EPA-600/2-76-059
March 1976
Environmental Protection Technology Series
COMPARISON OF WET CHEMICAL AND INSTRUMENTAL
METHODS FOR MEASURING AIRBORNE SULFATE
Interim Report
^£D ST4f(
-3
Environmental Sciences Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into five series. These five broad
categories were established to facilitate further development and application of
environmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The five series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
This report has been assigned to the ENVIRONMENTAL PROTECTION
TECHNOLOGY series. This series describes research performed to develop and
demonstrate instrumentation, equipment, and methodology to repair or prevent
environmental degradation from point and non-point sources of pollution. This
work provides the new or improved technology required for the control and
treatment of pollution sources to meet environmental quality standards.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
COMPARISON OF WET CHEMICAL AND
INSTRUMENTAL METHODS FOR MEASURING
AIRBORNE SULFATE
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 68-02-1660
Project Officer
Carole R. Sawicki
Atmospheric Chemistry and Physics Division
Environmental Sciences Research Laboratory
Research Triangle Park, North Carolina 27711
U.S. ENVIRONMENTAL PROTECTION AGENCY
OFFICE OF RESEARCH AND DEVELOPMENT
ENVIRONMENTAL SCIENCES RESEARCH LABORATORY
RESEARCH TRIANGLE PARK, NORTH CAROLINA 27711
-------
DISCLAIMER
This report has been reviewed by the Environmental Sciences Research Labor-
atory, 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 recommenda-
tion for use.
ii
-------
CONTENTS
Page
List of Figures v
List of Tables vi
Acknowledgements viii
SECTIONS
I. Introduction 1
£
II. Summary _ 4
III. Set Up and Preliminary Evaluation of Methods 7
A. Methylthymol Blue Procedure
B. Modified Brosset Procedure
IV. The Effect of Interferents on Sulfate Determinations 16
A. Description of the Experiment
B. Results of Interference Studies
C. Time Dependence of Interference Effects with Sulfide
and Sulfite
D. Recommendations to Eliminate Interferences
E. Summary and Conclusions
iii
-------
V. Precision, Equivalence and Accuracy of Four Sulfate
Procedures with Atmospheric High Volume Filter Samples 29
A. Description of the Experiment
B. Analytical Precision
C. Equivalence of Methods
D. Interference Effects
E. The Accuracy of the Sulfate Methods by Standard
Additions
F. Summary and Conclusions
VI. Equivalency of Wet Chemical and X-ray Fluorescence Methods
and Influence of Sampling Design with Atmospheric Low Volume
Filter Samples 38
A. Description of the Experiment
B. Results
C. Summary and Conclusions
VII. References 47
Appendix
A. The Turbidimetric Method. 101
B. The Technicon Industrial Method No. 118-71W. 106
C. The AIHL Microchemical Method. 109
D. The Modified Brosset Method. 132
iv
-------
FIGURES
1 Preliminary Comparison of Turbidimetric and Methylthymol Blue Methods
for Sulfate on Atmospheric Samples Collected on 24-Hour Glass Fiber
High Volume Filters
2-8 Interferograms
9-22 Recovery of Sulfate from Standard Additions to Atmospheric Samples
-------
LIST OF TABLES
1 Precision of Standard Curves for Methylthymol Blue (NASN) Analysis of
Sulfate with the Technicon Auto Analyzer II
2 Comparison of the Brosset and Micro AIHL Methods for Sulfate
3 Comparison of H2SC>4 and Na2S04 Standards in Analysis by the Modified
Brosset Procedure (in acetone) with 2.5 and 5.0 cm Cells Without Ion
Exchange
4 Precision of H2S04 and Na2S04 Standards in Analysis by the Brosset
Procedure in Dioxane Without Ion Exchange
5 Precision of the Modified Brosset Procedure with Ion Exchange Treatment
(three filter discs)
6 Recovery of Sulfate by Modified Brosset Procedure After Treatment with
Three Ion Exchange Filter Discs
7 Interference Effects in the Turbidimetric Method (yg/ml Observed Sulfate)
8 Interference Effects in the Technicon Methylthymol Blue Method (yg/ml
Observed Sulfate)
9 Interference Effects in the AIHL Microchemical Method (yg/ml Observed
Sulfate)
10 Interference Effects in the Modified Brosset Method, in Acetone (yg/ml
Observed Sulfate)
11 Interference Effects in the Modified Brosset Method, in Dioxane (yg/ml
Observed Sulfate)
12 The Effect of Aging of Interferent-Sulfate Solutions on Interference
Effects by the Modified Brosset Method (in dioxane)
13 The Effect of Aging of Sulfide-Sulfate Solutions on Interference Effects
by Four Methods (20 yg/ml added sulfate)
14 The Effect of Aging of Sulfite-Sulfate Solutions on Interference Effects
by Four Methods (20 yg/ml added sulfate)
15A Sulfate Analysis of Atmospheric Hi-Vol Samples by Four Methods, St.
Louis, MO (yg/m3)
15B Sulfate Analysis of Atmospheric Hi-Vol Samples by Four Methods, Durham,
NC (yg/m3)
vi
-------
15C Sulfate Analysis of Atmospheric Hi-Vol Samples by Four Methods, Pasadena,
CA (yg/m3)
16 The Influence of Ion Exchange Treatment on High Volume Filter Samples
Analyzed by the AIHL Microchemical Method
17 Protocol for Standard Addition Study-High Volume Glass Fiber Filter
Sample Extracts
18 Recovery of Addition of Sulfate to High Volume Glass Fiber Filter Sample
Extracts (%)
19 Mean Fractional Recoveries of Sulfate with Standard Additions
*5
20 Summary of Low Volume Fluoropore Filter Sulfate Determinations (yg/m
sulfate)
21 Summary of Low Volume Glass Fiber Filter Sulfate Determinations
22 Relative Results-Fluoropore Filters
23 Relative Results-Glass Fiber Filters
24 Comparison of- EPA and LBL X-ray Fluorescence Results on Low Volume
Fluoropore Filters
25 The Mean Fraction of Sulfate in Refined (0-2 ym) Particles
26 Summary of Comparison of Glass Fiber and Fluoropore Filter Sulfate
Results as a Function of Particle Size and Sampling Site
27 Comparison of Selected Metals Concentrations by XRFA in Total and
Refined Particle Samples
28 Comparison of Low (0-20 ym) and High Volume Glass Fiber Filter Sulfate
Results
29 Comparison of Fluoropore (0-20 ym) and Glass Fiber High Volume Filter
Sulfate Results
vii
-------
ACKNOWLEDGEMENTS
Other participants in this study included S. Twiss (data reduction and dis-
play) , Y. Tokiwa (field sampling) G. Buell (assistance with modified Brosset
determinations) and A. Alcocer (sample handling and logistics).
In addition, we wish to acknowledge R.D. Giauque, L. Goda and T. Dzubay of
the Environmental Protection Agency who provided the x-ray fluorescence
analysis discussed in this report. We also express our appreciation to
J. Frazer, F. Scaringelli and Dr. J. Stikeleather of the Environmental Pro-
tection Agency and G. Colovos of Rockwell International Science Center for
helpful discussions.
Mrs. Carole Sawicki served as Project Officer for this program. Her help-
fulness throughout this work has been sincerely appreciated.
-------
I. INTRODUCTION
As a consequence of the present energy crisis and the limited supply of
natural gas and low sulfur oil, use of fuels of higher sulfur content is
increasing. The adverse implications for human health of particulate
sulfate together with the likelihood of higher atmospheric loadings for
this material make essential the use of rapid, sensitive, accurate, speci-
fic, and precise analytical methods to determine the levels of atmospheric
sulfate. Validated methods are especially required for determining sulfate
in the range of 1 to 10 yg/ml such as are often obtained with extracts from
1 to 2 hour size-segregated aerosol samples. As part of existing programs,
the Air and Industrial Hygiene Laboratory (AIHL) has previously begun
evaluation of both instrumental and wet chemical methods for aerosol analysis
including sulfate and nitrate methods. The present study complemented this
work, with principal focus on comparison of methods for particulate sulfate
analysis.
The objectives of the program have been to evaluate and compare procedures
for sulfate determination in atmospheric particulate matter and to relate
sulfate values so obtained to total sulfur determinations by x-ray fluores-
cence analysis (XRFA) using samples collected in several geographic areas
differing in pollutant composition. The specific goals of the evaluation
effort include:
1. Evaluation of the precision and accuracy of each of the sulfate methods
employing synthetic and atmospheric samples from several locations.
-1-
-------
2. Evaluation of the influence of likely interferents on the apparent
sulfate values.
Since the degree of interference in analytical methods may be dependent on
particle size, as with XRFA, or on the concentration of interfering species
which are principally associated with small or large particles sizes in
atmospheric aerosols, size-segregated samples are included in this study.
The present study has examined five analytical methods for water soluble
sulfate or total sulfur:
1. The BaCl2 turbidimetric procedure (Appendix A).
2. The methylthymol blue procedure as automated for the Technicon Auto
AnalyEer II (MTB) (Appendix B).
3. The AIHL microchemical procedure (Appendix C)
A. A modified Brosset procedure (Appendix D)
5. X-ray induced x-ray fluorescence (XKFA).
Ambient air samples were collected at three locations, Durham, NC; St. Louis,
MO; and Pasadena, CA; to obtain varied particle matrices. Sampling in North
Carolina and Missouri was conducted by the EPA staff while that in Pasadena
was conducted by the AIHL staff. Sampling at each site was conducted for
four days for a total of 12 days of sample collection for the study. All
sample collections were on a 24-hour basis. Sampling equipment consisted
of a hi-volume sampler and two "T samplers". Each "T sampler" contained
both a total and a refined particle sampler. Filter media and sampling
conditions were as follows:
-2-
-------
1. Hi-volume sampler 8 x 10" Gelman Type A glass fiber filters.
2. Total filter 37 mm Gelman Type A glass fiber filters, sampling
at 12 1/min and collecting 0 to 20 ym particles.
3. Refined fraction filter 37 mm Gelman A glass fiber filters,
sampling at 12 1/min and collecting 0 to 2 ym particles.
4. Total filter 37 mm Fluoropore (1.0 ym pore size) filters, sampling
at 12 1/min and collecting 0 to 20 ym particles
5. Refined fraction filter 37 mm Fluoropore (1.0 ym pore size)
filters, sampling at 12 1/min and collecting 0 to 2 ym particles.
Wet chemical analysis of these as well as synthetic samples was carried out
at AIHL while XRFA was carried out both at the Environmental Protection
Agency's Research Triangle Park Laboratory (EPA-RTP), and at the Lawrence
Berkeley Laboratory (LBL) under subcontract to AIHL.
-3-
-------
II. SUMMARY
A program has been conducted to evaluate and compare four wet chemical methods
and x-ray fluorescence analysis for determination of sulfate in atmospheric
samples. The four chemical methods studied were the BaCl2 turbidimetric,
automated methylthymol blue (MTB), AIHL microchemical and a modified Brosset
procedure. The specific parameters evaluated included the equivalency of the
methods, analytical precision and accuracy and the influence of potential
interferents. The atmospheric samples used were collected in Durham, NC;
St. Louis, MO and Pasadena, CA. In addition, the sampling design permitted
a limited evaluation of sampling errors related to atmospheric sulfate deter-
mination. Specifically the influence of filter medium, sampling volume and
particle size were evaluated.
With 24-hour high volume filter samples analytical precision with the wet
chemical methods expressed as coefficients of variation, ranged from 1-5%.
The methods yielded nearly equivalent results with the range of results with
the three sets of samples being about 10%. While agreeing within the stated
range, the MTB and modified Brosset methods gave generally higher results
than the other methods. This proved to be consistent with studies of accur-
acy by standard additions suggesting a systematic positive bias rather than
negative interference effects with the remaining methods as the cause for
the higher results. The standard addition studies with the MTB method
revealed positive errors of 10-20% with 63% of the experiments, 5-10% posi-
tive errors in 25% of the experiments and -5 to 5% errors in the remaining.
-4-
-------
The degree of agreement found suggests that the choice of a method from
among these four, for analysis of high volume samples can be based upon
such factors as cost per determination or experimental convenience. The
restricted range of both the modified Brosset and AIHL micromethod clearly
makes them inappropriate for consideration with high volume samples.
The four wet chemical methods displayed widely varying sensitivities to
the 12 interferents studied. The modified Brosset procedure emerged as
the method least affected by the potential interferents; only sulfide and
sulfite caused significant interference with this method. The degree of
agreement between the four methods with high volume filter samples suggests
these interferents to be of minor importance with large air samples. In
what appears to be the worst case, the AIHL microchemical method results
for St. Louis and Durham were shown to be about 10% low due to interference,
probably by cationic interferents.
In contrast to the high volume sample results, low volume filters have
revealed differences between the three wet chemical methods able to analyze
these samples of up to a factor of 2 for individual samples and 1.6 when
pooled by sampling site. The greater dilution of interferents from the
filter medium with the high volume samples may account for the closer
agreement found between methods with these samples.
XRFA results for sulfur by the Lawrence Berkeley Laboratory and Research
Triangle Park differ by a factor of 1.7 for the same samples. The lack
-5-
-------
of good agreement for sulfur in this study indicates a need for better
standards and for more accurate correction for the attenuation of the soft
sulfur x-rays in the filter medium. The RTF XRF measurements of total
sulfur as sulfate are generally higher by 10 to 50% than the measurements
by the MTB method; the LBL XRF measurements are lower than the MTB measure-
ments by 20 to 50%.
The present data are consistent with significant artifact sulfate formation
from S02 on the low volume glass fiber filter samples with enhanced sulfate
formation in the presence of a large particle-related oxidation catalyst(s).
The equivalence of low volume Fluoropore total filter and high volume glass
fiber sulfate results as measured by the MTB method implies that an insig-
nificant percentage of the sulfate determined is due to artifact sulfate
formation with 24-hour high volume glass filters.
Finally, in comparing the three wet chemical methods with low volume samples,
the modified Brosset procedure has, at times, yielded what is considered
to be erratic behavior. In spite of the positive error in the method
revealed by studies with high volume samples, the MTB method is considered
the most reliable of the three for long term (e.g. 24-hour) low volume
samples. For short term, low volume samples requiring a micro sulfate
method we favor the AIHL microchemical method.
-6-
-------
III. INITIAL SET UP AND EVALUATION OF METHODS
Both the AIHL micromethod and the BaCl2 turbidimetric procedure have been
in use for some time at AIHL and, therefore, required no further prepara-
tory work before beginning the present program. However, the automated
MTB and Brosset methods were new to the laboratory and, therefore, required
both set up and preliminary evaluation to insure reliable data are obtained
before beginning the methods comparison study. The present section describes
these efforts.
A. The Automated Methylthymol Blue (MTB) Procedure
1. Introduction
The objective in setting up this method was to duplicate as closely
as possible the technique in use at Research Triangle Park for
analysis of samples from the National Air Sampling Network (NASN).
Starting point for this effort was the Technicon Industrial Method
No 118-71W, a copy of which is included as Appendix B. The Rock-
well International Science Center, which currently uses the MTB
procedure with a Technicon Auto Analyzer II, provided initial assis-
tance. An inherent limitation in duplicating the RTP system arises
from the use at RTP of the Technicon Auto Analyzer I rather than
the model currently available. It is believed that the more uniform
air bubbles (used to separate samples in the tubing) in the Auto
-7-
-------
JL
Analyzer II will produce somewhat improved precision.
The similarity of operating procedures used at the AIHL and EPA-RTP
was confirmed by site visits.
2. Operation and Calibration of the Auto Analyzer II for Sulfate Analysis
During initial operation, bubbles in the ion exchange column proved
to be a source of significant unreliability. When starting the
Auto Analyzer there is an initial surge of air. This air surge is
greater than the capacity of the debubbler preceding the ion ex-
change column thereby introducing air into that column. To overcome
this strong surge, a by-pass and valve were introduced between the
debubbler and the mixing coils. In starting up the instrument the
valve is opened and the air surge goes through the by-pass. When
the surge dissipates the valve is closed and the flow goes through
the column without introducing air. This modification does not
affect the operation but significantly decreases the down time of
the instrument.
The original range for sulfate as supplied (0-300 yg/ml) was adjusted
*RTP has recently changed to use of the Auto Analyzer II and has, indeed,
found improved precision (C.V. 5% vs 7%). Accordingly, the AIHL and RTF
methods are, at present, the same.
-8-
-------
to give a recorder span of 0-60 yg/ml by using an 0.42
sample suction tube and a 1.6 cnrVmin stream of dilution water.
The calibration curve is approximately linear over the 0-60
range.
As an initial step for running comparisons, the Auto Analyzer was
calibrated with sodium sulfate standards up to 100 ug/ml S04= in
different positions on the sampling tray to determine memory effects,
High-volume filter sample extracts which had been previously ana-
lyzed for sulfate by turbidimetry were available for an initial
comparison. Blanks were run separately by replacing the methyl-
thymol blue solution with 80% alcohol and extending the sensitivity
range.
3. Operating Parameters Influencing Reliability of the MTB Method Data
After one and a half hours warm up time of the Auto Analyzer, blanks
and standards were run on a 30 samples per hour mode. Since the
turntable has 40 positions (defined as a cycle), each cycle in the
mode lasts about one hour and 20 minutes. In order to compare the
cycles within one day and over more extended periods of time, the
regression lines from the standards were calculated and the follow-
ing parameters were obtained: intercept (a), slope (b), standard
deviation of intercept (Sa), slope (Sfo) and the standard error of
the estimate (Sy.x). (Table 1.)
-9-
-------
Table 1
PRECISION OF STANDARD CURVES FOR METHY1THYMOL BLUE (NASN)
ANALYSIS OF SULFATE WITH THE TECHNICON AUTO ANALYZER IIa
a)
b)
c)
a)
b)
c)
a)
b)
c)
a)
b)
c)
a)
b)
c)
Parameters
Linear range 0-200 jug S04=/Anl
No ion exchange
10 standard solutions at 5
concentrations + 30 water
blanks
Linear range 0-60 jug S04~/ml
No ion exchange
10 standard solutions at 5
concentrations + 30 water
blanks
Linear range 0-60 jug S04=/ml
2 ram I.D. ion exchange
10 standard solutions at 5
concentrations + 30 water
blanks
Linear range 0-60 jug S04=/ml
2 mm I.D. ion exchange
30 samples and 10 standard
solutions
Linear range 0-60 yg SO^/ml
2.5 inm I.D. ion exchange
30 samples and 10 standard
solutions
Cycle13
Wo.
1
h
1
3
1
2
1
2
1
2
Sy.x
0.145
OM
0.91
0.146
0.83
0.62
1.15
1.33
0.88
0.81
b
0.28
0.29
1.10
1.08
1.07
1.05
1.10
1.15
1.17
I.ll4
sb
.002
.002
.022
.011
.012
.009
.023
.030
.020
.018
ac
-1.U
-2.5
-U.3
-3-8
-2.8
-1.0
-1.1
-3-0
-3.6
-2.1
sa
.25
.25
.85
.1*3
.1*5
31*
1.1
1.3
.83
.77
a. The symbols are as follows:
Sy.x = standard error of the estimate
a = intercept
b = slope
Sa and St = standard error of intercept and slope, respectively.
b. Each cycle represents running of some combination of standards,
samples and water totaling ^0 determinations.
c. In strip chart units, 100 units per full scale.
-10-
-------
The results show small variations in slope and significant shifts
in the intercept over the linear portion of the working curves.
These findings indicated the need to either perform base line sub-
tractions or to intermingle standards and samples using the working
curve developed for a given cycle to calculate the unknown samples.
There are two relevant blanks of concern to the study of the auto-
mated MTB method, (1) the inherent color of the aqueous extract of
a particulate sample, and (2) the sulfate blank from extraction of
a clean filter. The sample blanks (1) were determined for a group
of samples collected in Riverside, CA. For these samples the sample
blank measured by substituting 80% ethanol* for the reagent repre-
sented < 1.0% of the measured sulfate and is considered within the
uncertainty of the method. Therefore, such corrections were not
determined during the remainder of this study. Filter blanks were
measured for Gelman A glass fiber and proved to be 2-3 vg/ml apparent
SC>4=**. Cellulose filters (Whatman 41) had a zero blank.
*This omits HC1 and BaCl2 as well as the dye. The effect of omission of
HC1 is negligible on the final pH since a large excess of NaOH is added,
relative to the acid normally present. The influence of. omitting BaCl2
is considered negligible as well.
^Extracting a 3/4" strip into 100 ml H20.
-11-
-------
4. Intermethod Comparison with the MTB and Turbidimetric Methods -
Preliminary Findings
Employing aqueous extracts from particulate samples collected on
glass fiber filters, sulfate determinations were conducted both
by the MTB and turbidimetric method. A comparison of results by
the two techniques is shown in Figure 1 indicating at most a 15%
difference with the MTB results consistently higher. The differ-
ence is especially marked at high concentration. Having estab-
lished by preliminary studies the precision and comparability of
the MTB to the turbidimetric method, it was judged adequate for
beginning interference studies.
B. The Modified Brosset Procedure
1. Introduction
The starting point for this method is the work by Brosset and
Ferm. '^ As detailed therein, the method is semi-automated and em-
ploys custom-made glassware. Since it was not considered feasible
to duplicate this system in the time available, methods referred to
here as "modified Brosset procedures" were established and evaluated.
The Brosset procedure is based on the change in color brought about
by the reaction:
-12-
-------
PRELIMINARY COMPARISON OF TURBIDIMETRIC AND METHYLTHYMOL BLUE METHODS
FOR SULFATE ON ATMOSPHERIC SAMPLES COLLECTED
ON 24 HOUR GLASS FIBER HIGH VOLUME FILTERS
80 r
60
00
_
CO
5 40
E
-------
SC>4 + Ba-thorin dye -* BaS04 + thorin dye complex
(a suspension) (a suspension)
The method can be carried out in several solvents. Procedures in
both acetone and dioxane were evaluated in this study. The final
medium containing 70% organic solvent, is necessary to reduce the
dissociation of the Ba-thorin complex.
A comparison of the Brosset and AIHL microchemical methods for
sulfate is shown in Table 2.
2. Dispensing Systems for Reagents and Samples
Several alternatives to the custom-made dispensing device of the
original method were tried including the use of a diluting-dispen-
sing syringe and a small volume reagent pipet all made by Lab-
industries. With this equipment, small variations in dispensed
volumes caused a change in the ratio of sample to diluent and of
reagent to diluent, which resulted in a change of the dissociation
of the Ba-thorin complex and a shift in the absorption maxima. By
employing dispensing pipets fabricated by Rainin, most of these
variations were reduced significantly. The volume delivered by
these dispensing pipets was shown to be within 1% of the nominal
values. Employing the Rainin pipets the sample solution and the
reagent mixture were measured independently. The diluent was added
-Ik-
-------
Table 2
COMPARISON OF THE BROSSET AND MICRO AEHL METHODS FOR SULFATE
Brosset
Ion exchange to avoid Ca, Pb, and
other interferents.
Sensitivity 0.1 to 0.2 jug/ml.
Two ml of sample used.
Dioxane or acetone 70$, remainder
water.
Organic solvent used to reduce
dissociation of Ba-thorin complex.
The complex is a colloid which
precipitates with time.
25 or 26 mm optical cell in sample
beam. 520 to 1480 nm filter or 520
nm monochromator, measuring decrease
in absorbance. A single beam spec-
trophotometer is advised by
C. Brosset. At AIHL a double beam
instrument is used with a grey filter
solution in the reference beam.
Barium diluent contains 1$ adipic
acid to overcome EgQz interference.
Diluent is a 1:100 mixture of 0.21$
Ba(C104)2 in 0.1 M HC104 and
dioxane or acetone. 5 ml per test.
Reagent is a 0.0025$ thorin solution
in 0.001 N H2S04. 0.125 ml per test
(6.125 jug S04 added per test).
Total volume 7-125 ml containing
312 VL& thorin.
Ba concentration 6.25 x 10 M in
diluent. 0.312 ^M Ba per test.
0.06U juM S04 per test.
Approximate working range 2 to
9 US/ml-
Micro-AIHL
Ca, Pb interferes only if in large amounts,
Sensitivity 0.1 to 0.2 jug/ml.
One ml of sample used.
Acetonitrile 75$? remainder water.
Organic solvent used to reduce solubility
of BaS04 and increase rate. The Ba-nitro-
chromeazo complex is soluble and stable.
10 mm optical cell in reference beam.
6h2 nm in monochromator and double beam
spectrophotometer necessary. A decrease
of absorbance is measured.
Not designed to work with solutions
containing H202 (used for collection of
S02).
Diluent-reagent mixture is a 2:100
mixture of 0.001 M BaCl2 with 85$
acetonitrile, containing 0.12$ pyridine,
0.12$ benzenesulfonic acid, 0.00375$
nitrochromeazo. 8 ml per test.
Total volume 9-0 ml containing 300 ^g
nitrochromeazo.
-5
Ba concentration 2 x 10 M in reagent.
0.160 juM Ba per test.
Range 5 to lU jug/ml.
-15-
-------
with a repetitive dispensing syringe. The reagent mixture contains
sufficient sulfate to react with 20% of the barium contained in the
diluent to repress the solubility of BaSCty and promote rapid equilib-
rium.
3. Ion Exchange Treatment of Samples
The function of the ion exchange paper is to remove cations which
might interfere in the determination; metal ions are replaced by IT".
To obviate the use of inconvenient ion exchange columns as suggested
in the original method, discs were punched from ion exchange paper
and inserted into disposable plastic syringe bodies for use.
The degree of removal of sodium ion by filter discs was used to
establish the efficiency of the system. The sample solution was
filled in to the syringe bodies and allowed to drip slowly through
the tip into small beakers. Flame photometric determinations were
used to monitor the extent of sodium penetration. With one filter
disc, penetration was evident. It was eliminated by employing two
filter discs in series. For added security three filter discs were
used routinely.
-16-
-------
4. Drifting of Reagent Blank Spectrophotometer Value
During use of a double beam Bausch and Lomb UV 200 Spectrophotometer
it became evident that the reagent blank plus excess sulfate placed
in the reference beam of the instrument caused drifting. This drif-
ting could be prevented by filling the reference beam cell with a
gray filter solution. A gray filter solution was prepared with
CoS04, CuS04 and NiCl2 which gave a reasonable flat absorbance
between 440 and 540 nm.
5. The Influence of Sodium on Analytical Precision
As summarized in Tables 3 and 4 a comparison was made between sul-
furic acid and sodium sulfate standards to assess the influence of
sodium on analytical precision (without ion exchange). The compari-
son was made with both acetone and dioxane as solvents. In both
cases the sodium sulfate standards yielded somewhat better precision.
The influence of cell size on precision can also be examined with
both sulfate reagents. No clear trend is evident so we infer that
the precision with these reagents is approximately equal for both
cell lengths.
6. Precision of the Modified Brosset Procedure with Ion Exchange Treatment
Table 5 lists the coefficients of variation (C.V.) for four replica-
-17-
-------
Table 3: COMPARISON OF HgS04 AND NaS04STANDARDS IN ANALYSIS BY THE
MODIFIED BROSSET PROCEDURE (in acetone) WITH 2.5 and 5.0 cm
CELIS WITHOUT ION EXCHANGE
Number of Cell length, Coef. of
Reagent Experiments cm Variation j>
HgS04 1 2.5 5-5
2 2.5 3.0
2 5.0 4.7
NagS04 1 5-0 4.0
a
Each experiment includes_three replications at each of five concentrations,
2,4,6,8 and 10 ug/ml
-18-
-------
Table h PRECISION51 OF H2S04 AND Na2S04 STANDARDS IN "DIEECT ANALYSIS"
BY THE BROSSET PROCEDURE IN DIOXANEb
Sulfatec
Reagent
Na2S04
H2S04
2 4
5.1 6.4
17.3 2.9
6 8
7-5 2.7
k.k 2.1
10
2.1
2.3
Overall
U.5*
5.2*
a. Expressed as coefficient of variation of four replications at
each concentration.
b. "Direct Analysis" indicates no ion exchange treatment.
c. The true concentrations with the Na2S04 standards are within
of the nominal concentrations shown. The H2S04 standards are
consistently lower by h% than the nominal values shown.
-19-
-------
Table 5 PRECISION3" OF THE MODIFIED BROSSET PROCEDURE WITH ION EXCHANGE
TREATMENT (three filter discs
Solvent 2
Acetone k.3
Dioxane 11.3
/-»
jug/ml Sulfate
h
5.0
1.9
6
2.8
5-1
8
0.9
2.9
10
0.8
2.6
Overall
2.2%
3-9$
a. Expressed as coefficient of variation of four replications at each
concentration.
b. Pretreatment of filter discs was the same in all cases; filter
discs were soaked in distilled H20 for several hours then drained
and dried overnight in a stream of clean air.
c. The true concentrations are within 1% of the nominal values shown
and employed Na2S04 as the sulfate source.
-20-
-------
tions at each of five sulfate concentrations employing standard
sulfate solutions and ion exchange. In both acetone and dioxane,
C.V. values generally decreased with increasing concentration of
sulfate, with dioxane results usually less precise than those in
acetone.
7. Recovery of Sulfate by a Modifed Brosset Procedure with Ion Exchange
Treatment
Table 6 summarizes results for recovery experiments with four repli-
cations at each of five sulfate levels in both acetone and dioxane.
Percent recoveries after ion exchange are calculated relative to
direct determination without such pretreatment. Recoveries were
generally slightly in excess of 100% with particularly large apparent
sulfate pickup at the lowest standard solution level. Since most
samples to be analyzed are expected to be in the 4-10 yg/ml range
the recoveries shown, ranging from 96 to 113% are considered ade-
quate. Acetone results appear to be somewhat better than those in
dioxane.
8. Conclusions
The acetone modification of the Brosset procedure with three pre-
treated ion exchange filter discs will be the one used for the
remainder of this text. The protocol followed is given in Appendix D.
-21-
-------
Table 6: RECOVERY3" OF SULFATE BY MODIFIED BROSSET AFTER TREATMENT
WITH THREE ION EXCHANGE FILTER DISCS'3
jug/ml Sulfate
Solvent 2 k 6 8 10 Overall
Acetone 113 96 !<& 100 10U 102%
+ 27 +9 +6 +3 +3 +1
Dioxane 136 113 HO 103 101
+ 7 + h +5 +2 +2 +1
a. Expressed as a percent of sulfate determined by direct analysis
without ion exchange treatment.
b. Pretreatment of filter discs as in footnote b. Table 5.
c. Same as footnote c, Table 5.
-22-
-------
IV. THE EFFECTS OF INTERFERENTS ON SULFATE DETERMINATIONS
A. Description of the Experiments.
The experimental design calls for an evaluation of the influence of a
series of potential interferents including:
a. sulfide
b. sulfite
c. persulfate
d. sulfur
e. phosphate
f. lead sulfate
g. calcium
h. lead
Based upon a more detailed evaluation of the interferents most likely
to be significant, the above list was modified and lengthened to the
following 12:
Anionic Parent Compound
sulfide 1:1 Na2S-9H20:NaOH
sulfite NaHS03
persulfate K2S208
thiosulfate Na2S203
bicarbonate NaHC03
phosphate Na2HP04«12 H20
silicate
-23-
-------
Cationic Parent Compound
barium BaCl2'2H20
calcium CaCC>3 + acetic acid
lead Pb(N03)2
Other
colloidal clay Kaolinite
p-benzoquinone
The interferents eliminated from this evaluation include elemental sul-
fur and lead sulfate. p-benzoquinone, was included to simulate the
yellow chromophores present in some aqueous extracts of atmospheric
particulate matter. Bicarbonate and silicate were added because these
are thought to be found in aqueous extracts from glass fiber filters
together with ions such as Na+ and Ca+ . Thiosulfate was included
because it is an end product of the reaction of sulfur and sulfite as
well as being produced during oxidation of many reduced sulfur species
under alkaline conditions. Finally, colloidal clay was added since it
may be easily obtained in filtered extracts from atmospheric particulate
matter.
It is recognized that interference effects relatable to the cationic
potential interferents may result from interactions with sulfate during
the sampling, extraction or analysis phase. However, the determination
of the point at which interference, if any, occurs was beyond the scope
of this study.
-2U-
-------
For turbidimetry and the MTB methods the interferents were examined at
the concentration levels 10 yg/ml and 30 yg/ml, at each of two sulfate
levels, zero and 20 yg/ml. Where significant interference was found,
i
an evaluation at 60 yg/ml of sulfate was also done. With the modified
Brosset and AIHL microchemical procedures the restricted range of the
methods required dilution of these concentrations by a factor of 2.5.
Thus interferents were actually evaluated with these methods at 4 and
12 yg/ml levels with 0 and 8 ug/ml sulfate. For ease in comparison of
findings, all results with the latter two methods were then multiplied
by 2.5.
Replicated data are distinguished by the quotation of the experimental
precision found (i.e. ± la). Other data presented were the result of
a single trial. Separate experiments established the time dependence
with unstable interferents.
B. Results of Interference Studies
The effects of the interferents have been expressed as "yg/ml observed
sulfate". Tables 7, 8, 9, 10 and 11 summarize such data for the inter-
ferents studied with the four wet chemical methods. Data for the
Brosset procedure were obtained with both acetone and dioxane modifica-
tions.
Tabulation and comparison of results for sulfide and siilfite are com-
-25-
-------
Table 7: Interference Effect with Turbidimetric Method (/jg/ml Observed Sulfate)
Sulfate level,
0
20
60
I
IV)
Interferent level, jug/ml 10
Interferent
Sulfide
Sulfite
Phosphate
Colloidal clay
Persulfate
Thiosulfate
Bicarbonate
Silicate
p-benzoquinone
Barium
Calcium
Lead
13
15
2.8
<1.0
2.8
-1.0 to 1.5
<1.0
-1.0 to 1.5
<1.0
-i.o to 1.5
3.0
<1.0
30
20
31
~L.5
3.3
6.9
-------
Table 8 : Interference Effect with Technicon Methylthymol Blue Method (/ug/ml Observed Sulfate)
Sulfate level, iug/ral
0
20
60
I
ru
Interferent level, ug/mL 10
Interferent
Sulfide
Sulfite
Phosphate
Colloidal clay
Persulfate
Thiosulfate
Bicarbonate
Silicate
p-benzoquinone
Barium
Calcium
Lead
3-5
5-5
1.0
<0.5
<0.5
<0.5
<0.5
-------
Table 9: Interference Effect with AIHL Micro chemical Method (jug/ml Observed Sulfate)
IX)
CD
Actual
Sulfate level, jug/ml
0
20
Interferent level, Aig/ml 10
Interferent
Sulfide
Sulfite
Phosphate
Colloidal clay
Persulfate
Thiosulfate
Bicarbonate
Silicate
p-benzoquinone
Barium
Calcium
Lead
1.7
8.2
<0.5
-------
ISable 10: Interference Effect with the Modified Brosset Method, in Acetone (wg/ml Observed Sulfate)a
Actual
Sulfate level, ug/ml
Interferent level, Mg/ml 10
Interferent
Sulfideb
Sulfiteb
Phosphate
Colloidal clay
Persulfate
Thiosulfate
Bicarbonate
Silicate
p-benzoqulnone
Calcium
Lead
0.8l .5
3.71 -u
< 0.2
< 0.2
1.2 jf .Oh
1.5 1 1
< 0.2
< 0.5
< 0.2
0.61 .1
2.21 '2
0
30
0.6l .1
7.91 -1
< 0.2
< 1.0
2.5 JH .1
1.81 -5
< 5
< 0.8
< 0.2
1.81 -8
4.5 1 .1
20
10
201 1
23 1 1
20 1 1.6
181 -3
191 -3
201 **
19 1 -8
191 -7
18 1 .5
191 -1
191 -1*-
30
231 U
28 J; .2
1910.9
19J; .Ofc
21 ± .1
22 JH .1
181 -8
20 1 1.6
20 _+_. .U
191 -3
191 -1
ro
VD
See footnote a, Table 11.
Results shown are means for trials conducted vith fresh solutions (i.e. about 1 hour old).
Results are highly dependent on aging time.
-------
Table 11: Interference Effect with the Modified Bros set Method, in Dioxane (ug/ml Observed Sulfate)*
Actual
Sulfate level, wg/ml
Interferent level, jug/ml 10
Interfere nt
8ulfideb
8ulfiteb
Phosphate
Colloidal clay
Per sulfate
Thiosulfate
Bicarbonate
Silicate
p-benzoquinone
Calcium
Lead
2 1 .7
h 1 .6
1.21 .3
< 0.7
2 1 .h
11-7
< 1
< 0.5
2 l .7
< 0.5
< 0.8
0
30
2 1 .7
11 1 .5
1.6 1 .2
1.2 1 .2
k 1 .3
51-4
< 0.6
< 0.8
< 0.6
< 0.8
< 1
20
10 30
21 1 1.3 25 l 5
2U i 2.5 > 29
19 1 1.6 19 1 1
181 1 20 1 .k
201 .5 201 .8
21 1-6 2Ul .2
21 1 .1* 20 1 .5
19 1-7 19 1 .6
20 1 .5 21 1 .8
19 1 .5 19 1 1
19 + .8 20 + .8
oo
O
aThe Brosset method cannot be used above ca 10 ng/ml SOA without dilution. As vith the AIHL
method all studies were done after diluting by a factor of 2.5. Thus the sulfate levels in
the solutions analyzed were 0 and 8 ng/ml and the interferents levels, U and 12 wg/ml.
Results shown are means for trials conducted with fresh solutions (i.e. about 1 hour old).
Results are highly dependent on aging time.
-------
plicated by significant aging effects. Results for fresh solutions of
these anions were generally available only with up to 20 yg/ml sulfate
since 60 yg/ml sulfate greatly exceeded the range for two of the four
methods. Thus for internal consistency the results tabulated are with
solutions of the same but indeterminate age. Values for fresh solutions
of these anions are listed by footnotes in these tables.
The laboratory data from these interference studies can be displayed
in a format which may be termed an "Interferogram" (Figure 2). The
Interferogram is based upon the standard working curve of sulfate ab-
sorbance vs. sulfate concentration. In the example the influence of
interferents at 30 yg/ml with sulfate at 20 yg/ml is shown with the
AIHL microchemical method. Since an 0.4 ml sample aliquot is diluted
to 1.0 ml as part of the analytical procedure, the concentrations shown
on the ordinate are lower than the above by a factor of 2.5.
The true concentration of sulfate, Co, is noted by the vertical line.
Horizontal lines, labelled to represent a given interferent, are plotted
at the absorbance found with the added interferent. For interferent y
the intersection of the horizontal line with the working curve is the
apparent sulfate level, Cy. The interference effect, in yg/ml apparent
SC>4=, is given by Cy-Co and may be either positive or negative as shown
by the x-shaped pattern. For a given sulfate level and method, the
Interferogram provides a graphical ranking of interferents by sign and
magnitude.
-31-
-------
Interferogram for AIHL Microchemical Method
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1 -
0.0
Pb
8/tg/ml
12/Jig/ml Interferent
p-benzoquinone^
HCO
Si02
y'day
10
Sulfate Concentration,
Figure 2
12 14
-------
Interferograms are included as figures 2 through 8. Results for the
unstable anions sulfide and sulfite are not shown in the figures because
of the time dependence of their interference effects. In figures 5 and
6, the working curve over the range 0-60 yg/ml sulfate has been approxi-
mated as a straight line. In fact it is slightly "S" shaped with very
marked deviations from linearity above ca. 60 yg/ml. Since for figures
7 and 8 apparent sulfate levels above 60 yg/ml are measured the working
curve extends to 80 yg/ml. In representing this as a continuous function
the more precise characterization of the working curve below 60 yg/ml is
shown. *
C. Time Dependence of Interference Effects with Unstable Anions
Two independent experiments were conducted to explore the influence of
aging of dilute sulfate-interferent solutions with sulfide, sulfite,
thiosulfate and persulfate. The first employed only one method, the
modified Brosset procedure (in dioxane) with the four interferents at
0 and 20 yg/ml sulfate. The second employed all four methods (with
both the acetone and dioxane variations of the Brosset procedure) with
the two most unstable interferents at 20 yg/ml sulfate.
*In section IVC, results obtained by linear regression in the 0-60 yg/ml
range are compared to those using a 3rd order polynomial to fit the S-
shaped function.
-33-
-------
Interferogram for BaC^ Turbidimetric Method
0.7
20/
-------
Interferogram for BaCl2 Turbidimetric Method
0.7r_
20/Kg/ml S04~
30/
0.3
0.2
Co
0.1
p-benzoquinone Si02-
and HCOj"
Ba
+2
0.0
Pb+2, Ca+2 and
10
20 30
Sulfate Concentration,
Figure k
-35-
40
50
60
-------
Interferogram for Methylthymol Blue Method
0.7
0.6
0.5
0.4
o>
o
c
CO
o
10
0.3
0.2
0.1
0.0
20/Mg/ml
10/ig/ml Interferent
Co
Ca+2 and Pb+2
3203=
Clay
Si02
0 10
20 30
Sulfate Concentration,
Figure 5
-36-
40 50 60
-------
Interferogram for Methyl thymol Blue Method
0.7
0.6
0.5
0.4
01
o
c
o
CO
0.3
0.2
0.1
0.0
Co
20/^g/ml S04~
30/xg/ml Interferent
-w'p-benzoquinone
HC03~ and P
Ca+2 and S102
I
I
10
20 30 40
Sulfate Concentration, Mg/ml
Figure 6
50
60
-37-
-------
Interferogram for Methylthymol Blue Method
0.9
0.8
0.7
0.6
0.5
0)
o
c
I ».*
0.3
0.2
0.1
0.0
10
60/Kg/ml S04~
10/(g/ml Interferent
20 30 40 50
Sulfate Concentration, /u.g/ml
Figure 7
60 70
80
-38-
-------
Interferogram for Methylthymol Blue Method
0.1 l_
0.0
0 10
60/^g/ml S04
SO^wg/ml Interferent
P04= and
p-benzoquinone
20 30 40 50
Sulfate Concentration, /Ug/ml
Figure 8
-39-
-------
Results of the study using the modified Brosset method with four inter-
ferents are shown in Table 12. In all cases the interference is positive
with increasing effects with time. Comparing solutions with and without
sulfate, the effects of sulfide, thiosulfate and sulfite appear to be
independent of the presence of sulfate. Note that the measured sulfate
in the presence of S= and 803" exceeded the range of the method after
extended storage periods. Assuming complete oxidation to sulfate, 30
yg/ml of S~ is theoretically capable of generating 90 yg/ml of S04=, and
30 yg/ml of 863", 36 yg/ml of S04=. Initial interference effects and
changes with time for sulfate-thiosulfate and sulfate-parsulfate were
relatively small compared to those for sulfide. While sulfite appears
to undergo only modest increases (where measurable) this reflects, in
part, the apparently rapid oxidation occurring in the first hour before
analysis.
Based upon these findings, studies with the four methods employed only
sulfide and sulfite. With these two interferents, the results have
been calculated using four replications for the MTB method and two
replications each for the other three methods. Sulfide time dependence
results are shown in Table 13. The results with the modified Brosset
method (in dioxane) for fresh solutions are substantially different
from those in Table 12. For example, 28 ± .4 yg/ml observed sulfate
with 30 yg/ml sulfide added to 20 yg/ml sulfate compares to 21 ± .9
reported in Table 12. These findings imply very marked aging effects
occurring within one hour of solution preparation, making reproducibility
-ko-
-------
Table 12: The Effect of Aging of Interferent-Sulfate Solutions on Interference
Effects by the Modified Brosset Method (in dioxane)
Sulfate level, ug/ml
0
20
Time, hours 0-5& 70-75 500-550
Interferents levels,
^M 10 30
Interfere nt
Sulfide 2 _+ .7 2 _+ .7
Thiosulfate 1 ± .7 5 ± **
Persulfate 2 _+ .k k +_ .3
Sulf ite U _+ .6 11 ± .5
10 30
9 + -U 25 jt .3
NDb ND
ND ND
81.5 16 +; .2
10 30
> 30° > 30°
3± -7 7± -5
7± .3 11 ± .U
12 1 .3 > 30C
0-5a 70-75 500-550
10 30
20 ± .5 21 _+ .9
21 i .6 2U _+ .2
20 i .5 20 ± .8
23 _+ .9 28 ± .2
10 30
28 ± .2 29 _+ 1
ND ND
ND ND
27 _* .3 >30C
10 30
> 30° > 30C
23 jf .8 27 _+; .2
2k _+ .6 27 j: .u
28 _+ .7 > 30°
More exact solution ages:
Sulfide
Sulfite
Thiosulfate
Persulfate
determined.
0.5 hours
1.0
5-0
5.0
cRange of method exceeded.
-------
Table 13: The effect of aging of sulfide-sulfate solutions on interference effects "by four methods
(20 f/g/ml added sulfate)
Method
Sulfide level
Kg/ml
Observed S04~
0-3 hours0
Observed So4~
h8 hours
Mg/ml
a
wg/ml
or
AIHL
10
21. U
.1
30.8
.1
Mlcrob
30
28.8
2.1
39-6
.2
Bros set,
10
20.6
.1
28.6
.2
acetone
30
25.6
.1
29.2
.2
Brosset, dloxane
10 30
21.5 28.0
.1* .k
> 30 > 30
Turbidimetrlc
10
22.8
.3
31.6
.U
30
25. U
1.2
5U.O
9
Technlcon MTBa
10
18.3
2.7
2T.3
.7
30
20.5
1.3
37.6
U.6
aData reduction by linear regression
See Table 11, footnote a
More exact solution age at time of analysis:
AIHL micro: 1 hour
Brosset (both
modifications): 1
Turbidimetric: 1
MTB: 3
-------
difficult. The interferents effects tabulated for sulfide in Table 11
at 20 yg/ml are the means of results for the 0-1 hour old solutions
reported in Table 12 and 13.
Of the four methods with aged sulfate-sulfide solutions the modified
Brosset method in acetone and the MTB procedure yielded the least inter-
ference effects with aged sulfate-sulfide solutions. The greatest
interference was found with the turbidimetric method.
The results for sulfite are shown in Table 14. Comparing these results
to those in Tables 7-9 reveals generally similar findings. However,
the present data indicate that the AIHL microchemical method is subject
to the greatest interference effect by sulfite whereas previously this
was observed for the turbidimetric method. The MTB method demonstrated
the least interference effect by sulfite.
Finally, comparing aged solutions of sulfide and sulfite with sulfate,
greater interference was found with sulfide. The generally smaller
changes between 0-3 and 48 hours for sulfite, and the quite substantial
initial interference effects suggest faster oxidation of sulfite to
sulfate than is found with sulfide.
D. Recommendations to Eliminate Interferences
The methods used in the present study are affected by interferents
-43-
-------
Table lU: The effect of aging of sulflte-sulfate solutions on interference effects by four methods
(20 fig/ml added sulfate)
Method
Sulfite level
Mg/ml
Observed 80^
0-3 hours
Observed 650
b8 hours
Mg/ml
CT
Mg/ml
CT
AIHL Microb
10 30
26.8 38A
.k .2
28.9 39-^
.3 -1
Bros set,
10
21.8
.k
26.0
.1
acetone
30
27.7
.1
29.0
.1
Bros set,
10
26.0
1.1
27.8
.8
b
dioxane
30
> 30
> 30
Turbidimetric
10
21.5
.1
25.2
.1
30
21.8
5
3^.6
.U
Technicon MFE*
10 30
21.1 2U.2
.5 i.o
23.U 3^.7
.k .6
aData reduction by linear regression
See Table 11, footnote a
°More exact solution age at time of analysis:
AIHL micro: 1 hour
Brosset (both
modifications): 1
Turbidimetric: 1
MTB: 3
-------
extracted from both the filters and particulate matter samples. As
previously demonstrated these interferents may affect the results from
the four methods quite differently. In the following, possible means
to overcome interference will be discussed.
1. Turbidimetric Method
Because of the substantial sample requirement (e.g. >^ ca. 20 ml of
20 pg/ml sulfate), high volume samples are usually necessary to
permit use of this method. Filter materials commonly available for
this purpose are glass fiber and cellulose. Preliminary findings
indicate that glass fiber filters release substantial amounts of
soluble silica which, upon subsequent acidification and addition of
BaCl2, co-precipitates with the BaSCty as silicate and acts as a
nucleating agent. As a result the apparent blank sulfate value for
a glass fiber filter may be anomalously high. Furthermore we
believe such apparent blank values to be dependent on the total
sulfate in the extract, the apparent sulfate ascribable to the
filter decreasing with increasing total sulfate. Therefore, calib-
ration curves for sulfate determination for samples collected with
glass fiber filters should-be made using blank filter extracts for
making up the calibrating solutions.
-1*5-
-------
2. MTB Method
The bright methylthymol blue-Ba complex, upon reacting with sulfate
in alkaline solution, splits off the grayish methylthymol blue dye
and forms a suspension of BaS04. The positive absorbance shift at
460 nm (orange region of the spectrum) is used for quantitation of
sulfate. Since many atmospheric particle extracts contain sub-
stances absorbing appreciably at this wavelength, especially at
higher values of pH, this produces a positive interference. The
significance of this effect was evaluated, as discussed in Section
II A4, for a group of samples from Riverside, California. In this
case this interference represented less than one percent of the
observed sulfate. Thus it is unclear if this source of interference
would be sufficient to warrant correction. With strongly absorbing
extracts, such interference could be eliminated by using a dual
channel Technicon system replacing the reagent by a colorless sol-
vent, e.g. 80% alcohol, in one channel or by subtracting blank
values obtained in a subsequent run of a single channel instrument,
without the reagent.
3. AIHL Microchemical Method
This method is significantly affected by calcium at levels found in
particulate matter extracts. It was found that this interference,
could be eliminated using two approaches: ion-exchange paper discs,
-1*6-
-------
as in the modified Brosset method or addition of citric acid to the
reagent solution (^ 0.2 g/liter). With the latter approach, ratios
of Ca /SO^" up to 4 could be tolerated with only a slight loss in
accuracy. No loss was experienced up to a ratio of one. The incon-
venience of the long reaction time (30 min. to an hour) can be
reduced to 10 min. by forming a suspension of 63864 corresponding
to 3 yg/ml in the reagent. The reagent containing both corrections
extends the range from the present 6-12 to 1-14 yg of sulfate.
4. Modified Brosset Method
The AIHL modification of this procedure employing paper discs coated
with ion exchange resin has significantly simplified the method and
reduced the sample requirement. Nevertheless, the sample volume
needed remains relatively high (ca. 3 ml) compared to a maximum of
1 ml with the AIHL method.
The only significant interferents are sulfur-containing anions.
Techniques for reducing the influence of these interferents have
not been investigated.
E. Summary and Conclusions
Based upon the interference study data the following observations are
made:
-------
1. The modified Brosset method is the technique least subject to inter-
ference effects with the interferents studied.
2. Sulfide and sulfite are generally strong positive interferents with
all of the methods studied. While very fresh solutions of these
anions with sulfate display reduced effects, these conditions are
considered unlikely to be relevant to extracts of atmospheric par-
ticles.
3. Persulfate can be a source of significant interference. However,
its presence in ambient air particulate matter is considered unlikely.
4. Colloidal clay, as expected, is a strong interferent in the tech-
nique relying on light scattering for quantitation, viz., the
BaCl2 turbidimetric method. More surprising is its substantial
interference in the MTB and AIHL microchemical methods with both
positive and negative interference.
5. The only species showing a consistent trend in the direction (or
sign) of the interference effect are as follows:
803"" (always positive)
S208~ (always positive)
2
Ba (always negative)
Barium is invariably a negative interferent since it reacts essen-
tially irreversibly with sulfate. The possibility of a significant
-------
barium extraction from glass fiber filters remains to be evaluated.
6. p-Benzoquinone is a significant interferent only for the MTB method
which employs a wavelength of 460 nm for quantitation, at which
yellow solutions absorb significantly. Since aqueous extracts from
ambient air particulate samples will often be yellow, small positive
errors by this method are possible unless results are corrected for
sample blanks.
7. Phosphate exhibited, at most, a minor effect. In only one case, the
modified Brosset method in acetone, was the interference effect
negative and this result was only slightly beyond experimental un-
certainty (i.e. beyond la). An acidic medium would be expected to
minimize interference by barium phosphate precipitation. Neverthe-
less, comparing the three techniques employing acidic conditions
(the AIHL microchemical, modified Brosset and turbidimetric methods
with the MTB procedure which is run in basic solution, no signifi-
cant differences were observed. The maximum interference was about
5%.
8. While cation exchange resins or complexing agents can effectively
reduce or eliminate interference from cationic interferents elimina-
tion of effects of anionic interferents is more difficult.
9. It is difficult to assess the impact of these findings on the accuracy
-1*9-
-------
of 864 analysis without knowledge of the concentration of these
potential interferents in aqueous extracts from atmospheric parti-
culate matter. For example calcium exhibits only a minor effect
in the present study, but it is found in relatively high concentra-
tion from water extraction of blank glass fiber filters and may
also be extracted in significant amount from particulate matter.
Thus its influence on the accuracy of sulfate determination may be
substantial, especially with the AIHL micromethod.
-------
V. PRECISION AND INTERMETHOD COMPARISON OF THE FOUR SULFATE PROCEDURES WITH
ATMOSPHERIC SAMPLES (High Volume Samples)
A. Description of Experiment
To obtain samples sufficient to permit replication and analysis by the
four methods, high volume samples were obtained from three sites, St.
Louis, MO; Durham, NC; and Pasadena, CA. Four 24-hour filter samples
were collected at each location plus two filter blanks which were used
to adjust the data reported here.
To conduct an intermethod comparison with techniques differing widely
in working range and for which the available sample was quite limited,
a scheme was adopted to provide aliquots of appropriate concentration
and volume for each method. One-inch discs were cut from each hi-vol
filter, extracted in 5 ml of H20 and analyzed by the AIHL microchemical
method to establish the available sulfate in each sample. Based upon
these results, the remainder of each filter was then extracted for 90
minutes in about 80 ml of boiling water under reflux and filtered.
Each sample extract was diluted sufficiently to obtain a concentration
of about 20 yg/ml sulfate except for filter blanks. These solutions
were analyzed directly by the MTB and turbidimetric methods. For the
The extraction procedure consisted of immersion in water for two hours in
a sealed test tube at 80°C.
-51-
-------
modified Brosset and AIHL microchemical methods, further dilution into
appropriate working ranges was necessary. All determinations were
conducted with three replications except with the modified Brosset
method. For the latter method, analysis was restricted to two replica-
tions because of insufficient solution.
B. Analytical Precision
Results of the measurements for all samples, corrected for blanks are
shown in Tables 15A, 15B and 15C.
Data for the MTB procedure are presented as obtained by working curves
fit with both a linear and a third order regression line. Each deter-
mination shown represents the mean of three replications (two replica-
tions with Brosset) ± one sigma. The variability of the internethod
mean value reflects the range of results by the four methods not the
variability of the individual mean values. The latter, however, have
been pooled for each method over the four samples from each location
and the resulting pooled sigma values are used to "estimate the precision
of each method.
The precision of all methods as measured by coefficients of variation
was 5% or better. Within this range, the most precise values at all
locations were obtained with the modified Brosset and AIHL micromethods
but these results were more variable with sampling location than was
-52-
-------
Table 15A: Sulfate analysis of atmospheric hi-vol samples by four methods, St. Louis, MO (jug/m3)£
i
\j)
U)
Sample No.
1/I23067
M23068
1C 3072
M23073
Mean
a pooled
Brosset
AIHL Micro acetone
33-7 + -5 37-0 + .2
3h.h + .5 37-6 + .oh
25.0 + .k 28.2 + .3
23.1 + .5 25.8 + .5
29.1 32.2
0.5 0.3
i 1.6 1.0
Turbidimetric
36.3 + 1.8
37.3 ± 1.3
27.7 + -9
25.1 + 1.1
31.6
1.3
U.2
Technicon Technicon, MTB
MTB, linear 3rd order
regression regression
37.8 + 1.0 38.5 + .9
39.8 + .8 hO.h +_ .8
27. 1+ + 1.1 28.3 + 1.2
25.3 + 1.2 26.2 + 1.2
32.6 33.^
1.0 1.0
3.2 3.1
b
Interme thod
mean value (C.V.,%
36.1 + 1.9 (5.2)
37-3 + 2.2 (6.0)
27.0 + 1.5 (5.M
2U.8 + 1.3 (5.^)
Results are mean values + 1 a
Mean excluding Technicon result by 3rd order regression. The cr value and coefficient of variation shown
reflect the range of results by the four methods, not the pooled variabilities of each mean.
-------
Table 15B: Sulfate analysis of atmospheric hi-vol samples by four methods, Durham, NC (ug/m3)a
Sample No. AIHL Micro
N23069 11.8 + .02
N23070 12.3 + .2
N23071 11.2 + .1
N23066 lk.0 + .7
Mean 12.3
CT pooled 0.36
Coeff. of Var. (%) 2.9
Bros set
acetone Turbidimetric
11.9 + .1 n.8 + .h
12.5 + .k H.9 + .5
11.6 + .1 11.1 + .7
Ik.h- + .2 lU.O + .8
12.6 12.2
0.2U 0.62
2.0 5.1
Technicon
MTB, linear
regression
12.6 + .3
13.5 ± .1
12. k + .k
lU.l + .6
13.2
0.39
3.0
Technicon, MTB
3rd order
regression
13.0 + .3
13.9 ± .1
12.8 + .k
Ih.6 + .6
13.6
o.ko
2.9
Intermethod
mean value (C.V.,$)
12.0 + .h
12.6 + .7
11.6 + .6
ll+.l + .6
(3.7)
(5.6)
(5.5)
(U.O)
8. a
Results are mean values +1
Same as in Table 15A
-------
Table 15C: Sulfate analysis of atmospheric hi-vol samples by four methods, Pasadena, CA (jug/m3)c
1
VJ1
VJ1
1
o ample No.
C23063
C23061+
C23076
C23077
Mean
cr pooled
Coeff . of Var.
AIHL Micro
6.53 ± .09
3-51 ± .09
2.32 + .08
3.15 + .I1*
3.88
0.10
(%} 2.6
Brosset
acetone
6.85 + .08
3.72 + .22
2.17 + .05
3.13 ± .07
3.97
0.13
3.2
Turbidimetric
6.1*5 + .20
3-39 ± .15
2.12 + .11
3.02 + .16
3.75
0.16
k.2
Technicon
MTB, linear
regression
6.7k + .28
3.70 + .20
2.2U + .13
3.39 ± .10.
U.02
0.19
U.7
Technicon, MTB
3rd order
regression
7.06 + .28
3-93 + .21
2.^3 + .lU
3.62 + .10
U.26
0.19
U.6
b
Intermethod
mean value (C.V.,%
6.62 + .23 (3.^)
3.59 ± -33 (5.6)
2.20 + .23 (5.2)
3.21 + .& (5.9)
Results are mean values + 1 a
Same as in Table 15A
-------
experienced with the other methods. The turbidimetric method yielded
the poorest precision except with the California samples.
It should be emphasized that in this study employing high volume samples,
solutions were diluted to yield sulfate values in the optimal range of
each method. In routine use with samples yielding a wide range of sul-
fate concentrations, somewhat poorer precision can be expected.
C. Equivalence of Methods
Considering next the degree of agreement between the methods, the range
in values for all methods was approximately 10% of the intermethod mean
for each sample. The coefficient of variation of the intermethod mean
was surprisingly constant, ranging between 3.4 and 6.0%, considering
the substantial variation in expected composition of the samples. While
still yielding agreement within about 10%, the St. Louis samples led to
the poorest agreement. In this case, the AIHL micromethod yielded con-
sistently low values relative to the other techniques. One aqueous
extract from St. Louis was subsequently shown to contain relatively high
calcium concentrations (about 32 yg/ml) providing a likely explanation.
In comparing the methods of data reduction for the MTB method we note
that all solutions analyzed were in the approximately linear portion
of the working curve (i.e. 10-50 yg/ml) although the filter blanks were
lower. Even in the "linear range", however, a third order line provides
-56-
-------
a somewhat better fit to the S-shaped curve. The resulting concentra-
tions (corrected for blanks) differ by up to 10% with third order results
always higher. The difference diminishes with increasing concentration
(in yg/nP) . Data by linear regression are invariably closer to results
by the three other methods. We conclude that, within the approximately
linear range, linear regression is the technique of choice.
D. Interference Effects
From each of the 12 high volume glass fiber filter samples, a 1" disc
was extracted and analyzed by the AIHL microchemical method. Three
aliquot sizes were taken from the aqueous extract to insure samples
in the optimal working range of the method, corresponding to absorbance
values 0 . 3 < A < 0 . 7 .
In contrast to the samples from Durham and Pasadena the apparent
values from the St. Louis samples proved to be strongly dependent on
aliquot size as shown by the example (M23067) :
Aliquot size, ml Absorbance SOA
0.10 0.612 28.6
0.20 0.718 16.9
0.40 0.640 7.5
Thus aliquot size dependence was observed even within the working range.
The influence of aliquot size suggested analytical interference as the
-------
cause. Since calcium was suspected as a probable interferent, the
aqueous extracts from one St. Louis sample and one Pasadena sample were
analyzed for calcium by flameless atomic absorption. Other interfer-
ents were possibly present but were not determined.
Sample Site (No.) Ca+2 yg/ml S0&= yg/ml** Ca+2/SOA=
St. Louis (23067) 32 111 0.29
Los Angeles (23063) 4 24.2 0.17
Q __
Based upon the observed ratio of Ca /S04 and previous interference
studies a negative error of about 15% in the sulfate value for the St.
Louis sample would be expected when operating in the working range of
the method.
The interference effects were further explored by subjecting portions
of each extract to the cation exchange treatment (Reeves Angel, strong
acid form) employed with the modified Brosset procedures. Table 16
lists selected results from this study comparing results with and with-
out ion exchange treatment. The data compared for a given filter were
obtained at approximately equivalent aliquot sizes.
The results suggest little effect of ion exchange on the California
By the AIHL micromethod on the extract obtained as described above. The
value for St. Louis obtained with an 0.1 ml aliquot, is considered a
minimum.
-58-
-------
Table 16: THE INFLUENCE OF ION EXCHANGE TREATMENT ON HIGH VOLUME FILTER
SAMPLES ANALYZED BY THE AIHL MICROCHEMICAL METHOD
S04
Sample ID
N 23069HV
N 23070HV
N 23071HV
N 23066HV
M 23067HV
M 23068HV
M 23072HV
M 23073HV
C 23063HV
C 2306UHV
C 23076HV
C 23077HV
With
Ion Exchange
10.7
11. It
12.9
1U.3
26.5
26.9
29.lt
22. k
7-50
3-75
2.55
3-53
Without
Ion Exchange
9.62
10.U
9-95
12. k
25.2
2lt.lt
29.7
21.9
7.56
3.95
2.57
3.51
Ratio
With/Without
Ion Exchange
1.11
1.10
1.30
1.15
1.05
1.10
0.99
1.02
0.99
0.95
0.99
1.01
Results shown are for a single trial in a.11 cases.
-59-
-------
samples. However, with samples from North Carolina and St. Louis, the
results were generally higher after ion exchange supporting the signi-
ficance of cationic interferents in these cases.
E. Accuracy of the Sulfate Methods by Standard Additions
The design of this study was complicated by the difference in working
ranges for the four methods and the differing concentrations of sulfate
available from the samples for a given degree of dilution. High-volume
filter sample extracts, diluted to approximately 20 pg/ml sulfate, were
the starting points for the standard addition studies. The concentra-
tions of sulfate without standard additions were obtained from the pre-
ceding intermethod comparison. Table 17 summarizes the experimental
protocol followed.
Per cent recoveries are listed in Table 18 for each sample extract at
the two levels of sulfate addition. Similar results are also included
for the blank glass fiber filters extracts. All experiments were run
with two or three replications.
While these tables are useful for detailing the results, trends are
difficult to discern from the inevitable experimental scatter. Inter-
pretation is aided by plotting observed against added sulfate for each
sample by each of the four methods. Such plots are presented in
e\
Figures 9-22. In place of concentration units in yg/ml, yg/cm of filter
-60-
-------
Table 17: PROTOCOL FOR STANDARD ADDITION STUDY-HIGH VOLUME GLASS FIBER FILTER SAMPLE EXTRACTS
Method
ALHL Micromethod
Modified Brosset
Turbidimetric
Methylthymol Blue
Volume ca. 20 ug/ml
S04~ extract, ml
1.00
1.00
20.0
2.00
Volumes 1 jug^/ml
standard B 04"
added, ju.1
10 and 30
10 and 30
200 and 600
20 and 60
Factor for Final approximate
further dilution concentration range
prior to analysis analyzed, /ag/ml
5 6-10
U and 7 7
none 30-50
none 30-50
o\ All extracts were initially diluted to obtain about the same final concentration, 20 /jg/ml S04 .
H
Solutions were divided for analysis by the four procedures following the standard additions.
-------
St. Louis, MO Samples
tmple AIHL Micro
No. 10 ng/ml S04" 30 jug/ml S04'
= 3067
23073
23069
23070
23071
23066
23063
23061+
23076
23077
PA 1
PA 2
76.3 + 30.5
98.0 + 13.9
83-5 + H.2
91+.0 + 11.5
87.6 + 13-9
83.6 + 11.3
88.6 + 10.5
81+.3 ± 8.6
92.8 + 13-5
89.1 + 12.1+
87.9 + n.6
87.2 ±12.7
90.7 + 2.8
92.2 ± 2.7
99.6 + 5-0
99.1 + 7.0
99-6 + 5.7
99-3 ± 5.1
96.3 + 3.6
102 + 1+.1+
102 +6.7
101 ± 5.0
103 + 1+.2
97.8 + 5-2
99.6 + 5-7
102 ± 7.2
90.7 + 2.1+
9!+. 8 + 2.7
Modified Brosset Acetone
10 jug/ml S04~ 30 Mg/ml S04
10
123 + *+.6
137 + U.7
110 + 8.3
129 ± 5-3
ii+o + 5.3
137 +7-1
ll+2 + 5-7
119 ±2.1+
133 +5-0
123 + 8.8
131 + 3-^
131 ± *+.9
101 + 2.8
ll+6 + it-.O
115 +2.0 100 + 7-9
115 ±2.7 102 ± 5-7
108 + l+.O 90.8 + 3-8
106 ± 1.8 92.1* ± 7-2
Durham, NC Samples
105 + 3.0 102 + 3-9
no + 2.3 91,3 + 5.2
109 + 1.1 39.1 + 6.9
151 ± 1.1 100 ± 7.1+
Pasadena, CA Samples
103 + 2.5 9!+. 7 +8.7
106 + 5-1 99- ^ + 5-9
97.8 + 3.2 81+.9 + 3.8
10k + 1.7 98.9 ± 7.0
Blanks
102 + 1.3 80.8 + 12.7
95.2 + 3.2 82.7 ± n.9
Turbidimetric
S04~ 30 ug/ml S04'
10
Technicon MTB
S04~ 30
S04'
103 ±2.5
99-5 ± 2.2
101 ± 2.5
103 + 2.1+
99-4 ± 3-0
100 ±2.0
96.5 ±1.3
100 + 5-0
116 + 6.1
123 + 13-0
116 + 12.8
lll+ ± 15.!+
90.5 + 5-2
96.7 + 3.7
107 + 5.9
108 ± 15.7
H7 + 15-7
116 + 9.8
112 + 13.1+
108 ± 15.!+
98.1+ + 8.5
9!+. 6 ± 11.1
121+ + U.a
12.2 + 3.7
122 + 16.1+
119 ± 6.9
108 + 10.5
109 + 0.8
113 + 2.9
118 ± 13-2
120 + 8.2
115 + 2.7
87.0 + 2.0
109 ± 1.9
101+ + 1.5
103 ± 1-9
-------
AIHL
80 60 40 20 0 20 40 60 80 100
BROSSET
SAMPLE
N23066
200 -,
180 -
160 -
140 -
120 -
T3
§ 100 -
o
CNJ 80 -
-? »
^ 60 7
40 /-
/20 -
/
/
/
/
/
/
/ b = 1.55
a = 59.6
80 60 40 20 0 20 40 60 80 100
ON
U)
TURBIDIMETRY
b = 1.03
a = 63.5
MTB
I I 1 T | I 1 | I
80 60 40 20 0 20 40 60 80 100
80 60 40 20 0 20 40 60 80 100
r\ f\
n added yKg/cm added
RECOVERY OF SULFATE FROM STANDARD ADDITIONS TO ATMOSPHERIC SAMPLES
Figure 9
-------
A1HL
120 -
ON
15-
BKOSSKT
SAMPLE
N23069
TURBIDIMETRY
60 40 20 0
b =- 1.10
a = 56.4
60
40
cm2 added
20 0 20
yt
-------
AIHL
BROSSET
I I I I I I I
60 40 20 0 20 40 60 80
TURBIDIMETRY
v_n
i
b = 1.00
a = 54.6
I I T ~\
60 40 20 0 20 40 60 80
ug/cm added
SAMPLE
N23070
MTB
b = 1.07
a = 61.8
I I I I I I I I I
60 40 20 0 20 40 60 80 100
140 -,
b = 1.10
a = 61.2
60 40 20 0
20 40 60 80 100
added
RECOVERY OF SULFATE FROM STANDARD ADDITIONS TO ATMOSPHERIC SAMPLES
Figure 11
-------
AIHL
BROSSET uo -,
120-1
SAMPLE
N23071
80 60 40 20 0 20 40 60 80
2
added
b = 1.05
a = 58.6
I I \
80 60 40 20
I I I 1
0 20 40 60 80
2
m added
ON
TURBIDIMETRY
b = 1.01
a = 50.6
MTB
b = 1.13
a = 56.8
80 60 4b 20 0 20 40 60
n added
6'0 40 20 0 2'0 40 60 80
added
RECOVERY OF SULFATE FROM STANDARD ADDITIONS TO ATMOSPHERIC SAMPLES
Figure 12
-------
AIHL
ON
-q
i
300
TURBIDIMETRY
BROSSET
300 _
SAMPLE
M23067
150 100 50 0 50 100 150
150 100 50 0 50 100 150
n2 added yUg/cm^ added
RECOVERY OF SULFATE FROM STANDARD ADDITIONS TO ATMOSPHERIC SAMPLES
Figure 13
-------
AIHL
BROSSET
SAMPLE
M23068
150 100 50 0 50 100 150200
i
ON
CD
i
TURBIDIMETRY 300
i i i i i i i i
150 100 50 0 50 100 150 200
rj
XXg/cm added
I i r IT i l
150 100 50 0 50 100 150 200
MTB
i i i i i i
150 100 50 0 50 100 150 200
/t(g/cm added
RECOVERY OF SULFATE FROM STANDARD ADDITIONS TO ATMOSPHERIC SAMPLES
Figure
-------
AIHL
T3
§
O
M-(
CM
e
o
00
350
300
250
200
150
100
/50
BROSSET
b = 1.00
a = 112
SAMPLE
M23072
150 100 50 0
I
50
i I 1
100 150 200
1
TD
C
3
O
U-l
CM
6
o
w>
x
/
1 1
JJU -
300 -
250 -
200 -
150 _
100 /-
y/50 -
1
/
/
/
/
/
/
/ b = 1.
a = 13
liil
/4g/cm added
O\
VO
TURBIDIMETRY 300 n
150 100 50 0
2
/
-------
AIHL
250-,
BROSSET
SAMPLE
M23073
150 100 50
50 100 150 200
/Kg/cm added
250-i
150 100 50
T
0 50
2
i i
100 150 200
added
o
TURBIDIMETRY
250
MTB
150 100 50
50 100
150 200
/t^g/cm added
250 _
150 100 50 0 50 100 150 200
s\
Mg/cm added
RECOVERY OF SULFATE FROM STANDARD ADDITIONS TO ATMOSPHERIC SAMPLES
Figure 16
-------
AIHL
b = 1.03
a = 20.4
20
H
i
IDIMETRY
50 _
-a 40 -
c
3
O
IJ ^
^ 30-
!/
T i
20 10 0
/
/
/ b = 1.00
>T a = 20.3
/
10 20 30 40
BROSSET
SAMPLE
C23063
b = 0.99
a = 23.5
MTB
b = 1.20
a = 21.2
/* T7
/
-------
AIHL
BROSSET
50 -
b = 0.98
a = 11.1
b = 1.00
a = 10.9
TURBIDIMETRY
50 -
SAMPLE
C23064
cm added
50-
20
RECOVERY OF SULFATE FROM STANDARD ADDITIONS TO ATMOSPHERIC SAMPLES
Figure 18
-------
AIHL
BROSSET
SAMPLE
C23077
40
I
20
CO
i
TURBIDIMETRY
MTB
13
C
O
M-l
E
CJ
40 -
30 -I
20 -J
10 -,
b = 1.01
a = 10.4
10
\
10
I
20
I
30
I
40
I
20
I
10
0 10 20 30 40 20 10 0 10 20 30 40
n2 added /^g/cm^ added
RECOVERY OF SULFATE FROM STANDARD ADDITIONS TO ATMOSPHERIC SAMPLES
Figure 19
-------
AIHL
b = 1.00
a = 7.30
BROSSET
SAMPLE
C23076
30
TURBIDIMETRY
MTB
I
20
10 0 10 20 30 20 10 0
/Mg/cm2 added jL(g/cm^ added
RECOVERY OF SULFATE FROM STANDARD ADDITIONS TO ATMOSPHERIC SAMPLES
Figure 20
-------
AIHL
T3
C
O
t>0
7-
6-
5-
4-
3-
2-
1-
SAMPLE
EPA 1 (Blank)
Slope = .91
Intercept = .32
\
8
Mg/cm added
VJl
I
TURBIDIMETRY
C
3
O
CN)
o
60
Slope = .81
Intercept = .82
)SSET 7 -,
6-
5-
c
1 4~
c,a 3-
o
60 2
6 4 2 (
/
/
/
/
y Slope =
/ Intercep
32468
MTB
- 0.77
^
yt(g/cm added
RECOVERY OF SULFATE FROM STANDARD ADDITIONS TO ATMOSPHERIC SAMPLES
Figure 21
-------
AIHL
7 _
BROSSET
TD
C
D
O
^e
o
x
6 -
C
4 -
3 -
2 _
1 -
III!
/
/
/
/
/ Slope =
/ Interce
/
/
III!
SAMPLE
EPA 2 (Blank)
20
2
added
TURBIDIMETRY
7
= 0.22
Slope = .83
Intercept = .65
\
8
MTB
o
C
o
o
60
7 -
6 -
5 -
4 -
3 -
2 -
\2
Slope = .887
Intercept = 1.283
0
2
added
1 1
8 6
7 _
6 -
-o 5 ~
c
1 4-
^a 3 -
o
60 2
s;
7
/
/
/ Slope =
f- Interce
/
II 1 1 1 1
42 024 68
rj
/i(g/cm added
RECOVERY OF SULFATE FROM STANDARD ADDITIONS TO ATMOSPHERIC SAMPLES
Figure 22
-------
are shown since this was an easier parameter to determine.
While these plots often appeared non-linear the number of data points
was judged inadequate to attempt anything but linear regression. For
such plots a least squares slope (b) of 1.00 represents a mean recovery
of 100%. These plots may be examined for their mean recoveries, the
degree of scatter about the least squares line, and non-linearity.
Finally, Table 19 summarizes the least squares slopes and the results
of an analysis of covariance to test for significant difference in
slopes for the samples from a given site.
From these data and displays we make the following observations:
1. Comparing recoveries at the low and high levels of added sulfate,
the modified Brosset procedure, with only one exception, yielded
lower recoveries at the higher level. The remaining techniques
generally exhibited the opposite trend. Such observations would
result in non-linear plots for sulfate found against sulfate added.
The plot for sample NC23066 by the Brosset procedure was tested and
found by analysis of variance to be significantly non-linear (F(l,5)
> 75).
2. Mean recoveries by the AIHL microchemical and turbidimetric methods
were close to 100% for all samples at all sites.
-77-
-------
Table 19: MEM FRACTIONAL RECOVERIES OF SULFATE WITH STANDARD ADDITIONS
Pasadena, CA
a
Sample ID
C-23063
6k
76
77
AIHL Micro
1.0
1.0
1.0
1.0
Turbidimetric
1.0
1.0
1.0
1.0
Technicon, MTB
1.2?
Modified
Brosset,
Acetone
1.0
1.0
0.9
1.0
M-23067
68
72
73
1.0
1.0
1.0
1.0
St. Louis, MO
1.0
1.0
1.0
1.0
1.2
1.2
1.2
1.2
1.1
1.1
1.1
1.0
N-23069
70
71
66
1.0
i.o
1.0
1.0
Durham, NC
1.0
1.0
1.0
1.0
1.1
1.1
1.1
1.2
i.c£
iio?
i.?
a
'As measured by the least squares slope of plots of sulfate found against
added sulfate.
Recoveries within the set of four samples are significantly different at
level by analysis of covariance.
-78-
-------
3. Recoveries by the MTB method were 10-20% high in 63% of the experi-
ments, 5-10% high in 25% of the experiments and -5 to 5% high in
the remaining. Considering mean recoveries for each sample, recov-
eries were, with one exception, high by 10-20%.
4. Recoveries by the Modified Brosset method were 10-50% high in 66%
of all experiments with remaining results between -2 and 10% high.
Mean recoveries by this method ranged from 90 to 110%. Mean recov-
eries were close to 100% with Pasadena samples and generally above
100% with St. Louis and Durham samples.
5. Recoveries from spiked blank filter extracts are somewhat lower than
those from the particulate samples. Comparing the two procedures
with ion exchange pretreatments (MTB and modified Brosset) to the
two lacking such treatment, recoveries with the blanks are system-
atically higher with pretreatment. These findings suggest the
importance of cationic interferents, such as calcium, extracted from
glass fiber filters. Failure to observe similar trends with the
atmospheric samples is not surprising considering the possibility
of positive interferents extracted from the samples.
6. Comparing the present results to the intermethod comparison reported
in Table 15 the Technicon MTB has, in both cases yielded somewhat
higher sulfate values. Furthermore these results are consistent
with preliminary findings summarized in II A and Figure 1.
-79-
-------
F. Summary and Conclusions
Employing high volume filter samples the four wet chemical methods yield
agreement within about 10% for all samples. Within this range the poorest
agreement was found with St. Louis samples; this appears to be relatable
to the influence of cationic interference. The degree of equivalence
found, in spite of the varying sensitivity of the methods to interfer-
ents, suggests that with such large samples, interference effects are
minimal. The coefficient of variation of the four methods was 5% or
less.
While agreeing within about 10% the modified Brosset and MTB methods
gave consistently higher values in analysis of the high volume filter
samples compared to the remaining procedures. Standard addition results
also demonstrated more than 100% apparent sulfate recoveries by these
methods in > 80% of all experiments implying that a systematic positive
bias is the cause of the higher results by the methods with the atmos-
pheric samples.
The degree of agreement found suggest that the choice of a method from
among these four, for analysis of high volume samples can be based upon
such factors as cost per determination or experimental convenience. The
restricted range of both the modified Brosset and ATHL micromethod
clearly makes them inappropriate for consideration with high volume
samples.
-80-
-------
VI. EQUIVALENCY OF WET CHEMICAL AND X-RAY FLUORESCENCE METHODS AND INFLUENCE
OF SAMPLING DESIGN WITH ATMOSPHERIC LOW VOLUME FILTER SAMPLES
A. Description of the Experiment
Details of the sampling design were discussed in the introduction. This
design permits sulfate analysis method comparisons employing both glass
fiber and Fluoropore filters each with two particle size fractions (i.e.
0-20 ym and 0-2 ym). Furthermore, since all sampling was conducted
simultaneously, results obtained on glass fiber and Fluoropore may be
compared as well as the influence of sampler (i.e. High volume vs. low
volume).
Fluoropore filters mounted in plastic frames were initially analyzed by
XRFA at Research Triangle Park. Following shipment to Berkeley, 1"
(25 mm) discs were removed from the center of each filter and the samples
hand-carried for XRFA at the Lawrence Berkeley Laboratory. Many of the
1" discs curled up badly after their removal. Since the LBL technique
requires constant distance between sample and detector it was necessary
for the analyst to press the samples flat between glassline paper.
Therefore, the loss of some S in handling is likely. Such loss would,
of course, influence the subsequent wet chemical analyses as well.
Following XRFA, the discs were returned to AIHL.
Since the 37 mm Fluoropore filters had been glued into their plastic
-81-
-------
frames it was not possible to include the remaining portion of the
filters in the wet chemical determinations. The 1" discs were extracted
by the micropercolation technique into 10 ml H20. The 37 mm glass fiber
filters were received unmounted and the complete discs were extracted
by the micropercolation technique into 10 ml H20. The resulting solu-
tions provided sufficient sample for a single replication by each of
three methods: the AIHL microchemical, MTB and modified Brosset methods,
Accordingly, results are quoted without a statement of precision. How-
ever, the precision of the methods as established with the high volume
filter extracts (i.e. C.V. <_ 5%) represents a reasonable upper limit to
the precision to be expected with low volume samples.
B. Results
1. Comparison of Wet Chemical Sulfate Analysis
Tables 20 and 21 detail results for Fluoropore and glass fiber
filters, respectively, with filters identified by number and sam-
pling date. For ease in interpretation of results, the data in
Tables 20 and 21 have been pooled and recalculated relative to
results with the MTB method as shown in Tables 22 and 23. The
ratio of means are shown by sampling site for the four total
particulate (0-20 ym) and refined particulate (0-2 ym) samples.
Considering first the pooled Fluoropore results (Table 22) the MTB
values are consistently higher than those by the AIHL microchemical
-82-
-------
Table 20: SUMMARY OF LOW VOLUME FLUOROPORE FILTER StJLFATE DETERMINATIONS (ng/m sulfate)
Sample ID
Date Sampled
03
U)
I
NC-1C
NC-1D
NC-2C
NC-2D
NC-3C
NC-3D
NC-4C
NC-4D
MO- 1C
MO- ID
MO-2C
MO-2D
MO- 3C
MO- 3D
MO-4C
MO-4D
CA-4ATF
CA-4HIF
CA-6ATF
CA-6BRF
CA-7ATF
CA-7BRF
CA-8ATF
CA-8BRF
7/15M
7/15M
7/16/74
7/16M
7/17M
7/17M
7/18M
7/18/74
8/5/7*
8/5/74
8/6/74
8/6/74
8/7/74
8/7/74
8/8/74
8/8/74
12/2/74
12/2/74
12/7/74
12/7M
12/8/74
12/8/74
15/9/74
12/9/7^
Brosset
11.T
13 .^
11.2
8.55
8.84
14.7
11.1
12.1
23.8
16.0
33-9
20.1
25-3
24.8
29.8
28.1
9.06
9.00
3-71
5.16
3-62
1.72
3.32
2.41
Technlcon
9.7^
1U.8
17.0
13.4
12.2
13.9
15.8
15.8
31.6
10.6
37.2
13.2
25.8
23.6
31-7
30.1
6.5
3
3
2
AIHL Micro
XRFA (RTP)
XRFA (LBL)
99
63
44
2.36
3-19
3.01
10.8
13.9
14.
12.
11.
12.
14.
9
.4
.1
.1
.3
14.0
24.6
8.72
28.9
11.5
21.9
21
26
3
7
26.7
61
05
50
55
13
07
2.17
1.90
17.7 + 4.28
20.1 + 4.36
23.6 + 5.82
19.0 + 4.13
18.0 + 4.33
19.3 + 4.19
22.31 5-37
21.4 l 4.64
44.7 + 10.7
9.52+ 2.07
47.1 + 11.2
11.5 + 2.52
31.0 + 7.40
18.9 + 4.09
40.91 9-73
32.01 6.90
8.09 + 2.06
6.58 + 1.52
4.16 + 1.23
3.47 + .90
1.71 + .53
1.63 + .61
3.33 + 1.02
2.99 + .81
9-42 + 1.9
12.0 + 2.4
13.4 + 2.7
10.3 + 2.0
9.81 + 2.0
11.1 -i- 2.2
11.6 + 2.3
12.5 12.5
26.0 + 6.5
5.67 + 1.1
25.9 16.5
7.86 + 1.6
20.9 + 4.2
12.2 + 2.4
24.7 + 6.2
19.2 1 3.8
4.23 + .84
4.68 + .93
2.28 H- .45
2.43 + .48
.78 + .15
.84 + .18
1.47 + .30
1.981 .39
1yg/m S expressed as S04
-------
Table 21: SIMMARY OF LOW VOLUME GLASS FIBER FILTER SULFATE DETERMINATIONS
(jug/m3 sulfate)
Sample ID
NC-lAa
NC-1B
NC-2A
NC-2B
NC-3A
NC-3B
NC-4A
NC-4B
MO-lAa
MO-IB
MO-2A
MO-2B
MO-3A
MO-3B
MO-^A
MO-4B
CA-4CTGb
CA-4DRG
CA-6CTG
CA-6CRG
CA-7CTG
CA-7DRG
CA-8CTG
CA-8DRG
Date SampI
7-15-74
7-15-74
7.16-74
7-16-74
7-17-74
7-17-74
7-18-74
7-18-74
8-5-74
8-5-74
8-6-74
8-6-74
8-7-74
8-7-74
8-8-74
8-8-74
12-2-74
12-2-74
12-7-74
12-7-74
12-8-74
12-8-74
12-9-74
12-9-74
Brosset
14.1
9.14-8
15-2
14.5
12.2
10.9
15.2
14.5
30.5
10.7
32.6
15.0
29.
22.
27-
.5
.7
.1
27.6
6.72
6.92
4.29
3-27
7-25
.98
.86
3-
3.
2.40
Technicon
16.8
11.8
18.6
18.8
16.6
15.6
19-3
19.1
40.3
n.i
17.3
35.8
35.0
30.4
6.86
7.60
3.
3.
1,
1.
2.
57
40
74
64
78
1.42
AIHL Micro
15.0
n.4
15.1
14.5
14.9
14.3
18.0
17-9
34.6
8.98
35.0
16.5
32.1
26.2
32.6
29-3
5.84
6.79
87
74
36
32
2.25
1.16
TTor North Carolina and Missouri samples A are total (0-20 /urn) and B
are refined (0-2 ^m) particle samples.
For California samples CTG indicates total and DRG, refined samples.
-84-
-------
Table 22: RELATIVE RESULTSFLUOROPORE FILTERS
a
Durham
St. Louis
Pasadena
MTB
Total 1.00
Refined 1.00
Total
1.00
Refined 1.00
Total 1.00
Refined 1.00
Modified
Brosset
.78 +
.84 +
.89 ±
1.15 +
1.22 +
1.11 +
.10
.09
.05
.14
.13
.13
AIHL
93 ±
.91 +
.81 +
.88 +
.70 +
.64 +
.04
.02
.02
.01
.08
.04
XRFA
1.49
1.38
1.30
93
1.07
.89
(EPA)
+ .08
+ .01
+ .04
+ .08
+ .10
+ .04
XRFA
.80 +
79 +
.77 +
.58 +
.54±
.60 +
(LBL)
.04
.01
.03
.04
.06
.05
Results are expressed as the ratio of the means of determinations on four samples.
Errors are calculated as the standard deviation of the ratio of two dependent
variables:
CTY OHr _2 cov(x,y) ~i
S.D/-") =v/var. (-^ and var. - = (-}£ [ -,
\y/ V \yj y \yj L x=
+ -.
This technique, we believe, eliminates the variability in the ratio due merely to
day to day changes in the sulfate concentrations for the four samples pooled.
-85-
-------
Table 23: RELATIVE RESULTSGLASS FIBER FILTERS (Low Vol)£
Durham
MTB
Total 1.00
Refined 1.00
Modified
Brosset
.80 + .02
.76 + .02
AIHL
.88 + .03
.89 + .05
St. Louis
Total 1.00
Refined 1.00
.76 + .03
.88 + .03
.85 + .Ok
.$k + .02
Pasadena
Total 1.00
Refined 1.00
1.18 + .25
.82 + .02
.85 + .03
as footnote a, Table 22.
-86-
-------
methods while the modified Brosset results were more variable and
highly site dependent.
Table 23 shows a similar comparison but using samples collected on
glass fiber filters. The MTB method yields results generally higher
than those by the modified Brosset and the AIHL microchemical method
while the latter two methods agreed well except for California samples,
2. X-ray Fluorescence Results
A comparison of XRFA results obtained at Research Triangle Park and
the Lawrence Berkeley Laboratory is shown in Table 24. While results
by the two laboratories are clearly highly correlated, they differ
by nearly a factor of two with LBL results lower. The difficulties
experienced at LBL because of the poor quality of the samples are
improbable sources of so constant a discrepancy in results.
Previous comparisons of LBL-XRFA sulfur with wet chemical sulfate
analyses used the AIHL micromethod with Gelman GA-1 cellulose ace-
tate membrane filters. The samples studied had been collected at
various locations in California's South Coast Basin. The mean ratio
wet chemical S04=/XRFA S as S04= was 1.01 ± .06 for 400 filters.3
In the present study the degree of agreement between XRFA and wet
chemical sulfate is significantly affected by choice of wet chemical
-87-
-------
Table 2k: COMPARISON OF EPA AMD LBL X-RAY FLUORESCENCE RESULTS ON
LOW VOLUME FLUOROPORE FILTERS
Sample ID
NC-1C
NC-1D
NC-2C
NC-2D
NC-3C
NC-3D
S(LBL)/S(EPA)
0.53
0.60
0.57
0.5^
0.55
0.58
0.52
0.58
MO-1C
MO-ID
MO-2C
MO-2D
MO-3C
MO-3D
MO-^C
MO-UD
0.58
0.60
0.55
0.68
0.67
0.6k
0.60
0.58
CA-lj-ATF
CA-1*BRF
CA-6ATF
CA-6BRF
CA-7ATF
CA-7BRF
CA-8ATF
CA-8BRF
Overall
0.52
0.71
0.55
0.70
0.1(6
0.52
0.1*
0.66
0.58
-88-
-------
method. When LBL-XRFA and the AIHL micromethod are compared, the
ratio of means S04=/XRFA S as SO^ =1.18. If MTB results are used
the ratio becomes even higher. Since XRFA results should be greater
than or equal to wet chemical findings (i.e. the ratio should be <
1.0) we conclude, based upon the present findings as well as prior
experience, that the current LBL results are too low.
Considering the comparison of RTP-XRFA and wet chemical analyses,
again the degree of agreement depends markedly on the choice of wet
chemical method. When compared against the MTB procedure as in
Table 21 Pasadena results are within about 10% of those obtained
wet chemically. At Durham, results are significantly higher by
XRFA with mixed findings at St. Louis.
Considering the influence of particle size on the XRFA results, as
shown in Table 21, in all cases the RTP-XRFA results, relative to
MTB, were higher with total particulate than with refined particu-
late samples. On average the ratio of total/refined XRFA sulfur was
1.23 which is suspiciously close to the ratios of correction factors
used in correcting the RTP-XRFA for self absorption. For total
filters 19% self absorption was assumed by T. Dzubay compared to
15% for refined samples, yielding a ratio of corrections of 1.27.
Thus the apparent differences in agreement between RTP-XRFA and MTB
for total and refined samples may not be real.
-89-
-------
3. The size distribution of sulfate
While not a direct objective of this study the proportion of refined
(0-2 ym) to total (0-20 ym) particulate sulfate may be seen in the
data listed in Tables 20 and 21. Results have been pooled by sam-
pling site for the Fluoropore filters as shown in Table 25. The
striking observation here is the importance of large particle sul-
fate with St. Louis samples.
4. Comparison of sulfate results on glass fiber and Teflon filters
The present study provides one of the few data sets in which sulfate
results on glass fiber filter samples may be compared to those on a
relatively inert filter of equivalent filtration efficiency.
Results for corresponding glass fiber and Fluoropore filters are
compared in Table 26 as the ratio of means. Glass fiber sulfate
values with total filters are systematically higher at Durham and
St. Louis but approximately equivalent at Pasadena. With refined
*The filtration efficiency of Fluoropore filters (FALP, 1 my pore size) was
recently studied by Prof. Benjamin Liu, University of Minnesota. Results
were provided by private communication from Prof. Liu, 1975.
It is believed by the EPA staff that much of this large particle sulfate
resulted from a nearby industrial source of CaS04.
-90-
-------
Table 26 : SUMMARY OF COMPARISON OF GLASS FIBER AND FLUOROPORE FILTER SULFATE RESULTS AS A FUNCTION OF PARTICLE
SIZE AND SAMPLING SITE
Glass Fiber/Fluoropore
Sampling Site/Date
NC 7-15-71*
NC 7-16-7U
NC 7-17-7^
NC J-±Q-7h
NC Mean
MO 8-5-7^
MO 8-6-jk
MO 8-7-7h
MO 8-8-7^
i MO Mean
CA 12-2-7U
CA 12-7-7U
CA 12-8-7^
CA 12-9-7U
CA Mean
Overall (for 3 sites)
MTB
1.72
1.09
1.36
1.22
1.30 + .12
1.28
1.25
1.39
1.10
1.25 + .05
l.Ol*
.89
.71
.87
.92 + .07
1.21* + .Oh
0-20 ium
Brosset
1.21
1.36
1.38
1.37
1.32 + .Oh
1.28
96
1.17
91
1.06 + .08
.7^
1.16
2.00
1.16
1.12 + .28
1.13 ± -07
AIHL
1.39
1.01
1.3!*
1.26
1.23 ±
1.1*1
1.21
1.1*7
1.22
1.32 +
l.Ol*
1.15
1.20
i.oi*
1.08 +
1.27 +
.09
.06
.03
.05
MTB
.80
i.Uo
1.12
1.21
1.13 + .12
1.05
1.31
1.16
1.01
1.11 + .06
1.02
.9^
.69
.^7
.86 + .12
1.09 + .05
0-2 urn
Brosset
71
1.70
7U
1.20
1.01 + .20
.67
.75
.92
98
.85 + .07
.77
.63
2.31
1.00
.91 + .07
.91 + .07
AIHL
.82
1.17
1.18
1.28
1.11 +
1.03
1.U3
1.23
1.10
1.19 +
1.3^
1.07
1.23
.61
i.lh +
1.15 ±
.11
.07
.16
.05
Mean ratio glass fiber/fluoropore (for 3 methods and 3 sites) 0-20 jum = 1.21 + .03
Mean ratio glass fiber/fluoropore (for 3 methods and 3 sites) 0-2 jum = 1.0U + .0^
Mean ratio glass fiber/fluoropore (for 3 methods, 3 sites, and 2 size cuts) = 1.1^ + .03
Overall results are expressed as the ratio of means. Errors are calculated as the standard deviation of the
ratio of two dependent variables.
-------
Table 25: THE MEAN FRACTION OF SULFATE
IN REFINED (0-2 jma) PARTICLES a*
Durham 1.07
St. Louis 0.69
Pasadena 0.96
8. ^.
Results obtained by pooling Fluoropore
results for analysis by the MTB, AIHL
microchemical and modified Brosset
methods.
-92-
-------
samples results at Durham and St. Louis are much closer to being
equivalent on the two filter types.
If oxidation-promoting species (e.g. Fe, Ni, Mn, Cu) were largely
restricted to particles sizes > 2 pm then the present data would
permit additional mechanistic interpretation. Table 27 lists the
XKFA results from T. Dzubay, EPA, for the metals mentioned above.
While some results are below limits of detection, results do, indeed
indicate a predominance of these elements in the total filters
(0-20 ym) compared to the refined (0-2 vim) particle samples. Thus
the present data are consistent with enhanced artifact sulfate for-
mation on glass fiber filters in the presence of oxidation-promoting
species.
5. The influence of sample size on sulfate results with glass fiber
filters
Sulfate results on low and high volume glass fiber filters are com-
pared in Table 28 . The trends are very similar to those for glass
fiber/Fluoropore ratios; Durham and St. Louis results are consis-
tently higher on the low volume samples. Total air volumes sampled
was 3 m^/cm^ for the high volume sampler compared to 2 m^/cm^ for
the low volume units. Thus the present results are consistent with
previous studies of the "sulfate anomaly ' which demonstrated
that apparent sulfate levels increased with decreasing air volume
sampled.
-93-
-------
Table 27: COMPARISON OF SELECTED METALS CONCENTRATIONS , a)
BY XRFA IN TOTAL AND REPIHED PARTICLE SAMPLES
Fe Mn Cu ' Ni
Sampling Sampling
Site Date Total Refined Tcytal Refined Total Refined Total Refined
7/15/74 1.607J*; .219 .150 ± .020 .030 _+ .010 <.017 .012 _+ .005 <-009 .012 J; .005 <.010
NC 7/16M 2. 664 J;. 391 .147 ± .020 .040 ± .012 <.017 <.012 <.009 <.013 <.010
7/17/74 1.109.+ -149 .077 ± .012 <.0l8 <.017 <.009 <.009 <.010 <.010
7/18M 1.^64 ± .198 .072 ± .011 .024 _+ .009 <.017 <.010 <.009 <.011 <.010
8/5/74 1.765 ± .237 .053 ± .010 .067 j: .013 <.oi8 .oi9_+.oo5 <.oo9 <.oii <.oio
2. MO 8/6/74 2.61H.+ .349 .15TJ: .021 .076 j; .013 <.0l8 .038 jt .007 .Ol4j»: .005 .Ol4j»: .005 <.010
8/7/74 2.976^; .394 .479J; .061 .I20jt .018 .045 ± .010 .035 _+ .006 .022 ± .005 <.010 <.010
8/8/74 1.940 J: .255 .145 jf .019 .086 jt .014 <.0l8 .030 ± .006 .014 JH .005 <.011 <.011
12/2/74 1.240 jt .168 .153J: .021 .030 J: .009 <.0l8 .Ol8_j; .005 .014 _+ .005 .026 J; .006 .020 jt .006
CA 12/7/74 1.373 ±.183 .171 ±.022 <.017 <.0l6 . 020 ± .005 <.008 .020^.005 .012^.005
12/8/74 .572 4; .076 .055 ± .010 <.oi7 <.oi6 <.O09 <.oo9 <.oio <.oio
12/9/74 .914^.121 .102^.014 <.017 <.016 . 017 ± .005 <.008 .019^.005 . 010 ± .005
a. Samples collected on Fluoropore filters with XRFA by T. Dzubay, RTP.
-------
Table 28 : COMPARISON OF LOW (0-20/jm) AND HIGH VOLUME GLASS FIBER
FILTER SULFATE RESULTSa
Sampling Site/Date
NC 7-15-71*
NC 7-16-7^
NC 7-17-7^
NC 7-18-7^
NC Overall
MTB
Low Volume/High Volume
Brosset
AIHL
1.33
1.38
1.37
1.36 + 0.01
1.18
1.22
1.05
1.06
1.12 + 0.0k
1.27
1.23
1.33
1.29
1.28
+ 0.02
MO 8-5-7^
MO 8-6-7^
MO 8-7-7^
MO 8-8-7^
MO Overall
1.07
1.17
1.31
1.38
1.21 + 0.07
.82
.87
1.05
1.05
0.93 + 0.06
1.03
1.02
1.28
l.lkL
1.16 + 0.09
CA 12-2-71*
CA 12-7-71+
CA 12-8-7^
CA 12-9-7^
CA Overall
Overall (for 3 sites)
1.02
96
.78
.82
.93 + 0.06
1.23 +0.05
.98
1.15
3.3^
1.23
1-39 ± 0-38
1.02 + 0.06
0.89
0.82
0.59
0.71
0.79 + 0.06
1.16 + 0.06
Mean ratio low vol/high vol = 1.13 + 0.0k
(for 3 methods and 3 sites)
Same as footnote a, Table 26
-95-
-------
6. Comparison of high volume glass fiber and low volume Fluoropore
filter sulfate result
The findings in Tables 26 and 27 suggest that the high volume and
Fluoropore data should agree well. These results are compared in
Table 29. The striking result here is the equivalence obtained with
these filters at the three sites as measured by the MTB method in
contrast to the more scattered results with the remaining methods.
The modified Brosset and AIHL micromethods indicate opposite trends
with location; the Pasadena site yielded the highest ratio with the
modified Brosset procedure and the lowest, with the AIHL method.
These trends are largely dictated by the relative results obtained
on Fluoropore filters by the three methods as summarized in Table 22.
C. Summary and Conclusions
Results with low volume filters have revealed differences between the
three wet chemical methods able to analyze these samples of up to a
factor of 2 for individual samples and 1.6 when pooled by sampling site.
This contrasts markedly with results obtained with high volume samples.
XRFA results by LBL and RTF differ by nearly a factor of two. The RTF
XRFA permit estimation of water soluble sulfate within 10-50% of that
obtained by the MTB method depending on location. Since Fluoropore and
high volume results by the MTB method agree well, XRFA on Fluoropore and
-96-
-------
Table 29 : COMPARISON OF FLUOROPORE (0-20 juin) MD GLASS FIBER HIGH VOLUME
FILTER SULFATE RESULTS8"
Fluoropore/Glass Fiber (Hi-Vol)
Modified
MTB Brosset AIHL
Durham
St . Louis
Pasadena
I. Ok + .10
.97 + .08
1.01 + .03
.85 +
.88 +
1.2k +
.05
.10
.11
i.oU + .06
.88 + .08
.7^ + .07
as footnote a, Table 22
-97-
-------
MTB analyses of high volume glass fiber filters, such as employed with
the NASN network, are expected to agree within 10-50%, as well, for these
sites.
Large particle (2-20 ym) sulfate was clearly observed in the St. Louis
samples; about 30% of the total sulfate observed was in this size range.
Such large particle sulfate was not observed in the Pasadena and Durham
samples. In fact at Durham, the "refined" samples often yielded higher
sulfate and sulfur than total filter samples suggesting possible sampling
error.
The present data are consistent with artifact sulfate formation from
S02 on the low volume glass fiber filter samples with enhanced S0^= for-
mation in the presence of a large particle-related oxidation catalyst(s)
in the aerosol sampled. The equivalence of low volume Fluoropore total
filter and high volume glass fiber sulfate results as measured by the
MTB method implies an insignificant percentage of the sulfate results
from sampling artifacts with 24-hour high volume glass filters.
Finally, in comparing the three wet chemical methods the modified Brosset
procedure has, at times, yielded what we consider to be erratic behavior
with the low volume samples. Of the remaining two only the MTB method
provides protection against cationic interferents. Thus, in spite of
the 10-20% positive error in the method revealed by studies with high
volume samples, the MTB method is considered the most reliable of the
-98-
-------
three for long term (e.g. 24-hour) low volume samples. For short term,
low volume samples requiring a micro sulfate method we favor the AIHL
microchemical method.
-99-
-------
VII. 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 4N18.
3. Appel BR, Wesolowski JJ, Alcocer A, Wall S, Twiss S, Giauque R, Ragaini
R and Ralston R: Quality assurance for the chemistry of the aerosol
characterization experiment. AIHL Report No. 169, Air and Industrial
Hygiene Laboratory, California State Department of Health, Berkeley, CA
94704, July 1974.
4. Lee RE and Wagman J, Amer. Ind. Hygiene Assoc. J. TT_ 266 (1966)
5. DuBois L, Zdrojewski A, Teichman T and Monkman JL, Int. J. Environ.
Anal. Chem 1 113 (1971).
-100-
-------
Appendix A
(from Selected Methods for the Measurement of Air Pollutants
Public Health Service Publication No. 999-AP-ll with modifications
by AIHL)
Determination of Sulfate in Atmospheric Suspended Particulates:
Turhidimetric Barium Sulfate Method *
INTRODUCTION
Suspended particulate matter is collected over a 24-hour period
on an 8- by 10-inch glass fiber filter by using a high-volume sampler.
A water extract of the sample is treated with barium chloride to form
barium sulfate. T^ie turbidity caused by the barium sulfate is a mea-
sure of the sulfate content. Aliquoting is adjusted so that samples
containing 1 to 20 p.g/m (the expected range of atmospheric samples)
can be measured. The sensitivity of the turbidimetric analytical pro-
cedure is 50 y.g of sulfate. Nephelometrically, as little as 2 |ig of
sulfate can be measured.
REAGENTS
All reagents are made from analytical-grade chemicals.
Hydrochloric acid (10 normal). Dilute 80 ml of concentrated
reagent grade hydrochloric acid to 100 ml with distilled water.
Glycerol-alcohol solution. Mix 1 volume of glycerol with 2
volumes of absolute ethyl alcohol (reagent grade).
Barium chloride. Use 20- to 30-mesh crystals.
Standard sulfate solution (100 ^g SO2 per ml). Dissolve 0. 148
g of anhydrous sodium sulfate (dry if necessary) in distilled water and
dilute to 1 liter.
EQUIPMENT
High-volume sampler. A motor blower filtration system with a
sampling head, which can accommodate an 8- by 10-inch glass fiber
filter web and is capable of an initial flow rate of about 60 ft per min-
ute, is used and is shown in Figure 6. These samplers are available
from General Metals Works, Box 30, Bridgetown Road, Cleves, Ohio;
and Staplex Company, 774 Fifth Avenue, Brooklyn 32, N. Y. , among
others.
Glass fiber filters. Use 8- by 10-inch size, Mine Safety Appli-
ances Company, 1106 BH or any comparable make.
Refluxing apparatus. Use 125-ml flask fitted with reflux con-
denser and hot plate.
Funnels and Whatman No. 1 filter paper.
Cuvettes. Use cuvettes with a 1-inch light path and plastic
stoppers.
* As used by the National Air Sampling Network.
Prepared by Norman A. Huey, Laboratory of Engineering and Physical Sciences,
Division of Air Pollution, Public Health Service. Approved by the Interbranch Chemical
A J *"* !» ,..!.. IO14
Advisory Committee. July 1964.
Turbidimetric Barium Sulfate Method
-101-
-------
*"£
\ -^ i
FIGURE 6. TYPICAL SAMPLER ASSEMBLY (ABOVE) AND HIGH-VOLUME AIR
SAMPLER (BELOW).
I-Z
SELECTED METHODS
-102-.
-------
Pipettes. Use 10-, 4-, and I-ml pipettes.
Spectrophotorneler or colorimeter. This device should be suit-
able for measurement at 500 m |i.
PROOF DURE
Sampling. Using the Ki-Vol sampler, collect the particulates
from approximately i, 000 m of air. Twenty-four hours is the usual
sampling period. The air volume is calculated from the sampling
time and the average of airflow measurements taken at the start and
end of the sampling period.
Sample preparation. The sample filter is folded upon itself
alone thf- 10-inch axis to facilitate storage and transportation. This
told may result in a nonhomogeneous area in the sample. All sample
a.liquoting is, therefore, made across the fold. Using a wallpaper
cutter or other suitable device and a straight edge, cut a 3/4- by 8-
inch strip from the filter. Place this in the refluxing apparatus with
23 ml of distilled water and reflux for 90 minutes. Filter through
Whatman No. 1 paper, rinsing with distilled water till 50 ml of fil-
trate is obtained.
Analytical procedure. Place a 20-ml portion of the prepared sam-
ple into a clean, dry, 1-inch cuvette. Add 1 ml of 10 N HC1, 4 ml of
the glycerol-alcohol solution, and mix. Determine the absorbance at
500 mu,a-2ainst a reference cuvette containing distilled water. T'vs
_ ___ i: __ :_ _ . . u * ___ *. j c --- »i __ £: i ---- 1: -- T*. ----- .._ c -- ..... --- - - i- - j
cuvettes and other impurities in the sample. Add approximately 0. 25 g
of barium chloride crystals and shake until dissolved. Let stand for
40 minutes at room temperature (20 to 30°C). Measure the absorbance
at 500 mu,against the reference cuvette containing distilled water.
A standard sulfate solution should be analyzed with each batch of
samples. Deviations up to 5% from the standard curve can be expected.
Occasionally it is advisable to determine the percentage recovery by
adding knov.-n amounts of sulfate to clean filters and determining the
amounts found, using the entire procedure including refluxing. The
percentage recovery should be close to 100.
Standardization. Obtain a standard curve by analyzing 20 ml
of a series of standards containing 2 to 60 ji g SO^ per ml. Coordinates
of the curve are absorbance and total sulfate in jig (40 to 1,200).
Calculations.
jig SO Per m
F = sample aliquot factor = 30
C = number of |j. g of SOT found
V = sample air volume in m
Turbid-metric Barium Sulfate Method
-103-
-------
DISCUSSION OF PROCEDURE
Collection media. Although most past data have been gathered
on Mine Safety Appliance Company glass fiber filters, other filters are
available from H. Reeve Angel &c Company, Union Industrial Equip-
ment Company, Carl Schleicher £. Scnuell Company, and the Gelman
Instrument Company.
Glass fiber filters are not sulfate free. The MSA filters have
been found to contain about 4 mg per 8- by 10-inch sheet. It is
advisable to check whatever sampling media are used. This sulfate
can be removed by water wash prior to sampling if desirable. When
this is not practical, results must be corrected accordingly.
The analytical method can also be applied to samples collected
on membrane filters and with electrostatic precipitators.
Modification of analytical method sensitivity. To decrease
sensitivity, use a smaller sample portion diluted to 20 ml with dis-
tilled water. To increase sensitivity, use a larger sample portion
or, in extreme cases, measure nephelometrically.
Critical variables and their control. Measurement is dependent
upon the amount, the size, and the suspension of the barium sulfate
particles. v~~
Parameters that must be controlled are stability of the suspension
of colloidal particles, suliate concentration, ua.iiun 1-.". strength, pH.
temperature, and aging of the barium chloride solution. Glycerol
acts as a stabilizer for the colloid, while alcohol promotes precipita-
tion of the sulfate. Use of solid crystals of BaC^ eliminates the prob-
lem of barium ion strength and solution aging. Sulfate concentration is
maintained within the limits of the method, and pH is controlled by
addition of HC1. Variations in temperature of ZO to 30°C do not appear
to have a significant effect.
Precision of method. The method has been shown to have an
i 1""' coefficient of variation.
REFERENCES
1. Air Pollution Measurements of the National Air Sampling
Network 1957-19rl. Public Health Service publication
No. 97S. U.S. Government printing Office, Washington,
D.C.
1. Pate, J. B. , Tabor, E.C. , Analytical Aspects of Glass Fiber
Filters, Am. Ind. Hys. Assoc. J. 23:145-150. 1962.
3. p.irr, S.W., Staley, Vr'.D. , Determination of Sulfur by
Means of the Turbidimeter, Ind. Eng. Chem. Anal. Ed. i:
of^-67. 1931.
4. Kcilv. H.J., Rodeers, L. B. , Nephelometric Determination
_" Sjlfnte ln-.purit-.es. ir, Certain Reagent Grade Sails, Anal.
fh.-.-n. 27:759. 1°55.
;-4 SF.LF.C1 F.D MFTHODS
-------
Appendix A
MODIFICATIONS OF THE TURBIDIMETRIC PROCEDURE AT AfflL
The reaction was carried out directly in a series of 36 cuvettes. This saved
time and avoided the transfer of the analyte mixture from other vessels into
cuvettes which would be expected to yield more erratic measurements. A zero
reading was done before adding the barium chloride. The liquid reagents were
measured with repetitive pipets (Repipets) which allowed higher precision and
better time control. The barium chloride was added at one minute intervals
from sample to sample. A Bausch and Lomb Model 20, single beam spectropho-
tometer was employed. A small vortex mixer was used to eliminate the bubbles
produced by shaking following addition of the BaCl2«
Measurement
Readings were made after exactly 40 minutes of the addition of the barium
chloride, mixing and shaking of each sample, by reading at one minute time
interval between individual samples. One minute proved ample time for doing
the reading and writing down the result.
Quality control
Standards were run simultaneously with the samples for each batch, regard-
less of any existing calibration curve prepared previously.
-105-
-------
APPENDIX B
TECHWICON AUTO ANALYZER II
INDUSTRIAL METHOD No. 118-71W/TENTATIVE
DATE RELEASED: DEC. 1372
SULFATE IN WATER AND WASTEWATER (Rcngo: 0-300 mg/l)
GENERAL DESCRIPTION
In this automated procedure for sulfate, the sample
is first passed through a cation-exchange column to
remove interferences. The sample cor.tajning suli'ate
is then reacted with barium chloride at a pH of 2.5-3.0
to form barium sulfate. Excess barium reacts with
methylthymol blue to form a blue-colored chelate at a
pH of 12.5-13.0. The uncomplexed methylthymol blue
color is gray; if it is all chelated with barium, the color
is blue. Initially, the barium chloride and methyl-
thymol blue are equirnolar and equivalent to the highest
concentration of sulfate ion expected; thus the amount
of uncomplexed methylthymol blue, measured at460 nm,
is equal to the sulfate present. *
PERFORMANCE AT 30 SAMPLES PER HOUR
USING AQUEOUS STANDARDS
Sensitivity at 300 mg/l
Coefficient of Variation
REAGENTS
0.35
obsorbance units
Ql fju my/ i
Detection Limit
10 mg/l
BARIUM CHLORIDE
Barium Chloride Dihydrate (Bad 2'2HzO) 1.526 g
Distilled Water, q.s. ' 1000 ml
Preparation:
Dissolve 1.526 g of barium chloride dihydrate in
500 ml of distilled water. Dilute to one liter with dis-
tilled water. Store in a brown polyethylene bottle.
METHYLTHYMOL BLUE
Methylthymol Blue* .
3', r-Bis-N, N-bis (carboxymethyl)-
Amino Methylthymolsulfonephthalein
Pentosodium Salt
Barium Chloride Solution
Hydrochloric Acid, 1.0A/(HC!)
Distilled Water
1 Lazrus, A.L.. Hill, K.C. ond Lo.i^c, J.P., "A New Colon-
metric MicroHotermmiition of Suluito Ion". Automat ion IP
Analytical r*hi-r.ii
-------
WORKING STANDARDS
ml Stock
1.0
6.'0
12.0
18.0
24.0
30.0
mg/l SC>4~
10
60
120
180
240
300
Preparation:
Pipette stock into a 100 ml volumetric flask. Dilute
to 100 ml with distilled water.
ION EXCHANGE COLUMN
. The column consists of a length of* glass tubing
7.5 inches long, 2.0 mm ID and 3.6 mm OD. The com-
mercial ion-exchange resin Bio-Rex 70,** 20-50 mesh,
sodium form is freed from fines by stirring with several
portions of deionized water and decanting the supernate
before settling is complete. Fill the column with resin,
taking care that air is not trapped in the column. Glass
wool plugs are placed at each end to prevent the resin
from escaping. Care should be exercised that excess
glass wool is not used which -.vill cause back pressure.
OPERATING NOTES
1. When running this system, it is very important
that no air bubbles enter the ion-exchange column at
any time. If air bubbles become trapped, it is advisa-
2. Cations, such as calciur.i, aluminum, and iron,
interfere by complexing the methylthymol blue. These
ions are removed by passage through an ion-exchange
column.
3. Before running this method, position the con-
trols of the Modular Printer as follows:
** Available from Bio-Rad Laboratories, Richmond California.
CONTROL POSITION
MODE Switch Normal
SAMPLING RATE Switch 30
RANGE Switch 300
DECIMAL Switch 000.
Details of Modular Printer Operation are provided
in Technical Publication No. TA1-0278-10.
4. Since this chemistry does not conform to Beer's
Law, a Technicon Linearizer is necessary in order to
obtain readings which are directly proportional to
concentration.
When using the Linearizer, a non-linearized
calibration curve is first plotted by placing the Linea-
rizer in the direct mode. Using the non-linear curve,
concentrations of standards are selected which fall.at
approximately 75% for each range of the Linearizer.
For example, a concentration which falls at 15% of
scale in the non-linearized mode should be selected as
the standard for the 0-20% range of the Linearizer.
These concentrations as calculated are then used to
set the Linearizer as directed in the Linearizer manual
(Technicon No. TA1-0279-00).
5. At the end of each day, the system should be
washed with a solution of EDTA. This may be dune by
placing the methyltnymol blue line and the sodium hy-
droxide line in water for a few minutes and then into
the tetrasodium EDTA for 10 minutes. Wash system with
water for 15 minutes before shutting down.
6. Alternate ranges may be obtained by utilization
of the Std Cal control, ton the Colorimeter.
7. When*using a Linearizer, the use of multiple
working standards is only to establish linearity. For
day-to-day operation, the 180 mg/l standard is recom-
mended for instrument 'Calibration.
-107-
-------
SULFATE IN WATER AND WASTEWATER
- (Range: ^SttOrng/l) 0- CaO
MANIFOLD NO. 116-D096-01
To Sampler IV
Wash Receptacle
GRN/GRN (2.00) WATER
ION EXCHANGE COLUMN
116-G006-01
170-0103-01
A2
Waste
r°
O
BLK/BLK (0.32) AIR
P.LO /FLU (/.£0) WATER
^ 5 Turns|116.^oa-u.^ORN/f>RN (Q ^ SAMPLE
0.110 Standard L/
Sleeving ivV,,,r ^^ GRY/GRY (1.00) WASTE
_* I ".
-O
I 157-B095 157-0370
p. BLK/BLK (0.32) AIR
Jli "
20 Turns
SAMPLER IV
22 Turnsj 116-0489-01 RED/RED (0.70) METHYLTHYMOL BLUE
COLORIMETER
460 nm To F/C
15 mm F/C x 2.0 mm ID ' Pump
199-B023-06 Tube
Waste
ORN/ORN (0.42) SODIUM HYDROXIDE
GRN/GRN (2.00) FROM F/C **
NOTE: FIGURES IN PARENTHESES
* 0.034 POLYETHYLENE SIGNIFY FLOW RATES IN
** SILICONS RUBBER ML/MIN.
TECHNICON INDUSTRIAL SYSTEMS / TARRYTOWN. NEW YORK 10591
A DIVISION OF TECHNICON INSTRUMENTS CORPORATION
-108-
r c Wl«. TtexMCOM rairmMfori cooKMUTion
-------
PREPRINT NOT FOR PUBLICATION
Limited Distribution
APPENDIX C
. A MICROMETHOD FOR SULFATE
IN ATMOSPHERIC PARTICIPATE MATTER
AIHL REPORT NO. 163
Prepared by:
E. Hoffer and E. L. Kothny
This work was partially supported by the
Air Resources Board, Division of Technical Services, Research Section
Mr and Industrial Hygiene Laboratory
California State Department of Health
Laboratory Services Program
2151 Berkeley Way
Berkeley, California 94704
July 1974
-109-
-------
INTRODUCTION
The increased concern with the biological effects of particulate
sulfate and the need to determine diurnal patterns spurred the study
for a fast and sensitive micromethod for analysis of sulfate in
atmospheric particulate matter.
Despite new instrumental methods, some cations and anions must still
be analyzed by wet chemistry. Hi-vol filter mats generally have a
substantial amount of material collected which allows the use of macro
techniques (mg amounts) for the determination of the analytes. How-
ever, if samples have been collected at low air flow rates and for
short periods, the techniques to be applied must be refined and selected
accordingly.
In this paper we present a review of existing procedures for sulfate
analysis, a number of which were considered and rejected as inadequate
for further development. Following, a micromethod is detailed which
extends the working range for S0^= down to nearly 1 ug/ml of extract.
SUMMARY OF SULFATE METHODOLOGY
Sampling and extraction are very important steps in the determination
of sulfate. When determining sulfate from particulate matter collected
on glass fiber filters, low values are to be expected because of
-110-
-------
retention of sulfate on glass (1) Analytical losses vary inversely
with the sulfate concentration per unit area. The retention of sulfate
when using glass fiber filters is believed to be caused by the barium
content of the glass. Therefore, organic filters are to be preferred
for low amounts of sulfate (< 50 yg). The collection efficiency of
membrane filters is very high even at relatively large nominal pore
sizes C2). A good recovery of sulfate from the filters has been ob-
tained by using a microextraction procedure which utilizes 5 ml of
distilled water in a small flask (3,4).
Analyzing low concentrations of sulfate has been accomplished in the
past by concentrating large samples and determining precipitated barium
sulfate gravimetrically (5,6,7). Gravimetry is still used where high
accuracy is of prime concern. A step toward simplifying these procedures
and expanding the capabilities on smaller samples has been made by using
turbidimetry or nephelometry of barium sulfate (6,8,9,10) or organic
sulfates such as benzidine (11), amino-chlorobiphenyl (12,13), and
aminoperimidine (11,13,14,15).
Titration methods were proposed for increasing the accuracy over turbi-
dimetric methods. The most popular method, refined to be usable in the
range of 2 to 10 yg/ml, titrates the sulfate in 80% ethanol with 0.005 M
Ba using Thorin as indicator (16). Because of the difficulty in ob-
serving the endpoint, the use of photometric titration has been
suggested (17). Sharper endpoints are obtained using Sulfonazo III (18)
and Nitrochromeazo (19) as indicator. In a study made comparing different
-111-
-------
indicators it has been shown that the indicator Dimethylsulfonazo III
is best suited for titrimetry with barium for visual endpoint estima-
tion (20). Dimethylsulfonazo III has been collaboratively tested as
an indicator in Ba titration of sulfate after separation of cations
with a cation exchange resin (21). It has been observed that pH plays
an important role in the formation of initial crystallization centers
which speed up the equilibrium of the reaction during titration. The
best buffer was a combination of pyridine with a strong acid (22) in
an almost anhydrous organic solvent. Pyridine seemed to produce
solvolysis of the indicators. A similar solvolysis effect has been
observed with surfactants (23).
A lead-sensitive electrode has also been used for determining the
endpoint in the titration of sulfate in dioxane using a dilute lead
solution (24,25). Exchange resins can be used to eliminate interfer-
ences for this technique (26). Some methods are based on the exchange
of the anion bound to Ba with sulfate. The popular chloranilate
method (27,28) has been modified to eliminate interferences caused
by some cations and to stabilize the variable sensitivity to changes
of pH (29). It has been adapted for use in automated systems (30).
Other exchange methods use barium iodate. After exchange with sulfate
the solubilized iodate is used to oxidize iodide to iodine. Either the
triodide ion is measured directly (31) or reacted with cadmium iodide
linear starch (32). Interference from calcium in these methods is
sometimes a problem, because only 15% of the calcium sulfate reacts (31),
-112-
-------
Two general sulfate reduction methods to ^S are available. One reduces
sulfate with HI and hypophosphorous acid (33,34,35) and the other uses
stannous phosphate in excess phosphoric acid (36,37). The hydrogen
sulfide so generated can be quantitated with very sensitive reagents
which offer a high degree of specificity. The most sensitive techniques
for analyzing t^S are fluorescence quenching (38,39,40), silver-dye
exchange (41,42), methylene blue formation (34,37,43), and molybdenum
reduction (44).
I |
Indirect titration methods, reacting an excess Ba with the sulfate
and determining the unreacted barium by atomic absorption (45) or by
a colorimetric procedure using methyl thymol blue has been proposed.
The latter method was automated and employed extensively for sulfate
analysis of samples collected by the National Air Sampling msi.wui.k.
Flame photometry (48) and thermogravimetry, which are still in the
developmental stage, are sensitive direct procedures for sulfate.
SUMMARY OF THE AIHL MICROCHEMICAL SULFATE METHOD
In selecting an appropriate method the aim was to achieve a compromise
between simplicity, specificity and accuracy. The limitation imposed
for the selection of a method in the present study was the amount of
sulfate encountered in aqueous extracts of membrane filters obtained
from sampling 10 m^ of air. Therefore, the method must be capable of
furnishing acceptable information in the low microgram range. Samples
-113-
-------
were collected on preweighed cellulose ester membrane filters. Ex-
tractions were performed by a micropercolation technique using a minimal
amount of liquid. The principle of the barium-dye exchange method was
selected for measurement. The substance used as barium salt for the
reaction was the complex formed with nitrochromeazo [2,7 - bis (4-nitro-
2-sulfo-phenylazo) - 1,8 dihydroxynaphthalene - 3,6 disulfonic acid
sodium salt] C19,49). A procedure based on the same principle has been
described recently (50) . The reaction employed may be symbolized as
follows:
Ba-dye + SO^ -> BaS04 + Dye
The barium-dye complex exhibits an absorbance maximum at 640 to 645 nm
in pH 5.4 buffered acetonitrile while the unassociated dye absorbs less
at this wavelength. By using an excess of the barium-dye complex
relative to the expected sulfate level, the decrease in absorbance at
640 to 645 nm can be directly measured against a reagent blank in a
double beam spectrophotometer and related to the sulfate concentration.
EXPERIMENTAL
A. Apparatus
A double beam spectrophotometer with 2 nm bandpass and
10 mm pyrex glass cuvettes are suitable.
Variable volume micropipets (0-20 yl, 0-200 yl, 0-1,000 yl) .
Glass stoppered test tubes 16 X 150 mm.
-------
B. Chemicals
Nanograde acetonitrile free of traces of sulfate and metals.
Each batch of this solvent must be tested for sulfate before
use.
Barium chloride. The chloride is preferable to the
perchlorate which is hygroscopic.
Nitrochromeazo (also called Nitrosulfonazo III) . This
product is available from different sources*. In the
formulation of the reagent a 50% molar excess over Ba is
used in order to compensate for the less than 100% dye
content .
Pyridine, AR.
Benzene sulfonic acid, monohydrate. The Eastman Kodak
No. 2313, low in heavy metals, is satisfactory.
Sodium sulfate anhydrous. Dry in an oven for one hour at
105°C in a weighing bottle. Close bottle with glass
stopper while hot and leave to cool in a dessicator.
C. Reagents
0.01 M Bad 2 aqueous.
0.001 M BaCl2 aqueous.
0.15% Nitrochromeazo, aqueous (approximately 0.0015 M) .
Buffer pH 5.4. Dissolve benzenesulfonic acid monohydrate
in water to obtain a 50% w/v solution. To 25 ml of 50%
and Bauer, Gallard-Schlesinger, K + K, Fluka, Aldrich.
-115-
-------
benzenesulfonic acid add 12 ml pyridine, then cool and
make up to 100 ml.
Reagent mixture. In a 1,000 ml volumetric flask place
850 ml acetonitrile and add:
20 ml 0.001 M BaCl2 solution
25 ml 0.15% Nitrochromeazo solution
10 ml buffer pH 5.4
Fill to mark with water. The apparent pH of this solution
is 5.4.
Sulfate stock solution, 1,000 yg/ml sodium sulfate:
Dissolve 1.479 g dried Na2SO^ in one liter of distilled
water.
Sulfate standard solution, 20 yg S0^= per ml. Dilute
2.00 ml sulfate stock solution to 100 ml.
D. Procedure
1. Sample preparation and approximate estimation of sulfate
content.
The air is preferably sampled with preweighed cellulose
ester membrane filters. After weighing, the approximate
amount of sulfate collected is estimated from XRF data
if available or from statistical information on sulfate
as a percent of total mass obtained from previous
studies in the same or adjacent areas under similar
sampling conditions. Generally, the range for sulfate
found on the west coast is between 2 and 20 yg/m
-116-
-------
with most values at the lower quartile, or between 3 and
10% of the total amount of particulate material collected.
2. Extraction.
Cut each filter with a scalpel holding it with tweezers and
insert the pieces into a fritted disc funnel. A flask
containing 5 ml of distilled water and stoppered with a
one-hole rubber stopper holds the fritted disc funnel
whose stem reaches the bottom of the flask (3). Heat
repeatedly with an aluminum block or a microburner. To
insure complete extraction 5 micropercolations were used.
The extract may be used for determination of sulfate,
nitrate and eventually other water-soluble products.
Alternatively, heat screw.cap test tubes containing 5 ml
. : 2»'«*
of water and filter to 80°C for two hours in an oil bath.
After shaking occasionally allow the tubes to cool over-
night. This technique gave 10% less recovery than the
micropercolation extraction technique.
..'. -.«"«» -
3. Sulfate determination.
Pipet aliquots of one ml or less of the aqueous extracts
blanks and standards containing approximately 6 to 10 yg
S0^~ into a glass-stoppered test tube (16 X 150 mm).
Dilute the aliquots to 1 ml with water with the 1.00 ml
variable volume micropipet. Then add 8 ml of the reagent
mixture, shake and allow one hour reaction time in a cool,
dark place. Read the absorbance in a 10 mm cell at the
-117-
-------
peak between 640 and 645 nm in the double beam spectre-
photometer against a blank prepared with 1 ml distilled
water.
4. Calibration.
Place aliquots from the dilute standard solution into glass
stoppered test tubes (16 X 150 mm) so as to cover a
measuring range between 1 and 14 yg, and bring each to
one ml with distilled water. Alternatively, add small
aliquots of the concentrated stock solution to one ml
of distilled water with the aid of micropipets. Then
add 8 ml of the reagent mixture and proceed as indicated
above. Construct a calibration graph with absorbance
on the ordinate and yg sulfate on the abscissa. The
origin o£ che line should go Chrougli zeiu. "A lectsL
square line is calculated.
5. Measurement.
The calibration curve was a straight line from the origin
up to absorbance 0.6, which corresponded to approximately
12 yg S04=. Above 14 yg/ml S04= the curve begins to
flatten, and below 6 yg S0^= there seems to be a larger
error on real samples although the standard curve is
linear to one yg or below.
6. Effect of interferents.
Calcium is a serious interference, since the exchange of
Ba by Ca in the complex does not affect the color of the
-lie-
-------
complex significantly. For 10 yg of sulfate and a molar
ratio of 0.1 Ca/SO^ the loss in response was 4%; at a
molar ratio of 1 Ca/SO^ the loss was 90 to 95%, whereas
at a molar ratio of 20 Ca/SO^ the loss in response was
total. Atmospheric samples collected in Los Angeles,
however, contained only a small amount of soluble Ca
which resulted in losses of 4 to 10%.
7. Internethod comparison.
Hi-vol aqueous extracts of atmospheric particulate matter
saved from previous turbidimetric analyses were used for
intermethod comparisons. These extracts were combined so
as to obtain four different concentrations in the range of
10 to 40 yg/ml and analyzed by two methods, i.e., turbidi-
metric BaSO^ and titrimetric with a photometric endpoint
using the same dye as for the colorimetric procedure.
The titration was made in quadruplicate with 0.001 M
BaCl2« The turbidimetry of BaSO, and this method were
made in triplicate. Results are shown on Table 1.
It can be observed that the mean relative standard deviation
is largest for the micromethod and smallest for the titri-
metric procedure. Both the micromethod and the turbidimetric
procedure have larger deviations at the lower concentrations,
whereas the titrimetric procedure has higher precision at
lower concentrations. Higher values are obtained for the
-119-
-------
titrimetric procedure at the lower levels, which is ex-
plainable by the relatively higher solubility of BaSO^
at low concentrations which affects more of those methods
based on a strict solubility equilibrium.
The micromethod has a consistently lower value (^ 4 pg/ml)
at all concentrations possibly due to solubility of BaSO^
which seems to be greater for atmospheric sample extracts
than for the solutions prepared from pure aqueous standards.
DISCUSSION
A. Other Methods
Aminoperimidine was interfered by coextracted organic substances
which promoted premature agglomeration and precipitation of the
suspension. This fact was not observed with pure solutions. Ex-
change methods which relied on the measurement of iodine were
not considered because of possible interference by reductants
and chromophores in atmospheric extracts.
Measurement of hydrogen sulfide obtained by reduction of sulfate was
considered cumbersome and inappropriate.
B. AIHL Methods
1. Limitations
Since this is a differential method, the precision is critically
dependent upon adjusting sample aliquot size to an optimum range.
-120-
-------
As conducted in the present study, the solution containing the
barium-dye complex had the capacity of reacting with a maximum
of 15 yg/SO^". The optimum aliquot size for this capacity
contained between 6 and 10 yg S0^~. This requirement repre-
sents the most serious limitation of the method.
Sulfate introduced spuriously shortens the range of the method
and affects precision. Glassware used for all tests should be
scrupulously clean. Reagents must be tested for sulfate
content. New batches of solvents (acetonitrile, pyridine) and
reagents must be compared to the old set of reagents in order
to establish performance. Solvents containing more than
100 yg SO^/liter or reagents which introduce more than one
yg/S04 per determination should be replaced.
The negative effect due to calcium could be reduced by addition
of 0.05 ml of a 20% citric acid solution. This observation was
made during titrimetry of CaSO^ but no experimental data was
obtained for the micromethod. Another approach used previously
is the removal of interfering cations by ion exchange.
2. Variations
From the various organic solvents proposed for the exchange
reaction, acetonitrile offered the lowest solubility product
for BaSO^ and fastest stabilization, thus gave more reliable
-12.1-
-------
information at the lowest levels of sulfate. Eighty percent
acetone could not be used because of large variations
apparently caused by a slower reaction rate.
It has been shown that at a pH of 5 to 6 all dyes react fastest
in organic-water medium. In principle, this value of pH can be
obtained in many different ways, e.g., with pyridine and acetic
acid (18). However, we have observed that equilibration be-
_ | |
tween available SO^ and Ba in an aqueous-organic mixture
works fastest when the anion ionizes readily. This approach
has been used unwittingly when using a buffer composed of
pyridine and perchloric acid (22). The hazardous nature of
perchloric and nitric acids are known when they come in contact
with organic solvents. Therefore, benzene sulfonic acid was
selected as a substitute although toluene sulfonic acid and
trichloroacetic acid worked as well.
Small changes in the water content of the final solution
between 15 and 30% affected the response at the level from
0 to 1 yg/ml, which is apparently caused by increased solu-
bility of the BaSO^ produced. If the water content of the
reagent or the final mixture falls below 15%, precipitation
of the Ba-dye complex may occur.
3. Measurement and standards
-122-
-------
The use of variable micropipets was very convenient for working
with aliquots under one ml. The accuracy of the pipets we
used was found to be better than 1%.
Better accuracy for the method is obtained in the upper range.
Below 6 yg S0^~ an asymptotic increase in % relative error has
been measured. This effect is a result of BaSO^ solubility
and reaction speed. The slope of the calibration curve depends
on the relative amount of sample, reagent and time but is
essentially a straight line when fixing these parameters. The
water content of the final mixture and the absolute content
of Ba and complexant also affects the slope. When the molar
concentration of Ba in the reagent is approached or exceeded
by the S0^= content, the calibration curve is no longer
linear and flattens out. Therefore, accuracy is highest
on a narrow range of 6-14 yg per determination.
Before final readings were made, a technique was tried which
consisted of reducing the final apparent value of pH from 5.4
to below 3.5 in order to reduce blank values. When comparing
the reacted mixture with the reagent blank on a double beam
spectrophotometer, no significant differences were observed on
real samples whether measuring the absorbance at one or another
pH value. In cases where the optical mechanism of an instrument
is not capable of measuring differences at high values of
-123-
-------
absorbance Cdifferences of tenths in a background of about
absorbance 3) the shift to a lower pH before readings may be
necessary.
4. Intermethod Comparisons
Several dyes of similar composition have been evaluated for
the titrimetry of sulfate in the microgram range (20).
Although not adequate for the visual titration step, nitro-
chromeazo was selected because of its highest molar absorp-
tivity (19) and fastest equilibration when reacting with Ba.
The titration was carried out using pyridine-benzenesulfonic
acid buffer and 66% acetone. The color change at the
equivalence point was compared to two identical Erlenmeyer
flasks containing all the reagents but Ba in one and 0.5 ml
excess of 0.001 M Ba in the other. The reason for using
66% acetone in preference to 80% propanol is because it is
less affected by variations of solvent concentration, and
less toxic.
ACKNOWLEDGMENT
The authors wish to acknowledge the assistance of Dr. B. R. Appel
in preparation of this report.
-------
REFERENCES
1. Private communication from E. Perry of Rockwell International,
1973.
2. Appel BR, Wesolowski JJ. Studies to select filter media for
particulate sampling with a Lundgren impactor. AIHL Report
#125, 1972.
3. Hermance HW, Russell CA, Bauer EJ, Egan TF, Wadlow HV. Relation.
of airborne nitrate to telephone equipment damage. Environ. Sci.
Technol. 5:781-85, 1971.
4. Appel BR, Wesolowski JJ, Alcocer AA, Wall S, Twiss S, Giauque R,
Ragaini R, Ralston R. Quality assurance for phase II chemistry of
the aerosol characterization experiment. AIHL Report #169, 1973.
5. Standard methods tor the examination or water and wastewater.
APHA 1965 (12th ed.). New York. 769 pages.
6. Kolthoff, Sandell, Meehan, Bruckenstein. Quantitative Chemical
Analysis. The McMillan Co. 1969, London. 1199 pages.
7. Intersociety Committee. Methods of air sampling and analysis.
APHA 1972. Washington, DC.
8. Air and Industrial Hygiene Laboratory: The alkaline plate
method for the determination of total sulfation. California
Department of Public Health. April 1971. Method No. 36.
9. Huey NA: Determination of sulfate in atmospheric suspended
particulates: Turbidimetric barium sulfate method. National
Air Sampling Networks, PHS, July 1964. In: Selected Methods
for the Measurement of Air Pollutants, USDH-PHS 1965.
Publ. 999-AP-ll.
-125-
-------
10. Intersociety Committee. Tentative method of analysis of the
sulfation rate of the atmosphere (lead dioxide plate method -
turbidimetric analysis). Health Laboratory Science 8:243-247,
1971.
11. Berger AW, Driscoll JN, Morgenstern P: Review and statistical
analysis of stack sampling procedures for the sulfur and nitrogen
oxides in fossil fuel combustion. Presented APCA meeting,
St. Louis, June 1970.
12. Coleman RL, Shults WD, Kelley MT, Dean JA. Turbidimetry via parallel'
photometric analysis Determination of sulfate. Anal. Chem. 44:1031-
1034, 1972.
13. Stephen WI: A new reagent for the detection and determination of small
amounts of the sulfate ion. Anal. Chim. Acta.50:413, 1970.
14. AuJou^s S, Scncsictti C. Automatic determination of sulfate "'<>" n-arps.
Inform. Chim. 107:223-226, 1972. From Chem. Abstr. 77, 69743b.
15. Jones PA and Stephen WI. The indirect spectrophotometric determination
of the sulphate ion with 2-aminoperimidine. Anal. Chim. Acta 64:85-92,
1973.
16. Fritz JS and Yamamura SS: Rapid microtitration of sulfate. Anal. Chem.
27C9):1461, 1955.
17. Rodes CE. A colorimeter system for determination of method 6 Thorin
titration endpoint. Presented 167th National Meeting of the ACS,
Los Angeles, April 1974.
18. Determination of sulfur trioxide and sulfur dioxide in stack gases.
Shell Method Series 62/69, 1969.
-126-
-------
19. Kuznetsov VI and Basargin NN. Indicator for Ba during a volumetric
determination of sulfates in the presence of phosphates and arsenates.
Zavodsk Lab 31(5):538-41, 1965. From Chem. Abstr. 63, 2364e.
20. Budesinsky B and Krumlova L. Determination of sulphur and sulphate
by titration with barium perchlorate. Anal. Chim. Acta 39:375,
1967.
21. Scroggins LH. Collaborative study of the microanalytical oxygen flask
sulfur determination with Dimethtlsulfonazo III as indicator. Journal
*
of the AOAC 57(1):22-25, 1974.
22. Archer EE, White DC, Mackinson R: An improved titration medium for
sulphate ion indicators. Analyst 96:879, 1971.
23. Toei K and Kobatake T. Surface-active agents in analytical chemistry.
I. Titrimetric determination of sulfate with Sulfonazo III (as an
indicator). Bunseki Kagaku 17(5):589-92, 1968.
24. Ross Jr JW and Frant MS: Potentiometric titrations of sulfate using
an ion-selective lead electrode. Anal. Chem. 41:967, 1969.
25. Heistand RN and Blake CT: Titrimetric determination of traces of
sulphur in petroleum using a lead-ion-selective electrode. Mikro-
chimica Acta 2:212-216, 1972.
26. Mascini M. Titration of sulphate in mineral waters and sea water by
using the solid state lead selective electrode. Analyst 98:325-328,
1973.
27. Bertolacini RJ and Barney JE. Colorimetric determination of sulfate
with barium chloranilate. Anal. Chem. 29:281-283, 1957.
-127-
-------
28. Bertolacini RJ and Barney JE. Ultraviolet spectrophotometric
determination of sulfate, chloride and fluoride with chloranilic
acid. Anal. Chem. 30:202-205, 1958.
29. Schafer HNS. An improved spectrophotometric method for the deter-
mination of sulfate with Ba chloranilate as applied to coal ash and
related materials. Anal. Chem. 39:1719-1726, 1967.
30. Gales Jr. ME, Kaylor WH, Longbottom JE: Determination of sulphate
by automatic colorimetric analysis. Analyst 93:97-100, 1968.
31. Klockow D and Ronicke G: An amplification method for the deter-
mination of particle-sulfate in background air. Atmos. Environ.
7:163, 1973.
32. Hinze WL and Humphrey RE: Spectrophotometric determination of
sulfate ion with barium iodate and the linear starch iodine system.
Anal. Chera. 45:814, 1973.
33. Davis JB and Lindstrom F: Spectrophotometric microdetermination of
sulfate. Anal. Chem. 44:524-532, 1972.
34. Gustafsson L: Determination of ultramicro amounts of sulphate as
methylene blue. Talanta 4(4):236-243, and 227-235, 1960.
35. Steinbergs A, lismaa 0, Freney JR, Barrow NJ. Determination of
total sulphur in soil and plant material. Anal. Chim. Acta
27:158-164, 1962.
36. Ramananskos E and Grigoniene K: Determination of trace amounts of
sulfate sulfur with crystal violet. Elem Mikrokiekiu Nustatymas
Fiz Chem Method, Liet TSR Chem Anal Mokslines Konf. Darb., 2nd
1969, 145-52. From Chem. Abstr. 76, 41584f.
-128-
-------
37. Hachino H: Determination of sulfur in silk protein (microquantitative
methods for sulfur in biologic materials). Hoi Kanshiki Shakai Igaku
Zasshi 1968, 6(1-2), 18-25. From Chem. Abstr. 70, 44720J.
38. Wronski M. Thiomercurimetric determination of t^S in natural waters
below and above microgram per liter level. Anal. Chem. 43:606, 1971.
39. Griinert A, Ballschmitter K, Tolg G: Fluoreszenzanalytische Bestimmung
von Sulfidionen im Nanogrammbereich. Talanta 15:451-457, 1968.
40. Vernon F and Whitham P. The spectrofluorimetric determination of
sulfide. Anal. Chim. Acta 59:155-156, 1972.
41. Deguchi M, Abe R, Okumura I: Spectrophotometric determination of
trace amounts of sulfide ion. Bunseki Kagaku 18:1248, 1969.
42. Kothny EL: 2-methyl - l,4-napthoquinon-(2 ben2o)-thiazolyl hydrazone
as a sensitive reagent for Ag and H2S. Lab Notebook No. 216. State
of California Department of Health. Page 11 and tt. April iy/z.
43. Bamesberger WL and Adams DF. Improvements in the collection of
hydrogen sulfide in Cd hydroxode suspension. Environ. Sci. Technol.
3:258-261, 1969.
44. Buck M, Gies H.: Die Messung von Schwefelwasserstoff in der Atmosphere.
Staub 26C9), 379-384, 1966.
45. Magyar B, Sanchez Santos F: Indirekte Bestimmung von sulfat neben
Phosphat mit Hilfe der atomaren Absorptiosspektrometrie. Helv.
Chim. Acta 52:820-827, 1969.
46. Lazrus AL, Hill KC, Lodge JP: A new colorimetric microdetermination
of sulfate ion. Technicon Symposia, New York 65. Automation in
analytical chemistry. Mediad Inc, New York 1966, p. 291.
-129-
-------
47. Altshuller AP: Atmospheric sulfur dioxide and sulfate. Environ.
Sci. Technol. 7(8):709, 1973.
48. Scaringelli FP and Rehme KA: Determination of atmospheric concen-
trations of H2S04 aerosol by spectrophotometry, coulometry, and
flame photometry. Anal Chem 41(6):707-713, 1969.
49. Basargin NN and Novikova KF: Totrimetric micromethod for determining
sulfur in organic phosphorous and arsenic compounds with the indicator
nitrochromeazo. Anal. Khim. 21(4):473-8, 1966. From Chem. Abstr.
65, 6297g.
50. Lukin AM, Chernyshova TV, Avgushevich IV, Kulikova ES. Spectro-
photometric determination of microamounts of sulfates using
Chlorophosphonazo III. Zavod. Lab. 40:22-23, 1974. From Chem.
Abstr. 80, 127825e, 1974.
ACKNOWLEDGEMENTS
This was supported partially by funds from the Air Resources Board.
-130-
-------
TABLE 1 COMPARISON OF MICRO AND MACRO SULFATE ANALYSES
WITH ATMOSPHERIC PARTICIPATE SAMPLES
Micro-SO^
Titrimetric S04~b
Turbidlmetric S04~c»a
1
H
OJ
M
i
Sample
Si
S2
j O
A
Ss
yg/Aliquot
3.710.2
3.510.3
4.610.3
8.110.1
6.010.2
yg/ml
7.310.4
7.010.6
11.610.7
20.310.4
30.011.1
yg/ Aliquot
57016
580112
855120
1145135
1535152
yg/ml
11.410.1
11.610.2
17.110.4
22.910.7
32.711.0
yg/Aliquot
21015
20517
295120
48418
72315
yg/ml
10.510.3
10.210.4
14.811.0
24.210.4
36.210.3
aData shown represent mean values and la valuta for 3 replications of each sample.
bData obtained from 4 replications of each sanple.
cReference (5).
-------
Appendix D
PROTOCOL FOR THE MODIFIED
BROSSET PROCEDURE FOR SULFATE DETERMINATION
A. Introduction
1 9
The original method ' was assembled and modified. This procedure was
employed with atmospheric samples collected on filter media. Sulfate
was removed by aqueous extraction techniques discussed in Sections V
and VI. In the present study, dilution of the extracts was necessary
to provide samples within the working range of this method, ca. 2-10
yg/ml sulfate. The degree of dilution necessary was established by
preliminary analysis with a second sulfate method.
The procedure contains two parts: ion exchange treatment for removal
of cationic interferents and reagent addition plus spectrophotometric
determination.
B. Chemicals and Equipment
1. Chemicals
Amberlite SA-2 ion exchange paper (Reeves Angel, strong acid form)
Adipic Acid (Fisher certified)
-132-
-------
Acetone (Fisher ACS certified)
Thorin (Baker's analyzed reagent)
Barium perchlorate, anhydrous (Pfaltz and Bauer)
72% Perchloric acid (Baker's analyzed reagent)
2. Diluent Solution
Dissolve 10 g adipic acid in about 500 ml acetone. Add 10 ml of
a solution containing 525 mg anhydrous 63(0104)2 in 250 ml 0.1 N
HC104 to the adipic acid in acetone and make up to a volume of 1
liter with acetone.
3. Reagent Solution
Dissolve 250 mg Thorin in 10 ml 0.01 N H2S04 and bring up to 100
ml with distilled water. In a red glass bottle, this solution
showed no signs of deterioration after 6 weeks.
4. Grey filter solution
100 ml of grey filter solution was prepared with 1.5 g CoS04'7H20,
1.0 g CuS04'5H20 and 4.0 g NiCl2'6H20.
-133-
-------
5. Ion exchange filter discs
Circular filter discs were punched out of the ion-exchange paper
sheets by means of a 5/8" "Arch" Punch. The filter discs were
soaked overnight in one liter of 4% HC1, then washed four times
with two liters of glass distilled water over an eight hour
period. The discs were spread out individually on a sheet of Saran
wrap and allowed to air dry overnight in a laminar flow clean bench.
Tweezers were used to insert three discs in each of a series of
plastic syringe bodies.
6. Equipment
Plastic syringe bodies ("Monoject" 12 cc, Catalogue No. 512S)
25 ml beakers
"Rainin" adjustable pipet (0-5 ml)
"Rainin" adjustable pipet (0-200 yl)
Repeatable 5 ml dispensing pipet (Repipet)
C. Ion Exchange Treatment
Add about 3 ml of the sample solution to the filter discs contained in
*(Sherma J: Combined ion-exchange - solvent extraction of metal ions on
ion-exchange papers, Separation Science 2_ 177 (1967)
-------
the syringe bodies and allow to slowly drip through the discs into
a 25 ml beaker. Replace the syringe plunger and depress to remove
an additional amount oif the solution if necessary. Separate sets of
ion exchange filter discs are used for each sample.
D. Reagent Addition and Spectrophotometric Determination
Transfer 2 ml of the treated sample solution, using the 0-5 ml Rainin
pipet to a 2.5 cm cylindrical quartz cell. Add 0.125 ml of reagent
solution using a 0-200 yl Rainin peipet followed by 5 ml of diluent
dispensed from the automatic pipet (Repipet). After capping the cell,
start a timer and shake 5 to 10 times. Read the contents of the cell
after one minute at 520 nm. The zero is set to absorbance 0.800 by
replacing the sample solution with distilled water. A double beam
spectrophotometer was used with a stable grey filter solution in
the reference beam light path. The quartz cell is drained, but not
rinsed, between samples.
The calibration was madSe by using 2 ml of standard solutions containing
from 0 to 10 yg/ml sulf'ate, prepared by stepwise dilution of a more
concentrated (1000 yg 804" per ml) solution. This solution was made
with oven-dried sodium sulfate.
-135-
-------
TECHNICAL REPORT DATA
(Please read Inuniktions on the reverse before completing)
1.
4.
7.
REPORT NO.
EPA-600/2-76-059
2.
TITLE AND SUBTITLE
COMPARISON OF WET CHEMICAL AND INSaiRDMENTAL METHODS
FOR MEASURING AIRBORNE SULFATE
Interim Report
AUTHORtS)
B. R. Appel, E. L. Kothny
J. J. Wesolowski
, E. M. Hoffer, and
9. PERFORMING ORG VJIZATION NAME AND ADDRESS
Air and Industrial Hygiene Laboratory
California Department of Health
2151 Berkeley Way
Berkeley, California 94704
12
15
16
17.
a.
*
*
*
*
it
*
. SPONSORING AGENCY NAME AND ADDRESS
Environmental Sciences Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, N.C. 27711
SUPPLEMENTARY NOTES
3. RECIPIENT'S ACCESSION NO.
5. REPORT DATE
March 1976
6. PERFORMING ORGANIZATION CODE
8. PERFORMING ORGANIZATION REPORT NO.
10. PROGRAM ELEMENT NO.
6NA442
11. CONTRACT/GRANT NO.
EPA 68-02-1660
13. TYPE OF REPORT AND PERIOD COVERED
interim, 6/22/74 - 8/22/75
14. SPONSORING AGENCY CODE
EPA-ORD
.
ABSTRACT
A ' »
Four techniques for determination of water soluble sulfate in atmospheric
samples were compared including the barium sulfate turbidimetric method, the
Brosset (barium-Thorin) method, the automated barium-methylthymol blue procedure
and a micrcchemical (barium-dinitro-sulfanazo lip color imetric method developed
at the Air and Industrial Hygiene Laboratory. These, in turn, were compared to
x-ray fluorescence for determination of total sulfur, obtained independently at the
Environmental Protection Agency's Research Triangle Park Lai-oratory. The parameters
studied included precision and accuracy employing standard solution and ambient
air samples, and the influence of twelve potential interf erents . The ambient air
samples studied were collected at different locations throughout the U.S. so that
the influence of different particle matrices could be evaluated. As supplementary
objectives, analyses of particulate matter samples collected simultaneously on
high volume and low volume glass-fiber filters and low volume Teflon filters,
with and without size segregation, were compared. Results of the study are presented
r
KEY WORD;; AND DOCUMENT ANALYSIS
DESCRIPTORS
Air pollution
Particles
Sulfates
Chemical tests
Chemical analysis
Comparison
«
13. 015 1 RIB'JTICX ST ATL VL \T
RELEASE TO PUBLIC
b. IDENTIFIERS/OPEN ENDED TERMS
-
19. SECURITY CLASS ; 1 ,::s ilepon)
UNCLASSIFIED
20. SECURITY CLASS flliis rwc>
UNCLASSIFIED
c. COSATI Field/Group
13B
14B
07D
21. NO. OF PAGES
14*
22. PRICE
EPA Form 2220-1 (9-73)
136
-------
INSTRUCTIONS
1. REPORT NUMBER
Insert the EPA report number as it appears on the cover of the publication.
2. LEAVE BLANK
3. RECIPIENTS ACCESSION NUMBER
Reserved for us* by each report recipient.
4. TITLE AND SUBTITLE
Title should indicate clearly and briefly the subject coverage of the report, and be displayed prominently. Set subtitle, if used, in smaller
type or otherwise subord.nate it to main title. When a report is prepared in more than one volume, repeat the primary title, add volume
number and include subtitle for ihe specific title.
5. REPORT DATE
Each report snail carry a date indicating at le.ist month and year. Indicate the basis on which it was selected (e.g., date of issue, date of
approve!, dzie of preparation, etc.).
t .
6. PERFORMING ORGANIZATION CODE
Leave blank.
7. AUTHORISI
Give name(sj in conventional order (John R. Doe, J. Robert Doe, etc.). List author's affiliation if it differs frpm the performing organi-
zation.
8. PERFORMING ORGANIZATION REPORT NUMBER
Insert if performing orrinbation wishes to as;.ign this number.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Give name, street, ciu. state, and ZIP code. List no more than two levels of an organizational hirearchy.
10. PROGRAM ELEMENT NUMBER
Use the program element number under which the report was prepared. Subordinate numbers may be included in parentheses.
11. CONTRACT/GRANT NUMBER
Insert contract or pant number under which report was prepared. ""
12. SPONSORING AGENCY NAME AND ADDRESS
Include ZIP code.
13. TYPE OF REPORT AND PERIOD COVERED
Indicate interim final, etc., and if applicable, dates covered.
14. SPONSORING AGENCY CODE
Leave blank.
i
15. SUPPLEMENTARY NOTES
Enter information not included elsewhere but useful, such as: Prepared in cooperation wifh, Translation of, Presented at conference of,
To be published in, Supersedes, Supplements, etc. ^
16. ABSTRACT
Include a brief i200 words or less) factual summary of the most significant information contained in the report. If the report contains a
significant bibliography or literature survey, mention it here.
17. KEY WORDS ANO DOCUMENT ANALYSIS
(a) DESCRIPTORS Selec; from the Thesaurus of Engineering and Scientific Terms the proper authorized terms that identify the major
concept of the research and are sufficiently specific and precise to be used as index entries for cataloging.
(b) IDENTIFIERS AND OPEN-ENDED TERMS - Use identifiers for project names, code names, equipment designators, etc. Use open-
ended terms written in descriptor form for those subjects for which no descriptor exists.
(c) COSAT1 FIFLD GROUP - Field and group .Tsignments are to be taken from the 1965 COSATI Subject Category List. Since the ma-
jority of documents a:e ~v,il;idiseiplinary in nature, the Prinary Field Group assignmentisi'will be specific discipline, area of human
endeavor, or tyre 01 physical ofject. The appiication(s) will be cross-referenced with secondary Field,Group assignments that will follow
the primary posting! s).
18. DISTRIBUTION STATEMENT
Denote re!c_s-'.'::H\ to the pul lie or limitation for reasons other than security for example "Release Unlimited." Cite any availability to
the put-he. \Mth address and price. <
19. & 20. SECURITY CLASSIFICATION
UO NO1 >librr.it classified reports to the National Technical Information service.
21. NUMBER OF PAGES
Insert Lie to.il r.urnoer of pages, including this one and unnumbered pages, but exclude distribution list, if any.
22. PRICE
Insert the price set by the National Technical Information Sen ice or the Government Printing Office, if known.
i Form 2220-1 (9-73) (Rever^c)
------- |