1161

Notes be taken as an estimate of Nova Scotia’s background level from sources such as long-range transport and natural releases from organic decomposition and rock weathering. The acidification trend does not appear to be cumulative. The surface water pH values will in any year reflect an approximate equilibrium between the rate of input of acid in precipitation and the rate of leaching of alkaline minerals from the sediments and rocks. They should therefore not decrease further unless there is a further increase in the rate of input of acid pollutants. W. D. Watt D. Scott S. Ray Department of Fisheries and Oceans Fisheries and Marine Service Resource Branch, P.O. Box 550 Halifax, Nova Scotia B3J 2S7

Limnol. Oceanogr., 24(6), 1979,1161-1165 @ 1979, by the American Society of Limnologyand

Acid precipitation:

ENVIRONMENT CANADA. 1974. Analtyical

methods manual. Inland Waters Directorate, Water Quality Branch, Ottawa. GORHAM, E. 1957. The chemical composition of lake waters in Halifax County, Nova Scotia. Limnol. Oceanogr. 2: 12-21. -. 1976. Acid precipitation and its influence upon aquatic ecosystems-an overview. Water Air Soil Pollut. 6: 457-481. HAYES, F. R., AND E. H. ANTHONY. 1958. Lake water and sediment. 1. Characteristics and water chemistry of some Canadian East Coast lakes. Limnol. Oceanogr. 3: 299-307. LIKENS, G. E. 1976. Acid precipitation. Chem. Eng. News 54: 2944. PERKIN-ELMER. 1976. Analytical methods for atomic absorption spectrophotometry. Norwalk, Conn. PETTIPAS, B. 1976. Atmospheric sulphur source emission inventory. Nova Scotia Dep. Environ. Halifax. Unpubl. Rep. SUMMERS, P. W., AND D. M. WHELPDALE. 1976. Acid precipitation in Canada. Water Air Soil Pollut. 6: 447456.

Submitted: 21 September Accepted: 10 April

1978 1979

IllC Oceanography,

Measurement

Abstruct-The pH and acidity of precipitation are difficult to measure accurately because of the low ionic strength of the samples. Use of measured pH to estimate hydrogen-ion concentration may err by as much as 50% if activity coefficients, junction and streaming potentials, and non-Nernstian behavior of the electrode system are ignored. The magnitude of the individual errors is assessed and procedures for measuring pH and acidity to reduce the total error to *5% are recommended.

Precipitation in most of the eastern United States has a hydrogen-ion concentration lo-500 times higher than would be expected if the pH were controlled by atmospheric CO, (Likens 1976). The low pH (~5.6) is caused by H,SO, and HNO, (Galloway et al. 1976). Procedures for ’ This research was performed tract 04-5-022-24.

References

under NOAA con-

of pH and acidity’ collecting, storing, and handling acidprecipitation samples have been given elsewhere (Galloway and Likens 1976, 1978). The sources and magnitude of errors associated with measuring pH in solutions of low ionic strength and procedures to reduce or eliminate these errors are presented here. We thank E. Edgerton for technical assistance and C. Brosset, T. Church, C. Culberson, and J. &Morgan for critical comments. pH is defined as the negative logarithm of the hydrogen-ion activity, aH (Stir-ensen and Linderstrgm-Lang 1924): pH = -logIOaH.

(1)

Operationally, electrodes are used to estimate the hydrogen-ion activity in solutions. The Nernst equation relates the potential as measured by such electrodes

1162

Notes

to the pH of the solution. Combining 1 with the Nernst equation yields PKC

= PHS +

a- - ESP 2.3RT

’

Eq.

(2)

where pH, is -log,,a,s, PH,~ is -log,,a,q, “H, is the activity of hydrogen ion in a test solution, aH, is the activity of hydrogen ion in a standard solution, E, is the potential measured with an electrode in a test solution, E, is the potential measured with an electrode in a standard solution, F is Faraday’s constant, R is the gas constant, and T is absolute temperature. To use the operational definition of pH (Eq. 2), we need a scale of standard values of pH,. Values of pH, are necessarily arbitrary and several scales have been suggested, each involving nonthermodynamic assumptions (e.g. Bates 1973; Stumm and Morgan 1970; Feldman 1956). However, at the low ionic strength of precipitation samples, the pH, values of the scales now in use differ by less than the uncertainty (+0.005 units) of National Bureau of Standards’ buffers. Operational pH measurements with glass electrodes estimate the hydrogenion activity, a,,. The hydrogen-ion concentration, [H+], can be calculated from [H+] = $,

+

(3)

where yH+ is the hydrogen-ion activity coefficient. yH+ is a function of the ionic strength, 1. For solutions where I < lo-” M (typical of precipitation), the simple DebyeHiickel expression for water at 25°C (Stumm and Morgan 1970) is used with Eq. 3 to generate [H+] = (lo-pH)( l()“~5~z”2)~

(4)

If ionic strength is ignored in converting from operational pH to [H+], the associated error varies from 0.001 to 0.015 pH units for ionic strengths in the range of 1O-5 to 1O-3 M. P recipitation samples collected near the ocean may contain appreciable amounts of sea spray, which would increase the ionic strength and the error

in calculated [H+] if corrections for activity coefficients are not made. Additional potentials may be produced and measured along with the potential generated by the hydrogen-ion concentration. In measuring operational pH, a liquidliquid boundary at the salt bridge of the reference electrode creates a potential (liquid-junction potential) measured along with the potentials developed by the reference and test electrodes (Bates 1973; Ives and Janz 1961; Eisenman 1967). The residual liquid-junction potential, Ej, represents the difference between the junction potentials generated when the reference electrode is in a test solution or in a standard solution. Ej is a function of pH, temperature, and the difference in composition between the test and standard solutions. Although Ej cannot be measured directly, attempts have been made to estimate its magnitude (Henderson 1907, 1908; see also Bates 1973). Residual junction potentials, which arise mainly from the large differences in ionic strength between buffers and samples, can give rise to errors as large as 0.04 pH units for various reference electrode-filling solutions and typical precipitation samples. Ej can be reduced by diluting the ionic strength of the buffers (however, the pH, values of such buffers would not be known), by using dilute solutions of strong acids as standards (these must be carefully and frequently calibrated by titration), or by increasing the ionic strength of the test solution by adding an inert salt such as KCl. Since the pH in acid-precipitation samples, which are dilute solutions of strong acids with very little buffering capacity, can be drastically changed by trace contaminants, we feel that the use of strong acid standards is the best technique to reduce the junction-potential errors. Stirring or agitating the test solution while pH is being measured can produce a streaming potential, E,, , the magnitude of which is a function of the type of electrodes used as well as of the ionic strength of the solution. Although stream-

Notes ing-potential changes in solutions of high ionic strength are usually negligible, we have measured changes >0.5 pH units while stirring precipitation samples. Such error is easily eliminated by making all measurements on quiescent solutions. Use of the operational definition of pH (Eq. 2) assumes that the probe is 100% Nernstian; that is that a plot of pH vs. E has a slope of Fs(2.3 RT)-l. However, most commercial glass electrodes in good condition have a slope that varies from 95 to 102% of the theoretical value. The slope of an individual electrode can be corrected on many pH meters by using a high- and low-buffer calibration procedure. In addition, the temperature response of some electrodes may not be Nernstian and using a temperature compensator or manually correcting the meter for temperature differences between buffers and test solutions may not be adequate. These uncertainties can be eliminated by equilibrating the buffers and test solutions to the same temperature and by determining the slope of the pH vs. E response of the electrode before each measurement. Combining Eqs. 1,2, and 4 and including potentials Ej and E,, gives the expression for hydrogen-ion concentration

1163

sphere to allow free flow at the junction. The electrode pair should be calibrated each day using strong acid solutions with a free hydrogen-ion concentration known to within 1%. These solutions may be standardized by titration (see below) with corrections made for complexing of H+ and the anion, if necessary. Between solutions, the electrodes should be rinsed copiously with distilled water. If buffers are used for calibration rather than strong acid solutions, copious rinsing will be necessary to prevent the buffer from contaminating the sample. After the electrodes are placed in the solution, the solution should be thoroughly agitated, then allowed to come to rest. Only after the meter reading is stable for 30 s should the potential of the electrode be recorded. The 5 min or more that may be necessary to achieve a steady reading can be reduced by increasing the ionic strength of the test solution with KCl; however, contamination from trace impurities in the KC1 must be avoided. All potential measurements should be made with a meter accurate to 20.1 mV. Although we found that storing the electrodes in a strong acid solution at pH 5 is better than storing them in a buffer or salt solution, prolonged storage in such a solution of low ionic strength leads to dilution of the filling solution in the reference eleclog [H+] = 0.5(P2) - pH, - & trode, necessitating frequent refills. Us. ing these procedures, we can measure UL - Es - Ej - Es,) (5) pH reliably and repeatedly in precipitation samples to within +0.02 units with that indicates the sources of uncertainty commercially available equipment and in estimates of [H+] from operational pH supplies. Such pH measurements may be measurements. If the associated errors interpreted in terms of hydrogen-ion conare additive, the uncertainty in [H+] may centration with an accuracy of 25%. be as large as 50% (or 20% if streaming The acidity in acid precipitation sampotentials are eliminated). The proce- ples is difficult to determine because of dure below should reduce this uncertainthe low values and the variety of potenty to itation for chemical analysis. Tellus 30: 71-82. -,-, AND E. S. EDGERTON. 1976. Acid precipitation in the northeastern United States: pH and acidity. Science 194: 722-724. GRAN, G. 1950. Determination of the equivalence point in potentiometric titrations. Acta Chem. Scan. 4: 559-577. -. 1952. Determination of the equivalence point in potentiometric titrations. Part 2. Analyst 77: 661-671. HENDERSON, P. 1907, 1908. Zur thermodynamik der fliissigkeitsketten. Z. Phys. Chem. 59: 118127; 63: 325-345. IVES, J. G., AND G. J. JANZ. 1961. Reference electrodes, theory and practice. Academic. KRUPA, S. V., M. R. COSCIO, JR., AND F. A. WOOD. 1976. Evidence for multiple hydrogen-ion donor systems in rain, p. 371-380. Zn Acid precipitation and the forest ecosystems. Proc. Int. Symp. (1st). USDA Forest Serv. Gen. Tech. Rep. NE-23. LIBERTI, A., M. POSSANZINI, AND M. VICEDOMINI. 1972. The determination of the non-volatile acidity of rainwater by a coulometric procedure. Analyst 97: 352-356. LIKENS, G. E. 1976. Acid precipitation. Chem. Eng. News 54: 2944. ROSSOTTI, F. J., AND H. ROSSOTTI. 1965. Potentiometric titrations using Gran plots. Chem. Educ. 42: 375-378. SORENSEN, S. P., AND K. LINDERSTRQIM-LANG. 1924. The determination and value of 7~~ in electrometric measurements of hydrogen-ion concentrations. C. R. Trav. Lab. C&slb&g 15: 40 p. STUMM, W., AND J, J. ~~ORGAN. 1970. Aquatic chemistry. Wiley-Interscience.

Submitted: Accepted:

14 June 1977 26 June 1979