refluxing for 90 minutes; final volume 100 ml.
c. Excluding blank samples.
-35-
-------
The working curve with ion exchange pretreatment vas eplicated. By
analysis of covariance, the results for three curves yere not sig-
nificantly different. The resulting pooled curve is shown in
Figure 9-
Table 9 summarizes results for the observed concentrations and the
concentrations of the undiluted solution calculated from each analy-
sis. In this case, the value calculated from the 11, 2k, and 38
jug/ml solutions were averaged and the resulting 60.6? ug/wl accepted
as the undiluted extract concentration. Expected values were then
calculated from this and the % dilution. Also shown is the variance
((T2) for the three determinations of each observed concentration.
The ratio of observed to expected concentrations are plotted against
observed concentrations for three determinations on the same par-
ticulate extract in Figure 10. Relative accuracy within + 5$ was
observed in the range 12-51 yg/ml.
Figure 11 plots the variance from three determinations at each con-
centration level for the Durham sample diluted to varying degrees.
The variance remains low and approximately constant up to 38 wg/ml;
at higher levels, the variance increases markedly. At 51 wg/ml, the
variance observed corresponds to a coefficient of variation of 6%.
Based on both the observed precision and accuracy, the working curve
covers the range from about 10 to kO wg/ml. However, if a coefficient
of variation of 6% is acceptable then a range from about 10 to 50 jug/ml
is indicated.
-36 -
-------
I
U)
1.0
0.8
S 0.6
O
W)
0.4
0.2
0
ABS = .0152 [Conc.l -.00784
r = .998
Syx = .0189
10 20 30 40
Sulfate Concentration, jug/ml
50
60
Figure 9. Chloranilate method calibration.
-------
TABLE 9
DETERMINATION OF RELATIVE ACCURACY ANg PRECISION
WITH AN ATMOSPHERIC EXTRACT
BARIUM CHLORANILATE METHOD
Calculated Expected Observed Cone.
Cone.
% of Undiluted
100
90
80
60
40
20
10
5
2.5
Observed
yg/ml
64.8
57.9
50.8
38.1
24.1
11.6
4.9
1.5
- 0.9
Undil. Cone.
yg/ml
64.8
64.4
63.5
63. 5d
60.3d
58. 2d
49.5
30.5
- 37.3
Diluted Cone.
yg/ml
60.7
54.6
48.5
36.4
24.3
12.1
6.1
3.0
1.5
Expected Cone.
1.07
1.06
1.05
1.05
.99
.96
.82
.50
- .61
c*C
14.5
17.0
8.9
1.9
2.6
3.3
0.3
1.7
0.8
Durham, NC, sample N23069.
The mean of the 3 observed concentrations.
O A
oz of the mean observed concentrations.
Tlean of these values accepted as the correct undiluted concentration.
-38-
-------
1.2
In
• vl
.8
o> .6
ea
a .4
03
•o-o .2
O 0)
> -M
1 *•* V
w 8 & °
^ 8^
-.2
-.4
-.6
-
-£X_ ® ® © ©
©
-
©
—
-
-
i i i i i • i i i
-10
10
20 40 40 50
Observed Sulfate, pg/ml
60
70
Figure 10. Accuracy of barium chloranilate versus sulfate concentration from a single atmospheric sample extract.
-------
»4
O
•H
?
c
o
o
a
0
0
0)
-M
«H
tH
CO
•a
o
^
-p- o
0 6
o
o
§
•H
>
£l\J
18
16
14
12
10
8
6
4
2
0
™* '
-
(•)
_
—
—
®
—
—
—
©
^-^ xTN
© ®
wi i i i i i i i
-10 0 10 20 30 40 50 60 70
Observed Sulfate, ug/ml
Figure 11. Variance of atmospheric sample extract sulfate analyses versus concentration for the barium
chloranilate method.
-------
VI. THE ANALYSIS OF ATMOSPHERIC SAMPLES
A. Introduction
Samples were collected by EPA personnel at two sites in the St. Louis
area, identified as numbers 106 and 12lf in Figure 12.^ The filter
media and samplers are described in Table 10. The resulting samples
permit a comparison of results between the three sulfate methods
described above as well as an evaluation of precision with air samples.
In addition, the Fluoropore samples were analyzed by x-ray fluorescence
(as described in the Appendix) by T. Dzubay of EPA/RTP, thus permitting
an additional method comparison. Selected samples were also analyzed
for potential interferents in wet chemical sulfate determination.
Table 11 summarizes the determinations performed on the field samples.
The analytical scheme for the field samples is diagrammed in Figure 13.
This section will emphasize the precision of analytical methods for
sulfate with atmospheric samples, the degree of agreement between
sulfate methods and the analysis of these samples for potential inter-
ferents in sulfate determination.
Discussion of filter media effects on collection of sulfate requires
knowledge of the effects of interferents extractable from each filter
type. Accordingly, such a discussion will be included following
presentation of results of the study of interferents with each sulfate
method.
^Samples from sites 106 and 12^ are referred to as "urban" and "rural",
respectively.
-------
10 bn
Figure 12. Sampling sites in the St. Louis area (1975 sampling sites = 124 and 106).
-1*2 -
-------
TABLE 10
SAMPLERS AND FILTER MEDIA EMPLOYED
Sampler
Hl-Vol
Manual • :. .
Dichotomous
Sampler
Size
Range, ym
—
0 - 3.5
3.5 - 20
0 - 20
Size
Code
Total
Fine
Coarse
Total
Filter
Diameter, mm
203 x 254d
37
37
126
Flow Rate
1/min
1133
13.7
14
200
Filter Medium
Glass
b
Fluoropore
Fluoropore
Glass and Quc
a. Gelman AE glass fiber.
b. Fluoropore FALP (1 ym pore size).
c. Gelman AE and tissue quartz 2500 QAO (Pallflex Products) were used in side-by-side samplers.
d. Commonly referred to as 8 x 10 inch.
e. Total filters sampled from the same inlet duct as the dichotomous sampler and is described in a
paper by R.K. Stevens and T.G. Dzubay, IEEE Transactions on Nuclear Science, Vol. NS-22, No. 2,
849 (1975).
-------
Table H
ANALYSIS OF FIELD SAMPLES, NUMBERS OF DETERMINATIONS
Samples
1
-IS*
I
Sample
Fluor opore,
fine
Fluor opore,
coarse
Total filter,
glass
Total f ilter
quartz
Hi-vol, glass
Total
BlaLks
2k
19
22
2k
19
108
P0£
0
0
22
Ik
38
Tk
soa° siQ3=
0 0
0 0
22 kk
0 k8
0 0
22 92
CO." + Br~ MB.
2k 2k
0 0
kk 66
Ik 72
0 57
82 219
Barium
Brosset Chloranilate
2k
19
66
72
0
181
2k
0
26a
38b
53
iiu
XRF
2k
19
0
0
0
U3
aDuplicate analyses made on estimated 9 samples (plus k blanks) for total of 26 determinations.
Because of insufficient sample, 7 samples were analyzed with 2 determinations. The remainder used 3
determinations.
°Analyses by T. Dzubay, EPA/RTP for total sulfur expressed as sulfate. Other elements were also measured.
-------
3-ldll. to 10.1
x~
riM Cinerafon y^
<23M 4lM) >.
CMIH flwnncn
(»•• titty
i'l" /
4laea /
Toul IllMr (|Ulf>
Toul fllttr (furti)
1/2 (lltor
\
•t«tic
•xtraetlon
til M HjO
111 of
film
txtriot
I7X In HjO
oxt
•title
•xtractlon
In ID.lHjO
cut In
Iwlf
5.1
oxtraot
Extract
In TCH
100*1
1 oxtraet
net
•71 In «JO
10ml ' 1
in.,. i... ,»';•«
./4 «!«., «""
^^^^^ Iroiftt
WJ (1 d.t.)
flron
^^**^ *>;.
IMrt
•xtract C"™"^
•1411. to 20.1 *°»l
•xtract
In TCH
ixtnotlon „. .....
HjO
«*»'t io i.e. *r MTB
•t (3 oat.)
k> « XMTB () aa^laa + 1 I
\
out In
hilf
1/2 filtor
•ctntt
tn HjO
100 •!
iXtfMt
J"MTB
^ »\liC> )
* fo;(i4,t.>
Figure 13. Analytical scheme for atmospheric samples.
-------
B. Sulfate Analysis of Atmospheric Samples
Tables 12, 13, 1*+, and 15 present sulfate values obt. led with
glass fiber hi-vol, 126 mm glass fiber, quartz, and "fine" (< 3-5
particle) Fluoropore filter samples, respectively. In addition to
these results, coarse Fluoropore samples were analyzed by the modified
Brosset method. Because of the substantial variability in the filter
blanks, the limit of detection corresponded to ca. 1.5 MS/m3 sulfate.
With these samples, sulfate was invariably below this level and,
therefore, no values are reported.
Hie degree of agreement between methods, expressed as the ratio of
means, and the analytical precision, expressed as coefficients of
variation (C.V.) are shown for each filter sample type in Table 16.
The results for the glass fiber hi-volume and quartz samples indicate
good agreement between methods while those for the 126 mm glass
fiber and Fluoropore samples differ by up to 1.6%. Such filter media
effects will be discussed in detail in Section VIII.
The coefficient of variation of the MTB method was low and relatively
constant (1.1-2.1*$) and compares to a C.V. of 3% for St. Louis glass
fiber hi-vol samples reported in the interim report. The modified
Brosset and barium chloranilate methods were less precise (C.V. = 5 to
especially with the 126 mm glass fiber filter samples.
-------
Table 12
SULFATE ANALYSIS OP GELMAN AE 8 X 10" HI-VOL FILTER SAMPLES (pg/m3)a
Sample _MTBC Barium Chloranilate
0601 GH 7.0 + .1 7.3 + .2
060l* GH 23.H +. .7 22.8 +_ .9
0605 GH 13.1 +_ .1 13.0 +_ .If
0606 GH 6.8 + .1 6.7 ± .k
06l6 GH 22.7 i. .3 21.6 + .5
0618 GH 12.6 + .1 12.6 + .5
0619 GH 6.1 + .2 6.5 + .2
1201 GH
1202 GH
1203 GH
1201* GH
1205 GH
1206 GH
1210 GH
1211 GH
1212 GH
1219 GH
3.2
3.9
2k. 6
22. U
9.7
fc.8
11.1
12.6
U.8
8.3
± -1
± -1
± A
± -6
± -1
± -1
± -1
± -1
± -2
± -1
3.5
U.I
22.6
21.6
10.2
5.3
11.1
12.9
5.2
9.1
± -3
± -3
± -8
+.1.3
± •1
± .2
± -5
± -6
± -2
± -3
aMean + la for three determinations .
first two digits indicate sampler number and the second pair of
digits, the sampling day in numerical sequence. "GH" indicates
"Glass Hi-Vol". Sampler 06 used at urban site and 12, at rural site.
cFilter blank below working range and taken to be zero.
dFilter Blank = 296 ±165 yg/filter (mean of 2) or 0.2 + .1 pg/m3.
-------
Sample ID
0201TG
0202TG
0203TG
020UTG
0205TG
0206TG
0219TG
0801TG
0802TG
0803TG
OSOltTG
0805TG
0806TG
0810TG
0811TG
0812TG
0819TG
MTBC
8.2 +
11.2 +_
23.9 ±
28.2 +_
15.9 ±
7.2 +
7.6 +
3.1 +
3.9 ±
25.8 +_
25.0 +
10.8 +_
5.1 +
13.8 +_
15.7 +
5.1 +
11.2 +
.1
.1
.1
.2
.2
.03
.Olt
.1
.1
.2
.3
.03
.1
.1
.1
.1
.1
Table 13
SULFATE ANALYSIS OF 126 mm GELMAN AE GLASS FIBER FILTL SAMPLES (yg/m3)
a A J .p j
Brosset BC
7.1 + .6 7.0 + .3
11.9 ± 1.8
21.3 +_ 1.7 19.0 +_ l.lt
25.5 i 2.it
1U.5 + .3 12.k +1.5
6.5 +_ .3 5.U ± .3
6.9 + .2
1.2 + .1
2.0 +_ .3
23.1 + 1.6 23.8 +_ 1.2
22.0 +_ 2.0 22.6 +_ .k
9.9 1 .2 7.9 ± .02
3.8 + .1
12.5 ± .It
lU.lt +_ .It 12.9 1 .5
3.7 ± .3
9.9 + .5 9.0 + 1.0
Values shown are means for three determinations, except as noted, and
the standard deviation.
"TG" indicates "glass fiber total filter." Sampler 02 used at urban site
and 08, at rural site.
°Uncorrected for filter blanks since blanks below the working range.
Corrected for filter blank: Ul9 +_ 19k yg/126 mm filter or 1.6 + .8 yg/m3
limit of detection: 1.6 yg/m3
eCorrected for filter blank: klk +_ 111 yg/filter (mean of It) or 1.6 +_ .k yg/m3
limit of detection: 0.8 yg/m3
Mean + la for two determinations.
-------
Table
SULFATE ANALYSIS OP 126 mm PALLPLEX 2500 QAO QUARTZ FILTER SAMPLES (yg/m3)
Sample ID
0101TQ
0102TQ
010 3TQ
OlOlfTQ
0105TQ
0106TQ
0116TQ
0118TQ
0119TQ
0701TQ
0702TQ
070 3TQ
070l*TQ
0705TQ
't
0706TQ
0710TQ
0711TQ
0712TQ
0719TQ
MTBC
7.6 +
9.7 ±
22.3 +.
27. U +_
lU.6 +.
6.0 +_
28.1 +_
lU.9 ±
6.5 1
2.7 +
3.7 ±
2U.9 +
23.2 +
10.1 +_
U.5 +
10. U +
12.1 +
5.1 ±
8.0 +
.3
.3
.5
.3
.k
.1
.U
.2
.1
.1
.1
.2
.2
.1
.2
.1
.1
.2
.2
Brosset
8.0 £ .2
10.6 +_ .3
23.2 +_ 1.0
27.7 1 .2
15. »!• + .1*
6.3 + .1
29.2 + .u
15.8 +_ .2
6.9 + .3
1.5 ± .3
3.1 + .3
2h.k +_ .7
22.5 +1.3
10.2 +_ .U
3.9 ± .1
11.0 +_ 1.8
11.7 ± .3
U.U + .1
8.0 + .1
BC
11.6 +_ .36
9.3 1 -6
22.2 + 1.6e
5.9 + -l
25.6 + 1.8
15-1 + .9
3.8 + .3
21.0 ± .5
10.6 + .5
U.2 1 .U
9.2 + .5
10.5 ± .5
3.9 ±
7.2 +
.8
.1
Values shown are means for three determinations except where noted, and
the standard deviation.
TQ indicates "quartz fiber total filter." Sampler 01 used at urban site
and 07, at rural site.
cUhcorrected for filter blank since blanks below working range.
Corrected for filter blank: 131 +_ 191 yg/fliter (mean of 3) or 0.5 + .7 yg/m3
Two trials
Corrected for filter blank: ikk +_ 1*5 ug/filter or 0.6 ± .2 yg/m3.
Limit of detection: 0.^ yg/m3.
-------
Table 15
SULFATE ANALYSES OF "FINE" FLUOROPORE FALP FILTER SAMPLES (pg/m3)8"
XRF S as Sulfate
U.9
9.6
18.7
2U.6
13.9
6.U
2U.O
1U.1
6.5
6.5
__ f
3.6
20.9
22.1
9.U
U.6
10.6
11.7
U.6
7.6
Sample ID
301FF
302FF
303FF
30UFF
305FF
306FF
316FF
318FF
319FF
U19FF
901FF
902FF
903FF
90UFF
905FF
906FF
910FF
911FF
912FF
919FF
MTBC
(7.8)
9.7
20.6
2U.8
1U.2
(7.9)
28.5
13.5
(7.8)
8.3
(ca. 0)
(7.2)
25.5
23.0
11.1
(7.2)
12.0
12.6
(7.6)
9.9
Brosset
3.8
8.3
20.2
26. U
13.5
U.5
30.3
11.9
U.9
U.6
< 1.6
2.6
25. u
22.8
8.8
3.5
9.5
10.9
3.U
6.9
BC6
5.1
8.2
11.9
23.9
5.6
< U.8
27. U
1U.8
5.7
< U.8
< 5.1
< U.8
25.1
18.7
5.8
< U.6
< U.8
10.2
< U.7
15.6
a. Results for single trials because of insufficient sample.
b. FF indicates fine Fluoropore filters.
c. Values shown in parentheses result from sulfate readings below 6 jug/ml
and are probably too high. Filter blanks were below 6 /vig/ml and taken
to be zero.
Mg/m3 = >/*er where vol(m3) = °- 98 x time (mi. ) x flow
d. Corrected for filter blank: 2U.9 + 23. U Mg/37 mm filter (mean of 6)
Limit of detection: hi /ug/37 mm filter (1.5 ug/m3)
e. Corrected for filter blank: 97.0 + U5.U- /ig/37 mm filter (mean of If)
Limit of detection: 91 jug/37 mm filter (U.8 jug/m3)
f . No value reported.
- 50 -
-------
Table 16
PRECISION AND METHOD COMPARISONS WITH ATMOSPHERIC SAMPLES
(Relative to the MTB Method)
MPB
Rel. Results
1.0
1.0
1.0
i.oe
Brosset
C.V. (£) Rel. Results C.V. (%)
2A N.D.b
1.1 0.89 + .02 9-3
1.8 1.0 + .01 IK?
0.936
BC"
Rel. Results
1.0 + .01
.81* + .02
.97 + .02
.8f
CV f<&\
* Vr c
k.6
6.7
6.8
c
XRF
Rel. Results
N.D.
N.D.
N.D.
.91d
Filter
Hi-vol (glass)
126 nm glass
126 nm quartz
Fluoropore (fine)
a. Values shown are ratios of means excluding values below the working ranges. Coefficients of variations
pooled for each filter type.
b. Not determined.
c. Since only single determinations were possible, precision not obtainable.
d. Results obtained with both 1 irmar and third order regression of MTB data.
-------
A more detailed comparison of methods is given in Figures lU-18 which
plot results by filter type for MTB vs. Brosset, r ^ 0.98). The poorest correlation was found with the MTB-BC
comparison for fine Fluoropore samples. This is thought to reflect
primarily the imprecision of the BC method for low level sulfate samples.
For Fluoropore samples which provided sulfate levels near the bottom of
the MTB methods working range the results by that technique varied signi-
ficantly depending on the procedure used for fitting the working curve.
Figure 15 illustrates the substantial improvement in agreement between
the MTB and Brosset techniques for these samples using a third order
polynomial fit rather than linear regression. With the other samples
the choice of curve fitting procedures had minimal impact. Comparing XRF
and wet chemical results by the Brosset and MTB methods (third order
regression), XRF results are consistently lower at high concentrations.
A general comparison of results between the first and second year's
program is complicated by the use of different methods, a different
version of the Brosset technique and different sampling techniques.
Table 17 lists the feasible comparison. Brosset results for the current
program reflect somewhat better agreement with the MTB method.
C. The Analysis of Atmospheric Samples for Potential Interferents
Previous studies, reported in the Interim Report for this contract,
have demonstrated that ions such as silicate, phosphate, halogen and
sulfite can interfere in various wet chemical sulfate methods. The
significance of these observations remains unclear until the concen-
trations of these species in real sample extracts are determined.
The present study provided for such determinations on selected
- 52 -
-------
30-.
en
O4
I26ou GLASS (pg/m3)
15 20 25
KTB (Linear Regression)
15 20 25
MTB (Linear Regression)
Brosaet - 0.93 JHTB] -0.61
r - 0.996
Sy-x - 0.75
Brosset - 1.03(MTB] -0.27
r - 0.997
Sy»x * 0.66
Figure 14. Comparison of MTB and modified Brosset results for St. Louis samples.
-------
I
in
s
g
30
25
20
15
10
BROSSET
r
Syx
10 15 20 25
MTB (LINEAR REGRESSION)
1.2l[MTB]-4.62
0.997
0.70
30
5 10 15 20 25
MTB (3rd ORDER REGRESSION)
BROSSET
r
Sy*x
0.97 [MTB]-0.49
0.996
0.77
30
Figure 15. Comparison of Brosset and MTB results using linear and 3rd order regression for MTB data.
-------
Ul
tn
30
25
13 20
w
15
co
I 10
0
10
15
MTB
20
XRF
r
Sy«x
0.710 [MTB] + 3,37
0.978
1.35
25
30
5 10 15 20
BROSSET METHOD, jug/m3
XRF = 0.730 [BROSSET] +3.74
r = 0.981
Sy«x =1.26
25
30
Figure 16. Comparison of XRF, MTB (3rd order regression), and Brosset results.
-------
\J1
ON
30
25
20
15
10
HI-VOL GLASS, ;ig/m3
30
25
20
15
10
I
FINE FLUOROPORE,
I
I
I
I
BC
r
Sy*x
5 10 15 20 25
MTB VALUES (Linear Regress ion)
0.92JMTB]+0.89
0.999
0.38
I
5 10 15 20 25 30
MTB VALUES (Linear Regression)
BC - 0.95[MTB]-1.49
r - 0.86
Sy.x - 4.19
Figure 17. Comparison of barium chloranilate (BC) and MTB sulfate values for St. Louis samples.
-------
126na FILTER (GLASS), jig/i
VJ1
-3
25
20
01
1
I
10
20
15
10
10 20 30
MTB VALUES (Linear Regression)
BC - 0.93JMTB]-1.56
r - 0.99
Syx - 1.03
126n» FILTER (QUARTZ), jig/m3
I
10 20 30
KTB VALUES (Linear Regression)
BC - 0.9l[MTBl40.68
r - 0.98
Syx - 1.45
Figure 18. Comparison of barium chloranilate (BC) and MTB sulfate values for St. Louis samples.
-------
Table 17
COMPARISON OF RELATIVE SULFATE RESULTS IN THE
FIRST AND SECOND YEARS OF THIS STUDY WITH ST. LOUIS SAMPLES
a
Sample MTB Brosset XRF
197*+
Fluoropore, "fine" 1.0 1.15 °-93
Glass fiber 1.0 0.76 —
1212
Fluoropore, "fine" 1.0 0.93 0.91
Glass fiber0 1.0 0.89
a. Data of T. Dzubay, EPA/RTP.
b. 37 mm Gelman AA glass fiber filters (Batch 8188) at 12 1pm.
c. 126 mm Gelman AE glass fiber filters at 200 1pm.
-------
interferents for samples collected at urban and rural sites in the
Gt. Louis area. Aqueous extracts were analyzed for silicate, phos-
phate, halide (primarily chloride), turbidity and "yellowness"
(e.g. absorbance at ^20 nm) and filter sections were analyzed for
sulfite.
It was recognized that in some cases the filter medium would be the
dominant^ source of the interferent. In such instances, a measure-
ment of ambient concentrations would probably not be feasible.
However, the important parameter to be obtained was the gross ex-
tract concentration of the interferent from all sources. Since such
concentrations depend on the size of the filter section extracted
and the extract volume, observed concentrations were adjusted to
correspond to an EPA extraction protocol for hi-vol filter sections
in designing the interferent study (Section VI, A).
1. Silicate Analysis*
The 126 mm glass and quartz filters, less 3 - l" discs were ex-
tracted into a final volume of 100 ml. These aqueous extracts
were analyzed for silicate using the technique detailed in
Appendix F. This procedure measures "reactive silicate" by
which is meant those soluble silicate species which react with
molybdic acid. This restricts the determination to monomeric
*It should be noted that these results relate only to the water-soluble portion
of the total silicate in the sample and filter medium. No relationship is
expected between these results and soil levels in the particulate samples.
- 59 -
-------
and/or dimeric silicates. Since silicates dep ymerize at high
pH, the results can be expected to be pH-dependent to some
degree.
Table 18 lists results, expressed as /ug/m3, for both filter
types. The results for glass fiber are all below detectable
limits because of a high blank. The samples in all cases yielded
silicate values less than those of the mean blank value. It is
concluded that the glass fiber filter itself is the source of
most of the silicate in the extract and that the contribution of
silicate from the particulate sample is relatively small.
For the tissue quartz filter samples, the blank values are low
and measurable atmospheric silicate levels were obtained in some
cases. Silicate levels from the rural site were generally below
detection.
2. Phosphate Analysis
Phosphate analyses were done by the molybdenum blue method as
detailed in Appendix G. Analyses of glass fiber filters were
done with one trial because of insufficient sample. The results
given in Table 19 indicate low but measurable atmospheric values
in all cases. Results for selected tissue quartz samples are
also given. The calculated atmospheric concentrations from
samples collected on quartz are about a factor of 10 lower than
_ 60 _
-------
Table 18
SILICATE ANALYSES OF AQUEOUS EXTRACTS FROM 126 mm GLASS
AND QUARTZ FIBER TOTAL FILTER SAMPLES8-
Glass Fiber
Quartz Fiber
Sample ID
201TG
202TG
203TG
Silicate (as Si02)
Silicate (as Si02)
205TG
206TG
219TG
801TG
802TG
803TG
8C4TG
805TG
806TG
810TG
811TG
819TG
< 1.1
< 1.1
< 1.1
< 1.1
< 1.1
< 1.1
< 1.1
< 1.1
< 1.1
< 1.1
< 1.1
< 1.1
< 1.1
< 1.1
< 1.1
< 1.1
Sample ID
101TQ
102TQ
103TQ
105TQ
106TQ
116TQ
118TQ
119TQ
701TQ
702TQ
703TQ
70UTQ
705TQ
706TQ
710TQ
7HTQ
712TQ
719TQ
5.1
14.3
U.I
1.8
1.9
2.6
< 1.2
9*8
2.0
< 1.2
< 1.2
< 1.2
< 1.2
< 1.2
< 1.2
< 1.2
< 1.2
< 1.2
Mean Blank = 8U.O + 1.3 iug/ml
or 9655 + 1^9 jug/filter
Mean Blank =5.2 + 1.5 jug/ml
or 598 + 172 Mg/filter
a. Total filter area 109.5 cm2 (88$ of 126 mm filter) extracted with
final volume 100 ml.
b. The gross /ug/ml silicate were, in a].l cases, less than the value for
the filter blank.
c. Results are calculated for the second of two trials. The first trial
was rejected because of experimental difficulties.
- 61 -
-------
TABLE 19
PHOSPHATE ANALYSES OF 126 mm GLASS AND
QUARTZ FIBER FILTER SAMPLES a
Glass Fiber
Quartz Fiber(
Sample ID P0i+- (yg/ml)d P0»*r (ug/m3)d Sample ID PO^ (ug/ml)d POt*" (U8/m3)d
201TG
202TG
203TG
204TG
206TG
219TG
801TG
802TG
803TG
804TG
805TG
806TG
810TG
811TG
812TG
819TG
Mean Blank
.35
.41
..47
.31
.33
.09
.03
.10
.20
.16
.19
.10
.17
.16
.40
.30
.17
0.15
0.17
0.20
0.13
0.14
0.04
0.01
0.04
0.09
0.07
0.08
0.04
0.07
0.07
0.17
0.12
0.07
10UTQ
105TQ
219TQ
701TQ
T03TQ
.03
.03
.00
.01
.01
.01
.01
< .005
< .005
< .OCA.
-.02 ± .01 ug/ml
Mean Blank
.01 ± .005
a
footnote a, Table 18.
Sample sufficient for only one determination.
orrected for filter blank.
of two trials.
- 62 -
-------
those obtained on glass fiber. The cause of the higher apparent
atmospheric phosphate levels from glass fiber filter samples is
unclear. The atmospheric concentrations observed on quartz
filters are considered more reliable.
Chloride Analysis
Samples were analyzed using the AgNOs turbidimetric method.2
Figure 19 illustrates the working curve with duplicate standards.
This technique responds to chloride, bromide, and iodide and is
subject to several other interferences not likely to be signifi-
cant in atmospheric samples. The intended strategy was to employ
XRF Br values from filter analyses as a measure of bromide ions
in the extract. By subtracting these from the apparent chloride
(after correction for atomic weight) and assuming negligible
iodide, the remainder would approximate the true chloride con-
centration.
Tables 20 and 21 list results for observed halide ion (as chloride)
>
on glass and quartz filters, respectively. In both cases, the
relatively high and variable filter blanks resulted in atmospheric
concentrations below detectable limits. Table 20 also lists XRF
bromine values which are 2 or 3 orders of magnitude below the de-
tectable limit for chloride by the turbidimetric method for these
samples.
- 63-
-------
Date: 7-15-76
ABS
Sy«x
50 100 150
Concentration, ;ig/ml
.002 + .003 [Cone]
.0191
200
Figure 19. Chloride calibration curve.
-------
TABLE 20
HALIDE (AS CHLORIDE) ANALYSIS OF 126 mm GLASS FIBER
FILTER SAMPLES
Sample ID
202TG
204TG
219TG
Observed Cl~, yg/mla
14.8
6.1
9.3
Cl", yg/nf
< 10.9
< 10.9
< 10.6
Br, yg/m"
.057
.042
.150
801TG
802TG
803TG
804TG
806TG
810TG
812TG
21.4
31.2
5.1
4.0
17.3
7.2
14.5
< 10.
< 11.
< 10.
< 10.
< 10.
< 11.
< 10.
5
1
3
5
6
1
3
—
.012
.008
.012
.006
.010
.016
Filter Blank: 21.8 ± 12.7 yg/ml
or 2500 ± 1450 yg/filter
a. Not corrected for filter blank.
b. By XRF of the fine particle fraction on Fluoropore filters
collected by the dichotomous sampler.
-65-
-------
TABLE 21
HALIDE (AS CHLORIDE) ANALYSIS OF 126 mm QUARTZ
FIBER FILTER SAMPLES
o
Sample ID Observed Cl", yg/ml Observed Cl . yg/m
104TQ 2.2 < 2.8
105TQ 5.4 < 2.8
119TQ 5.0 < 2.8
701TQ 5.3 < 2.7
703TQ 3.8 < 2.7
Filter Blank: 0.11 ± 3.2 yg/ml (mean of 3)
or 12.3 ± 371 yg/126 mm filter.
«n
a. Mean of two trials.
b. Not corrected for blank.
-66-
-------
For the purpose of estimating upper limits to interferent levels
for St. Louis samples, the chloride levels, in i^g/ml, will be
used uncorrected for filter blanks.
The unused halves of the one-inch discs from the fine Fluoropore
samples (20 samples and h blanks) were extracted in 10 ml H20
and analyzed for chloride. Results for samples, filter blanks
and reagent blanks were not significantly different. Therefore,
the data were not reduced.
Sulfite Analysis
Spptions cut from the exposed 126 mm Gelman AE glass fiber
filters were extracted in tetrachloromercurate solution and
analyzed by the West-Gaeke procedure. The extraction and ana-
lytical procedures are detailed in Appendix H. The quantity of
sample proved to be sufficient for only a single trial. The
results are summarized in Table 22. Also shown is the sulfite
expressed as a fraction of the water soluble sulfate for the same
glass fiber sample. The latter probably includes both sulfate
and sulfite since oxidation of S03= in solution appears to be
rapid (at least using synthetic solutions).1
The filter samples had been stored at -10°C for about 6 months
preceding analyses without protection from oxygen. Therefore,
any sulfite surviving on the filter probably exists as a metal-
sulfite complex or salt stabilizing the S (IV) from oxidation.
- 67-
-------
TABLE 22
SULFITE ANALYSES OF 126 mm GLASS FIBER FILTER SAMPLES3
b 1b =
Sample ID S0a , yg/ml S0a , yg/m SOg /SOi, , %
201TG 0.068 0.038 0.50
202TG 0.042 0.023 0.21
203TG 0.017 0.010 0.04
204TG 0.005 0.003 0.01
205TG < 0.004 < 0.002 < 0.01
206TG < 0.004 < 0.002 < 0.03
219TG 0.005 0.003 0.04
801TG 0.089 0.05 1.6
802TG < 0.004 < 0.002 < 0.05
803TG < 0.004 < 0.002 < 0.01
804TG < 0.004 < 0.002 < 0.01
805TG < 0.004 < 0.002 < 0.02
806TG < 0.004 < 0.002 < 0.04
810TS 0.004 0.002 0.01
811TG < 0.004 < 0.002 < 0.01
812TG 0.028 0.015 .29
819TG 0.013 0.007 .06
Filter Blank: - 0.002 ± .003 yg/ml (mean of 5)
or .3 ± .5 yg/126 mm filter.
Limit of Detection: .0023 yg/ml .
2
a. Analysis of 15.2 cm of filter area extracted into tetrachloromercurate
solution. Total volume: 20 ml.
b. Corrected for blank.
c. Sulfate values by the MTB method.
d. Basis twice the CT of the intercept of the working curve.
-68-
-------
Accordingly, it is reasonable to seek correlations between sul-
fite levels and metals which might serve to stabilize sulfite.
It has been suggested3 that zinc may be involved in fixation of
S02 with the result being zinc sulfite. Accordingly, sulfite
levels were examined for possible correlations with XEF Zn
values determined by T. Dzubay. Figure 20 shows a scatter dia-
gram of Zn and S03= for the cases where measurable data are
available for the same location and date. No correlation is
evident.
5. Turbidity
Since colloidal particles can remain in suspension in spite of
filtration of aqueous extracts, the resulting light scattering
can interfere in some sulfate determinations, especially the
BaClg turbidimetric procedure.1 Other, more subtle, phenomena
(e.g. nucleation of BaS04 precipitation) may also lead to inter-
ference in sulfate measurements.
Although not a part of the initial study design, a limited effort
was undertaken to estimate the maximum turbidity to be expected
in aqueous extracts. For this purpose, 3/V x 8" strips were cut
from the Gelman AE hi-vol filter samples and extracted following
the EPA/EMBL methylthymol blue Technicon II procedure which in-
cludes vacuum filtration with a fine porosity fritted disc. Two
urban and two rural samples were chosen; the samples used had the
- 69-
-------
e
c
160
140
120
100
80
60
©
40
20
0©
©
8
12
©
I
I
16 20
Sulfite,
24
28
32
Figure 20. Scatter diagram of zinc versus sulfite concentrations.
©
36 40
-------
two highest sulfate levels experienced at each location. Col-
loidal clay was selected as a convenient and plausible model for
the suspended particles in the extracts. Colloidal solutions
were prepared by shaking a portion of bentonite clay in water,
allowing the suspension to settle for three hours and filtering
through a Whatman No. 1 filter. Ninety degree light scattering
was measured at 600 run using an Aminco Spectrofluorimeter cali-
brated using the bentonite clay suspensions. The wavelength
600 nm was chosen to avoid fluorescence observed in atmospheric
sample extracts at shorter wavelengths. It should be noted that
the relationship between clay concentration and light scattering
depends on the particle size distribution and may vary with the
age of the suspension, between suspensions prepared at different
times or with use of different types of clay.
Table 23 summarizes the results expressed as /ug/ml clay. These
determinations indicate that kO /ug/ml clay approximates the
highest level of light scattering particles observed.
6. Yellow Chromophore Concentration of Particulate Extracts
Aqueous extracts of atmospheric samples can be decidedly yellow.
This corresponds to absorbance in the UOO-500 nm region. Since
the MTB method employs 460 nm for quantitation some interference
would be expected if no blank correction is made.
- 71 -
-------
TABLE 23
TURBIDITY OF AQUEOUS EXTRACTS (AS pg/ml COLLOIDAL CLAY
OF 8 x 10 GLASS FIBER HI-VOL SAMPLES FROM ST. LOUIS3)
Sample ID Colloidal Clay (yg/ml)
604 GH 28
616 GH 36
1203 GH 15.4
1204 GH 22
Filter Blank: 10 ± 1 (mean of 2).
a. A 3/4" x 8" strip extracted by the EMSL/MTB procedure with final
volume 50 ml.
-72 -
-------
Extracts absorbing in the UOO-500 run region will typically absorb
even more strongly in the UV region in the vicinity of 312 nm,
the wavelength used for quantitation with the BC method. While
a blank correction is included in the method, its large value is
expected to diminish analytical precision for sulfate.
The study of this interferent required, first, a model chromophore
and second, a determination of the relevant concentration for
such a model.
UV-Visible scans were obtained for the four St. Louis extracts
analyzed for turbidity. The scan for the solution with greatest
absorbance in the i4-00-500 nm is shown in Figure 21 compared to
scans for two concentrations of p-benzoquinone (p-bzq) and for a
solution of coffee bean extract. Five jug/ml p-bzq approximates
the absorbance of this extract both at 312 nm and at 560 nm.
- 73 -
-------
0.2
Absorbance
per
cm 0.1
I
I
I
300
350 400
Wavelength (ran)
450
A - 3/4" strip of #616 extracted in 50ml
(most yellow sample)
B » Blank filter extract + 5 ^g/ml p-benzoquinone
C = Blank filter extract + 20 ;ig/ml freeze-dried coffee
D * 15 ;ug/ml p-benzoquinone in water
Figure 21. Comparison of UV-visible scans of atmospheric sample aqueous
extracts and candidate model chromophores.
-------
VII. EFFECTS OF INTERFERENTS
A. Study Design
In the first year of this program, interferent concentrations used
were 10 and 30 jug/ml with 20 and 60 ng/ml sulfate. This year's con-
centrations were to be dictated based upon the "highest observed
concentrations". However, the observed concentrations, in jug/ml,
depend upon the filter size, type, sampled air volume, filter area
extracted, and final volume of extract.
Since glass fiber filters, themselves, proved to be a significant
source of interferents, the EPA protocol for hi-vol sampling was
adopted in calculating the interferent concentrations for this study.
The conditions assumed were 21f-hour sampling (2000 m3 of air)
through an 8 x 10" filter and extraction of a 3/V x 8" strip into
a final volume of 50 ml HgO. It was further assumed that the inter-
ferents extracted per unit area from the 126 mm glass fiber and
8 x 10" glass fiber filters were equal. The interferent concentra-
tions observed with the glass fiber 126 mm total filters (which were
always greater than those from the tissue quartz filters, loaded or
unloaded) were used to calculate concentrations expected under the
above-assumed conditions. Although studies of single interferents
would be restricted to 5 substances and, for studies with inter-
ferent pairs and quartets, to only U, Table 2*4- calculates the maximum
concentrations expected for 7 potential interferents as determined
with atmospheric samples. The variation with analytical method re-
sults from the dilution required for analysis by the modified Brosset
method.
- 75 -
-------
Table 2k
CALCULATED MAXIMUM CONCENTRATIONS OF POTENTIAL INTERFERENTS
UNDER CONDITIONS SIMULATING EPA PROCEDURES3
(Ug/ml)
Species
S103=
Cl"
Br"
Colloidal clay
p-benzoquinonec
METHOD
MTB
0.6
60
0.1
20
0.5
40
6d
Brosaet0
0.15
15
0.025
5
0.1
10
1.5
BC
0.6
60
0.1
20
0.5
40
6
a. Extracting a 3/4 x 8" strip from a 24-hour hi-vol sample
collected on an 8 x 10" glass fiber filter.
b. Concentrations are 1/4 of the values calculated
for the MTB and BC methods based on the need for
an approximately four-fold dilution into the
working range of the Brosset Method.
c. A model yellow chromophore.
d. 6 yg/ml used instead of 5 for experimental convenience.
- 76 -
-------
Excepting for phosphate, these concentrations served as the basis
for interference studies. The phosphate concentration was increased
by a factor of ca. 20 at the request of the Project Officers to
evaluate the: influence of glass fiber filters with high phosphate
blanks.*
Based on consultations with the Project Officers, phosphate, Si03=,
Cl~, colloidal clay and p-benzoquinone were chosen for single inter-
ferent investigation and for the pairs and quartet studies, chloride
was eliminated from this set. Sulfate concentrations were selected
throughout the working range of each method. Table 25 lists the
concentrations chosen for single interferent work, while Tables 26,
27, and 28 provide corresponding values for interferent pairs and
quartets.
B. Effect of Single Interferents with Known Sulfate Concentrations
Following the protocol given in Table 25, studies were completed with
each of the three sulfate methods, with three determinations for each
level. Tables 29, 30? and 31 summarize the interferent results for
the MTB, modified Brosset and BC methods, respectively.
*The glass fiber filters (Gelman No. 651^4-) recently distributed
to the California Air Resources Board from the EPA were determined
at AIHL to have 2222 + 77 Ug/8 x 10" sheet of phosphate.
- 77 -
-------
It may be noted in these three tables that the observed sulfate
levels with added interferents differ significantly from the nominal
sulfate values. These differences for the MTB and Brosset data
parallel findings for the accuracy of these techniques using the
EPA audit strips. The MTB results, on average, were 3$ low compared
to J% low for the audit strips. With the Brosset method, both the
zero interferent level samples and the audit strip results averaged
about ty%> high. With the BC method the relatively poor precision
obtained probably accounts for some of the discrepancy between
nominal and observed sulfate values. The results with the audit
strips indicate values averaging only 1$ low. Thus inaccuracy may
not be a significant contributor. With the MTB and Brosset data
interpretation of interference effects should not be hampered
if results are compared to those for the zero interferent levels.
With the BC data this same comparison is the most reasonable pro-
cedure but the poor precision greatly reduces the reliability of
conclusions regarding interference effect.
- 78 -
-------
TABLE 25
CALCULATED CONCENTRATIONS FOR SINGLE INTERFERENT STUDIES
Method
MTB
(S04 )
t*q/ml
10
30
50
(P04S)a
Eg/ml
(SiO.=)
t*q/m
Colloidal clay
10. 5. 2.5
same
same
120, 60, 30 80, 40, 20
same same
same same
p-benzoquinone
tig/ml
40, 20, 10 12. 6, 3
same same
same same
Brosset
5
8
11
0.3, 0.15, 0.08 30, 15, 8 20, 10, 5
same same same
same same same
10,5, 2.5 3, 1.5, 0.75
same same
same same
Barium
Chloranilate
15
25
40
1.2, 0.6, 0.3 120, 60, 30 80, 40. 20
same same same
same same same
40, 20, 10 12, 6, 3
same same
same same
Maximum atmospheric concentrations: 0.2 fig/m3 P04 , 0.0^ /ag/rn3 S0^~ and 0.15 jug/oi3 Br. Assuming collection of
2000 m3 on an 8" x 10" filter and extract 3/V x 8" strip into 50 ml HaO, concentrations : 0.6 jig/ml PO^S,
0.12 (ig/ml 80s and 0.1)6 ng/ml Br. For MTB msthod the concentrations were increased above those calculated
from marijmnn observed concentrations at the request of the Project Officers.
Extraction of 110 cm2 of glass fiber filter in 100 ml HaO gave 85 jjg/ml SiOs~ (as Si02) and < 32 ug/ml total
halogen (ca 98£ d"). Data indicate filter the principal source o£ these ions. For a 3A" x 8" atrip of
glass fiber extracted In 50 ml HeO> concentrations: 60 yg/ml SiOa (as Si02) and 22 jig/ml Cl~.
-------
Table 26
CALCULATED CONCENTRATIONS FOR STUDIES OF INTERFERENT PAIRS
Concentrations of Interferent for Experiment No.
Interferent
Si03=
Clay
P04=
p-benzoquinone
Blank
0
0
0
0
I
60
ho
0
0
II
60
0
10
0
in
60
0
0
6
IV
0
ho
10
0
V
0
ho
0
6
VI
0
0
10
6
OD
o
_
Concentrations for use with MTB and BC Methods. With Brosset method divide all concentrations by
-------
TABLE 27
CALCULATED CONCENTRATIONS FOR STUDIES OF INTERFERENT QUARTETS (Mg/ml)*
Concentrations of Interferent for Experiment No.
Interferent Blank
0
Clay 0
po4E o
p-benzoquinone 0
I
00
I
60
40
5
6
II
30
40
5
6
III
60
20
5
6
IV
60
40
2.5
6
V
60
40
5
3
VI
90
40
5
6
VII
60
60
5
6
VIII
60
40
7.5
6
IX
60
40
5
9
*Concentrations for MTB and BC methods. For the Brosset method divide all concentrations by 4.
-------
TABLE 28
SULFATE CONCENTRATIONS FOR STUDY OF
INTERFERENT PAIRS AND QUARTETS (yg/ml)
Method Working Range SOu Level la^ SOu Level 2
MTB 7-75 25 40
Modified Brosset 3-13 5 10
Barium chloranilate 10-50 25 40
a) Approximately one third of the working range
b) Approxiamtely two thirds of the working range
-82-
-------
TABLE 29
INTERFERENCE EFFECT WITH THE METHYLTHYMOL BLUE METHOD (yg/ml OBSERVED SULFATE)1
Nominal
Sulfate Level, yg/ml
Inter ferent. Level ,
yg/ml
Interferent
Silicate
-
Chloride
Clay
p-benzoquinone
Nominal
Sulfate Level, yg/ml
Interferent, Level ,
yg/ml
Phosphate
0
14.0
.4
14.2
.3
14.3
.2
14.3
.1
0
9.3
.2
A
14.8
.1
14.4
.4
14.7
.1
14.5
.2
A
9.7
.3
15
B
15.0
.03
14.3
.4
15.3
.2
14.7
.1
10
B
10.0
.3
C
15.6
.05
14.3
.4
16.9
.3
14.9
.5
C
10.2
.3
0
24.1
.3
24.1
.3
24.6
.3
24.7
.4
0
29.1
.1
A
25.4
.3
24.4
.7
25.1
.1
24.8
.2
A
30.6
.1
25
B
25.6
.2
24.6
.7
25.7
.2
25.3
.2
30
B
31.0
.2
C
26.1
.1
24.6
.5
26.9
.03
25.6
.2
C
31.6
.3
0
40.0
.05
39.3
.7
39.9
.1
40.4
.2
0
49.5
.1
A
41.5
.2
39.4
.8
40.6
.2
40.5
.2
A
52.5
.1
40
B C
42.2 42.5
.04 .0
39.6 39.9
.6 .8
41.4 42.7
.1 .1
41.0 41.9
.2 .1
50
B C
53.0 53.7
.2 .2
I ;
00
9.3
.2
a.
b.
9.7
.3
Mean of
10.0
.3
10.2
.3
29.1 30
.1
.6 31.0 31.
.1 .2
6
3
49.5 52.5
.1 .1
53.0 53.7
.2 .2
3 determinations ± la value shown below mean.
Interferent concentration, yg/ml:
A
P0i»
Si02
Cl
Clay
p-bzq
2.5
30
10
20
3
B
5
60
20
40
6
C
10
120
40
80
12
-------
Table 30
INTERFERENCE EFFECT WITH MDDIFIED BROSSET METHOD (yg/ml OBSERVED SULFATE)
a
Nonlnal
Sulfate Level, jug/ml
Inter ferent b
Level, /*g/«l
INTERFERENT
Phosphate
Silicate
Chloride
Clay
p-benzoqulnone
5
0
5.5
.1
5.5
.2
5.5
.1
5.0
.3
5.2
..1
a.
b.
-
A
5.3
.2
5.3
.2
5.5
.1
5.4
.2
4.8
.1
Mean of
B
5.3
.1
5.0
.1
5.4
.1
5.7
.2
4.7
.1
C
5.3
.2
5.1
.1
5.3
.2
6.0
.2
4.6
.1
8
0
8.0
.1
7.6
.04
7.9
.03
8.1
.1
7.8
.1
ABC
7.6 7
.2
'7.7 7
.1
7. 9 7
.2
8.3 8
.1
7.8 7
.1
three determinations with
Interferent
concentrations, /*g/ml
•
.7 7.8
.4 .1
.6 7.5
.2 .2
.9 7.9C
.03 .1
.6 8.8
.1 .1
.7 7.6
.1 .1
11
0
11.2
.2
11.6
.2
11.2
.1
11.0
.2
11.8
.4
+ !<* value shown
r A
P04
SiOa
Cl
Clay
p-bzq
.08
8
2.5
5
.75 1
A
11.2
.04
11.6
.1
11.2
.1
11.6
.2
11.5
.3
below
B
.15
15
5
10
.5
B
11.2
.2
11.7
.1
11.0
.2
12.1
.2
11.5
.4
mean.
C
.30
30
10
20
3
C
11.2
.1
11.6
.2
11.0
.2
12.3
.2
11.3
.4
C. The sulfate value for the "zero" level sample to be compared
with this was 8.1 + .1.
-------
Table 31
INTERFERENCE EFFECT WITH THE BARIUM CHLORANIIATE METHOD (Mg/ttl OBSERVED SULFATE)a
Nominal
Suifate Level, /(g/ml
Interferent
Level, /*g/ml
INTERFERHiT
Phosphate
Silicate
Chloride
Clay
Nominal
Suifate Level, uq/ml
Interferent
Level, ug/ml
p-benzoquinone
15
0 A B C
14.3 14.9 17.5 16.5
1.7 .1 1.3 4.2
14.4 14.8 14.7 14.6
2.9 3.1 2.7 1.7
14.4 15.0 16.5 15.6
.3 .8 1.7 .3
14.9 15.1 15.8 16.4
.1 .3 .7 .2
10
O A B C
9.9 9.5 11.3 11.7
.3 .8 1.3 .8
0
23
2
25
2
24.
1.
25.
•
0
29.
•
.9
.0
.0
.3
0
3
1
6
6
8
a. Mean of 3 determinations ±
b. Interferent concentration.
A
25
2.
26
2.
24
25
25
B C
-
.4 26.5 26.2
6
5.3 2.1
.6 24.8 28.0
9
.8
.4
.5
1.5
A
29
30
.1
.3
1
1.3 5.0
24.2 25.1
.4 ,7
25.5 26.4
.3 1.0
B C
29.7 30.8
.7 .8
0
38.0
1.1
38.6
.6
40.5
.6
38.8
.6
O
49.1
.9
40
A
38.0
1.0
37.6
2.2
39.2
1.8
40.1
.8
50
A
49.9
.6
B
37.8
1.4
38.9
.9
39.8
1.3
40.6
.7
B
50.4
1.0
C
38.3
.9
38.7
2.3
37.9
.4
40.5
.9
C
49.1
2.3
a value shown below mean.
M,g/ml:
P04
SiOa
Cl
Clay
p-bzq
A
.3
30
10
20
3
B
.6
C
1.2
60 120
20
40
6
40
80
12
CD
in
-------
For the MTB data, the study with phosphate was inadvertently run at
10, 30 and 50 jug/ml sulfate. Aside from the awkwardness of data pre-
sentation, this error seemed insufficient to warrant repetition at the
correct level. Observed interferences with this method were either
negligible or positive in direction. Chloride did not show inter-
ference beyond experimental error while silicate and phosphate gave
the most interference (up to 11$). While the maximum interference from
p-benzoquinone was only +**•$, it was observed at »"n sulfate levels.
With the modified Brosset method, the results with single interferents
suggest negligible (i.e. < 5$) effects for phosphate and chloride.
Silicate exhibits up to a 10$ negative interference at 5 Mg/ml sulfate,
clay a 10-20$ positive interference at 5 and- 8 iJ.g/aH. sulfate and p-
benzoquinone, up to a 12$ negative interference at the 5 Mg/ml sulfate.
In the latter case, small but consistent negative interferences are
seen at 8 and 11 jug/ml sulfate as well.
For the BC study, p-benzoquinone was inadvertently run at 10, 30 and
50 jug/ml sulfate instead of 15» 25 and kO. Again, aside from the
awkwardness of data presentation, this error seemed insufficient to
warrant repetition at the correct sulfate levels.
Interpretation of interference effects with the BC method is made
difficult by the relatively poor precision of the method. Co-
efficients of variation up to 25$ at the lowest sulfate levels were
seen for triplicate determinations. Nevertheless, all of the
- 86 -
-------
interferents except silicate caused measurable (> % positive)inter-
ference at the lowest sulfate level. At higher sulfate, interferences
are either < 5% or uncertain "because of large standard deviations in
the results. The largest interference is shown for p-benzoquinone
with 10 ug/ml sulfate. The "bentonite clay suspension generally
yielded a negligible effect with possibly a small positive error at
the highest clay level. Phosphate exhibited a 10-15$ positive inter-
ference at 15 and 25 jug/ml sulfate. Such interference was expected,
based on the approximately neutral pH of the method and the resulting
formation of insoluble barium phosphate.
C. Effect of Interferent Pairs with Known Sulfate Concentration
These studies as well as those with interferent quartets were done
seeking evidence of interactions between interferents. Tables 32,
33, and 3^ summarize results for studies of interferent pairs for
the three sulfate methods. As was the case for single interferents
the MTB results without added interferents averaged 3$ low probably
reflecting the same trend as found for EPA audit strips. However,
the Brosset results in Table 33 (averaging ty> low) differ in this
respect from those with audit strips (b% high) without apparent cause.
The BC results for zero interferents were within 2$ of the nominal
values. Again comparison of results against the blank values is the
most reliable strategy for interpretation of results.
With the MTB data, greater percentage interference was observed at
the lower sulfate concentration but results at the higher level were
qualitatively similar. A comparison of paired results to those
- 87 -
-------
TABLE 32
INTERFERENCE EFFECT WITH THE MTB METHOD USING INTERFERENT PAIRS
Experiment
Interferent
Interferent
Concentration (yg/ml)
Observed Sulfate at
Nominal~Sulfate Level
«a
(yg/ml)
Blank
I
II
III
IV
V
VI
A
Si03 =
Si03 =
=
Si03
Clay
Clay
"*"
B
Clay
POi^"
b
p-bzq
POli"
p-bzq
p-bzq
A
0
60
60
60
40
40
10
B
0
40
10
6
10
6
6
25
23.7 ±
26.5 ±
26.7 ±
25.1 ±
26.6 ±
26.2 ±
27.7 ±
.03
.5
.4
.4
.3
.2
.2
40
39.8 ±
41.7 ±
43.7 ±
41.9 ±
43.0 ±
41.6 ±
43.6 ±
.4
.6
.3
.3
.07
.4
.6
a. Mean ± 1 a for three determinations.
b. p-benzoquinone, a model yellow chromophore.
c. A fresh clay suspension was prepared from kaolinite clay in place of the
bentonite clay which was used up in the single interferent work. The light
scattering of a 60 yg/ml kaolinite suspension was equivalent to that from
100 yg/ml bentonite.
-88 -
-------
Table 33
INTERFERENCE EFFECT WITH MODIFIED BROSSET METHOD USING INTERFERENT PAIRS
Experiment
Interferent
Interferent
Concentration (jug/ml)
Observed Sulfate at
Nominal Sulfate Level (jag/ml)*
Blank
I
II
III
IV
V
VI
A
Si03=
Si03=
Si03=
Clay
Clay
P04-
B
—
Clayc
P04-
p-bzq
P04=
p-bzq
p-bzq
A
0
15
15
15
10
10
2.5
B
0
10
2.
1.
2.
1.
1.
5
5
5
5
5
5
lf.9±
lf.9±
5.1 +
5.0 +
5.0 +
lf.9±
lf.9 +
— _ _ _ __ _ \.r—\^*t /
10
.1
• 3
.2
.2
.1
.2
.1
9
9
10
9
10
9
9
.if
.7
.if
.8
.2
.8
.9
+
+
+
+
4-
+
+
.2
.2
.1
• 3
•2
•2
.If
a. Mean + 10 ^or three determinations.
b. p-benzoquinone, a model yellow chromophore.
c. A fresh clay suspension was prepared from kaolin! te clay in place of the
bentonite clay which was used up in the single interferent work. The
light scattering of a 60 /itg/ml kaolin!te suspension was equivalent to
that from 100 Mg/ml bentonite.
-89-
-------
TABLE 3^
INTERFERENCE EFFECT WITH BARIUM CHLORANIIATE METHOD USING INTERFERENT PAIRS
Experiment
Interferent
Blank
I
II
III
IV
V
VI
A
Si03=
Si03=
Si03=
Clay
Clay
P04=
B
Clay
P04=
h
p-bzq
P04=
p-bzq
p-bzq
Interferent
Concentration
A
0
60
60
60
ko
ko
10
B
0
40
10
6
10
6
6
Observed Sulfate at
Nominal Sulfate Level (jug/ml)a
25
24.9 + .5
24.7 + .7
26.1 + .9
25-3 ± -3
26.8 +1.2
25.7 + .2
27.8 + .4
39-3 + .2
39-3 ± .8
in.9 +1-0
4o.i + .8
Ul.4 +l.l
39.6 + .2
Ul.7 1.1
+ 1 C7 for three determinations.
p-benzoquinone, a model yellow chromophore.
A fresh clay suspension was prepared from kaolinite clay in place of the bentonite
clay which was used up in the single interferent work. The light scattering of a
60 wg/ml kaolinite suspension was equivalent to that from 100 ug/ml bentonite.
-90 -
-------
obtained individually is given in Table 35 for the lower sulfate
level. The paired results are seen to often exceed 10$ interference
and are roughly additive from the results for the individual inter-
ferents (i.e. no interaction is evident).
With the modified Brosset method, greater interferences were observed
at the higher sulfate level. Table 36 compares results for inter-
ferents singly and in pairs at 10.5 + .5 MS/ml sulfate. The results
for clay-phosphate, clay-p-benzoquinone and silicate-clay appear to
be somewhat less than additive. With silicate-phosphate, the pair
result exceeds significantly the sum of the interference effects
individually. However, as may be seen comparing Tables 25 and 26,
the phosphate concentration used in the paired interferent studies
for the Brosset (and BC) method was about a factor of 10 higher than
that used in the single interferent work. Thus, no conclusions are
possible regarding non-additivity for paired interferents involving
phosphate.
For the BC method, the study with interferent pairs was done at
sulfate levels for which only phosphate proved to be a significant
interferent in single interferent experiments. The pair results
summarized in Table 3^ indicate, in all cases, interferences < 12$.
in contrast to the corresponding Brosset data, similar interference
is observable at both sulfate levels studied. Table 37 compares these
results to those for the interferents individually. The paired inter-
ferent results are roughly predictable from those for the single inter-
ferents (i.e. no interaction is evident). The scatter in the data
make conclusions tenuous, however.
- 91 -
-------
TABLE 35
COMPARISON OF RESULTS OF SINGLE AND PAIRED
INTERFERENTS WITH THE MTB METHOD3
Interferent Pair
A + B,
Result
Individual
Results
A
Si03 =
Si03 =
Si03 =
clay
clay
P0t4~
B_
clay
POit"
p-bzq
POi+=
p-bzq
p-bzq
A + B Result
I Indiv. Results
x 100
o
o
o
75
67
66f
100f
129
Mean: 87 ± 24
a. 0 equals <_ 5% interference
+ equals > 5% positive interference
++ equals > 10% positive interference
b. Results at 25 yg/ml sulfate. At 40 yg/ml sulfate, interferences are
less pronounced.
c. Results in the range 25-30 yg/ml sulfate with interferent concentrations
as used in paired study.
d. p-benzoquinone
e. A consistent positive interference of up to 4% is seen at all sulfate
levels studied.
f. Corrected for use of kaolinite rather than bentonite.
-92-
-------
TABLE 36
COMPARISON OF RESULTS OF SINGLE AND PAIRED INTERFERENTS
WITH THE BROSSET METHOD
A + B
Interferent Pair Resulta
A
Si03"
Si03"
SiO^
clay
clay
P04=
£
clay 0
P0it= ++
p-bzq° 0
POiT +
p-bzq 0
p-bzq 0
Individual A + B Result nf.
Results 1 Indiv. Results X
A B
0 +
0 0
0 0
od
+ 0
0 0
25
—
—
67d
44
— —
a. Results at 10 pg/ml sulfate
0 equals <_ 5% interference
- equals > 5% negative interference
+ equals > 5% positive interference
-H- equals > 10% positive interference
b. Results at 11 yg/ml sulfate
c. p-benzoquinone
d. Results of questionable value since phosphate concentration 10 times
lower than used in paired experiment.
-93 -
-------
TABLE 37
COMPARISON OF RESULTS OF SINGLE AND PAIRED INTERFERENTS
WITH THE BARIUM CHLORANILIATE METHOD
A + B
a
Interferent Pair Result
A
Si03"
Si03 =
Si03 =
clay
clay
P0if=
B_
clay 0
POtf." +
p-bzq 0
P0k= +
p-bzq 0
p-bzq +
Individual A + B Result .-..
Results I Indiv. Results x
A
0
0
0
0
0
+d
B_
0
+d 50e
0
+d 56e'f
0 80f
0 -.12le
Mean: 77 ± 32
a. Results at 25 and 40 yg/ml sulfate. See Table 35 for definition of
symbols.
b. Results in the range 25-50 ug/ml sulfate with interferent concentrations
as used in paired study except for P0i|~.
c. p-benzoquinone
d. At 25 pg/ml only
e. Assumes a ceiling on phosphate interference of 2.4 yg/ml apparent
sulfate at > 1.2 pg/ml phosphate.
f. Corrected for use of kaolinite rather than bentonite clay
-94 -
-------
D. Effect of Interferent Quartets with Known Sulfate Concentrations
The intent of this study was first, to evaluate the combined effect
of four interferents on sulfate determination and secondly, to
evaluate the influence of individual interferents in the presence of
a more realistic matrix. The procedure was to systematically vary
the concentration of one interferent at a time to concentrations
equal to 1/2 and 3/2 times the level chosen as relevant for study.
Tables 38, 39 and hO report results of this study done at two sulfate
concentrations with the three sulfate methods. Just as with the single
and paired interferent studies the zero interferent level sulfate
values for the quartet studies often differed from the nominal values.
With the MTB method, results averaged k% low; with the Brosset tech-
nique, results averaged 2% high and with the BC procedure 1$ low.
These MTB and Brosset results parallel the trends observed with the
EPA sulfate audit strips. The BC results here indicate substantially
greater error than was obtained with the audit strips but the poor
precision of the analyses hamper conclusions (e.g. all results at
25 jig/ml are within 2 a of the zero solution). Calculation of inter-
ference effects by comparisons with blank solution results should
provide reasonably accurate and precise results, at least for the
MTB and Brosset techniques.
For the MTB method, similar relative interference effects were ob-
served at both 25 and UO Mg/ml. As with the pairs, combined inter-
ferences often exceeded 10$. A comparison of quartet and individual
interferent results, shown in Table *H, reveals that the quartet
- 95 -
-------
TABLE 38
INTERFERENCE EFFECT WITH MTB METHOD USING INTERFERENT QUARTETS
Experiment
Interferent Concentration (yg/ml)
a. p-benzoquinone
b. Mean ± 1 o for 3 trials
c. See footnote c, Table 32
Observed Sulfate at Nominal
Sulfate Level (pg/ml)
Blank
I
II
III
IV
V
VI
VII
VIII
IX
Si03
0
60
30
60
60
60
90
60
60
60
ft
Clay0
0
40
40
20
40
40
40
60
40
40
_
P0u=
0
5
5
5
2.5
5
5
5
7.5
5
0
p-bzq
0
6
6
6
6
3
6
6
6
9
25
23.5 ±
26.4 ±
26.4 ±
26.1 ±
26.4 ±
25.9 ±
26.7 ±
27.4 ±
26.9 ±
26.6 ±
.3
.2
.2
.1
.2
.1
.3
.1
.4
.3
40
38.9 ±
43.1 ±
43.2 ±
43.1 ±
43.0 ±
41.8 ±
43.8 ±
44.0 ±
44.0 ±
44.0 ±
.2
.2
.2
.2
.2
.2
.3
.4
.2
.2
-------
TABLE 39
INTERFERENCE EFFECT WITH MODIFIED BROSSET METHOD USING INTERFERENT QUARTETS
Experiment
Interferent Concentration (pg/ml)
Observed Sulfate at Nominal
Sulfate Level (yg/ml)
Blank
I
II
III
IV
V
VI
VII
VIII
IX
SiQ3
0
15
7.5
15
15
15
22.5
15
15
15
ClayC
0
10
10
5
10
10
10
15
10
10
~ = .a
POit p-bzq
0
1.
1.
1.
.
1.
1.
1.
1.
1.
25
25
25
63
25
25
25
88
25
0
1
1
1
1
1
1
1
2
.5
.5
.5
.5
.75
.5
.5
.5
.25
5.
5.
5.
5.
5.
5.
5.
5.
5.
5.
5
2 ±
0 ±
2 ±
4 ±
0 ±
1 ±
1 ±
2 ±
2 ±
1 ±
.2
.1
.3
.1
.1
.3
.2
.1
.02
.4
10
10.0 ±
10.2 ±
10.6 ±
10.7 ±
10.6 ±
10.7 ±
10.9 ±
10.9 ±
10.8 ±
10.1 ±
.2
.2
.2
.3
.3
.5
.2
.1
.1
.2
b-benzoquinone
mean ± 1 c for 3 trials
°see footnote c, Table 33
-------
TABLE 40
INTERFERENCE EFFECT WITH BARIUM CHLORANILATE METHOD USING INTERFERENT QUARTETS
vo
00
Experiment
Interferent Concentration (pig/ml)
ap-benzoquinone
Tfean ± 1 cr for 3 trials
Observed Sulfate at Nominal
Sulfate Level (yg/ml)
SiQ3
Blank 0
I 60
II 30
III 60
IV 60
V 60
VI 90
VII 60
VIII 60
IX 60
Clay
0
40
40
20
40
40
40
60
40
40
P0u=
0
5
5
5
2.5
5
5
5
7.5
5
p-bzq
0
6
6
6
6
3
6
6
6
9
25
23.9 ±
25.8 ±
25.6 ±
26.0 ±
25.5 ±
24.0 ±
25.7 ±
25.5 ±
25.7 ±
25.6 ±
1.4
1.2
1.6
.9
.7
.3
.3
1.4
.5
.9
_40
38.7 ±
40.1 ±
40.4 ±
41.8 ±
41.0 ±
40.0 ±
41.4 ±
41.3 ±
42.0 ±
42.5 ±
.9
.7
.7
.2
1.0
.9
1.0
.9
1.4
.9
-------
TABLE 41
COMPARISON OF RESULTS OF SINGLE AND INTERFERENT QUARTETS
WITH THE MTB METHOD
Individual Result Combined
Experiment
I
II
III
IV
V
VI
VII
VIII
IX
SiOa Clay
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
+ o
+ 0
+ 0
POiT p-bzq
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
I Individual
49
51
73
53
44
52
57
55
50
Mean 54 ± 8
a. At 25 and 40 yg/ml sulfate, See Table 35 for definition of symbols.
b. In the range 25-50 yg/ml sulfate. A double underscore denotes an
interferent concentration cut in half. A single underscore denotes
an interferent concentration increased by 50%.
c. See footnote e, Table 35.
-99-
-------
results are about half the sum of the individual results. Systemati-
cally increasing the concentration of single interferents reveals
either a negligible change or an increase in observed sulfate.
For the Brosset results with interferent quartets the following ob-
servations can be made:
1. In no case does the observed interference exceed 10$ of the
nominal sulfate concentration.
2. As in the study with interferent pairs, interference is more
pronounced at the higher sulfate level, and »n interferences
are positive.
3- Except for p-benzoquinone, the results of varying one interferent
at a time show no consistent trend. At 10 jug/ml, both increasing
and decreasing interferent concentrations gives higher values than
the initial experiment (l). With p-bzq, results are consistent
with a negative interference.
The combined results are compared to those of the individual inter-
ferents in Table h2. The combined results range from 11 to 150$ of
the effect predicted. Relatively small interference effects and a
somewhat poorer precision of the method compared to the MTB probably
account for the scattered results.
For the BC method, Table UP reports results of this study from which
we make the following observations:
1. In no case does the observed interference exceed 10$ of the
nominal sulfate level.
- 100 -
-------
TABLE 42
COMPARISON OF RESULTS OF SINGLE INTERFERENT QUARTETS WITH
THE BROSSET METHOD AT 10-11 yg/ml SULFATE*
Individual Result
Experiment
I
II
III
IV
V
VI
VII
VIII
IX
SiQ3 Clay
0 +
0 +
0 +
0 +
0 +
0 +
0 +
0 +
0 +
POiT
0
0
0
0
0
0
0
_0
0
P-bzg
0
0
0
0
0
0
0
0
0
Combined
I Individual
x 100
22
67
86
150
130
100
122
113
11
Mean 89 ± 48
a. See Table 35 for definitions of symbols.
b. At 11 ug/ml sulfate. A double underscore denotes an interferent con-
centration cut in half. A single underscore denotes an interferent
concentration increased by 50%.
-101-
-------
2. As in the study of interferent pairs, interference effects are
similar at both sulfate levels studied.
3. Varying systematically the concentrations of single interferents
reveals no consistent pattern. Possibly the most significant
result is the increase in apparent sulfate on increasing the
p-benzoquinone level from 6 to 9 jug/ml at 40 jug/^l sulfate. The
increase, about 6%, is similar to the *4$ interference observed
in single interferent studies.
The quartet results are compared to those found in studies of inter-
ferent individually in Table ^t-3. As with the MTB method, the quartet
results are generally about half of those predicted from the indivi-
dual interferents.
E. Summary and Discussion of Interference Studies
Table kk- summarizes results of the single interferent studies for the
three sulfate methods and generally reflects the sensitivity of each
method around its mid-range. The modified Brosset method is subject
to the fewest interferents and the MTB, the greatest number from the
interferents selected.
Interference effects for the MTB and a manual modified Brosset method
have been previously reported. For the MTB method, a positive inter-
ference from phosphate was observed only at the upper end of the work-
ing range. The present results indicate substantial sensitivity to
phosphate at lower sulfate levels as well. Previously, colloidal clay
- 102 -
-------
TABLE 43
COMPARISON OF RESULTS OF SINGLE AND INTERFERENT QUARTETS WITH
THE BARIUM CHLORANILATE METHOD AT 25 AND 40 yg/ml SULFATE* ..
b c
_ Individual Result Combined
Experiment
I
II
III
IV
V
VI
VII
VIII
IX
SiQ3
0
0
0
0
0
0
0
0
0
Clay
0
0
_0
0
0
0
0
0
0
POtT p-bzq
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
Z Individual
56
50
75
47
3
53
38
53
41
Mean 46 ± 19
a. See Tables 35 and 42 for definition of symbols.
b. In the range 25-50 pg/ml sulfate.
c. Assumes a ceiling of 2.4 yg/ml apparent sulfate for phosphate
interference above 1.2 yg/ml phosphate.
-103-
-------
TABLE 44
SUMMARY OF SINGLE INTERFERENT RESULTSa)b>
Interferent MTB Brosset BC
silicate +00
phosphate + 0 +
colloidal clay + + 0
p-benzoquinone Oc Oe 0
chloride 0 0 +d
aO equals <_ 5% interference
+ equals > 5% positive interference
-H- equals > 10% positive interference
°Sulfate level at mid to upper range of each method
except as noted
CA consistent + 4% interference observed at all sulfate levels
dobserved only at £ mid-range of this method
eUp to a 4% negative interference is observed at the mid and
upper range and up to a 12% negative interference
near the bottom of the sulfate range.
-------
exhibited substantial positive interference at mid-range sulfate
levels which is consistent with present results. The small positive
interference by a yellow chromophore is consistent with the absence
of any blank correction with this method. The positive interference
by silicate was not previously observed in this study. While the
cause of the differences in observations is unknown, the present
results, reflecting three independent trials rather than a single
trial as used earlier, are considered more reliable.
The previous studies with a modified Brosset method using dioxane and
acetone as solvents did not show interference by colloidal clay, in
contrast to the current results using isopropanol. A small negative
interference by p-benzoquinone was also observed in the earlier work.
Interference effects for the BC method were previously reported by
workers at TRW who observed substantial interference by chloride at
100 jug/ml levels of interferent. The present results indicate a
greater sensitivity to chloride.
Studies of interferent quartets indicate about half the expected
effect with the MTB and BC methods implying some degree of interaction.
However, studies of interferent pairs has generally failed to reveal
clear evidence of such interactions; the results with interferent
pairs are, with one exception, roughly predictable as the sum of the
effects of the interferents individually. The exception is the case
of the phosphate-silicate pair with the modified Brosset method. The
- 105 -
-------
results for the pair reveal > 10$ positive interference while ex-
hibiting negligible effects individually. However, the much higher
phosphate concentration used in the paired interferent studies
vitiates comparisons of single and paired interferents results.
F. Correction of Atmospheric Sample Results for Interference Effects
Having analyzed at least selected samples for a number of interfer-
ents, these data may be used to estimate the maximum likely effects
of these interferents on the analysis of a given filter type by a
given method. Table k$ compiles calculated maximum interferent con-
centrations for each filter type and analytical method. The concen-
trations differ by analytical method since differing degrees of
dilution were used to achieve sulfate concentrations in the working
range of each method. Approximate average dilution factors were used
for the calculation. The assumptions necessary to construct such a
table are listed as footnotes.
Using these estimated maximum concentrations and the results for the
single interferent studies (Tables 29, 30? 31)? the m^vinmm estimated
errors are calculated for the mid-range sulfate level of each method
in Table k6.
The principal conclusions of these calculations are:
1. Significant errors due to the interferents studied are possible
with all filter media.
2. Due primarily to the small air volume sampled, the error, in jug/m3,
is substantially greater for the 126 mm glass fiber compared to the
glass fiber hi-vol filters.
- 106 -
-------
Table ^5
MAXIMUM CALCULATED INTERFERENT CONCENTRATIONS UNDER
CONDITIONS USED FOR MTB, BROSSET AND BC ANALYSES (jug/ml)a
Fluoropore
Silicate
Phosphate
Chloride and
Bromide
Sulfite
Colloidal Clay
p-benzoquinone
MTB
Brosset
BC
MTB
Bros set
BC
MFB
Brosset
BC
MTB
Brosset
BC
MTB
Brosset
BC
MTB
Brosset
BC
11
k
11
.2
.05
.2
Z o
~ 0
.01*
.01
.0*
xD
Q.
2T
6b
1
.k
1
8 x 10" Glass
66
_
66
.5
.5
21*
2k
.11*
-
.11*
1*0
-
1*0
6
-
6
126 mm Glass
81*
17
k9
.5
.1
• 3
31
6
18
.09
.02
.05
38
8
22
5
1
3
126 mm Quartz
1*1
ll*
29
•03
.01
.02
5
2
.09
.03
.06
2l*b
8.
17
5
1
3
a. Based on observed maximum concentrations corrected for average degree of dilution
used for each method.
b. Calculated assuming filter blank =» 0.
- 107 -
-------
Table k6
ESTIMATED MAXIMUM ERROR FOR ATMOSPHERIC SAMPLES
AT MID-RANGE SULFATE CONCENTRATIONS3"'13
Combined Error
Filter
Fluoropore
8 x 10" aiass
126 mm Glass
126 mm. Quartz
MEB
Brosset
BC
MCB
Brosset
BC
MTB
Brosset
BC
MTB
Brosset
BC
Si03~
+ .5
0
0
+1.5
-
0
+1.7
0
0
+1.1*
0
0
POi
+ .1
- .3
+1.0
+ -3
_
+2.2
+ -3
. .l|
+1.5
0
0
0
cr
0
0
0
+ .5
-
0
+.5
0
0
+.1
0
0
SOj-
0
0
0
+.1
-
+.1
+.1
0
+.1
+.1
0
+.1
Clay
+ .2
+ .1
+ .1
+1.1
-
+ .U
+1.0
+ .u
+ .4
+ .6
+ .k
+ A
p-bzq
0
0
0
+ .6
-
0
+ .5
0
0
+ -5
0
0
Ug/ml
+ .8
- .2
. +1.1
+4.1
-
+2.7
+!*.!
0
+2.0
+2.7
+• .4
+ .5
^^
+1.1
- -3
+1.5
+ -5
-
+ -3
+1.6
0
+ .8
+1.0
+ .2
+ .2
a. Estimations based on Tables 29-31 and l<5
b. ug/ml sulfate, except as noted.
c. Errors taken as approximately additive.
- 108 -
-------
3« The modified Brosset method results are subject to tnin-imn.1
interference effects due to the greater dilution employed and
the lesser sensitivity of this method to interferents.
- 109 -
-------
VIII. EVALUATION OF ARTIFACT SULFATE FORMATION WITH ATMOSPHERIC SAMPLES
The phenomenon of fixation of S02 on a filter surface by interaction with
the filter medium and particulate matter already collected, and subse-
quent analysis as water soluble sulfate, is referred to as artifact sul-
fate formation. Nearly all studies to date have concluded that S02-
filter interactions predominate over S02-particulate interactions.
The significance of filter choice is illustrated in Table V7, taken from
studies "by R. Coutant, Battelle Columbus Lab. Sampling was done from a
manifold drawing ambient air into two equivalent tubes. On each tube,
four filter holders were installed permitting simultaneous sampling. For
the work described in Table Vf, 150 ppb S02 was introduced into one of
the manifold tubes upstream from the samplers. Thus, the sulfate sampled
from the tube with added S02 should reflect any incremental increase in
artifact sulfate formed due to this S02. For the two high pH glass fiber
filters, the increase in observed sulfate was 7 to 8 jug/m3. Comparing
the glass fiber with the neutral cellulose ester and Teflon filters, the
observed sulfate levels were substantially higher. While filter pH may
not be the only factor controlling artifact sulfate formation via S02
sorption, it appears to be dominant.
The filter media used in the two years of sampling for the joint AIHL-EPA
study are the following:
Summer
Fluoropore FALP, 37 mm
Gelman glass fiber batch
8l88 (similar to Gelman A)
8 x 10"
Summer 1975
Fluoropore FALP, 37 mm
Gelman AE, 8 x 10"
Gelman AE, 126 mm
Tissue quartz 2500 QAO, 126 mm
(Pallflex)
- 110 -
-------
TABLE 47
OBSERVED SULFATE CONCENTRATIONS IN 24-HOUR SAMPLING IN COLUMBUS, OHIO3
o u
Sulfate (yg/m )
Filter
Filter pH
Without added SO
(47-99 ppb SO?)
With added SO,,
(199-250 ppb SO?)
MSA 1106BH (glass) 9.2
Gelman AE (glass) 9.4
Celotate
(cellulose acetate) 6.7
Mitex (Teflon) 7.0
29
31
19
18
35
39
21
19
Source: R. Coutant, paper presented at 172nd American Chemical Society
National Meeting, San Francisco, August 1976, and Progress
Report to EPA, Contract No. 68-02-1784, September 1975.
""Sampling done at ca. 13 cfm with 142 mm filters.
"The range in ambient S0_ concentration.
-------
Table ^8 summarizes filter pH values for the media used in the current
program based on the Battelle Study. Where experimental values are not
available, they are estimated based on values for similar filters
evaluated at Battelle.
These data suggest that there is relatively little difference between
the glass fiber filters used in the first and second year's sampling.
In the presence of significant levels of SOa, artifact sulfate formation
should follow the order:
Gelman Batch 8l88 ~ Gelman AE > Pallflex Quartz > Fluoropore
Recent studies by Pierson et al4 confirm that Pallflex Quartz 2500 QAO
and Fluoropore filters do provide minimal artifact sulfate, with the
Fluoropore filter the more inert.
Based on these studies, efforts at AIHL were directed toward evaluating
the extent of artifact sulfate formation on Gelman AE filters. Such an
evaluation requires, at least, knowledge of ambient S0g concentrations.
Using monitoring network results (one-hour average values obtained by the
conductivity method) 2^-hour average S02 concentrations could be calcu-
lated for particulate sampling on days 1-5 at the urban site (106) follow-
ing the same sampling schedule (1300-1300 hours of the following day) as
used for particulate collection.
- 112 -
-------
Table Us
THE pH OF THE FILTER MEDIA USED IN
THE TWO YEAR EPA-AIHL STUDY3"
Filter Type pH
Fluoropore, FALP EST. 7°
Pallflex Tissue Quartz, EST. 8
2500 QAO
Gelman Batch 81886 8.9
Gelman AE 9-3
a. Data from R. Coutant, Progress Reports to EPA, Contract No.
68-02-178U.
b. By ASTM D-202 .
c. Based on value 7.0 for Mitex (Teflon) filters.
d. Based on value 8.1 for Pallflex tissue quartz QAST.
e. This filter type is designated Gel man AA in the Battelle study.
-113-
-------
Coutant's studies of artifact sttLfate formation are summarized in
Figure 22. Using this information, his measured alkalinity for
mmS
Gelman AE (h x 10" meq/g of filter or .3 ^g/cm2) and the flow rate
for the hi-vol (U.O m3/cm2), Table 49 summarized the extent of arti-
fact sulfate formation expected from the SOg levels given. Since
relative humidity data were not available, an average humidity of
6O% was assumed. Using Figure 22, an absolute error of 20% in the
average R.H. would lead to an error of ca. 0.2 jug/m3 in the estimated
sulfate error. These data indicate a sulfate error in the range
1.3-l.T Mg/m3 for 2U-hour, 8 x 10" Gelman AE hi-vol filter samples.
A shortcoming of this calculation is the reliance on SOs. determination
by the conductimetric measurement method "which is subject to substan-
tial positive interferences. Thus, use of the present S02 data may
lead to some overestimates of the expected artifact sulfate levels.
In addition, interpolation is necessary to calculate results for the
flow and alkalinity of the present samples and, for SOa concentrations
below 21 ppb, extrapolation of the curve is necessary.
-------
TABLE 1^9
CORRECTION TO BE APPLIED TO MTB SULFATE RESULTS
FOR S02 CONVERSION ON GELMAN AE 8 X 10" HI-VOL FILTERS
Artifact
Mean S02 Sulfate
Sampling Day ppb ug/m3
1 15 1.6
2 21 1.7
3 4 1.3
4 11 1.5
5 6 1.4
- 115 -
-------
Extreme
C">
-4L
60
M
O
fc
h
W
a
0)
4-1
i-l
3
CO
10
1/3
F-3, A-0.3
F-4.5, A=0.3
Presented by R.W. Coutant, A.C.S. Meeting, San Francisco,
August 1976, "Effects of Environmental Variables on the
Collection Efficiency of Atmospheric Sulfate11.
F = flow, nr/cm2
2
A = alkalinity, ueq/cm of filter
PS02 = ppb S02
R.H. = relative humidity expressed as a fraction
Figure 22. Artifact sulfate in filter sampling.
- 116 -
-------
IX. COMPARISON OF SULFATE RESULTS ON DIFFERENT FILTER MEDIA
The observed concentration of sulfate extracted from a given filter
medium reflects at least five factors:
A. The correct atmospheric sulfate level
B. Artifact sulfate formation
C. Interferents extracted from the filters
D. Interferents extracted from the particulate matter
E. Errors in flow calibration of samplers
Factor E cannot be directly measured with the data available. However,
such effects should result in a systematic bias in results for a given
filter type regardless of the analytical method used. Sulfate results
are compared by filter type in Tables 50, 51 and 32. These compile re-
sults by sampling day obtained on the different filter media using the
MDB, Brosset and BC methods, respectively. No systematic bias in results
is evident suggesting that flow calibration errors are relatively
insignificant.
To test the significance of differences in results among the different
filter types as implied by the ratios of means, a subset of samples
obtained on the same days with all filter media was tested for mean dif-
ferences and the significance of the difference by the Wilcoxon signed
rank test. The results, shown in Tables 33? *&• and 33 are consistent
with implications from the ratios of means; with the MCB method only the
126 mm quartz and "fine" Fluoropore are not significantly different at
> 95$ significance level. Similarly, with the Brosset method only the
- 117. -
-------
glass and quartz results are not significantly different. Finally, with
the BC method, none of the paired results are significantly different
at > 95$ confidence level probably reflecting the poor precision of the
method.
With the MTB method, the 126 mm glass fiber filters yield 10-15$ higher
results than the glass fiber hi-vol, quartz and Fluoropore filters.
Using the Brosset technique for the comparison, Fluoropore results are
lower by about 10$ relative to the 126 mm glass and quartz filters.
However, the latter results include only sulfate on particles < 3-5 jura-
Since "coarse" (i.e., > 3.5 jum) sulfate was below the detection limit
of 1.5 Wg/m3, the error introduced by omitting large particle sulfate
cannot be accurately assessed. Assuming a value of 0.8 yg/m3 (i.e., one-
half the detection limit) for coarse sulfate, this amount added to the
fine Fluoropore results, would decrease the ratio of means quartz/total
Fluoropore by the Brosset method to l.OU.
If this argument is valid, then Fluoropore results should also be too
low by the MEB method. However, four of the Fluoropore samples included
in this comparison were in the range 6.0-7-2 /jg/ml sulfate, just at, and
slightly below, the working range as determined in Section III. Positive
errors are likely with these samples (see Figures 1^-1?). A quartz/
Fluoropore ratio of 0.98 by the MTB method may reflect the combined in-
fluence of omission of coarse sulfate and the positive errors at low sul-
fate concentrations.
- 118 -
-------
An important application of inter-filter comparisons is to determine the
extent of artifact sulfate formation. However, the ratio 126 mm glass/
quartz of 1.1 for the MTB method but 0.9^-0.99 for the BC and Brosset
techniques is not consistent with artifact formation as the cause. Such
a source of added sulfate (or sulfite) would contribute about equally to
the three methods. These observations are, however, consistent with the
significance of analytical interferents; as shown in Table U63 the maxi-
mum error by the MTB method for the 126 mm glass fiber samples is 0.6
jug/m3 greater than with the quartz samples. Furthermore, since silicate
from the filter medium is the most significant single interferent with
the glass fiber sample compared to phosphate from the particulate matter
(which varies from day to day) for the quartz sample, the glass fiber
results should, on average, be even greater than this 0.6 jL
-------
Table 50
COMPARISON OF MTB SULFATE VALUES FOR ST. LOUIS SAMPLES
ON DIFFERENT FILTER MEDIA (jug/m3)
Sampling Day
Hi-Vol
Glass
1
2
3
k
5
6
16
18
19
7.
23-
13-
6.
22.
12.
6.
0 + .1
—
—
U + .7
1 + .1
8 + .1
7± -3
6 + .1
1 + .2
126 mm Filter
(Glass)
URBAN
8.2 + .1
11.2 + .1
23.9 ± -I
28.2 + .2
15.9 ± -2
7-2 + .03
7.6 + .01+
126 mm Filter
(Quartz)
7-6+ .3
9-7 ± .3
22.3 ± -5
27.U + .3
1U.6 ± A
6.0 + .1
28.1 + .U
1U.9 + .2
6.5 + .1
Fine
Fluoropore
(7.8)a
9-7
20.6
2k.B
111-.2
(7.9)
28.5
13-5
(7.8) 8.3
1
2
3
5
6
10
11
12
19
Ratio of means:
3.2 + .1
3-9± -1
2U.6 + .k
22.U + .6
9-7 ± .1
1».8 + .1
U.I + .1
12.6 + .1
U.8 + .2
8.3 + .1
RURAL
3-1 ± .1
3-9 + -1
25.8 + .2
25-0 + .3
10.8 + .03
5.1 + .1
13.8 + .1
15-7 + .1
5.1 + .1
11.2 + .1
2.7 ± .1
3.7 ± .1
2U.9 + .2
23.2 + .2
10.1 + .1
U.5 + .2
10.U + .1
12.1 + .1
5.1 ± .2
8.0 + .2
126 mm glass filter/glass hi-vol = 1.15 '
126 mm glass filter/quartz filter = 1.12° . .
126 mm glass filter/fine Fluoropore = l-10Cl
126 mm quartz filter/fine Fluoropore = 0.98*
Hi-vol (glass)/fine Fluoropore = 0.91
(ca. 0)
(7.2)
25.5
23.0
11.1
(7.2)
12.0
12.6
(7.6)
9-91*
Values in parentheses result from concentrations below 6 /jg/ml and are probably
too high.
Excludes days 2> 3, 16 and 18, urban samples.
CExcludes days 16 and 18, urban samples .
nSxcludes values in parentheses.
- 120 -
-------
TABLE 51
COMPARISON OF MODIFIED BROSSET SULFATE VALUES FOR ST. LOUIS
SAMPLES ON DIFFERENT FILTER MEDIA (ug/m3)
126 mm Filter 126 mm Filter "Fine"
Sampling Day Glass Quartz Fluoropore
URBAN
I 7.1 + .6 8.0 + .2 3.8
2 11.9 + 1.8 10.6 + .3 8.3
3 21.3 + 1.7 23.2 + 1.0 20.2
4 25.5 + 2.4 27.7 + .2 26.4
5 14.5 + .3 15.4 + .4 13.5
6 6.5 + .3 6.3 + .1 4.5
16 29.2 + .4 30.3
18 15.8 + .2 11.9
19 6.9 + .2 6.9 + .3 4.8 + .:
RURAL
1 1.2 + .1 1.5 + .3 <1.6
2 2.0 + .3 3.1 ± .3 2.6
3 23.1+1.6 24.4+ .7 25.4
4 22.0 + 2.0 22.5 + 1.3 22.8
5 9.9 + .2 10.2 + .4 8.8
6 3.8 + .1 3.9 + .1 3.5
10 12.5+ .4 11.0+1.8 9.5
11 14.4 + .4 11.7 + .3 10.9
12 3.7 + .3 4.4 + .1 3.4
19 9.9 + .5 8.0 + .1 6.9
Ratio of means glass/quartz = 0.99 + .03a
glass/fine Fluoropore = l.la
quartz/fine Fluoropore =1.1
"Excludes day 16 and 18, urban samples.
- 121 -
-------
TABLE 52
COMPARISON OF BC SULFATE VALUES FOR ST. LOUIS
SAMPLES ON DIFFERENT FILTER MEDIA (yg/m3)
Sampling
Day
Hi-Vol
Glass
126 mm Filter
(Glass)
126 mm Filter
(Quartz)
Fine
Fluoropore
1
2
3
4
5
6
16
18
19
7.3 +
-
-
22.8 +
13.0 +
6.7 +
21.6 +
12.6 +
6.5 +
.2
.9
.4
.4
.5
.5
.2
1
2
3
4
5
6
10
11
12
19
3.5
4.1
22.6
21.6
10.2
5.3
11.1
12.9
5.2
9.1
+ .3
+ .3
+ .8
+ 1.3
+ .1
+ .2
+ .5
+ .6
+ .2
+ .3
7.0
19.0
12.4
5.4
URBAN
± -3
± I-*
+ 1.5
± -3
11.6
9.3
22.2
5.9
25.6
15.1
+ .3
+ .6
+ 1.6
+ .1
+ 1.8
+ .9
5.1
8.2
11.9
23.9
5.6
< 4.8
27.4
14.8
5.7
RURAL
23.8 + 1.2
22.6 + .4
7.9 + .02
12.9 + .5
9.0 + 1.0
3.8 +
< 5.1
< 4.8
21.0 +
10.6 +
4.2 +
9.2 +
10.5 +
3.9 +
7.2 +
.5
.5
.4
.5
.5
.8
.1
25.1
18.7
5.8
< 4.6
< 4.8
10.2
< 4.7
15.6
126 mm glass / hi-vol (glass) = 0.98
126 mm glass / 126 mm quartz = 0.9415
= l.la»b
quartz / Fine Fluoropore
hi-vol (glass) / Fine Fluoropore
1.0*
a. The poor agreement between BC and Brosset results
for the fine Fluoropore samples suggests the BC values are
suspect. Little credence is given to these ratios.
b. Ratio of means based on 9 sampling days.
- 122 -
-------
Table 53
STATISTICAL EVALUATION OF MEAN DIFFERENCES IN SULFATE RESULTS
ON DIFFERENT FILTER MEDIA BY THE MTB METHOD
Filter Media Compared
126 glass/hi-vol glass
126 glass/126 mm quartz
i
ro 126 glass/fine Fluoropore
126 mm quartz/fine Fluoropore 8
Hi-vol glass/fine Fluoropore
No. of
Samples
8
8
8
8
8
Mean Diff . 95%
(/jg/m3) Conf. Inter v.
2.65 + 0.97
1.96 + 1.01+
1.66 ± 1.05
0.30 + 1.19
0.99 + .M*
Significance8-
of Difference
P > -99
P > -99
P > -99
P = .85
P > -99
a. By the Wilcoxon signed rank test.
-------
I •
Table
STATISTICAL EVALUATION OF MEAN DIFFERENCES IN SULFATE RESULTS
ON DIFFERENT FILTER MEDIA BY THE MODIFIED BROSSET METHOD
Filter Media Compared
126 mm glass/126 mm quartz
126 mm quartz/fine Fluoropore 16
!\3 126 mm quartz/fine Fluoropore 16
No. of
Samples
16
16
16
Mean Diff .
O.lU
1.23
1.37
95%
Conf. Inter v.
± 0.73
+ 0.96
+ 0.66
Significance8-
of Difference
P = .75
P > -99
P > -99
a. By the Wilcoxon signed rank test.
-------
Table 55
STATISTICAL EVALUATION OF MEAN DIFFERENCES IN SULFATE RESULTS
ON DIFFERENT FILTER MEDIA BY THE EC METHOD
Filter Media Compared
126 glass/hi-vol glass
126 glass/126 ram quartz
ro 126 mm quartz/fine Fluoropore 5
Hi-vol glass/fine Fluoropore
VJ1
i
No. of
Samples
5
5
5
5
Mean Diff . 95%
(jug/m3) Conf. Inter v.
•3k ± 1.50
.30 ± 3.90
1.10 + 7-22
l.lli. ± 5-^0
Significance
of Difference
P = -83
P = .71
P = .79
P = .79
a. By the Wilcoxon signed rank test.
-------
REFERENCES
1. Appel, B.R., Kothny, E.L., Hoffer, E.M. and Wesolowski, J.J.,
"Comparison of Wet Chemical and Instrumental Methods for Measuring
Airborne Sulfate, Interim Report", Contract No. EPA 68-02-1660.
Environmental Protection Technology Series.
2. Luce, E.N., et al., Ind. Eng. Chem. Anal. Ed. 15, 0. 365 (1943).
3. Dyson, W.L. and Quon, J.E., Environ. Sci. and Tech. J.O 476 (1976).
4. Pierson, W., et al., Anal. Chem. 48_ 1808 (1976).
5. TRW Report 24916-6017-RU-OO, Prepared under EPA Contract 68-02-1412
(1975).
-126-
-------
Appendix A
DIFFERENCES IN EPA AND AIHL LABORATORY PROCEDURES FOR THE
AUTO TECHNICON II MTB SULFATE METHOD*
1. Principle and Applicability
1.1 A linear regression fit of the working curve from 5-60 jug/ml
was employed for data reduction except where noted.
2. Range and Sensitivity
2.1 Using linear regression, the working range of the MTB Method
was determined to be 7-75 jug/ml on atmospheric extracts with
an accuracy of + 5$ relative to the accuracy in the optimal
range of the method.
5. Apparatus
5.2.1.1 A Technicon Auto Analyzer Sampler IV with a 3° sample/hr. cam
having a 6:1 sample to wash time ratio was used.
5.2.1A Ion - Exchange Column: A small piece of stainless steel mesh
was used as plugs at the ends of the ion-exchange column in-
stead of glass wool.
5.2.1.7 Linearizer: No linearizer was employed.
5.2.1.8 A single channel Recorder was used.
5.2.1.9 Modular Digital Printer: Not provided.
•^Section numbers used here are taken from the EMSL/RTP Draft
Auto-Technicon II Procedure, 9/19/75-
- 127 -
-------
5.2.1.10 Pump tubing: The tubing used was as given in the EMSL pro-
cedure with the exception of the pump tube from the sample
probe which was flowrated at 0.^2 ml/min instead of 0.32
ml/min.
6. Reagents (The AIHL procedures followed those given by Technicon Industrial
Method 226-72W)
6.2.12 Sodium Hydroxide solution O.l8 N instead of 0.08N.
Dissolve 7-2 g of sodium, hydroxide in distilled water and make
to 1000 ml in volumetric flask.
6.2.13 Hydrochloric Acid 10.0 N instead of l.ON.
Add 83 ml of concentrated hydrochloric acid to water in
volumetric flask and make to 100 ml.
6.2.1*1 Barium Chloride - Hydrochloric Acid Stock.
Solution: Add 16 ml of ION HCl and 1.526 g BaCl22H20 to a
volumetric flask and make up to 1000 ml with distilled water.
6.2.15 Methylthymol Blue Solution:
To 0.1182 g instead of 0.1352 g of MTB in a 500-ml volumetric
flask, add successively 25 ml of stock solution from 6.2.1^,
75 ml of distilled water and make to 500 ml with 95$ ethanol.
Prepare fresh daily.
6.2.18 Stock sulfate solution (1000 jug S04"2/inl). Dissolve 1.U79 g
anhydrous sodium sulfate in a volumetric flask and make up to
1000 ml with distilled water.
6.2.19 Blank Reagent Color Solution: Not prepared.
6.2.20 Potassium Chloride Solution: Not prepared.
-128-
-------
7. Procedure
7.2.2.1 General:
The sample turntable rate used was 30 samples per hour with a
6:1 sample to wash time ratio.
7.2.2.2 Sample and Quality Control Standard Loading:
For all standard sulfate solutions and atmospheric extracts,
two cups of solution are analyzed in sequence and only the
second cup value is considered valid. This protocol minimizes
the analytical error (memory effect) which occurs when suc-
cessive samples differ greatly in sulfate concentration
(^ > 20 jug/ml). Each day before the analysis of atmospheric
samples, standard solutions of 60, 50, Uo, 30, 20, 10, 5 jug/ml
are analyzed and the working curve compared with a historical
data base for linearity as a quality control measure. Stand-
ards are interspersed with the atmospheric samples such that
each sample set (15-20 filter extracts) can be reduced with a
separate calibration curve encompassing the range 5-60 wg/ml
S04~2. This compensates for the small drift in the baseline
(1%/k hrs.)> No correction is made for highly-colored sample
extracts because preliminary results suggest a maximum inter-
ference of 2% from this source.
7.2.3. System Maintenance:
The sample manifold was equipped with a column by-pass valve
which removes the ion-exchange column from the flow circuit
during washing (see figure) and permits removal of air bubbles
during the start-up operation.
-129-
-------
8. Calibration
8. P.I Flow Jtetes: Degradation in flowrated pump tubing is monitored
daily by evaluation of the calibration curve per 7-2.2.2.
8.2.3 Concentration Standards: Transfer about 50 ml of the 1000
/og/ml standard sulfate solution into a smaller container which
allows easy access with a digital pipet. With a 5 ml digital
pipet, transfer the following volumes of the 1000 jug/ml stand-
ard into 100 ml volumetric flasks and fill to the mark with
distilled water:
Standard /jg/ml mis of
(100 ml Volumetric) 1000 jug/ml Stock
5 0.5
10 1.0
20 2.0
30 3-0
lj-0 h.O
50 5.0
60 6.0
8.2.6 Baseline Adjustment: No baseline adjustment is performed since
standards are interspersed with the samples.
8.2.7 Calibration Standards: The set of standards run at the be-
ginning of the day is compared with the standards interspersed
in each sample set of the day. If the variation between sets
of standards is more than 3 Mg/ral? that run is considered in-
valid and must be repeated.
-130-
-------
Sampler IV/"40/hr. 2:1 ^/4.
—t
IGH EXCHANGE COL.
116-GOO'S-Ol _ ,
170-0103-01
A2
5 turns 116-0^39-01
BLK/BLK(0.32) A!R. I
JHUL/i2ii(^LJ)II)J!AT.EJl.|
PRC3E
O.llp Std.Sleeving
| Column b^-plSS
-I ^
^
XB. Y / n H Y ( LJ?.OJl'M T £ .
?
QJXB. Y / n
157-0370
-Q-3 LK AB.LK.( 0.3?). AIR
22 turns'^
Silicon
116-0483-01 ^^ REO/RED(Q.7Q) METHLTHYKOL BLUE
.rx
WASTE
IN(0.42) SODIUM HY03QX
.11 icon
MOTE: l)F1gurcs In parenthesis signify flow rare
(snl/mln).
2)P'jr.p tubes are tygon unless othervMso narked.
Figure A-1. Sulfate in water and waste water (range 0-100 jug/ml
-------
Appendix B
PROTOCOL FOR DETERMINING WORKING RANGE OF THE MTB METHOD
1. Linearity of working curve and reproducibility
Determine, on a single day, the working curve with three separate
determinations each with the concentrations 75? 70, 65, 60, 50> ^0>
30, 20, 10, 8, 6, h jug/ffll sulfate. For each determination use two
cups at each level, but only report the second cup value.
Trial A
stepwise change in concentration
Trial B
random order of concentrations
Trial C
a different random order
Data Display
A. Construct a single curve plotting the mean instrument response
(chart units) + 2 or for three determinations at each concentration
level listed above. Determine the range in concentration which
provides the best linear regression fit as a function of scatter
(Sy.x) and plot the least squares line.
B. Plot the variance (or2), in recorder chart units, against the true
jig/ml sulfate.
2. Precision and accuracy with sulfate from atmospheric samples
Using an extract from a St. Louis Hi-vol sample with an initial S04~
concentration of > 85 Mg/rol prepare dilutions at the same concentrations
- 132 -
-------
shown in (1). Analyze with three determinations using separate working
curves. Determine mean and a for each determination in Mg/ml employing
a linear regression fit of the working curve. Assume that the calcu-
lated undiluted concentration previously obtained from analysis of the
hi-vol extract in the optimal range of the method (ca. 35 ug/ml) is the
"expected" concentration. Calculate the ratios observed/expected
concentrations.
All analyses are performed by the same chemist.
Data Display
A. Plot observed/expected concentration as a function of the jug/ml
when analyzed.
B. Plot the variance calculated for the diluted solutions against the
observed iUg/ml when analyzed.
- 133
-------
Appendix C
The Modified Brosset Method for Sulfate Analysis '
1. Principle
1.1 Aqueous extracts of particulate matter are freed from cationic
interferences "by contacting the solution with a strong acid ion
exchange resin. Sulfates are converted into sulfuric acid.
1.2 The dilute sulfuric acid solutions are mixed with an alcoholic
solution containing excess barium. Sulfuric acid reacts with
barium which precipitates as insoluble BaS04.
1.3 Excess barium is reacted with Thorin (2-(OH)2AsOC6H4N:N-l-C10H4-
2-OH-3j6-(S03Na)2) and the colloidal complex is measured at
520 nm in a spectrophotometer.
2. Range and Sensitivity
2.1 The concentrations of the reagents are selected to cover a range
from 0 to 12 /jg/ml. The useful range is approximately 3 to
10
2.2 The limit of detection is about 0.03
Modifications based upon procedures employed by Dr. J. Stikeleather,
Northrup Service, under contract to the EPA.
•frfr
The working range, defined as the region of constant variance, is to
be evaluated as part of the present AIHL study.
-------
3- Interferences
3.1 Cationic interferences in solution are eliminated with the ion
exchange treatment. However some cations like Ba, Sr, Ca, Fe,
Pb may react with sulfate prior to ion exchange, thus producing
a negative interference. The interference for the elements listed,
except Ba, was found to be less than % for sulfate;interferent
ratios of 2:1 w/w and 0.7:1 w/w when using acetone as sol vent .^^
3-2 Anionic interferents, such as phosphate, may react with the
barium thus producing a negative interference. The interference
was found to be less than 10$ under the conditions given in 3.1.
3.3 Colloidal clay and a yellow organic material caused generally
less than a 10$ interference effect when using acetone as the
solvent.
h. Precision and Accuracy
lj-,1 Under careful conditions the reproducibility of the method is
within +0.2 u,g/til.
k.2. The accuracy of the method in isopropyl alcohol was evaluated
with aqueous extracts of filter strips loaded with known quantities
of K2S04. On average the results differed by 1$ from those reported
by EPA. The accuracy of the method using extracts from atmospheric
samples has not been evaluated with this solvent. In acetone,
recoveries of sulfate were typically 100-110$ with standard
additions to atmospheric particulate extracts. However such
recoveries may not correlate with those attainable from atmospheric
samples.
7Wnirlnterference effects using isopropyl alcohol as solvent are being
evaluated as part of the present AIHL study.
- 135 -
-------
5. Apparatus
5.1 Ion exchange columns 200 mm long and 5 to 6 mm in diameter or
similar such as those described in AOAC 6th Ed. 19^5? P- 609
available from Kontes as Chromaflex columns cat. no. K^-20150.
5-2 Automatic pipet. Consists of a three syringe system, operated
pneumatically on a two cycle loading and unloading sequence.
The cycles are triggered by a foot operated air release valve.
Each syringe has a stop on its upper portion which can be set
with an hexagonal Allen wrench after releasing the lock pin
which holds the stop in place. The setting screw is barreled
and has four vernier divisions, each of which corresponds to
0.1 ml. One syringe, the sample syringe, pulls liquid from a
tip through a long tubing which holds the sample solution. The
second cycle pushes the sample through the same'sample inflow
opening. The second syringe, pulls the diluent out of a container
and empties it into the sample syringe, thus flushing its content
through the same tubing as the sample and pushes its content
through the same sample inflow opening. The third syringe is
independent and pulls the Thorin reagent solution out of a
container and then pushes it through a separate delivering port.
Both delivery ports are positioned together, which allows a
simultaneous delivery into a single spectrophotometric cell.
5.3 Spectrophotometric cells. Square 20 mm cells are preferable if
adapters for thespectrophotometers are available. Cylindrical
25 nun pathlength cells can be used, provided that they are kept
clean and free of bubbles.
-136-
-------
5.4 Spectrophotometer. 2 nm slit width with output for recorder or
voltmeter.
5.5 Digital Voltmeter. 5-1/2 digits, capable of reading the absorbance
output of the Spectrophotometer such as Fluke Model 8800 A.
6. Reagents
6.1 Air or Nitrogen. The Autopipet requires 65+5 psi air pressure.
If not available from the laboratory compressed air line, cylinder
air or nitrogen may be used.
6.2 Distilled water. Water distilled from a very dilute KMn04 solution
in an all glass system is preferable.
6.3 Perchloric acid, 0.1 M, as alternate to distilled H20 for extracting
sulfate from particulate samples.
6.4 Ion exchange resin. Type Dowex 50W X8 hydrogen form 50-100 mesh
or equivalent such as BioRad AG 50W X8, hydrogen form 50-100 mesh.
6.5 Sulfate stock, 1000 ppm S04~. Dilute 10.42 ml of commercial 1 N
H2S04 to 500 ml with distilled water ( solution is 0.0104 M H2S04).
6.6 Barium stock. Dissolve 550 mg anhydrous Ba(Cl04)2 in 6 ml 72$
HC104 and fill to 250 ml with distilled water.
6.7 Diluent. Take 10 ml of barium stock and fill to 1 liter with
isopropanol.
6.8 Thorin reagent. Dissolve 200 mg Thorin in 7 ml 0.01 N H2S04
(or 3.5 ml sulfate stock, 1000 ppm S04~) and fill to 250 ml with
distilled water. This solution is stable for four weeks if stored
in a brown bottle.
-137-
-------
7. Procedure
7.1 The extraction technique varies with the sample size. Either
distilled water or 0.1 M HC104 may be used. References 3 and k
discuss extraction methods. Depending on the extraction technique,
extracts may be filtered after cooling.
7.2 Ion exchange treatment.
7.2.1 Resin preparation. Stir resin into water in a beaker. Discard
the fines by decanting several times. Soak for one or two hours
and renew the water a few times. Then let soak overnight and
fill the columns.
7.2.2 Column preparation. Fill ion exchange columns to 12-13 cm height
with resin. The tip of the column is plugged from the inside
with a small glass wool swab before filling the resin. Attach
a piece of rubber tubing and a pinch-off clamp to the tip for
flow control. Fill with aid of a Pasteur pipet slurrying the
resin in a beaker. Avoid air entrapment. Flush the columns with
distilled water before use and keep covered with water. If air
gets entrapped, re-make the column, flush water from the tip
up to the reservoir, or stir with a Pasteur pipet containing
water from the top down. Cover with a plastic foil. Renew the
columns after 20 cycles.
7.2.3 Procedure. Drain off the distilled water from the reservoir
through the column. Add 10 ml fresh distilled water and flush
through. Just before air reaches the top of the resin bed, add
5 ml of sample extract. Discard this first portion and add a
second 5 ml portion of the sample extract. Collect this fraction
in a 25 ml beaker or Erlenmeyer. Cover after collection. Flush
-138-
-------
the column with 10 ml of distilled water before using it for
a new sample. For storage, flush with about 10 ml distilled
water, then fill the reservoir, pinch-off the flow and cover
with a plastic foil.
7-3 Analysis
7.3-1 Operation of the automatic pipet (see 5.3)
The syringe delivering the sample is set at 1.8 ml.
The syringe delivering the diluent is set at U.8 ml.
The syringe delivering the Thorin reagent is set at O.k ml.
Total volume = 7.0 ml.
The volume of 7 ml is equal to the capacity of the 25 mm pathlength
cylindrical cell and can also be used with the 20 mm square cells.
If other volumes are necessary or desired, the formulations of
the individual reagent solutions must be modified accordingly.
Proportional amounts of Ba, Thorin and sulfate, as well as the
same ratio of water to alcohol must be maintained.
7.3.2 Adjustment of the spectrophotometer.
When using a double beam instrument, one beam must be balanced
by inserting a grey wedge, filter or solution, at the operational
wavelength, i.e. 520 nm.
7.3.3 Zeroing of the spectrophotometer. Operate the filling mechanism
of the automatic pipet first using distilled water a few times,
until stabilized. Set the digital voltmeter to read 0.800
absorbance units. With intervals of 5 minutes make several blank
measurements, and record the readings. Drifting should be less
than 0.5$ or O.OOU.
-139-
-------
7-3.4 Measurement procedure. Submerge the syringe port tubing into
the sample beaker. Press foot pedal and wait until the sample is
pulled into the long holding plastic tube. Withdraw the beaker
and place the ceH under the syringe port tubings. Eress foot
pedal and wait until all the three syringes have delivered their
contents. Cap well, shake to free bubbles, insert in spectro-
photometer, wait until stabilized and read at 520 nm.
7.3.5 Measurement. Samples are run with two determinations within
the working range of the method. If the initial value exceeds
the range, use twofold dilutions using a repetitive pipet and
repeat the measurements. If dilutions of the extracts are
necessary these should be made following the ion exchange
treatment. Continue diluting until readings fall in the working
range.
8. Calibration
8.1 Standards. With a 1 ml repetitive pipet measure 0.3, 0.4, 0.5,
0.6, 0.7, 0.8, 0.9 and 1.0 ml of the 1000 ppm sulfate stock
solution into 100 ml volumetric flasks and fill to mark.
8.2 Quality control.
8.2.1 After a days run, clean the quartz cell with alkaline acetone
(one drop 5 M ammonia, a shot of acetone from a squeeze bottle),
cap the cell and shake. Add a little water and continue shaking,
then rinse a few times and drain. An orange deposit may form
occasionally which is a Ba-Thorin insoluble salt.
-------
8.2.2 Run new standards in triplicate when making up fresh reagent,
when changing solvents and every day before analyzing samples.
Check the calibration line; points should be almost co-linear
between 3 and 10 jug/ml if standards have been properly prepared
and if spurious sulfate introduced by reagents is not excessive.
8.3 Calibration. Calibrate the instrument running standards in
duplicate or triplicate within the range of 3 to 10 jug/ml. If
desired, duplicates can be run before and after measuring sample
solutions.
9- Calculation
9.1 Calculate a least square regression line with the data obtained
by running the standard calibrating solutions. If standards
are run both before and after the samples, pool all data for a
single regression line.
9-2 Concentrations of the sample are calculated in jug/ml using the
regression line obtained above.
9.3 Multiply the concentration values by the dilution factor in order
to obtain original extract concentrations. Multiply these values
by the extract volume to obtain the amount of sulfate extracted
from the filter samples.
10. Storage effects. Sulfate standards and samples are stable for
several weeks at room temperature, and up to a year if stored
in a refrigerator and protected from light.
-------
11. References
1. Brosset C and Ferm M: An improved spectrophotometric method
for the determination of low sulfate concentrations in aqueous
solutions. To "be published in Atmospheric Environment.
2. Air Quality-sulfur dioxide concentrations in air- analysis by
the thorin (spectrophotometric) method. International
Organization for Normalization, 1974-05-15, ISO/TC 146/WG
1/TG ltfO.8.
3. Hoffer EM and Appel BE: A comparative study of extraction
methods for sulfate and nitrate from atmospheric particulate
matter. A1HL Report No. 181, Air and Industrial Hygiene
Laboratory, California State Department of Health, November 1975-
k. Jutze GA and Foster KE: Recommended standard method for atmos-
pheric sampling of fine particulate matter by filter media—
high-volume sampler. JAPCA 1J_ 17 (1967).
-1U2-
-------
Appendix D
BARIUM CHLORANILATE METHOD FOR DETERMINATION
OF SULFATES IN THE ATMOSPHERE!/
March 1976
U.S. Environmental Protection Agency
Environmental Monitoring and Support Laboratory
Research Triangle Park, North Carolina 27711
&/ This method has been carefully drafted from available experimental
information. The method is still under investigation and therefore,
Is subject to revisions.
-------
BARIUM CHLORANILATE METHOD FOR DETERMINATION
OF SULFATES IN THE ATMOSPHERE
1. Principle and Applicability
1.1 Ambient sulfates are collected by drawing air through a glass fi-
ber filter with a high-volume pump. The filters are extracted with water
and the extract treated with excess barium chloranilate. ' The released
chloranilic acid equivalent to the sulfate content of the sample is then
measured at a pH of 2.0^ ' spectrophotometrically at 312 run. If the absor-
bance is too high the absorbance may be measured at 530 nm without dilut-
ing the sample.
1.2 The method is applicable to the collection of 24-hr samples in
the field and subsequent analysis in the laboratory.
2. Range and Sensitivity
2.1 The range of the analysis at 312 nm is 1 to 60 ug SO^/ml. By
using the 530 nm absorption peak, the range may be extended to 1,500 ug/ral.
With a 50-ml extract from 1/12 of the exposed high volume filter collected
at a sampling rate of 1.7 m-Vmin (60 cfm) for 24 hr, the range of the method
is 0.2 to 300 ug/tn . The lower range may be extended up to 12-fold by
increasing the portion of the filter extracted.
2.2 Using the procedure outlined, a concentration of 1.6 ug/ml will
produce an absorbance of 0.02 at 312 ran.
3. Interferences
3.1 Water soluble chlorides, fluorides and phosphates produce a
positive interference which is dependent on the concentrations.(3) Chloride
at 20 times the sulfate concentration produces a positive error of o. 10%.
Fluorides produce only a slight interference except when present as fluosil-
icate. The absorbance produced by 50 ug of fluoride as fluosilicate is
0.03, The absorbance produced by 100 ug of phosphate is equal to 0.01.
3.2 The interferences from water soluble cations are removed by
contacting the sample with a hydrogen form ion-exchange resin. Cations
such as Ca+2 or Pb , which form insoluble sulfates and which are present
in concentrations which exceed the solubility product of the respective
compounds cause a negative interference.
-------
4. Precision and Accuracy
A.I A single laboratory's relative standard deviation for the anal-
ysis is 2.5%.(3> Overall precision is not presently known.
4.2 Adequate data for accuracy determinations is not presently
available.
5. Apparatus
5.1 Sampling; Apparatus as specified in Appendix B - "Reference
Method for the Determination of Suspended Particulates in the Atmosphere
(High Volume Method),"(4) shall be used.
5.2 Analysis
5.2.1 Spectrophotometer; Capable of measuring absorbance at
312 and 530 ntn.
5.2.2 Spectrophotometer Cells; Matched set of cells with a
10 mm path length constructed of high silica material transparent in the
ultraviolet to visible reagon (165 to 2,600 nm).
•
5.2.3 Filter Paper; Whatman No. 4 or equivalent 55 mm in diam-
eter.
5.2.4 Filter Paper; Whatman No. 42 or equivalent 55 mm in di-
ameter.
5.2.5 pH Meter; Capable of measuring the pH to nearest 0.1 pH
units over a range of 0 to 14.
5.2.6 Mechanical Shaker: Capacity for shaking the required
number of samples. For a small number of samples a magnetic stirrer may
be used in place of the shaker.
5.2.7 Erlenmever Flask; 125 ml with 24/40$ joint.
5.2.8 Condenser; Water jacketed, 300 mm length with 5 24/40
joints.
5.2.9 Hot Plate; Suitable for sample extraction (7.21).
5.2.10 Volumetric Flasks; Class A - 50, 100, 500, 1,000 ml
capacity.
5.2.11 pjpets; Class A-l, 5, 10, 20, 50 ml volumetric; 10 ml
graduated in 1/10 mi intervals.
-------
5.2.12 Buchncr Funnels: Buchner style 150 ml capacity with fine-
pore fritted glass filter.
5.2.13 Buchncr Funnels: Buchner style with a perforated plate
and 150 ml capacity for 55 mm filter paper.
5.2.14 Vacuum Filtering Apparatus: Device which permits
vacuum filtering directly into receiver. This consists of a bell jar with
a top opening, a side tubulation and a bottom plate. The Buchner funnel
passes through the rop opening and is sealed to the bell jar with a stopper.
The bell jar should be tall enough to contain the graduated tubes used for
collecting the samples. The vacuum connection is made using the side tubu-
lation. The filtering apparatus is shown in Figure D-l.
5.2.15 Vacuum Pump: Any device which can maintain a vacuum of
at least 64 cm of Hg. Mechanical pumps or water aspirators may be used.
5.2.16 Polythylene Bottles: Bottles with a capacity of 60 ml
(2 oz) fitted with polyseal caps.
5.2.17 Standard Scoop: Spatula with small (1/8 x 1/2 in.) spoon
on one end. Practice with barium chloranilate and an analytical balance so
that one scoop of approximately 25 mg can be measured out.
6. Reagents
6.1 Sampling
6.1.1 Filter Media: Filter media as specified in Appendix B -
"Reference Method for Determination of Suspended Particulates in the
Atmosphere (High Volume Method),"^) shall be used. Each lot of filter
should be analyzed for background sulfate content and pH using a statis-
tically valid sample from that lot.
6.1.1.1 Determination of Filter pH; Cut a 9-in? (58 cm2)
section of a glass fiber filter with pizza cutter. Place the filter in a
125-ml Erlenmeyer flask. Add 15 ml of 0.05 M KCl and stopper the flask.
Stir with a magnetic stirrer for 10 min at 60 RPM. Determine the pH of
the extract.
Obtain the pH for a given lot of filters, and report the
mean and standard deviation. pH has an effect on accuracy of the collection
procedure. The optimal pH value of the filter extract is not presently
known but pH information will be useful for historical purposes. Filters
currently used in the National Air Sampling Network (NASN) have a pH of
9.74 + 0.89.
-------
6tl>1'2 Determination of Filter S04 Content: Measure the
sulfate content of each lot of filters. Cut a 3 x 8 in. (7.6 x 20.3 cm)
strip from each filter using a pizza cutter and a template. Follow the
procedures for extraction and analysis given in Sections 7.2.1 and 7.2.2.
Calculate the mean and standard deviation in ug SO,/in.2 . The SO, content
Of glass fiber filters may vary significantly from lot to lot. A low S04
content is desirable, but it is more important that the value be constant
within a given lot.
6.2 Analysis
6.2.1 Sodium Hydroxide: ACS Reagent Grade.
6.2.2 Barium Chloranilate; Trihydrate Reagent Grade. The ma-
terial must, be crystalline; the amorphorus material forms a colloidal sus-
pension which is difficult to remove. The product supplied by J. T. Baker
Chemical Company has been found acceptable.
6.2.3 Cation Exchange Resin: Dowex 50W-X8, hydrogen form, or
equivalent, 300 to 850 pm (20 to 50 mesh). The resin should be stirred
into distilled water and the fines discarded before they can settle. The
resin should be in a fully swollen condition before use. After soaking,
remove excess water by filtering with section and pressing between sheets
of filter paper.
6.2.4 Isopropyl Alcohol; ACS Reagent Grade.
6.2.5 Chloroacetic Acid: Minimum purity 997.; m.p. 61 to 62°C.
6.2.6 Sodium Sulfate; ACS Reagent Grade, anhydrous.
6.2.7 Potassium Chloride; ACS Reagent Grade.
6.2.8 Sodium Hydroxide Solution (1.0 N): Dissolve 20.0 g, of
sodium hydroxide in distilled water and make to 500 ml in a volumetric
flask, transfer to a polyethylene bottle.
6.2.9 Bufft nif-2.0: Dissolve 18.9 g of chloroacetic acid in
50 ml of distilledTwater. Adjust the pH to 2.0 by adding 1.0 N sodium
hydroxide solution. Make the solution to 100 ml and recheck the pH.
6.2.10 Stock Sulfate Solution (1,000 ug SO^/ml): Dissolve
1.4789 g of sodium sulfate (Na2S04), which has been heated at 105°C for
a minimum of 4 hr and cooled in a desiccator over anhydrous magnesium
perchlorate, and dilute to 1,000 ml with distilled water. Store under
refrigeration.
-------
6.2.11 Distilled Water; ACS Reagent Grade, having a specific
conductance of 2 microhms or less.
6.2.12 Potassium Chloride Solution (0.05 M): Dissolve 3.7 g of
KC1 in 1,000 ml of C02 free distilled water. The pH of this solution
should be 7.0+0.3.
7. Procedure
*
7.1 Sampling: Sampling procedure as specified in Appendix B - "Ref-
erence Method for the Determination of Suspended Particulate in the Atmos-
phere (High Volume Method),"(^) shall be used. Quality Assurance Guidelines
for use with the High Volume Method are applicable to the collection of
samples for sulfate determination.^'
7.2 Analysis
7.2.1 Sample Extraction: Remove the filter from the folder,
open flat, and cut a 3/4 x 8 in. (1.9 x 20.3 cm) strip using a pizza
cutter and filter cutting template. The filter should be cut with the
particulates face up. The filter strip is folded and placed in a 125 ml
Erlenmeyer flask. Add 35 ml of distilled water to the flask and connect to
a 300 mm water jacketed condenser. Place the flask condenser assembly on
a hot plate and boil gently for 30 min. Maintain cold water circulation
through .the condenser while the sample cools to room temperature. Rinse
the walls of the condenser with 5 ml of distilled water and disconnect
the flask. Decant the liquid in the flask directly into the Buchner
funnel of the filtering apparatus and filter into a glass graduated tube
with a 50 ml graduation mark. Rinse the filter in the flask with a 5 ml
portion of distilled water and add the rinse to the funnel. Squeeze the
filter with a glass rod to remove the remaining extract and collect the
filtrate. Repeat the rinse with a second 5 ml; portion of distilled water.
Collect the filtrate and dilute to a volume of 50 ml with distilled water.
Transfer the sample to a 60 ml (2 oz) polyethylene bottle and cap with a
polyseal cap. Mix thoroughly. These samples are stable at room tempera-
ture for at least 2 weeks.
A random 5 to 10% of the filters should be extracted in duplicate
for purposes of qualifying the precision of measurement.
7.2.2 Sample Analysis; Transfer 20 ml of the sample to a 60 ml
plastic bottle and add 1.5 g of ion exchange resin. Prepare a reagent blank
in the same manner using 20 ml of distilled water. Seal the bottle with a
polyseal cap and agitate for 15 min using either a mechanical shaker or mag-
netic stirrcr. Vacuum filter the mixture using a Whatman No. 4 or equivalent
- 1U8 -
-------
filter paper. Do not wash. Pipet a 2-ml aliquot of the supernatant solu-
tion into a 60 ml plastic bottle containing 16 ml of isopropyl alcohol and
mix. Transfer a fraction of the solution to the spectrophotometer cell and
measure the absorbance at 312 nm against 80% isopropyl alcohol. Record the
absorbance as the sample blank. Add one scoop of barium chloranilate to
the solution remaining in the plastic bottle, seal the bottle and agitate
mechanically for 15 min. Vacuum filter the mixture using a Whatman No. 42
or equivalent filter paper. Add 2 ml of buffer, mix and transfer the sol-
utions to the spectrophotometer cell and measure the absorbance at 312 nm
against 807, isopropyl alcohol. If the absorbance is too high for practi-
cal measurement at this wavelength, read at 530 nm. The amount of barium
chloranilate specified is sufficient for samples containing up to 400 ug
SO^/ml.
8. Calibration
8.1 High Volume Sampler; The high volume air samplers shall be cali-
brated as specified in Appendix B - "Reference Method for the Determination
of Suspended Particulates in the Atmosphere (High Volume Method)."^4^
8.2 Calibration Curve: Dilute 50.0 ml of stock sulfate solution con-
taining 1,000 ug SO^/ml to 500 ml with distilled water. This intermediate
sulfate solution contains 100 ug SO^/ml. Pipet 5, 10, 10, 15, 20, 50, and
60 ml of the 100 ug S0=/ml solution into 100, 100, 50, 50, 50, 100, and 100
ml volumetric flasks and dilute to the mark with distilled water. These
solutions contain 5, 10, 20, 30, 40, 50, and 60 ug SO^/ral, respectively.
If samples contain sulfate concentration greater than 60 ug/ml, anal-
ysis is possible (without dilution) by measuring at 530 nm. This will re-
quire preparing a calibration curve for the 530 wavelength. Pipet 5, 10,
10, 15, 20, 50, and 50 ml of the 1,000 ug soj/ml stock solutions into 100,
100, 50, 50, 50, 100, and 100 ml volumetric flask and dilute to the mark
with distilled water. These solutions contain 50, 100, 200, 300, 400,
500, and 600 ug SO^/ml respectively.
Pipet 2 ml of the standard to a test tube and add 16 ml of isopropyl
alcohol and one scoop of barium chloranilate. Agitate mechanically for
15 min and vacuum filter using a Whatman No. 42 or equivalent filter paper.
A blank consisting of 2 ml of distilled water should be included with the
calibration series. Add 2 ml of buffer, mix and transfer the solution to
a 1 cm cell and read the absorbance at 312 nm against 80% isopropyl alco-
hol. Subtract the blank absorbance from 'chc standard absorbance and plot
net absorbance versus ug S0£/ml. A straight line with a slope of 0.29 +
0.1 absorbance units/ng SOj/ml, passing through the origin, should be ob-
tained.
-------
9, Calculations
9.1 Air Volume; The volume of air sampled shall be calculated accord-
ing to Section (9.2.2) of Reference (3).
9.2 Sulfntc Concentration: Add the absorbance of the reagent blank
and the sample blank and subtract from the absorbance of the sample. Using
the calibration curve from 8.2, calculate the net ug S07 in the sample as
follows:
9.2.1 ug SO" From Sample Extract (7.2.1)
Jig SO/gv - tig SO/ml x 50 ml
where
U8 SOTv^v = micrograms of sulfate in the sample extract
Ug SO^/ml = analyzed micrograms of sulfate from sample (7.2.2)
50 ml = total volume of extract
•
9.2.2 ug SO^ From Filter Content (6.1.1.2)
ug SO^Wpv = ug SO^ m2 - filter x in2 - filter used
where
Ug SOv = micrograms of sulfate from filter
e 0
Ug SOA/in - filter = micrograms of sulfate in each square inch
of filter (6.1.1.2)
2
in filter used » square inches of filter used for the analysis,
usually 6 (3/4 in. x 8 in.)
- 150 -
-------
9'3 Calculate Concentration of Ambient Sulfatcs
where
C » concentration of ambient sulfates, ug/ra3
12 «= total width of exposed filter -=- width of strip used (9 in.-r 3/4 in.)
N « net ug S0°, ug from (9.2)
V « total volume sampled, ra3 from (9.1)
10. References
1. Bertolacini, R. J., and J. E. Barney, II, "Colorimetric Determination of
Sulfate with Barium Chloranilate," Anal. Chem.. 29_: 281-283 (1957).
2. Bertolacini, R. J. , and J. E. Barney, II, "Ultraviolet Spectrophotometric
Determination of Sulfate, Chloride, and Fluoride with Chloranilic Acid,"
Anal. Chem.. 3^:202-205 (1958).
3. Schafer, H. N. S., "An Improved Spectrophotometric Method for the Deter-
mination of Sulfate with Barium Chloranilate as Applied to Coal Ash and
Related Materials," Anal. Chem.. 39j 1719-1726 (1967).
4. Appendix B - "Reference Method for the Determination of Suspended Par-
ticulates in the Atmosphere (High Volume Method)," Federal Register,
.36(84) :8191-8194, April 30, 1971.
5. "Guidelines for Development of a Quality Assurance Program - Reference
Method for the Determination of Suspended Particulates in the Atmos-
phere (High Volume Method," EPA Environmental Monitoring Series,
EPA-R4-73-028b, June 1973.
- 151-
-------
BUCHNER FUNNEL
WITH A PERFORATED
OR FRITTED DISC
TO VACUUM
PUMP
GRADUATED TUBE
BEAKER
•BELL JAR
BASE PLATE
Figure D-l Vacuum filtering apparatus,
- 152 -
-------
Appendix E
AIHL Procedure for the EPA-MRI Barium Chloranilate Method
1. Transfer 20 ml of each sample or standard solution (standards cover the
range between 6 and 60 jug/ml) and a convenient number of water zeroes
(generally three) into 2 oz. plastic containers. Add one scoop (about 1.5 g)
of prewashed, slightly damp ion exchange resin to each container, cap and
shake on a shaker for 20 minutes.
2. Take 2 ml of the supernatant and place in a new set of 2 oz. plastic con-
tainers containing 16 ml of isopropyl alcohol. The alcohol is delivered
from a repetitive pipet shortly before (no solution should be stored in the
containers for more than four hours).
3. Measure the blank value of the mixture of sample plus alcohol at 312 nm.
Eeturn the liquid from the measuring cell back into the container.
U. Add a scoopful (about 15-20 mg) of barium chloranilate to the containers
with the alcoholic solution. Put on a shaker for 20 minutes.
5. Filter a portion (about one-third) through a fine frit funnel unwashed from
the previous sample. Discard. Pour in remainder and collect. Measure 9 ml
from the filtrate with a glass pipet. Scrape and rinse the funnel with IPA
after every four solutions.
6. Put the 9 ml in a glass tube containing 1 ml of buffer, mix and measure
absorbance at 312 nm in the same spectrophotometrie cell used above for
blanks. Do all measurements against distilled water in the reference beam.
?. Deduct all blank readings from the corresponding sample, standard or zero
readings.
8. Construct a calibration curve including the three values for the zero and
calculate the least squares line. Calculate concentrations with slope and
•Intercept for each day's run.
- 153 -
-------
APPENDIX F
Determination of Reactive Silicate
Reactive silicates are those forms of silicate which react with molybdic
acid to give the silicomolybdate complex. Generally, only monomeric or
dimeric silicates react whereas higher polymers do not. Silicates easily
depolymerize at higher pH. Since glass fiber filters are believed to be
the dominant source of silicate in the current program, and these are made
out of relatively alkaline compositions, it is assumed that most of the
soluble silicate in the extracts is reactive.
Critical step in the method is the formation of silicomolybdate which
involves a rapid adjustment of the acidity of the sample in the presence
of molybdic acid. This is accomplished by adding the sample extract to a
mixture of molybdic acid and hydrochloric acid (and not vice versa). After
10 to 15 minutes, the reaction is assumed to be essentially complete. Once
the silicomolybdate complex has formed, it can be either measured directly
or changed to molybdenum blue and measured at ca. 800 nm. The silicomolybdate
complex absorbs at 320/330 nm, but molybdic acid also absorbs strongly in
this region. This overlap makes the determination imprecise especially if
a tungsten filament light source is used. Therefore, conversion to the
blue compound is preferable.
The method used is based on information detailed in the book "Fisheries
Research Board of Canada, Bulletin No. 125, 2nd Ed. 1965: A Manual of
Sea Water Analysis, by JDH Strickland and TR Parsons." Modifications,
discussed here were made in the reducing agent and volumes of reagents
and samples.
VW Truesdale and CJ Smith: The Spectrophotometric Characteristics
of Aqueous Solutions of a + p Molybdosilicic Acids. Analyst 100, 797-8059
1975.
-------
Sensitivity for the blue compound is about 2 to 3 times higher than for the
species read in the w.
The reducing substance recommended in the paper is p-methylaminophenol.
Since this compound was not available to us, we tried several other reducing
compounds generally used for molybdate blue methods. In our trials, h com-
pounds tested gave essentially the same results:
Absorbance at 810 nm
with 1 ppm Si02
p-aminophenol (most similar to the 0.58
substance recommended)
ascorbic acid 0.5^
1, 2, if aminonapthol sulfonic acid 0.5^
o-phenylenediamine 0.56
hydrazine 0.^3
In all cases except with hydrazine, sodium bisulfite was added as a pre-
servative. Hydrazine was eliminated because of the slowness of the reaction.
In most methods using hydrazine, this substance is heated to hasten reaction.
AH the organic substances except ascorbic acid were eliminated because of
their yellow color, either before or after the reaction. A stock solution
containing ascorbic acid and bisulfite in equal amounts has been found to
remain stable for about one month. This same reducing substance was also
p
suggested by Murphy and Riley for the analysis of phosphate for reduction
of phosphomolybdic acid to the corresponding Mo-blue. Other modifications
to the original method involved proportionate reductions of volumes to scale
the method to the samples analyzed.
2Murphy and JP Riley: A modified single solution method for the determination
of phosphate in natural waters. Anal. Chim. Acta, 27(1962), 31-36.
- 155 -
-------
Method for the Determination of Reactive Silicate in
Samples of
Aqueous Extracts from Atmospheric Particulate Matter
Reagent A: 80 ml of 2% ammonium molybdate solution
9 ml of 6 M hydrochloric acid
Make up to 100 ml. Hie solution is usable indefinitely unless
a precipitate is seen.
Reagent B: In a 100 ml stoppered cylinder mix:
55 ml of 5 IT H2S04 (70 ml cone. HeS04 to 500 ml with water)
15 ml of saturated oxalic acid solution (approx. 10$)
25 ml of 1$ ascorbic acid containing 1% of sodium bisulfite
Make up to 100 ml. Discard after 2h hours.
Procedure: In 10 ml Erlenmeyer flasks, place 1 ml of reagent A and add 5 ml
of sample solution. Let react for 15+5 min.
Add k ml of reagent B and let react for one hour + 15 min.
Measure at 810 nm against a blank made with distilled water.
Compare with a calibration curve made with standards containing
a maximum of 2 ppm S±02.
Standard: A 100 ppm Si02 solution is made by diluting h.6j ml of a 1000
jug/ml Si standard (commercially available).
Range: With a 20 mm (or 25 mm) cell, the range is 0.2 to 2 jug/ml Si02.
With a 10 mm cell, the range is 0.5 to if Mg/ml Si02.
With a 1 mm cell (10 mm cell + 9 mm spacer), the range can be
extended to kO MS/ml Si02«
Precision: The precision is better than 5$ at all levels, when standards
are run concurrently with the samples.
Working Figures A-l and A-2 show the working curves used for silicate
Curve: prepared from standards run before and after the samples.
Where only one point is visible, the two determinations were
indistinguishable.
- 156 -
-------
1.1
1.0
.9
.8
S
i
Date Analyzed: 4-8-76
.4
.3
.2
.1
a
ABS - .218fconc.l + .00183
r = .9999
Sy»x = .00276
1234
Silicate Concentration, jig/ml as Si02
Figure F-1. SiO2 determinations on quartz filters (date analyzed: April 8, 1976).
-------
APPENDIX G
Determination of Phosphate
Introduction
The chemistry of molybdenum blue formation was exhaustively studied in
the past three decades and as a result many methods for phosphate, silicate,
arsenate, germanium and other metals were described. Generally, these methods
are not specific and are cumbersome. Higher specificity can be obtained,
however by employing low pH, organic reducing substances and complexing
agents. The method for phosphate was improved by selecting appropriate
conditions which enable the use of a single solution (1, 2) for the analysis.
The procedure used here is a minor modification of the method in references
1 and 2. Changes involved a proportional reduction of the amounts of reagent
which was necessary to handle the small samples available.
Interferences
In this method, the only interferent of any significance is arsenate, which
is O.lj- times as sensitive as phosphate. This element is only rarely occur-
ring in concentrations affecting the phosphate results (the natural ratio
is < 1:20). The color is measured either at 710 or 882 run. About 20$
higher sensitivity can be achieved when measuring at 882 nm. However, this
wavelength is near the cut-off limit of most instruments and causes large
instabilities. The second peak at 710 nm was chosen for use with our instru-
mentation. The concentration of the complexing metal, antimony, must be
below 8 jug/ml to avoid clouding. The blue color contains phosphorus and
antimony in a ratio of about unity.
- 158 -
-------
Reagents
Mix 32 ml of 5 N H2S04 (made by diluting 70 ml cone, acid to 500 ml)
20 ml of 2% molybdate
35 ml of 1% ascorbic acid solution containing 1% Na bisulfite
3 ml of 1 mg/ml solution of Sb (0.2?U g of K Sb tartrate to 100 ml)
the mixture to 100 ml with distilled water and discard after 2h hours.
Procedure
To three ml of sample solution add one ml of the reagent mixture. When
using a 25 mm cell, these volumes must be doubled. Measure at 710 nm
after 10 to 15 min.
A range of 0.3 to k ppm can be covered with a 25 mm cell. With shorter
pathlengths, e.g. 10 mm, up to 80 ppm can be measured directly. Higher
concentrations need aliquoting and dilution of samples.
Standards
The 1000 ppm P04~ standard is made by dissolving 0.286 g KH2P04 in 200 ml
of distilled water. Dilute 1 ml to 100 ml with water for the working
standard of 10 ppm. With a repetitive pipet make dilutions covering the
desired range.
1. J. Murphy and J. P. Riley: A modified single solution method for the
determination of phosphate in natural waters. Anal. Chim. Acta 27,
31-36, 1962.
2. S. J. Eisenreich, R. T. Bannerman, D. E. Armstrong: A simplified
phosphorus analysis technique. Environmental Letters 9(1), ^3-53, 19T5-
- 159 -
-------
Working Curve
Figure B-l illustrates the working curve used as prepared from standards
run "before and after the samples. Where only a single point is shown the
two determinations are indistinguishable.
- 160 -
-------
0>
1
.Q
u
o
M
.338 [Cone.] + .0067
.999 J
.005
.4 .6
Phosphate Concentration, jug/ml
Figure G-1. Calibration curves for phosphate determination on selected quartz and glass total filters
(May 14, 1976 calibration).
-------
Appendix H
DETERMINATION OF SULFITE
Three on-inch filter discs, previously removed from glass total filter samples,
were placed into 100 ml plastic screw cap vials containing 10 ml of 0.0k M
tetrachloromercurate. The containers were capped and immersed into a minimal
depth of water inside a sonicator bath for one minute.
After removal from the bath, the vials were uncovered and the following re-
*
agents were added:
1 ml of 0.6% sulfamic acid. After 10 min. stabilization,
2 ml of a 1:200 dilution of k-0% formaldehyde,
5 ml of pararosaniline solution (20 ml purified 0.2$ pararosaniline
diluted with 200 ml 3 M H3P04 and 30 ml distilled water) and
2 ml of distilled water. The total volume was 20 ml.
After 20 min. reaction time, each filter sample was filtered by gravity through
a 7 cm No. ho Whatman filter into a dry test tube 16 x 200. The absorbance was
measured at 580 nm in a 5 cm cell against water. A calibration line was meas-
ured concurrently. This was prepared by measuring aliquots of a precalibrated
10 jug/ml So2 solution in tetrachloromercurate. Figure H-l shows the resulting
working curve.
Adapted with the modifications indicated for maximum sensitivity from Method
^^Ol-Ol-So/T, Intersociety Committee: Methods of Air Sampling and Analysis,
American Public Health Association, 1972, Washington, B.C., pp
- 162 -
-------
SULFITE CALIBRATION CURVE
.8
.6
0)
o
9
.0
JH
O
CO
-4
.2
Date: 5-18-76
Total Glass
1 .2 .3
Concentration, jug/ml
.5
ABS
Syx
,007 + 1.53
,0021
[Cone]
Figure H-l
-163-
-------
Table H-l
INTERLAB COMPARISON OF SULFITE DETERMINATIONS USING SMELTER DUST SAMPLES
Sample
1
2
3
Sulfite, weight
Smelter Type
Cu
Pb
Pb
Cu
Pb
BYU
March Oct. 1976
0.69 + .1 1.03 + .oU
none 2.16 + .62
none none
1.5U + .1 1.1? + .1
0.85 + -11 0.77 + .09
AIHL
Sept. 1976
not determined
0.021
0.002
0.85
-------
Appendix I
X-Ray Fluorescence Analysis of St. Louis
Aerosol Collected on Fluoropore Filters
Thomas G. Dzubay
Environmental Sciences Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
An X-ray fluorescence spectrometer was used for elemental analysis of the
aerosol particles that were collected on Fluoropore filters using dicho-
tomous samplers that were operated in St. Louis. The spectrometer was
equipped with a secondary fluorescence-type excitation source and a Si (Li)
1 2
type energy dispersive detector. ' Elements with atomic numbers between
13 and 20 were analyzed using the nearly monoenergetic X-rays from a
titanium fluorescer; elements with atomic numbers between 22 and 38 and Pb
were analyzed using a molybdenum fluorescer.
The X-ray spectra for each sample were analyzed using a sequential stripping
1 2
technique. For this technique, a library of single element spectra
corresponding to thin standards for all elements to be analyzed was stored
in the memory of a minicomputer, which was used for the analysis. Also
stored were spectra for a clean filter to represent the blank. Using the
stripping technique, a linear combination of the stored single element
spectra and blank was found which accurately fit the unknown spectrum. The
amount of each component in this fit was assumed to be proportional to the
concentration of each element in the sample.
- 165 -
-------
The X-ray fluorescence spectrometer was calibrated using thin film standards
obtained from Micromatter Co., Seattle, Washington. Each standard consisted
of a 28 mm diameter deposit of 1 or 2 elements on a thin Mylar film. The
deposits were prepared in a vacuum using vapor deposition. For each stan-
dard, the mass per unit area was determined from the measured weight gain
of the substrate after the deposition and from the known area of the deposits.
The sulfur standard consisted of sulfur and copper deposits of 28 Mg/cm2 and
91 fig/cm2, respectively, on the Mylar film.
For analysis of sulfur, the Ka line was used. For this line there is an
interference from the M X-rays of lead. To correct for this interference,
the true sulfur concentration was obtained from the lead concentration and
from the uncorrected sulfur concentration using the relationship:
S(true) = S(uncorr) - K Pb
For the correction coefficient, the value K = 0.50 + 0.05 was deduced by
analyzing a thin film lead standard as if it were sulfur. With a 10$
uncertainty in K, one can estimate the resulting uncertainty in the sulfur
concentration due to the presence of lead. For samples collected at
Regional Air Pollution Study (RAPS) Site 106 between August 18 and September 7,
19T5> the mean lead and sulfur concentrations were 0.6 and 3«7 Atg/m3, respec-
tively. Thus, the lead causes an uncertainty of 0.03 Mg/ni3 for the sulfur,
which amounts to only O.Q% of the mean sulfur concentration. At RAPS Site
12h, the mean lead and sulfur concentrations were 0.13 and 3-0 jug/m3,
respectively, resulting in a 0.2% contribution to the uncertainty in the
mean sulfur contribution.
- 166 -
-------
For sulfur in the fine particle fraction, a correction was made for a
theoretically predicted attenuation of the X-rays by the filter medium in
which particles were collected.3 This correction consisted of dividing the
sulfur K X-ray yield by the factor 0.85.
The results from the X-ray fluorescence analysis of the samples from
St. Louis were submitted in March 1976 to Dr. Bruce Appel of the Air and
Industrial Hygiene Laboratory. More recently, it has been learned that the
attenuation effect for the membrane filter is much smaller than the earlier
theoretically predicted value, and that the appropriate attenuation factor
k
should be 0.97 + 0.03. In addition, there is a need to make a small
correction for the attenuation of the sulfur in the layer of particles
collected on the filter. The net effect of the revised filter attenuation
factor and the added layer attenuation factor will be to yield an overall
factor that is close to the value of 0.85, which was used in the original
analysis. A. report describing these correction's in greater detail is now
being prepared.
References
1. Goulding, F.S. and J. M. Jaklevic. "X-Ray Fluorescence Spectrometer for
Airborne Particulate Monitoring," EPA Report No. EPA-R2-73-182, April 1973-
2. Jaklevic, J.M., F.S. Goulding, B.V. Jarrett and J.D. Meng. "Applications of
X-Ray Fluorescence Techniques to Measure Elemental Composition of Particles
in the Atmosphere," in Analytical Methods Applied to Air Pollution Mea-
surement, R.K. Stevens and W.F. Herget, Eds. (Ann Arbor, Michigan:Ann Arbor
Science Publishers, 197*0» PP- 123-lk6.
3. Dzubay, T.G. and R.O. Nelson. "Self Absorption Corrections for X-Ray
Fluorescence Analysis of Aerosols, " in Advances in X-Ray Analysis, Vol 18,
(New York: Plenum publishing Corp., 1975), PP- 619-631.
k Loo, B.W., R.C. Gatti, B.Y.H. Liu, C.S. Kim, and T.G. Dzubay. "Absorption
Corrections for Submicron Sulfur Collected in Filters, " in X-Ray Fluorescence
Methods for Environmental Samples. (Ann Arbor, Michigan: Ann Arbor
Science, l9YY)«
- 167 -
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-600/7-77-128
3. RECIPIENT'S ACCESSION-NO.
4. TITLE AND SUBTITLE
COMPARISON OF WET CHEMICAL AND INSTRUMENTAL METHODS
FOR MEASURING AIRBORNE SULFATE
Final Report
5. REPORT DATE
November 1977
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
B. R. Appel, E. L. Kothny, E. M. Hoffer, and
J. J. Wesolowski
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Air & Industrial Hygiene Laboratory
California Department of Health
2151 Berkeley Way
Berkeley, California 94704
10. PROGRAM ELEMENT NO.
EHE 625 EB-08 (FY-76")
11. CONTRACT/GRANT NO.
68-02-2273
12. SPONSORING AGENCY NAME AND ADDRESS
Environmental Sciences Research Laboratory - RTF, NC
Office of Research and Development
U. S. Environmental Protection Agency
Research Triangle Park. N.C. 27711
13. TYPE OF REPORT AND PERIOD COVERED
•FTNAT.
14. SPONSORING AGENCY CODE
EPA/600/09
15. SUPPLEMENTARY NOTES
This contract was partially funded (47%) by EMSL - RTP.
16. ABSTRACT
The methylthymol blue (MTB), modified Brosset, and barium chloranilate sulfate methods
were evaluated for precision, accuracy, working range, interference effects, and
degree of agreement with x-ray fluorescence analysis (XRF) using atmospheric particu-
late samples. The samples used were collected simultaneously with glass fiber, quartz
fiber and Fluoropore filters, the latter being used in a dichotomous sampler. Studies
of interference effects were based upon measured concentrations of potential inter-
ferents extractable from the particulate matter as well as the filter media.
The results demonstrated agreement within 16% for determining atmospheric sulfate
concentrations by the three wet chemical procedures with all the filter media. XRF
results on the "fine" Fluoropore samples agreed within 10% of those obtained by wet
chemical procedures on the samples and were, on average and within experimental
error, equivalent to results obtained by the MTB method on 8 x 10" glass fiber high
volume samples. Small differences in results obtained with different filter media
in the present study are more consistent with the effects of analytical interferents
rather than artifact sulfate formation as the cause.
17.
KEY WORDS AND DOCUMENT ANALYSIS
a.
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
COSATI Field/Group
*Air pollution
*Particles
*Sulfates
*Chemical analysis
*Comparison
*X-ray fluorescence
13B
07B
07D
20F
18. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19: SECURITY CLASS (ThisReport)
UNCLASSIFIED
21. NO. OF PAGES
178
20. SECURITY CLASS {This page)
TTWrr.ASSTPTF.n
22. PRICE
EPA Form 2220-1 (9-73)
-168-
------- |