EPA-650/2-73-050
December 1973
Environmental Protection Technology Series
Jffi
-'.•'.•
m
I
|
:'•;•:
-------
EPA-650/2-73-050
METHODS FOR
THE RAPED AND ACCURATE
MEASUREMENT
OF NITRATE AND SULFATE
IN ATMOSPHERIC PARTICULATES
by
D. Williams, J. Driscoll,
C. Curtin and R. Hebert
Walden Research Division of Abcor Inc.
359 Allston Street
Cambridge, Massachusetts 02139
Contract No. 68-02-0564
Program Element No. 1AA010
EPA Project Officer: Eva Wittgenstein
Chemistry and Physics Laboratory
National Environmental Research Center
Research Triangle Park, North Carolina 27711
Prepared for
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460
December 1973
-------
This report has been reviewed by the Environmental Protection Agency and
approved for publication. Approval does not signify that the contents
necessarily reflect the views and policies of the Agency, nor does
mention of trade names or commercial products constitute endorsement
or recommendation for use.
ii
-------
TABLE OF CONTENTS
Page
I. SUMMARY AND RECOMMENDATIONS 1
A. PROJECT OBJECTIVES 1
B. SUMMARY 1
1. Nitrate Procedure 1
2. Sulfate Procedure 2
3. General Conclusions 3
C. RECOMMENDATIONS 4
1. Nitrate 4
2. Sulfate 4
II. INTRODUCTION 6
A. BACKGROUND 6
B. NITRATE MEASUREMENT 11
C. SULFATE MEASUREMENT 11
III. PRELIMINARY STUDIES 17
IV. LEACHING STUDIES 23
V. NITRATE ELECTRODE METHOD 27
A. ALTERNATE METHODS OF NITRATE ANALYSIS 27
B. EVALUATION OF INTERFERENCES 28
1. Evaluation of Halide Interferences 28
2. Effect of Selenate on the Nitrate Electrode 33
3. Effect of Vanadate on the Nitrate Electrode 33
4. Removal of Selenium and Vanadium as Interferences with
the Nitrate Electrod?~~ 33
C. STABILITY STUDY OF THE NITRATE ELECTRODE 37
D. NITRATE PROCEDURE DEVELOPMENT 37
E. DEMONSTRATION OF THE NITRATE PROCEDURE 41
1. Calibration Curve Method 41
2. Standard Addition Method 45
F. COLORIMETRIC REFERENCE METHODS 49
G. COMPARISON OF ELECTRODE AND COLORIMETRIC NITRATE METHODS
ON HIGH VOLUME FILTER SAMPLES 49
VI. SULFATE ELECTRODE METHOD 62
A. SULFATE ELECTRODE APPROACHES 62
B. SOLUBLE SULFATE INTERFERENCES 62
C. PRELIMINARY SULFATE STUDIES 64
1. Solvent Studies 64
2. Known Subtraction Study ; 67
3. Comparison to Tltratton Method • 67
Ulalden
-------
TABLE OF CONTENTS
(cont.)
Page
STANDARD SUB™CT<°"
,
F. SULFATE SENSITIVITY BY STANDARD SUBTRACTION 77
I: ^^^^^f^f^i^mtis,, 77
G- GRAN'S PLOT PROCEDURE - - ''
H. GRAN'S PLOT EVALUATION o?
I. SULFATE MEASUREMENT OF HIGH VOLUME FILTERS 87
APPENDIX A - RECOMMENDED PROCEDURE FOR NITRATE ANALYSIS
APPENDIX B - RECOMMENDED PROCEDURE FOR SULFATE ANALYSIS
REFERENCES
•j
-------
I. SUMMARY AND RECOMMENDATIONS
A. PROJECT OBJECTIVES
1. To develop rapid and accurate analytical methods for the determina-
tion of nitrate and sulfate in atmospheric participate samples.
2. To compare the new methods developed with current spectrophoto-
metric methods for nitrate and sulfate.
B. SUMMARY
Ion selective electrodes are used for the measurement of nitrate and
sulfate leached from high volume filter samples. The nitrate method uses a
direct measurement with a nitrate electrode. The sulfate method uses a lead
ion precipitation of sulfate, and a lead ion-selective electrode.
1. Nitrate Procedure
An Orion nitrate electrode is used to measure nitrate leached from
High Volume filter samples. A known amount of silver fluoride is added to the
extract, and a fluoride electrode is used as the reference. The stiver ion
precipitates possible interferences of the nitrate electrode (CT, Br~, I",
VOp SeO.~). It is believed that bromide is the most probable interference of
this electrode in these samples.
In the experimental evaluation of this method, nitrate was leached
2
from 48 cm strips of filters. The filter pulp need not be removed, and as
little as 10 ml of HLO can be used using an ultrasonic bath. However, for com-
parison with colorimetric methods the filter was leached with 25 ml H?0 and the
filter pulp removed. A calibration curve method was found to be preferable to
the standard addition method. Statistical analysis of random nitrate standards
shows that this calibration curve method is accurate to ±10% (C.V.) from
_r o
3 x 10~ molar to 10" molar, and presumably higher. The standard deviation in
this analysis was ±2.0 millivolts. Greater precision should be possible since
other investigators using a similar nitrate electrode have obtained a 0.36% C.V.
with a flow system and temperature control. Since the response of the electrode
lUlalden,
-------
is logarithmic (56 mv/decacie 2 broad concentration range of extracts can be
assayed without dilution. In 15 ml, the range from 3 x 10'5 to 10~3 molar
covers nitrate from 28 ug to 930 yg.
Ten filter strips were analyzed for nitrate by the electrode,
Sawicki and Scaringeili, and phenol disu.lfonlc acid (PDS) methods. Recovery of
added nitrate was also determined by each of these methods. One sample was out-
side the range,of the Sawicki method. One of the remaining samples gave an
electrode value which was 70% high. For the remaining samples we obtained a
good correlation with both methods (r = 0.986 and 0.945). Also, the slopes of
the regression lines were 1.0 ±10%. Standard deviations between the methods
for these samples were 13% (Electrode-Sawicki) and 22% (Electrode-PDS).
2. Sulfate Procedure
A sulfate procedure was developed using a lead precipitation tech-
nique and measuring the excess lead ion with an Orion lead electrode. .Increased
sensitivity was obtained by adding an equal volume of methyl cellosolveRacetate
(MCA) to the High Volume filter extracts. In this mixed solvent the solubility
of PbS04 is lowered several orders of magnitude. Methods of measuring the ex-
cess lead included simple potentiometric titration, determination of equivalency
by extrapolation of Gran's plots, and calculation of excess lead using a standard
subtraction method. Although the other methods appeared promising the standard
subtraction method was found to be most reliable. Lead solutions of relative
concentrations (0.3, 1, 3, 10, etc.) were chosen such that no less than 30% of
the lead should be precipitated. At this point the relative error per mv is 19%.
It decreases rapidly, being 8% mv at 50% precipitation, and 2% mv at 80% pre-
cipitation.
In a study of random order sulfate standards a standard deviation
of 1.5 mv was obtained. In this test, quantities from 0.04 to 4 mg of sulfate
were added in 5 ml of aqueous solution.
Nine filter sections were analyzed for sulfate by the electrode and
barium chloranilate procedures. The methods agreed for seven of the samples
(r = 0.992, y = 0.945x + 0.039) for sulfate from 0.65 to 0.225 mg (5% of extract).
lUlalden,
-------
The percent standard deviation for these samples was 10%. The remaining two
samples were from Birmingham, Alabama. The electrode gave higher values and
we hypothesize that metal ions may have decreased the chloranilate values.
3. General Conclusions
The nitrate method is the most direct method we know of for nitrate
analysis. It is also procedurally very simple. No interferences are anti-
cipated. However, because of the logarithmic electrode response, precision is
achieved with some difficulty (4%/mv). With recently developed electrodes, and
careful handling, this limitation should be unimportant.
The sulfate method measures soluble sulfate only, and is pro-
cedural^ more difficult than the nitrate method. With the procedure presented
in the recommendations section, however, the method would be simpler than
other soluble sulfate procedures. This is especially true if an ion exchange
step is required for the chloranilate procedure. The precision of the sulfate
procedure can be improved by using more lead standards (this study used two per
decade). A good correlation with the chloranilate method was obtained for most
samples. The question of interference in the remaining samples remains to be
resolved.
lUlaldeni
-------
C. RECOMMENDATIONS
1. Nitrate
A larger number of filter sections should be analyzed by
the electrode method, and compared to the two colorimetric methods. A
statistical analysis should be performed to test if the methods are identical,
and also to determine the precision of the electrode method. If the methods
are not identical a study of the interferences of each method should be under-
taken.
The nitrate method can be further simplified by leaching with AgF
solution. This method is presently used with the nitrate monitor under develop-
ment in EPA Contract No. 68-02-0591. Limited comparison of the methods indicates
no differences. A smaller quantity of solution should be used with a specially
constructed solution holder.
Nitrate precision should be improved by careful attention to
electrode upkeep, solution temperature and stirring, Also, newer Orion nitrate
electrodes may improve the precision of the method.
2. Sulfate
a. Determination of Discrepancy of Methods for Samples from
Birmingham, Alabama (S-7, S-9)
This task takes priority because if the electrode method is in
error on these samples, then the method may require major modification to elimi-
nate the interference. (We believe adding an ion exchange step to the chloranilate
procedure will result in equivalent responses for the electrode and colorimetric
assays).
We would recommend the duplicate analysis of a large number of
filter samples by the electrode method, and a comparison to both the chlor-
anilate and thorin procedures.
-------
b. Simplification of the Procedure
The method, as developed in the present program and recommended
in the Appendix, uses an aliquot of filtrate and adds an equal volume of MCA.
If a 50/50 MCA/H20 solution can be used for leaching, then an unmeasured volume
of filtrate can be used for analysis. Furthermore, a smaller volume and there-
fore more concentrated sulfate sample can be obtained in this manner. Calculations
can be made on the mv deflection of a small sample which will indicate whether
more sample or blank should be added. This will allow a decade response from one
lead solution.
c. Titration Method
Although a titration method was not successful in the early
stages of this investigation, with the use of MCA and with proper instrumenta-
tion this method should yield accurate results. For electrode stability we
would recommend adding S04 = to an initial excess of Pb++. With an automatic
titrator (with differentiator, if necessary) the time to reach equivalence could
be correlated to the sulfate level in the sample.
If this concept is valid we would recommend that this procedure
be extended to include continuous titration from a Leap Sampler washed with 50%
MCA in water.
lUIaldeni
-------
II. INTRODUCTION
A. BACKGROUND
Sulfate and nitrate in participate matter are clearly important
constituents since these species represent about 12% and 3% by weight of
the total suspended particulate matter in the ambient air. Furthermore,
the sulfate and nitrate in particulates are presumably the products of
oxidation of sulfur and nitrogen oxides, which are of importance as gas-
eous pollutants. A summary of data from the National Air Sampling Network
(NASN) is given in Table I. Note the small (factor of 2) range in concen-
tration between the arithmetic average for the urban and non-urban values,
and also the standard deviation of values from a particular site ( Tables II a
Since sulfate is to be determined by addition of excess lead, its modest
standard deviation indicates that the analyst can usually use the correct
lead concentration the first time.
The methods in use by NASN for NO" and S0= involves aqueous ex-
traction of lot Of the high volume filter and subsequent wet chenical
analysis. ,he technique for sulfate is the methyl thymol blue method and
a Technicon auto-analyzer is required. The nitrate method involves reduc-
tion to N02, and then determination by a colorimetric diazotization
(Saltzman) method. Both methods are time consuming and involve addition
of a number of reagents and require considerable calculations. Hence, we
have attempted to develop a more rapid and simplified procedure These
methods employ ion selective electrodes. The method is direct for nitrate
(nitrate electrode) and indirect for sulfate (lead electrode). The direct
measurement of nitrate is expected to have minimum interferences. The sul-
fate measurement by lead sulfate precipitation is similar in concept to
the barium precipitation methods now in use, and will have the same gen-
eral interferences. In these cases only soluble Sulfate is measured.
In order to determine the required sensitivity needed for an
analytical method, we can calculate the volume of air sampled with a 24-
hour high volume sampler. If we assume a sampling rate of 566 liter/min.for 24
hours, we obtain a total volume of 815 m3 (see Table IV). For a non-urban
lutuai
-------
TABLE I
SUMMARY OF DATA FROM NASN AND CONTRIBUTING
STATE AND LOCAL NETWORKS
(Values in ug/m3)
Pollutant
Suspended Parti culates
Nitrates
Sul fates
Nitrogen Dioxide
Sulfur Dioxide
No. of
Stations
217
132
132
47
45
Urban
Arith.
Average
102
2.9
10.7
141
62
Non-Urban
Max. Station
Average
254
13.5
28.8
333
346
No. of
Stations
30
29
29
Arith.
Average
38
1.3
5.6
Max. Station
Average
79
2.5
12.6
-------
3
TABLE II
SULFATES, NON-URBAN FREQUENCY DISTRIBUTIONS
I ; . .r '
r \l ,', V.
AR1 <"<• ".•>
iifjAf; i CANYON PK
C»l "V' '] A
CCl i" '• 0
"U-MJ UlHA CnUNTY
;.'f !'"•'< r
'-ENT C.-.IJNTY
* i U> I - <•
it'- !'
i.U'lt COUNTY
PA«n.t COUNTY
utLA»ARE COUNTY
•4 j 1 r l
ft ( A u i A NAT PARK
M » ' ' > I f v [)
IAI.VI «T COUNTY
JA(,',tON COUNTY
iitC'ir-UN COUNTY
UOI.*CIF
*9
63
64
63
63
63
e»
63
69
63
.65
63
65
63
69
6»
»5
65
65
63
24
^
23
23
21
21
29
22
26
29
26
23
26
2)
29
29
0
26
2*
29
26
M.n
1.2
4.0
2 3
1.1
l.C
1 a
•>,<>
2.0
,7
3.6
».<
1.6
1.3
1.7
3,9
:n
1 .4
4.9
2.2
2.3
2.0
3.8
2.4
Z.7
1.9
1.9
1.8
4.3
.',<
1.7
6.3
3.2
2.6
6. a
4.3
"'*
2.1
2.9
4.8
3.0
.0
l.e
T.O
4 , f*
3.0
6.6
6.8
2.3
2. a
6.8
ft. 2
in
1.9
T.T
4.T
3.0
8.J
7.3
6,6
2.6
3.0
7.6
6. A
f
PI'.I- hm •
'«
1*9
B.y
3.1
4.9
9.4
9.4
9.6
3.2
3.1
8.3
7.1
,,qf.,,m
I'lTll-
"!
9.6
6.2
6.2
12.0
3.3
3,6
12. U
10. i
3.7
3./
9.1
10. B
V 1 full '
i
;,-,
10.2
6.7
7,3
14.1
6,0
3,7
6,Z
13, J
M.I
4.0
4.0
U. 1
12.0
Mi-l. i
I'-
ll.!
T.O
12, U
16,1
'..9
4.U
U.7
*.•
4.3
13.1
M.4
''0
l».l
U.I
19.7
7.7
4,5
4,3
12,0
18.2
16.7
3.3
8.0
16,7
16,7
KJ«
25, '
14,1
10,1
16. «
28. e
10. C
3.1
'*'
17.1
2* 1
21, n
17.7
10.3
19.2
90,0
Mh
MW
9.5
8,0
6,T
U.fr
S.J
3.}
S.«
6.1
10.0
9.2
»,,
i.a
9.0
12tt
(•TO
Mf^^n
8.7
4.7
3.J
10,4
3.0
3.4
3.6
9.2
a.'
r.B
3,4
3.4
7,8
9.1
su
fl-1)
l.'l
1.68
2.0'
l.fti
, >n
1,3"
1.36
1.T4
!i & 1
1,6«
1.76
1.92
1.03
1,62
1,78
'•22
-------
TABLE in
NITRATES, NON-URBAN FREQUENCY DISTRIBUTIONS
Iot.ii>.
o- S' .' >«
•q| n. 'I"
liHAflU t»NYON PK
<'.AL !,r (,"'(! I *
MJ"a.JLDT COUNTy
( Ot. o't'.i u
'"CfMttUHA COUNTY
9SL»i"Ai«(
"^fT COJNTY
ruCi>t'.:»
'LUWlJ* KEYJ
TU'Ht.i
BUlK CCUNT7
•Nl'I'M
Mfc«llE COUNTY
•Or*
ULLAwARE COUNTy
"Al'-t-
AtAQIA NAT PARK
"APyl.»"|i
CULVERT COUNTY
'ISIIiSjPPl
J«tH.2GN COUNTY
Htl-N-VM
SHANNON COUNTY
MQflT'N*
tjL*CIEI» NAT PARK
NEL'n'SKA
iHUHAS COUNTY
•'IVAIJA
"HUE PINE CO
*!EW MMpjHIfif
coos COUNTY
SI*- "EMCO
"1" *RR1BA COUNTY
•(i« YO«K
CAPE VINCENT
NQBTH CAROLINA
C'^t MAlTtKAS
OKI ANQMA
c-tROKtt COUNTY
onf CON
IIJ«RY COUNTY
Bf"NSYUVAN|A
CLARION COUNTY
^001 ISL'NO
"•»»HIN6TON CO
Vr,i's
»»
45
«)
»3
IV*
6)
ftj
4»
»J
AS
69
63
(.3
OS
«»
69
69
69
• 3
63
6»
69
69
Nii"ihci
of
S,vn|ik'S
24
2*
2)
It
21
23
Z»
2»
22
2*
29
2ft
21
16
21
23
29
Z9
25
26
2*
29
26
HMMMM^V
M ircqi.i'.'s 'i'i fubit Mrl'''
Mi'i
.1
.1
,2
.2
.1
• 6
.1
• 2
••>
.1
.1
.3
.T
1 ffliiriii i PAiiitiiiiion IVr
.4
.»
.9
.4
1.*
l.l
.1
t
.»
.»
.T
.2
1.1
1.4
^HWOH
n
.9
.4
1.2
.6
1.1
1.4
.1
.»
• 6
i.l
.8
.1
1.4
l.T
''I1
.6
.9
l.s
• •
1.4
l.«
l.«
.1
.*
.7
1.4
.8
.1
l.»
l.V
61
.6
.»
1.9
*9
2.*
I.'
.2
.1
• '
».T
.9
.1
l.«
«.o
,« M—.^— i
11
.9
.8
2.2
1.0
I.I
2.1
.2
.4
1.0
1.9
i*9
1.0
.4
2.0
2.9
K^H*«M
«n
.9
1.0
2.6
1.6
».»
'.»
.2
1.4
3.1
••»
Itl
.4
2.T
».T
KHB^BIH
«n
UZ
f*4
1.0
1.9
Z.8
1.0
.1
2.1
4.4
4.4
l.T
.9
l.T
4.2
•^MV^BI
Hti
».»
l.t
S.T
2.»
».l
>.'
.t
1.6
*.»
».•
».*
.1
4.1
».»
i^ ^
l/'th
hVan
.7
.T
• 4
U7
• 9
1.1
l.»
.2
1.1
.}
2.9
l.t
1.0
.1
1.1
J.s
fr«
Mc*i
.6
.1
.6
.3
1.1
«T
1.6
1.6
.2
.»
.8
.9
1.*
1.9
• 9
.3
1.9
2.1
Sid
On
OPV
1,86
1.92
1.9*
2.02
!.««
!.»'
2.11
2.12
1.T9
!.»»
1.6»
l.TO
!.«»
2. or
l.Tl
2. IB
l.Ow
l.«T
l.?l
Z,04
l.«4
-------
10
TABLE IV
REQUIRED SENSITIVITIES FOR SO" AND NO
*
3
Total Volume (m3) = 20 ft /mln x 60 mln/hr x 24 hrs x 28 3 I/ft3 3
! =815m
The weight (total) of NO" and S0= is as follows:
NO" = 1.3 yg/m3 x 815 m3 = 1.06 mg or 0.106 mg for 10% of sample
SO" - 5.6 yg/m3 x 815 m3 = 4.56 mg or 0.456 mg for 10% of sample
If the sample (above) is dissolved in 10 ml of solution, the concentration
is:
= 1.4 x 10"4 moles/liter
SO" = 5.8 x 10"4 moles/liter
-------
11
- q
atmosphere, we take average N03 and SO^ values of 1.3 and 5,6 yg/m , respec-
tively. The weights of SOT values and NCU are given for both the total fil-
ter sample and 10% of the sample. If we assume 10% of the sample is taken
and dissolved in 10 ml of solution, this gives us a measure of the sensitiv-
ity required for the average sample. The detection limit should probably be
1/10 of these values.
-1 -5
The range for the nitrate electrode is 10 to 10 M and for the
lead electrode is 10 to 10 M according to the manufacturer's literature.
B. NITRATE MEASUREMENT
The nitrate electrode is linear over > four orders of magnitude.
Thus, any concentration range desired can be measured without dilution of
the sample. A list of interferences for the Orion liquid ion exchange elec-
trode is given in Table V. For this electrode (the k value for NO^ = 1.0).
CIO* gives 1000 times the nitrate response for an equal concentration.
Perchlorate, I", S=, Br", CIOZ are not expected to be significant interfer-
ences since these species are not present at high concentrations in the at-
mosphere. Chloride ion which can be present in high concentrations in sea
coast areas has a selectivity constant of 4 x 10"3. Thus, with an equimolar
mixture of nitrate and chloride, the potential (mv value) due to chloride
would be 0.4% that of nitrate. We see that the effect of chloride is quite
small. Similar considerations apply to the other known possible interfer-
ing anions.
C. SULFATE MEASUREMENT
Since there is no ion selective electrode commercially available
for sulfate, our effort has been placed on using the lead electrode and
measuring the excess lead after PbS04 precipitation. This procedure has
been described in Orion literature (Orion 94-82A). The sample, in 50%
methanol, is titrated with lead perchlorate, and the electrode potential
is monitored throughout the titration with a millivolt meter. The end-
point is determined by plotting ml of titrant added versus the electrode
potential. The major difficulty with the potentiometrie titration is that
nearly 20-25 points are needed to obtain an accurate curve.
IttUleni
-------
Interfering
cio-
r
C103
Br"
HS"
NO;;
CN"
HCO-
12
TABLE V
ANIONIC INTERFERENCES AND THEIR
COEFFICIENTS FOR THE NITRATE
Approximate Anion Selectivity
SELECTIVITY
ELECTRODE
Constants
I°ns KX Interfering Ions
103
20
2
1.3 x 10"1
4 x 10"2
4 x 10"2
1 x 10"2
9 x 10"3
cr
OAc"
C03
P0l
F"
H2PO-
soj
HPO=
Kx
4 x 10"3
4 x 10"4
2 x 10"4
1 x 10"4
6 x TO"5
5 x 10"5
3 x 10"5
3 x 10"5
-------
13
An alternate method of assimilating data is the use of Gran's plots,
These plots involve plotting ml of titrant vs potential (mv) on semi -anti log
paper which is designed to accommodate the volume change which occurs during
the titration. The basis for this selection of axes is Nernst's equation:
E = E° + BI log [A']
Rearrangement yields:
nFE nFE°
- = -
The ion sensed can be made a linear function of the electrode potential by
taking the antilog of both sides:
antilog (E/S) = E] + [A"]
where S = nF/RT and E] = antilog (E°/S).
The change in volume is corrected for by shifting the axis up 10% from left
to right making additions of up to 10% of the original sample volume pos-
sible. Gran's plot paper has a slope of 58 mv for a monovalent electrode
and 29 mv for a divalent electrode. Corrections for electrodes whose
slopes differ from 58 mv can be achieved by the use of a blank.
There are two major advantages to the use of Gran's plot paper.
Since the plot is linear, fewer points (4-6) are needed and the time per
determination is reduced considerably. A straight line is drawn through
the points and the endpoint is found by extrapolation to the x axis. Sec-
ondly, this method is well suited for small amounts of sulfate, since
points well beyond the endpoint can be used to obtain the straight line
and hence the endpoint, thus eliminating the difficulties encountered
with electrode instability.
A third technique involves addition of sulfate to an excess of lead
ifld determination of the difference in potential (mv). The reduction in
potential between the initial lead solution and the solution containing
Ulalden
-------
14
sulfate is a direct measure of the sulfate content. The advantage of this
technique is the rapidity of the determination. It also has the advantage
that a direct reading sulfate meter could be constructed.
The interferences in the lead electrode are given in Table VI
below.
TABLE VI
INTERFERENCES IN THE SOLID STATE LEAD ELECTRODE
Must be absent
Cdl !f0 1nterference unless the concentration is
Lcu greater than that of lead
The "interference" for the lead electrode involves a poisoning of the elec-
trode surface. PbS is the active matrix of the electrode and has a K of
10 . Any element which as a K$p < lO'26, e.g., CuS, HgS, AgS, will5"
"poison" the electrode. Thus, these metals must be absent from the solu-
tion. CdS and Fe?S3 have about the same K as PbS. Therefore, Cd and Fe
can be tolerated in high concentrations. Sulfide ion is an example of an
anion which poisons the Pb electrode. We have, however, used the lead
electrode in solutions from a kraft mill which had high concentrations of
S . Good results were obtained, but the electrode had to be cleaned
frequently.
In order to determine the applicability of the electrode for sul-
fate determination in atmospheric Hi Vol samples, the interfering metals
which might be present must be considered further. In Table VII we have
taken data from NASN (1968) for New York City which is one of the most ex-
treme cases, and calculated the average concentrations of various "problem
metals" and compared them to the average sulfate levels present in the
city. Silver is not present in the ambient air and presents no potential
problem. Little, if any, mercury is collected on the Hi Vol filter
-------
15
TABLE VII
COMPARISON OF SULFATE AND METAL ION CONCENTRATIONS
FOR NEW YORK CITY
Ag
Hg
Cu
Pb
Fe
Cd
S0=
yg/m3 (N.Y.C.)
NA
NA
0.3
1.6
2.3
0.069
30
Cone. (M/l )*
t
< 4 x 10"bf
< 6 x 10"5f
< 4 x 10"4f
< 6 x 10"4
3 x 10"3M
4
0.013
0.020
0.10
0.002
_ _— _
Assume Hi Vol sample 815 m
10% of sample taken and dissolved in 10 ml
''"Note the assumption is made that all of the metal salts are_
soluble; this is not true but these concentrations represent
the maximum attainable values.
-------
16
(Thompson, 1972). The concentration of copper, however, is quite signifi-
cant, <1% of the S0= level. It is unlikely that much of the copper dis-
solves in the aqueous solution, however, since most of the copper is pres-
ent as CuO which has a low solubility in aqueous solutions. However, even
a small amount of Cu could be a problem over the long term. If the elec-
trode must be polished too often, a copper complex forming agent could be
added to the solution. Amine complexes of copper should remove this pos-
sible interference, if this is necessary.
-------
17
III. PRELIMINARY STUDIES
Some preliminary studies were performed on particulate samples col-
lected in Cambridge, Mass, by Hi Vol samplers placed near Wai den. Filters
#1 and 2 represent two different Hi Vol samples collected over a 24-hour
period. Filters #3 and 4 were collected at a later date but also repre-
sent two Hi Vol samples run side by side.
The first two filters were collected just after a rainstorm. The pre-
vailing winds were off the ocean. The latter two samples were collected
with the wind in an easterly direction, off the land.
The 8"x 10" Hi Vol sample papers were divided into 10 equal areas and
cut into strips using a razor blade and straight edge. Nitrate and soluble
sulfate were leached from the strips (10% of the area) with approximately
25 ml of boiling water. The samples were cooled to room temperature, the
liquid was decanted into a 50 ml volumetric flask and the filter paper
residue was rinsed several times with distilled water. The volumetric
flask containing the sample was then diluted to volume (50 ml) with water.
Note that no filtration step is necessary since the electrodes can measure
the potential in turbid solutions with no loss in accuracy. This is one
inherent advantage of electrode methods over spectrophotometric techniques.
The nitrate, chloride, iodide, and copper results were determined by
direct potentiometric measurement with ion selective electrodes. The
chloride determinations were checked by potentiometric titration with stan-
dardized Ag+ solution. The titration was monitored with an ion selective
silver electrode. The Cl" electrode and Ag+ titration results agreed to
within approximately 5%. The sulfate measurements were made by titration
with standard Pb(C104)2 in a 50% alcohol-50% aqueous solution. The prog-
ress of the titration was determined with a Pb electrode.
The nitrate electrode measurements on the filter samples yielded the
results in Table VIII. These values are in good agreement with the NOj
levels in urban areas. Since halides may interfere with the nitrate elec-
trode determination, and no levels were readily found in the literature,
UlMen
-------
TABLE VIII
ANALYSIS OF SUSPENDED PARTICULATE MATTER IN CAMBRIDGE
AIR WITH ION SELECTIVE ELECTRODES
Filter No.
1
2
3
4
Species Concentration yg/m
Total Particulate NO" SO^ Cl" I" Cu++
57.6
54.4
59.3
53.4
4.7
3.6
4.7
4.4
0.3
0.3
11.3
11.6
23.2 BD BD
38.8 BD BD
4.2 BD BD
3.1 BD BD
Remarks
Wind blowing off ocean
(from the east)
Light westerly winds
BD = below detectable by electrode - < 106 M/l I", < 10"5 M/l Cu"1"1"
00
-------
19
the concentrations of Cl" and I" were also measured. The Cl"/N0l concentra-
O
tion ratio for filters #1 and 2 is approximately 15. This concentration of
Cl" presents no error in the nitrate measurement. These filters probably
represent one of the worst cases encountered for halide interferences. The
wind was directly off the ocean for filters #1 and 2. This is confirmed by
the extremely low S07 values. SOJj levels approximately 1 yg/m or less are
typical of non-urban areas. However, no major sources of S02 were present
between the ocean and the sampling site. No I" was detected in the samples.
A sea water sample was collected and analyzed. The C1"/I" ratio measured
was 3.6 x 105. On this basis, the I" levels from sea spray would be ap-
proximately two orders of magnitude below the detectable level of the I
electrode. Although I" has a selectivity constant of 20, the low levels
present in the atmospheric samples indicate no interference. No Br" mea-
surements were made but Cl"/Br" measurements of particulate samples col-
lected over the ocean indicate a ratio of approximately 200-500. For fil-
ters #1 and 2, the predicted Br" concentrations are more than an order of
magnitude less than the NO^ levels. With a selectivity coefficient of 0.3,
Br" does not appear to be a problem from salt spray.
/
Filters #3 and 4 are more typical air samples. These values show ap-
proximately the same NO^ levels but the Cl" (sea salt aerosol) is decreased
by an order of magnitude.
The sulfate measurements (potentiornetric titration with the Pb elec-
trode as a monitor) in filters #1 and 2 are nearly at the limit of the
method. The titrant was 10"3M Pb(C104)2 standardized against_H2$04. Fil-
ters #3 and 4 are more representative and typical of urban SO" levels. No
problems were encountered with the titration at this level. Copper ion is
an interference for the Pb electrode, therefore soluble copper measurements
were made for these filter extracts. As expected, no soluble copper was
found in these samples.
Lininger, et al. (1966) collected atmospheric particulate samples
(Cambridge, Mass.) in a cascade impactor. The material on the first three
stages was combined and determined by neutron activation analysis. The
halogen concentration ranges found for Cambridge, Mass, are given below:
lUlakkni
-------
20
Cl" Br" I"
330
ng/m ng/m
1-6 20-200 2-10
Note that the Cl" levels are approximately the same as samples #3 and 4 in
Table VIII.
The halide/nitrate ratios using Lininger's halogen values and our
measured NO" data are 1, 0.025, and 0.001 for Cl", Br" and I", respec-
tively. If we consider the halide selectivity coefficients for the nitrate
electrode, we find that the percent interference for Cl" (k = 6 x 10"3) and
I" (k = 20} is essentially zero because of the high rejection of the elec-
trode for Cl" and the very low concentration of I" in the atmosphere. The
Br (k = 0.13) interference, however, is perhaps measurable and may account
for as much as 0.3% error if all the Br" were soluble.
In a second study, ten Hi Vol samples were collected in Leominster and
Natick, Massachusetts. These sites are approximately 35 miles from the
ocean. Hence, the halide/NOg ratios for non-coastal areas were observed.
The data are presented in Table IX. Most of the samples were non-urban.
The high value for filter #9 is due to its close vicinity to a heavily
traveled gravel area. Note that NO", SO^, Cl" and Pb in this sample are
low. The high suspended parti cul ate level is, therefore, due mainly to dust.
The low value in sample #7 is the result of a site which is distant from
any automotive or other type of pollutant emissions. No analytical prob-
lems were encountered with these levels of NOZ or SOT with the electrodes.
The Br'/NOg ratio for samples 5-10 would give an average 2% error in
the measurement with the nitrate electrode (assuming Br" = 0.13, and NO" =
o
1 for the electrode). A 6% average error would be expected on the basis of
the Br'/NO^ ratios observed in samples 11-14. This indicates that bromide
removal is necessary since the interference with the NO- measurements could
be significant. Statistical analysis of samples 5-14 gives the results be-
low in Table X.
UJalden
-------
TABLE IX
SUSPENDED PARTICULATE SAMPLES FROM THE LEOMINSTER-
FITCHBURG AND NAT1CK AREAS
Site
Leomi nster ,
Mass.
Natick,
Mass.
Filter No.
5
6
7
8
9
10
11
12
13
14
Suspended
Parti cul ate
39.
58.
15,
37
91
46
19
34
57
101
5a
,4
,6b
.4b
.3b'c
.2b
.64
.75
.2
.1
Concentration {pgm/m )
NO" SO^ Cl" Br"
0.77
0,
0
1
0
0
0
0
0
1
,87
.76
.74
.86
.78
.84
.66
.64
.12
1.42
1.47
0.26
2.15
0.86
0.57
1.38
0.57
1.02
0.44
1.58
0.54
0.28
0.20
0.10
0.10
0.30
0.21
0.05
0.38
0.09
0.18
0.07
0.17
0.18
0.10
0.41
0.32
0.13
0.63
r
BD
BD
BD
BD
BD
BD
BD
BD
BD
BD
Pb
(dithizone)
0,
1
0
0
1
0
1
0
0
2
.53
.73
.28
.21
.07
.58
.11
.87
.54
.80
(a) urban
(b) non-urban
(c) near heavily traveled gravel area
r-o
-------
22
TABLE X
Species Correlated Correlation Coefficient
Pb vs Br' Oi82
Cl vs Br" OJ3
Suspended particulate vs Pb Q.69
Suspended paniculate vs NOl o.ll
This analysis shows that the particulate is automotive related, as is the
source of bromide (high correlation between suspended particulate and Pb,
also Pb and Br).
-------
23
IV- LEACHING STUDIES
The primary step in the determination will involve leaching the SO" or
NO" from the filter sample. This can be done by percolating an aqueous
solution through the filter or by dissolving the anions directly into the
aqueous solution to be used for analysis.
Many analytical methods require only a minimal amount of sample. The
sensitivity of a method can therefore be improved if the filter is leached
with a lesser amount of water, thus increasing the ion concentration.
For any leaching method, it is necessary that all of the Teachable
sample be dissolved. However, it is not necessary that all of the leached
sample be recovered. In our experiments we found that the limiting volume
of solution was determined by the amount of filter pulp which could be sus-
pended and shaken in the container. The best sample container for effective
removal of the leaching solution was a test tube. A filter press could be
used with this system and only 0.3 ml of solution remained in the pulp of a
7-inch2 (45 cm2) piece of filter. However, because of bumping, this system
could not be used with a boiling technique.
Using electrodes for measurement, it is possible that the filter need
not be removed. However, for our later studies, comparing colorimetric
and electrode methods the filter was removed.
In an early test using the nitrate electrode, the filter was not re-
moved and the system was leached for 5 minutes. This data was compared to
values obtained as the solution reached boiling, and again after 2 hours
boiling. These data are shown as Series 1 in Table XI. Ten different
measurements were made on sections of two Hi Vol filters.
A more comprehensive leaching study was then undertaken for both nitrate
and sulfate. The colorimetric methods of analysis were used to determine
the efficacy of the leaching methods. The filter strip (2.5 cm x 17.8 cm) was
rolled with gloves on and placed in a 16 mm test tube. Ten or 15 ml of dis-
tilled water were added at time zero and the filter was broken up using a
clean metal spatula. It was found necessary to leach the filter presses
lUMkn,
-------
24
TABLE XI
NITRATE LEACHING STUDIES
(Values in ug)
I: Averages^ ^Sections of F1,ter Ana,y2ed Consecutively;
5 min after at boiling after 2 hr
1 mmay»c •{ nt\ . . J al l*cl *• '"
immersion Point _ boiling
640 750
740
Series II: Four Sections of Filter; Comparison of Broken and Whole
Sections by Two Leaching Methods
Not Broken
Leaching Method Broken
30 min, 90°C 129
30 min, 25°C 200 160
Series III: Comparison of Eight Leaching Methods Arranged by Rank
Time ^m1n) A91tat1Q" Temp. (°Q NO" (No. of Tests!
10 None 25 106 (1)
30 None 90 126 (2)
30 None 25 150 (2)
10 Shaking table 25 192 (2)
30 Shaking table 25 235 (2)
30 Ultrasonic 25 240 (2)
10 Ultrasonic 25 350 (1)
120 None 100 370 (i)
-------
25
with dilute HC1 before use. At the specified time, these presses were in-
serted with a slow twisting motion into 16 mm test tubes and the clear
solution decanted from the stem of the press (Filter/Sampler, Unichem,
P5190-6, Scientific Products, McGraw Park, 111.).
In the initial study, 15 ml of distilled water were used and some fil-
ters were not broken up with the spatula. Comparison of similar leaching
methods indicated that breaking up of the filter was advisable (Table XI,
Series II). Comparison of the 10 and 15 ml tests indicated no improvement
due to increased fluid volume.
Table XI, Series 3 shows the results of filter samples broken with the
spatula. Ten and 15 ml leaching samples are averaged. The samples for
each leaching method were initially chosen at random. For the second group
the same order was maintained, but the strips were offset by five -- plac-
ing the previous end sections in the middle. We found no effect of posi-
tion on the amount of material leached out. The filters were 24-hour
high volume filter samples taken at Walden during the winter.
A similar study was performed on leaching of sulfate, using the barium
chloranilate procedure. These data are shown in Table XII.
From the data of these studies, it is evident that the ultrasonic bath
is very effective in solubilizing the sulfate. The nitrate studies are
less conclusive. However, we currently believe the ultrasonic method to
be a viable alternative to heating.
Walden
-------
26
TABLE XII
SULFATE LEACHING STUDY
Six Leaching Methods Arranged by Rank (yg)
Time (min)
30
30
10
120
10
30
Agitation
None
Shaking Table
Shaking table
None
Ultrasonic
Ultrasonic
Temp. (°C)
90
R.T.
R.T.
100
R.T.
R.T.
SO^ Concentration
670
758
1495
1496
1517
1539
-------
27
V. NITRATE ELECTRODE METHOD
The method developed for measurement of nitrate in this program is a
direct measurement, using an Orion ion-selective nitrate electrode. Since
the method is direct, there are a minimum number of possible interferences.
The non-specificity of the electrode, however, does generate possible posi-
tive errors. These we have attempted to eliminate by addition of complex-
ing agents. Mil ham (1970) and Mil ham, et al. (1970) describe nitrate
analyses of plants and soils using an Orion nitrate electrode. They also
use a solution which includes silver ion in order to reduce interferences.
These investigators report a limit of sensitivity which is similar to our
tests. They also indicate an excellent coefficient of variation (C.V. =
1.5%). With a flow system and temperature control, they indicate a 0.36
(C.V.) for standards and 0.63% (C.V.) for plant extracts. Hence, we be-
lieve that this method should provide a useful method for analysis of
nitrate in air pollution samples.
A. ALTERNATE METHODS OF NITRATE ANALYSIS
Although the measurement of nitrate with the nitrate specific ion
electrode is direct and has minimum interferences, a method with increased
sensitivity would be advantageous.
Alternate methods of ion selective electrode measurement of nitrate
include reduction of the nitrate, and measurement of nitrite or ammonia.
An electrode which responds to nitrite has been briefly investi-
gated at Orion. However, it was not sufficiently developed to warrant
evaluation in this program. If it were to be used, reduction of nitrate
to nitrite could be required. This could be accomplished by the methods
Presently used to produce nitrite in the colorimetric diazotization pro-
cedures for the measurement of nitrate. The method would measure nitrate
and nitrite, unless a blank were run without the reduction step.
The ammonia electrode is commercially available and appears to be
more sensitive than the nitrate electrode. This may be due to the lack of
interferences, since the electrode employs a gas barrier. The electrode
lUlaklen,
-------
28
has been used for the measurement of nitrate using a reduction by aluminum
powder. The method is described in the Orion instruction manual for the
ammonia electrode. The method does not give complete conversion of nitrate
to ammonia, but this might be improved by using an active metal reducing
agent (Zn) in a column (M. Frant, Orion, personal communication). The
method includes several steps (acidification with HC1 and NaF, addition of
aluminum, and addition of NaOH to release NH,).
Obviously, the ammonia electrode will measure ammonium and nitrite,
plus nitrate. Removal of endogenous ammonium might be accomplished by
prior addition of base. However, this would make the method difficult and
time consuming. The endogenous ammonium concentration is comparable to the
nitrate concentration in air pollution samples.
We therefore preferred to develop the nitrate measurement using the
nitrate electrode method but reducing the sample size by reducing the leach-
ing volume. If necessary, the volume could be further reduced by evapora-
tion, followed by the electrode method using a small volume in a microcell
or flow-through cell configuration. Such cells are commercially available
from Orion.
B. EVALUATION OF INTERFERENCES
The major interferences appear to be from the response of the elec-
trode to halides. However, the response to metal oxide anions such as
selenate and vanadate were also studied.
1. Evaluation of Halide Interferences
Iodide, which would be the major halide interference with the
nitrate electrode, was not found in any of the extracted Hi Vol filters.
Bromide, which is an interference with the nitrate electrode, although less
of one than iodide, was found in the extracted samples. Chloride, although
present in the extracted samples, is not a major interference with the
nitrate electrode. In any case, however, silver fluoride addition has been
found to eliminate any of these halide interferences, if and when they exist
in the extracted samples.
-------
29
a. Determination of the Selectivity Constant for Bromide
The selectivity constant indicates the nitrate electrode
error when bromide is present in extracted samples. The selectivity con-
stant of the nitrate electrode for bromide was determined using potassium
nitrate standards and three different bromide concentrations (4 x 10 ,
4 x 10"4, and 4 x 10"5M NaBr). A 25 ml aliquot of each nitrate standard
with 1 ml of known concentration of bromide was run. The millivolt read-
ings were plotted (Figure 1) and the selectivity constant was determined at
each concentration of bromide. The selectivity constant can be related to
the activity (concentration) of the interfering ion by the equation:
where: Z = charge on ion being measured - Z = -1 for N03
Z. = charge on interfering ion - Z1 = -1 for Br"
A = concentration (activity) of ion being measured
A
A. = concentration of interfering ion
K. = selectivity constant for the interfering ion
Ax is determined from the plotted curve by extrapolating the Nernstian re-
sponse part of the curve and the plateau part of the curve. The point on
this curve where AE is 3£ mv from the extrapolated Nernstian response line
gives a value for A which is equal to AX (Orion Newsletter, 1969). K. can
then be calculated from this information. The experimental selectivity
constants with bromide as the interfering anion were found to be:
K = 0.3 0 4 x 10"5M Br"
K = 0.18 0 4 x 10"4M Br"
K = 0.17 0 4 x 10"3M Br"
The average is K = 0.21. The reported selectivity constant of the nitrate
electrode for bromide is K = 0.13 (Orion Manual, 1971). The observed dis-
crepancy between the experimental and reported values is not unexpected.
lUMdeni
-------
-50
4-50r
4 150-
r\>
AE= \ = -18
100-
•MUD
FIGURE i DELTEIPMIMATIOM
A.IMT
-------
31
It is reported that K.'s calculated this way could vary by as much as a fac
tor of two depending on the particular electrode and the level of the back-
ground interference (Orion Newsletter, 1971).
b. Removal of Interferences from Nitrate Measurement by AgF
Addition
Bromide, chloride, iodide, and several other potential In-
terferences are precipitated by sHver ion. This Palpitation is ,uff,-
cient to remove the electrode interferences, without ftlt«t,on of the
precipitant.
Silver fluoride was chosen as the silver salt to be added
since fluoride 1s not a significant interference, and the salt Is readily
soluble.
The fluoride electrode was investigated as the reference
for the nitrate electrode since an excess of r wiH be addec i to 1 *e so u-
.; , r
It should be noted that Milham (1970) also used Ag+ to re-
j u,n=h>n (1970) used the fluoride electrode as a
move Interferences, and Manahan (1970) "seo tn . Urate
reference electrode for the determination of mtrate with an Onon
electrode.
c. Demonstration of Bromide Interference Removal Using AgF
To den»nstrate the removal of bromide Interference, the
r.
UJMe/i
-------
TABLE XIII
COMPARISON OF FLUORIDE AND CALOMEL ELECTRODES AS REFERENCE
ELECTRODES FOR THE NITRATE SELECTIVE ION ELECTRODE
(a)
Concentration KNO, mv
Mud (no KNO-)
ID'6
2.5 x 10"6
5 x 10"6
7.5 x 10"6
10~5 +257
2.5 x 10~5
5 x 10"5
7.5 x TO"5
10"4 +215
2.5 x TO"4
5 x 10"4
7.5 x 10~4
10~3 +156
10~2 +103
10"1 + 53
slope (my/decade)
10-4M-10-2M) + 56
(a) Nitrate electrode and
(b} Hitrate electrode and
9/27/72
(b) (a)
mv with AgF
Addition mv
+319 +278
+261
+277 +261
+264 +214
+194
+178
+168
+210 +160
+153 +108
+96 +58
+ 55.5 + 53
9/28/72
(b)
mv with AgF
Addition
+314
+317
+306
+269
+245
+227
+217
+210
+151
+ 96
+ 59
9/29/72 10/2/72
(a) (b) (a) (b)
mv with AgF mv with AgF
IIV Addition mv Addition
+286
+252
+270
+267
+269
+240
+263
+260
+219
+197
+180
+170
+163
+107
+ 53
+ 56
+324
+309
+309
+309
+313
+300
+284
+276
+269
+246
+229
+219
+212
+154
+ 98
+ 61
+290
+284
+279
+276
+271
+268
+250
+235
+226
+217
+162
+106
+ 55
+ 55.5
+331
+330
+327
+325
+319
+318
+301
+289
+280
+270
+211
+151
+ 95
+ 59.5
fiber tip calomel reference electrode
f 1 uori de el ectr ode as
reference electrode
CO
-------
33
and silver. The curves obtained were identical (Figure 2), thus indicating
that the bromide interference is removed by the silver ion addition, and
that the electrode system is indeed responding to the nitrate concentration
relative to the fluoride ion concentration.
2. Effect of Selenate on the Nitrate Electrode
The effect of the presence of selenate on the measurement of
nitrate concentration was also investigated. The presence of selenium as
an atmospheric contaminant originates with the incineration of paper, the
coloring of paints and inks, the production of glass and plastics, and the
refining of copper, lead, and zinc. Since selenium often acts like a non-
metal , we investigated its effect on the response of the nitrate electrode.
It was found that the interference of selenium would be minimal. The data
are presented in Figure 3 as apparent nitrate at various selenium to nitrate
ratios.
3. Effect of Vanadate on the Nitrate Electrode
The effect of the presence of vanadate on the measurement of
nitrate concentration was also investigated. It was found that at high con-
centrations, vanadium appears to affect the nitrate concentration measure-
ment by lowering the millivolt readings of the nitrate standards (Figure 4).
This experiment was performed by addition of different amounts of a 10" M
sodium orthovanadate solution to nitrate standards.
It should be noted that the normal leveling effect of an in-^
terference does not occur until the nitrate concentration is down to 10" M.
At this point, the vanadate concentration is an order of magnitude higher
than the nitrate concentration.
4. Removal of Selenium and Vanadium as Interferences with the
Nitrate Electrode
The effect of selenium and vanadium on the nitrate electrode
using the nitrate-fluoride electrode system with silver fluoride addition
was investigated after the development of this system. As reported above,
lUUkni
-------
-50
150
200
MUD
10
10
^4"
10"
?3~
25mls NO7i- Iml IO~M AqF
4- Iml >3 ^
25mls NCTt Iml JO~'MAgF
+ Iml IO~2M NaBr
CO
10
10
k-l
•RGUREZ EUM\N£Y\OH OF HAL\DE \NTERFERENCE WITH NOT ELECTRODE
-------
-J
10
— 4
10
10
-1
iO
10
10
10
10
-MEASURED [l\lO:]
-CALCULATED [NOj]
10
„..!„.........J_
tos
FIGURES EFFECT OF SELENATE ON NITRATE DETERMINATION
CO
on
-------
300
230
200
ISO
t JOO
*• 50
NOj CALIBRATION POINTS
A 1,05 x IO~M NcaV04»xHJD
MUD
FIGURE f EFFECT OF VANADATE ON ^0^] DETERMINATIONS
10
-------
37
the interference due to selenium would be minimal. Work with the addition
of selenate in the presence of silver fluoride proved that any potential in-
terference due to selenate is removed by the silver fluoride except when the
ratio of selenate to nitrate is of the order of 38:1 or greater (Figure 5).
Taking into account the large concentration of selenate needed to interfere
2
minimally with the determination of the nitrate concentration ([SeO. ] >_ 38
x [N03]), it is reasonable to consider any selenate interference to be re-
moved by the addition of silver fluoride.
Vanadium also might be an interference with the nitrate elec-
trode. Thus, it was decided that the vanadate interference with the nitrate
electrode in the nitrate-fluoride electrode system with silver fluoride pres-
ent should be investigated. It was found that any potential interference due
to vanadate was eliminated by the presence of silver fluoride except when the
ratio of vanadate nitrate was equal to or greater than 3.8:1. Since the
highest reported ratio of vanadate to nitrate is 1:3 (NASN Air Quality Data,
1966), the presence of silver fluoride in nitrate standards and samples should
eliminate the vanadate interference (Figure 6).
This removal of interference is due to the insolubility of sil-
ver vanadate and silver selenate.
C. STABILITY STUDY OF THE NITRATE ELECTRODE
The stability study of the nitrate electrode was performed over a
Period of two months. The data are given in Table XIV.
These data can be used to estimate the probable precision of the
method, and to set tentative standardization requirements.
D. NITRATE PROCEDURE DEVELOPMENT
A tentative procedure was developed for the nitrate measurement.
This procedure included addition of AgF and used the fluoride reference
electrode. This procedure was tested on standards, and the precision of
the calibration curve method and standard addition calculation method were
compared.
UlMen
-------
+300
+250
+200
+150
+ 100
+50
STANDARDS WITH AgF
A NO" STANDARDS WITH AgF AND
IO~4M N
MUD
10
-5
10
-4
10
-3
10
-2
10'
FIGURE5 ELIMINATION OF SELENATE INTERFERENCE WITH
THE NITRATE ELECTRODE
-------
+300
+ 250
4200
+ 150
NOT STANDARDS WITH
A N0~ STANDARDS WITH AgF AND
3.84 XlO 4M
xH20
CO
+ 100
+501
MUD
FIGURE
10"
10
-4
JNG
10
-3
10'
icy
ELIMINATION OF VANADATE INTERFERENCE
THE NITRATE ELECTRODE
WITH
-------
40
TABLE XIV
STABILITY OF NITRATE ELECTRODE* AS A FUNCTION OF TIME
uate
— • — . »
9/1/72
9/7/72
9/8/72
9/11/72
9/15/72
9/18/72
9/27/72
9/28/72
9/29/72
10/2/72
10/4/72
10/11/72
10/19/72
10/25/72
Mean
Standard Deviation
mv/decade (10 -10~4M)
— —
54.0
53.0
55.0
53.5
56.0
53.5
56.0
51.2
56.0
55.5-
56.0
54.5
C 1 r™
51.5
56.0
54.4
1 7
1 . /
mv ([N0~] =
106
104
106
108
108
104
103
108
107
106
108
108
109
110
106.8
2.0
With fiber tip calomel reference electrode
-------
41
The method was written as a precursor to the final method which is
described in Appendix A. For these tests on nitrate standards, 25 ml of
KNO^ solutions were used.
E. DEMONSTRATION OF THE NITRATE PROCEDURE
1. Calibration Curve Method
For the analysis of this method, nitrate standards with a known
amount of silver fluoride added were run in random order (Table XV) in
order to determine the accuracy and precision of the electrode method. The
results of the statistical analysis of the data obtained on two different
days are given in Table XVI. The results obtained on the two different days
were then combined despite a possible small difference in the concentration
of silver fluoride in the samples. The slopes used to calculate the per-
centage error were the average of the slopes for the two days. The results
of the statistical analysis of the combined data are given in Table XVII.
The percent error was found as follows:
E = S log C + K
dE _ S d&nC
dE _ / S Wl\
HC - \OJ\CJ
(AE) x L|03 (100) „ £ (100)
(AE) [230.3\ = % error in terms of concentration
Sl0pe between l
-------
42
TABLE XV
TABLE OF RANDOM NUMBERS
M
—*————_
^
10"b
3 x 10"5
ID'4
3 x 10~4
lO'3
1
— —— — — — — — . — __
n
16
i
25
8
2
*
20
18
2
15
9
No.
3
5
14
21
22
23
4
7
12
10
17
19
5
4
3
6
24
13
-------
43
TABLE XVI
RANDOM ORDER NITRATE CONCENTRATION DETERMINATIONS
[NO']
Day I
10" 5M
3xlO"5M
10"4M
3xlO"4M
10' 3M
Day II
10" 5M
3xlO"5M
10' 4M
3xlO"4M
10" 3M
*
% error found
**
% error found
Average
(mv)
+159.66
+150.32
+125.66
+ 99.46
+ 69.02
+163.58
+150.20
+124.56
+96.66
+67.42
Standard
Deviation
(mv) (AE)
+2.61
+1.71
+1.65
+0.17
+0.33
+1.81
+1.46
+3.55
+0.89
+2.78
*
Slope % Error
(S)
7.66 62.0**
34.5 11.4
56.6 6.7
56.6 0.7
56.6 1.3
24.0 17.4
45.5 7.4
57.1 14.3
57.1 3.6
57.1 11.2
_ — ••
by AE x 22|il = % error
by graphical method
-------
44
TABLE XVII
COMBINED RANDOM ORDER NITRATE CONCENTRATION DETERMINATIONS
[NO']
— — — — — — -___
10~5M
3xlO~5M
10" 4M
3xlO"4M
10' 3M
Average
(mv)
"
+161.62
+150.26
+125.11
+ 98.06
+ 68.22
— — — .
Standard
Deviation
(mv)
• .
±2.59
±1.81
±2.38
±1.49
±2.21
Slope
(S)
15.8
40.0
56.9
56.9
56.9
% Error
—
37.7
10.4
9.6
6.0
9.0
-------
45
2. Standard Addition Method
In addition to running nitrate concentration determinations in
random order on Day II, the standard addition method was also run in random
order on the same nitrate samples. A 25 ml aliquot of nitrate standard with
AgF was run to determine its initial millivolt reading. The reading was
taken two minutes after the electrodes were inserted in the solution, after
which time a 1 ml aliquot of a sample 30 or 33 times as concentrated as the
original sample was added to the 25 ml aliquot. Two minutes after the ad-
dition, the second millivolt reading was taken. From the millivolt change
before and after the 1 ml aliquot addition, one is able to obtain the con-
centration of the initial 25 ml sample as follows:
V M
where: C. is the change in concentration on addition to the sample
V is the volume added
M is the molarity of solution added, and
V is the original volume of solution
To find the total original concentration, CQ, the equation:
co = CA (antilog At/S) -
was used, with
AE = the millivolt change on addition of the known concentration
to the original sample, and
S = the slope of the calibration curve at the CA point.
The results of the statistical analysis of the data obtained are presented
in Table XVIII. Comparison of these data with that of Table XVI, Day II,
indicates no advantage to the standard addition method.
/UlaUeni
-------
46
TABLE XVIII
STANDARD ADDITION IN RANDOM ORDER
Initial CA An s
[NO'] A E
10'5M 1.2xlO"5M 10.2 24
10.8
12.1
10.3
9.1
Means yAE=10.5 mv
Std. Dev. a F=l .1 mv
3xlO"5M 4xlO"5M 16.4 45.5
16.7
16.5
17.4
15.6
yAE=16.5 mv
oAE=0.7 mv
°Cal ciliated
7xl.O"6M
6xlO'6M
5xlO"6M
7xlO'6M
8xlO"6M
yc =6.6x10"^
0
ac =l.lx!0"6M
3.09xlO"5M
3.01xlO"5M
3.07xlO"5M
2.8x10"5M
3.33xlO"5M
yr =3.1xlO"5M
Lo
ar =1.8xlO"5M
>o
-------
47
TABLE XVIII (continued)
Initial CA
[NO'] A
10"4M 1.2xlO"4M
3xlO"4M 4xlO"4M
AE S
17.9 57.14
17.9
17.8
18.3
18.2
yAE=18.0 mv
aAE=0.22 mv
20.0 57.14
20.0
19.7
19.8
19.6
yAE=19.8 mv
aAE=0.2 mv
c
Calculated
1.14xlO"4M
-4
1.14x10 *M
-4
1.14x10 14
-4
1.10x10 11
l.llxlO"4M
_4
yc =1.12x10 M
CT_ =0.02xlO"4M
Co
3.23x10"4M
-4
3.23x10 ^M
3.30xlO"4M
-4
3.28x10 >l
-4
3.32x10 *M
-4
yc =3.27x10 ^M
-4
ac =0.04x10 HM
-------
48
TABLE XVIII (continued)
Initial r
[NO'] A
10"3M 1.2xlO"3M
AE S
19.6 57.14
18.7
17.7
18.7
18.6
yAE=18.7 mv
0^=0.7 mv
C
°Calculated
9.98x10~4M
1.07xlO"3M
1.15x10~3M
1.07xlO"3M
1.08xlO"3M
uc =1.07xlO"3M
°Cal
ac =0.06xlO"3M
o~ ,
-------
49
F. COLORIMETRIC REFERENCE METHODS
In early tests, nitrate was measured by the electrode and phenol
disulfonic acid (PDS) methods. Unfortunately, a high nitrate blank of the
filter and problems with the PDS method vitiated these results. However
an attempt was made to correct the PDS values for the NO^ levels, by
running the Griess-Saltzman reaction on these samples. The amount of N02
averaged 6% of the nitrate level on these six samples.
In later tests we used the Sawicki and Scaringelli (1971) nitrate
method. In this method NOg is reduced to NOg by hydrazine and the color
reaction uses sulfanilamide and N- (1-naphthyl) ethylenediamine as in the
Griess-Saltzman reaction.
G. COMPARISON OF ELECTRODE AND COLORIMETRIC NITRATE METHODS ON HIGH
VOLUME FILTER SAMPLES
Filter sections were obtained from EPA, courtesy of Ms. Eva
Wittgenstein. Random filters were taken from this packet and were desig-
nated as S-l to S-9. The source and weight of these filter sections is
shown in Table XIX. All filters were taken at standard conditions (36-47
cfm for approximately 24 hours). Portions of these filters had been
Previously analyzed by EPA.
These nine sections (48 cm2) of high volume filters were analyzed
for nitrate by the electrode calibration curve method, and by the wet
chemical referee method (Sawicki) for nitrate.
The filtrate was diluted from the required 10 ml to 25 ml to allow
b°th the electrode and colorimetric method to be run.on the same filtrate.
' eliminated variables in the filter sections and leaching. However it
both methods to be performed on extracts with half the normal nitrate
concentration.
The results of this study are shown in Table XX. The test was
done using only the electrode and Sawicki methods. A portion of
extract was also used for the sulfate analysis (see later
/Ulalden,
-------
50
TABLE XIX
SOURCE OF FILTERS FOR COMPARISON DETERMINATION
Wai den
No.
Series 1
S-l
S-2
S-3
S-4
S-5
S-6
S-7
S-8
S-9
Series 2
s-n
S-l 2
S-l 3
S-14
S-15
S-l 6
S-l 7
S-18
S-19
S-20
EPA
No.
06 3 384
065669
064918
064919
065888
065228
065972
065478
065956
066000
065907
065889
064899
065672
064917
063385
065689
065310
064248
Site
Particle Wt (g) -
f)
Total Filter (406 cmj.
Bronx, N.Y.
Anaheim, Cal.
Fairlawn, N.J.
Fair!awn, N.J.
Elizabeth, N.J.
Riverhead, N.Y.
Birmingham, Ala.
Chattanooga, Tenn.
Birmingham, Ala.
Birmingham, Ala.
Birmingham, Ala.
Elizabeth, N.J.
Ridgewood, N.J.
Anaheim, Cal.
Fairlawn, N.J.
Bronx, N.Y.
Santa Monica, Cal.
Magna, Utah
Charlotte, N.C.
0.252
0.142
0.041
0.043
0.135
0.142
0.297
0.049
0.225
0.216
0.034
0.090
0.103
0.163
0.071
0.218
0.067
0,126
0.121
-------
51
TABLE XX
INITIAL NITRATE ANALYSIS
Nitrate in Filter Extract (ug/48 cm )
Sample
No.
S-l
S-2
S-3
S-4
S-5
S-6
S-7
S-8
S-9
Blank
Blank
Electrode
1 2
486
3180
230
197
736
408
962
226
470
310
2320
149
67
496
102
666
194
248
34
33
Sawicki
1 2
313
1880*
150
59
433
144
603
81
279
195
1270*
116
58
337
69
485
57
220
25
25
PDS
(2)
170
1880*
16
20
307
80
134
11
157
29
* Outside range of colorimetric procedure
-------
52
section). The test was repeated using a separate section of the filter, and
using the electrode, Sawicki, and phenol disulfonic acid (PDS) procedures.
Sample S-2 was not measured accurately by the Sawicki method and perhaps not
by PDS. Also it was too highly weighted in correlation coefficients. Cal-
culated correlations, without S-2, are given in Table XXI.
The correlation between the three methods was not as good as anti-
cipated, presumably because of laboratory inaccuracy. Correlation within
each method on two sections of the same filter also showed large variability.
In these tests the electrode method gave a 30-40% higher value, which would
also be a problem of the method.
Therefore the test was repeated, adding a known quantity of nitrate
to each sample. The initial quantity and the incremental increase were
analyzed by the electrode, Sawicki and PDS methods. For this study new filters
had to be used. These are described as S-ll through S-20 in Table XXII. The
2
equivalent of two strips (97cm ) were leached in 50 ml of water.
The results of these tests are described in Tables XXIII through XXV
and Figures 7 through 9 . In these tests there is an excellent one-to-one
correlation between the methods. One sample (S-ll) appears to have an
unusually high nitrate value according to the electrode method. This is one
of two samples from Birmingham, Alabama. We have no explanation for this
difference. However, since the value is outside of 2.5 standard deviation
the values are recalculated without this entry. The standard deviations be-
tween the methods are shown in Table XXIV. There is a ±46 rug [N03~) error
between the electrode and the Sawicki methods (including the 207 mg error on
the 2903 mg S-15 value). Similarly, there is a 13% relative standard devia-
tion (including the 9 mg error on the 24 mg S-20 sample).
Recoveries of nitrate by the three methods are also satisfactory (Table
XXV). The low value of the Sawicki method is presumably due to the lack of
reagent for the high nitrate values. Thus we are outside of the range of the
method. The large recovery (125%) by the electrode method is unexplained.
However, a lower recovery accuracy of the electrode method is due to the 124 mg
lUlaldeni
-------
53
600
M 500
00
j"
§ 300
zoo
loo
U4
600
Sawlckl Method, yg/48c«
FrGURE 7 . COMPARISON OF ELECTRODE AND SAJttCKI MfiTH©BS
(Perfect Correlation Line Drawn)
-------
54
600 ~
500 -
o
CO
o>
* 400
Ul
300 -
200 ~
100 -
PDS Method, ug/ 48cm
FIGURE fi. COMPARISON OF ELECTRODE AND PDS METHODS
-------
55
Swrtcfct Method, yg/ 48cm*
FIGURE 9 . COMPARISON OF PDS AND SAWICKI METHODS
-------
56
TABLE XXI
CORRELATIONS OF INITIAL NITRATE DATA*
Methods parm . .
J:orJS a:lon. . Linear Regression. . \
Coeffielent(r) (y , electrode x = colorimetric^
Electrode - Sawicki (!) 0.978 y-1.40x+104
Electrode- Sawicki (2) Q.971 y= i.31x+28
Electrode - PDS 0.660 y = K34x + 129
SI - S9 excluding S-2 which is outside of colorimetric
range
-------
57
TABLE XXII
ELECTRODE DATA FOR HI VOL NITRATE MEASUREMENT
E vs F"Reference
— •••ff 1 V,
-
s-n
S-12
S-13
S-14
S-15
S-16
S-17
S-18
S-19
S-20
JO \J 1 U/\ Wl <-*W W
(for 48 cm2)
100
100
100
100
100
100
100
80
100
100
(mv)
226
250
245
256
183
229
247
224
245
298
(xlO4)
3.65
1.40
1.70
1.12
19.4
3.15
1.58
3.90
1.70
0.13
-------
TABLE XXIII
REPEAT OF NITRATE ANALYSIS OF HI VOL SAMPLES
Sample
11
12
13
14
15
16
17
18
19
20
Nitrate in
Electrode
566
217
264
174
3007
488
245
484
264
20
Filter Extract (uq/48 rm2
Sawicki
342
194
256
160
2800
477
232
543
221
29
PDS
-i •
324
114
224
150
2824
414
220
524
Z54
68
-------
59
Methods
Electrode-Sawicki
Electrode-PDS
Sawicki-PDS
Electrode-Sawicki**
Electrode-PDS**
TABLE XXIV
flTTONS OF NITRATE ANALYSIS
Correlation*
Coefficient
0.896
0.869
0.971
0.986
0.945
Linear Regression*
(Y=electrode, x=colorimetric)
y *
y =
y =
y =
y =
0.989x + 33
l.OBx + 36
l,06x + 3
0.911x + 29
0.954x + 35
p
Std. Dev. = Ave. Dev. x ——
VTT
Assumes: Electrode (y) = Colorimetric
Electrode-Sawicki
Electrode-PDS
Electrode-Sawicki**
Electrode-PDS**
mgL
69
89
46
69
17
34
13
31
* Analysis without S-15
** Analysis without S-ll, which is outside 3 Std, Dev.
-------
60
TABLE XXV
RECOVERY OF NITRATE ADDED TO HI VOL SAMPLES
Sample
S-ll
S-12
S-13
S-14
S-15
S-16
S-17
S-18
S-19
S-20
Average
Std. Dev.
% Recovery + Std. Dev.
Recovery of 124
Electrode
(1543)
160
155
135
(3743)
190
164
(2953)
145
139
155.4
18.6
125+12
V 3l»n ^ ^ T V /
w u V* 1 w IN (
108
107
104
97
(214)
80
86
C554)
95
100
97.1
9.3
78±10
PDSJ4)
128
127
130
127
120
129
129
124
123
122
125.9
3.4
101+3
(1) Data in parentheses not used in calculation.
(2) ?SraJ/!L?Itr trjct"sed (5ml). Hence these recoveries are 5 times
larger relative to the sample concentration.
(3) Increment too small to measure accurately (< 25% initial value).
(4) Value outside range of method.
-------
61
of nitrate being added to five times the amount of sample used for the colon-
metric analyses.
lUtttn,
-------
62
VI. SULFATE ELECTRODE METHOD
A. SULFATE ELECTRODE APPROACHES
Early in the program a sulfate electrode was prepared by Orion Research
Inc. according to Rechnitz et. al. (1972). This electrode was reported by Rechnitz
to have a Nernstian response for 50= down to 10"4 molar. We found ^25 mv/decade
between 10" and lO^M and ^5-10 mv/decade below 10'2M. Although a reading is ob-
tained, it is not clear that the electrode is actually responding to sulfate.
This approach did not appear very promising since the electrode had to be soaked
prior to analysis, the response time was slow, and the sensitivity inadequate.
A later article by Rechnitz and Mohan (1973) prompted us to contact
Prof. Rechnitz to ask if we could borrow an electrode for testing on this program,
and also whether or not there were cation interferences (e.g. Pb^). Prof. Rechnitz
replied that he had received numerous requests for electrodes since 1972, and was
unable to meet these requests. Thus they published the second paper giving complete
instructions for making the electrode. He also mentioned that cation interferences
had not been determined.
We did not pursue this approach any further.
It should also be mentioned that sulfate electrodes have been prepared
by other investigators, using an insoluble precipitate in a polymer matrix.
Saunders has developed such a system (BaS04) and this is published as U. S. Patent
No. 3,709,811 (Government-owned invention).
Jasinski and Trachtenberg also have a government-owned application (No. 350-
444) which describes a determination of sulfate using a ferric ion-selective electrode*
The method which we pursued was the titration of the sulfate with Pb (C104)2
and measurement of the excess lead with an Orion lead electrode. This concept is
described by Orion (1972), and by Ross and Frant (1969).
B. SOLUBLE SULFATE INTERFERENCES
In the determination of sulfate by the precipitation with lead, two types
r^\
UUP
-------
63
of interferences are possible. These are:
(1) Metals which respond at the lead electrode.
(2) Other anions which precipitate with lead.
However, only those materials which would be soluble in air pollution samples need
be considered. Fortunately, materials which precipitate with sulfate, such as lead,
are not potential interferences since the precipitate will form and these metallic
sulfates are not expected to be found in the leached sample. If we were attempting
to measure total sulfate, these would be interferences.
However, potential anion interferences should be considered in view of
Possible lead precipitates with the titrant. If a significant concentration, rela-
te to sulfate, exists in the solution, then titration of these anions with Pb(C104)2
"HI cause a positive error in the sulfate determination. Lead halides, fortunately,
^ relatively soluble, and lead vanadate is slightly soluble. However, the error
f'°m other anions (e.g. Sb-Of? AsO^3, HBO^2, H^. SeO,'2, CrO," , MoO, . Mn04 )
«*'ch form lead precipitates should be considered further. We believe that these
«"1ons will be found in lower concentrations than the endogenous lead in most instances,
Tl*s. these may exist only as insoluble precipitates. However, there is more sulfate
«*n lead, hence, much of the lead is tied up in the sulfate precipitate. The quantl-
* of interfering anions that are soluble is probably best determine experim tal y.
^dynamically anions which form more insoluble lead salts than PbSO shouId pr -
Cl'Pitate in preference to PbS04 during the leaching process, and lead salts that
"tore soluble than PbS04 should not interfere.
The only practical test of this form of interference Is by the correlation
of electrode and chemical measurements of exposed filters.
The lead electrode 1s relatively Insensitive to other metals.
lof :I per mercury, and silver poison the electrode, «-'-^
">terfere at concentrations above the concentration of lead. Most
*1.1. fon, insoluble sulfates and will therefore not be found i. the leached
Si»»P1e.
lUhUenj
-------
64
4-4*
The remaining ions (especially Cu ) could be complexed with citrate or
ammine complexers, if necessary.
C. PRELIMINARY SULFATE STUDIES
1. Solvent Studies
A problem appeared when low levels of sulfate (< 5 x 10" M) were
analyzed due to the solubility of PbS04 in aqueous solution (K = 1.6 x 10" ).
Therefore, a study of different solvents was undertaken so the K could be suffi-
ciently lowered to determine low levels of sulfate. The concentrations of the
solvents were varied and the time to reach equilibrium was noted.
The lead electrode was set up with the double junction reference
electrode. The reference electrode had 1M NaN03 in the outer chamber. To two milli-
liters of 10"3M lead, water and solvent were added to bring the sample volume to
18 ml. The reference potential was recorded (10~4M Pb++) and 2 ml of 10" M S04"
were added. When the potential reading stabilized, the time and potential were
recorded again (Pb =V^D^'
The data of Table XXVI summarize this study. The equilibration time is
the minutes required to come within 2 mv of the final reading. The A mv value is
the difference, between the solvent system and water, for the change occurring after
the addition of sulfate. Thus,
A mv = (EPb =v/Ts ~ EPb = 10"4)solv " (EPb =^p "EPb * KT4)H20'
Based on this study, which is summarized by Figure 10, a series of
calibration curves (known subtraction study) were prepared using solutions of 50%
methyl cellosolve acetate in water and 50% methanol in water. The methanol calibra-
tion is included to show the increased sensitivity obtained by using the cellosolve
acetate which lowers the K of PbS04 by at least one order of magnitude over that
of methanol at the 50% solvent concentration. This K is at least two orders of
10
magnitude lower than aqueous solution, or less than 10. Sulfate in the range
of 10 to 500 ug may therefore be determinable.
lUlaldeni
-------
65
TABLE XXVI
EFFECT OF SOLVENT ON THE K$p OF PbS04
Methanol %
Equilibration Time
A mv
Acetone %
Equilibration Time
A mv
Dioxane %
Equilibration Time
A mv
Methyl Cellosolve 1
Equilibration Time
A mv
Methyl Cellosolve
Acetate %
Equilibration Time
A mv
0
*
_
o
2
£
0
10
o
0
10
10
3
4
10
1
2
10
6
-21
10
*
-1
10
9
14
20
12
6
20
4
9
20
14
-8
20
2
*•
-8
20
10
27
30
7
13
30
5
23
30
7
4
30
3
-1
30
2
36
40
4
22
40
2
28
40
3
12
40
4
13
40
1
40
50
3
38
50**
4
42
50
3
31
50
4
30
50
3
57
60
2
58
60**
5
69
60**
3
58
60
3
49
60
2
76
70
1
74
70**
2
91
70**
9
63
70
2
72
70**
2
96
80
3
89
80**
3
128
80**
7
81
80**
106
80**
2
122
"NO change of reading (< 2 mv)
*No precipitate evident
-------
Figure 10- Solvent Effect on the K of PbSO..
sp 4
-140 -
-1201-
s
0HO
£
i
UJ
O.
Z
-80
-60
-40
-20}-
LE6END
METHYL CELLOSOLVE ACETATE
ACETONE
METHYL CELLOSOLVE
• METHANOL
O DIOXANE
0
10
30
40
50
VOUiVAE
OF
6O 70
SOL.VE.NT
80
90
-------
67
2. Known Subtraction Study
The major effort was placed on developing the method of known sub-
traction as a means of determining sulfate. This involves the addition of sulfate
to a known amount of lead and monitoring the change in potential with an ion-selective
]ead electrode. The quantity of lead must be greater than the quantity of sulfate.
The decrease in lead concentration after addition of sulfate gives the measurement
of sulfate.
Initially, varying amounts of standard aqueous sulfate were added to
* standard lead solution in SOX-methanol. The results were inaccurate since various
dl'lution factors were introduced and the solvent (methanol) was not kept at a
constant 50%. It was also found that about 8 minutes were required between additions,
in order to reach equilibrium.
In later experiments the sulfate solution was also prepared with 50%
solvent. The data of these experiments are plotted in Figures 11-14. The data are
M°tted linearly, since this approximation is sufficient for a logarithmic response
°vei" a small range of concentrations.
Figures 11 and 12 represent the upper calibration range, that Is, 48 to
480 ug of SO/2 The only difference Is the solvent used. As can be seen, the
Wentfal change fro, 3.8 x lo"»M to 2 x lO^M 1s greater when the metnyl ce oso ve
a«tate 1s used than when methanol 1s used (16.3 « versus 10.9 w). * u^13 "
14 show the lower range, 4.8 to 48 Mg SO,'2. Again the cellosolve acetate resp ds^
"th a greater potential change when the lead concentration changes from 7.8 x. 10
to 6.0 x 10"5M. This 1s 7.2 mv compared to 4.4 mv.
The difference In sensitivity Is primarily due to the use of different
<"ot decade different lead solutions for these tests (I.e., 4 x 10 «N for high
fan9e and 8 x 1(T5M Pb(C104)2 for the low range).
3. Comparison to
Method
The known subtraction Kthod 1s better than the potentlometHc tltratlon
the deterl n, t oTof a*b1ent levels of SO,"*. A .1.1.. of 10 P™^^
°btam a tltratlon curve while only one determination i. required for known sub-
-------
10
£-158
o
d -162
-166
M Pb*2= U3x IO~5(MV) * 2.15x10 3
COR COEF. = 0.985
B
PO 1 NT
A
B
C
D
E
F
0
V9 S04
46
96
192
288
384
480
NONE
jj MOLES
0.5
1
2
3
4
5
-170
-174
4x10
I
- I
3.5x10
-4
s-4
3.0x10 " 2.5x10
M Pb*2 \?i FINAL SCLU TIOU
Figure 11. Use of Known Subtraction Method for S(h
Detection.
J
-4
2.0x10
-------
-158
-160
M
COR COEF.- 0,976
B
POINT
A
B
C
0
E
f
0
so;2
48
96
192
288
384
480
NONE
pMOLES 504 2
0.5
I
2
3
4
5
-I62|
0-1641
-166
-168!
-170
vo
-I721-
4X IO
J_
_L
-4
3.5x10'
3.0x10
M Pb"2 \N FINAL SOLUTION
2.5x10'
Figure 1Z. Use of Known Subtraction Method for SO^ Detection,
Methanol
2.OXIO
-4
-------
M Pb+2 =2.525xfO~6(MV)
COR COEF. = 0.992
5.055x10
-4
2
-166
-176
-178
8.0x10
-5
7.5x10
7.0xlO"5
FINAL SOLUTION
6.5x10
M Pb IN
Figure 13. Use of Known Subtraction Method for S0^c Detection.
HethyA CeUosolMe
6.0x10
-------
-170
-171
COR COEF. = O.995
-J72
-173
-174
-175
8.OxlO"D
7X>XIO":J 6.5x10"
. _ IN FTNAL SOLUTION
Figure 14. Use of Known Subtraction Method for SO"2 Detection.
Methanol
-------
72
traction method. Another disadvantage with the titration is a very shallow end-
point break, which requires that the first derivative be taken to accurately des-
cribe the endpoint. Figure 15 shows such an example where the sulfate level was
found to be 8.3 ug/m3 or 3.0 x 1(T4M. This experiment was performed in 50% methanol.
D. ANTICIPATED SULFATE ACCURACY BY STANDARD SUBTRACTION
Measurement of sulfate by measuring lead ion decrease requires that the
quantity of lead be greater than that of sulfate. However, if the sulfate quantity
is much smaller than the lead, then the precision of the method will be poor. In
this case, a large change in sulfate will result in only a minor change of the lead
concentration.
Calculations were therefore made to determine how many lead solutions would
be reared to achieve reasonable accuracy In the sulfate determination. The pr«ry
equation is:
E - E0 = 29 log ([Pb]o - [so4]) - 29 log [Pb]Q
[Pb]
0
where:
[Pb]0 is the initial quantity of lead, and
[S04] is the quantity of sulfate.
We assume no volume change or dilution in tMe „ •• . A
IT*, ununon in this analysis. We will let FPbl = 1 and
™ P "9
:, ..5rations are °f «° rM° °-i
incre.ents is shown in Tab,e x^n 'Ind F ^ ^^ ™»°™ «"
The concentration error associatpH in-m -, u
, . , associated with an electrode error ran he found by
analyZ1ng the slope of the curves of Figure 16. Alternately t , " nb* dete
nnned mathematically by differentiating the above equation ?hus
-------
-220-
15 ML ALIQUOT OF
50 ML SAMPLE
-340
2£> 3.0 4.O 5.0 6.0 O I 2 3 A
juLS OF I0~3 M
Figure 15. Potentiometric Titration and First Derivative for SOT2
Detection. 4
CO
-------
TABLE XXVII
CALCULATION OF ELECTRODE RESPONSE WITH VARIOUS RATIOS OF S04/Pb
(S04)/(Pb)
Pb=l
0.05
0.10
0.15
0.20
0.25
0.30
0.40
0.50
0.60
0.70
0.80
0.90
(Pb-S0.)/Pb
H .
Pb=l Pb=.2 Pb=.3 Pb=.5
0.95 0.95 0.83
0.90 0.50 0.67
0.85 0.25 0.50
0.80 0.33 0.60
0.75 0.17 0.50
0.70 0.40
0.60 0.20
0.50
0.40
0.30
0.20
0.10
AE (mv)
Pb=l Pb=.2 Pb=.3
0.65 3.62 2.35
1.33 8.73 5.04
2.05 18 8.73
2.81 13.96
3.62 22.32
4.49
6.43
8.73
1 1 . 54
15.16
20.27
29.00
Pb=.5
4.49
6.43
8.73
11.54
20.29
-------
28
24
20
IB
12
8
(XI
O2
03 0.4 O5 0.6 0.7
0.9
FIGURE /6 VOLTAGE RESPONSE FOR VARIOUS
r -3 c +3
N/hi
RATIOS
/ 0
-------
/b
dE 29 1
"" 2.3
d[SOJ
-lE^-"1 7.9 ([Pb]Q - [S04])
This data is tabulated in Table XXVIII for 80% precipitation and for the maximum error
points.
TABLE XXVIII
[S04]/[Pb]o
0.8
0.5
0.3
0.2
0.1
[Pb]Q
1.0
1.0
1.0
0.5
1.0
d[S04]
X
dE(mv) [S04]
(% error/mv)
2%
8%
19%
12%
72%
100
We therefore decided to use the 0.1:0.3:1:3:10 increments of lead solutions.
E. SULFATE PROCEDURE DEVELOPMENT
An initial procedure was written in which the sulfate solutions were not
premixed with an equal volume of methyl cellosolve acetate (MCA). The 50% MCA was
maintained by having a higher MCA concentration initially (62,5%) which was diluted
to 50% with the aqueous solutions. The blank was obtained by addition of water to
the 62.5% MCA solution.
Initial results using this procedure were not as reproducible as had been
obtained earlier. Hence a modified procedure was developed. The procedure uses
the 0.1:0.3:1:3:10 lead ratios as described above.
This method is described in Appendix B.
-------
77
F. SULFATE SENSITIVITY BY STANDARD SUBTRACTION
1. Calculations of Voltage Response of Sulfate Addition
The above calculations (Section D) had neglected the dilution effect
of the sulfate solution, and were based on 20 ml of Pb(C104)2. The calculated
voltage shift for the sulfate additions used in the study to be described below
are given in Table XXIX. In these tests, 10 ml of sulfate solution was added to
15 ml of lead solution. Since the ratios of Pb/S04 solutions remain constant,
the calculations can be used at each lead level.
TABLE XXIX
VOLTAGE SHIFT FOR VARIOUS SULFATE INCREMENTS
ml (S04) *
0
1.0
1.5
2.0
2.5
3.0
3.5
* Balance of 5 ml is water.
Pb
(Pb-S04) V^Vg
1.67
2.28
2.78
3.33
5.00
8.30
21.3
Also, 5 ml of MCA is added.
mv
Change
6.4
10.4
12.8
15.0
20.3
26.6
38.5
2. Results of Sulfate Analysis
The sulfate procedure described in Appendix B was tested on sulfate
standards. The experiment was performed on a random order design.
The data of the sulfate study are presented in Figures 17-20. It
1s evident that in many Instances the experimental curves do not follow the
theoretical curve. We believe this is due to an error in.the equivalence of the
Ulalden
-------
38
34
30
26
o
•* 22
o
o>
ua
n
18
14
10
Theory
co
3.5 ml Na2S04
0.5 1.0 1.5 2.0 2.5 3.0
Figure 17, Addition of 4 x 10"4M Na2S04 Solution to 10"4 Pb(C10A)2 Solution
-------
38
34
30
± 26
_j»
o
rt-
5 22
18
14
10
<£>
0.5
1.0
1.5
2.0
2.5
3.0
3.5
ml
Figure 18. Addition of 1.2 x 10'3M Na2$04 Solution to 3 x 10'4M Pb(C104)2 Solution.
-------
A mv
38
34
30
26
22
18
14
10
CO
o
0.5 1.0 1.5 2.0 2.5 3.0 3.5 ml
Figure 19, Addition of 4 x 10"^ Ha2S04 Solution to 10"^ Pb(C104)2 Solution.
-------
38
34
30
26
22
18
10
03
*t
1 »
1 1 1 1 L
0.5 1.0 1.5 2.0 2.5 3.0 3.5 ml
Figure 20. Addition of 1.2 x 10"2M Na2S04 Solution to 3 x 10"3M Pb(C104)2 Solution.
-------
82
lead and sulfate solutions. We anticipate that the theoretical curve can be
used for calibration purposes when this problem is eliminated.
The experiment was performed in a random order. However, the ex-
periments using 3.5, 2.5 and 1.5 ml of Na^SO. solution were performed on days
1-2 and the 3.0, 2.0 and 1.0 ml quantities later, on days labeled day 3-4.
The average difference in millivolts of the non-consecutive duplicates was
1.7 mv. This can be converted to a standard deviation of 1.49 mv (Moroney, 1956'
for the method. The source of this error is analyzed in Table XXX. It is evident
that the millivolt error is independent of the fraction of lead precipitated.
Hence, the above analysis of the sulfate error, as a function of the millivolt
error, is valid. There does appear to be a difference between day 1-2 and day
3-4. Also, the error at the low sulfate (lead) concentration is greater than
at the higher concentration levels.
There is no effect of test order on abnormally high and low mv values-
This was analyzed on days 3-4 and the number in Figures 17-20 refer to this test
order.
TABLE XXX
EVALUATION OF SULFATE METHOD
•
Day 1-2
Day 3-4
High frac.
Low frac.
Medium
Low cone.
Low cone.
Low cone.
High. cone.
Overall
3.5-2.5-1.5 ml
3.0-2.0-1.0 ml
3.5-3.0 ml
1.0-1.5
2.0-2.5
4 x 10"4M
1.2 x 10"3M
4 x 10"3M
1.2 x 10"2M
N
02)
(12)
(8)
(8)
(8)
(6)
(6)
(6)
(6)
(24)
Range Cmv)
1.19±0.88
2,17±1.49
1.86±1.03
1.64±1.86
1.55±0.98
2.40±1.81
1.7511. 01
1.42+1,39
0.83±0.50
1.68±1.3
S.D. (mv)
1.05
1.92
2.13
1.55
K26
0.74
1.49 , — •
-------
83
G. GRAN'S PLOT PROCEDURE
The major problem with the standard subtraction method is the loss of
sensitivity due to the larger volume of lead solution versus the unknown sulfate
solution.
Another method, which might improve the precision, as well as increase
the sensitivity of the analysis, is the use of Gran's plots to obtain a titration
end point. In this method small volumes of increasingly concentrated Pb (C104)2
solutions would be added until excess lead is sensed. The system could then be
adjusted to a standard volume and increments of the final Pb (C104)2 solution
would be added. This method was simulated by calculating the response to an
arbitrary random sulfate solution. The method and results are described below.
We used the same solutions which would be required for the standard subtraction
method.
In the simulated procedure, 9 ml of unknown (1.23.x 10" M equals l.lly
moles sulfate) were placed in test vessel with 9 ml MCA, and the following
solutions were added:
0.1 ml 1 x 10"4M Pb (C104)2 + 0.1 ml MCA = O.Oly moles added
•' 3 x 10'4M " • O-O4
" 1 x 1Q-3M " " °-14
n 3 x 10-3M " = O-44
„ ! x 10-2M n = 1.44 - excess Pb** indicated
by voltage jump
Following tMs we assume addition of 0.5 ml HgO + 0-5 ml MCA to bring volume
to 20 ml. Read meter (e.g., 15.1 mv).
Add 0.1 ml of 1 x 10"2M Pb (C104)2 + 0.1 ml MCA Read (e.g., 32,6 mv}
0.1 ml of 1 x 10"2M Pb (C104)2 + 0.1 ml MCA Read (e.g., 39.6 mv}
0.1 ml of 1 x 10'2M Pb (C104)2 + 0.1 ml MCA Read (e.g., 44.2 mv}
0.1 ml of 1 x 10'2M Pb (C104)2 + 0.1 ml MCA Read (e.g., 47.3 mv)
Plot data on 10 percent volume corrected Gran's plot paper. Extrapolate to
Illlalden
-------
84
equivalence (0.067 ml - Figure 21) and add y moles added of more dilute
solutions (0.44 y moles).
Calculate quantity
0.067 ml x 1 x 10"2M = 0.067 y moles + 0.44 y moles = 1.11 y moles/9 ml
In practice the error involved in extrapolating Gran's plots would be
expected to be small compared to the back calculation from two points and a cal-
culated slope (standard subtraction). Also the Gran's plot method eliminates a
search for the correct lead solution. Simplification of the above procedure woul
be addition of 10 ml of unknown. The error in extrapolating 0.1 ml additions to
10.1 - 10.7 ml volumes, versus a 10 ml volume, is minimal on Gran's paper. The
data calculated for 10.5 ml of solution above is also plotted on Figure21 . The
extrapolated value showed no significant error.
H. GRAN'S PLOT EVALUATION
Twelve solutions of five different concentrations of Na2$04 were ^
analyzed using the procedure described above. Sulfate solutions between 1 x 10
to 3 x 10"3M were measured. The average percent error of these 12 analyses was
35%. Eliminating the 1 x 10"5M concentration the error was 30%. These data are
shown in Table XXXI (Series 1).
Since the Oxford pipette (0.2 ml, 50% unknown + 50% MCA) appeared some-
times to give false volumes the study was repeated at three concentrations using
20 times the volume and using glass pipettes. However the error remained about
the same (24% average error, Series 2).
Hence we cannot recommend this method of analysis. Therefore the
standard subtraction method described above was used for the analysis of high
volume sample filters, since this method showed better precision on known
samples.
-------
10 ml + 0.1 ml addltio
intercept = 0.067 fll x 10-2M = |0.67y moles
to 10.5 ml +0.1 ml addition
A = 0167 + 0.44 -p.llp moles
B = OL79 + 0.44 = 1.23u moles
-------
86
TABLE XXXI
EVALUATION OF GRAN'S PLOT METHOD FOR
DETERMINING SULFATE IN KNOWN STANDARDS
Molar Concentration % Error
Calculated
Series 1:
IxlO"5
IxlO"5
IxlO"4
IxlO"4
5x1 O"4
5x1 O"4
5x1 O"4
IxlO"3
IxlO"3
IxlO"3
3x1 O"3
3x1 O"3
Series 2:
5x1 O"5
2.5xlO"4
5x1 O"4
Observed
10 ml total sample (0.1
solution)
1.6xlO"5
1.6xlO"5
7.0xlO"5
8.6xlO"5
3.5xlO"4
5.3xlO"4
6.3xlO"4
6.5xlO"4
6. IxlO"4
6.5xlO"4
1.5xlO"3
Z.OxlO"3
50 ml total sample
6.3xlO"5
3.3xlO"4
4.4xlO"4
ml unknown + 0.1 ml MCA, to 10 ml lead
+60
+60
-30
-14
-31
+6
+26
-35
-39
-35
-50
-35
+26
+32
-13
-------
87
I. SULFATE MEASUREMENT OF HIGH VOLUME FILTERS
Separate sections of filters described in Section V.G, labeled S-l
through S-9, were analyzed for sulfate by the electrode and barium chlor-
anilate (Driscoll et. al., 1972) procedures. A 2.5 cm x 19 cm section of filter
was used and the filtrate was diluted to 20 ml. This allowed both procedures
to be run on the same extracted sample. However, it lowered the quantity of
sulfate in each procedure by 50%. Three concentrations of Pb(C104)2 solution
were required, and each concentration was titrated with sulfate to produce a
standard curve. However the data indicated that this is not required.
The data are presented in Tables XXXII and XXXIII and Figure 22. It is prob-
able that the filter blanks as determined by the barium chloranilate procedure are
too high. We tend to believe the electrode blank determination. However there
is also a discrepancy between the methods for S-7 and S-9. The electrode gives
the higher value in these cases. However it is not known which method is
correct. From an analysis of the sample locations, both filters came from
Birmingham, Alabama. We believe that the error may be caused by metal ions
which precipitate the chloranilate ions and generate low values for sulfate by
this method. Al+++, Ca++, Fe+++, Cu++ and Zn++ are known to interfere in this
manner:
Ba Chi + S04= -* BaS04 + Chl= (530 nm color)
Chl= + M++ •* M Chi
Adding a cation exchange step is reported to eliminate this problem. Al-
ternately, we may be precipitating an anion with lead which is soluble with
barium ion. This appears to be less likely, although solubilities of lead
salts in 50% MCA are not known.
With these two exceptions we have obtained an excellent one-to-one
correlation between the electrode and colorimetric methods. The percent
standard deviation between the methods is 9.5% (Table XXXIV). This indicates
that in general we are measuring total soluble sulfate in spite of adding 50%
methyl cellosolve acetate.
UlMen
-------
88
TABLE XXXII
ELECTRODE DATA FOR HI VOL SULFATE MEASUREMENT
Sample
S-l
S-2
S-3
S-4
S-5
S-6
S-7
S-8
S-9
Blank
Blank
% of
Extract
5
5
5
5
5
5
5
5
5
25
25
Initial Pb++
Molar ity*
3x1 O"4
IxlO"4
IxlO"4
IxlO"4
IxlO"4
3x1 O"4
IxlO"4
IxlO"4
3x1 O"4
3x1 O"5
3x1 O"5
AE
mv
13.4
17.2
15.4
14.6
16.9
16.0
12.9
13.8
13.8
23
18
% of Pb++
Precipitated
44
60
54
50
59
52
45
49
46
96
85
* 15 ml of 50% MCA Solution
-------
89
Sample
S-l
S-2
S-3
S-4
S-5
S-6
S-7
S-8
S-9
Blank
Blank
TABLE XXXIII
COMPARISON OF SULFATE ON HI VOL FILTERS BY
ELECTRODE AND BARIUM CHLORANILATE METHODS
Sulfate Quantity (ug/48
Electrode
3800
1710
1570
1440
1690
4490
1300
1400
3930
164
140
cm2)
Colon' metric
3910
2050
1650
1260
1620
4700
552
1560
1760
628
1080
-------
90
5 -
2 3 4
Barium ChlorantUte, mgCSO^l/46 cm2
FIGURE 22. COMPARISON OF ELECTRODE AND COLORIMETRIC METHODS
-------
91
TABLE XXXIV
CORRELATIONS OF SULFATE ANALYSIS
Data
all
excluding
S-7, S-9 and
Blanks
Correlation
Coefficient
0.82
0.992
Linear Regression
y=electrode in mg
x=Chloranilate in mg
y = 0.805x + 665
y = Q.945x + 0.039
Std. Dev. * Ave. Dev. x
Assumes: Electrode (y) = Colorimetric (x)
S-l through S-9
Excluding S-7 and S-9
Std. Dev. (tng)
510
185
Std. Dev.
27.0
9.5
-------
92
With the use of ultrasonic agitation in a test tube and use of an
acid washed filter press the method is relatively simple. By use of a third
of the required sample, and a preliminary electrode measurement, the range of
the method may be expanded to a decade without loss of sensitivity. The final
reading is taken after addition of more sample or solvent (see Procedure and
Recommendations). This method will also indicate whether the sample is outside
the decade range, by use of only a third of the normal sample.
Obviously the method could be made more precise by using a smaller
range, with more lead standards. Also the method can be made more sensitive
by decreasing the total volume. The present system uses 25 ml of solution.
If the samples contain 1 mg or more sulfate however, increased sensitivity is
not required.
-------
APPENDIX A
RECOMMENDED PROCEDURE FOR NITRATE ANALYSIS
Preparation of Standards
1. One liter of KNOg solutions
lx!0"5M, 3xlO"5M, lxlO~4M, 3xlO~4M, lx!0"3M
using KNO^ Baker Analyzed Reagent and deionized water.
2. Add 40 ml of 10 M AgF solution (Alpha Organics, Ventron, Beverly,
Mass.) to each solution.
3. Store in brown polyethylene bottles.
-------
Preparation of Samples
1. Cut 10% of the exposed area of Hi Vol filter. Handle the filter
with gloves.
2. Place in 16 mm test tube.
3. Add 15 ml of HgO.
4. Break sample with metal spatula.
5. Place in ultrasonic bath for 5 minutes.
6. Press out liquid using Filter/Sampler, Unichem P5190-6, Scientific
Products, Evanston, 111.
7- Transfer this sample into 50 ml Tri Pour Beaker.
8- Add 0.6 ml of 0.1M AgF.
9. Use same- procedure for unused Hi Vol filter for blank determination.
Test Procedure
1. Place M5 ml of each standard (lx!0'5M to lx!0"3M), sample, blank,
etc., Into 50 ml disposable beaker (Tri Pour, VWR).
2. Insert clean magnetic stirrer.
3. Stir to release air bubbles.
4. Insert Orion nitrate and fluoride electrode assembly.
5. Read Orion Model 701 meter on millivolt scale after 2 minutes.
-------
Wash Procedure
1. After each determination wash the electrodes and stirrer and blot
dry with a Kim-Wipe.
Calculations
Plot standard curve from 10"5 - 10"3 M KNOg on semi-log paper. Read
unknown concentration from eurve. Calculate mg from concentration
mg = M x 930, and multiply by 10 for total mg on filter. Or calibrate
ordinate of curve in required units.
-------
APPENDIX B
RECOMMENDED PROCEDURE FOR SULFATE ANALYSIS
1. Solutions
a- Stock Solutions (in water)
IxlO'2 m Pb(C104)2
3x1O"4 m Na2S04
9x1O"4 m Na2S04
3xlO"3 m Na2S04
9x1O"3 m Na2S04
b. Pb(ClQ4)2 Solutions (prepared daily)
IxlO"4 m Pb(Cl04)2 = 2 ml 10"2 m Pb(Cl04)2
+ 98 ml H20
+100 ml MCA
3xlO~4 m Pb(C104)2 = 6 ml 10"2 m Pb(C104)2
+ 94 ml H20
+100 ml MCA
IxlO"3 m Pb(C104)2 = 20 ml 10"2 m Pb(Cl04)2
+ 80 ml H20
+100 ml MCA
3x1O"3 m Pb(C104)2 = 60 ml 10"2 m Pb(C104)2
+ 40 ml H20
+100 ml MCA
c. Na2S04 Solutions (prepared as knowns for calibrations as necessary)
From each of the four Na2S04 solutions prepare standards to
precipitate 0.3, 0.5, 0.7 and 0.9 of Pb++.
-------
Thus for 15 ml of IxlO"4 m Pb(C10J2, prepare
4.5 ml of 3xlO"4 m Na2S04 + 0.5 ml HgO + 5 ml MCA
3-5 ml 1.5 ml 5 ml
2-5 ml 2.5 ml 5 ml
1 prepare
4.5 ml of 9xlO"4 m Na2$04 + 0.5 ml Hg) + 5 ml MCA
3-5 ml 1.5 ml 5 ml
2-5 ml 2.5 ml 5 ml
1<5 ml 3.5 ml 5 ml
For 15 ml of IxlO"3 m Pb(C104)2, prepare
4.5 ml of 3x1O"3 m Na2S04 + 0.5 ml HgO + 5 ml MCA
3-5 ml 1.5 ml 5 ml
2-5 ml 2.5 ml 5 ml
]-5 ml 3.5 ml 5 ml
For 15 ml of 3x1O"3 m Pb(Cl04)2, prepare
4.5 ml of 9x1O"3 m Na2S04 + 0.5 ml H20 + 5 ml MCA
3-5 ml 1.5 ml 5 ml
2-5 ml 2.5 ml 5 ml
T-51"1 3.5ml 5ml
2, Samples
Identical to nitrate, steps 1-6.
Take aliquot of filtrate and add an equal volume of MCA.
3. Test Procedure
Pipette 15 ml of Pb(C104)2 solution into 50 ml Tri Pour beaker-
-------
Insert clean magnetic stirrer.
Stir to release air bubbles.
Insert Orion lead and double junction reference electrodes.
Read meter (mv) after 2 minutes.
Add 10 ml of sample (5 ml sample + 5 ml MCA)*.
Read meter (mv) after 2 minutes.
If deflection <10 mv use less concentrated PbCClO^
If deflection >38 mv use more concentrated Pb(C104)2
d. Wash Procedure
Same as nitrate procedure.
e. Calculations
Standard curve(s) of AE vs sulfate as fraction of lead used,
* In the analysis of unknowns, 1 ml of solution was added and an addition of
4 ml water or 4 ml sample was based on this initial deflection.
-------
REFERENCES
1 Driscoll, J., J. Becker, R. Hebert, K. Horbal, M. Young, (1972),
"Validation of Improved Chemical Methods for Sulfur Oxides Measure-
ments from Stationary Sources", EPA-R2-92-105.
2. Manahan, S.E., (1970), Anal. Chem.. 42, 128.
3. Milham, P.O., A.S. Awad, R.E. Paul!, J.H. Bull, (1970), Analyst, 95, 751.
4. Milham, P.O., (1970), Analyst. 95., 758.
5. Moroney, (1956), "Facts from Figures", 3rd. Ed., Penguin Books,
New York, P. 155.
6. Orion Newsletter, "Selectivity Constants and Electrode Interferences",
October 1969, Vol. 1.
7. Orion Research Instruction Manual, "Nitrate Ion Electrode", (1971), p. 10.
8 Orion Research Instruction Manual, "Lead Electrode Model 94-82", (1972),
pp. 17-19.
9. Rechnitz, G.A., G.H. Fricke and M.S. Mohan, (1972), Anal. Chem.,
44, 1098.
10. Rechnitz, G.A., Mohan, M.S., (1973), Anal. Chem., 45, 1323.
11. Ross, J.W., M.S. Frant, (1964), Anal. Chem.. 41_, 967.
12. Sawicki, C.R., R.P. Scaringelli, (1971), Microchemical Journal, 16^657.
13. Thompson, R., (1972), Private Communication, Environmental Protection
Agency, Research Triangle Park, North Carolina 27711.
------- |