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 RESULTS—FLUOROPORE 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 RESULTS—GLASS 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)

-------