702 Analyst, Juune, 1981, Vol, 106, pp . 702-709

Determination of Anions in Atmospheric Precipitation by Ion Chromatography Joan Crowther and Jenifer McBride Ontario Ministry of the Environment, Laboratory Services Branch, Water Quality Section, Box 21 3, Rexdale, Ontario M9W 5L1, Canada

The semi-automated ion-chromatographic system studied permitted analyses of 60 samples per day for the three major anions found in atmospheric precipitation, viz., sulphate (0.06-10.0 mg l-l), nitrate (0.022-2.00 mg 1-1 as nitrogen) and chloride (0.028-1.50 mg 1-I). The sensitivity was too low to determine nitrite, orthophosphate, bromide and sulphite under the selected operating conditions ; however, the expected concentrations of these species were less than 0.1 mg 1-1. Fixed point calibrations were invalid as cali- brations were non-linear. When the subject system was compared with alternative procedures, the resultant data for sulphate, nitrate and chloride agreed to within 2%. Ion-chromatographic data for fluoride were un- acceptable, as shown by the inter-comparison with two reference procedures.

Keywords : Atmospheric precipitation analysis ; ion chromatography ; sulphate determination ; nifrate determination ; chloride determination

Small et a1.l introduced an ion-chromatographic system, which was based on the use of ion-exchange resins and conductimetric detection, for the determination of ionic species in aqueous solution. Since then, commercial units, produced by Dionex Corp., have gained ac~eptance ,~r~ particularly for sulphate determinations. Moreover, these ion-chromatographic instruments have been employed for determining some of the ionic species found in atmos- pheric precipitation4 where ion concentrations are significantly lower than those found in surface waters. As current studies on the effects of acidic rain require accurate and precise analysis, a detailed study was undertaken to evaluate the suitability of this instrumentation for determining anions in precipitation samples on a routine basis.

Experimental For Dionex ion-chromatographic systems, the eluent is pumped continuously, either

through the sample loop or directly to the columns. When determining anions, these columns consist of a pre-column (optional but advisable to screen foreign matter from the sample before it reaches the expensive separator column), the separator column packed with a proprietory ion-exchange resin and the suppressor column packed with a strong acid ion- exchange resin. By selecting an appropriate eluent, the anions of interest are separated according to size and charge in the separator column, converted into their acid form in the suppressor column and determined by comparing the resultant conductivities of these acids with those obtained for mixed standards. As the eluent flows continuously, the acid form of the eluent must be a comparatively poor conductor.

Apparatus The Dionex, Model 10, ion-chromatographic unit with pre-packed columns was tested as

received, and also after modification to include the addition of an automated sampling and eluent-spiking train. This train consisted of a sampler (a Technicon large industrial model modified to accept timing signals from a Supergrator 111), a peristaltic pump (Technicon, Model MI), a relay box (in-house design) and a programmed timer (Columbia Scientific Industries Supergrator 111). Automation also entailed replacing the three-way sampling valve of the Dionex unit with a four-way valve. The manifold for automated sample pre- treatment is illustrated in Fig. 1. Two lengths of separator columns were tested, 250 and 500 mm. All valves in the Dionex instrument were activated by air (80 lb in-2). Results were recorded on a two-pen Linear, Model 585, chart recorder with the signal to the second pen expanded two-fold to obtain a second analytical range. AutoAnalyzer systems were utilised as support systems to evaluate ion-chromatographic data.

Publ

ishe

d on

01

Janu

ary

1981

. Dow

nloa

ded

by U

nive

rsity

of

Cal

ifor

nia

- Sa

nta

Cru

z on

26/

10/2

014

01:4

9:49

. View Article Online / Journal Homepage / Table of Contents for this issue

CROWTHER AND MCBRIDE

20 turns 2.00

TO3

Re-sample 0.32 b To IC analyser

Air

Sample

Eluent spike B; 0.10 1 1

Fig. 1. Manifold for automated sampling train of ion-chromatographic Air supplies to proportioning pump and eluent spike B reservoir system.

were drawn through sodium hydroxide solution to remove carbon dioxide.

Reagents

were prepared with de-ionised, distilled water. 0.3 to 0.7 pS cm-l.

EZuent. carbonate.

EZuent spikes. eluent. and 0 . 2 4 ~ in sodium carbonate. sodium hydrogen carbonate and 0.048 M in sodium carbonate.

All chemicals were of analytical-reagent grade, unless otherwise stated, and all reagents The conductivity of this water ranged from

This solution was 0.003 M in sodium hydrogen carbonate and 0.0024 M in sodium

For sample pre-treatment these were more concentrated solutions of the For manual operation, eluent spike A was 0.30 M in sodium hydrogen carbonate

For automated sampling, eluent spike B was 0 . 0 6 ~ in

Szdphuuric acid, 1 N. The acid regenerant for the suppressor column. S o d i u m hydroxide solution, 6 N. Utilised to scrub carbon dioxide from the air supply to

the sampling manifold. Standard solutions. Prepared from sodium fluoride, sodium chloride, sodium nitrite ,

potassium dihydrogen orthophosphate, sodium sulphite nonahydrate, potassium nitrate and sodium sulphate.

Procedure The general procedure for the determination of anions using Dionex instruments is described

in the 1iteraturelJ and in the manufacturer’s manual, and the selected operating conditions for this ion-chromatographic system are listed in Table I.

The eluent was pumped continuously, either through the sample loop to the columns (when analysing) or directly to the columns (between analyses). To ensure that the back- ground concentrations of hydrogen carbonate and carbonate remained relatively constant in the system, samples were pre-treated by spiking with more concentrated solutions of the eluent. For manual operation, 0.50 ml of eluent spike A was mixed with 50.0 ml of sample. When the sampling train was automated, this spiking function was handled by the manifold illustrated in Fig. 1. The system was calibrated with one blank and six mixed standards containing all anions of interest except sulphite, which required a separate calibration, as described later. Although blank corrections were not required in the determination of anions, blanks of de-ionised, distilled water were analysed daily to check that the ion-chromato- graphic system was not contaminated. To check the calibrations, two quality control solutions were analysed with each run.5 These quality control solutions were mixed standards prepared from different batches of chemicals ; fresh solutions were prepared before old stocks were exhausted, and both sets of solutions were analysed duiing a 3-day changeover period. To monitor in-run sensitivity changes, one mixed standard was analysed after each group of five samples. Anion concentrations were determined from manually plotted calibration graphs based on peak height. After about 10 h of operation in the analytical mode, the suppressor column had to be regenerated, and the system then re-calibrated.

Publ

ishe

d on

01

Janu

ary

1981

. Dow

nloa

ded

by U

nive

rsity

of

Cal

ifor

nia

- Sa

nta

Cru

z on

26/

10/2

014

01:4

9:49

. View Article Online

704

Time cycles . . Eluent ..

Sample . .

Pre-column . .

CROWTHER AND MCBRIDE: DETERMINATION OF ANIONS IN Analysf, Yo;. 106

TABLE I OPERATING CONDITIONS FOR ION-CHROMATOGRAPHIC SYSTEM

Separator column . . Suppressor column . .

Detector . . . . Chart recorder . . Calibration . . . .

..

..

..

..

.. ..

..

. .

..

Analytical mode: 9-10 h Regeneration mode: 0.5-1 h Composition: 0.003 0 M in NaHCO, and 0.0024 M in Na,CO, Flow-rate: 165 and 180 ml h-l for 500- and 250-mm separator

Manual pre-treatment : mix 50.0 ml of sample with 0.50 ml of eluent

Automated pre-treatment : mix sample with eluent spike B, 2.00 + Sample volume: 0.30 ml. Size: 3 mm i.d. x 126 mm Purpose: trap foreign matter Size: 3 mm i.d., length 250 or 500 mm Size: 9 mm i.d. x 250 mm Regenerant: 1 N H,SO, Regeneration cycle: acid a t flow-rate of 280 ml h-' for 10 min

followed by water a t flow-rate of 280 rnl h-l for a t least 30 min Conductivity meter: 10 $3 cm-l full scale Pen 1 : high analytical range, full scale equivalent to 10 pS cm-l Pen 2: low analytical range, full scale equivalent to 5 pS cm-l Blank plus 6 mixed standards Calibration graphs based on peak height

columns, respectively

spike A

0.10 v/v

When sulphite was being determined, fresh sulphite standards were prepared and analysed. Calibrations had to be corrected for the presence of sulphate as this was invariably present in the sulphite standards.6 Each sample was analysed twice: as received and after pre- treatment with permanganate (1 OOO mg 1-1 as manganese). One drop of the permanganate solution was mixed with a 50-ml sample; the second analysis differentiated between bromide and sulphite.

Results and Discussion Preliminary Study

Calibrations for individual anions were non-linear, and the errors arising from fixed-point calibrations were approximately 10%. The commercial ion-chromatographic unit was equipped with a multi-range conductivity detector, and full scale for the system could be varied from 0.1 to lo00 pS cm-l. The 3, 10 and 30 pS cm-1 scales were compared and found to be compatible to within 2%. Thus, one could calibrate the 10 scale and then determine concentrations using either the 3 or the 30 scale by applying the appropriate factor and by correcting for the observed change in base line on the chart recorder. This facet of the ion-chromatographic unit was not pursued further, as one scale was found to be adequate for the major anions in question. In this laboratory, sensitivity fluctuations due to atmospheric temperature changes were adequately controlled by keeping the system away from direct sunlight and draughts.

As sulphate is the dominant anion in atmospheric precipitation, the operating conditions for the ion-chromatographic system (Table I) were selected such that one line on the chart- recorder paper represented approximately 0.05 mg 1-1 of sulphate. As full scale for the conductivity detector was maintained at 10 pS cm-l, these operating conditions determined the calibration ranges for all of the anions. Calibration ranges and retention times for two lengths of separator column are summarised in Table 11, and Figs. 2 and 3 show typical chromatograms. Note the comparative loss in sensitivity for fluoride and chloride relative to nitrate and sulphate when the shorter separator column was employed.

Bromide and sulphite had the same retention time (6.2min) when the length of the separator column was 500mm. These anions were differentiated by analysing the sample twice, as received and after mixing with permanganate. Under the specified conditions, bromide was not affected by this pre-treatment, but sulphite was quantitatively oxidised to subhate.

Publ

ishe

d on

01

Janu

ary

1981

. Dow

nloa

ded

by U

nive

rsity

of

Cal

ifor

nia

- Sa

nta

Cru

z on

26/

10/2

014

01:4

9:49

. View Article Online

June, 1981 ATMOSPHERIC PRECIPITATION BY ION CHROMATOGRAPHY 705

TABLE I1 ANALYTICAL RANGES AND RETENTION TIMES

Analytical range/ Anion mg 1-1

F- . . . . . . 0.014-0.400 c1- . . . . 0.028-1.50 NO,- as" . . . . 0.01-0.40 P043- as P . . . 0.08-0.40 Br- . . . . . , 0.04-2.50

. . . . 0.08-10.0

. . . . . . 0.056-10.0 NO,- as" . . . . 0.021-2.00

Retention times for different column lengths/min

250-mm column 500-mm column 1.6 1.6 2.2 2.5 - 3.1 - 4.4 - 6.2 - 6.2 4.5 7.3 6.1 10.4

I A 1

Approximately 200 precipitation and snow samples from southern and central Ontario Most of these samples were collected from ,areas affected by acidic

All of the peaks of the chromatograms were well formed and cleanly separated; there were then analysed. ain.

80

60 4- .- C 3

2 E

+?

E 40

c. .-

0,

0 1

Y m 0

.-

20

0

CI i so4

80

ul

C +- .-

60 L- E +? c. .-

4 2 1 0, W .- 5 40 m a

20

0

CI

U

N

0 4 8 12

Retention tirne/min

0 4 8

Retention time/ min

Fig. 2. Ion chromatogram obtained using Fig. 3. Ion chromatogram obtained 500-mm separator column.

were no extraneous peaks. The distribution of anion concentrations in these samples (Table 111) showed that more than 96% m/V of the anions were in the sulphate, nitrate and chloride forms, and that the selected concentration ranges for these three anions were appropriate. For nitrite, orthophosphate, bromide and sulphite, the sensitivity of the

using 250-mm separator column.

Publ

ishe

d on

01

Janu

ary

1981

. Dow

nloa

ded

by U

nive

rsity

of

Cal

ifor

nia

- Sa

nta

Cru

z on

26/

10/2

014

01:4

9:49

. View Article Online

706 CROWTHER AND MCBRIDE: DETERMINATION OF ANIONS IN Analyst, Vol. 106 designed system was inadequate as the measured concentrations of these anions were less than 0.1 mgl-l. Changing the scale of the conductivity detector in mid-analysis was not attempted. To continue analysing for all eight anions was not economical as analysis of one sample required 15 min. By reducing the length of the separator column from 500 to 250mm, fluoride, chloride, nitrate and sulphate still could be determined, but the analysis time was reduced to 8min. This column change was therefore made, the sampling train was automated and then this modified system was evaluated.

TABLE I11 DISTRIBUTION OF ANION CONCENTRATIONS I N PRECIPITATION SAMPLES

Distribution of concentrations, %

Anion F- . . .. c1- . . .. NO,- as N . . Br- . . . . NO,- as N . .

PO,,- as P .. so,'- . . .. SO,%- . . ..

Detection criteria/ mg I-'

. . 0.014

. . 0.028

. . 0.01 . . 0.08

. . 0.04

. . 0.08

. . 0.021

. . 0.056

<0.1 mg 1-l

83.7 3.5

100 100 100 99.0 3.0 0

0.1-0.5 mg 1-l 16.3 44.4 0 0 0 1 .o

39.4 0.8

0.5-1.0 mg 1-l

0 18.4 0 0 0 0

42.0 4.2

1.0-2.0 mg 1-1

0 18.4 0 0 0 0

13.6 28.4

2.0-5.0 mg 1-l

0 8.3 0 0 0 0 2.0

48.1

> 5.0 mg 1-l

0 7.0 0 0 0 0 0

18.5

Calibration artd precision The daily calibration graphs for the four test parameters varied slightly in sensitivity

(5-10yo full scale) and in degree of curvature. To check the validity of manually drawn calibration graphs, two quality control solutions were included in each run (Table IV). Over a 6-week period, the average recovery of the four anions at two concentration levels ranged

TABLE IV CALIBRATION CONTROL

Parameter

Anion

F- C1- NO,- as N SO,%- A

f \

No. of measurements . . . . . . . . 18 33 33 33 Quality control mixed standard, 80% full scale-

Theoretical concentration/mg 1-' . . . . 0.320 1.20 1.60 8.00 Average measured concentration/mg 1-l . . 0.320 1.20 1.60 8.00 Standard deviation/mg 1-l . . . . . . 0.0055 0.012 0.006 0.036 Relative standard deviation, % . . . . 1.7 1 .o 0.4 0.45

Quality control mixed standard, 20% full scale- Theoretical concentration/mg 1-l . . . . 0.080 0.300 0.400 2.00

Relative standard deviation, yo . . . . 3.6 3.8 1.7 2.0

2.03 Average measured concentration/mg 1-l . . 0.081 0.299 0.406 Standard deviationlmg 1-l . . . . . . 0.0029 0.0113 0.0067 0.039

,--

from 99.8 to 101.5y0, and the relative standard deviations ranged from 0.4 to 3.7%. Precision was also estimated by analysing routine precipitation samples in duplicate. Table V shows within-batch standard deviations for four anions with the results grouped according to sample

TABLE V DUPLICATE ANALYSES OF PRECIPITATION SAMPLES

Anion F- . . .. c1- . . .. NO,- as N . .

SO,%- ..

Sample concentration range/mg 1-l

. . (0.100

. . <0.20

.. (0.20

. . (2.00

0.20-0.50

0.20-0.50 0.50-1 .OO

2.00-5.00 5.00-10.0

No. of samples analysed

49 28 26 10 29 25 11 45 Q

Standard deviation/ mg 1-1

0.008 69 0.0170 0.0165 0.0126 0.0124 0.0159 0.033 7 0.077 5 0.167

Publ

ishe

d on

01

Janu

ary

1981

. Dow

nloa

ded

by U

nive

rsity

of

Cal

ifor

nia

- Sa

nta

Cru

z on

26/

10/2

014

01:4

9:49

. View Article Online

June, 1981 ATMOSPHERIC PRECIPITATION BY ION CHROMATOGRAPHY 707

concentration. From the standard deviations for samples in the lowest concentration range and the appropriate Student's t-value, detection criteria (95% confidence limits) were calcu- lated as follows: sulphate, 0.058 mg 1-l; nitrate, 0.022 mg 1-1 as nitrogen ; chloride, 0.028 mg 1-l; and fluoride, 0.014 mg 1-l.

Interference study De-ionised, distilled water and mixed standards (50% full scale) were spiked with other

chemical species a t concentrations in excess of those expected in precipitation samples and most surface waters. The data (Table VI) showed that the recovery of anions was not affected when the pH of mixed standards was varied over the pH range 3-10. Solutions of a technical-reagent grade of lignin sulphonate produced peaks on the chromatograms, but the shape of these peaks indicated that they were impurities in the test chemical; all four anions were recovered from lignin sulphonate solutions provided corrections were applied for the presumed impurities. The retention time for formic acid was identical with that for fluoride, and hence this acid is an interferent; fluoride could be recovered from solutions of formic acid provided that a suitable correction was applied.

TABLE VI INTERFERENCE STUDY IN DETERMINING ANIONS USING ION-CHROMATOGRAPHIC PROCEDURE

Potential interferent Measured anion concentrations -7 in test material/mg l-l Recovery of added anions, yo"

Concen- I A \ I A \

trationl NO,- NO,- Test material mgl-' F- C1- as N F- C1- as N SO,*-

Acid or base?- p H 3 . . . . . . - 0 0 0 0 98 98 99 100 p H 6 . . . . . . - 0 0 0 0 102 100 99 100

- 101 p H 9 . . . . . . - 0 0 0 0 pH 10 . . . . - 0 0 0 0 100 100 100 100

Aluminium(II1)~ . . 2 0 0 0 0 100 100 100 99 5 0 0 0 0 93 101 100 102

Iron(II1): . . . . 2 0 0 0 0 98 - 98 100 Manganese(VI1): . . 2 0 0 0 0 98 98 100 100 Carbonate asCaCO, . . 50 0 0 0 0 104 99 100 99

100 0 0 0 0 100 100 100 101 Formic acid . . . . 1 0.43 0 0 0 100 100 100 100 Oxalic acid . . . . 10 0 0 0 0 102 100 100 100 Humic acids . . . . 10 0 0 0 0 98 101 100 100

20 0 0 0 0 100 101 100 100 50 0 0 0 0 100 101 99 101

Lignin sulphonates . . 25 0.100 0.09 0 0.5 101 102 101 98 50 0.200 0.17 0 0.9 96 99 101 100

Tannic acids . . . . 10 0 0 0 0 102 101 100 101 20 0 0 0 0 102 - 101 101 50 0 0 0 0 105 100 101 102

- -

* Anion concentrations added were: 0.200 mg 1-' F-, 0.75 mg 1-' C1-, 1.00 mg 1-1 NO,- as N and

7 p H values for test solutions were adjusted with dilute sodium hydroxide solution and either hydro-

The salts of aluminium(III), iron(II1) and manganese(VI1) tested were aluminium sulphate or aluminium

5.00 mg 1-'

chloric or sulphuric acid.

chloride, iron( 111) chloride and potassium permanganate. 5 Test materials were of technical grade.

Comparison of methods Precipitation samples were analysed by the ion-chromatographic system and by alternative

methods (Table VII). The resultant data were evaluated by linear regression analyses: y on x where y referred to the ion-chromatographic results. Next, the data were tested for bias (y - x) associated with anion concentration level.

For sulphate, 218 precipitation samples were analysed by the ion-chromatographic system and by an automated version of the methylthymol blue procedure.' The range for the last method was 0.69-50.0 mg 1-1 of sulphate but that for the ion-chromatographic system was 0.06-10.0 mg 1-1 of sulphate. For spectrophotometric analyses only, samples were pre- treated by neutralisation whenever their pH was less than 5. Given the incongruity of the

Publ

ishe

d on

01

Janu

ary

1981

. Dow

nloa

ded

by U

nive

rsity

of

Cal

ifor

nia

- Sa

nta

Cru

z on

26/

10/2

014

01:4

9:49

. View Article Online

708 CROWTHER AND MCBRIDE: DETERMINATION OF ANIONS IN A%d$st, vd. 106

calibration ranges, linear regression analysis showed that the agreement between the two procedures was satisfactory. Evidence of bias associated with sample concentration level was minimal.

For nitrate, 118 precipitation samples were analysed by ion chromatography and by an automated procedure similar to that described by Kamphake et aZ.,8 i.e., hydrazine reduction of nitrate to nitrite followed by formation of an azo dye. Full scale for the nitrate plus nitrite system was 1.00 mg 1-1 as nitrogen and that for the nitrite channel was 0.100 mg 1-1 as nitrogen.

For chloride, 78 precipitation samples were analysed by ion chromatography and by an automated version of the mercury(I1) thiocyanate - iron(II1) spectrophotometric proced~re .~ The manifold, for which full scale was 2.00 mg 1-1 of chloride, included a reference stream to compensate for sample matrix reactions. Linear regression analysis indicated that the methods were compatible, but chloride results from ion chromatography were higher by an average of 0.02 mg 1-1 of chloride, which is only 1% of full scale.

TABLE VII

The agreement between the two procedures was satisfactory.

INTER-COMPARISON OF ION-CHROMATOGRAPHIC SYSTEM WITH ALTERNATIVE PROCEDURES BY ANALYSIS OF PRECIPITATION SAMPLES

Anion

Parameter F- Linear regression analysis, y on x*-

Number of samples . . . . . . . . . . 55 Correlation coefficient . . . . .. . . 0.653 Slope . . .. . . . . . . . . . . 1.874 Standard deviation of slope . . . . . . . . 0.298 Interceptlmg 1-1 . . . . . . . . . . 0.0067 Standard deviation of intercept/mg 1-1 . . . . 0.0041 Standard error of estimate . . . . . . . . 0.0004

Bias associated with anion concentration level ( y - x)*- Full-scale concentration (100%) of ion-chromato-

graphic systemlmg 1-1 . . . . . . . .

Number of samples .. . . . . . . Average difference/mg 1-' . . . . . . . . Standard deviation of diff erenceslmg 1-'

Number of samples . . . . . . . . Average differencelmg 1-l . . . . . . . . Standard deviation of differenceslmg 1-1

Number of samples . . . . . . . . Average differencelmg 1-' . . . . . . . . Standard deviation of differenceslmg 1-1 . .

Number of samples . . . . . . . . . . Average differencelmg 1-I . . . . . . . . Standard deviation of differenceslmg 1-I . .

o-10yo-

. . 10-20y0-

. . 2&50%-

. . 50-100y0-

. .

0.400

48 0.015 0.019

7 0.025 0.032

0 - -

0 -

-

c1-

78 0.992 2 0.9968 0.0144 0.021 8 0.003 1 0.000 3

1.50

50 0.020 0.018

23 0.023 0.016

5 0.014 0.015

0 - -

NO,- as N

118 0.993 4 0.996 1 0.010 7 0.013 4 0.007 1 0.001 3

2.00

1 0.01 -

41 0.002 0.017

60 0.018 0.034

13 0.009 0.029

S0,Z-

218 0.967 8 0.996 9 0.017 7 0.191 6 0.062 9 0.2440

10.0

7 0.13 0.05

76 0.15 0.30

82 0.12 0.44

50 0.25 0.70

* y refers to ion-chromatographic data. For chloride, nitrate and sulphate, x refers to spectro- photometric data.

For fluoride, 55 precipitation samples were analysed by the ion-chromatographic system, and an automated ion-selective electrode systemlo (Table VII). Later a second group of 33 precipitation samples were analysed by the ion-chromatographic system, the ion-selective electrode system and an automated version of the distillation - alizarin visual method'l (Table VIII); f d l scale for both reference procedures was 0.400 mg 1-1 of fluoride. As can be seen from the linear regression analyses, determination of fluoride by ion chromatography was not compatible with either reference procedure whereas the agreement between the two reference procedures was adequate. When the data were scanned, it was apparent that most fluoride results using ion chromatography were acceptable, i.e., 0.004-0.008 mg 1-1 higher than those for either reference procedure. However, a significant proportion (24 out of 88) of the ion chromatography results were 0.015-0.100 mg 1-1 of fluoride higher than

For fluoride, x refers to ion-selective electrode data.

Publ

ishe

d on

01

Janu

ary

1981

. Dow

nloa

ded

by U

nive

rsity

of

Cal

ifor

nia

- Sa

nta

Cru

z on

26/

10/2

014

01:4

9:49

. View Article Online

Jztne, 1981 ATMOSPHERIC PRECIPITATION BY ION CHROMATOGRAPHY 709

those for the ion-selective electrode ; these results were confirmed by re-analysing the samples. For 12 of the 33 samples taken for the three-way comparison (Table VIII), both reference procedures indicated that the fluoride concentrations were equal to or less than 0.004 mg 1-1 whereas the ion-chromatographic data showed fluoride concentrations ranging from 0.012 to 0.036 mg 1-l. As these data are characteristic of an intermittent interference problem, the ion-chromatographic system studied is not considered acceptable for the determination of fluoride in atmospheric precipitation.

TABLE VIII INTER-COMPARISON OF FLUORIDE PROCEDURES BY ANALYSIS OF PRECIPITATION SAMPLES

Linear regression analysis, y on x*

Number of samples . . . . . . . . Correlation coefficient . . . . . . Slope . . . . . . . . . . . . Standard deviation of slope . . .. Interceptlmg 1-’ . . .. . . . . Standard deviation of intercept/mg 1-1 . . Standard error of estimate. . . . ..

Ion chromatography versus distillation -

alizarin visual 33

0.748 7 0.898 6 0.142 9 0.0100 0.001 9 0.000 08

Ion-selective electrode versus

distillation - alizarin visual

33 0.952 7 0.9002 0.050 7 0.001 7 0.000 7 0.00001

Ion chromatography versus ion-selective

electrode 33

0.7758 0.9945 0.145 3 0.008 0 0.001 9 0.000 07

* For the first two columns of results x refers to distillation - alizarin visual procedure. For the third column x refers to ion-selective electrode data.

References 1. 2.

3.

4. 5.

6. 7 .

8. 9.

10.

11

Small, H., Stevens, T. S., and Bauman, W. C., Anal. Chem., 1975, 47, 1801. Sawicki, E., Mulik, J. D., and Wittgenstein, E., Editors, “Ion Chromatographic Analysis of Environ-

mental Pollutants,” Ann Arbor Science Publishers, Ann Arbor, Mich., 1978, p. 53. Sawicki, E., Mulik, J - D., and Wittgenstein, E., Editors, “Ion Chromatographic Analysis of Environ-

mental Pollutants.” Ann Arbor Science Publishers, Ann Arbor, Mich., 1978, p. 65. Wetzel, R., Environ. Sci. Technol., 1979, 13, 1216. La Fleur, P. D., Editor, “Accuracy in Trace Analyses : Sampling, Sample Handling, Analysis,”

Volume 1, N B S Special Publication 422, National Bureau of Standards, Washington, D.C., 1976, p. 141.

Stevens, T. S., and Turkelson, V. T., Anal. Chem., 1977, 49, 1176. American Public Health Association, American Water Works Association and Water Pollution

Control Federation, “Standard Methods for the Examination of Water and Wastewater,” Fourteenth Edition, American Public Health Association, Washington, D.C., 1976, p. 628.

Kamphake, L. J., Hannah, S. A., and Cohen, J . M., Water Res., 1967, 1, 205. American Public Health Association, American Water Works Association and Water Pollution

Control Federation, “Standard Methods for the Examination of Water and Wastewater,” Fourteenth Edition, American Public Health Association, Washington, D.C., 1976, p. 613.

American Public Health Association, American Water Works Association and Water Pollutio: Control Federation, “Standard Methods for the Examination of Water and Wastewater, Fourteenth Edition, American Public Health Association, Washington, D.C., 1976, p. 391.

American Public Health Association, American Water Works Association and Water Pollution Control Federation, “Standard Methods for the Examination of Water and Wastewater,” Fourteenth Edition, American Public Health Association, Washington, D.C., 1976, p. 614.

Received July 24th. 1980 Accepted December 2nd, 1980

Publ

ishe

d on

01

Janu

ary

1981

. Dow

nloa

ded

by U

nive

rsity

of

Cal

ifor

nia

- Sa

nta

Cru

z on

26/

10/2

014

01:4

9:49

. View Article Online