United State*
Environmental Protection
Agency
Environmental Monitoring and Support EPA-600/4 79-028
Laboratory           April 1979
Research Triangle Park NC 27711
Reaaarch and Development
Improvement  and
Evaluation of
Methods for
Sulfate Analysis

-------
                RESEARCH REPORTING SERIES

Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology.  Elimination of traditional grouping was  consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:

      1.  Environmental  Health Effects Research
      2.  Environmental  Protection Technology
      3,  Ecological Research
      4.  Environmental  Monitoring
      5.  Socioeconomic Environmental Studies
      6.  Scientific and Technical Assessment Reports (STAR)
      7.  Interagency  Energy-Environment Research and  Development
      8.  "Special" Reports
      9.  Miscellaneous Reports

This report has been assigned to the ENVIRONMENTAL MONITORING series.
This series describes research conducted to develop new  or improved methods
and instrumentation for the identification and quantification of environmental
pollutants at the lowest conceivably significant concentrations. It also includes
studies to determine the ambient concentrations of pollutants in the environment
and/or the variance of pollutants as a function of time or meteorological factors.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.

-------
IMPROVEMENT AND EVALUATION OF METHODS FOR SULFATE ANALYSIS
                      Final Report
                           by

    B.R. Appel, E.M. Hoffer, M. Haik, W. Wehrmeister,
             E.L. Kothny and J.J. Wesolowski
      Air and Industrial Hygiene Laboratory Section
         California Department of Health Services
                     2151 Berkeley Way
                Berkeley, California
                 EPA Grant No. 805-UVf-l
                    Project Officer

                     John C. Puzak
               Quality Assurance Branch
    Environmental Monitoring and Support Laboratory
     Research Triangle Park, North Carolina  27711
    ENVIRONMENTAL MONITORING AND SUPPORT LABORATORY
        OFFICE OF RESEARCH AND DEVELOPMENT
       U.S. ENVIRONMENTAL PROTECTION AGENCY
   RESEARCH TRIANGLE PARK, NORTH CAROLINA  27711

-------
                                DISCLAIMER






This report has been reviewed by the Environmental Monitoring and Support




Laboratory, U.S. Environmental Protection Agency, and approved for




publication.  Approval does not signify that the contents necessarily




reflect the views and policies of the U.S. Environmental Protection




Agency, nor does mention of trade names or commercial products constitute




endorsement or recommendation for use.
                                    ii

-------
                                 ABSTRACT



A simpler and faster procedure for the manual turbidimetric analysis of


sulfate has been developed and evaluated.  This method as well as a

                                      •o
turbidimetric procedure using SulfaVer , automated methylthymol blue (MTB)

procedures for analysis in the 0-100 yg/ml and 0-10 ug sulfate/ml ranges,


and the Dionex Ion Chromatograph were evaluated for accuracy, precision,

working range, interference effects, and degree of agreement using atmospheric

samples.  Using EPA sulfate audit strips, all methods showed accuracies within

8$ of the accepted value, and coefficients of variation with atmospheric

samples of <^ 6%.  Colloidal clay and yellow, water soluble organics isolated

from atmospheric samples caused interference with all methods.  All the

methods studied provide reliable analyses for 2k hour hi-vol filter samples.

The automated MTB method, modified for use in the 0-10 yg/ml range as

suggested by Colovos,  offers excellent potential for analysis of low volume

samples such as those  provided by a dichotomous sampler network.




This work is submitted in fulfillment of Grant No. 805-M*7-1 by the

California Department  of Health Services under the- sponsorship of the

U.S. Environmental Protection Agency.  This report covers the period

October 1, 1977 to September 30, 1978, and work was completed as of


September 30, 1978.
                                   iii

-------
                                 CONTENTS

Abstract	ill
Figures	  Y
Tables	vii
Acknowledgements	  x'
    I.  Introduction	  1
   II.  Summary and Conclusions	  2
  III.  Improvement and Evaluation of a BaSOij Turbidimetric Method  	  7
   IV.  Evaluation of an Alternative Turbidimetric Technique Using
        SulfaVer?	  59
    V.  Evaluation of the Midwest Research Institute Version of the
        Automated Methylthymol. Blue Method	  73
   VI.  Evaluation of the Colovos Modification of the MTB Method	  90
  VII.  Evaluation of the Dionex Ion Chromatograph	102
 VIII.  Effect of Interferents	115
   IX.  Intermethod Comparison with Atmospheric Samples	12U
References	130
Appendices
    A.  AIHL Method 6l-Determination of Sulfate in High Volume Particulate
        Samples :  Turbidimetric Barium Sulfate Method	131
    B.  AIHL Method 75 (Draft)-Determination of Sulfate in High Volume
        Particulate Samples:  A Simplified and Improved Barium Sulfate
        Method	ikk
    C.  Protocol for the Ruggedness Test for Determination of Sulfate by
        the Turbidimetric Method	155
    D.  Protocol for SulfaVei^IV Procedure	l62
    E.  Tentative Method for the Determination of Sulfates in the
        Atmosphere (Automated Technicon II Methylthymol Blue Procedure)—
        prepared by Midwest Research Institute for EPA-EMSL	163
    F.  Protocol for Ion Chromatographic Analysis of Sulfate and Nitrate.. 180
    G.  Protocol for Extraction of Organics from Aqueous Extracts of
        Atmospheric Particulates for Use in Interference Studies of
        Sulfate Methods	187
                                   iv

-------
                                    FIGURES


Number

   1    Comparison of Curve Pitting Procedures with AIHL Method 6l (Trial I)   9

   2    Relative Accuracy as a Function of Sulfate Concentration with AIHL    18
        Method 6l (Without Sulfate Additions) Third Order Regression for the
        Working Curve

   3    Coefficient of Variation as a Function of Sulfate Concentration       19
        with AIHL Method 6l (Without Sulfate Additions) Third Order
        Regression for the Working Curve

   U    Relative Accuracy as a Function of Sulfate Concentration with AIHL    22
        Method 6l (With Sulfate Additions) Third Order Regression for the
        Working Curve

   5    Coefficient of Variation as a Function of Sulfate Concentration with  23
        AIHL Method 6l (With Sulfate Additions) Third Order Regression for
        the Working Curve

   6    Comparison of Curve Fitting Procedures                                26

   7    Relative Accuracy as a Function of Sulfate Concentration with AIHL    33
        Method 75 (Without Sulfate Additions) Third Order Regression for
        the Working Curve

   8    Coefficient of Variation of the AIHL Method 75 with Atmospheric       3U
        Samples  (Without Sulfate Additions)

   9    Relative Accuracy as a Function of Sulfate Concentration with AIHL    37
        Method 75 (With Sulfate Additions) Third Order Regression for the
        Working Curve

   10    Coefficient of Variation of AIHL Method  75 With Atmospheric Samples  38
         (With Sulfate Additions)

   11    Comparison of Working Curves with Varying BaCl2 Samples Using AIHL    U7
        Method 75
   12     The  Effect  of  BaC^^^O Particle Size on the Slope for the Working    52
         Curve  AIHL  Method 75

   13     Micrographs of Three  Samples of 1*0-50 Mesh Barium Chloride             57
                                                     •p
   Ik     Reproducibility of Working Curve for SulfaVer  Method                  66

   15     Relative Accuracy as  a Function of Sulfate Concentration with          71
         SulfaVer  (Third Order Regression for the Working Curve)

   l6     Coefficient of Variation as a Function of Sulfate Concentration        72
         (Third Order Regression for the Working Curve) SulfaVer  Method

   17     MTB  Assay by Midwest  Research Institute Procedure-February 2,  1978     75

-------
18    Working Curve for MRI-MTB Procedure-February 27, 1978                 T9

19    Relative Accuracy as a Function of Concentration for the MRI-MTB      88
      Method

20    Coefficient of Variation as a Function of Concentration for the       89
      MRI-MTB Method

21    Comparison of Working Curves as a Function of Ba/MTB Ratio            9^

22    Typical Working Curve for Colovos-MTB Method in 0-10 yg/ml Range      96

23    Relative Accuracy as a Function of Sulfate Concentration with the    100
      Colovos-MTB Method

2k    Coefficient of Variation as a Function of Sulfate Concentration with 101
      the Colovos-MTB Method

25    Resolution of Sulfate and Nitrate at Equal Concentrations by Ion     103
      Chromatography

26    Reproducibility of Working Curve for Dionex Ion Chromatograph        106
      (Range:  10 ymho)

27    Comparison of Working Curves for Dionex Ion Chromatograph on Ranges  107
      1, 3 and 10 ymho

28    Regression Equation Slope Against ymho for Full Scale with Dionex    109
      Ion Chromatograph

29    Relative Accuracy as a Function of Sulfate Concentration with the    113
      Dionex Ion Chromatograph  (.Range 10 ymho)

30    Coefficient of Variation as a Function of Sulfate Concentration      llU
      with the Dionex Ion Chromatograph (Range 10 ymho)

31    Scatter Diagrams of Results With Atmospheric Sample Using Five       129
      Sulfate Methods
                                   VI

-------
                                    TABLES


Number                                                                     Page

   1     Summary of Methods' Evaluations                                       5

   2     Curve Fitting as a Function of Concentration for AIHL Method 6l    11
         BaSOij TurMdimetric Method Without Sulfate Additions Using
         Log-Log and Linear Regression

   3     Curve Fitting as a Function of Concentration for AIHL Method 6l    12
         BaSO^ Turbidimetrie Method Without Sulfate Additions Using
         Third Order Regression

   U     Accuracy as a Function of Concentration for AIHL Method 6l BaSOit   lU
         TurMdimetric Method with 300 yg/20 ml Sulfate Addition and
         Linear Regression

   5     Accuracy as a Function of Concentration for AIHL Method 6l BaSO^   15
         TurMdimetric Method with 300 yg/20 ml Sulfate Addition and
         Third Order Regression

   6     Relative Accuracy  and Precision of AIHL Method 6l BaSO^            IT
         TurMdimetric Method (without Sulfate Additions) with an
         Atmospheric Sample  (yg S0it=/20 ml solution)

   7     Relative Accuracy  and Precision of AIHL Method 6l BaS04            21
         Turbidimetric Method (with Sulfate Additions) with an
         Atmospheric Sample  (yg S0tt=/20 ml Solution)

   8     Curve Fitting as a Function of Concentration AIHL Method 75        27
         Without Sulfate Additions Log-Log and Linear Regression

   9     Curve Fitting as a Function of Concentration AIHL Method 75        28
         Without Sulfate Additions Third Order Regression

   10     Accuracy as a Function of Concentration for the AIHL Method 75     29
         with 300 yg/20 ml  Sulfate Addition and Linear Regression

   11     Relative Accuracy  and Precision of the AIHL Method 75  (without     32
         Sulfate Additions) with Atmospheric Samples  (yg S0tt=/20 ml
         solution)

   12     Relative Accuracy  and Precision of AIHL Method 75  (with Sulfate    36
         Additions) with Atmospheric Samples (yg S0it=/20 ml solution)

   13     Analysis of EPA Sulfate Audit Strips by the AIHL Method 75         39

   lU     Summary of Results of Turbidimetric Methods                        Hi

   15     Factors for Evaluation in Ruggedness Test                          1*3

   16     Results of Ruggedness Test of Sulfate Analysis by Turbidimetry     kh

   17     Description of BaCl2 Used for Turbidimetric Method                 k&


                                     vii

-------
18
       Size Distribution of J. T. Baker BaCl2-2H20 (% by Weight)
19     Working Curves for Method 75 Turbidimetric Method With Size          51
       Fractionated BaCl2-2H20 (Linear Regression)

20     Comparison of Observed Working Curve Slopes for AIHL Method 75       53
       With Those Calculated from Individual Fractions

21     Analysis of BaCl2-2H20 for Trace Metals (ppm by Weight)              55

22     Comparison of Dissolving Times for Three Samples of BaCl2            56

23     Conditions Studied in Developing Protocol for SulfaVer  IV           60
       Procedure

2U     Influence of Protocol Variables on the Working Curve Using  -         62
       SulfaVerR IV (Third Order Regression)

25     Influence of Protocol Variables on the Working Curve Using           63
       SulfaVerR IV (Linear Regression)
                                                                    j>
26     Curve Fitting as a Function of Concentration Using a SulfaVer  IV    65
       Procedure
                                                    TJ
27     Reproducibility of Working Curve for SulfaVer  Method  (Third Order   67
       Regression)
                                                         R
28     Analysis of EPA Sulfate Audit Strips by a SulfaVer  Method           68

29     Relative Accuracy and Precision of SulfaVer  Turbidimetric Method    70
       With an Atmospheric Sample (yg SO^2 /20 ml)

30     Calculation of MTB Purity                                            77

31     Assay  of MTB by Midwest Research Institute Method                    78

32     Reproducibility of Calibration:  Sulfate by MRI-MTB Method           80

33     Curve  Fitting as a Function of Concentration for the MRI-MTB Method  82

3^     Effect of  a 5$ Error in MTB Assay Upon the Calibration Curve for     83
       the MRI-MTB Method

35     Analysis of EPA Sulfate Audit Strips by the MRI-MTB Method           8U

36     Relative Accuracy and Precision of MRI-MTB Method With an            86
       Atmospheric Sample  (yg SO^ /ml)Linear Regression

37     Relative Accuracy and Precision of MRI-MTB Method With an            87
       Atmospheric Sample  (yg SO^ /ml) Third Order Regression

38     Straight Line Fitting as a Function of Concentration for the         92
       Colovos-MTB Method  (% Error)


                                   viii

-------
39    Reprodueibility and Linearity of the Working Curve for the            95
      Colovos-MTB Method (0-10 yg/ml)

UO    Analysis of Sulfate EPA Audit Strips by the Colovos-MTB Method        98

Ul    Relative Accuracy and Precision of Colovos-MTB Method With an         99
      Atmospheric Sample (yg
H2    Reproducibility of Working Curve for Dionex Ion Chromatograph        105
      (Range:  10 ijmho)

U3    Analysis of EPA Sulfate Audit Strips by the Dionex Ion Chromato-     110
      graph

UU    Relative Accuracy and_Precision of the Dionex Ion Chromatograph,     112
      Range 10 ymho  (yg S0
U5    Sulfate and Interferent Levels for Interference Studies              Il6

U6    Interference Effect With the AIHL Method 75 (wg/ml Observed Sulfate) 118

^T    Interference Effects With the MRI-MTB Method                         119

kS    Interference Effect With the Colovos MTB Method (yg/ml Observed      121
      Sulfate)

1*9    Interference of Nitrate in Sulfate Determination With the Dionex     123
      Ion Chromatograph

50    Description of 2U Hour Hi-Vol Filter Samples for Intermethod         125
      Comparison

51    Results of Intermethod Comparison With Atmospheric Samples (yg/ml)   127

52    Average Agreement and Precision of Methods With Atmospheric Samples  128
                                    IX

-------
                             ACKNOWLEDGEMENTS









Other participants in this study included Ms. S. Twiss who supervised data




reduction and Ms. L. Raftery who provided capable assistance in the laboratory.









The atmospheric samples used in this study were provided by Mr. J. Wendt,




California Air Resources Board.  His cooperation and assistance are




gratefully acknowledged.









Mr. J. C. Puzak served as Project Officer for this program.  His helpfulness




throughout this work has been sincerely appreciated.

-------
I.   INTRODUCTION



    In preceding EPA-sponsored programs, a series of wet chemical sulfate



    methods was evaluated and compared to total sulfur determinations by


                                1 2
    x-ray fluorescence analysis. '   These methods were:

                                                                              2

    —A barium sulfate turbidimetric procedure (Public Health Service Version)



    —An automated methylthymol blue (MTB) procedure



    —The AIHL microchemical method


                                                                       7
    —Two modifications of the thorin method as developed by C. Brosset

                                         o

    —A manual barium chloranilate method .
    The current program continued these studies employing improved or



    alternative versions of some of these procedures.  The specific objectives



    included,  (l) improving and evaluating for ruggedness a barium sulfate



    turbidimetric method (AIHL Method 6l) for sulfate analysis, and (2) an



    evaluation of a series of sulfate methods for comparability, accuracy,



    precision, working range and effect of interferences.  These methods were:

                                                        •p

    a  barium sulfate turbidimetric method using SulfaVer , the Midwest Research



    Institute  version of the MTB method  , the MTB method as modified by Colovos


                                        11                                   12
    for analysis in the 0-10 yg/ml range  , and the Dionex ion chromatograph.
                                    - 1 -

-------
II.  SUMMARY AND CONCLUSIONS


     Modification of the AIHL Method 6l, a version of the BaSO^ turbidimetric

     method, to permit reaction of samples in inexpensive, capped test tubes

     directly usable for measurement of turbidity eliminates the need for

     multiple transfers of samples.  The resulting procedure, designated AIHL

     Method 75, shoved slightly improved precision and increased accuracy in

     the non-linear range below 300 ug sulfate/20 ml solution when using third

     order regression techniques.  The addition of fixed quantities of sulfate

     to low concentration samples  (to permit their analysis in the nearly

     linear region of the working  range) was found inferior to direct analysis

     using third order fitting of  the working curve.  Defining the working
                                                                     *
     range as  that providing approximately constant relative accuracy  and

     coefficient of variation, for Method 75 the working range was from 1^0

     to at least lUOO yg S04~/20 ml samples  (7 to 70 yg S04~/ml) using third

     order regression.




     An eleven parameter ruggedness test was performed utilizing the pro-

     cedures  of both AIHL Methods  6l and 75 to reveal areas contributing to

     variability in  sulfate determinations by turbidimetric methods.  From

     this test only  the age of the pre-mixed glycerin-HCl-water reagent was

     found  to be a statistically significant contributor to such variability.




     Methods  6l,75,  and similar procedures call for use of crystalline barium

     chloride dihydrate specified  by its manufacturer as being "For Parr turbi-

     dimeter" with or without a specified mesh size (e.g. 20 mesh in the Old

     U.S.  Series mesh  size).  The  current unavailability of barium chloride
  *The analytical results of the method with  an  atmospheric sample of varying
   dilution relative to that obtained for  the same sample in the optimal
   concentration range of the method.
                                     — 2 —

-------
graded for use in turbidimetric sulfate analyses prompted an investigation



of the influence of mesh size and impurities in BaCl^-SI^O on its suita-


bility for these analyses.  The ideal barium chloride was considered to


be one yielding linear working curves with relatively high slope values,


small intercepts and little•scatter.  It was determined that size-


fractionated Bad2 is not mandatory for turbidimetric analyses.  For a


given reagent, the resulting working curve slope increased with increasing


BaCl2-2H20 particle size.  Differences in results between reagents with the


same mesh size were explained by differences in the crystals in the


aggregates composing the particles, e.g. smaller crystal size promoted


more rapid dissolution rates with correspondingly increased slopes.




                                                                    •p
A study of an alternative turbidimetric procedure utilizing SulfaVer ,


a mixture of barium chloride crystals and a proprietary colloid stabilizing


agent, was included in this program.  After optimizing the analytical


protocol, an evaluation of this procedure yielded similar results to


those for Method 75.






Studies of the automated methylthymol blue (MTB) procedures included


evaluations of the Midwest Research Institute (MRl) method for 0-100 yg/ml


sulfate and the modification made by G. Colovos and co-workers for 0-10


yg/ml sulfate.  The working range for the MRI-MTB procedure, while ex-


panded at the upper end, was notably compressed at the lower end relative

                                                         2
to the version of the MTB procedure previously evaluated.   The latter


permitted analysis down to 7 yg/ml (linear regression) and to ca. H yg/ml


with third order regression.  Modification of the MRI procedure to eliminate


a sample memory effect as a cause of its reduced range is here recommended.




                               - 3 -

-------
Both the MRI and Colovos methods employ relatively time consuming




procedures for achieving more nearly linear working curves.  The




procedure yielded a working curve which was fit substantially better




by two straight lines or third order regression than by a single line.




The Colovos procedure yielded working curves in the 0-10 vg/ml with




acceptable linearity (e.g. correlation coefficient = 0.997 for a least




squares line).








The Dionex 1C chromatograph, thermostated for use at 35°C, provided




sulfate determination with generally lower accuracy and precision compared




to the other methods.  Peak heights provided somewhat better precision




and accuracy compared to integrated peak areas.  A small positive inter-




ference from nitrate was identified as one source of decreased accuracy.








Colloidal clay, water soluble yellow organics isolated from atmospheric




particulate matter, phosphate and bicarbonate were evaluated as potential




interferents with turbidimetric and methylthymol blue methods.  Colloidal




clay  and  organics caused interference in all cases studied but with differing




sign.








Table 1 summarizes  results for the methods studied with respect to accuracy




and precision,  the  concentration ranges providing relative accuracy and




precision within 10$  and  5$, respectively, and the effect of interferences.

-------
    Method

Method 75



SulfaVerR

MRI-MTB (linear regression)

MRI-MTB (third order)



Colovos-MTB



Dionex-IC (Range 10 vimho)
                                                      Table 1

                                          Summary of Methods Evaluations
                                       EPA Audit Strips
                                    Concentration Range
                                   for Relative Accuracy
                                     Within 10% and
                                      Significant Interferent
Mean Obs/Theoret.
l.OU
1.06
1.06
l.OU
0.98
Mean C.V. (%)a
3.8
3.6
2.8
3.U
1.3
C.V. <
10
9
17
lit
2
5$ (pg/ml)
to >_ 70b
to >_ 70b
to >_ 90
to >^ 90
to >_ 10
Positive
	
Negative
colloidal
clay,
organics
not determined
not determined
colloidal
clay,
organics
colloidal
clay,
organics
	
	
1.08
6.2
7 to 130
nitrate
a.  Coefficients of variation for samples within working range of method.

b.  Employing 20 ml samples.

c.  Working range as defined by concentration range providing approximately constant relative accuracy and
    C.V. was 7 to 111 ug/ml.

-------
   An  intermethod  comparison with the methods  cited in  Table 1 plus the AIHL



   microchemical sulfate method  * using 2k  hi-vol  filter  sample extracts




   demonstrated, on average, agreement within  10% for all  methods.
    The regression equations  relating results between the Method  75  &&& the




    other procedures are as follows :




    SulfaVer =1.10 (Method 75)  - 0. 570      r  =  0.998      Sy>x =3-00




    MRI-MTB = 1.06 (Method 75)  - 0.759      r  =  0.997      Sy>x = 3.^3




    Colovos MTB = 1.01 (Method  75) +  0.77^   r  =  0.999      Sy>x = 1.72




    Dionex 1C = 1.08 (Method 75) - 0.022     r  =  0.998      Sv Y = 2.76
                                                           jf * -X.



    AIHL Micro = 1.06 (Method 75) +  0.851   r  =  0.998      Sy>x = 2.51*







                                                                  •p

    The present study indicates that  Method 6l, Method  75,  SulfaVer  ,  Dionex




    1C (range 10 ymho) and the  automated  MRI-MTB  methods are suitable  for




    analysis of hi-vol samples.   Choice between methods should  reflect equip-




    ment availability and number of  samples.  The MBI-MTB procedure  can




    probably be improved by decreasing the  samples per  hour.  For low  volume




    samples the present study suggests that the Colovos-MTB method is  suitable



                                                                         *
    if at least 4 ml of solution in  the range 2 to 10 yg/ml is  available.




    As such, it may be the wet  chemical method  of choice for sulfate analysis




    of samples from a dichotomous sampler network.
*

 For comparison, the manual AIHL microchemical method requires < 1 ml of sample

 in the range from about 1 to 15 yg/ml.                         ~
                                   - 6 -

-------
III.  IMPROVEMENT AND EVALUATION OP BaSO^ TURBIDIMETRIC METHOD




      A.  Introduct ion




          Efforts to improve the turMdimetric method were focused on three




          aspects:




          —The addition of a known amount of barium sulfate to all samples to




            eliminate reliance on the non-linear portion of the working curve.








          —Mixing of reagents and samples in inexpensive, screw-capped test




            tubes suitable for measuring turbidity, eliminating the need for




            multiple transfers from graduated cylinders prior to determining




            turbidity values.








          —Use of a relatively inexpensive spectrophotometer (B&L 20) in place




            of a B&L TO.








          The specific parameters included in the evaluation methods were,




          l) reproducibility and linearity of the working curve, 2) accuracy and




          precision using EPA audit samples (glass fiber filter strips spiked




          with known quantities of sulfate), 3) working concentration range, and




          k) addition of fixed quantities of sulfate to samples to permit analyses




          of low sulfate concentration samples in the linear region of the working




          curve (see below).








          The standard of comparison for the current efforts to provide an




          improved turbidimetric procedure was the Air and Industrial Hygiene




          Laboratory Method 6l (Revised July 1976).  The procedure is included




          as Appendix A.
                                      - T -

-------
   B.  Evaluation of AIHL Method 6l


       1.  Evaluation of Working Curves Without Sulfate Addition


           In AIHL Method 6l the reaction of sulfate with barium is carried


           out in graduated cylinders with transfers to a 20 mm diameter


           cuvet for measurements using a B&L Spectronic TO.




           Working curves were prepared, in three trials covering the con-


           centration range UO-2000 yg sulfate/20 ml.  Standards and samples


           were dispensed in 10 ml of water followed by 10 ml of distilled

                 *
           water.   This contrasts with the usual procedure which employs


           20 ml of samples or standards.  In the present study, observed


           concentrations expressed as yg/20 ml are equal to the yg/10 ml


           of the initial sample.  As written, Method 6l employs linear


           regression in the range 300-1000 yg/20 ml and restricts the working


           curve to levels <_ 1500 yg/20 ml because of reported non-linearity.


           No evidence of non-linearity was observed in the present study up


           to 2000 yg/20 ml.  Below 300, the curve is non-linear and a


           graphical method of data reduction is recommended.  The present


           evaluation employed non-linear regression techniques to provide


           more  objective results for method comparison.  Two curve fitting


           procedures were evaluated:  l) A log-log fit for the range UO-2^0


           yg/20 ml followed by linear regression in the range 300-2000


           yg/20 ml, and 2) a single third order polynomial fit throughout


           the range.  Figure 1 illustrates the results for one trial.  The


           regression equations and the average goodness of fit by the


           two procedures as a function of concentration are given
£
For trials  with sulfate  addition, the  10 ml of distilled water was replaced

by  10 ml of solution containing  a known amount of sulfate.
                                   -  8  -

-------
                                             Figure i


                COMPARISON OF  CURVE  FITTING PROCEDURES WITH AIHL METHOD 61 (TRIAL I)
    1.25
    1.00
    0.75
HI
o
c
to
.o
n
o
01
    0.50
    0.25
             3rd Order



             Log-Log (0-2^0)

             Linear (300-2000)
                200
400
600      800     1000    1200     1400


    Sulfate Concentration (pg/20ml)
1600
1800
2000

-------
in Tables 2 and 3.  These results show that:




1) third order regression curve fitting is adequate down to 80 yg/20 ml




and the log-log fit, to ^0 yg/ml, 2) considering both mean % error




and coefficient of variation, the third order fit is superior to the




log-log fit + linear regression for the range 80-2000 yg/20 ml




sulfate, and 3) in the range 300-2000 yg/20 ml, the % error by both




third order and linear regression averaged + .2 and -.2, respectively.




In general, the two curve fitting procedures differed by less than 3%.









Although the results by the log-log fit imply that analysis down to




kQ yg/20 ml might be feasible, additional perspective on the lower




end of the working range with standards was obtained by considering




the variability in the initial turbidity reading (read as % trans-




mission, TI , and subsequently recalculated as absorbance).  Our




experience with the turbidimetric method suggests that an uncertainty




of up to +_  .5% transmittance exists due to such variables as residual




bubbles, colloidal matter in the sample and slight variations in




positioning the cell in the holder of the spectrometer.  This




sharply increases the coefficient of variation at low levels.  For




example, assuming no imprecision in value for T2 (after BaCl2 addi-




tion) , the results indicate at 160 yg/20 ml a C.V. of kO% due only




to this estimated maximum uncertainty in Tj, the value increasing




to 80$ at kO yg/20 ml.  This source of imprecision may be especially




significant when the method is not carried out with maximum care.




The present work is being carried out with the greatest possible




effort to minimize such sources of imprecision.  The above cal-




culation, therefore, is not considered relevant to the current






                        - 10 -

-------
H
I
                                                         Table  2

                Curve Fitting as a Function of Concentration for  AIHL Method  6l  BaSO^ Turbidimetric Method
                              Without Sulfate Additions  Using Log-Log and  Linear Regression
True Sulfate
yg/20 ml
Uo
80
120
160
200
2k°
300
500
700
900
1200
1600
2000

Mean Observed SOit C.
^5-1 + 5.5
69.3 + 6.8
113.8 + 9-5
156.3 + 3.3
201.2 + 11.1
266.9 + 21.U
318.9 + 13. U
1*82.0 + 8.2
672.7 + 12.8
890.1 + 32.0
1255 +105
1591* +11.1
1988 +37.8

.V. (%}
12.2
9.8
8.3
2.1
5-5
8.0
U.2
1.7
1.9
3.6
8.U
0.7
1.9
Mean Obs. SO^
True S0u=
1.13
0.87
0.95
0.98
1.01
1.11
1.06
0.96
0.96
0.99
1.05
1.00
0.99
T_
Mean % Error0
12.8
- 13.U
- 5-2
- 2.3
0.6
11.2
6.3
- 3.6
- 3.9
- 1.1
U.6
- O.U
- 0.6
Coefficients for Regression Equations
For S0^~ < 300 yg/20
Trial a

I .000077
II .000051
III .000157
ml (S0k ) = a (Absorbance)
b r S
y.x
1.23 .998 .028
1.29 .995 .QUO
1.08 .957 .106
For



^ .
— .
301+ >_ 300 yg/20 ml (f
c d r

137 .000722 .9996
123 .000712 .9995
127 .0007UU .9920
SQjt ) = c + d (Absorbance)
S
yx
0.0136
0.01U5
0.0635
          a.  Calculated from the regression equations, yg/20 ml.
          b.  [(mean obs. - true)/(true sulfate)] x 100

-------
                                         Table 3

Curve Fitting as a Function of Concentration for AIHL Method 6l BaSO^ Turbidimetric Method
                   Without Sulfate Additions Using Third Order Regression
True Sulfate
pg/20 ml
Uo
80
120
160
200
2UO
300
500
700
900
1200
1600
2000


Trial e

I - 1.89 x
II - 2.19 x
III - U.87 x

Mean Observed SO^
75- U + U.7
90.6 + 8.5
121.6 + U.9
151-5 + 2.1
182. U + 10.6
226.3 + 11.3
291.6 + 7-9
U95.5 + 6.9
697.2 + 11.9
906.3 + 39.9
1235 +60.5
1555 +37.3
2038 +_ U6.9
Coefficients for
SO^" = e + f x + gx2 + hx3
f • g

10~2 3.33 x lO"4 3.U2 x
10~2 3.35 x ID"4 3.69 x
10~3 1.66 x 10"4 6.97 x

C.V.(*)
6.2
9.U
U.O
l.U
5.8
5.0
2.7
l.U
1.7
U.U
U.9
2.U
2.3
Regression
For Uo


10~7
ID'7
10~7
Mean Obs.
True S0i4
1.89
1.13
1.01
0.95
0.91
0.9U
0.97
0.99
1.00
1.01
1.03
0.97
1.02
Equation
<_ S0it~ < 2000
h

0.87U x 10~10
1.0U x 1(T10
2.26 x 10~10
S04
Mean %
88.5
13.2
1.3
- 5-3
- 8.8
- 5.7
- 2.8
- 0.9
- o.U
- 0.7
2.9
- 2.8
1.9

yg/ml
r

.99997
.99996
.9997

Error















S
y *x
.0119
.0138
.0378

-------
    program, but may offer insight relevant to ruggedness testing.









2.  The Accuracy of AIHL Method 6l with Sulfate Addition




    In place of 10 ml of distilled water three hundred yg of sulfate




    in 10 ml water was added to each of a set of pure sulfate solutions




    covering the range UO to 1600 ug/20 ml.  The resulting solutions




    were analyzed by AIHL Method 6l using both linear and third order




    regression for working curves in the range 300-2000 yg/20 ml.




    The mean results for three trials, after subtracting the sulfate




    added, are compared to the true sulfate values in Tables U and 5.








    The accuracy with standards, adding fixed quantities of sulfate,




    was generally excellent.  The accuracy using third order regression




    for standards ranging from UO to 500 ug/20 ml was significantly




    better than with linear regression but the precision was generally




    poorer in this range as well as above 500 yg/20 ml.








    These results may be compared to the goodness of fit to the non-




    linear working curves below 300 pg/20 ml, given in Table 3, as a




    measure of the accuracy of the method without addition of fixed




    amounts of sulfate.








    In the range 80-2^0 yg/20 ml the mean error (using absolute values




    for the values given in the Table) with addition of fixed quantities




    of sulfate is 1.2$ and 3.9$ by third order and linear regression,




    respectively.  This compares to a mean error of 5-1$ for the log-




    log and third order fit of the working curve (without addition of




    sulfate) in the range 80-2^0 yg/20 ml.





                           - 13 -

-------
                                Table  fc

             Accuracy as a Function of Concentration for
            AIHL Method 6l BaS04 Turbidimetric Method with
         300 Mg/20 ml Sulfate Addition and Linear Regression
True Sulfate
  less 300
 (MR/20 ml)
     80

    120

    160

    200

    2*4-0

    300

    500

    700

    900

   1200

   1600
    Mean
Observed S04
=b
                  Mean Obs. S04"
 Trial

   I
   II
   III
(Mg/20 ml)
1*7.0 + 6.9
80.7 + 6.1
118.0 +6.2
153 A ± 15-6
186.0 + 10.5
225.6 + 14.0
291.7 + 14.1
^82.1 + 16.1
693.6 + 10.2
900.3 + 21.2
1218 + 9.2
1597 + 20.1
Coefficients for the
a b
-.123 0.000737
-.13^ 0.000729
-.138 0.000731
C.V.
1U.6
7.5
5-3
10.2
5-7
6.2
4.8
3.3
1.5
2.k
0.8
1-3
True Sulfate
1.17^ + .172
1.009 + .077
0.983 ± .053
0.958 + .098
0.930 + .053
0.940 + .059
0.972 + .048
0.964 + .034
0.991 + .018
1.000 + .026
1.015 ± -013
0.998 + .016
Linear Working Curve
r
0.9925
0.9996
0.9997
sy.x
O.C608
0.0131
0.0124
 Mean ,
i Error

 +  17 = 5
 +   0.9

 -   1.7
 -   4.1
 -   7.0
 -   6.0
 -   2.8
 -   3.6
 -   0.9
    0.0

 +   1.5
 -   0.2
 a.   A linear regression  curve was prepared from 6 points in the range
     300-2000 Mg/20 ml.   An independent set of samples was prepared from
     sulfate standard solutions  covering  the range shown.  To each sample,
     300 Mg sulfate was added, the total  sulfate determined from the
     working curve and 300 ug subtracted  from the resulting values.

 b.   Obtained for three independent  trials.

 c.   A precision of + 1$  was estimated for the "true  sulfate" concentration
     to calculate a values.

 d.   Mean Obs.-True
          True
                                    -14- '

-------
                                Table 5

             Accuracy as a Function of Concentration for
            AIHL Method 6l BaS04 Turbidimetric Method with
       300 Mg/20 ml Sulfate Addition and Third Order Regression2
True Sulfate
  less 300
    Mean
Observed S04"
                            =b
Mean Obs. S04   . ^c
                                                                      Mean
(Mg/20
MP
80
120
160
200
2 1*0
300
500
700
900
ml) (Mg/20 ml)
1*1.1* + 13-1
80.5 ± 11.2
122.5 ± 11.3
162.0+ 1*.8
196.9 ± 2.6
239.0 ± 7.9
307.3 + H.7
1*95.1 + 23.7
69!*. 5 ± 27.0
886.5 ± ^3.0
1200 1187,9 ±32.2
1600
Trial
I.-.
II
III
1597.3 ± 7.6
Coefficients for
a b
-8.10l*xlO~ 1*. 635xlO~
-l*.865xlO~2 l*.253xlO"4
-5.830xlO"2 1*. 1*28x10
C.V.
31-7
13-9
9.2
3-0
1.3
3.3
3.8
I*. 8
3-9
i*.9
2.7
0.5
Third Order
c
3.820x10
2.893x10
2.803x10
True Sulfate -
1.036 + .328
1.006 ± .11*0
1.020 + .095
1.012 + .032
0.985 ± .016
0.996 + -.031*
1.021* + .Ql*0
0.990 + .01*8
0.992 + .oi*o
0.935 ± .oi*9
0.990 ± .029
0.998 ± .oil
Working Curve
d
"I -1.333x10"^
-0.796x10" g
"7 -0.78Uxlo"10
°lo Error
+ 1.0
+ 0.6
+ 2.1
+ 1-3
- 1.6
— .1*
+ 2 .1*
- 1.0
- 0.8
- 1.5
- 1.0
0.2

0.9982 .0675
0.9999 -0025
0.9999 .0022
a.  A third order regression curve was prepared from 6 points in the range
    300-2000 jug/20, ml.  An independent set of samples was prepared from
    sulfate standard solutions covering the range shown.  To each sample,
    300 Mg sulfate was added, the total sulfate determined from the working
    curve and 300 Mg subtracted from the resulting values.

b.  Obtained for three independent trials.

c.  A precision of + 1.% was estimated for the "true sulfate" concentration
    to calculate a values.
    Mean Obs.-True
          True
                   x 100
                                   -15-

-------
3.   Working Range Without Sulfate Additions




    A single atmospheric sample extract was prepared by pooling a




    large number of single filter extracts.  The resulting solution




    was centrifuged cold at 6000 rpm to remove suspended particles and




    diluted to varying degrees to permit evaluation of the "relative




    accuracy" and precision as a function of concentration.   The term




    relative accuracy reflects    the accuracy of the analysis at a




    given degree of dilution relative to the analytical results from




    analyses done in what is considered the optimal concentration range




    of the method.  Relative accuracy and precision were used to




    evaluate the working range of the sulfate method (i.e.,  the




    concentrations which show both approximately constant relative




    accuracy and coefficient of variation).








    The results of the Method 6l, without addition of standard amounts of




    sulfate, are compiled in Table 6, employing third order  regression




    of the working curve.  The calculated, undiluted concentration of




    the extract showed no apparent bias between 15 and 100$  concentration.




    Thus, the correct undiluted concentration was taken to be the mean




    of seven results, lUl8 +_ ^9 Vg/20 ml.  The ratio of observed to




    expected concentrations are plotted against expected concentration




    in Figure 2.  The results indicate accuracy within J% of the optimal




    value in the concentration range from 71 to lUl8 yg/20 ml.  The




    coefficient of variation of three independent trials is  plotted




    against expected concentration in Figure 3.  The C.V. values remain




    less than 8% in the range 106 to 1^18 yg/20 ml.
                            - 16 -

-------
                                                       Table 6

                   Relative Accuracy  and  Precision of  AIHL Method 6l BaS04 Turbidimetric Method
                   (without Sulfate Additions) with an  Atmospheric Sample  (jug S04"~/20 ml solution)


Concentration,
1











i
H
—3
1
b of Undiluted
2.5
5.0
7.5
10.0
12.5
15.0
18.75
31.25
43.75
56.25
75.0
100

Observed
Mean of
3 Trials
67.7
75-8
108.1
139-4
160.1
206.9
250.5
436.1
631.0
827.1
1091.9
1448.3



g.
4.8
13.7
2.5
9.2.
12.5
11.6
8.1
11.7
2.9
8.1
18.8
8.4




-------
 I

l->
cx>
o

T3
            0)
            CO
               •H

               4-J
               0)

               O
   •u
   01
   4J
   O


   a

   w
                                                   Figure  2


                           RELATIVE ACCURACY AS A FUNCTION OF SULFATE CONCENTRATION

                               WITH AIHL METHOD 61 (WITHOUT SULFATE ADDITIONS)

                                THIRD ORDER REGRESSION FOR THE WORKING CURVE
                   2.00.-
                   1.60
        1.20
                   0.80
                   0.40
                                           I
                                         I
I
I
I
I
                                         400                 800               1200


                                             Expected Concentration  (ug  SOr720ml)
                                                                                       1600

-------
          40
                                                      Figure 3
                          COEFFICIENT  OF  VARIATION  AS A  FUNCTION OF  SULFATE CONCENTRATION
                                  WITH AIHL METHOD  61 (WITHOUT SULFATE  ADDITIONS)
                                    THIRD ORDER  REGRESSION FOR THE WORKING CURVE
          30
       e
       o
H
MD
>
M-t
O
       cs
       
-------
   Based on both relative accuracy and precision as measured by




   coefficient of variation, the working range of Method 6l, without




   sulfate additions, is from about 100 to >_ ll*00 yg/20 ml using




   third order regression.








 .  Working Range with Sulfate Addition




   The same atmospheric sample used above was analyzed by the Method 6l




   after addition of 300 yg sulfate to each sample.  Since the data




   analysis without sulfate addition was done with a third order




   regression fit for the working curve, the same procedure was used




   with added sulfate.  The results, after subtracting 300 from the




   observed sulfate, are shown in Table 7.  The correct, undiluted




   sulfate concentration for the atmospheric sample was taken to be




   lUl8 yg/20 ml  (Table 6).  The ratio of observed to expected con-




   centrations and the C.V. are plotted against expected concentrations




   in Figures U and 5, respectively.  The results indicate relative




   accuracy within 10% in the range from about TO to lUOO yg/20 ml.




   The C.V. values remained constant within 8% over the range 106 to




        yg/20 ml.
5.   Conclusions




    Using AIHL Method 6l,  both with and without addition of fixed




    quantities of sulfate, third  order regression of the working curve




    data provides a working  range from about 100 to at least lUOO  yg




    sulfate/20 ml.   Within this range, the procedure is not improved




    by the addition of fixed quantities of sulfate to each sample.









                            - 20  -

-------
                                                      Table  7

                   Relative Accuracy  and Precision of AIHL Method 6l BaS04 Turb.idimetric Method
                     (with Sulfate Additions) with an Atmospheric Sample  (ug SQ^/ZQ ml Solution)
i
ro
     Concentration,
     % of Undiluted

           2.5
           5-0
           7-5
          10.0
          12.5
          15.0
          18.75
          31.25
          43.75
          56.25
          75.00
         100.0
Observed Mean
Expected
Observed Cone.
       lirial

        I
        II
        III
of 3 Trials8- a o2















•8
'9
6
35.3
71.1
101.7
132.1
163.3
199.2
251.6
426.5
617.4
810.4
1098.8
1464.7

a

.075 x 10"2
.096 x 10"
.227 x 10"2
8.6
9.2
7.8
6.1
3.6
2.6
6.5
9-5
U.7
10.4
32.9
29.0
Coefficients
b
^ggggMM
5.621
5.828
4.588
74.4
85.1
61.2
37.4
13.2
6.9
42.1
89.4
137.5
107.8
1084.8
842.5
for Third


x 10"4
X 10~t
x 10"
C.V. do]
24.4
13.0
7.7
4.6
2.2
1.3
2.6
2.2
1.9
1.3
3.0
2.0
Order Working^
c

1.242 x 10"7
1.242 x 10",;
2.476 x 10
Diluted Cone. Expected Cone.
35-5
70.9
106.4
141.8
177-3
212.7
265.9
443.2
620.5
797.7
1063.7
1418.2
Curves
d

-2.738 x 10"";
-3.429 x 10"
-6.555 x 10
0.995
1.002
0.956
0.931
0.921
0.936
0.946
0.962
0.995
1.016
1.033
1.033
..... .. 	
Str v
y.x
0.00518
0.0111
0.00485
       a.  After subtraction of 300 Mg S04  .

       b.  The correct, undiluted concentration was taken from Table 6 to be l4l8 + 49 Mg/20 ml  solution.

-------
ro

i
 e
 o
•H
4-1
 (0
4J
 C
 O)
 O
 c
 o
o

•o
 CU
      2.00 r-
      1.60
                  1.20
 c
 o
•H
4J
 re
 M
4J
 C
 (U
 O
 C
 O
o

•o
 01
   o>-  0.80
   P-
      0.40
         0 4
                                                    Figure  It

                           RELATIVE ACCURACY AS A FUNCTION OF SULFATE  CONCENTRATION

                                 WITH  AIHL METHOD 61 (WITH SULFATE ADDITIONS)

                                   THIRD ORDER REGRESSION FOR THE WORKING CURVE
                                                                     I
          0
                                         400
                                              800
1200
                                                                                                   1600
                                           Expected Concentration
                                                           SO^/20ml)

-------
           40
                                                Figure 5
                     COEFFICIENT OF VARIATION AS A FUNCTION OF  SULFATE  CONCENTRATION
                            WITH AIHL METHOD  6l  (WITH SULFATE ADDITIONS)
                            THIRD ORDER REGRESSION FOR THE WORKING CURVE
           30
I
ro
      6-S
       C
       o
«
•H
CO
       c
       d)
       0)
       O
           20
           10
                                     400
                                                      800
1200
                                                                                                      1600
                                            Expected Concentration  (^ig  S0^/20ml)

-------
  C.   Evaluation of a Modified BaSOu Turbidimetric Method (AIHL Method 75)


      1.   Introduction


          AIHL Method 6l specifies that 20 ml sulfate samples are mixed^ with


          a glycerol-alcohol-acid solution in a graduated cylinder, an aliquot


          is transferred to a 20 nun diameter cuvet to determine its initial


          turbidity, then poured back into the graduate and, finally, BaCl2


          is added.  After agitation and standing for kO minutes, an aliquot


          is again transferred to the cuvet and the final turbidity measured.





          In the current study, this procedure was modified to eliminate the


          multiple transfers by mixing the sample and glycerol-alcohol^-water


          and, finally, BaCl2 crystals in inexpensive 2^.k x 150 mm Teflon


          lined screw capped test tubes also used to determine both initial <
                                                                      i

          and final turbidity.  Each sample requires a separate test tube.


          A B&L 20 spectrophotometer was employed since this accepts 25.1* mm

                               *
          diameter glass cells.  The potential advantage of this technique


          includes reduced chance of contamination and/or loss of samples and


          increased speed in analysis.  This modified procedure is hereafter


          referred to as AIHL Method 75 or more simply, Method 75.  A de-


          tailed protocol is included as Appendix B.





      2.  Evaluation of Working Curves Without Addition of Sulfate


          Working curves were prepared, in three trials, covering the con-


          centration range UO-2000 yg sulfate/20 ml.  Two curve fitting


          procedures were evaluated:  a log-log fit for the range UO-300 yg/


          20 ml followed by linear regression in the range 300-2000 yg/20 ml,
*
 Test tubes are evaluated for optical matching as described in Appendix B.


                                  - 2U -

-------
    and,  secondly,  a third order polynomial fit  throughout  the  range.




    Figure 6 illustrates the results for one trial.   The  regression




    equations and the average goodness of fit by the two  procedures




    as a  function of concentration are given in  Tables 8  and 9.




    These results show that (a)  below 80 yg/20 ml,  curve  fitting  is




    poor  by third order regression and,  below 120 yg/20 ml,  relatively




    poor  by log-log fit, (b) the third order fit is  superior to the




    log-log fit + linear regression for  the range 1*0-2000 yg/20 ml




    sulfate, (c) in the range 300-2000 yg/20 ml, the third  order  and




    linear fit differ, in general, by less than  2%,  and (d)  the log-




    log (UO-260) regression line shows substantial variability from




    day-to-day in contrast to the linear (300-2000)  and third order




    (UO-2000 yg/20  ml) fit.








3.   Accuracy of the Method 75 with Added Sulfate




    Three hundred yg of sulfate  were added to each  of a set  of pure




    sulfate solutions covering the range ho to 1600  yg/20 ml.  The




    resulting solutions were analyzed by Method  75 using  linear re-




    gression for working curves  in the range 300-1600 yg/20  ml.   The




    mean  results for three trials, after subtracting the  sulfate  added,




    are compared to the true sulfate values in Table 10.








    The results indicate below 300 yg/20 ml an average error of -10.9$.




    This  may be compared to -tine  goodness of fit  to  the non-linear working




    curves below 300 yg/20 ml given in Table 9.   The mean percent error




    is substantially lower by either a log-log (average of  -0.2%) or




    third order regression (average of 2.1%). We can infer that  for
                           - 25 -

-------
I
ro
ON

I
        1.00.-
                            Figure 6  COMPARISON OF CURVE FITTING PROCEDURES
        0.80U
        o.eoh-
o»
o
C
CO

•B
o
to
,0
                                                                                     3rd Order

                                                                                     Log-Log  (0-300)

                                                                                     Linear (300-2000)
        0.40H
        0.20H
                               400
                                               800                  1200


                                           Sulfate Concentration  (jig/20ml)
1600
2000

-------
                                                  Table 8
I
ro
         True Sulfate
           yg/20 ml
             80
            120
            160
            200
            2UO

            300
            500
            700
            900
           1200
           lUOO
           1600
           1800
           2000
                               Curve Fitting as a Function of Concentration
                                 AIHL Method 75 Without Sulfate Additions
                                       Log -Log and Linear Regression
                                                                   Mean Obs.
    For SO^  < 300 yg/20 ml

Trial           a
  I
  II
  III
                 0.000327
                 0.000209
                 0.000013
















Mean Observed ;
1+2 + 6.7
75-8 + 15.3
122.2 + 5.1
152.3 + 11.7
208.0 + 5-0
253.7 +. 32,5
318.3 + 9^9
1+81+.5 + 7.3
671.3 + 15.1+
883.8 + 16.8
1217 +15-8
11+20 + 1+.3
161+0 +37.7
1793 +16.1
1961+ +23.6
r±r\ /I TT
DUlf. 0 « V .
16.0
20.2
U.2
7-7
2.1+
12.8
3.1
1.5
2.3
1.9
1.3
0.3
2.3
0.9
1.2
Coefficients for Regression
ml (SO^") = a

0.
1.
1.
b r
921 0.979
01 0.986
5!+ 0.988
^
(Absorbance) For
Sy.X
0.0617 -o.
0.0552 -o.
0.0671+ -o.
True Sulfate ;
1.05
0.95
1.02
0.95
1.01+
1.06
1.06
0.97
0.96
0.98
1.01
1.01
1.03
1.00
0.98
Equations
S0lt~ — 300
c
0893 o.
101 0.
0971+ o.
















yg/20 ml
d
000528
000528
00051+3
Mean % Error

    5.0
  - 5-3
    1.8
  - 1+.8
    l+.O
    5.7

    6.1
  - 3.1
  - l+.l
  - 1.8
    1.1+
    l.U
    2.5
  - 0.1+
  - 1.8
    0.9991
    0.999k
    0.9975
= c + d (Absorbance)

      Sy.x

     0.0136
     0.0113
     0.0227

-------
                                          Table 9
                        Curve Fitting as a Function of Concentration
                          AIHL Method 75 Without Sulfate Additions
                                   Third Order Regression
I

OO
I
   True Sulfate
     pg/20 ml

       Uo
       80
      120
      160
      200
       300
       500
       TOO
       900
     1200
     lUOO
     1600
     1800
     2000

i Observed SOU
U5.3
81.8
126.1
155.7
20k. 6
235-7
283.2
^99. 5
698.6
902.7
1205
1392
1608
1778
2016
Coefficients

C.V.
79.2
5-2
k.5
6.5
10.3
0.6
1.0
0.9
2.2
1.3
0.5
1.0
2.2
1.2
0.1
for Regression
Mean Obs . SO^
True SOU=
1.13
1.02
1.05
0.97
1.02
0.98
0.9^
1.00
1.00
1.00
1.00
0.99
1.01
0.99
1.01
Equations
                                                               Mean % Error

                                                                  13.3
                                                                   2.3
                                                                   5.1
                                                                 - 2.7
                                                                   2.3
                                                                 - 1.8

                                                                 - 5-6
                                                                 - 0.1
                                                                 - 0.2
                                                                   0.3
                                                                   O.U
                                                                 - 0.6
                                                                   0.5
                                                                 - 1.2
                                                                   0.8
Trial

  I

  II

  III
         ") = e + fx + g x2 + h x3  for Uo <_ SO^  <_ 2000 yg/ml

     e              f               g                h              r
-2.U8 x 10~3    1.78 x 10"4     3.71 x 10~7     -1.12 x 10~10    0.99998      0.00593

-2.28 x 10~3    1.83 x 10~4     3.30 x 10~7     -0.922x 10~10    0.99999      0.00538

 2.0U x 10~3    1.36 x lO"4     U.U2 x 10~7     -1.36 x 10~10    0.9998       0.00911

-------
                           Table 10

       Accuracy as a Function of Concentration for the
             AIHL Method 75 with 300 vg/20 ml
           Sulfate Addition and Linear Regression8"
True Sulfate            Mean    _b
  less 300          Observed S04~       Mean Obs. S04~    c        Mean ,
 (Mg/20 ml)          (ng/20 ml)         True Sulfate   -         
-------
analysis of sulfate below 300 yg/20 ml by AIHL Method 75, at least




with pure sulfate solutions, it is preferable to construct a non-




linear working curve fit by either a log-log or third order re-




gression technique than to add sufficient sulfate to permit use of




the linear portion of the working curve.  A more definitive test of




the use of addition of standard amounts of sulfate will be given




below.








Above 300 yg/20 ml, the sulfate addition procedure yields better




accuracy than below this concentration, but the results are still




less accurate than without added sulfate (comparing Tables 9 and 10).








The cause of the decreased accuracy (at all levels) is unclear.




The Repipet for dispensing the 300 yg S0jt~/10 ml used for the




addition of standard amounts of sulfate was calibrated gravi-




metrically by dispensing sulfate solution.  In addition, the




repeatability of dispensing this quantity of sulfate was checked




by sulfate analysis as part of several experiments.  The mean




standard deviation for repeatability was 3.6 yg/ml (C.V., 1.2%).




The apparent reproducibility of the standard addition from one




experiment to another reflects both the repeatability of the Repipet




and the uncertainty of the working curve, since each experiment em-




ploys a different working curve.  Employing linear regression, the




observed mean for two experiments (three trials per experiment) was




306.U +_ 13.9 yg/ml (C.V. of U.5#) which shows that the repeatability




of the Repipet, ca. +_ 1%, is adequate for the sulfate addition.  The




reproducibility of the sulfate addition from experiment to experiment








                        - 30 -

-------
    (C.V. = k,5%) seems to reflect primarily the uncertainty and/or




    inaccuracy of the regression equation.








U.  Working Range of the AIHL Method 75 (without Sulfate Additions)




    The results for AIHL Method 75, without addition of standard amounts




    of sulfate, are compiled in Table 11 employing third order re-




    gression of the working curve.  The calculated, undiluted con-




    centration of the extract showed no apparent "bias between 10 and




    100$ concentration.  Thus, the correct undiluted concentration was




    taken to be the mean of nine results, 139^ +. ^7 yg/20 ml.   The




    ratio of observed to- expected concentrations are plotted against




    expected concentrations in Figure^ 7.  The results indicate accuracy




    within 1% of the optimal value in the concentration range from 100




    to lUOO yg/20 ml.  The coefficient of variation of three independent




    trials is plotted against expected concentration in Figure 8.   C.V.




    values remain less than 6% in the range lUO-lUOO yg/20 ml.








    Thus, based on both relative accuracy and precision (as measured




    by coefficient of variation), the working range of the modified




    barium sulfate method, without sulfate additions, is from 1^0 to,




    at least, 1^00 yg/20 ml using third order regression.








5.  Working Range of AIHL Method 75 with Sulfate Addition




    The same sample used in Section 3 was analyzed by the AIHL Method




    75 turMdimetric procedure after addition of 300 yg sulfate to each




    sample.  Since the data analysis without sulfate additions was done




    with a third order regression fit for the working curve, the same







                             - 31 -

-------
                                                          Table 11
                                  Relative Accuracy and Precision of the AIHL Method^ 75
                       (without Sulfate  Additions) with Atmospheric Samples  (jig S04=/20 ml solution)
                                                                                             a
    Concentration,
    % of  Undiluted
U)
ro
           .5
           .0
  2,
  5.
  7.5
 10.0
 12.5
 15.0
 18.75
 31.25
 J+3.75
 56.25
 75.00
100.0
Observed
Mean of
3 Trials
55.6
84.0
111.8
141.6
177.5
197.5
248.6
426.1
623.3
787.0
1077.6
1440.4


a
17.5
15.5
14.5
1.5
9.8
11.0
7.1*
8.2
17-0
12.5
17.5
9-1


a2
305.9
24l.l
211.4
2.2
95.8
120.2
55.0
67.7
287.6
157.0
305.7
82.4


C.V. (%)
50.2
22.2
13.9
1.1
5.6
5-3
2.8
1.9
2.8
1.6
1.7
0.65
Calculated
Undiluted
Concentration

  2223.3
  1679.7
  1409.2
  1416.5
  1420.2
  1316.5
  1326.0
  1363.5
  1424.8
                                                                        1399.
                                                                        1436
l44o.4
                                            Mean undiluted cone. = 1394 + 47
                                                                                Expected
                                                                                Diluted          Observed Cone.
                                                                                Concentration?    Expected Cone.
  34.8
  69.7
                                                                                        139-^
                                                                                        209.1
                                                                                        261.3
                                                                                        435 <
                                                                                        609,
 784.0
1045.3
1393-7
                                       595
                                       205
                                       069
                                       016
                                       019
                                     0.9^5
                                     0.951
                                     0.978
                                     1.022
                                     1.004
                                     1.031
                                       1.033
                    Trial

                    I
                    II
                    III
                               Coefficients  for Third Order Working Curves

                                     ABC
                               -7.^33 x 10'
                               ^1.323
                               •3.071 x 10
 -3
 I

•>~3
 -3
                                                             -4
2.050 X 10
1.681 x 10"J
1.943 x 10"
                                                                                 -7
                -7
        3.506 X 10
        4.026 x 10
        3.33^ x 10"7
               D

         -1.110 x 10"10
         -1.366 x 10"1
         -9.603 x 10"11
           a.  The samples used were prepared by dilution of the same pooled extract used to evaluate AIHL Method 6l
           b.  (:a/mean) x 100.    <

           c.  Equals (Mean undiluted cone.) x (Concentration, % of undiluted) x .01.

-------
           1.60r-
                         Figure 7

RELATIVE ACCURACY AS A FUNCTION OF  SULFATE  CONCENTRATION

  WITH AIHL METHOD 75 (WITHOUT SULFATE ADDITIONS)

    THIRD ORDER.REGRESSION FOR THE  WORKING  CURVE
           1.20
u>
u>
         c  0.80
         0)
      .£>

      O
           0.40
                                                                                        I
                                      400
                             000
1200
1600
                                               Expected  Concentration (jigS04/20ml)

-------
                                                      Figure 8


                                    COEFFICIENT OF VARIATION OF THE AIHL METHOD 75
                                  WITH ATMOSPHERIC SAMPLES  (WITHOUT SULFATE ADDITIONS)
          60
          40
U)


I
      (3
      O
      «8
     •1-1
      M
      4/2Oml)
                                                                                                       1600

-------
    procedure was used with  sulfate  additions.   The results after




    subtracting 300 from the observed sulfate, are shown in Table 12.




    The correct, undiluted sulfate concentration for the atmospheric




    sample was taken to be 139^ Pg/20 ml (Table U).   The ratio of




    observed to expected concentrations and the coefficient of




    variation are plotted against  expected concentrations in Figures 9




    and 10, respectively.  The results indicate relative accuracy within




    10$ in the range from 170 to 1^00 yg/20 ml.   This is significantly




    poorer than that found without sulfate  additions.   C.V.  values




    remained approximately constant from lUO to lUOO yg/10 ml.








    The results indicate that the  modified method with sulfate addition




    hag a working range from lUo to lUOO yg/20 ml.  However,  within




    this range accuracy is significantly inferior to that without




    addition of sulfate.








6.  Accuracy Using EPA Audit Strips




    Glass fiber filter strips loaded with five different levels of




    sulfate, four strips per level,were extracted at room temperature




    for 30 minute by ultrasonic extraction and filtered through a glass




    frit.  Sulfate analyses by Method 75 are shown in Table 13.  The




    results indicate a fairly consistent positive bias averaging \%




    throughout the range.  The greatest error, +7%, was found at the




    lowest solution concentration, U yg/ml (80 yg/20 ml).
                            - 35 -

-------
                                                Table 12


                             RELATIVE ACCURACY AND PRECISION OF AIHL METHOD 75
                (WITH  SULFATE ADDITIONS) WITH ATMOSPHERIC SAMPLES (pg SOi4~A°0 ml SOLUTION)
Concentration,
Observed Mean
Expected
                       Observed Cone.
% of Undiluted
2.5
5-0
7-5
10.0
12.5
15.0
18.75
31.25
ii U3-75
O"\
, 56.25
75-00
100.0
of 3 Trials a
27-7
57-1
87.0
118.9
156.8
188.9
238.1*
1*29-3
620.1*

823-1*
1086.9
11*31-7
a
12.1
6.3
6.8
2.9
9-8
6.8
8.0
8.1*
29.6

17-5
1*5.3
2U.2
a2
11*7.3
39.6
1*5.8
8.2
96.6
U6.1
63.1*
71.1
877.1*

306.2
2052
586.3
C.V.(JJ)
1*3.7
11.0
7-8
2.5
6.3
3.6
3.1*
2.0
1*.8

2.1
1*.2
1.7
Diluted Cone .
31*. 8
69-7
10l*.5
139. ^
171*. 2
209.1
261.3
1*35-5
609.7

781*. 0
10U5
1391*
Expected Cone .
0.79^
0.820
0.832
0.853
0.900
0.901*
0.912
0.986
1.02

1.05
l.OU
1.03
                              Coefficients for Third Order Working Curves
                                    A
                                    B
     c
                              -U.150 x 10
                              -6.060 x 10'
                     -2
                     i
                     -2

                     -2
      x 10 '
4.118 x 10
5.171 x 10"4
    r7
    -7
1.909 X 10
1.179 x 10'
0.521 x 10"7
                 D
       x 10
-3.823 x 10'
       x 10"
          Trial

          I

          IIIC                -7.88U x 10

a.  After subtraction of 300  jug S04 .

b.  The correct, undiluted concentration was taken from Table h  to be 139^- _+ ^7 Mg/20 ml  solution.

c.  Employed a new bottle of  BaCl2 (sample D as described in Section D).
-11
-11
-11
 Sy.x

0.0091^
0.00712
0.00791

-------
     1.60
                                                         Figure 9


                                 RELATIVE  ACCURACY AS A FUNCTION OF SULFATE CONCENTRATION


                                     WITH AIHL METHOD 75 (WITH SULFATE ADDITIONS')

                                     THIRD ORDER REGRESSION FOR THE WORKING CURVE
          1.20
I

u>
e

-------
                                                   Figure 10


                                COEFFICIENT  OF  VARIATION OF AIHL METHOD 75
                             WITH ATMOSPHERIC SAMPLES  (WITH SULFATE ADDITIONS)
     60
 c
 o
 to
     40
Oo

I
4J
c
01
vi
o
•r-l
M-l
U-l
(U
O
O
         20
                                I
                                                                                                               I
                               400                      800



                                      Expected Concentration (pg SO4/?Oml)
                                                                                    1200
                                                                                                             1600

-------
                              Table 13
     ANALYSIS OF EPA SULFATE AUDIT STRIPS  BY THE AIHL METHOD 75
Approximate
Sample
1000 Series
9000 Series
8000 Series
5000 Series
, 7000 Series
VO
' T?AOT»PHST rm
Concentration
Analyzed
(yg/ml)
u
15
35
53
85

Tilrma +. i nn • v s _1
Theoretical
Value
(yg/strip)
200.2
7^5.6
1737.2
2670.1
U2UO . 2

o£ v- in~2 4. o Tt
Observed
Value
(yg/strip)
215 ± 16
7U8
1832
2812
UU03

•> v in-lt Y j-
+ 39
± 11
+ uo
+ 117

O 7li v
c.v. (JO
7.U
5-2
6.0
l.U
2.7

nn~7 v2 _ n T A -u-
Observed
Theoretical
1.07
1.00
1.05
1.05
l.OU

in-11 v3
                              iX = 1.U5  x 10"2
                               r = 0.9998

-------
    7.   Conclusions




        Using AIHL Method 75  (i.e.  the Method 6l version of the barium




        sulfate turbidimetric procedure modified by carrying out the




        reaction and reading  turbidity values in capped test tubes)  with




        third order  regression of the working curve provides a working




        range from lUo to at  least  ihOO yg sulfate/20 ml.   Method 75 is




        not improved by the addition of standard quantities of sulfate




        to each sample.   EPA  audit  strips  indicate  an accuracy within 6%




        for samples  in the working  range.








D.  Comparison of Results with AIHL Methods 6l and  75




    A summary of results with Methods 6l and 75 is  given in Table lU.




    These results indicate that:




    —Method 75 has  a slightly narrower working range than Method 6l.








    —Method 75 is more accurate  with standards in  the range 80-2UO  yg/20 ml




      relative to Method 6l.   In  the 300-2000 yg/20 ml range, the accuracy




      is about the same.








    —For Method 75  with atmospheric samples, relative accuracy is about




      the same as with Method 6l  averaged  over the  working range. In the




      range < 300 yg/20 ml, Method  75 is ca. 3% more accurate and 2.h%



      more precise than Method 6l.

-------
                                                     Table  ik

                                   SUMMARY  OF RESULTS OP  TUEBIDIMETRIC METHODS6
AIHL Method 6l
AIHL Method 75
AIHL Method 6l (with
added sulfate)
AIHL Method 75
(with added sulfate)
i
H AIHL Method 6l
i
AIHL Method 75
AIHL Method 6l (with
added sulfate)
AIHL Method 75
(with added sulfate)
Mean % Error
with Standards13 Working Mean % Mean Mean % Error Mean C.V.
80-21*0 300-2000 Range Error c C.V.W) < 300 < 300 (%}
6.9
2.6
1.2
9.3
Not
feasible
Not
feasible
3.9
13.8
Comparisons with Third Order Regression
1.8 106-lUoo 3.3 3.2 - 3.6 5-1
1.0 liiO-1400 2.2 2.6 - 1.7 3.7
1.2 140-1400 k.3 2.3 - 6.7 2.7
2.3 170-lUOO 5.5 3.5 - 9-5 4.4
Comparisons with Linear Regression
2.9
2.3
1.5
4.4
a.  Except as noted, all results were_obtained with a pooled atmospheric extract diluted to varying degrees.
    Concentrations in units of yg 30^ /20 ml.
b.  Results calculated from absolute values of errors to avoid cancellation effect.  Error without added sulfate
    calculated from goodness of fit to working curve.
c.  Error relative to that in optimal range of method.  Results calculated with absolute values of errors to avoid
    cancellation effect.

-------
   —The addition of  fixed  amounts  of  sulfate  to Method 75 increases




     error and  decreases precision  and working range.








   —The addition of  fixed  amounts  of  sulfate  to Method 6l decreases the




     error with standards < 300  yg/20  ml "but decreases  the working range




     and increases  the  error with atmospheric  samples < 300 yg/20 ml.




     The precision  with atmospheric samples is generally better with




     added  sulfate.








   Conclusion:   Considering the  results in Table 1^  and its greater




    simplicity and speed, the modified  turbidimetric  method (Method 75)




   with third order regression and  without addition  of  fixed amounts of




    sulfate  is preferable for analysis  of atmospheric samples.








E. Ruggedness Test  for  Turbidimetric Sulfate Analysis




    To  obtain additional information on the influence of several variables




    on  turbidimetric procedures for  sulfate, a  ruggedness test was carried




    out utilizing the  procedures  followed in both AIHL Methods 6l and 75.




    The protocol followed is included as Appendix C.








    The factors evaluated are given  in  Table 15.  The results of each of




    the twelve runs, calculated by third order  regression for the working




    curve,  were expressed as the  ratio  of the observed to the theoretical




    sulfate  level.   The  effect  of each  factor was evaluated as the difference




   between  mean results for the  runs with high (or plus) and low (or minus)




    levels.   Table  l6  ranks  the observed effects, squares the effects to

-------
                                     Table 15

                     FACTORS FOR EVALUATION IK RUGGEDNESS TEST
A = Level of

B = Level of Addition of S04=

C = Age of Reagent Mixture

D = Varying Strength of Reagent
    Mixture

E = Time After Adding Reagent
    Mixture Until Mixing

F = Mixing before Absorbance
    Reading lc'd

G = Addition of BaCl2

H = Mode of Shaking, after
    Addition of BaCl2

I = Timing before Second
    Absorbance Readinga

J = Spectrophotometer, cell length
    and mode of mixing'3

K = Dummy
  Low (-)

100 pg/10 ml

300 yg/10 ml

New (0-1 mo.)

(k ml + 1 ml of
95$ ethanol)

1 min.


Gentle Mixing6


\ scoop (ca. 0.125 g)

Gently for ^5 sec.
(l shake/second)

20 min.
                                         B & L 20, 25-^ mm
                                         separate cells
                                                                     High (+)

                                                                   1000 pg/10 ml

                                                                    600 yg/10 ml

                                                                   Old (2 years)

                                                                   5 ml


                                                                   10 min.
                                                                   Vigorously for
                                                                   20 sec.

                                                                   1 scoop  (ca.  0.25  g)

                                                                   Vigorously for
                                                                   U5 sec.

                                                                   kO min.
                          B & L TO,  20 mm
                          graduated  cylinder
a.  Following

b.  Sample and reagents mixed either in separate cells used for turbidity readings
    or in graduated cylinders with tranfers to a single 20 mm cell for reading
    turbidity.

c.  The absorbance due to initial turbidity of sample plus the glycerin-alcohol-
    HC1 "mixed" reagent before addition of BaCl2.

d.  After transfer of solution to cuvet, entrapped bubbles can cause appreciable
    error in absorbance.

e.  Slowly add glycerol-alcohol-acid solution minimizing air entrainment.  Mix
    solution by inverting graduated cylinder slowly (about 2 seconds to invert).
    After 10 seconds, invert again.  Repeat for total 6 inversions.

-------
                                                Table l6

                    RESULTS OF RUGGEDNESS TEST OF SULFATE ANALYSIS  BY  TURBIDIMETRY
Factor    Abbreviated Identification
(Effect)2 as  % of Total
c
G
A
D
E
I
H
i
£ K
1 F
J
B
Age of reagent
Amount of BaCl2 added
Sulfate level
Reagent strength
Wait-time until mixing
Wait-time until A2
Shaking mode (after BaCl2)

Dummy
Shaking mode (before BaCl2)
Spectrophotometer , cells
Sulfate addition
-o . in 7
-0.127
-0.117
-0.087
0.073
-0.033
-0.030

0.030
0.027
0.020
0.013
o.m
0.016
0.0136
0.0075
0.005^
0.0011
0.0009

0.0009
0.0007
0. 000*1
0.0002
78. 8a
7-3
6.2
3.1*
2.U
0.5
O.U

o.U
0.3
0.2
0.1
  a.   By the  procedure  of R. A. Stove and R. P. Mayer, Ind. and Eng. Chem.  5j3  (2)  36  (1966),  only
      the variance  for  the age of the reagent exceeds that of the dummy factor at  the 95$  confidence
      level.

-------
    estimate the variance of the method due to that effect and determine
    the proportion of the total variance due to each factor.


    The results indicate that the age of the prefixed reagent (i.e. the
    solution containing glycerin, water .and HCl) is the dominant source of
    variance in the method.  Clearly the shelf life of the pre-mixed reagent
    is less than two years.  Four other factors each contributed > 1% of
    the total variance, including amount of ZaCl2 used, sulfate concen-
    tration, reagent concentration and the time following addition of
    reagent until the solution is mixed.  The use of separate cells
    (i.e. test tubes) for each sample and a different spectrophotometer
    and cell thickness had a negligible effect onlfae variance.   These
    changes have been used in the modified BaSOif method (Method 75) as
    described in Section III B in an effort to make the procedure simpler
    and faster.


    Further studies are needed on the effect; of reagent age to establish
    the shelf life of the reagent.


F.  Influence of Bad2 Mesh and Crystal Size and Impurities on Sulfate
    Determination by Tni»bidimetry
    1.  Introduct ion
                                                                          13
        Literature on BaSOit turbidimetric methods dating to the mid-1930's
                                    3 lU
        as well as published methods '   have generally recommended use
        of 20-30 mesh BaCl2.   This was based on still earlier unpublished
        studies by Parr.    The latter developed an instrument, referred
        to as the "Parr turbidimeter", for use in determining sulfate in

-------
    vater  using barium chloride.   While manufacture of this turbidimeter




    was discontinued about 10 years ago, the only U.S. supplier of




    BaCl2  crystals  for turbidimetric sulfate analysis, J. T. Baker,




    continued to specify "For Parr Turbidimeter".  Recently, Baker




    has discontinued supplying this grade of BaCl2 .   Accordingly, an




    evaluation of alternate grades of BaCl2 was  performed.








2.   Effects on the Working Curves with Method 75




    The use of four different sources of BaCl2 (as BaCl2'2H20)  was




    observed to alter significantly the working curves obtained,  in




    some cases (Figure 11).  The  four samples are described in Table 17.




    The studies reported above employed samples A and D.   The initial




    hypothesis adopted to rationalize differences was based on particle




    size distributions.  While sample A had been  exhausted, it was




    labelled "20-30 mesh".  The mesh size distribution of the remaining




    samples was determined.








    The results are given in Table 18 together with the mean particle




    size calculated for each range.  Comparison of Table  18 and Figure 11




    shows a large difference between working curves for samples B and D




    consistent with a large difference in mesh size.   However samples




    B and C yielded similar working curves in spite of a comparably




    large difference in mesh size.








    Where sufficient BaCl2 was obtained, the individual size fractions




    were used to prepare working  curves (single trial each) with seven




    data points between 300 and 2000 yg/20 ml.  The results, using

-------
     1.00 ,-
     0.80
     0.60
  O
  CO
  ,0
1  <"
I  bQ
  c
     0.40
     0.20
                               Figure 3:1
COMPARISON OF WORKING CURVES WITH VARYING BaCl2 SAMPLES
   USING AIHL METHOD 75
                                                  BaCl	SAMPLE

                                                      A(2)      y
   REGRESSION EQUATION

  -0.092 + 0.000526X
    r  - 0.9995   Syx « 0.0108
                                                                                           -0.081 + 0.000509X
                                                                                             r = 0.9997  Sy.x - 0.
                                                                    -0.115 + 0.000430X
                                                                      r - 0.9955  Sy-x - 0.0274

                                                                    -0.091 + 0.000430X
                                                                      r = 0.9992  Sy-x = 0.0118

                                                                    -0.0824 + 0.000550X
                                                                      r " 0.9998  Sy-x = 0.00796
                         400
                  800             1200

                          Standard Concentration
2000
2400

-------
                                       Table  IT

                DESCRIPTION  OF  BaCl2  USED FOR TURBIDIMETRIC METHOD
                 J.  T.  Baker  Specified
                     Crystal  Size
Sample
Designation
A
B
C
D
(Old U.S. Series
Mesh Size)
20-30
None
None
None
Stated Purpose
For Parr
Turbidimeter
None
None
For Parr
Lot No.
28T93
32561
353T1
528858
Catalog No.
09 Ik
09 TO
09TO
1-09 T^
                                           Turbidimeter
a.  The specifications  for this  product were  quoted as  <  5$  retained on 20 mesh,
    < 10$ smaller than  TO mesh (i.e.,  85$ between 20 and  TO  mesh).

-------
                                Table  18

        Size Distribution of J. T. Baker BaCl2-2H20 (% by weight)0


Old U.S. Series    Mean Particle Size,
Mesh Size
< 20
20-30
30-^0
Itf)- 50
50-70
> 70
Mm
1015a
725
512
363
256
125b
Sample B
50.2
13.3
11.9
7 = 9
5-7
10.9
Sample C
O.k
3.^
28.7
IK). 2
16.6
10.8
Sample D
2.5
2.6
5-9
37.7
1(2.6
8.6
a.  All particles assumed to be between 16 and 20 mesh (1180 to 850 ium).

b.  All particles assumed to be between kOO and 70 mesh (38  to 212 jum).

c.  Sieving was done in a particle free hood to minimize contamination
    using pre-cleaned stainless steel sieves.
                                     -1*9-

-------
linear regression, are given in Table 19.  The results suggest




that for a given batch the slope increases with increasing BaCl2




particle size.  However, sample D shows substantially higher slopes




for a given mesh size compared to B and C.








The relation between particle size and slope of the turbidimetric




sulfate working curve is shown in Figure 12.  If it is assumed




that the slope for the working curve observed for the unfractionated




samples of BaCl2 is a linear combination of those resulting from




the individual size fractions, the slope for the "total sample"




can be calculated as:




              Slope for unfractionated BaCl2 = £ (px).




where         p = weight percent for fraction i




              x = turbidimetric slope for fraction i








Values for x were experimental or were obtained from the regression




lines for each sample in Figure 12.  A comparison of calculated




and observed slopes for the unfractionated BaCl2-2H20 is given in




Table 20.  The results indicate agreement within 5%.








These results provide some understanding of factors determining




the working curve for a given sample of BaCl2, and help to explain




why Batches B and C, which differ so greatly in particle size




distribution, can still yield similar slopes.  These results left




unanswered, however, the source of differences in behavior observed




for the same particle sizes from different batches.








                        - 50 -

-------
Old U.S. Series
Mesh Size
< 20
20-30
30-Ho
Ho- 50
50-70
1 > 70
a
-.OH67
-.0876
-.0881
-.0909
-.0779
-.'0722
Sample
b
.000*169 «
.000*K31 .
.000*1-07 .
.oooHoi .
.000368 .
.000388 .
B
r
999^
9986
9978
9973
9971
9982
.01121
.01510
.01810
.01982
.01891
.01580
a
___**
-.08*1-
-.091
— . 090
-.089
-.01*6
                                                Table 19

                       Working Curves for Method  75  TurMdimetrie Method with
                          Size Fractionated BaCl2-2H20 (Linear Regression)*
 Sample C
b       r    s\
                                             .08*0.  .000502  .9997  .00877

                                             .0911  .0001)67  .9991  .01332

                                                    ,000*|.*l"8  .9993  .0112U

                                             .0891  .000*123  .9989  .01338

                                             .0*^69  .000383  .9971  .01965
                                                                                          Sample  D
                        .0655   .000525  .9990  .015*).*).

                        0575   .000508  .9991  .01*t60
*   For ease in data handling,  standards  were restricted to 21 300 yg/20 ml permitting linear regression.

**  Dashes indicate insufficient sample for  construction of working curve.

-------
                                                          Figure  12



                         THE EFFECT OF BaCl2'2H2©  PARTICLE SIZE  ON THE  SLOPE FOR THE WORKING CURVE



                                                      AIHL METHOD 75
I


ro
        0.0006
     -a
     o
s   0.0005


cf
O)
to
PQ



"S   0.0004
•H
•o

I


<„"  0.0003


t-i
51
O
     s
     8.
     o
    i-H
    co
        0.0002
        0.0001 -
        0.0000
                                                      Slope = a + b (average particle  size)
                                                   Sample
                                                          a
                                               (B)    0.000360




                                               (5)    0.000369



                                               fD)    0.000467
                                                                            -7

                                                                   1.024 x  10


                                                                            -7

                                                                   1.903 x  10


                                                                            -7

                                                                   1.596 x  10
0.9473



0.9845
                                                                                                               B
                                                                                                                      I
                               200
                                           400              600


                                               Average  Particle Size
                                                                             800
            1000
1200

-------
                           Table 20

        Comparison of Observed Working Curve Slopes for
AIHL Method 75 with those Calculated from Individual Fractions
Sample     Calculated       Observed      Calculated/Observed

  B         0.000^36        0.000^30            1.01

  C         O.OOOUU5        0.000^30            1.03

  D         0.000520        0.000550            0.95
                               -  53  -

-------
G.  Impurity Levels in BaCla'SI^O Samples
    Batches B, C and D were analyzed for Ca, by flame atomic absorption




    and for Sr and Fb using a heated graphite analyzer.  No effort was




    made to overcome possible interelement interference in these analyses.




    The results in Table 21 ,which should be considered approximate, indicate




    no substantial differences between the samples which might account for




    the observed differences in the working curves.








 H.  Dissolving Times  for BaGl2 Samples in Water




    In an  effort to determine the source of the difference in behavior




    between different batches of barium chloride a simple experiment was




    performed to compare rates of solution in water.  For this purpose




    one scoop (ca. 0.25 g) of barium chloride of ^0-50 mesh size was




    dropped into 25 ml of distilled water at room temperature in a large




    test tube (l inch diameter) being stirred at a standard rate with a




     small  magnetic stirring bar.  The time required for dissolving of all




     crystals  was measured with a stopwatch.  The results are shown in




     Table  22  and indicate essentially identical dissolving times for




     batches B and C while batch D dissolved in approximately half this




     time.  Thus differing rates of solution between D and the other two




     batches might be related to differences in performance in turbidimetric



     sulfate analysis.








 I.   Optical Microscopy of BaCl2'2H20 Particles




    A comparison of the UO-50 mesh fractions was made by optical microscopy




     scanning  over a large number of particles.  Photomicrographs considered




    representative of the samples viewed are shown in Figure 13.  The pictures

-------
                        Table  21



Analysis of BaCl2'2H20 for Trace Metals  (ppm by weight)
Batch
B
C
D
Ca*
2 + 0.5
1+1
1 + 1
Sr
0.1
0.05
0.05
Pb
0.8
1
1.2
   All results  close to detection limit.
                            -  55 -

-------
                        Table 22

Comparison of Dissolving Times for Three Samples of BaCl2


                                                   £*
     Batch               Time to Dissolve (seconds)

B, UO-50 mesh                    35.0

C, UO-50 mesh                35-3 ± O.U

D, UO-50 mesh                l6.5 + 2.5
a.  Mean of two trials +_ 1 a employing 25 ml distilled 1^0 at
    room temperature, constant speed magnetic stirring and
    1 scoop (ca. 0.25 g) BaCl2-
                           - 56 -

-------
BaCl2-2H20 Sample B
    UO-50 mesh
                   BaCl2-2H20 Sample C
                       UO-50 mesh

              Grid Size = 60 /^m x 60 Mm
Micrographs of Three Samples of UO-50 Mesh Barium Chloride
BaCl2-2H20 Sample D
    UO-50 mesh
                                             Figure 13

-------
    suggest that for batches B and C the mesh size and the average crystal




    size are nearly equal.  In contrast to these, batch D exhibited a larg




    number of aggregated smaller crystals.  The faster rate of  solution of




    batch D is consistent with its smaller average crystal size.








J.  Conclusions



    The ideal BaCl2-2H20 batch is one which yields high slopes, small




    intercepts and little scatter for the working curve.   Well  size-



    fractionated particle sizes are not required to achieve good results




    with the turbidimetric method.  For a given batch  of BaCl2, its



    behavior in sulfate analysis is predictable from the  properties of



    its particle size fractions (i.e. crystal aggregate  size as obtained



    by  sieving).  Working curve slopes increase with increasing crystal



    aggregate size  (or decreasing Old U.S. Series Mesh Size).  However,



    the mesh size is a poor indicator of individual crystal size which,



    in  turn controls  dissolution rate.  For a given mesh size, smaller



    crystals, which dissolve faster, yield higher working curve slopes.



    Thus  a coarse mesh size barium chloride (e.g. 20-30 mesh) which



    consists of aggregates of small crystals appears to be  ideal.

-------
IV.  EVALUATION OF AN ALTERNATIVE TURBIDIMETRIC TECHNIQUE USING SULFAVERR

     A.  Introduction
                                             •o   #
         An evaluation of the use of SulfaVer  IV  for sulfate analysis of

         hi-vol extracts was conducted.  This material consists of BaCl2

         blended with a proprietary colloid-stabilizing material and en-

         capsulated in a plastic pillow-shaped.container.  One pillow contains
                                                                  •D
         the amount required for a single determination.  SulfaVer  is also

         available in bulk.  This evaluation has increased significance because

         of the current unavailability of crystalline barium chloride graded

         for use in turbidimetric analyses.
         The  evaluation  included development of an experimental protocol to

         optimize performance  of the method, an evaluation of the accuracy,

         precision  and working range of the method and lastly, an intermethod

         comparison.   The  latter will be reported in Section IX.




     B.  Development  of  an Experimental Protocol
                                                                   T>
         The  method requires addition of the contents of a SulfaVer  pillow to

         an aqueous sulfate sample, agitation by some means, allowing the

         solution to  stand for a specified period and reading of the light

         extinction due  to turbidity at 500 nm.  Conditions studied are

         summarized in Table 23.   The influence of these conditions on the

         resulting  working curve in the range 100 to 1600 yg/20 ml was

         evaluated  using both  linear and third order regression curve fitting

         procedures.
  * Hach  Chemical  Company



                                      -  59  -

-------
                               Table 23

Conditions Studied in Developing Protocol for SulfaVer%V Procedure
Agitation (in 25 x 150 mm Teflon-lined             Code
           screw cap test tubes)

Machine for 60 sec                                  Sj
Hand (vigorous) 55 sec                              82
Machine Shaking Speed (oscillations/minute)

277                                                 B!
183                                                 s2
 90                                                 s3
Standing Time Before Reading (min)

 5                                                  wj
10                                                  w2
20                                                  w3
30                                                  w,,
Final Volume (ml)

20                                                  vx
25        •                                          v2
 Spe ct rophot bmet er

 B and L 20 (read in 2.5^ cm test tubes)             Bj
 B and L 70 (read in 2 cm cuvet)                     B2
                                 - 60 -

-------
Regression curve parameters for each trial are given in Tables 2k and
2_5 for linear and third order regression, respectively.  Trials 1-5
employed a. B and L Model 20 spectrophotometer.  Since this unit
accepts one inch diameter test tubes, turbidity readings were obtained
directly in the screw cap test tubes used for mixing the sample and
reagent.  Trials 6-11 followed a protocol analogous to that in AIHL
Method 6l:  initial turbidity readings were obtained before addition
of SulfaVerR.  After addition of SulfaVerR followed by mechanical
shaking in test tubes the samples were transferred to a 2 cm cuvet for
reading in a B and L Model 70.


Comparing linear and third order regression, the latter gives a
better fit in every case over the range 0 to l600 yg/20 ml (for linear
regression, zero point excluded).  From trials 1-5 the minimum in
S     (third order) for trial k was used to select mechanical shaking
 y »x
with  90 oscillations per minute over hand shaking or mechanical shaking
at higher speeds.  Trials 6-11 employed these conditions but varied
waiting time.  Trials 7 and 8 were replicates and demonstrated a
reproducibility within +_ 10% of the working curve.  Based on the
minimum in S    (third order) in Trial 9» a 20 minute wait was chosen.
            yx
It is clear, however, that the difference in S    using 5, 10 and 20
                                              y*x
minutes waiting times, is relatively small.  A protocol based on this
evaluation is included as Appendix D.


The use of the B and L 70 with transfer of samples to a cuvet was
chosen to minimize differences from the AIHL Method 6l turbidimetric
method.  Reading of turbidity directly in the test tubes used for

                            - 61 -

-------
ON
ro
                                                         Table 2U


                           Influence of Protocol Variables on the Working Curve Using SulfaVer IV

                                                (Third Order Regression)
Trial
1
2
3
1*
5
6
7
8
9
10
11
Conditions a
Si,
Si,
Si,
Si,
S2,
Si,
Si,
Si,
Si,
Si,
Si,
si, wls
S2, Wj,
s2, wx,
83, wx,
W2, vi,
s3, wl9
S3, »2»
83, W2,
83, w3,
83, W4,
33, w5,
vi, BI
vi, B!
v2, B!
vi, B!
BI
vi, B2
vi, B2
vi, B2
vi, B2
vi, B2
vi, B2
a0 a! a2 a3
2.17
-5.36
0.1*0
-3.5l*
7.91
-8.16
-2.1*8
-3.08
-2.50
-13.5
-1.20
x 10" 3
x 10" 3
x 10" 3
x 10~3
x 10" 3
x 10~3
x 10" 3
x 10" 3
x 10" 3
x 10" 3
x 10" 3
1.86 x lO"4
3.68 x 10~4
0.55 x 10"^
2.08 x 10"1*
0.13 x ID"4
2.93 x 10~4
1.92 x 10"1*
1.85 x 10"^
2.37 x 10"^
3.72 x 10"1*
2.16 x I0~k
9.55 x 10~7
6.1*7 x 10~7
9.59 x 10~7
8.10 x 10~7
1.05 x 10~7
6.08 x 10~7
8.88 x 10~7
8.92 x 10~7
7.36 x 10~7
U.38 x 10"7
7.80 x 10~7
-3.81 x
-2.55 x
-U.28 x
-3.06 x
-3.87 x
-1.65 x
-2.81 x
-2.82 x
-2.23 x
-0-62 x
-2.1*3 x
10" 10
10"10
10"10
10"10
10"10
10"10
10"10
10'10
10~10
10"1Q
10'10
r
0.9999
0.9999
0.9995
0.9999
0.9999
0.9998
0.9999
0.9999
0.9999
0.9999
0.9999
s
y-x
rt
0.96 x 10"
— o
1.10 x 10 z
1.86 x 10~2
•M O
0.90 x 10
1.1*2 x 10~2
2.13 x 10~2
1.06 x 10"2
_o
0.99 x 10
^9
0.72 x 10 *
2.03 x 10~2
1.19 x 10"2
           a.   See Table 23

-------
                                              Table 25
               Influence of Protocol Variables on the Working Curve Using SulfaVerxV
                                         (Linear Regression)
       Trial           Conditions8"
         1         Si,  si, vi, vls BI
         2         Si,  s2, vi, v1} BI
         3         Sls  s2, vls v2, Bl
         U         Sls  s3, wl5 vi, BI
         5         S2,  v2, vls B!
         6         Si,  s3, WJL, v1} B2
         7         Si,  s3, w2, vi, B2
         8         Si,  33, w2, v1} B2
         9         SL  s3, v3t VL B2
        10         SL  83, vi,, vi, B2
        11         Sls  s3, w5, vj, B2
a
-0.83 x 10"1
-0.69 x 10"1
-0-81 x 10~l
-0-93 x 10"1
-1.12 x 10"1
-1.3U x 10""1
-1.U6 x 10" 1
-1.U9 x 10"1
-1.30 x 10"1
-1.55 x 10~!
-1.30 x 10"1
b
8.08 x 10"1*
8.02 x 10~k
6.U5 x W~k
1.82 x I0~k
1.15 x 10"1*
8.93 x I0~k
9.6k x 10"^
9.60 x 10"1*
8.98 x 10~4
9.53 x 10~4
9.00 x 10"^
r
0.997
0.999
0.996
0.998
0.995
0.996
0.997
0.996
0.996
0.993
0.996
sy.x
3-73 x 10~2
2.25 x 10~2
2.80 x 10~2
2.8k x 10~2
U.U3 x 10~2
U.85 x 10~2
U.85 x 10~2
5.26 x 10~2
U.77 x 10~2
6.9^ x 10~2
It. 86 x 10"2
a.  See Table 23

-------
   mixing is a  simpler and  faster procedure  since  transfers  are avoided;


   this was set aside for examination  of precision and accuracy at a


   later time with a spectrophotometer which can accept a 25 mm test


   tube and provide stability  and precision  at  least  equivalent to the


   B  and L TO.




   Curve fitting as a function of concentration for Trial 9  is  shown in

   Table 26.  At 100 yg/20  ml  curve  fitting  is  relatively poor  by both

   linear and third order regression.  In  the range 200 to 1600 yg/20 ml

   third order  regression yields residuals below 2% in all cases.




   The reproducibility of the  working curve in routine use is illustrated

   in Figure lU and Table 27.   In four trials,  using  the protocol given


   in Appendix  D, the greatest variability is observed at the highest

   concentration.




C. Accuracy with EPA Sulfate Audit Strips

   Glass fiber  filter strips loaded  with five different levels  of sulfate,

   four strips  per level, were extracted by  ultrasonic agitation in water

    for 20 minutes, followed by filtration  through  a glass frit.  Sulfate
                            •p
   analyses by  the SulfaVer method  are shown in Table 28.  The results

   indicate a positive error of 0 to 10$ in  the range 15 to  85  yg/ml


    (300 to  1700 yg/20 ml).   The coefficient  of  variation remained at 6%

   or less  for  all samples.

-------
                            Table 26
     Curve Fitting as a Function of Concentration Using a
                     SulfaVerRlV Procedure
True Sulfate         Linear Regression       3rd Order Regression
  yg/20 ml             Mean % Error              Mean % Error
100
200
Uoo
600
800
1200
1600
69.72
11. l^r
- 9-38
-10.16
-U.98
0.08
2.76
-16. W
- 1.92
- 0.06
- O.Ul
1.18
-0.52
0.12
_ i7_u m_j „ i n mni^i n €>£, \t,*r.— tf ~~-..,^« — ~ v IAD


-------
                                  REPRODUCIBILITY OF WORKING  CURVE  FOR SULFAVERR METHOD
a\
a\
            0)
            u
            c
o
CO
Trial I


Trial II


Trial III
                                                                       1200
                                                                            1600
                           2000
                                                Sulfate Concentration  Oig/20ml)

                                                           Figure 1^

-------
                                                         Table  27
                                 Reproducibility of Working Curve for SulfaVer  Method
                                                (Third Order Regression)
Trial a
1 -1.6l x 10"2
2 -0.31 x 10~2
3 -2.33 x 10~2
k -3.19 x 10~2
*
2.84 x 10"*
1.72 x 10"4
3.56 x 10-^
U. 21 x lO-1*

7.00
8.UU
5. 71
U.85
c
x 10- 7
x 10" 7
x 10- 7
x 10~7
f      *Regression equation:  y = a + bx + ex2 + dx3
                                                                                               y.x
                                                                            -2.10 x 10-10     O.OlUs        0.9999

                                                                            -2.65 x 1Q-10     0.0035        0.9999

                                                                            -1.U3 x 10~10     0.0155        0.9999

                                                                            -1.19 x lO'10     0.0212        0.9999

-------
                                               Table  28

                                                                       R        Q,
                     Analysis of EPA Sulfate Audit  Strips  by a SulfaVer  Method
   Sample

1000 Series

9000 Series

8000 Series

5000 Series

7000 Series
 Approximate
Concentration
  Analyzed
   (pg/ml)

      1*

     15

     35

     53

     85
Theoretical
Value
200.2
71*5-6
1737.2
2670.1
1*21*0.2
Observed
Value b
260 +.
7U3 +
1905 +
2879 ±
1*1*39 +
16
1*3
27
52
2l*3
C.V.
6.1
5.8
1.1*
1.8
5.5
Observed
Theoretical
1.30
1.00
1.10
1.08
1.05
a.  Results are mean values +_ 1 o for four strips from each series extracted by 30 minutes ultrasonic  extraction.

b.  Using third order regression data analysis.

-------
D.  Working Range

    The atmospheric  sample extract previously used  for  evaluating

    Methods 6l and 75 was diluted to varying  degrees to  permit  evaluation

    of relative accuracy and precision as  a  function  of concentration.

    As before, the working range of the sulfate method  was considered to be

    the concentration range which showed 'both approximately constant

    relative  accuracy and coefficient  of variation.   The results for the
            •p
    SulfaVer  method are compiled in Table 29, employing third order

    regression of the working  curve.   The  calculated, undiluted concentration

    of the  extract showed no apparent  bias between  10 and 100$ concentration.

    Thus , the correct undiluted concentration was taken to be  the mean of

    nine results, 1^35  +. Vf yg/20 ml.   This  compares  to the values 139^ +. ^7

    and lUl8  +_ U9 previously found with the  modified  BaSOi* method and Method

    6l, respectively.   The ratio of observed to expected concentrations are

    plotted against  expected concentration in Figure  15.  The  results

    indicate  accuracy within h% of the optimal value  in the concentration

    range from 11*3 to 1^35 yg/20 ml.   The  coefficient of variation of

    three independent trials is plotted against expected concentration in

    Figure  16.  The  C.V. values remain less  than 6% in  the range 180 to

         Pg/20 no-
     Based on both relative accuracy and precision as measured by coefficient

                                                    T)
     of variation, the working range of the SulfaVer  method is from about


     180 to >_ ikOO yg/20 ml using third order regression.
                                - 69 -

-------
                                           Table 29

             Relative  Accuracy and Precision of SulfaVerR Turbidimetric Method
                        with an Atmospheric  Sample (pg SO^2  /20 ml)

Concentration ,
% Undiluted
2.5
5-0
7-5
10.0
12.5
15.0
18.75
31.25
U3.75
56.25
75-0
, 100.0
o
i
Observed
Mean
3 Trials
51.8
93.8
117.3
1U2.9
171.8
201*. 9
258.9
1*56.5
61*7.0
830.6
1101.6
ll*66.9




a
21.8
9.6
8.8
12.7
U-5
10.6
15-1
19.5
26.1*
10.7
19-7
1.1*




a2
1*75.0
91.3
78.3
161.2
20.2
113.1
229.3
380.0
697-8
113.8
387.1
1.9
Mean



C.V. 3
1*2.0
10.2
7.5
8.9
2.6
5.2
5-8
U . 3
|| j_
1.3
1.8
0.1
undiluted

Calculated
Undiluted
I Concentration
2071.8
1875.6
1561*. 5
ll*29 . i*
137^.3
1366.0
1380.6
11*60.9
ll*78.9
ll*76.7
11*68.8
11*66.9
cone. = ll*35 ±. ^7

                                                                       Expected
                                                                       Diluted
                                                                    Concentration

                                                                          35.8
                                                                          71. 7
                                                                         107.5
                                                                         179.2
                                                                         215-0
                                                                         268.8
                                                                         1*1*8.0
                                                                         627.2
                                                                         806. U
                                                                        1075.2
                                                                        11*33.6
                                                                    Observed Cone.
                                                                    Expected Cone.
                                                                        1.1*1*5
                                                                        1.308
                                                                        1.091
                                                                        0.997
                                                                        0.959
                                                                        0.953
                                                                        0.963
                                                                          019
                                                                          032
                                                                          030
                                                                          025
                                                                        1.023
Trial

 I
 II
 III
     Coefficients for Third Order Working Curves

abed
-1.100 x 10~2
-2.081* x 10~2
-1.881 x 10"1*
2.656 x 10~k
2.856 x 10"1*
1.209 x 10"1*
                          7.598 x 10~7
                          7.502 x 10~7
                          9.351 x 10~7
-2.1*09 x 10~10 8.581* x 10~J
-2.1*70 x 10~10 ll*.l  x 10"
-2.901* x 10~10 3.101* x 10~3
0.9999
0.9999
0.9999

-------
                        2.0,-
                               RELATIVE ACCURACY AS A FUNCTION OF SULFATE CONCENTRATION

                               WITH SULFAVERR (THIRD ORDER REGRESSION FOR THE WORKING CURVE)
                        1.6
-3
H
                        1.2
                  01
                    
-------
                               COEFFICIENT OF VARIATION AS A FUNCTION
                          OF SULFATE CONCENTRATION  (THIRD ORDER REGRESSION FOR THE WORKING CURVE)

                                             SULFAVERR  METHOD
ro
i
             C
             o
             
-------
V.  EVALUATION OF THE MIDWEST RESEARCH INSTITUTE VERSION OF THE AUTOMATED


    METHYLTHYMOL BLUE METHOD


    A.  Introduction


        The working curve for the MTB procedure is distinctly non-linear at


        its extremes.  Non-linearity is thought to result from the lack of


        purity of commercial grades of methylthymol blue.11  Thus, vhen a


        reagent containing 1:1 mole ratio of Ba to MTB is prepared based on


        an assumed MTB purity of 100$, the resulting reagent, in fact, has a


        Ba to MTB ratio > 1.0.  Because of the formation of a 2:1 Ba-MTB


        complex with absorbance differing from that of the 1:1 complex, a


        non-linear working curve results.  Accordingly, if the purity of the


        MTB can be determined, a reagent can be prepared providing known


        Ba:MTB ratios.





        The Technicon was set up according  to the MRI method's specifications


        as detailed in Appendix E, with the following exceptions.  A Uo sample


        per hour timing disc was not initially available so a 30 sample  per


        hour timing disc with a 6:1 sample to wash ratio was installed and


        used for assaying the MTB.  Later studies employed hO samples per hour


        as specified.  Instead of the linearizer and digitizer (which was not


        available), a chart recorder was used.





        Using Uo samples per hour as specified, the analyses were observed


        to exhibit a marked memory effect.  This was particularly noticeable


        when successive samples differed greatly in concentration.  A significant


        improvement in accuracy and precision was found at lower sample rates


        (e.g., 30/hour).  The memory effect has been eliminated by using two

                                                         2
        cups for each sample and reading only the second.   In spite of the


        deficiency, the method was evaluated as written (excepting the MTB


                                    - 73 -

-------
    assay procedure as noted above).   Samples were analyzed in random




    order to provide a realistic measure of the memory effect.








B.  Determination of MTB Reagent Purity




    The measurement of MTB purity was performed according to MRI's




    directions using a 3:1 BaCl2 to MTB mole ratio and sulfate standards




    ranging from 20 to 200 yg/ml.  This was replicated on three separate




    days.  During all trials, the baseline remained stable but the span




    increased despite a one hour or longer warm-up period.  This increase




    was  also noted by MRI personnel (F. Bergman, Private Communication, 1978).








    When peak height (ordinate) was plotted against sulfate concentration




    (abscissa), the result in every case was a smooth curve having a




    continuously increasing slope as illustrated in Figure 17.  To measure




    the  purity of the MTB reagent, three straight lines are fit through




    experimental points of the curve and the intersection points determined.




    The  lines were obtained by selecting points which visually provided the




    best three linear segments and calculating the respective linear




    regressions; the intersections were determined mathematically.  In




    the  first three trials we experienced the problems of upward drift




    and  of  selecting the data points within the three linear regions of




    the  curve.  To overcome these difficulties the fourth trial included




    modifications to the MRI procedure:  eighteen (in place of 13)




    standards were run three times on a given day.  The average of these




    three consecutive runs compensated for the drift and provided more




    clearly defined linear regions having at least four data points each.




    (The MRI procedure requires that the standards be run in triplicate

-------
 100 r-
                                          Figure 17


                     MTB ASSAY BY MIDWEST RESEARCH INSTITUTE  PROCEDURE

                                      FEBRUARY 2,  1978
  75
                            Calculated Purity: 78.5%
4J
Tl
c
  50
                                                               Second
                                                               Intersection:
                                                                 163.5
  25
                                          ®.
                                     ©
                                       First Intersection:  96.9 jug/ml
                                                    I
                                            _L
                                        I
   20
40
60
80       100       120        140

 Sulfate Concentration  Qig/ml)



            - 75 -
160
180
200

-------
   without  specifying the protocol).






   The first  (i.e. lower sulfate)  intersection  defines  the  sulfate



   concentration required to react with the  excess "barium and the second



   intersection indicates the concentration  required to react with  the



   complexed  barium.  Purity of the MTB is calculated as shown in Table 30.



   Table  31 compares the line segments and calculated purity of the MTB



   found  in the trials.  Despite the difficulties experienced in the


   early  tests, the Midwest assay  procedure  was reproducible within 5-5$


    (coefficient of variation).  Because of the  greater  reliability  of the



   last purity determination made, the value 78.5$ "was  used rather  than


   the mean value of all trials, jk.5%.






   The MRI  procedure utilizes a ratio of Ba/MTB of 1.13 m/m for MTB of


   100$ purity.  With the current  lot of MTB, the weight of impure  MTB



   was increased by the ratio 1/0.785 to provide a Ba/MTB value as


    specified.






C.  Linearity  arid Reproducibility of the Working Curve



    Thirteen sulfate standards -ranging from 2 to 100 yg/ml were run  in



   triplicate on three separate days to evaluate the shape  and reproduci-


   bility of  the calibration curve.  The graphs obtained consisted  of two


   approximately linear regions, one from 2  to  about 30 yg/ml, and  the



   other  from 40 to 90 yg/ml (dashed lines in Figure 18).  Table  32



    compares the correlation  coefficient  (r)  and the  standard error  of the



    estimate (S) for three options  in data handling:   l)  linear regression
               j *•

   covering the 2-90  yg/ml range,  2) two linear regressions, one for the



                                -  76 -

-------
                                 Table  30


                         Calculation of MTB Purity
           [yg SOt, /ml(2) - MS SO^/ml^J Fj x C x 866.73

  Purity = 	W x F2 x 2	j	x 10°
•where:



        yg SOij /ml/2} = yg/ml reading from second intersection point



        yg 30^ /ml/  > = yg/ml reading from first inter sectors point



        F! = flow rate in ml/min for sample line (normally 0.32)


        C  = molar concentration of stock solution  (normally 1.0^1 x 10  )



        866.73 = molecular weight of MTB



        ¥  = weight  of MTB in mg



        F2 = flow rate in ml/min for MTB line  (normally 0.7)


        2  = value to calculate MTB concentration on basis of equivalents/liter
        Calculation  of Weight  of MTB


                        .  n  *      ,   •     0.16U  x 100
        Weight of material  for analysis. =   $ Puritv—
                                    - 77 -

-------
                                             Table 31
                        Assay of MTB by Midwest Research Institute Method'
                                                                         a,b
Trial





Mean

1
2
3
1*
Slope :
Line
a
1.80
-1.U2
-0.79
0.60
0.251 +
1
b_
0.21*6
0.253
0.272
0.232
.017

Line
2
a
-13.
-18.
-16.
-22.
0.
03
93
17
27
1*37 ±
0
0
0
0

b_
.1*00
.M*3
.1*35
.1*68
.028

Line 3
a.
-125
-133
-117
-ll*7
1
.98
.7
.7
.5
.16 +. .

b
1.13
1.17
1.10
1.23
06
Purity ,

68.9
76.2
7U. 1*
78.5

                                                                  Mean Purity:    7!*.5 + U.I
a.  Eastman Kodak methylthymol blue, Lot  Fo. E3X



b.  Intercept - a,  Slope = b

-------
  100
   75
w
4J
•H
c
D
o
   50
   25
                                       Figure 18


                         WORKING CURVE FOR MRI-MTB PROCEDURE

                                  FEBRUARY 27, 1978
                                                  	data fit with two straight lines,


                                                  	data fit with one straight line.
                         25
50
                                                                  75
100
                                Sulfate Concentration  (pg/ml)
                                          -79-

-------
                                                                   Table  32

                                          REPRODUCIBILITY OF CALIBRATION:  SULFATE BY MRI-MTB METHOD


       Trial         One Straight Line        	Two Straight Lines	    	Third Order
                 *      **
                a      b        r     S       Range    a      b        r^y.x      a      ^        c         d
         1     1.77   1.01   0.9988   1.1*7     2-30   0.2l*   1.12   0.9996   0.358     -0.12    1.2   -2.9-10"3   6.5-10"6   0.9999   0.386
                                              1*0-90   7-90   0.92   0.9993   0.598

         2     3.1*1*   0.98   0.9967   2.1*6  •   2-30   1.36   1.13   0.9868   2.01       0.95    1.2   -3.5'10~3   8.1-10"6   0.9999   1-78
                                              1+0-90  10.0    0.89   0.991*2   1.72

,         3     2.66   l.Ol*   0.9985   1-72     2-30   1.13   l.lU   0.9995   0.351*      0.98    1.2   -0.1*-10~3  -0.lU.10~6  0.9999   0.535
^                                             UO-90   9.78   0.93   0.9975   1.07
o
i
       *a = intercept

      **b = slope

       +y = a * bx + ex2 + dx3

-------
    2-30 and the other for the HO-90 yg/ml range, and 3) third order



    regression for the 2-90 yg/ml range.  This comparison based on


    working curves constructed on three separate days, confirms the


    presence of two approximately linear regions, with reproducible


    slopes, but third order regression over the entire range gives a


    slightly superior fit, as measured by S
                                           y.x





    The goodness of fit of the three regression techniques as a function


    of concentration is illustrated for a single trial in Table 33.


    Fitting a single line in the- range 2-100 yg/ml yielded large errors


    at low concentrations.  The difference between the two straight  line


    fit and third order regression was not significant.






    An experiment was performed to evaluate the effect of a 5% error in


    the MTB assay on the working curve since a 5$ C.V. was found for the


    assay.  The calibration standards were run using a reagent with a


    5% excess MTB in one case and a 5$ deficiency in the second.  Table 3^


    shows that neither the slope nor the shape of the curve are affected.






D.  Accuracy with EPA Sulfate Audit Strips


    Glass fiber filter strips loaded with five different levels of sulfate,


    four strips per level, were extracted following the MRI 30 minutes


    ultrasonic extraction procedure.  Sulfate analyses by the MRI-MTB


    method are shown in Table 35.  The results indicate a positive error


    of 9 and lU$ at 15 and U yg/ml, respectively, using third order



    regression.  At higher concentrations, the error remains <_ \%.  Except


    at U yg/ml, the coefficients of variation remained at or below 5%-



                               - 81 -

-------
                             Table 33




CURVE FITTING AS A FUNCTION OF CONCENTRATION FOR THE MRI-MTB METHOD





                                       % Error
Sulfate Standard
(yg/ml)
100
90
TO
50
30
20
10
5
3
2
Single
Straight Line
-7-5
-2.9
-0.7
+U.O
+5-7
+5.6
+1.3
-23.1
-U3.0
-71.3
Third Order
-1.6
+1.2
+0.3
+2.0
+1.7
+2.2
+5.0
+0.3
+6.3
+11.3
Two
Straight Lines
-U.2
-0.9
-O.U
+0.8
0
+2.1
+5.0
-3.0
-3.3
-6.3
                               - 82 -

-------
                                                             Table 3k




                        EFFECT OF A 5% ERROR IN MTB ASSAY UPON THE CALIBRATION CURVE FOR THE MRI-MTB METHOD
Error One Straight Line
* **
b r Sy.x
+5$ 1.77 1.00 0.9985 1.6U
-5% 2.23 1.01 0.9988 1.52
i
OO
i
Two Straight Lines
Range
2-30
1*0-90
2-30
kQ-90

a
0.02
8.05
O.Ik
8.75

b
1.12
0.91
1.11
0.92

r
0.9991
0.9983
0.9997
0.9992

Third Order
Sy-X a b c d r
0.1+95 -0.19 1.2 -1.5-10"3 -5.0-10"6 0.9999
0.91*6
0.290 0.62 1.1 -0.75-10"3 -9.6-10~6 0.9999
0.651*


0.595
0.270

 *a = intercept



**b = slope



 +y = a + t>x + ex2 + dx3

-------
                                                   Table 35




                          ANALYSIS OP EPA SULFATE AUDIT STRIPS BY THE MRI-MTB METHOD"
                Approximate


Sample
1000 Series
9000 Series
8000 Series
5000 Series
7000 Series
i
O3
Concentration
Analyzed
( yg/ml )
1*
15
35
53
85

yg/strip
Theoretical
Value
200.2
71*5.6
1737-2
2670.1
1*21*0.2

Single Line
Regression
182 +_ 53
812 + 36
189!* +. 80
2829 ± 35
1*261 + 57

Two Line
Regression
233 ± 1*9
837 ± 32
187!* + 75
251*9 ± 79
1*393 +. 61

Third Order
Regression
229 +_ 1*0
810 +_ 1*3
1780 +_ 8U
2666 _+ 39
1*1*06 +_ 92

C.V.
(Third Order
Regression )
17
5.3
1*.7
1.5
2.1


Third Order
Theoretical
1.1U
1.09
1.03
0.999
l.Ol*

a.  Results are mean values +_ 1 a for four strips from each series extracted by  30 minutes  ultrasonic extraction.

-------
E.  Working Range




    The concentrated atmospheric  sample  extract previously used for




    evaluating turbidimetric methods was diluted to varying degrees




    to permit evaluation of relative accuracy and precision as a function




    of concentration.  Because the working  curve in the 0-100 yg/ml range




    only approximated a straight  line, the  data was analyzed both by




    linear and third order regression.   The results are given in Table 36




    and Table 37.  The calculated, undiluted concentration of the extract




    showed no bias in both cases  at 12.5$ and higher concentration.  Thus




    the correct undiluted concentration  was taken to be the mean of eight




    results 139 +, U.5 yg/ml  (linear regression) or 137 +.3.6 yg/ml (third




    order).  These compare to 139 +_ ^-7  yg/ml by the AIHL Method 75 and




    Ih2 +_ k.9 yg/ml by AIHL Method  6l.








    The ratios of observed to expected concentrations are plotted against




    expected concentration in Figure 19.  The results indicate accuracy




    within 1% of  the optimal value  in the concentration range from 17 to




    91 yg/ml using linear regression.  With third order regression,




    accuracy at low sulfate concentrations  is significantly better compared




    to linear regression.  In the range  17  to 89 yg/ml the results are




    within 5% of  the optimal*value.  The>coefficients of variation of




    three trials  are plotted against concentration in Figure 20.  Little




    difference in precision is noted between linear and third order




    regression except for samples < 5 Ug/ml.  The C.V. values are < k%




    by both linear and third order  regression in the range 10-90 yg/ml.








    Based on both relative accuracy and  precision as measured by C.V.,




    the working range of the MRI-MTB method is from 17 to >_ 90 yg/ml by both





                                - 85 -

-------
                                                       Table  36
Concentration
% of Undiluted

     2.5
     5-0









1
OO
1
7.5
10.0
12.5
15.0
18.75
22.0
31.25
!*3.75
56.25
65.0


Observed Mean
 of 3 Trials

    2.7
    5.0
    8.5
   12.1
                      16.
                      20.
                      27-9
                      31.0
     .7
     -5
   60.6
   77.6
   89.1
                                   Relative Accuracy  and Precision  of MRI-MTB Method
                                        with  an Atmospheric Sample (pg SO^  /ml)
C.V.
Linear Regression

        Calculated
        Undiluted
      Concentration
0.23
0.32
0.28
0.39
O.U5
0.1*3
0.73
0.17
0.10
0.50
0.1*1
0.10
8.8
6.3
3.3
3.2
2.7
2.1
2.6
0.6
0.2
0.8
0.5
0.1
                106.1*
                100.5
                113.5
                121.3
                133.6
                136.U
                11*8.6
                11*1.0
                ll*0.6
                138.5
                138.0
                137.1

       Mean undiluted = 139 + 1*.5
   Expected
   Diluted
Concentration

     3.5
     7.0
    10.1*
    13.9
    17.1*
    20.9
    26.1
    30.6
                                                                                   1*3,
                                                                                   60,
                                                                                   78,
                                                                                  90.5
Observed Cone,
Expected Cone,

   0.761*
   0.722
   0.815
   0.872
   0.960
   0.980
   1.067
   1.013
   1.010
   0.995
   0.992
   0.985
                                   Coefficients for Linear Regression Working Curves
                                   Trial

                                   I
                                   II
                                   III
                                a
                              •^^^^MM«*(*V>

                               1.75
                               2.29
                               0.311
                 0.978
                 0.978
                 1.000
                            y.x

                          1.983
                          1.786
                          2.002

-------
                                                        Table  37
                                    Relative Accuracy and Precision of MRI-JfTB Method
                                         with an Atmospheric Sample (yg SO^ /ml)
 Concentration
 % of Undiluted

      2.5
      5.0
      7.5
     10.0
     12.5
     15.0
     18.75
     22.0
     31.25
     1*3.75
i     56.25
oo   65.0
•—3
Observed Mean
 _of 3 Trials
                                                 Third Order Regression
c.v.W
  Calculated
   Undiluted
Concentration
  Expected
  Diluted
Concentration
                Trial
                ^*MV-«V*V

                I

                II

                III
3
6
9
12
16
20
26
29
1*2
59
77
91
• 9
.0
.2
• 5
.6
.1
.9
.9
.3
.1
.5
.1
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
It*
35
22
31*
1*2
33
65
29
29
72
1*1
06
3.
5.
2.
2.
2.
1.
2.
1.
0.
1.
0.
0.
5
8
1*
7
6
7
1*
0
7
2
5
1
156.0
120.1*
122.3
12 1*. 5
132.9
133.8
ll*3.7
135.9
135 A
135-0
137.9
ll*0.2
3.
6.
10.
13.
17.
20.
25.
30.
1*2.
59.
77.
88.
1*
8
3
7
1
5
7
1
8
9
0
9












Mean undiluted = 137 ±.3.6
Coefficients
0.







a
1.105 x 10
T.536 x 10
-15A29 x 10

-1
-1
-1

1
1
1
for
b
.092
.081*
.135
Third Order Working Curves
c
-3.297 x I0~k -1.
-5.891* x 10'1* -0.
-lO.lM* x 10"1* -0.

d
073 x
670 x
5^5 x


10"
10"
10"


5
5
5
Observed Cone.
Expected Cone.

   l.lUO
   0.880
   0.891*
   0.910
   0.971
   0.978
   1.050
   0.99^
   0.989
   0.986
   1.007
   1.025

-------
        2.00,-
        1.60
                            RELATIVE ACCURACY AS  A FUNCTION OF CONCENTRATION FOR THE MRI-MTB METHOD
                                                         • 3rd Order Regression for the Working Curve



                                                         O Linear Regression for the Working Curve
oo
00
    4-1

    Ctj
    C
    0)
    u
   -o
    
-------
                    COEFFICIENT OF VARIATION AS A  FUNCTION OF CONCENTRATION FOR THE MRI-MTB METHOD
   10.Oi-
B-5
                                                     •  3rd  Order Regression for the Working Curve




                                                     O Linear Regression for the Working Curve
                        20
40                60                80




Expected Sulfate Concentration  <)ig/ml)




             Figure  20
100

-------
         linear and third order regression.  These results compare to a


         working range 7-75 yg/ml (linear regression, accuracy +_ 5-5$)
                                                                           2
         found by this procedure for another modification of the MTB method


         which did not correct for the impurity of the MTB.  Thus, as regards


         accuracy or working range, the MRI procedure offers no discernible


         advantage over a simpler procedure evaluated earlier.





         The diminished working range, at low sulfate levels, for the MRI


         procedure results , we believe , primarily from a sample-to-sample


         memory effect.  This may be reduced as described previously.





VI.  EVALUATION OF THE COLOVOS MODIFICATION OF THE MTB METHOD


     A.  Introduction


         G. Colovos et al  have described an empirical procedure for linearizing


         the MTB working curve in the range 0 to 100 yg/ml sulfate.    By


         decreasing sample dilution, this technique allows measurement in the


         range 0-10 yg/ml.  For purpose of the present report the resulting


         procedure is referred to as the Colovos modification of the MTB


         method ( Colovos-MTB ) .





         Implicit in the Colovos procedure is the presumption that a 1:1 mole


         ratio of Ba:MTB provides the most linear working curve.  The rationale


         for this presumption is given in Section V-A.  Colovos et al deter-


         mined the linearity of a series of working curves prepared with


         varying Ba-impure MTB ratios.  If for example, a Ba-impure MTB mole


         ratio of 0.6  resulted in the most linear working curve, this ratio


         is presumed to provide a Barpure MTB mole ratio of 1:1 and is used
a.  Calculated using the molecular weight for pure MTB.  This is therefore,
    only an approximation.

                                 - 90 -

-------
    for analyses.  A spectrophotometric estimation of MTB purity can also




    be performed.  However, this accounts for non-absorbing impurities



    only.









B.  Linearizing the Working Curve




    The Technicon system was set up as in the Midwest Research procedure




    described in Appendix E except that the rate used was 30 rather than




    UO samples per  hour.   The spectrophotometric estimation for Eastman




    Kodak MTB, lot E3X, gave a 72.5$ purity compared to the 78.5% obtained




    by the Midwest procedure.  There was, however, insufficient MTB




    remaining from this lot for use with Colovos' empirical technique for




    determining MTB purity.








    With Eastman Kodak Lot C5F, spectrophotometric analysis yielded a




    purity of 6k.9%.  This batch was used to prepare working curves




    ranging from 0.65 to 1.2^ in Ba/impure MTB mole ratio.  The degree




    of curvature for each working curve was evaluated by measuring the




    goodness of fit of a straight line regression for the 2-100 pg/ml




    data (Table 38).  Based on the minimum in SywX for a Ba/MTB mole




    ratio of 0.78, this ratio was selected for use in evaluating the




    Colovos procedure.  This suggests a purity of ca. 78$.








    Since this procedure was not performed on the same MTB lot as used




    with the Midwest procedure no direct comparison is possible.  However,




    both lots were analyzed by spectrophotometry.  Based on analyses of




    lot E3X (655?) a purity by the MRI method about 6% above that by




    spectrophotometry would be expected with lot C5F, or about 71$-  This








                               - 91 -

-------
                             Table  38

             Straight Line Fitting as a Function of
       Concentration for the Colovos-MTB Method (% error)
Sulfate
Cone .
(yg/ml)
100
90
80
70
60
50
UO
30
20
10
7
5
3
2
Intercept
Slope
r
Sy.X
Obs.-
1.2U
- O.U
3,2
2.8
1.1
0.0
- U.9
-10. U
-12.5
-17.7
- U.6
13.7
35-5
92.9
169.5
-3.8U9
1.011
0.997
2.12k
True v inn«/
1.00
- U.U
1.1
3.2
3.5
3.3
2.0
- 1.0
-18.0
- 5.7
- 0.7
7.8
15-3
39.2
—
-2.211
1.008
0.998
2.5U6
Ba/MTB
0.92
- 5.0
0.0
2.6
U.I
U.3
2.1
- 0.8
- 1.0
- 8.3
- 8.2
-10.1
- U.l
0.1
25.3
Regression
0.185
l.OOU
0.998
2.050
•\T*
(mole ratio)
0.78
- 3.U
- 2.0
1.3
3.0
5-2
3.U
1.6
1.6
- U.U
-11. U
-1U.3
-17.5
- 5.2
7.7
Parameters
-0.382
1.013
0.999
1.792
0.71
- 5.0
- 2.U
2.U
3.5
5.6
U.3
3.6
U.5
- 2.7
-10.9
-20.2
-21. U
-21.1
-U0.2
1.075
1.02U
0.998
2.387
0.65
- U.7
- 2.6
1.1
3.U
5.6
7.0
U.2
U.5
- l.U
-10.1
-17.9
-28. U
-36.5
-U9.0
1.66U
1.016
0.998
2.U28
True
                             - 92 -

-------
    is  1% below the observed purity by the Colovos procedure,  a result




    which we consider equivalent within experimental uncertainty.








    The working curves for Ba/MTB mole ratios 1.00, 0.78 and 0.71  are




    compared in Figure 21.  A ratio 1.00 corresponds to a reagent  prepared




    assuming the MTB to be 100$ pure (note that Colovos and the Midwest




    methods differ in the specified Ba/100$ MTB ratios; Colovos specified




    1.00 while MRI specified 1.13).  In all cases the curves remained




    distinctly S-shaped with an obvious increase in linearity  for  the




    mole ratios 0.71 and 0.78 ratios compared to 1.00.








C.  System Change for Operation in the 0-10 ug/ml Range




    In normal operation of the MTB method, the sample is diluted 1 to 6.




    As noted by Colovos, the sensitivity can be increased by reducing the




    dilution.  The system was modified to permit the analysis  of the




    undiluted sample by using a 2 ml/min pump tubing for the sample,




    eliminating the water line, the sample air bubbler and the mixing




    coil but keeping the debubbler.  The waste line pump tubing was




    changed to reduce the flow from 2 ml/min to 1.6.  This was found




    necessary because the higher flow rate produced air bubbles in the




    colorimeter cell.








    The reproducibility of the working curve from day to day using this




    system is shown for six trials in Table 39.  The slope remained




    constant within about 5% but the intercept varied markedly.  To




    illustrate the linearity of the working curve Figure 22 plots results




    for Trial 3; both the slope and the S    for this trial closely
                                         «y • x



    approximate the mean values for six trials.






                              - 93 -

-------
                                               Figure 21
                                 Comparison of Working Curves as a
                                 Function of Ba/MTB Ratio
U)
•X
o
    so

-------
                      Table  39

         ReproduciMlity and Linearity of the
           Working Curve for the Colovos-MTB
                   Method (0-10 yg/ml)
ial
1
2
3
U
5
e
Slope
6.00
6.13
6.1*
6.75
6.78
6.79
Intercept
0.173
-1.20
0.039
0.0638
-1.68
-l.lU
r
0.9967
0.9870
0.997^
0.9979
0.998U
0.9972
^y.x
1.U60
3.062
1.636
1.185
1.099
1.U63
mean    6.^8 + .35                                    1-65
                          - 95 -

-------
                                                    Figure 22
vo
ON
                                    TYPICAL WORKING CURVE  FOR COLOVOS-MTB METHOD
                                                IN  0-10 jug/ml RANGE
                                                                          y  =  6.44 x +0.039
                                                                        Sy.x  •  1.636
                                                                          r  «  0.997
                                                    4                6

                                             Sulfate Concentration  (pg/ml)
_J
 10

-------
D.  Accuracy and Working Range




    Glass fiber filter strips loaded with five levels of sulfate, four




    strips per level, were extracted following the MRI ultrasonic ex-




    traction procedure.  Extracts were diluted to cover the range from




    1 to 10 yg/ml.  The results, therefore, are subject to error because




    of differences in accuracy and precision over this concentration range.




    As shown in Table Uo, the results are 30$ low at 1 yg/ml but are




    within 5$ of the true value for concentrations of U to 9.8 yg/ml.




    Above 1 yg/ml the C.V. values remain <_ 2%.








    The concentrated atmospheric sample extract previously used for




    evaluating the turbidimetric and MRI-MTB methods was diluted to




    varying degrees to permit evaluation of relative accuracy and preci-




    sion as a function of concentration and from these, the working range.




    The results are given in fable Ul.  The calculated, undiluted concen-



    tration of the extract .showed no bias at 2 yg/ml and higher .concentration.




    Thus the correct undiluted concentration was taken to be the mean of




    ten results, 1^3 +_ 5 yg/ml.  These compare to values from 139 to lU8




    found by five other procedures as given in footnote b to Table Ul.








    The ratios of observed to expected concentrations are plotted against




    expected concentration in Figure 23.  The results indicate accuracy




    within 5$ of the optimal value in the concentration range from 2 to




    10 yg/ml.  The coefficient of variation of three trials are plotted




    against concentration in Figure 2k.  The C.V. values remain < 5%




    (with one outlier) in the range 2-10 yg/ml.  We conclude that, based




    on relative accuracy and precision as measured by C.V., the working
                               - 97 -

-------
                            Table 1*0




 Analysis of Sulfate EPA Audit Strips by the Colovos-MTB Method






Approximate
Sample
9000 Series
5000 Series
1000 Series
8000 Series
7000 Series
i
00
t
Concentration
Analyzed
(yg/ml)
1
2.7
i*
7
9.8


Theoretical
Value
(yg/strip)
7^5.6
2670 . 1
200.2
1737.2
1*21*0.2

Regression Equation:
Observed
Value
(pg/strip)
511* + U3
2360 + 1*
193 + 3
1770 + 38
1*1*56 +_ 57

y = 0.17 + 6.00X
r = 0.9967
c.v. (%)
8.1*
0.2
1.6
2.2
1.3


Observed
Theoretical
0.69
0.88
0.96
1.02
1.05


                                    S    = 1.U60
                                     y.x

-------
                                                  Table Ul

                              Relative Accuracy and Precision of Colovos-MTB Method
                                   with an Atmospheric Sample (yg SO^/ml)
Concentration
% of Undiluted

    0.36
    0.72
    1. U*4-
     .80
    2.16
.88
.60
.32
    2.
    3.
    1*.
    5-
    5.
    6.
    7.19
.76
              Observed Mean
               of 3 Trials
                  0.83
                  0.82
                    07
                    53
                   .97
                   .35
                  6.17
                  7.22
                  8.18
                  9.75
                 10.65
                                             C.V.
O.U6
O.lU
0.08
0.05
0.07
o.oU
0.39
0.25
0.35
0.29
0.3U
55.85
16.58
3.95
1.81
2.56
1.02
7.25
I*. 12
U.78
3.55
3.51
                                                            Calculated
                                                            Undiluted
                                                          Concentration

                                                               231
                                                               llU
                                                               lUH
                                                               133
                                                               138
                                                               1U9
                                                               1U3
                                  O.U
                                               k.oQ
                                                               1U2
                                                               151
   Expected
   Diluted
Concentration
    0.
    1.
    2.
    2.
    3.
    U.
    5-
   .52
   .03
   .06
   .58
   .09
   .12
   .16
 6.19
 7.22
 8.25
 9.27
10.30
Observed Cone,
Expected Cone.

    1.59
    0.79
    1.00
    0.98
    0.93
    0.96
    1.03
    0.99
    1.00
    0.99
    1.05
    1.03
                                               Mean undiluted = 1U3.2 +5.3
a.  Data reduction by linear regression.
b.  Compares to results (ug/ml):
        AIHL Method 6l          :
        Optimized turbidimetric:
        SulfaVer                :
        Midwest MTB (3rd order):
        Dionex 1C               :
                                  lUl.8 +_ U.8
                                  139- ^ +. H.7
                                  1U3-5 ± ^-7
                                  137 ±3.6
                                  1U8.2 + 1

-------









1

o
o
,



c
0
•<-t
CO
4J
c
a)
u
C
0
o
-o
(U
fc
(U
(0
t*\
o
a
o
•l-l
CO
4J
C
cl)
U
C
o
u
•X)

-------
                                                       Figure 2k



                            COEFFICIENT OF VARIATION AS  A FUNCTION OF SULFATE CONCENTRATION

                                               WITH  THE  COLOVOS-MTB METHOD
        60  r-
        50
H
O
H
    C
    O
    a
    •H
    H
    to
4J
a
0)
    40
        30
    0)
    O
    O
        20
        10
                                                 4567


                                               Expected Concentration  (jig  SO^/ml)
                                                                                                       10
11

-------
          range of the Colovos modification of the MTB method, (0-10 yg/ml)

          is from 2 to > 10 ug/ml sulfate.
VII.  EVALUATION OF THE DIONEX ION CHROMATOGRAPH

      A.  Introduction

          On recommendation of the EPA Project Officer a Dionex Model 10 ion

          chromatograph (1C) was modified by water jacketing and thermostatting

          the column and detector and insulating these as well as the Teflon

          solution lines.  The unit was operated with a Varian A-25 recorder

          and Spectra-Physics Autolab Minigrator.  Operating parameters were

          as follows:

          Temp:  35°C

          Eluent:  0.0030 M NaHC03 + 0.002** M Na2C03

          Column:  3 x 100 mm precolumn + 3 x 500 mm anion separator
                   6 x 250 mm anion supressor column

          Eluent flow rate:  2.5 - 3-3 ml/minute

          Ranges:  10, 3, and 1 ymho

          Recorder:  1 volt input (full scale)

          Minigrator:  peak width = 99 seconds

          Sample loop:  30 yl



          A detailed protocol is included as Appendix F.



          In addition to peak area, peak height was determined manually by

          subtracting the middle of the baseline trace from the middle of

          the peak trace.  The nitrate and sulfate peaks were not completely

          resolved when present at similar concentrations (Figure 25) .  The

          effect of this small overlap on sulfate determination is reported

          in Section VIII.

                                    - 102 -

-------
                                  Figure 25

                    Resolution of Sulfate and Nitrate at

                 Equal Concentrations by Ion Chromatography

Range:  10 ymho (linear) -----
3 x 100 mm precolumn
3 x 500 mm anion
           separator
Temp:  35°C
Eluent Flow: 2.5 ml/min
Eluent: 0.003M NaHC03
        0.0024M Na2C03
Recorder/Dionex
  Response 1:1
                                 •»•« n NI
                                                     oo-seiooo-i£
                                   - 103 -

-------
B.  Reproducibility of Working Curves and Linearity of Range Switch




    Initial objectives in evaluating the 1C were l) to establish the




    reproducibility of the working curve set on a range suitable to




    provide analysis in the 0-150 yg/ml range, 2) to determine if




    calibrations obtained on a given range were directly applicable to




    other ranges after correction for the change in umhos corresponding




    to full scale and 3) to compare peak height and peak area techniques




    for data reduction.








    Range 10 ymho was selected for analysis in the 0-150 ug/ml sulfate




    range.  The day-to-day variation in the working curve for three




    trials is shown in Table U2 and Figure 26.  The table includes results




    employing both peak height and area and for the figure, peak heights




    only.  An analysis of covariance established that results by the




    peak height method were significantly different on the three days.




    Inspection of the results for peak areas suggests this is true here




    a.3 well.  Accordingly, the determination of a working curve must be




    a part of any analytical protocol for daily use of the ion chroma-




    tograph.








    A comparison of working curves for ranges 1, 3, and 10 ymho is shown




    in Figure 27.  No effort was made to attenuate or amplify the output




    voltage at the recorder.  Range 1 umho was found suitable for working




    curves extending to about 30 yg/ml, whereas the 3 ymho, range was good




    for working curves up to about 80 vig/ml.  The curve obtained for  10 ymho




    includes all results for the three trials illustrated in Figure 26.




    No allowance could be made for day-to-day variability in the data for




    ranges 1 and 3 ymho.







                             - 10k -

-------
I
H
O
       Trial




         1



         2



         3
                                                         Table h2



                        Reproducibility of Working Curve for Dionex Ion Chromatograph (Range:  10 ymho)

pk. ht.
-0.38
-0.09
-0.2.8
a

pk. area
-0.90
-1.12
-2.81
x 105
x 105
x 10 5

pk.
0.
0.
0.
1
ht.
332
293
3k6
D
pk.
1.59
1.52
1.68

area
x 105
x 105
x 105

pk
0
0
0
r
. ht.
.9997
.9997
.9997
by.x
pk. area
0.9996.
0.9999
0.9997
pk. ht.
0.361
0.3U6
0.1*07
pk. area
2.18 x 105
1.20 x 105
2.03 x 105

-------
50 r
                          REPRODUCIBILITY OP WORKING CURVE FOR DIONEX ION CHROMATOGRAPH
                                                (Range:   lOpmho)
                                                                              A 5/3/78

                                                                              D 5/11/78

                                                                              O 5/19/78
                  20
40
60
80
100
                                       Sulfate  Concentration (jig/ml)
                                                Figure 26

-------
   lOOr-
COMPARISON OF WORKING CURVES FOR DIONEX  ION  CHROMATOGRAPH
  ON RANGES 1,  3  AND lOjimho
•H   °UH
4J
VI
oo
                                                  60             80

                                           Sulfate Concentration (ug/ml)

                                                     Figure  27
                                             100
120
                                                                            140

-------
   Since the difference in intercepts  for the three  curves  is  small,




   the  linearity  of the range  switches can be evaluated by  plotting




   the  regression equation slopes  against 1/ymho  for full scale




    (Figure  28) .   These data  suggest that calibration curves run  on




    a given  range  can be used  to determine,  approximately,  samples




    run  on the  same day using another range.  For  example, using  the




    regression  equation from  range  10 ymho, with slope multiplied by




    10/3 and the peak height  observed on range 3 for  a 25 yg/ml sample,




    a concentration of 26. 7 yg/ml is calculated.   This represents an




    error of +1%.








    Based on these initial results  there appeared  to  be  relatively




    little distinction between  use  of peak heights and peak  areas.  An




    additional  comparison of  these  procedures is included below.








C.  Accuracy with  EPA Sulfate Audit Filter Strips




    Extracts of 20 EPA sulfate  audit strips (four  at  each of five levels),




    were analyzed  by 1C using range 10  ymho only (Table  U3).  The analyses




    were performed on three separate days for which individual  working




    curves were prepared with three or  four standards .  The  concentration




    of the extracts ranged from U to 85 yg/ml.  The results  indicate a




    positive bias  in all cases  with less accuracy  and precision at




    <_ 15 yg/ml.  At higher concentrations the results were accurate




    within 1% using peak heights with a C.V.  of £ 9%. The persistent




    positive bias  may reflect,  in part, the effect of slight overlap of




    nitrate  and sulfate peaks.   This would be particularly significant




    for  the  9000 series samples which,  had a  relatively,  large




   proportion  of  nitrate.  (SO^ /WOa"  = 0.6  w/w) . The peak  height
                             - 108 -

-------
           REGRESSION EQUATION SLOPE AGAINST
        FOR FULL SCALE WITH DIONEX ION CHROMATOGRAPH
   4r~
s
8  2
   0
                                                          10
                        jumho FULL SCALE
                          Figure 28


                           -109-

-------
                                   Table h3




    ANALYSIS OF EPA SULFATE AUDIT STRIPS BY THE DIONEX ION CHROMATOGRAPH8
Approximate
Concentration
Analyzed
Sample (yg/ml)
1000 Series 1*
9000 Series13 15
8000 Series 35
5000 Series 53
, 7000 Series 85
I i
H
O
a. Results are mean values +_ 1
Theoretical Observed Value
Value (yg/strip)
( UK/strip) Pk. Ht. Pk. Area
200 . 2 282 1*36
71*5.6 865 999
1737-2 1862 1929
2670.1 2787 2681
1*21*0.2 1*1*25 1*221*
C.
Pk. Ht.
19-6
11.1
9.0
3.7
0.8
a for four strips from each series extracted by
b. The weight ratio sulf ate /nit rate in this series was 0.6 compared-

Dav_
1
2
3
Regression Equations
(peak height)
a b
-0.233 0.1*1*1
-0.709 0.1*65
-0.527 O.l*3l*
to 2.2 to

r
0.9999
0.9995
0.9999
v.W
Pk . Area
166
37.7
15.5
3.0
7.9
30 minutes
12 for the

Sy.X
0.122
0.600
0.275
Observed
Theoretical
Pk. Ht. Pk. Area
1.1*1 2.18
1.16 1.3l*
1.07 1.11
l.Ol* 1.00
l.Ol* 1.00
ultrasonic extraction.
other series.






-------
    technique yielded results with generally better accuracy and precision


    compared to the peak area technique.




D.  Working Range


    The concentrated atmospheric sample extract previously used for


    evaluating the turbidimetric and MTB methods was diluted to varying


    degrees to permit evaluation of relative accuracy and precision


    as a function of concentration.  The results are given in Table Mt.


    The calculated undiluted concentration was taken as the mean of


    five results between 12.5 and U3.75$ dilution, lU8 +_ 1 yg/ml.  This


    compares to 139 ± U.7, 139 +. ^-5, 1^2 +_ U.9 and lUU +_ k.J yg/ml by

                                              •o
    Method 75, MRI-MTB, Method 6l and SulfaVer  IV procedures, respectively.


    The peak height ratio nitrate/sulfate for this series of samples was


    2.0 +_  .2.  Thus some positive-interference from nitrate might be


    expected.  The ratio of observed to expected concentrations are


    plotted against expected concentration in Figure 29.  The results


    indicate relative accuracy within 5% in the range 7 to 111 yg/ml.


    The coefficients of variation of three independent trials are plotted


    against expected concentration in Figure 30.  The C.V. values


    generally decreased with increasing sulfate concentration but


    remained below 5% at >_ 7 yg/ml-




    Based on both relative accuracy and precision, as measured by co-


    efficients of variation, the working range of the Dionex ion


    chromatograph on range 10 ymho is from about 7 to ca. Ill yg/ml.





                              - Ill -

-------
                                                   Table
                 Relative Accuracy and Precision of the Dionex Ion Chromatograph ,  Range 10 ymho
                                                 (pg
Concentration,
% of Undiluted







1
H
H
ro
2.5
5.0
7-5
10.0
12.5
15.0
18.75
31.25
U3.75
56.25
75.0
87.5
100
Observed Mean
 of 3 Trials
                           12
                           U3
     k
     1
    10.99
    lU.31
    18. U9
    22.19
    27-58
    U6.78
    6U. 88
    85.36
   115.63
   1^2.25
   167.13
C.V. %
0.37
0.27
0.30
0.65
O.U7
0.67
0.25
0.9^
1.7k
0.30
0.28
0.80
1.30
8.88
3.61
2.72
U.5U
2.52
3.02
0.91
2.01
2.68
0.35
0.2U
0.56
0.78
  Calculated
Undiluted Cone.

    16U.67
    1U8.67
    1U6.53
    1U3.13
    1U7.89
    1U7.96
    lVf.08
    1U9.71
    1U8.30
    151.75
    151*. 18
    162.57
    167.13
                                               Mean Undiluted = 1U8.2 + 1
     Expected
   Diluted Cone,
                                      3.
                                      7-
         .70
         Al
       11.11
       1U.82
       18.52
       22.23
       27.79
       U6.31
       6U.83
       83.36
                                                                                    129.66
                                                                                    1U8.19
Observed Cone.
Expected Cone.

   1.112
   1.003
   0.989
   0.966
   0.998
   0.998
   0.993
   1.010
   1.001
   1
   1
   1
                                                        ouo
                                                        097
                                                      1.128
                   Trial


                    I
                    II
                    III
           Coefficients  for Linear Regression Working Curves

           Intercept           Slope              r          S
           -0.3037
           -0.0581
           -0.1358
                                                             .x
  O.U06U
  O.U192
  O.U397
     0.9993
     0.9999
     0.9999
1.07U
0.390
0.289

-------
                                               Figure 29


                       RELATIVE ACCURACY AS A FUNCTION OF SULFATE CONCENTRATION

                           WITH THE DIONEX ION CHROMATOGRAPH  (Range  lOpniho)
   2.0
   1.6
CJ
•a
ai

fc

-------
                                              Figure  30


                    COEFFICIENT OF VARIATION AS A FUNCTION OF SULFATE CONCENTRATION
                           WITH THE DIONEX ION  CHROMATOGRAPH (Range Ityimho)
     10
*•*
d
o
 CO
c
Q>
M-l
0)
O
O
       0
40
 8 0             120             160


Expected Concentration  tyig  SOT/20ml)
200
240

-------
VIII.  INTERFERENCE EFFECTS


       A.  Introduction
           Previous AIHL studies have included effects of potential interferents


           singly, in pairs and quartets in relation to a series of sulfate

                   1 2
           methods.  '   Interferents selected for study and their concentrations


           were based, in part, on measurements of atmospheric participates.


           The present study includes four potential interferents:  phosphate,


           colloidal clay (to simulate turbid extracts), water-soluble organics


           contributing absorbance in ^00-500 nm range as isolated from Cali-


           fornia atmospheric samples, and bicarbonate.  The isolation of the


           organics  is described in Appendix G.





           The scheme for the study is shown in Table U5.  Phosphate was studied


           only with MTB procedures based on previous work indicating significant


           interference and high phosphate levels (ca. 2000 yg/8 x 10" filter)

                                                          2
           in early batches of Gelman "EPA Grade" filters.   As with clay,


           concentrations were decreased tenfold for the Colovos MTB.  The levels


           of colloidal clay used were chosen based on prior evaluation of  the


           turbidity of atmospheric extracts prepared from 3A x 8" strips from

                                                                2
           2U-hour hi-vol filter samples collected in St. Louis.   The maximum


           turbidity observed was equivalent to that of a colloidal suspension


           of 1*0 yg/ml bentonite clay or 25 yg/ml kaolinite.  For analysis based


           on use of lA of a 8 x 10" filter the maximum turbidity expected was


           calculated as 65 yg/ml kaolinite.  This calculation was used to set


           concentrations for the Midwest MTB and Method 75-  Because of smaller


           samples anticipated with use of the Colovos method in the 0-10 yg/ml


           sulfate range, concentrations of clay were decreased by a factor of 10.




                                     - 115 -

-------
                                                           Table ^5

                                  Sulfate  and Interferent  Levels for Interference Studies
   Method.


   MR I MTB
   (0-100 (ug/ml S04=)
   Colovos MTB
   (0-10 jug/ml S04~)
(S04~)           (P04=)
Ug/ml             jug/ml
  10        10, 5, 2.5, 1-25
  35        10, 5, 2.5, 1.25
  60        10, 5, 2.5, 1.25
   1.5      l.O, 0.5, 0.25, 0.125
   5        1.0, 0.5, 0.25, 0.125
   8.5      l.O, 0.5, 0.25, 0.125
     Colloidal
        clay
       (iug/ml)


75, 50, 25, 10
75, 50, 25, 10
75, 50, 25, 10


7.5, 5, 2.5, l.O
7-5, 5, 2.5, l.O
7.5, 5, 2.5, l.O
        Organics
     from Atmos. Ext-
   Abs./cm at ^00 nm
.1, .05, .025, .0125
.1, .05, .025, .0125
.1, .05, .025, .0125
.05, .025, .0125, .006
.05, .025, .0125, .006
.05, .025, .0125, .006
                                                                                                                (HC03~)
   Method 75
a\
i
  10
  35
  60
75, 50, 25, 10
75, 50, 25, 10
75, 50, 25, 10
.1,  .05,  .025, .0125
.1,  .05,  .025, .0125
.1,  .05,  .025, .0125
30,15,7.5,3.8
30,15,7.5,3.8
30,15,7.5,3.8
   a.   Using kaolinite clay

   b.   Extracted by the procedure given in Appendix G.

-------
    Similarly, based on the previous study  the maximum level of organics




    expected, as measured by absorbance at UOO nm for extraction of one




    quarter filter is A = 0.07.  The bicarbonate concentrations selected




    were based on preliminary results  which showed a slight positive




    interference with 10 yg/ml bicarbonate for one version of the BaSO^



    turbidimetric method.








B.  Method 75




    The results of the interference study with Method 75 using third order




    regression are shown in Table U6.  While analyses are performed on




    20 ml samples, results have been expressed per ml of solution for




    ease in comparison with other methods.  Colloidal clay showed the




    largest interference.  Its interference was clearly observable at all




    sulfate levels but was largest at low sulfate.  Increasing concen-




    trations of clay produced a decreasing observed sulfate value.




    However, at low levels of sulfate and high clay concentrations results




    became erratic.








    Organics produced a measurable (up to 17$) negative interference at




    the lowest sulfate level.  At higher sulfate values no interference




    was observable.  Bicarbonate did not interfere at any sulfate level.








C.  Midwest Research Ihstitute-MTB Method




    Results with the MRI-MTB method are shown in Table ^7.  At 10 yg/ml




    sulfate, results are shown without added interferent because, using




    linear regression, the method yields low values at this level.  At




    10 ug/ml sulfate no interferences are observable in part because of










                              - 117 -

-------
                                                         Table U6
                            Interference Effect vith  the AIHL Method 75  (yg/ml Observed Sulfate)'
Sulfate Level
UK/ml
Interferent Level b
Inter ferent
Colloidal clay
Organics
bicarbonate (HCOs"")
10
A

8.8
1.6
9.4
0.9
10.2
0.2
B C

5.2 1.6°
2.3
9.9 9.4
0.3 0.3
10.1 10.0
0.7 0.6
D

2.9d
1.0
8.3
0.3
10.4
0.3
35
A

35.3
0.5
35.4
1.9
37.0
0.3
B

33.2
0.3
35.8
0-5
36.1
0.2
C

31.2
0.2
35.2
0.7
36.5
0.5
D

30.0
0.6
35.2
0.4
36.0
0.6
60
A

58.3
0.9
60.0
0.7
60.0
0.3
B

56.2
1.2
60.5
0.5
59.5
0.9
C

53.0
0.7
59.0
1.1
60.2
0.8
D

50.7
0.7
59.1
2.4
60.0
0.6
 I
H
H
co
 I
a.  Mean of three determinations, third order regression, vith +_ 1 o values shown below mean.
b.  Interferent concentrations:                               .
                                                              A      a      0      D
                                Colloidal clay (yg/ml)
                                Organics (Abs/cm at UOO run)
                                bicarbonate (ug/ml)
  10     25     50     75
0.0125  0.025  0.05   0.1
3.8     7.5     15     30
                           c.  Single value.  Two trials yielded negative results.
                           d.  Mean of two results.  One value yielded negative results.

-------
                                                        Table
                                      Interference Effects with the MRI-MTB Method
Sulfate Level
UK/ml
Interferent Level ^ Oc
Interf erent :
Colloidal clay ®'-?
n • 8-°
Organics 2
Phosphate

A

10.1
1-7
9.7
0.8
6.5
0.5

B

7.8
0.7
7.3
1.0
6.7
0.8
10
C

7.6
0.7
7.3
0.7
7.7
o.4

D A

7.7 35.6
0.6 0.7
7.3 33.5
0.7 0.6
9.2 36.0
1.5 0.3

B

35.7
0.6
35.4
0.5
35.7
0.4
35
C

33.1
4.2
35.9
0.4
36.4
0.4

D

36.7
0.7
36.2
0.3
36.1
0.4
60
A

60.7
1.3
61.2
1.2
61.8
1.3
B

62.2
0.4
61.9
0.5
63.0
0.5
C

62.7
1.0
62.5
1.3
62.8
0.9
D

63.2
0.5
63.8
1.1
63.1
1.9
I
H
                         a.  Mean of three determinations, third order regression, with +_ 1 a values shown below mean.

                         ID.  Interferent concentrations:
                                                                                                   B
                                                         Colloidal clay (yg/ml)
                                                         Organics (abs/cm at 400 nm)
                                                         Phosphate (Mg/ml)
10
0.0125
1.25
25
0.025
2.5
50
0.05
5
75
0.1
10
                          c.  Deviations from true sulfate level reflect poor fit of linear regression equation at this level

-------
    the relatively poor precision of the method observed at this




    concentration.  At 35 and 60 yg/ml sulfate, colloidal clay and




    organics exhibit small positive interference.   Interference from




    phosphate at <_ 10 yg/ml was not clearly established at any of the




    three sulfate levels.








D.  Colovos MTB Method (0-10 yg/ml range)




    Results with the Colovos MTB method are given  in Table U8.  At




    1.5 yg/ml sulfate only organics exhibit a measurable interference.




    The positive interference by yellow organics is more easily seen at




    5 and 8.5 yg/ml sulfate.  Colloidal clay shows evidence of positive




    interference at 5 and 8.5 yg/ml sulfate.  No clear evidence of




    phosphate interference is seen.  As expected,  these results are




    similar to those observed with the MRI-MTB method in the 0-100 yg/ml.








E.  Dionex Ion Chromatograph




    1.  Introduction




        Results obtained using EPA audit strips containing both sulfate




        and nitrate, the evaluation of relative accuracy, and the inter-




        method comparison results in Section IX suggest that the ion




        chromatograph sulfate results may be up to 5$ high.  A potential




        source of some of this error is the slight overlap of the nitrate




        and sulfate peaks observed during the current program.  In




        principle, the mutual influence of such ions could be eliminated




        by employing standards prepared with both sulfate and nitrate at




        about the same levels as in the samples.  However, because of the




        variability in the sulfate to nitrate ratio in atmospheric samples,




        this approach appears impractical.







                              - 120 -

-------
                                                      Table U8
                       Interference Effect with the Colovos MTB Method (yg/ml Observed Sulfate)
Sulfate Level
yg/ml
Interferent Level
Interferent
Colloidal clay
Organic s
Phosphate (P0i,.~)

A

1.67
0.16
1.57
0.18
1.31
0.06
1.
B

1.1*6
0.02
1.1*6
0.06
1.68
5C
c

1.1*7
0.00
1.57
0.03
1.23
0.01

D

1.1*6
0.02
1.83
o.o>*
1.23
0.01
5.0 8.5
ABCD ABCD

5.00 5.05 5.11 5.19 8.69 8 .-91 9.17 9.05
0.23 0.12 0.06 0.05 0.21* 0.15 0.07 0.08
5.10 5.17 5-26 5-59 8.72 8.97 9-22 9.M-
0.15 0.13 0.03 0.03 0.35 0.09 0.03 0.04
5.03 U.98 5.70 5.05 9-51 8.91 8.8U 9.03
0.02 0.11 0.60 O.Qi* 0.91* 0.05 0.07 0.09
ro
H
                 a.  Except  as  noted,  mean of three  determinations,   linear  regression with ^ 1 a values
                    shown below mean.
                b.   Interferent  concentrations:
                                                  Colloidal clay  (yg/ml)
                                                  Organics  (Abs/cm at 1*00 nm,
                                                  Phosphate (yg/ml)
                                                                                           B
D
1.0
0.006
0.125
2.0
0.0125
0.25
5.0
0.025
0.50
7.5
0.05
1.0
                 c.  Mean  of  two results.   One  set  of results yielded values which were too high because of a
                    poor  fit of the working  curve  at 1 yg/ml using linear regression.

-------
    The current  study employed a 3 x 500 mm anion separator column




    which was close to the end of its useful life.  Accordingly,




    the extent of nitrate interference with this column was evaluated




    together with that for a new column.  While no evaluation of the




    interference of sulfate in nitrate determination was made, similar




    results would be expected.








2.   Results
    Standards containing from 20 to 80 yg/ml sulfate and U to 80 yg/ml




    nitrate were analyzed and compared to results for sulfate standards




    of equal concentrations without nitrate.  The latter were run in




    succession with the mixed standard of equal sulfate level.  The




    interference was judged by a change in the peak height for sulfate




    compared to that for the sulfate standard.  Since peak heights are




    used for constructing working curves (which have negligible inter-




    cepts) the percentage change in peak heights equals the per-




    centage change in observed concentration.








    Results are given in Table k$ for both the old and new anion




    separator column.  Flow rates were varied from 2.5 to 3.3 ml/min




    and the temperature from 23 to 35°C.  Retention times for sulfate




    and nitrate were markedly lower for the old column.  However, the




    retention time difference between sulfate and nitrate had decreased




    by only 19 seconds (ll$).  At 20 yg/ml, sulfate and nitrate




    interference was not reliably measured.  Results at Uo and 80 yg/ml




    sulfate with equal concentration (by weight) of nitrate, showed




    increased sulfate signal with both the old and new columns.  For
                         - 122 -

-------
                                                    Tattle  h9

                 Interference of Nitrate in Sulfate Determination with the Dionex  Ion  Chromatograph
H
ro
                       Eluent Flow Rate
                         (ml/min)

                            3.26
                            3.26
                            2.50
                              ,26
                              .26
                              ,26
                              ,26
                              ,26
                            2.50
,26
,26
,26
,50
.50
                            2.50
               Column Temp.
35
35
23

35
35
35
35
35
23

35
35
35
35
35
23
                      True Cone,  (yg/ml)
           Column     SO^"          N03~
                                                           20
                                                           20
                                                           20
                                     ho
                                     ho
                                     ho
                                     ho

                                     80
                                     80
                                     80
                                     80
                                     80
                                     80
                                                     Nitrate
                                                  Interference
Old
New
New
Old
Old
Old
Old
New
New
Old
Old
New
New
New
New
20
20
20
ho
ho
ho
1*0
ho
ho
80
80
80
80
80
80
                                                   0.0
                                                   0.0
                                                  +5.7
                                                                       +1.1
                                                                       +3.6
                                                                       +3.2
                                                                       +2.
                                                                       +0.
   • 5
   .7
+1.8

+3.2
+h.o
+3.0
+U.3
+1.7
+1.8
Retention Times (seconds):

               Old column
               New column
                                                                               380
                                                                               657
                                                           NO 3

                                                           220
                                                           478
                                                  At

                                                  160
                                                  179

-------
             the  old column the  increase averaged 3.3 +_ .6% and for the new




             column, 2.2  + 1.355.  We conclude,  therefore,  that a small positive
                         • • *



             bias in sulfate results is unavoidable using  the ion chromatograph




             with the eluent, flow  rates and temperatures  studied.  These




             findings provide at  least  partial  explanation for the high results




             noted in evaluating  the ion chromatograph for sulfate determination.








IX.  INTERMETHOD  COMPARISON WITH  ATMOSPHERIC SAMPLES




     A.  Introduction




         Twenty-four hi-vol filter  samples  collected on Gelman EPA Grade




         filters  were selected for  use  in comparing analytical results among




         the sulfate methods evaluated  in the current study.   These include


                                                  T>

         Method 75, Colovos MTB,  MRT-MTB, SulfaVer  and the Dionex ion chroma-




         tographic methods.  In  addition, the samples were analyzed by the



                                                                  7 9
         AIHL microchemical method  with ion exchange pretreatment.








         The samples analyzed were  obtained from the California Air Resources




         Board,  They were selected from a  set  of 50, with five collected




         at each  of 10 sites throughout California.  Samples were visually




         graded into categories  "light", "medium" and "heavy" loadings.




         Where possible,  samples  from at least  two of these loading categories




         at each  site were used  for this study  (Table 50).  The samples in-




         cluded an equal  number  collected within and outside of the South



         Coast Air Basin.









         One-half of each 8 x 10" filter was extracted by  30 minute ultrasonic




         extraction with  final volume 200 ml.  The filtered, aqueous extract
                                 - 12U -

-------
                                Table 50
  Description of 2k Hour Hi-Vol Filter Samples for Intermethod Comparison8"

                                          Number of Filters
      Site                          Light       Medium       Heavy
                                    ^™^^*^aBt^B^^B~*       ^^^^•^^••^MMHM       *-*4II^MdMlblp-
Sacramento                            11            i
Fresno                                01            2
San Diego                             021
San Francisco                         030
Los Angeles                           002
Long Beach                            002
Ontario                               Oil
Pasadena                              002
Santa Ana                             Oil
San Bernardino                        002
a.  All samples collected between 7/12/77 and 9/22/77 using Gelman EPA
    Grade glass fiber filters.
                                  - 125 -

-------
    was analyzed successively by each method with three determinations



    made on separate days.







B.  Results
    The results, expressed as yg/ml of aqueous extract, are given in



    Table 51.  The undiluted extracts covered the concentration range



    from ca. 10 to 150 yg/ml.  The values shown for Method 75, Dionex

                            TD

    1C, MRI-MTB and SulfaVer  methods were obtained without dilution



    except for samples >_ 85 yg/ml which were diluted by a factor of 2



    before analysis.  Analyses by the Colovos MTB and AIHL microchemical



    method employed samples diluted into the range 1-10 and 1-15 yg/ml,



    respectively.







    The results are compared as ratios, of means, relative to results by



    Method 75, in Table 52.  Average results agreed within 10$ for all



    methods, with all methods showing somewhat higher values, compared



    to Method 75 •  The precision of the methods with atmospheric samples



    is expressed by the median and range of the coefficients of variation.


                                                           R
    All methods show median C.V. values < 6%.  The SulfaVer  method showed



    both the highest ratio of mean and the poorest precision.







    A more detailed comparison for the methods evaluated in this program



    is given in Figure 31.  The results are nearly identical to those



    by comparing ratios of means since intercept values are small.  All



    methods show slightly higher results compared to Method 75, throughout



    the concentration range.  Except at the lowest concentrations there is



    little scatter in the results.





                              - 126 -

-------
                                       Table  51

         Results of  Intermethod  Comparison with Atmospheric Samples  (yg/ml)
Sample
_JD
2B
7D
1C
IE
ItC
U
2A
ItF
IB
10D
2D
3D
3A
9F
7C
8C
3C
6G
10Aa
' 5Ba
6Ba
9Ea
5Aa
8Ea

Method 75
11.5 ± 0.3
13.5 ± 1.1
13.7 + 0.3
16.6 ± 1.0
17.5 ± 1.2
20.6 ± 0.3
21.3 ± 1.3
22.2 ± 0.6
23.2 ± 0.6
28.8 ± 0.8
32.2 ± 0.9
35.8 ± 0.6
38.5 ± 0.8
39.2 + 0.3
51. U ± 0.6
53.3 ± 0.3
5U.6 ± 0.8
67.8 ± 2.2
85.8 ± 0.9
92. U +_ 2.5
102.8 ±1.0
103-5 ± 1-5
107,9 i i.o
150.0 ± 2.U

SulfaVer
13.1 ± 0.5
lU.3 ± 1.3
13-5 ± 1.0
12.3 + 0.3
15.0 ± i.o
22.0 ± 2.0
2U.2 ± 1.3
23.1 ± 1.6
25.2 ± 1.9
29.8 ± 2.7
38.1 ± 1.6
39. U ± 0.9
Ul.6 ± 2.9
hk.2 + 0.9
60.3 + 3.1
61.5 + 2.2
59.1 4. 1.6
71.1 +_ 0.7
102. U +_ 3.6
97.^ +_ 5-1
111.8 + 7-3
110.7 ± 6.5
117.9 ± 6.2
163.0 +_ 2.6

MRI-MTB
11.7 ±0.5
12.5 ± 0.7
10.5 + 1.0
8.0 +_ 0.5
12.9 + 2.0
21.2 + O.U
23.9 ± U.7
23.0 + 0.1
27.0 +_ 1.6
28.7 ± 1.5
36. U +_ 0.1
37.6 +_ 0.5
Mt.8 +_ 0.8
U3.7 i 0.6
53.8 +_ i.U
57-6 + 0.6
60. U +_ 2.2
70.3 + l.l
89.7 ± 0.9
103.1 + 0.9
106.2 + 1.1
111.8 +_ l.O
113.8 + 2.8
153.0 +_ 2.9

Dionex 1C
1^.0 +_ 1.0
15-3 ± 0.8
lU.U +_ 1.0
9-5 ±1.3
15.3 ± 0.8
21.8 ± 1.5
2U.7 ± l.l
25.1 ± 1.3
25-3 ± 1.2
30.9 ± 0.2
36.2 ± 1.6
Uo.O ± 1.5
h3.2 ± 1.7
kl.3 ± I.U
5U.8 ± 2.1
58.7 ± 0.8
6l. U ± 2.U
71.9 ± 0.8
9U.2 ± 0.9
97. U ± l.l
107.8 ± i.U
113.1 ± 0.7
113.1 ± 1.8
16U.O ± 3.1
Colovos-
MTB
lU.9 ± 0.6
15.1 ± 0.8
lU.l ± 1.2
13.7 ± 1.3
15.1 ± o.u
20. U ± 2.0
23.2 ± 0.9
2U.U ± 0.1
2U.9 ± 0.2
29. U ± 0.7
31.8 ± 2.5
37-9 ± 0.6
39-9 ± U.9
U2.0 ± 1.6
5U.1 ± 2.1
55-9 ±1.6
57.0 ± 2.8
68.9 ± 3.1
86.6 ± 1.3
97.5 ± 3.8
105.9 ± i.o
10U.2 ± 3.2
109.1 ± U.2
151.1 ± 5-U
AIHL
Micro
13. U ± 2.7
15.0 ± 1.5
13.7 ± 1.5
10.6 ± 3.8
15. U ± 2.0
21.3 ± 2.U
2U.2 ± 2.5
2U.O ± 2.5
25.3 ± i.o
30.1 ± 1.1
36. U ± 0.8
U0.6 ± 0.1
U3.0 ± O.U
U5.0 ± 0.6
57.6 ± I.U
58.8 ± i.o
59.5 ± O.U
72.2 ± 1.6
92.7 ± 2.0
98.7 ± 0.5
109-7 ± 2.U
109.7 ±2.5
111.0 ± O.U
15U.U ± 1.5
                                                                              _
a.  Samples diluted by factor of two before analysis by Method 75, the SulfaVer ,
   MRI-MTB and Dionex 1C methods.
                                        - 127 -

-------
                                 Table  52

    Average Agreement  and Precision of  Methods with Atmospheric Samples
    Method 75

    SulfaVerR

    Dionex 1C

    MRI-MTB

    Colovos-MTB

    AIHL Micro
Ratio of
1.00
1.09 +
1.08 +
1.05 +.
1.03 +.
1.06 +
Means

.01
.01
.01
.01
.01
Median
C.V. (%}
2.2
5-3
3.9
1.9
3.9
2.3
Range
c.v. (%r
0.6 to 8.1
1.0 to 9.1
0.6 to 13.7
0.3 to 19.7
O.U to 12.3
0.3 to 35.8
a.  Results expressed relative to those for AIHL Method 75-   Errors are
    calculated as the standard deviation of the ratio of two dependent
    variables:
    S.D. (   ) .  ^var.  (    )  and
b.  From three determinations on each of 2k samples.
                                   - 128 -

-------
              16Or- SULFAVJEB = -O.57O+1.10  (TURBID.)
                           r =  0.998
                        Sy.x =  3.00
              120
     SULFAVER  80
      (jig/ml)
               40
                                               160 r- MRI-MTB = -0.759+1.06 (TURBID.)
                                                           r =  0.997
                                                        Sy.x =  3.43
                                               120
                                        MRI-
                                        MTB    80
                                       Oig/ml)
                                               40
8
              160 r-
              120
       DIONEX
         1C    80
               40
      40

 DIONEX-IC
         r
      Sy.x
                                   80
           120
160
        40
80
120
160
-0.022+1.08 (TURBIDJ
 0.998
 2.76
             160
                                              120
                                      COLOVOS
                                        MTB    80
                                      (jug/ml)
                                               40
       I
                                                      I
r-COLOVOS MTB = 0.774+1.01 (TURBID,)
            r = 0.999
         Sy.x = 1.72
      40       80       120      160                      40       80       120

TURBIDIMETRIC, METHOD 75 tyig/ml)                     TURBIDIMETRIC, METHOD 75

                                  Figure  31

SCATTER DIAGRAMS OF RESULTS WITH ATMOSPHERIC  SAMPLE USING FIVE  SULFATE METHODS
                                                                                                        160

-------
                                 References
 1.   B.  R.  Appel,  E.  L.  Kothny, E.  M.  Hoffer and J. J. Wesolowski, Comparison
     of  Wet Chemical  and Instrumental  Methods for Measuring Airborne Sulfate,
     Interim Report.   EPA-600/2-76-059 (1976).

 2.   B.  R.  Appel,  E.  L.  Kothny, E.  M.  Hoffer and J. J. Wesolowski, Comparison
     of  Wet Chemical  and Instrumental  Methods for Measuring Airborne Sulfate,
     Final  Report. EPA-600/7-77-128 (1977).

 3.   Selected Methods for the Measurement of Air Pollutants, Public Health
     Service Publication No.  999-AP-ll (196U).

 it.   Technicon Industrial Method 226-72W.

 5-   E.  M.  Hoffer, E. L. Kothny and B. R. Appel, Simple Method for Microgram
     Amounts of Sulfate in Atmospheric Particulates.  Accepted for publication
     in Atmospheric Environment, 1978.

 6.   E.  M.  Hoffer and E. L. Kothny, A Micromethod for Sulfate in Atmospheric
     Particulate Matter, Air and Industrial Hygiene Laboratory Report No. 163,
     California Department of Health,  July 197^-

 7.   C.  Bros set and M. Ferm, "An Improved Spectrophotometric Method for the
     Determination of Low Sulfate Concentration in Aqueous Solutions", Swedish
     Water and Air Pollution Research Laboratory, ^02, 2U Gothenburg, Sweden
     (197*0.

 8.   Barium Chloranilate Method for Determination of Sulfates in the Atmosphere,
     March 1976.  Prepared for U.S. EPA Environmental Monitoring and Support
     Laboratory, Research Triangle Park, KG.

 9.   Hach Chemical Company.

10.   Tentative Method for the Determination of Sulfates in the Atmosphere
     (Automated Technicon II Methylthymol Blue Procedure) Environmental Protection
     Agency, Quality Assurance Branch, July 15, 1977.

11.   G-  Colovos, et al, Anal. Chem, hQ_ 1693 (1976).

12.   H.  Small,et al,  Anal. Chem. Uj_ 1801 (1975).

13-   R.  T.  Sheen, et al, Ind. Eng.  Chem. (Anal. Ed.) J_ 262 (1935).

lit.   Standard Methods for the Examination of Water and Waste Water, lUth Ed. (1975).

15.   Brochure prepared by the Parr Co., Moline, 111.  (Circa 1900), unpublished.
                                     130

-------
                                                               APPENDIX A

    AIHL Method
               DETERMINATION OF SULFATE IN HIGH VOLUME PARTICULATE
                  SAMPLES:   TURBIDIMETRIC BARIUM SULFATE METHOD*
    Analyte;       Sulfate                       Method No;       61

    Application;   Air Pollution                 Range;          2 to 40 yg  sulfate/m3

    Matrix:        Air                           Lower Detection
                                                  Limit:         50 ug sulfate/20  ml
    Procedure;     Collection on filter by high-                  solution
                  volume  sampler,  extrac-
                  tion with water  followed       Precision:       5 to 11% coefficient
                  by turbidimetric analysis   ,                  of variation

    Date First
      Issued;      December 1974

    Dates Revised;  April   1975,  February  1976 and July 1976
1.   Principle  of  the Method

    1.1  Atmospheric suspended particulate  matter  is  collected  over a  24-hour

        period on  a 20 by 25-cm (8  by 10-inch)  glass fiber  filter by  using  a

        high-volume sampler.   This  method  can  also be used  for samples  collec-

        ted on cellulose  filters.

    1.2  A water  extract of the filter sample is treated with barium chloride

        to form  a  barium  sulfate colloidal suspension.  The turbidity of  the

        suspension is measured spectrophotometrically at  500 nm.

    1.3  The water  extract can also  be used for  analysis of  other water  soluble

        species, e.g. N03~, Cl"~, Na+, NH4+.
   *AIHL modification of Public Health Service Method  (Ref.  1).
    Underlined sentences are changed from the February 1976 revision of Method 6l.
   Prepared by staff of the Air and Industrial Hygiene Laboratory,  State Depart-
   rcnt of Health, Berkeley, Californ^


                                     - 131 -

-------
                                                            No. 61
2.  Range and Lower Detection Limit




    2.1  Aliquoting is adjusted so that samples containing 2 to 40 ug sulfate/ra?




         (the expected range of atmospheric samples) can be measured.




    2.2  The lower detectable concentration of the turbidimetric analytical




         procedure is 50 yg of sulfate in 20 ml solution.




3.  Interferences




    3.1  Measurement is dependent upon the stability of the suspension of col-




         loidal barium sulfate particles, size of particles, sulfate concentra-




         tion, barium ion strength, pH and temperature.  These parameters must




         be closely controlled to avoid obtaining low and inconsistent results.




         Addition of glycerol acts as a stabilizer for the colloid, while alco-




         hol promotes precipitation by reducing the solubility of barium sulfate.




         Use of solid crystals of BaCl2 eliminates the problem of barium ion




         strength and solution aging.  Sulfate concentration is maintained




         within the limits of the method, and pH is controlled by addition of




         dilute HC1.  Variations in temperature between 20 to 30°C do not appear




         to have a significant effect.  Vigorous shaking of the solution during




         the addition of barium chloride promotes the formation of finer par-




         ticles that stay in suspension longer.




    3.2  Sulfur-containing anions are generally strong positive interferents




         probably due to air oxidation to sulfate.  Colloidal clay interferes




         strongly but in a direction apparently dependent on the sulfate/inter-




         ferent ratio.  Table 1 summarizes results of a recent AIHL study (ref. 2).




    3.3  Glass fiber filters are not sulfate free.  The amount of sulfate depends




         on the manufacturer's filter types and lots.  For example, the MSA Type 1106 BH
                                      - 132  -

-------
                                                            No.  61
3.  (cont)




         filters have been found to contain up  to  4 mg  sulfate per  20 by  25-cm




         sheet.  This sulfate can be removed by water wash prior  to sampling  if




         desired.  When this is not practical,  as  is the  case for many monitoring




         purposes, determine the background sulfate concentration (section  7.3.6)




         for every lot and type of sampling filter used and correct the results



         accordingly.




    3.4  Frequently the aqueous extract  containing the  sulfate shows background




         coloration and/or turbidity which may  interfere  with the analysis.



         These interferences are accounted for  by  measuring the absorbance  (A^)




         of the filtrate and glycerol-alcohol-acid mixture before the addition



         of the barium chloride.  This value is subtracted from the absorbance




          (As)  °f "t^6  sample  after the addition of barium chloride.



 4.  Precision and Accuracy



    4.1  The precision of the method was established at AIHL by three determina-



         tions on each of the extracts from 12  high-volume atmospheric samples



         (ref. 2).  Extracts were diluted into  the optimal concentration  range




         of the method.  Under carefully controlled conditions, a coefficient



         of variation of 5% was found.   This compares  to  the value  5% given in




         ASTM method D516 (ref. 3) and to the value 11% reported  in the Public



         Health Service method (ref.  l).




    U.2  The extraction efficiency of the analytical method was recently evalu-



         ated at AIHL utilizing  exhaustive extraction of ten 2U_-hour samples



         collected in Southern California. The  mean recovery was  96.! + 1.2$.
                                      - 133 -

-------
                                                            No.  61
4. (cont)

         Also MSA 1106 BH and Staplex TFA 810 filters were spiked  with known


         amounts of sulfate salts and recoveries close  to 100% were established,


5.  Equipment

    5.1  High-volume sampler.  A motor blower filtration system with a sampling


         head which can accomodate a 20 by 25-cm glass  fiber  filter and capable

         of sampling at an initial flow rate of about 1.7 nrVmin (60 ft-Vmin).

    5.2  Filters.  Use 20 by 25-cm (8 by 10-inch) glass fiber filters of low


         sulfate content.

    5.3  Refluxing Apparatus.  Use 125 ml flask with ground glass  joint, a

         reflux condenser and a hot plate.


    5.4  Filtering Funnel and Whatman No. 42 filter paper.

    5.5  Cuvettes.  Matched 20 mm or 1 inch cuvettes.

    5.6  Pipettes.  Use 20, 5 ml and other sizes as required.


    5.7  Spectrophotometer.  Capable of measuring at 500 nm with a bandpass  of 16 nm
         or less.
    5.8  Filtrate Receivers.  100 ml glass-stoppered graduated cylinders.

6.  Reagents


    All  reagents should be made from ACS analytical-grade chemicals.


    6.1  Hydrochloric acid (10 N) .  Add carefully 80 ml of concentrated hydro-

         chloric acid to 20 ml of distilled water in 100 ml glass  stoppered

         graduate cylinder and dilute to mark with water.


    6.2  Glycerol - alcohol - acid solution.  Mix 20 ml of glycerol with 40  ml

         of reagent grade 95% ethyl alcohol and  15 ml of 10 N  HCl.

         Caution:  Do not use denatured alcohol.

-------
                                                             No.  61
6. (cont)




    6.3  Barium chloride.  BaCl2  •  2 H20  Crystals,  20-30  mesh.   J.  T.  Baker



         Cat. No. 0974 or equivalent.




    6*4  Standard sulfate solution  (100 ug  sulfate/ml).   Dry  anhydrous sodium




         sulfate at 105°C for 4 hours  and cool  in a desiccator.   Dissolve




         0.148 g of the dried anhydrous sodium  sulfate in distilled water  and




         dilute to 1 liter.  This solution  contains 100 pg sulfate  per ml.



7.  Procedure




    7.1  Sampling.  Using the high-volume sampler,  collect the particulate matter




         from approximately 2,000 m3 of air.  Twenty-four hours  is  the usual




         sampling period.  Note and record  the  airflow rates  at  the start  and



         end of the sampling period.






   7.2   Sample Preparation - The sample filter  should be delivered to the



         laboratory unfolded in a glassine envelope.  An aliquot of the




         filter is taken for analysis.  (Appendix 1  discusses sectioning



         the  filter for various analyses).   One-fourth of the filter



         {Quadrant A as shown in Appendix  1) is  cut  into about 5-cm lengths




         for  ease in handling and placed into the 125 ml boiling flask  con-




         taining 50 ml of distilled  water.   Hie  sample is refluxed for 60




         minutes.  The hot solution  is  filtered  through a Whatman No. 1*2




         filter paper which has been previously  rinsed free of sulfate with



         at least 50 ml of boiling distilled water.   The filtrate is collected




         in a 100 ml glass-stoppered graduated cylinder.  Both the boiling flask and
                                      -  135 -

-------
                                                            No. 61
7.  (cont)




         sample filter are rinsed 3 times with about 10 ml each of boiling




         distilled water.  After cooling, the final filtrate volume is brought




         up to 100 ml with distilled water.




    7.3  Analytical Procedure




         7.3.1  Pipette an aliquot of the filtrate, normally 20 ml, containing




                50 to 1000 yg sulfate into a clean dry 25 ml glass stoppered




                graduate cylinder.  When a smaller aliquot is used, dilute to




                20 ml with distilled water.  Add 5 ml of the glycerol-alcohol-




                acid solution and mix.



         7.3.2  Pour a portion of the solution in a dry cuvette.  Determine




                the absorbance at 500 nm against distilled water and record as




                A!.



         7.3.3  Pour the sample solution in the sample cuvette back in the




                cylinder.  Add approximately 0.25 g of barium chloride crystals,




                stopper the flask and shake it vigorously for 45 seconds to



                dissolve the crystals.




         7.3.4  Let the sample stand for exactly 40 minutes at room temperature




                (20° to 30°C).  Then gently mix the suspension by inverting the



                graduated cylinder once and measure the absorbance (A2> of the




                solution against distilled water as above.  When large number




                of samples are to be analyzed, add the BaCl2 crystals at timed



                intervals (e.g. 1 min) with similar intervals between spectro-



                photometer readings.




         7.3.5  Analyze a standard sulfate solution as a control with each batch
                                      '-.136 -

-------
                                                            No. 61
7.  (cont)




                of samples to detect gross variations in the analysis.  Devia-




                tions up to 5% from the standard curve are acceptable.




         7.3.6  A correction for the background concentration of sulfate in the




                filters must be made for each new lot of filters.  This value




                (B) must be the average of at least 5 determinations using 5




                filters from each lot of 100 filters using the entire analytical




                procedure  (sections 7.2 and 7.3) and must be subtracted as



                filter blank (section 9.2).



 8.  Standards and Calibration




    8.1  Prepare standard  from the 100 yg sulfate/ml standard solution



         by pipetting respectively 0, 0.5, 1.0, 2.0, 3.0, 4.0, 6.0, 8.0, and




         10.0 ml into 25 ml glass stoppered graduated cylinders.  Bring the




         volumes to 20 ml  with distilled water.  Analyze the standard solutions




         as in section 7.3, but the determination of AI is unnecessary since




         negligible values are always obtained.



    8.2  Plot the absorbance readings (A2) on the vertical axis versus the corres-




         ponding yg of sulfate on the horizontal axis using a rectilinear graph




         paper.



    8.3  The relation between absorbance and amount of sulfate should be linear




         between 300 and 1000 yg.  The slope of the curve in this  range is cal-




         culated by least  squares and used in determining the amount of sulfate




         in each sample.   Below 300 yg, the relationship is non-linear and the




         sulfate concentration must be determined graphically.
                                       - 137 -

-------
9.  Cc-icula tier.::


    9.1  Air volu.T.0 calculation


         a.  For 3?jrplcc collected  at  altitudes less than 3000 feet a cove mean


             sea level, uce the  cnli'L-rated air flew rate, which is approximately


             equsl to the flow rate under standard conditions of 760 Torr and


             25°C.

         b.  For sar.ples collected  at  altitudes of 2000 feet or greater, cali-


             brate the hirh-volune  sc-npler using the A?3 procedure (U) vhich


             corrects the flow rate to standard sea level conditions.


         c.  Using the flow rate determined in (a) or (b) above, calculate the


             air volune from  the sanplinr tine and the averare of the airflow


             rates at the start  and end of the sanplir.; period.




                   y - £^_Q2 x t
                          2


                   Where:  Qx =  airflow rate at start of sampling period (n3/2iin)


                                 cubic  feet per narrate x O.C2S32 = n3/rdr.


                           Qs =  airflow rate at end of sanpling period  (si2/2iin)


                                 cubic  feet per minute :<0.02832 = m3/nin


                           t  =  sampling period (ain)


                           V  =  sample volune in cubic meters (n3) at standard


                                 conditions


     9.2  Subtract Ai  free. Ao  and, calc^olatin? from the calibration ciarve


          obtained in section  8,  deternir.e the equivalent ^g of sulfate  (C|


          in the aliquot.   Calculate the concentration of sulfate in the 20 by


         25 cm filter sanple  in  Mg/E3  as follows:
                                       - 138 -

-------
                                                             No.  61
9. (cent)
                yg sulfate/m* =  (Fl  * F2  x C)  -  B
                Where:    Fi  »  total ml  of  filtrate _
                                 ml  of filtrate  taken  for  analysis
                                       sample  area  of  filter  sample
                                 sample  area  of  filter quadrant analyzed

                          C   -  yg  sulfate in aliquot of sample taken

                          B   -  yg  sulfate/20 x 25-cm filter blank

                          V   »  air sample volume in m^( determined

                                 as in T 9.1.)

10.  References

    1.  Public Health Service Publication No. 999-AP-ll, Selected Methods for

        the Measurement  of Air Pollutants; "Determination of Sulfate in Atmos-

        pheric Suspended Particulates:   Turbidimetric Barium Sulfate Method",

        pp 1-1, U. S. Dept.  of Health,  Education & Welfare, PHS, Div. of Air

        Pollution, Robert A. Taft Sanitary Engineering  Center, Cincinnati,

        Ohio  45226, May 1965.

    2.  Final Report, EPA Contract  No.  EPA 68-02-1660,  "Comparison  of Wet

        Chemical and Instrumental Methods for Measuring Airborne Sulfate",

        B.R. Appel, E.L. Kothny, E.M. Hoffer and J.J. Wesolowski, February  1976.

    3.  ASTM Method D5 16-68  (Method B) ,  "Standard Methods of Test for Sulfate

        Ion in Water and Waste Water";  1975  Book of ASTM Standards; Part  31;

        Water, ASTM; Philadelphia,  PA.

    4-  Standard Procedure for the Calibration of Hi-Vol Samplers and Plotting of

        Flow Calibration Curves Corrected for Altitude,  California Air Resources

        Board, Sept. 1975,  Sacramento, California.


                                     - 139 -

-------
                                 Appendix 1




                    Cutting of Glass-Fiber High-Volume Filters




1-  Remove the glass-fiber filter from the shipping envelope.




2.  Using a clean cutting tool, preferably stainless steel, cut the filter in



    half.  Then cut one half into two equal Quadrants as shown in Figure 1.
                   cut
                                 B
 t
!0  c


 I
20 cm
                                   25 cm
                                  Figure 1
3.  Use quadrant "A" for the determination of sulfate (AIHL Method 6l ) and



    nitrate (AIHL tfethod 66).




k.  Use quadrant "B" for the determination of lead (AIHL 1-fethod 5^).
5-  Use half "c" for the determination of benzene-soluble organics (AIHL



    Method 67).

-------
                                 Table 1




  The Effect of Potential Interferents in the BaCl2 Turbidimetric Method3
Interferent






Sulfide




Sulfite




persulfate



thiosulfate



bicarbonate




phosphate



silicate



calcium



lead




colloidal clay



p-benzoquinone (a yellow




  chromophore used to simu-



  late a yellow extract)
                                             Sulfate/Interferent Ratio
2:1 (w/w)
                                                            0.66:1 (w/w)
                                              •f



                                              0



                                              0



                                              0
                  •H-




                   0




                   0




                   0




                   0




                   0
a.  + indicates > 10% increase in sulfate



    44- indicates > 25% increase in sulfate




    « indicates > 25% decrease in sulfate




    0 indicates < 10% change in sulfate



b. . sulfate concentration 20 pg/ml








Source:  Ref. 2
                                   - lUl  -

-------
                METHOD 61.  TURBIDIMETRIC SULFATE - ADDENDUM








1.  Limit of Detection




    The method specifies that the lower detection limit is 50  tig per  aliquot.




    This corresponds to a net absorbance of 0.03, i.e., A2 - AJ..  Since A^




    may be as large as 0.07, the chemist and section leader should carefully




    examine any AS value smaller than 0.10 absorbance units, with the thought



    that the net absorbance may be less than the limit of detection.   In any



    case, all A2 - Aj values of 0.03 units or less should be considered as




    being less than 50 yg sulfate.








2.  Range of the Method



    The range of the method is given as 2 to kO jug/m3, equivalent to  200 to



    UOOO^g per aliquot.  Actually, the curve is linear between 300 and'1500




    Mg sulfate, corresponding to absorbances from 0.10 to 0.90.  (Note that




    the standards described in Method 6l do not covor this range.  It should



    be modified to include 1200 and 1UOO ug standards.)  The optimum  reading,




    from an instrumental standpoint, is 0.^0 absorbance units.  The following



    protocol should be adopted:




    a.  Aliquots shall be adjusted to an A2 value of Q.kO.  Samples giving A2




        values less than 0.10 or greater than 0.90 must be rerun, using an



        appropriate aliquot size.




    b.  Standards greater than 1500 wg shall not be run, since they adversely



        affect the linear regression line.




    c.  Chemists shall not record, as a reportable result, A2 - Aj. values > 0.90



        nor < 0.10, unless in the latter case, a 20 ml aliquot was used.
                                    - 1U2 -

-------
3«  Analytical Procedure




    There should be no deviation from the written procedure.  If circumstances



    should require the section leader to temporarily modify the procedure, he



    should instruct the chemist to flag these results and should, by memo, ex-



    plain the circumstances to his group leader, the QA chairman and the data



    handling section.








    This memo should be attached to, and become a part of, the chemist's and



    section leader's copy  of Method 6l.

-------
                                                               APPENDIX B


AIHL Method


            DETERMINATION OF SULFATE IN HIGH VOLUME PABTICULATE
              SAMPLES:  SIMPLIFIED AND IMPROVED TURBIDIMETRIC
                        BARIUM SULFATE METHOD*


Analyte:      Sulfate                    Method Ho:       75

Application:  Air Pollution              Working Range;   1^0 to lUOO yg
                                                          sulfate/20 ml
Matrix:       Air
                                         Detection Limit; 50 yg sulfate/20 ml
Procedure:    Collection on filter by                     solution
              high-volume sampler,
              extraction with water      Precision;       <_ 6% coefficient of
              followed by turbidi-                        variation in working
              metric analysis                             range

Date First                               Accuracy:        Within k%, on  average,
   Issued:     September 1978                              using EPA Audit Strips
 1.   Principle  of the Method

     1.1 Atmospheric suspended particulate matter is collected over a 2^-hour

         period on  a 20 by 25-cm  (8 by 10-inch) filter by using a high-volume

         sampler.

     1.2 A water extract of the filter sample is treated with barium chloride

         to form a  barium sulfate colloidal suspension.  The turbidity of the

         suspension is measured spectrophotometrically at 500 nm.

     1.3 Results are dependent upon the stability of the suspension of colloidal

         barium sulfate, particle size, barium chloride crystal and aggregate

         size, pH and temperature.  Glycerol acts as a stabilizer for the colloid

         while alcohol promotes precipitation by reducing the solubility of

         barium sulfate.

     l.U Barium sulfate formation and turbidity measurement are done in test

         tubes, thereby eliminating all sample transfers.
*The procedure was developed by E. M. Hoffer and is a revised version  of
 AIHL Method 6l.  Evaluation of the procedure is given in Reference  1.

Prepared by staff of the Air and Industrial Hygiene Laboratory,  State  Department
of Health Services, Berkeley, California.

-------
2.  Range and Detection Limit

    2.1  The detection limit concentration of this turbidimetric analytical

         procedure is 50 yg of sulfate in 20 ml solution.  The working range
         is from 1^0 to lUOO yg  sulfate/20 ml solution.
3.  Interferences

    3.1  Variations in temperature "between 20 to 30°C do not appear to have
         a significant effect.

    3.2  Sample coloration and/or turbidity may interfere with the analysis.
         These interferences are minimized by measuring the absorbafcce (AI)
         of the filtrate plus glycerol-alcohol-acid mixture before the addition
         of the barium chloride.  This value is subtracted from the absorbance
         (A2) of the sample after the addition of barium chloride.  In spite
         of this correction, interference may persist.  Colloidal clay, used
         to simulate the source  of turbidity in some samples, is a strong
         negative  interferent.   Yellow, water-soluble organics, isolated from
         atmospheric particulate matter, show a small negative interference
         at low sulfate concentration (Table 1).
    3.3  Sulfur-containing anions are generally strong positive interferents
                                                  o
         probably  due to air oxidation to sulfate  .
    3.U  Glass fiber filters contribute to observed sulfate both from a "blank"
                                                2
         value and by artifact sulfate formation  .   Artifact
         sulfate can be minimized by employing pH  neutral filters  (e.g. quartz

         fiber).   With all filter types, the background or blank sulfate
         concentration should be measured (Section 8.3.3) for every lot and
         type of sampling filter used and the results corrected.

-------
1*.  Precision and Accuracy




    U.I  The precision of the method was established "by three determinations




         on each of the extracts from 2k high-Trolume atmospheric samples




         ranging in concentration from 2hO to 1500 ug sulfate per 20 ml




         solution.  The median coefficient of variation was 2.2% (range 0.6




         to 8.1$).  This compares to the value 5% given in ASTM method D5l6




         and to the value 11$ reported in the Public Health Service method.
    k.2  Accuracy was established by analyzing EPA audit strips (i.e. filter




         strips loaded with known quantities of sulfate).  For solutions in




         the range 300 to 1700 ug/20 ml, the ratio of observed to theoretical




         concentration ranged from 1.00 to 1.05 with mean value l.OU.




5.  Working Range




    5.1  Working range is defined as the sulfate concentration range providing




         approximately constant coefficient of variation and "relative accuracy".




         The latter indicates the accuracy of the method relative to the value




         obtained in the optimal concentration range of the method.  This is




         determined using a pooled, concentrated atmospheric sample extract,



         diluted to varying degrees.




    5.2  This procedure yielded a relative accuracy within J% in the con-




         centration range ikO to 1^00 ug sulfate/20 ml solution with a




         C.V. of <_ 6%.




6.  Equipment




    6.1  High-volume sampler.  A motor blower-filtration system with a sampling




         head which can accomodate a 20 by 25-cm filter and capable of sampling




         at an initial flow rate of about 1.7 m3/min (60 ft3/min).




    6.2  Filters.  20 by 25-cm (8 by 10-inch) filters.




    6.3  Ultra Sonic bath.  For example, Bransonic Model 32.

-------
6.U  Fine glass frit, 30 ml capacity




6.5  Fisher filtrator




6-6  25 x 150 mm Teflon-lined screw capped test tubes.  The tubes must be




     unscratched.  Add a fiduciary mark to permit reproducible positioning




     in the spectrophotometer.  Test tubes are further pre-selected as




     follows:  a) fill all test tubes with distilled water, b) using a




     single beam spectrophotometer, select any tube for an initial




     reference and determine the tube with highest % T, c) label this




     tube and use as blank for all sample determinations, and d) match




     all other test tubes with the blank within 2%.




6.1  Pipettes.  20, 5 ml and other sizes as required.




6.8  Spectrophotometer.  Capable of measuring at 500 nm with a bandpass




     of 20 nm or less, using 1 inch cells.




6.9  Filtrate receivers.  60 or 100 ml plastic bottles with tightly




     fitting caps.




Reagents




Make all reagents from ACS analytical-grade chemicals.




7.1  Hydrochloric acid (10 N). Add carefully 80 ml of concentrated




     hydrochloric acid to 20 ml of distilled water in 100 ml glass




     stoppered graduate cylinder and dilute to mark with water.




7.2  Glycerol-alcohol-acid solution.  Mix 20 ml of glycerol with kO ml




     of reagent grade 95$ ethyl alcohol and 15 ml of 10 N HC1.  This




     solution is stable for up to 6 months.  Caution;  Do not use




     denatured alcohol.




7.3  Barium chloride.  BaCl2-2H20, 20-30 mesh if available.  Reference 1




     details an evaluation of barium chloride mesh and crystal size, both




     of which influence the calibration curve.  In general, use of BaCl2-2H20




     not graded for turbidimetric analysis yields a calibration curve with




     reduced slope and increased scatter.

-------
    T.k  Standard sulfate solution (100 pg sulfate/ml).   Dry anhydrous sodiuti




         sulfate at 105°C for U hours and cool in a desiccator.  Dissolve




         0.1^8 g of the dried sodium sulfate in distilled water and dilute




         to 1 liter.   This solution contains 100 yg sulfate per ml.




8.   Procedure




    8.1  Sampling.   Using the high-volume sampler, collect the particulate




         matter from approximately 2,000 m^ of air.  Twenty-four hours is




         the usual sampling period.  Note and record the airflow rates at




         the start and end of the sampling period.




    8.2  Sample Preparation.  The sample filter should be delivered to the




         laboratory unfolded in a glassine envelope.  Take an aliquot of the




         filter for analysis.  (Appendix 1 discusses sectioning the filter




         for various analyses).  Cut one-fourth of the filter (Quadrant A




         as shown in Appendix l) into about 5-cm lengths for ease in handling.




         Place it in a 50 ml glass stoppered Erlenmeyer  flask containing 50 ml




         of distilled water and then in an ultrasonic bath for 30 minutes.




         A batch of 18 samples may conveniently be handled at once.  Filter




         the solution through a clean, fine porosity fritted glass filter




         (30 ml volume) into a 60 ml polyethylene bottle using a Fisher




         filtrator.  After discarding filter pulp, carefully wash the frit




         free of sulfate with distilled water and suction it dry.  Washings




         are not added to the extract.




    8.3  Analytical Procedure




         8.3.1  Pipet aliquots of the filtrates, normally 20 ml, into a




                series of clean and dry screw capped test tubes.  When a




                smaller aliquot is used, dilute to 20 ml with distilled




                water.  Add 5 ml of the glycerol-alcohol-acid solution and




                mix slowly, inverting a batch of samples 6 times, taking








                                  - 1U8 -

-------
       about 2 seconds to invert, and waiting about 10 seconds




       before re-inverting.  Wait 10 minutes to permit escape of



       bubbles.




8.3.2  Determine the absorbance at 500 run against distilled water



       and record as Aj.




8.3.3  Add one scoop (approx. 0.25 g) of barium chloride crystals




       at one minute intervals to the batch of samples, following




       each addition by mechanical shaking.  After adding BaCl2 to




       the first sample, cap tightly and shake for 60 seconds at




       90 oscillations/minute on an Eberhard (or equivalent) shaker,




       with the tube lying on the plane of oscillation.  While




       shaking the first sample, time the addition of the BaCl2 to




       the second sample and, when shaking of the first is complete,




       immediately shake the second.  Continue in this way for all




       samples.




8.3A  Let the first sample stand at 20 to 30°C for exactly Uo minutes




       from the time of the initial BaCl2 addition.  Then, gently mix




       the suspension by inverting the sample test tube once and




       measure the absorbance (A2) of the solution against distilled




       water as above.  Read each successive sample at 60 second




       intervals to give each sample ^0+^1 minutes standing time.




8.3.5  A correction for the background concentration of sulfate in the




       filters must be made for each new lot of filters.  This value




       (B) must be the average of at least 5 determinations using 5




       filters from each lot of 100 filters using the entire analytical




       procedure (Sections 8.2 and 8.3) and must be subtracted as




       filter blank (Section 10.2).

-------
 9.   Standards  and  Calibration




     9.1  For each  batch of samples,  prepare calibrating solutions using the




          100 yg sulfate/ml standard  solution.   Pipet 0, 1.0, 2.0, U.O, 6.0,




          8.0,  10.0,  12.0 and lU ml into separate test tubes and bring the volumes




          to 20 ml  with distilled vater. Analyze the calibrating solutions as




          in Section  8.3.




     9.2  Plot  the  difference in absorbance  readings  (A2~Aj) on the vertical




          axis  versus the corresponding  pg of sulfate on the horizontal axis




          using a rectilinear graph paper.




          The relation between  absorbance and amount  of sulfate should be




          approximately linear  between 300 and 1000 ug/20 ml.   By restricting




          samples to  this range, linear  regression can be employed.




          For analyses in the range 1^0  to lUOO  yg/20 ml, a  third order




          regression  equation is used of the form y = a + bx +  ex2 + dx3



          where x = yg/20 ml  sulfate  and y = absorbance.




10.   CalcillatiOns




     10.1  Air  Volume Calculation




           a.   For  samples collected  at  altitudes  less than  2000 feet  above




               mean sea level,  use the calibrated  air flow rate, which is




               approximately  equal to the flow rate under standard conditions



               (760 Torr and  25°C).




           b.   For  samples collected  at  altitudes  of  2000 feet  or greater,




               calibrate the  high-volume sampler using the ARB  procedure




               which  corrects the flow rate  to standard sea  level conditions.




           c.   Using  the flow rate determined in  (a)  or (b)  above, calculate the




               air  volume from  the sampling  time and  the average of the airflow




               rates  at  the start and end of the sampling period.







                                   - 150 -

-------
          Where:  QL = airflow rate at start of sampling period (m3/min)



                       cubic feet per minute x 0.02832 = m3/min



                  Q2 = airflow rate at end of sampling period (m3/min)



                       cubic feet per minute x 0.02832 = m3/min



                  t  = sampling period (min)



                  V  = sample volume in cubic meters (m3) at standard



                       conditions



10.2  Subtract AI from A2 and, calculating from the regression equation



      obtained in Section 8, determine the equivalent yg of sulfate (c)



      in the aliquot.  Calculate the concentration of sulfate in the 20



      by 25 cm filter sample, in yg/m3 as follows:



                yg sulfate/m3 = (F* * *~2 x C) - B
      „       „    total ml of filtrate
      Where*  r
                   ml of filtrate taken for analysis
                 _ total sample area of filter sample	

               2   sample area of filter quadrant analyzed
              C  = yg sulfate in aliquot of sample taken





              B  = yg sulfate/20 x 25-cm filter blank





              V  = air sample volume in m3 (determined as in Sec. 10.l)
                             - 151 -

-------
11.  References
     1.  Final Report, EPA Grant No. 805-M7-1 "Improvement and Evaluation of




         Methods for Sulfate Analysis",  B. R. Appel, E. M. Hoffer, M. Haik,




         W. Wehrmeister, E. L. Kothny and J. J. Wesolowski, 1978.









     2.  Final Report, EPA Contract No. EPA 68-02-1660, "Comparison of Wet




         Chemical and Instrumental Methods for Measuring Airborne Sulfate",




         B. R. Appel, E. L. Kothny, E. M. Hoffer and J. J. Wesolowski,




         February 1976.









     3.  ASTM Method D516-68 (Method B), "Standard Methods of Test for Sulfate




         Ion in Water and Waste Water", 1975 Book of ASTM Standards, Part 31,




         Water, ASTM, Philadelphia, PA.









     U.  Public Health Service Publication No. 999-AP-ll, Selected Methods for




         the Measurement of Air Pollutants; "Determination of Sulfate in




         Atmospheric Suspended Particulates:  Turbidimetric Barium Sulfate




         Method", pp. 1-1, U.S. Dept. of Health, Education and Welfare, PHS,




         Div. of Air Pollution, Robert A. Taft Sanitary Engineering Center,




         Cincinnati, Ohio ^5226, May 1965.









     5.  Standard Procedure for the Calibration of Hi-Vol Samplers and




         Plotting of Flow Calibration Curves Corrected for Altitude, California




         Air Resources Board, September 1975, Sacramento, California.
                                   - 152 -

-------
                                Appendix 1




                Cutting of Glass-Fiber High-Volume Filters








1.  Remove the filter from the shipping envelope.




2.  Using a clean cutting tool, preferable stainless steel, cut the filter




    in half.  Then cut one half into two equal quadrants as shown in




    Figure 1.
              cut
\

1

T
20 cm
i
                                ,	25  cm	* I




                                   Figure 1




 3.   Use quadrant "A" for the determination of sulfate.
                                 - 153 -

-------
                              Table  1

 Interference Effect vith, the AIHL Method 75 (yg/ml Observed Sulfate)
Sulfate Level
UE/ml
Inter ferent Level b
Inter ferent
Colloidal clay
Organica
bicarbonate (HC(>3~)
10
A

8.8
1.6
9.4
0.9
10.2
0.2
B C

5.2 1.6C
2.3
9.9 9.4
0.3 0.3
10.1 10.0
0.7 0.6
D

2.9d
1.0
8.3 •
0.3
10.4
0.3
35
A

35.3
0.5
35.4
1.9
37.0
0.3
B

33.2
0.3
35.8
0,5
36.1
0.2
C

31.2
0.2
35.2
0.7
36.5
0.5
D

30.0
0.6
35.2
o.u
36.0
0.6
60
A

58.3
0.9
60.0
0.7
60.0
0.3
B

56.2
1.2
60.5
0.5
59.5
0.9
C

53.0
0.7
59.0
1.1
60.2
0.8
D

50.7
0.7
59.1
2.U
60.0
0.6
a.  Mean of three determinations, third order regression, vith +. 1 ° values shown below mean.
b.  Interferent concentrations:                               »
                                                                     B
                                Colloidal clay (vg/ml)
                                Organics (Abs/cm at UOO nm)
                                bicarbonate (yg/ml)
c.  Single value.  Two trials yielded negative results.
d.  Mean of two results.  One value yielded negative results.
10
0.0125
3.8
25
0.025
7.5
50
0.05
15
75
0.1
30

-------
                                                                  APPENDIX C
                   PROTOCOL FOR THE RUGGEDNESS TEST FOR
           DETERMINATION OF SULFATE BY THE TURBIDIMETRIC METHOD
Introduction

Ten factors were selected to evaluate as variables in conducting the

turbidimetric method.  The factors and the levels of each variable to be

studied are shown in Table C-l.  In all trials, standard quantities of

sulfate are to be added to the samples to permit analysis in the approximately

linear region of the working curve.



The ruggedness test is designed to determine the sensitivity of the method

to variation in each of the factors studied, following AIHL Method 6l.



Experimental Plan

The design is a balanced incomplete block as described by Plackett and

Burman, with 12 runs for the ruggedness test, and three replicates for each

run.  Each run will have 10 factors and a dummy factor, and each factor may

be either of two levels.  The scheme for all runs is shown in Table C-2

with the level of each factor designated by a "+" or "-" as previously

defined in Table C-l.



A standard calibration curve with six points ranging from 300 yg SOit/10 ml -

2000 yg SO^/10 ml, plus 10 ml of water for each point was included for each

run.  Following the design in Tables 1 and 2, half of the runs were read

with a B&L 70 spectrophotometer and a 20 mm cell, and the other half, with

a B&L 20 and 25. U mm cell.


                                 - 155 -

-------
                                     Table  C-l

                     FACTORS FOR EVALUATION IN RUGGEDNESS TEST
A = Level of SO^

B = Level of Addition of S0^=

C = Age of Reagent Mixture

D = Varying Strength of Reagent
    Mixture

E = Time After Adding Reagent
    Mixture Until Mixing

F = Mixing before Absorbance
    Reading lc»d

G = Addition of BaCl2

H = Mode of Shaking, after
    Addition of BaCl2

I = Timing before Second
    Absorbance Reading8-

J = Speetrophotoiaeter, cell length
    and mode of mixing^

K = Dummy
                                           Low (-)

                                         100 jjg/10 ml

                                         300 yg/10 ml

                                         Nev (0-1 mo. )

                                         (U ml + 1 ml of
                                             ethanol)
                                         1 min.
                                         Gentle Mixing6
  High (+)

1000 yg/10 ml

 600 yg/10 ml

Old (2 years)

5 ml


10 min.
Vigorously for
20 sec.
                                           scoop (ca. 0.125 g)     1 scoop (ca. 0.25  g)
                                         Gently for U5 sec.
                                         (l shake/second)

                                         20 min.
                                         B & L 20, 25. U mm
                                         separate cells
Vigorously for
U5 sec.

UO min.
B & L 70, 20 mm
graduated cylinder
a.  Following BaCl2

b.  Sample and reagents mixed either in separate cells used for turbidity readings
    or in graduated cylinders with tranfers to a single 20 mm cell for reading
    turbidity.

c.  The absorbance due to initial turbidity of sample plus the glycerin-alcohol-
    HC1 "mixed" reagent before addition of BaCl2.

d.  After transfer of solution to cuvet, entrapped bubbles can cause appreciable
    error in absorbance.

e.  Slowly add glycerol-alcohol-acid solution minimizing air entrainment.  Mix
    solution by inverting graduated cylinder slowly (about 2 seconds to invert).
    After 10 seconds, invert again.  Repeat for total 6 inversions.
                                       -156-

-------
                                    Table C-2
     5



     6



     7



     8



     9



    10



    11



    12
                         DESIGN  OF  RUGGEDNESS TEST&
Rim/Factor     ^5.£p_E_F_G_HI_JK





     1         +     +      -      +     +     +     ___     +     _




     2         +-      +      +     +     ___     +     _     +




     3         _     +      +      +     ___     +     _     +     +
 a.  Plus  and  minus  indicate levels of variable as given in Table C-l.
                                    - 157 -

-------
Data Evaluation

Since the runs involve the deliberate addition of sulfate in each case, the

result for a given run is considered to be:

           Run Result = Sulfate (observed)/Sulfate (expected)

The effect of variation of a given factor (e.g. A) over the range corre-

sponding to the "+" and "-" levels is determined by subtracting the mean of

all runs with "-" values for A from the mean results of all runs with "+"

values for A:


           Effect of A = Run (1,2,U,5,6,10) - Run (3,7,8 9,11,12)
                                  6                     6

For evaluation, the following steps will be used:

           1.  Rank the effects (highest first)

           2.  Sum the variance (i.e., square of the effect) and calculate

               the % of the variance in the turbidimetric method due to each

               factor.




Quality Assurance

The addition of standard amounts of sulfate to a sample, while permitting use

of the nearly linear portion of the working curve, introduces a potential

error because of imprecision and/or inaccuracy in pipetting the standard

addition.  For example, a 1% error in pipetting 300 ug sulfate leads to an

error of about Q% for a sample initially containing kO yg sulfate.   Accord-

ingly, careful calibration of the volumetric equipment is essential.
                                 - 158 -

-------
The standard addition/10 ml and the addition of the 10 ml of water and the

                                  *
mixed reagent are added by Repipet  , while the level of sulfate of 100 yg/


10 ml and 1000 yg/10 ml are by conventional pipet.  The various concentra-


tions of standard sulfate for the turbidimetric calibration curve are also


pipetted with a volumetric 10 ml pipet.  The standard addition/10 ml was


calibrated by weighing the 10 ml that are expelled by the Repipet.  The


Repipet was  adjusted to dispense 300 or  600 yg 30^"/10 ml with an accuracy


of +  0.5$.
 *A device which attaches to a reagent bottle permitting variable and repro-

  ducible dispensing of calibrated volumes of reagent.
                                  - 159 -

-------
                                Bibliography
1.   Plackett and Burman, "The Design of Optimum Multifactorial Experiments",
    Bionetrika 33., 305 (19^6).


2.   "Efficient Screening of Variables", Stowe and Mayer, Ind & Eng Chem 58.:
    No. 2, 36 (1966).


3.   "Statistical Techniques for Collaborative Tests", published by A.O.A.C.
    (1973).
U.  "Measurement of Atmospheric Sulfates:   Evaluation of the Methylthymol
    Blue Method", EPA-600A-76-015,  March 1976.
                                   - 160 -

-------
                                  GLOSSARY







       observed sulfate sample in experimental run including the level of



       sulfate from a standard or field sample plus the standard addition



       for the run.








      Sn 9  Sr, -3  =  observed sulfate values for 3 replicates for Run n.
       J-l • £- •$  II • J}
S   =  average sulfate values from 3 replicates for Run n.
A   "=  standard addition for run n calibrated by weighing solution at ambient



       temperature.
A  .  A  0  A  _  =  standard addition for three trials for Run n.
 n.l,  n.2,  n.3
A   =  average of standard additions for three trials for Run n.
Ks   =  calculated value for the level of sulfate and standard addition for
  n

        Run n  (factors A plus B in Table 1).
R   =  result for Run n as obtained by  subtracting calculated  sulfate  from
 n


       that observed:



              R   =  S"  - Ks
               n      n     n
                                    - 161 -

-------
                                                               APPENDIX D









                   Protocol for SulfaVer IV Procedure






1.   Determine initial transmittance (Tj) of sample in a 2 cm cuvet at 500 nm




    using a B and L TO Spectrophotometer.






2.   Tap SulfaVer IV pillows to settle contents.   Cut off tops and carefully




    transfer contents simultaneously to a set of 12 samples and/or standards




    contained in 25 x 150 mm Teflon-lined screw cap test tubes.   Simultaneous




    transfer can be facilitated by using a tray in which pillows are clipped




    at intervals corresponding to the space between test tubes.   Inverting




    the tray thereby empties contents of each pillow into a row of test tubes.




    A set of standards is run daily.






3.  Tighten caps on each, mount horizontally on a mechanical shaker which




    oscillates inahorizontal plane (e.g., as supplied by Eberbach Corp.,




    Ann Arbor, Ml) and shake for one minute at 90 oscillations per minute.






U.  Beturn samples to a rack in a vertical position and allow to stand




    for 20 minutes.






5.  Transfer to a 2 cm cuvet and read transmittance (T2).  Reading a set




    of 12 samples and/or standards requires 6 minutes or less.






6.  To reduce analysis time a second set of 12 samples can be started




    during the 20 minute reaction time.
                                 - 162 -

-------
                                                                     APPENDIX E
           TENTATIVE METHOD FOR THE DETERMINATION OF SULFATES
                IN THE ATMOSPHERE (AUTOMATED TECHNICON II
                      METHYLTHYMOL BLUE PROCEDURE)
     A tentative method is one which has been carefully drafted from available
experimental information, reviewed editorially within the Quality Assurance
Branch, and has undergone extensive laboratory evaluation. The method is
still under investigation and therefore, is subject to revision.
             ENVIRONMENTAL MONITORING AND SUPPORT LABORATORY
                  U.S. ENVIRONMENTAL PROTECTION AGENCY
              RESEARCH TRIANGLE PARK, NORTH CAROLINA  27711

                               March 1978
                                  - 163 -

-------
1.  Principle and Applicability

     1.1.   Ambient sulfates are collected by drawing air through a  glass
fiber filter with a high volume pump. The filters are extracted  with water
sonically or by refluxing and the extract is treated with barium chloride
and methylthymol blue (MTB) at a pH of 2.8. After the barium  sulfate pre-
cipitates, the pH is increased to 12.4 and the unreacted barium  forms a
chelate with the MTB. The uncomplexed MTB remaining is then measured colori-
raetrically at 460 nmJL'


     1.2.  The method is applicable to the collection of 24-hr samples in
the field and subsequent analysis in the laboratory.

2.  Ran&e and Sensitivity

     2.1.  The range of the analysis is 3 to 95 fig SC^/ml. With  a 50-ml  ex-
tract from one-twelfth of the high volume filter collected at a  sampling
rate of  1.7 m-Vmin  (60 cfm) for 24 hr, the range of the method is 0.74 to
23.3^tg/m . The lower range may be extended up to 12-fold by  increasing
the portion of the  filter extracted. The upper limit may be increased by
diluting the sample with distilled water.

     2.2.  Using the procedure outlined, a concentration of 3 pig SO^/ml
will produce a scale deflection of 3% of full scale (signal/noise ratio
of 2).

3.  Interferences

     3.1.  Sulfides, sulfites, and phosphates produce a positive interference
which is dependent  on the concentration of sulfate and the interfering ion.

     3.2.  The interferences from cations are eliminated by passing "the sample
through  an ion-exchange column.

4.  Precision and Accuracy

     4.1.  A single laboratory's relative standard deviation  based on the
analyses of duplicate strips taken from several thousand filter  samples
is +  	.

     4.2.  Adequate data for accuracy in the determination are not currently
available.

-------
5»  Apparatus

     5»1»  .Sampling;  Apparatus as specified in "Appendix B - Reference
Method for the Determination of Suspended Particulates in the Atmosphere
(High Volume Method),11 shall be used.1/

     5.2.  Analysis

          5.2.1.  Technicon II Analyzer;  An automated analytical system
must be used for the determination of water soluble -.sulfate by this method.
Alkaline solutions of methylthymol blue decompose on exposure to air. The
method, therefore, cannot be adapted to a manual procedure. The Technicon
II Automated Analyzer System (manufactured by  Technicon Instruments Corp.,
Tarrytown, New York 10591) consists of the following components:

               5.2.1.1.  Technicon Autoanalyzer Sampler IV

               5.2.1.2.  Proportioning Pump III;  Either single speed or
two speed pump for rapid flushing.

               5.2.1.3.  Mixing Coils;  One double 11-loop mixing coil 43 mm
(1.75 in.) long and 30 mm  (1.25 in.) wide. One double 10-loop mixing coil
41 mm (1.63 in.) long and 30 mm (1.25 in.) wide. One 5-loop mixing coil
30 mm (1.25 in.) long and 30 mm (1.25 in.) wide.

               5.2.1.4.  Ion-Exchange Column;  Interfering heavy metals
are removed by the use of an ion-exchange resin. Use Dowex 50W-X8, sodium
form, 300 to 850/im (20-50 mesh).  The resin should be stirred into distilled
water and the fines discarded before they can  settle. The resin should be
soaked before use, at least overnight, and may be stored under distilled
water until used. To pack a column, a small piece of glass wool is inserted
in one end of a piece of plastic  tubing 10 cm  (4 in.) long and 2.3 mm
(0.09 in.) I.D. A rubber pipette  bulb is attached to the end of the tubing
containing the glass wool plug. The other end  of the tubing is placed in
the soaked resin container and the rubber bulb operated until the tubing
is filled with resin. The column  must be free  of trapped air after filling
with resin. The resin column should be replaced after a full day's use.

               5.2.1.5.  Single Channel Colorimeter;  A stable colorimeter
suitable for use at 460 nm with a band width of no greater than 18 nm at
half height.

               5.2.1.6.  Flow Cell;  15-mm path length flow cell.

               5.2.1.7.  Linearizer;  The use  of a linearizer is optional.
The sulfate response does not always conform to deer's Law. If the MTB solu-
tion (6.2.15.) is prepared with the optimum molarity and a barium chloride

                                  - 165 -

-------
ratio a close approximation to Beer's Law is obtained, a linearizer is not
required. However, if a linearizer is available, it may be used.

               5.2.1.8.  Single Channel Recorder;  Strip chart recorder
matched to the analyzer output.

               5.2.1.9.  Modular Digital Printer;  Converts analog signal
from the recorder to digital printout. The use of the printer is convenient
but optional.

               5.2.1.10.  Pump Tubing;  Flow rated tubing of the capacities
shown in Figure 1. Deviations from these flow rates are acceptable only
to the extent that a proper calibration curve and acceptable quality con-
trol checks are obtained. The use of silicone rubber tubing in place of the
standard pump tubing is highly recommended for the MTB and flow cell waste
lines. Pump tubing should be replaced every 21 days used.

          5.2.2.  Volumetric Flasks;  50, 100, 500, 1,000 ml capacity.

          5*2.3.  Pipettes;  2, 3, 4, 5, 6, 7, 8, 10, 15, 20, 25, 50 ml
volumetric; 10 ml graduated in 1/10 ml intervals.

          5.2.4.  Pyrex Glass Wool

          5.2.5.  Plastic Tubing;  10 cm (3.94 in.) and 2.3 mm (0.09 in.)
I.D.

          5.2.6.  Rubber Pipette Bulb

          5.2.7.  Buchner Funnels;  Buchner style 150 ml capacity with fine-
pore fritted glass filter.

          5.2.8.  Vaccum Filtering Apparatus;  Device which permits vacuum
filtering directly into the receiver. This consists of a bell jar with a
top opening, a side  tabulation, and a bottom plate. The Buchner funnel pas-
ses through the top  opening and is sealed to the bell jar with a stopper.
The bell jar should  be tall enough to contain the polyethylene bottles used
for storing the samples. The vacuum connection is made using the side tubula-
tion. The filtering  apparatus is shown in Figure 2.

          5.2.9.  Vacuum Pump;  Any device which can maintain a vacuum of
at least 85 KPa (64  cm of Hg). Mechanical pumps or water aspirators may
be used.

          5.2.10.  Polyethylene Bont??s;  Bottles with a capacity of 60 ml
(2 oz) fitted with polyseal caps.

                                   - 166 -

-------
H
Connector
ir~!
n
22 Turn
•Mixing
Coil

1
f*~—

Debt
Ion Exchange
Column


jbbler
W
5 Turn
Mixing


Connector 1 1
B— 1
Coi
Wash
1
^ 1


\
20 Turn
• Mixing
SCoil

Waste
Waste


. cf

4
r


GRY GRY
BLK BLK
GRN GRN
BLK BLK
GRN GRN
BLK BLK
RED RED
ORN ORN
GRN GRN
Proportioning
Pump
1
(1.00)
« (°'32) S3
* (2-00) w,
* <°'32> 'a

« (°'32) Pec
(0.70)* Me
t
mple Air
jrer
mple
iter
agent Air
thy 1 thymol
Blue
(0.42) Sodium
""" Hydroxide
_ (2.00)*

fFlow Rates,
ml/min

                                                                                          *Silicone Pump Tube
                                           Colorimeter
                                           15 mm Tubular Flow Cell
                                           460 nm
                           Figure 1.  Autoanalyzer Flow Diagram  for  Sulfate  Analysis.

-------
                           •Buchner Funnel with
                            Fine Pore  Fritted Disc
                                     To Vacuum  Source
                                     Polyethylene
                                     Bottle

                                     Bell Jar


                                     Baseplate
Figure 2.   Vacuum Filtering Apparatus,
                -  168 -

-------
          5.2.11.  Glass Bottles  (brown);  500-ral glass bottles with poly-
seal caps*

          5.2.12.  Graduated Cylinder:  10 and 100-ral capacities.

          5.2.13.  pH Meter;   Capable of measuring pH to nearest 0.1 pH
unit over a range of 0 to 14.

     5*3.  Extraction;  The samples may be extracted using either the reflux
or ultrasonic procedure.

          5.3.1.  Ultrasonic Extractions

               5.3.1.1.  Ultrasonic Gleaner;  Of suitable size to process
the required number of samples and at least 7.6 cm (3 in.) deep.

               5.3.1.2.  Glass Bottles (clear);  60-ml (2-oz) glass bottles
with snap polyethylene covers.

          5.3.2.  Reflux Extraction

               5.3.2.1.  Erlenmeyer Flask;  125 ml with 24/40 £ joint.

               5.3.2.2.  Condenser;  Water jacketed, 300 mm length with
24/40 S joints.

               5.3.2.3.  Hot Plate;  Suitable for sample extraction (7.2.1.2. )<

               5.3.2.4.  Teflon® Sleeves;  Sized to fit 524/40 joints.

6.  Reagents

     6.1.  Sampling

          6.1.1.  Filter Media;   Filter media as specified in "Appendix B -
Reference Method for the Determination of Suspended Particulates in the
Atmosphere (High Volume Method)/1 shall be used*!/ At least six randomly
selected filters from each lot should be analyzed for sulfate, and those
lots producing a blank value >1 /Ltg/ml should be rejected.

               6.1.1.1.  Determination of Filter pH;  Cut a 58 cm2 (9 in.2)
section of glass fiber filter with a pizza cutter. Place the filter in a
125-ml Erlenmeyer flask. Add 15 ml of 0.05 M KCl (6.2.2.) and stopper the
flask. Stir with a magnetic stirrer for 10 min at 2=60 rpm. Determine the
pH of the extract.

                                    - 169  -

-------
               Obtain the pH for at least six randomly selected  filters
of a given lot of filters and report the mean and standard deviations.  The
pH value has an effect on accuracy of the collection procedures.  Optimal
pH value of the filter extract is not presently known, but pH information
will be useful for historical purposes. Filters currently used in the National
Air Sampling Network (NASN) have a pH of 9.74 + 0.89.

               6.1.1.2.  Determination of Filter _SO&= Content;   Measure the
sulfate content of at least six randomly selected filters from each  lot
of filters. Cut a 7.6 x 20.3 cm (3/4 x 8 in.) strip from each filter using
a pizza cutter and a template. Follow the procedures for extraction  and
analysis given in either sections 7.2.1.1. or 7.2.1.2. and 9.2.2.  Calculate
the mean and standard deviations in p.g SOA=/in.2. The mean standard  deviations
should not exceed 2.5 and 1.2 jug SO^/in.S respectively. The sulfate content
of filters in current use by NASN have mean and standard deviations  of 2.1
and 1.1 pig S04~~ in.^, respectively.

     6.2  Analysis

          6.2.1.  Sodium Hydroxide;  ACS Reagent Grade.

          6.2.2.  Barium Chloride;  ACS Reagent Grade.

          6.2.3.  Methylthvmol Blue (MTB); 3',3"-bis[S, N-Ms(carboxymethyl)
aminoj methyl thymolsulfone-phthalein pentasodium saltj 96% minimum  by spectro
analysis; Eastman No. 8068 or equivalent. The purity of commercial MTB may
vary considerably. Each lot of MTB must therefore be analyzed following
the procedure described in 8.1.8.

          6.2.4.  Ethanol;  957, U.S.P.

          6.2.5.  Ammonium Chloride;  ACS Reagent Grade.

          6.2.6.  Concentrated Ammonium Hydroxide;  ACS Reagent  Grade, 28
to 307. as NH3.

          6.2.7.  EDTA Tetra  Sodium Salt;  Tetra sodium ethylenediamine
tetraacetate, Technical Grade.

          6.2.8.  Sodium Sulfate;  ACS Reagent Grade, anhydrous.

          6.2.9.  Concentrated Hydrochloric Acid;  ACS Reagent Grade, 36.5
to 38.07. HCl.

          6.2.10.  Distilled  Water;  ACS Reagent Grade, having a specific
conductance of 2/u£2~l cm"1- or less..!/

                                    - 170 -

-------
          6.2.11.  Potassium Chloride;  ACS Reagent Grade.

          6.2.12.  Sodiun Hydroxide Solution (0.1 N);  Dissolve 2.0 g of
sodium hydroxide in distilled water and make to 500 ml in a volumetric flask.

          6.2.13.  Hydrochloric Acid Solution (1.0 N);  Add 8.3 ml of concen-
trated hydrochloric acid to water in a volumetric flask and make to 100 ml.

          6.2.14.  Barium Chloride Solution (0.0086 H):  Dissolve 2.090 g
of barium chloride dihydrate (BaCl2 • 2H20) in' distilled water and make
to 1,000 ml in a volumetric flask.

          6.2.15.  Methylthymol Blue Solution (0.00038 M);  To an amount
equal to 0.164 g of 1007. pure MTB (see section 8.2.8.) in a 500-ml volumetric
flask add successively 25 ml of barium chloride solution, 4 ml of 1.0 N
hydrochloric acid, and make to 500 ml with 95% ethanol. Store in a brown
glass bottle. Reagent prepared one day may be used on the following day
if stoppered and stored in the dark.

          6.2.16.  Buffer oH 10.1;  Dissolve 6.75 g of ammonium chloride
(NH4C1) in 500 ml of distilled water. Add 57 ml of concentrated ammonium
hydroxide (Nt^OH) and dilute to 1,000 ml with distilled water. Adjust the
pH to 10.1 with additional NH^OH.

          6.2.17.  Buffered EDTA  (wash solution)i  Dissolve 40 g of tetrasodium
EDTA in the pH 10.1 buffer solution.

          6.2.18.  Stock Sulfate  Solution (1,000 qg SO/rVml);  Dissolve
1.4787 g of sodium sulfate (^2804), which has been heated at 105°C
for 4 hr and cooled in a dessicator over anhydrous magnesium perchlorate,
and dilute to 1,000 ml with distilled water. Store under refrigeration.

          6.2.19.  Blank Reagent  Color Solution;  To  a. 500-ml volumetric
flask add 4 ml of 1.0 N hydrochloric acid and make to 500 ml with 95%  ethanol.

          6.2.20.  Potassium Chloride  Solution  (0.05  M);  Dissolve 3.7  g
of KCl in 1,000 ml of C02 free distilled water. The pH of this solution
should be 7.0+0.3.

          6.2.21.  Wettine Agent;  A 30% solution of  polyoxyethylene ether
of lauryl alcohol (BRIJ-35) or other suitable non-ionic  wetting  agent  should
be added at the  rate of 0.5 ml/liter to the wash and  sample dilution water
supply.
                                    - 171 -

-------
7.  Procedure                    ,

     7.1  Sampling:  Sampling procedures as specified in "Appendix B - Reference
Method for the Determination of Suspended Particulate in the Atmosphere
(High Volume Method)" shall be used.1/

     7.2  Analysis

          7.2.1.  Sample Extraction;  The filters are removed from the fol-
der, opened flat, and cut into 1.9 cm x 20.3 cm (3/4 in. x 8 in.) strips
using a pizza cutter. The filter should be cut with the particulates face
up.

               7.2.1.1.  Ultrasonic Extraction;  One or more filter strips
are placed in a 60-ml (2-oz) glass bottle. A random 5 to 10% of the filters
should be extracted in duplicate for use as quality control samples.^'  Fifty
milliliters of distilled water is pipetted into each bottle. The bottles are
then closed with snap polyethylene caps. The samples are placed in the ultra-
sonic bath, which should be refilled before each set of extractions with
fresh cold tap water to the level of the liquid in the bottles. The ultra-
sonic bath is operated for 30 min. The extracts are immediately vacuum filtered
using the Buchner funnels and the vacuum filtering apparatus. The samples
are filtered directly into polyethylene bottles. The filters should not be
washed or squeezed, and the filtrates are not diluted. After filtering is
complete, the polyethylene bottles are capped with polyseal caps and stored
upright until analyzed. The samples are stable at room temperature for at
least 2 weeks.

               7.2.1.2.  Reflux Extraction;  One or more filter strips are
folded and placed in a 125-ml Eerlenmeyer flask. Add 35 ml of distilled
water to the flask and connect to a 300-mm water jacketed condenser. Place
the flask condenser assembly on a hot plate and boil gently for 30 min.
Maintain cold water circulation through the condenser while the sample cools
to room temperature. Rinse the walls of the condenser with 5 ml of distilled
water and disconnect the flask. Decant the liquid in the flask directly
into the Buchner funnel of the filtering apparatus and filter into a glass
graduated tube with a 50 ml graduation mark. Rinse the filter in the flask
with a 5-ml portion of distilled water and add the rinse to the funnel.
Squeeze the filter with a glass rod to remove the remaining extract and
collect the filtrate. Repeat the rinse with a second 5-ml portion of distilled
water. Collect the filtrate and dilute to a volume of 50 ml with distilled
water. Transfer the sample to a 60-ml (2-oz) polyethylene bottle and cap
with a polyseal cap. Mix thoroughly. These samples are  stable at room  tempera-
ture for at least 2 weeks.
                                  - 172 -

-------
          7*2.2.  Sample Analysis;  A Technicon II Analyzer is employed
for analysis. A flow diagram and reagent flow rates are shown in Figure 1,
page 5. The absorbance is measured at 460 nm, and a flow cell with a path
length of 15 mm is employed. The sample turntable rate is 40 samples/hr
with a 12-sec wash time. The elapsed time between sample pickup and the
corresponding peak is approximatley 6 min. The instrument should be zeroed
and spanned following the manufacturer's directions.

          The automatic analyzer is operated at the beginning of each day,
after a fresh ion-exchange  column is installed, and prior to the first sample
analysis until a drift-free baseline is obtained. Analysis should be con-
ducted in a laboratory with reasonable temperature control since the method
is moderately affected by temperature.

               7.2.2.1.  Sample and Quality Control Standard Loading:  Fill
the test cups with Camples  and place on the turntable. One quality control
sample, a 50 ^g S04~/ral calibration standard (8.2.3.), and one or two water
blanks are included after every 10 samples. Establish a historical data
base and construct a quality control chart indicating 3(7 lower and upper
control limits for the 50 /ig S04=/ml standard*^/ The distilled water is
included after each 10th sample cup to correct for baseline drift. If the
sample extracts are highly  colored or contain suspended particulate, a blank
must be run of the samples. This may be accomplished by replacing the MTB
solution (6.2.15.) with ethyl alcohol plus acid solution (6.2.19.) and per-
forming the analysis a second time. The sulfate values from the sample
blanks should then be used  in calculating the final sulfate concentration.

               If a linearizer is used (5.2.1.7.), the various standards
used for calibration (8.2.3.) are programmed through the linearizer so that
sulfate concentrations are  printed directly in micrograms per milliliter
by the digital printer.

          7.2.3.  System Maintenance;  After completing the final analysis,
the system should be cleaned with the EDTA solution. With the analyzer operat-
ing, place the MTB line and NaOH line in distilled water for 2 to 3 min.
Remove the ion exchange column and replace it with transmission tubing.
Then transfer the MTB, NaOH lines, and the sample dilution lines to the
EDTA solution container for 10 min. All liquid lines are then finally washed
with distilled water for 15 min before shutting down the analyzer. The sample
line may be conveniently washed during this operation by shutting off the
turntable when the sample probe is in the wash position. All liquid lines
should be left filled with  water after the system has been washed if daily
use is anticipated. If, however, the system is idle for 1 week or more,
all lines must be drained and dried. During analysis a coating will slowly
build up on the flow cell windows which is not completely removed by the
EDTA wash. This buildup is  indicated by a loss in colorimeter sensitivity
and may be removed by washing the cell with 1 N HGl followed by an acetone
and then a water wash.
                                    - 173  -

-------
8.  Calibration

     8.1.  Fa.eh Volume Sampler;  The high volume air sampler shall be cali-
orated as specified in "Appendix B - Reference Method for the Determination
of Suspended Particulates in the Atmosphere (High Volume Method)."Z'

     8.2.  Technicon II Autoanalyzer

          8.2*1.  Flow Rates;  The flow rates in the Autoanalyzer system
should be checked when the system is originally set up. It should also be
checked when any system substitutions are made. Disconnect the specific  .
line as it leaves the pump and insert the line in a 10-ml graduated cylinder.
Operate the pump for 2 min. If the flow rate is in error by more than 5%,
change the pump tubing and recheck the flow.

          The flow rates indicated in Figure 1 page 5 will produce the recom-
mended reagent ratios. Minor variations in flow can be tolerated as long
as they are constant, since they will be corrected for in the calibration
procedure.

          8.2.2.  Colorimeter Wavelength;  The uncomplexed MTB at a pH of
12.4 has a maximum absorbance at 460 nm. The colorimeter wavelength accuracy
should be checked prior to use and annually thereafter. Maximum transmis-
sion of  the filter should occur at 460 + 15 nm.

          8*2.3.  Concentration Standards;  Dilute 50 ml of stock sulfate
solution containing 1,000 jxg S04=/ral to 500 ml with distilled water. This
intermediate  sulfate solution contains 100 ^g S0^~/ml. If a linearizer
is used, pipette 15, 35, 50, 65, 80, and 95 ml of 100 ^g S04.-/ml solution
into separate 100-ml volumetric flasks and dilute to the mark with distilled
water. These  solutions contain 15, 35, 50, 65, 80, and 95 M8 S0^~/ml,
respectively. If a linearizer is not used, pipette 5, 10, 10, 15, 20, 50,
60, and  75 ml of the 100 /ig S04=7ml solution into 100, 100, 50, 50, 100,
100, and 100 ml volumetric flasks and dilute to the mark with distilled
water. The solutions contain 5, 10, 20, 30, 40, 50, 60, and 75 ^tg S04~/ml,
respectively.

          8.2.4.  Autoanalyzer Start Up;  Start up the analyzer and start
reagents flowing through the system. The sample in the flow cell must be
free of bubbles during operation. Refer to manufacturer's instructions for
general operating procedures. Set the sampling rate at 40, the wash time
at 12 sec, the range switch at 100, and the decimal switch at 000.

          Operate the instrument until a stable baseline is obtained. This
normally requires a minimum of 30 rain. The turntable is loaded with dupli-
cate stanc'^rds in the following sequence;  0, 100, 0, followed by low to
high sulfate standards.

-------
               8*2*4.1.  Operation with Linear!zer:  The linearizer switch
is set in the direct mode and the recorder is  zeroed using the baseline
control and water blanks and spanned to 100% full scale using the standard
calibration control and a 100 ^g/ral'standard.  The recorder is rezeroed when
the second set of water blanks reaches steady  state. The linearizer is then
switched to the linear mode. When the 15 jig/ml standard reaches steady state,
adjust the 0 to 20 calibration control until the concentration indicator
reads 15 and activate the printer.  The second  15 /ng/ml standard should validate
the setting by indicating the correct concentration. As the 35 /^tg/ml standard
reaches steady state, adjust the 20 to 40 calibration control until 35 jig/ml
is indicated. Continue to adjust each calibration control using the standard
which lies within its range until all calibration controls have been set.
The calibration controls are locked in position after they have been adjusted.
The adjustment of the linearizer needs to be repeated only when changing
lots of MTB or when a standard varies more than 3 ^g/ml from its true value.

               8.2.4*2.  Operation  Without Linearizer;  The recorder is
zeroed using the baseline control and water blanks and spanned to 100% full
scale using the standard calibration control and the 100 ^g/ml standard.
The recorder is rezeroed when the second set of water blanks reaches steady
state. The remaining standards are  then processed and the recorded values
compared with their true values. The differences should be less than 3 /ng/ml.
Repeat span and zero  adjustments if required. If acceptable readings cannot
be obtained, check analyzer operation, flow rates, and reagent concentrations
to locate problem.

               When operating the automatic analyzer, air bubbles should
not be allowed to enter the ion-exchange column. If air bubbles become trapped,
the ion-exchange column should be replaced with a new column. A broadening
of the colorimeter output with a corresponding loss in peak height usually
indicates a performance decay in the pump tubing. At the first indication
of peak broadening the pump tubing  should be replaced. Pump tubing normally
is good for at least 190 hr of operation.

          8.2.5.  100% Adjustment;  The full range of the recorder is used,
from 0 to 100%.

          8.2.6.  Baseline Adjustment;  The baseline is adjust to  zero  at
the beginning and after every 10th  sample using a distilled water  blank.

          8.2.7.  Calibration Standards;   Standards are run at the beginning
and end of each day. If at the end  of  the day, the  standards vary more  than
3 jug/ml from their true value, the  run is considered invalid  and must be
repeated.
                                    - 175  -

-------
          8.2.8.  Assay of Methylthmol Blue;   The purity  of  commercial  MTB
may vary considerably. It is, therefore, necessary  to determine  the barium
equivalency for each lot of MTB using the following procedure*

         To 0.164 g of MTB in a 500-ml volumetric flask,  add successively
75 ml of 0.0086 M barium chloride  solution  (6.2.14.), 4 ml of 1.0 N hydro-
chloric acid (6.2.13.), and make to 500 ml  with  95% ethanol.

          Pipette 2, 3, 4, 5, 6, 8, 10, 12, 14,  16, 17, 18,  19,  and 20  ml
of 1,000 /^g S04=/ml solution  (6.12.18.) into  separate 100-ml  volumetric
flasks and dilute to the mark with distilled  water. Oiese solutions contain
20,  30, 40, 50, 60, 80, 100,  120,  140, 160, 170, 180, 190, and 200 fjig
S04~/ml, respectively.

          Start up the analyzer using the above  MTB solution in  place of
the  normal MTB  solution. Operate the analyzer until a stable baseline is
obtained.

          Span  the analyzer using  the 200 //g  S04~/ml solution and proceed
to analyze in triplicate all  of the above sulfate solutions.  The analysis
should be conducted with the  linearizer in  the direct mode.  Plot the data,
micrograms S04=/ml versus peak height, on linear graph paper. The plot  will
consist  of three  intersecting straight lines  as  shown in  Figure  3. The  inter-
section  of the  first  and second portions of the  plot (1)  indicates the  sulfate
concentration required to react with the excess  barium. The  intersection
of the  second and  third portions of the plot  (2) indicates the sulfate  concen-
tration  required  to react with the complexed  barium. Determine these two
intersects and  calculate the  purity and amount of MTB to  be  used for reagent
preparation  as  described in 9.1. and 9.2.

9.  Calculations
               [Mg S04/ml(p) - Oug S04/ml(c) + ^ug S04/ml(b))] x 50 x 12
 MS
 where :
     MS SO^/ml, N  =  printout for sample (7.2.2.1.)
     ug S04=/ml/c)  =  color correction (7.2.2.1.)
     jtig S04=/ml/[3)  =  blank correction.  This is the S04= content
                        of the filter reduced to  g/ral for the mean 3/4 in.
                        x 8 in. filter strip (6.1.1*2.).
                                   - 176 -

-------
0   10   20  30   40   50  60   70   80   90  100  110  120  130  140  150  160  170  180   190
                                         SO4=/ml
                        Figure 3 -  Plot  of MTB Assay
                                    - 177 -

-------
                    50  =  correction to toal volume of the sample  (7.21.1).
                    12  =  correction to entire filter SC^ content  (9 -f
                           3/4) (7.2.1.).
                    V  =   total volume of air through filter, nH (8.1.).
     9.1.  Calculation of MTB Purity:
             [>g S04/ml(2) - MS S0£/»l(l)] FL x C x 866.73
% Purity =   -  x  10°
                             W x F2 x 2


where:
        S04~/ml/2}  =  jug/ml reading from second intersection  point 8.2.8
                       /tig/ml reading from first intersection point 8.2.8.
     F-^  =  flow rate in ml/min for sample line (normally 0.32)
     C  =  molar concentration of stock solution (normally 1.041 x 10"*)
     866.74  =  molecular weight of MTB
     W  =  weight of MTB in mg
     F£  —  flow rate in ml/min for MTB line (normally 0.7)
     2  =  value to calculate MTB concentration on basis of equivalents/liter.

     9.2.  Calculation of Weight of MTB;

     Weight of material for analysis =  0.164 x 100	
                                        7, Purity from (9.1.)
                                   - 178 -

-------
10.  References
        Lazrus, A. L., K.  C.  Hill,  and  J.  P.  Lodge,  "A New Colorimetric
        Microdetermination fo Sulfate  Ion." Presented  at  the  Technicon Sym-
        posium, "Automation in Analytical  Chemistry,"  New York,  New York,
        September 8,  1965.

        "Appendix B -  Reference Method  for the Determination  of  Suspended
        Particulates  in  the Atmosphere  (High Volume  Method)," Federal
        Register, 36(84):8191-8194,  April  30,  1971.

        ASTM Standards (Water, Atmospheric Analysis),  Part 23, October 1969
        (p.  225).

        "Guidelines for  Development of  a Quality Assurance Program - Reference
        Method  for  the Determination of Suspended Particulates in the Atmos-
        phere (High Volume Method)," EPA Environmental Monitoring Series,
        EPA-R4-73-028b,  June 1973.

        ASTM Manual  on Quality Control of Methods, Special Technical Pub-
        lication 15-C, January 1951. Obtainable from ASTM, 1916  Race Street,
        Philadelphia, Pennsylvania 19103.
                                    - 179 -

-------
                                                               APPENDIX F


     Protocol for Ion Chromatographic Analyses of Sulfate and Nitrate


1.   Equipment

    1.1  lon-Chromatograph.   Dionex System 10 Ion Chromatograph, Dionex Corp.,

         1228 Titan Way, Sunnyvale, CA  9^087.

    1.2  Varian A-25 Recorder and Spectra-Physics Autolab Minigrator Model

         23000-010.  (Spectra Physics, 2905 Stender Way, Santa Clara, CA 95051)

    1.3  Containers.  To hold extraction solution and subsequent storage of

         samples, U oz. polypropylene containers with plastic screw-caps.

    l.U  Filters (extraction).  0.^5 y disposable filter unit (Swinnex-25

         Filter Unit, Millipore , or equivalent).

    1.5  Syringes.  12 cc disposable syringes, without needle, graduations

         0.2 cc.  (Monoject Sterile Disposable Syringe, Cat.  No.  512S, or

         equivalent)



2.  Reagents

    2.1  Deionized, filtered water.  (D.F. H20 for all reagents and suppressor

         column rinse.)  To minimize introduction of background ions into the

         system, the water should be deionized to a resistance of approximately

         15 megohms, or conductivity of 0.1 to 1.0 micromho/cm.  The water

         should be  free of particles larger than 0.20 ym to avoid the accumula-

         tion of residue or debris in the ion-exchange beads or flow system.

         Fill reservoir labelled "H^O" in the chromatograph.
* When filling reservoirs, avoid air bubbles which may cause pumps to lose
  their prime - see instruction manual for this procedure.


                                 - 180 -

-------
    2.2  Eluant.   Prepare 0.003 M NaHC03 - 0.002U M Wa2C03 solution as

         follows:  In a 2-liter volumetric flask, dissolve 1.008 g NaHC03

         (sodium bicarbonate, MCB, Cat. No. SX320 or equivalent) and

         1.0175 gNa2C03 (sodium carbonate, MCB, Cat. No.  SX395-CB705 or

         equivalent) with deionized filtered water prepared as in 2.1

         above.  Invert gently to dissolve, make to the mark with deionized

         water, mix.  Transfer to the eluant reservoir labelled "EI" or
                                   %
         "E2" in the chromatograph.   Add an additional 2 liters to make

         a total of 1* liters, and mix well.

    2.3  Regenerant.  Prepare 1 N E2SO^ as follows:  Into a 2 liter volumetric

         flask containing approximately 1 liter of deionized filtered water,

         introduce 111 ml of concentrated sulfuric acid,  mix, cool.  Make

         to the mark with deionized water, mix.  Fill reservoir labelled

         "Regenerant" in the chromatograph.  Add an additional 2 liters

         to make a total of U liters, and mix well.

    2.k  Stock Standard Sulfate Solution (1000 yg sulfate/ml).  Dry (NHtt)2SOtt

         (ammonium sulfate, NBS certified) powder at 105°C for U hours, cool

         in a desiccator.  Dissolve 1.376 g of the dried  ammonium sulfate

         in deionized water and dilute to 1 liter.

    2.5  Stock Standard Nitrate Solution (1000 pg nitrate/ml).  Dry KN03

         (potassium nitrate, NBS certified) powder at 105°C for k hours,

         cool in a desiccator.  Dissolve 1.631 g of the dried potassium

         nitrate in deionized filtered water and dilute to 1 liter.
* When filling reservoirs, avoid air bubbles which may cause pumps to lose
  their prime - see instruction manual for this procedure.
                                  - 181  -

-------
   2.6  Nitrate-Sulfate Working Standards.  Prepare working standards of

        0, 5, 10, 20, 1*0	160 pg/ml nitrate and sulfate concentrations.

        To obtain the best accuracy and precision, weigh out the required

        amounts of stock standards into small beakers, and then transfer

        the contents to Class A volumetric flasks.  The samples are then

        bracketed by at least four standards during analysis.  Samples

        of higher concentration (outside the 10 ymho range) are set aside,

        diluted, and analyzed at a later time.
         Working Standard          ml or gms of Nitrate     ml or gms of Sulfate
         yg/ml Hitrate-Sulfate     Stock Std. Added         Stock Std. Added
0
5
10
20
	 160
0
0.50
1.00
2.00
16.00
0
0.50
1.00
2.00
16.00
         In each case,  after addition of both stock standards to a 100 ml

         Class A volumetric flask, add sufficient double distilled water

         to the mark.



3.   Analytical Procedure  (Note:  Read the chromatograph and integrator

    instruction manuals before proceeding).

    3.1  Integrator-Recorder Operation.  If the range of the sample is known

         beforehand (e.g., 0 to 80 yg/ml), it is possible to optimize the

         recorder response by adjusting the integrator attenuation and

         recorder span.   For general operation on the 10 ymho to scale, so

         that  the recorder response corresponds 1 to 1 with the Dionex meter,

         the controls are set as below.

                                - 182 -

-------
     Varian A-25  Recorder




     Span:   1.0 volt.   Adjust  the  recorder  zero and  span so that they




     coincide with the Dionex  meter  readings.




     Chart  Speed:  10  inches per hour, when the 3 x  500 mm anion column



     is used.








     Minigrator Model  23000-010




     The most important parameters are set  as below:




     Peak Width (PW) = 99




     Slope  Sensitivity (SS)  =  99999




     Baseline (BL) =1.0




     Trailing Peak (TP) = 1








     The following parameters  are  inactivated:  MA,  SP, PL, BI, 62, and




     T2. TI is set at a time  which  eliminates early peaks, just prior




     to elution of the nitrate curve.  (Determine by experience)




3.2  Put the toggle switch on  the  front  panel of the chromatograph




     in the LOAD position; using a syringe, inject 2 ml of sample solution




     into the injection port.   Leave the syringe in  place during chroma-




     tography.




3.3  Using  the OFFSET  COARSE or FINE knobs, adjust the indicator needle




     on the SPECIFIC  CONDUCTANCE meter to 0.0.




3.U  Flip the toggle  switch  to the INJECT position,  at the same time




     start  the integrator or strip chart.  After 15-30 seconds, flip




     the toggle switch back  into the LOAD position.




3.5  Record the sample I.D.  on the chart.








                             - 183 -

-------
    3.6   After  the  run is  completed,  rinse  the  sample  loop with  3 ml of



         deionized  filtered water with  the  toggle  switch  still in the



         LOAD position.



    3.T   Inject the next sample  as  described  in Sections  3.2  through 3.6.



    Note:  The  system will use approximately  k  liters  of  eluant  per  day.







U.   Chromatograph Start-Up (Review  lon-Chromatograph Instrument  Manual)



    U.I  Using  the  regulator on  a dry air or  nitrogen  compressed gas cylinder,



         adjust the pressure to  90-100  psi  for  the air actuated  valves  in



         the chromatograph.  Flip the toggle  labelled  AIR, on the front



         panel  of the chromatograph,  to the ON  position.



    U.2  Flip the toggle labelled POWER to  the  ON  position.



    U.3  Place  the  toggle   labelled FLUSH down.


                                   INJECT
    U.U  Place  the  toggle  labelled    0/.     down.


                                     Ei
    U.5  Place  the  toggle  labelled     l     up.

                                   WATER

    k.6  Place  the  toggle  labelled  E2 up if EX  is  empty.



    U.7  Place  the  toggle  labelled  E2 down  if reservoir EX is full.



    H.8  Place  the  toggle  labelled  ANALYT up.



    U.9  Place  the  toggle  labelled  SUPPRESS up.



    U.10 Flip the toggle labelled PUMP  to the ON position.  Adjust the



         vernier dial on the front  pump so  that the flow  rate is appropriate,



         e.g. ,  2.5  ml min. (The  flow  ate may  be checked by disconnecting the



         output tubing from the  suppressor  column  and  placing the tubing in



         a graduated cylinder).  Allow  the  system  to run  for  30  minutes



         before beginning  the integrator calibration,  or  until the baseline



         drift  is reasonably stable.  Check for leaks  in  the  tubing  connections.



         Wear safety glasses when opening the column door.  Check reservoir



         levels.


                                   -  18U -

-------
    U.ll Turn MODE switch to LIN.




    H.12 Turn yMHO FULL SCALE to 10.




    U.13 Set SPECIFIC CONDUCTANCE needle to 5.0 or 10.00 with the OFFSET




         COARSE and FINE knobs.   (Allow sufficient positive baseline to




         account for any negative drift).








5-  Standards and Calibration




    5-1  Inject 3 ml of each of the standards described in paragraph 2.6.



         Record -the reading by automatic integration and by measuring the




         recorder trace (chromatogram)  peak height.




    5.2  Calculate regression lines separately for nitrate and  sulfate




         plotting recorder response versus concentration.








6.  Regeneration of Suppressor Column




    6.1  At the end of each day's run,  the suppressor column required




         regeneration as indicated by a color change in the column  resin




         bed from tan to whitish tan, or by a swift rise in the conductance.




    6.2  On the chromatograph, make the following settings:




         6.2.1  Flip toggle switch labelled PUMPS to the OFF position.  Set




                the switch labelled MODE to ZERO.




         6.2.2  Flip toggle switch labelled SUPPRESS to the down position.




                Check the liquid levels in the regenerant and  "H20" reservoirs.




         6.2.3  Set TIME MIN indicators to 10 on REG side and  50 on RIN side.




         6.2.^  Depress the green button labelled START, the rear pump should




                begin pumping.  Set the vernier on the rear pump to approximately




                90.



         6.2.5  The cycle of regeneration is now automatic.  At the end of




                the cycle, the pump will stop and the suppressor column will




                be tan colored.   The cycle may be stopped prematurely by




                                 - 185  -

-------
                depressing the red colored button labelled RESET.








T.   Chromatograph Shut Dovn




    T.I  Flip the toggle switch labelled PUMP to the OFF position.




    7.2  If the regeneration cycle is in process, and premature termination




         is necessary, depress the red button labelled RESET.




    7.3  Make sure the integrator is off (LINE button has been depressed




         and indicator lights are off).




    7.U  Flip toggle switch labelled POWER to the OFF position.




    7.5  Flip AIR toggle switch to the OFF position.




    7.6  Turn off regulator on compressed air or nitrogen cylinder.




    7-7  Protect the integrator from dust using a plastic cover.




    7.8  If the chromatograph is to be shut down for a long period  of time




         (e.g., 2 months) rinse both columns with distilled water beforehand.
                                 - 186 -

-------
                                                             APPENDIX G






 Protocol for Extraction of Organics from Aqueous Extracts of Atmospheric

     Partieulates for Use in Interference Studies of Sulfate Methods




Objective



To extract the .organics  from aqueous extracts from atmospheric particulate



matter -without co-extraction of sulfate.
                                            X1



Procedure



About 2kOO ml of an aqueous extract obtained from extraction at 85-100°C



of 60 loaded 8 x 10" hi-vol filter samples was filtered through Whatman #2



filter paper.  The filtrate was transferred into three 2 liter beakers and



evaporated on a steam bath to about 0.6 liter (total).  This was acidified



with HC1 and the 0.6 liter extracted with three 100 ml portions of 1:1 v/v



butanol-chloroform.  The aqueous phase was evaporated to near dryness and



extracted with 100 ml of butanol-chloroform.  The UOO ml of organic extract



was washed twice with 100 ml distilled water.  No sulfate could be detected



in the wash water (tested with aminoperimidine).  The organic solvent was



then evaporated to dryness on the steam bath.  The brown, resinous residue



was dissolved in 20 ml  water containing 0.5 ml of k N NHi+OB in a petri dish.



The petri dish was heated for 5 minutes to evaporate excess ammonia.  The



solution was diluted and filtered with washing into a 250 ml volumetric



flask and made up to 250 ml with distilled water.






Result



The pH of the solution  so obtained was 6.5 and the color intensity about 2



at \ = UOO nm.  Comparison of this absorbance with that of the remaining



yellow aqueous solution indicated 6U% extraction of the chromophoric organics.



The colored organics in the remaining solution was hydrophilic at all pH



values and could not be easily extracted or freed of sulfate.





                                    - 187 -

-------
                                   TECHNICAL REPORT DATA
                            (Please read Instructions on the reverse before completing)
        NO.
  EPA  60Q/4-79-D?R
                                                            3. RECIPIENT'S ACCESSIOWNO.
   c. I.E A\O SUBTITLE
 IMPROVEMENT  AND EVALUATION OF METHODS  FOR SULFATE
 ANALYSIS
                                                       5. REPORT DATE
                                                        April  1979
                                                       6. PERFORMING ORGANIZATION CODE
  AUTHOR(S)
                                                           8. PERFORMING ORGANIZATION REPORT NO.
 B.
 E.
R. Appel,
L. Kothy,
E. M. Hoffer, M.  Haik,
and J. J. Wesolowski
W. Wehermeister,
 9, PERFORMING ORGANIZATION NAME AND ADDRESS
 Air and  Industrial Hygiene Laboratory Section
 California  Department of Health
 2151  Berkeley Way
 Berkeley.  California 94704	
                                                       10. PROGRAM ELEMENT NO.

                                                          1AD883
                                                       11. CONTRACT/GRANT NO.
                                                        Grant R-805-447-1
 12. SPONSORING AGENCY NAME AND ADDRESS
 Office  of Research and Development
 Environmental  Monitoring and Support Laboratory
 U.S.  Environmental Protection Agency
 Research Triangle Park, North Carolina 27711
                                                       13. TYPE OF REPORT AND PERIOD COVERED
                                                        Final Report  1977-78	
                                                       14. SPONSORING AGENCY CODE
 15. SUPPLEMENTARY NOTES
 16. ABSTRACT
       A simpler and faster procedure for the manual turbidimetric analysis of sulfate
  has  been developed and evaluated.   This method as well as  a  turbidimetric procedure
  using SulfaVer , automated methyl thymol blue (MTB) procedures  for analysis in the
  0-100 yg/ml and 0-10 yg sulfate/ml  ranges, and the Dionex  Ion  Chromatograph were
  evaluated for accuracy, precision,  working range, interference effects, and degree of
  agreement using atmospheric  samples.   Using EPA sulfate audit  strips, all methods
  showed accuracies within 8%  of  the  accepted value, and coefficients of variation with
  atmospheric samples of <_ 6%.  Colloidal clay and yellow, water soluble organics
  isolated from atmospheric samples caused interference with all  methods.  All the
  methods studied provide reliable analyses for 24-hour hi-vol filter samples.  The
  automated MTB method, modified  for  use in the 0-10 yg/ml range as suggested by Colovos.
  offers excellent potential for  analysis of low volume samples  such as those provided
  by a dichotomous sampler network.
                                KEY WORDS AND DOCUMENT ANALYSIS
                   DESCRIPTORS
                                          b.lDENTIFIERS/OPEN ENDED TERMS  C. COSATI Field/Group
  Air  Pollution and Control

  Environment
  Air Monitoring
                                           Measurement Methods

                                           Sulfates
                                                               68A

                                                               43F
 8. DISTRIBUTION STATEMENT

 RELEASE TO PUBLIC
                                          19. SECURITY CLASS (ThisReport)

                                            IIMPI
                                                           21. NO. OF PAGES

                                                               197
                                               20. SECURITY CLASS (Thispage)
                                                                         22. PRICE
EPA Form 2220-1 (9-73)

-------