Notes

TRAVERSY, W. J. 1971. Methods for chemical analysis of waters and wastewaters. Water Quality Div., Inland Waters Branch, Dep. Fish. Forestry, Ottawa. -, P. D. GOULDEN, AND R. LEACOCK. 1974. Levels of arsenic and selenium in the Ontario Region. Environ. Can., in press. VON

707
SCHOUWENBURG, J. C., AND I. WALUNGA. 1967. The rapid determination of phosphorous in the presence of arsenic, silicon and germanium. Anal. Chim. Acta 37: 271-274.

Submitted: 28 January 1974 Accepted: 16 May 1974

An automated method for determination of chloride and sulfate in freshwater using cation exchange and measurement of electrical conductance
Abstract-A simple automated adaptation of Mackereths manual method for determining chloride and sulfate in freshwater is described, using hydrogen ion exchange to convert salts of chloride and sulfate to their free acids with detection by electrical conductance. The use of silver-saturated exchange resin to precipitate chloride permits distinction between chloride and sulfate. Detection limit is good at 2 peg liter- with excellent precision ( t I.So/, at 200 peq liter-). High levels of nitrate, orthophosphate, and fluoride give positive interference for sulfate; bromide and iodide similarly interfere with chloride estimates.

Ion exchange techniques in the analysis of freshwater have been known for some time (Nydah11951; Samuelson and Schramm 1951) and were used by Mackereth (1955a, b, 1963) in several methods noteworthy for their sensitivity and simplicity. Though Mackereths ion exchange technique for chloride and sulfate is elegantly simple and has particular advantage in the field of limnology (most interfering species being either absent or present at insignificant levels ) , it has not been widely used. The method outlined below is an adaptation of Mackereths manual technique to an automated continuous-flow analytical system having a capacity of 30 samples per hour using 5 cc of sample. Its sensitivity is adequate for most natural waters while its precision is excellent, being suitable for both routine and research purposes. I wish to thank J. Prokopowich and M. Cape1 for help with the method, and F. A. J. Armstrong for discussion of the manuscript. A filtered sample is divided into two

portions. One is passed through a strong acid cation exchange resin on hydrogen cycle ( Fig. 1, column 1 ), the other through a two-stage column (Fig. 1, column 2) consisting of cation exchange resin on silver cycle followed by resin on hydrogen cycle. In column 1 sample cations are exchanged for hydrogen ions yielding an acidic solution of HCl, H2SOh, and usually a small amount of HN03, all of which are dissociated, and a small amount of H3P04 having only the first hydrogen dissociated. Also present are various weak acids, both inorganic and organic which remain essentially undissociated at low pH. Typically, effluent from column 1 has a pH of less than 4. Table 1 shows that at this pH only HCI, HzS04, HN03, and HsPO, are significantly dissociated. Usually HN03 and HSPOA are present in only trace amounts so that the hydrogen ion concentration in column 1 effluent is equivalent to the sum equivalents of Cl- plus so42-. In the first stage of column 2, sample cations are exchanged for silver ion with chloride preciptating as silver chloride and remaining in the resin bed. Remaining silver ions then pass to the second stage of the column where silver is exchanged for hydrogen ion. Effluent from column 2 is the same as for column 1 except that HCl has been quantitatively removed, with hydrogen ion concentration being equivalent to S04- ( assuming NOs- and HzPO*- levels are negligible) . The difference in hydrogen ion concentration between column 1 and column 2 effluent is equivalent to Cl- concentration.

708
WASTE t DISTILLED WATER FOR WASH

Notes

LIOUID

WASTE

Fig. 1. Pump manifold

and equipment

assembly

for automated

chloride

sulfate

analysis.

Hydrogen ion can be determined either by titration with base or by measurement of electrical conductance, the latter being quicker and more sensitive. From Table 2, which lists the equivalent ionic conductance for the major ions of freshwater, it can be seen that hydrogen ion has five times the equivalent conductance of other cations and that all acid anions have similar equivalent
Table 1. Percent dissociation of various strong and weak acids at pH 4.

Acid HCl clm"3 NO; H2S04 HSOso424 H3p04 H2P;4 HP04. H2C03 HCO; CO;H4Si04

Dissociation Constant ----------1.2 x --7.5 x 6.2 x 2.2 x 4.3 x 5.6 x --2.2 x

Percent of Total at pH 4 0 100 0 100 0 0.83 99.17 1.32 98.68 0 99.98 0.02 0 100

conductances. Therefore the equivalent conductances of strong acids are high and essentially the same ( HN03 = 421, HCl = 426, A H2S04 = 429 pmhos meq-l ), giving a mixture of strong acids a conductance that is relatively large and related only to the concentration of strong acids and not to the kind of strong acid. The conductance of column 1 effluent is then directly proportional to equivalents of Cl- plus SOa2while for column 2 it is proportional to equivalents of SOh2-. Apparatus-automatic sampler (Carlo Erba SD3) ; proportioning pump (Carlo Erba) ; two conductivity meters ( Radiometer CDM3 ) ; two flowthrough conductivity cells ( Radiometer CD 3314) ; two-channel strip chart recorder ( Hewlett Packard 7100
Table 2. Equivalent ionic conductance some ions at 25C and infinite dilution.
Anion Cl1/2W4) NO; OHA 75.5 79.0 70.6 192 Cation Na+ K+ x 50.9 74.5 45.0 60.0 350

for

1o-2 1o-3 lo-' lo-l3 1o-7 lo-l1 lo-lo

l/2 Wg2+)
1/2(Ca2+) H+

Notes B ); pump manifold as diagramed in Fig. 1; and ion exchange columns (Fig. 2). Reagents-strong acid cation exchange resin (Rexyn-101 or equivalent having a wet volume exchange capacity of 2.0 x 10m3 eq ml-l, mesh size M-50); 0.1 N I-ICI; 0.1 N AgN03; chloride standard (5.845 g of analytical reagent grade NaCl were dissolved and made to 1 liter with distilled and deionized water, 1.00 cc = 100 peq Cl-); sulfate standard (7.102 g of analytical reagent grate anhydrous Na2S04 were dissolved and made to 1 liter with distilled and deionized water, 1.00 cc = 100 peq SOb2-). Columns were packed with resin as shown in Fig. 2; each column contained about 2 cc of resin. The manifold in Fig. 1 requires three of these columns on hydrogen cycle and one on silver cycle. Hydrogen cycle columns were prepared by pumping 0.1 N HCl through a column at 2 cc min-l for 20 min. Columns were then rinsed with distilled water for 20 min at 2 cc min-l. Silver cycle columns were prepared similarly using 0.1 N AgN03. Life of the columns depends on sample size, exchange capacity of resin used, and the ionic strength of samples analyzed but should be around 2,000 samples. Column 1 can be regenerated in the same way it was prepared. The first stage of column 2 cannot easily be regenerated and is discarded, but the second stage contains much silver and should be used to prepare a silver cycle column. With the equipment assembled as in Fig. 1, the sampler was set at 30 hr-l with a 1: 1 sample-towash ratio, the conductivity meter was set at 500 pmhos crnw2 full scale and the recorder was set to read 500 mV full scale with a chart speed of 0.51 cm min-l. These settings gave a recorder full-scale reading of 250 pmhos cm-l and a useful working range of O-600 peq liter-l chloride or sulfate (O-21 mg liter-l Cl or O-10 mg liter-l SO4 - S ) . The system was calibrated by running replicate 500 peq liter-l sulfate standards. From the peak heights obtained on each channel a mean scale factor ( peq liter-l chart unit-l) was calculated for each channel. These factors were slightly different, as the conductivity cell constants were not

709

,RA

Male Luer Coupling Tygon Tubing Sleeve Glass Wool Plug

100 mm

Strong Acid Cation Exchange Resin 16-50 Mesh Either Hydrogen or Silver Saturated Depending on Use

Female Luer Coupling

Fig. 2. Ion exchange fold of Fig. 1.

II-

8mm 0 D Glass Tubing column required for mani-

identical, but were usually around 7 peq liter-l chart unit-l. It was often necessary to dilute samples before analysis, both to prolong ion exchange column life and to obtain on-scale readings. It was convenient to measure specific conductance of samples before analysis and if necessary dilute samples to a conductance of less than 100 pmhos cm-l with distilled and deionized water. Sample peak heights for channel I were multiplied by the channel 1 scale factor to give peq liter-l of SOJ2- + Cl- (plus N03and H2P04- where significant). Similarly sample peak heights from channel 2 gave peq liter-l S04- (plus NO,- and H2POawhere significant) with the difference in equivalents between channel 1 and 2 being Sulfate concentrations Cl- concentration.

710

Notes
Table 4. Patallel determination of chloride and sulfate using ion exchange (Z.E.) technique and conventional methods (T-titrimetric G-gravimetric). Samples from Char, Meretta, and Resolute lakes (Cornwallis Island, N.W.T.).
Sample T B118 B119 8120 B121 B122 B123 B124 B125 17 19 2.5 25 25 11 10 10 -1 Cl- mg R I.E. 16 19 25 25 25 11 10 10 SOi- mg R-1 G 1.3 1.3 6.3 6.6 6.6 3.6 4.0 3.6 I.E. 1.4 1.7 6.6 6.6 6.3 4.0 4.0 3.6

Table 3. Precision estimates (relative standard deviation) for synthetic and natural samples.
Distilled Level peq k-1 50 100 150 200 250 Level ueq 11-l 235 76 water standards Chloride Sulfate Error (n=6) Error (n=6) % % % % % natural + 2.2 + 1.9 '_ 1.4 " 1.5 + 1.3 water Sulfate Error (n=20) % % % % %

+_ 1.3 + 1.3 " 1.2 + 1.5 : I.2 Canadian Shield Chloride Error (n=20)

+ 1.0 % "_ 1.5 %

were calculated either directly from channel 2 peaks or by correcting for equivalents of N03- and H2PO3- present. Chloride and sulfate concentrations calculated above are expressed as microequivalents per liter and are multiplied by appropriate gravimetric factors for values expressed in milligrams per liter. Table 3 lists precision data obtained for a range of distilled water standards while Table 4 compares the ion exchange technique and conventional (Am. Public Health Assoc. 1965) methods for chloride and sulfate. Both precision and agreement with conventional methods were excellent. Recovery tests with 500 peq liter-l chloride and sulfate added to various types of freshwater yield recoveries ranging from 98101%. Response was linear over the range O-600 peq liter-l. Since conductance is a function of temperature it is important to prevent temperature fluctuations in the conductivity cell between calibration and sample analysis. It was convenient to operate the system at ambient temperature and this presented no difficulty if samples were brought to room temperature before analysis and if the apparatus was located away from sources of heat. Nitrate and orthophosphate both can

cause positive sulfate interference but are usually found at levels low relative to sulfate and can generally be ignored or adequately corrected for with rough estimates of their levels. Fluoride can also positively interfere as it does not form an insoluble silver salt; when present, fluoride interference is significant and must be corrected for. High levels of weak acids can interfere to varying slight degrees depending on their dissociation constant and level relative to sulfate. Table 5 lists some observed positive interferences with sulfate determination. Bromide and iodide both positively interfere with chloride estimates but usually are not found at significant levels in freshwater. The method outlined above has been found to be simple, quick (30 samples per maintenance free hour ) and relatively ( columns regenerated after 2,000 samples). Detection limit is good at 2 peq liter-l of chloride or sulfate while precision is excelTable 5. Positive sulfate error produced various anions. by

Interfe_rfng 1 mg L.

species level

Observed zylfate error so4-s mg R 1.1 0.8 0.5 0.02 no interference

N03-N F H2P04-P HC03-C H3Si04-Si

Notes lent at 2 1.5% at the 200 peq liter-l level. Though indirect, the method gives good agreement with conventional direct estimates of chloride and sulfate, providing proper correction is made for any interferences present. M. P. Stainton Department of the Environment Fisheries and Marine Service Freshwater Institute 501 University Crescent Winnipeg, Manitoba R3T 2N6 References
AMERICAN PUBLIC HEALTH ASSOCIATION. 1965. Standard methods for the examination of water and wastewater, 12th ed.

71.1

MACKERETH, F. J. H. 1955a. Ion exchange proceedures for the estimation of (I) total ionic concentration, (II) chlorides, and (III) sulfates in natural waters. Mitt. Int. Ver. Theor. Angew. 4, p. 1-16. 1955b. Rapid microestimation of the -. major anions of freshwater. Proc. Sot. Water Treatment Examin. 4: 27-42. -. 1963. Some methods of water analysis for limnologists. Freshwater Biol. Assoc. Sci. Publ. 21. 68 p. NYDAHL, F. 1951. Sampling and analysis of lake waters by means of ion exchange. Mitt. Int. Ver. Theor. Angew. Limnol. 2, p. 276-290. SAMUELSON, O., AND K. SCHRAXM. 1951. Determination of total salt concentration by means of anion exchange resins. Svensk. Kern. Tidskr. 63: 307-311.

Submitted: 2 November 1973 Accepted: 20 May 1974

Thermometric titration and alkalinity1

analysis of seawater for chlorinity,

sulfate,

Abstract-Total sulfate, chlorinity, and total alkalinity in seawater solutions have been determined by thermometric titration. The precision of the analyses was 0.3% for sulfate, 0.04% for chlorinity, and 0.1% for total alkalinity. Because of their simplicity, speed, and high precision, thermometric titrations are ideally suited for a wide range of applications in the marine environment (e.g. pore water analysis ) .

Conventional methods for the quantitative determination of many of the ions in seawater and similar naturally occurring multicomponent electrolyte solutions (e.g. pore and estuarine waters ) requires complicated and long procedures such as digestion of Such procedures (including precipitates. weighing) are not suitable for shipboard use and must be made in a land based laboratory. It would be most useful, therefore, to
The support of the Office of Naval Research ( N00014-67-A-0201-0013) and the Oceanographic Section of the National Science Foundation (GA17386) is acknowledged. Scientific Contribution No. 1759 from the Rosenstiel School of Marine and Atmospheric Science, University of Miami, Miami, Florida 33149.

have available a technique that is rapid, precise, simple to use, and easily suited to automation. Thermometric titration meets these criteria and has the added virtue of being suitable for running a large variety of analyses on the same piece of equipment. We will discuss here the use of thermometric titrations to determine the total sulfate, total alkalinity, and chlorinity of seawater. Many workers have used the cessation of heat on completion of a reaction as an analytical end point. With the recent development of motor-driven burets and highly sensitive thermistors, thermometric titration has become established as a routine analytical procedure ( Bark and Bark 1969; Tyrell and Beezer 1968; Jordon 1968). The use of a thermometric end-point detector has many advantages over the more conventional potentiometric end points. For example, the sensitivity of the thermometric method is a linear function of concentration (instead of logarithmic) and it is often possible to separate simultaneous reactions if the reaction heats differ (Hansen and Lewis 1971) .