158 Analyst, March, 1968, Vol. 93, j5j5. 158-165 Determination of Low Levels of Sodium in Water by Using a Sodium-ion Responsive Glass Electrode BY D. HAWTHORN AND N. J. RAY (Central Electricity Generating Board, Midlands Region, Regional Research and Development De#artment, Hams Hall Power Station, Lea Marston, Sutton Coldjield, Warwickshire) An E.I.L. sodium-ion responsive glass electrode has been tested under laboratory conditions. The electrode responded linearly to sodium-ion concentrations over the range 0.25 to 25 p.p.m. of sodium, and the slope of the potential - concen- tration curve approximated to the theoretical value calculated from the Nernst equation. By correcting for the level of sodium estimated to be present in the pure water used to prepare the standards, a linear relationship could be obtained covering the range down to the level of sodium in the pure water.High precision could then be achieved for measurements made over the 0.004 to 25 p.p.m. range of sodium. A t low sodium levels the electrode is capable of detecting extremely small changes in sodium concentration. The possible errors involved in measuring sodium levels below 0.025 p.p.m. would tend to give results biased slightly high with respect to the true concentration. For plant control purposes this is generally preferable to readings biased low. A sodium monitor would be capable of responding quickly to the changes in sodium concentration likely to be encountered in the water - steam circuit of a power plant. A RELIABLE, continuous method of measuring sodium in the 0.005 to 25 p.p.m.range is required in the power industry. In recent years, glass electrodes have been developed that are capable of responding selectively to sodium ions. Basically, the instrument used is a pH meter, but a sodium- responsive glass electrode is used in place of a pH glass electrode in a novel cell assembly. The manufacturers claim that sodium-ion concentrations in the 0.001 to 25 p.p.m. range can be accurately measured. Analysers have now been developed that appear to be suitable for both laboratory and plant applications, and these would, obviously, be of value in deter- mining and controlling the level of sodium in the water - steam circuit of a modern power plant. This report describes the work carried out under laboratory conditions with apparatus borrowed from Electronic Instruments Ltd.APPARATUS ELECTRODE SYSTEM- During the past few years, various types of glass have been formulated that are primarily responsive to sodium ions, with suppression of other cation resp0nse.l The theoretical equation for glass electrodes responding solely to sodium ions is- 2.3026 RT E = Eo + logaNa+ where aNa+ is the sodium-ion activity. At low levels of sodium, the activity coefficient can be maintained at a constant value by controlling the environment around the electrode, e.g., excess ammonium ions, and therefore the above equation can be written- 2-3026 RT log KCN,+ F where CNaC is the concentration of sodium ions in gram ions per litre and K is a constant for the environment selected. 0 SAC and the Central Electricity Generating Board.E = E o +HAWTHORN AND RAY 159 As with pH measurements, potential measurements can be made with reference to a conventional calomel electrode with a potassium chloride salt bridge. Industrial, unscreened sodium-responsive glass electrodes manufactured by E.I.L., and designated GEA 28, were used throughout the work. The glass used in the fabrication of the sensitive bulbs is designated B.H.104. No specific details of the composition of the glass could be obtained from the manufacturers, but it is thought to be a sodium - lithium - aluminium - silica glass, possibly containing some boric oxide, the level of aluminium oxide or boric oxide, or aluminium oxide and boric oxide being the important factor determining selectivity to sodium ions.The inner reference system of the electrode comprised a silver - silver chloride electrode dipping into a hydrochloric acid - sodium acetate solution. Before the tests were started, the electrode was activated by allowing the sensitive membrane to stand in 0.1 M sodium chloride for 72 hours (the manufacturers recommend at least 24 hours’ immersion). Over the three months’ period of the tests, no drifting or sluggishness of the electrode occurred and therefore no “re-activation” was necessary. An E.I.L. R. J.23 calomel electrode, with a remote liquid junction of saturated potassium chloride, was used as the reference electrode. BUFFER SYSTEM- Mattock2 has shown that with solutions buffered to a pH above 8.0 with triethanolamine, the sodium-responsive electrode gives a linear response over the range 0 to 4 pNa (23,000 to 2.3 p.p.m.of sodium), but some curvature is obtained from 4 to 5 pNa (2.3 to 0.23 p.p.m. of sodium). The presence of sodium impurities in the triethanolamine could have been responsible for this departure from linearity. In an attempt to overcome this difficulty, E.I.L. now recommend that ammonia vapour be used instead of a liquid buffer. In this work, ammonia vapour was produced by bubbling air through ammonia solution, and the ammonia vapour was subsequently absorbed by the solution being tested (see Fig. 1). By this method, the test solution could be buffered to a pH of about 11.0, and this further increased the possibility of hydrogen-ion interference. Ammonia solution Cell ‘D’ t Waste Sample Fig.1. General arrangement of apparatus GENERAL ARRANGEMENT- Figs. 1 and 2 show a general arrangement of the apparatus and details of the cell used to measure sodium-ion concentration. To minimise contamination from glassware, the cell assemblies were made in Perspex and connecting tubing was flexible PVC. A Technicon peristaltic pump, A, was used to deliver 2-8 ml of sample per minute and 1.85 ml of air per minute into the apparatus. The air was bubbled through about 20 per cent. w/v ammonia solution in a Winchester quart and ammonia vapour was carried through to the cell, C, through inlet, F, and into chamber, G, where it was absorbed by the sample entering from E. A Perspex-covered magnetic stirrer was rotated in the chamber to ensure adequate mixing of vapour and liquid.The alkaline solution was passed over the sodium- responsive glass electrode, L, and then on to the porous plug of the salt bridge of a calomel160 HAWTHORN AND RAY: DETERMINATION OF LOW LEVELS OF SODIUM [Analyst, Vol. 93 reference electrode, M. Excess of ammonia vapour and air were removed through hole, H. To minimise the back-diffusion of potassium ions from the salt bridge into the measuring cell, the head of potassium chloride was never higher than 1 inch above the overflow, 0, and the tip of the remote liquid junction just touched the surface of the solution as it over- flowed from the cell. The fine bore of the flow path, J, also helped to restrict back-diffusion of potassium chloride. The electrodes used to determine sodium concentration were connected to a high impedance electrometer (E.I.L.Vibron 39A), which in turn was connected to a Kent “Dyna- master” recorder. G E K Fig. 2. Details of cell “C” (Lettered parts of the apparatus are referred to in the text) On leaving cell, C, via K, the solution was passed into a second cell, D, which was used to monitor the pH. The cell was comprised of an E.I.L. pH glass electrode G.H.S.23 and an E.I.L. calomel reference electrode R.J.23. The pH electrodes were connected to an E.I.L. 28A pH meter and a Cambridge D.E. recorder. Slight variations in pH were observed when the porous plug of the reference electrode was placed downstream of the glass electrode. When the positions were reversed, however, a much steadier pH record was obtained.The pump, cells and ammonia solution were kept in a cabinet, which was fitted with a thermostatically controlled heater. No provision was made for cooling, however, and the temperature of the cabinet varied slightly (27” to 30” C) with changes in ambient conditions. PREPARATION OF STANDARD SOLUTIONS- A 1000 p.p.m. sodium stock solution was prepared from sodium chloride, and standard solutions down to 0.025 p.p.m. of sodium were prepared by progressive dilution with distilled water. The presence of traces of sodium in pure distilled water and from contamination arising from apparatus during handling makes the preparation of accurate standards below 0.025 p.p.m. extremely difficult. During this work, therefore, standards below 0.025 p.p.m. were considered to be water containing a small unspecified amount of sodium, to which controlled additions of sodium had been made.The amounts of sodium added were pro- gressively decreased by factors of 10 until additions equivalent to 0.00025 p.p.m. of sodium were reached. Distilled water subjected to an identical handling procedure, but without an addition of sodium, suffered an increase in sodium content, and is therefore referred to in this paper as “contaminated” distilled water. A single batch of distilled water was used for preparing the above solutions. The same experimental procedure was adopted when water treated by powdered ion- exchange resins was used to make up the solutions in an attempt to obtain an over-all lower level of contamination than with distilled water. PREPARATION OF CALIBRATION GRAPHS- The standard solutions were pumped into the instrument, as previously described, in increasing order of strength, starting with distilled water and finishing with a 25 p.p.m.standard solution, and the potential reading was noted for each addition. These results were treated as follows to obtain a calibration graph. EXPERIMENTS AND RESULTSMarch, 19681 I N WATER BY USING SODIUM-ION RESPONSIVE GLASS ELECTRODE 161 A plot of observed potential against concentration of sodium added to distilled water followed the form described by Mattock2 and is shown in Fig. 3. Practically, the effect on the, 0.25, 2-5 and 25 p.p.m. standards of traces of sodium in distilled water, and of the slight contamination arising from apparatus, is negligible ; consequently they can be con- sidered as “true” standards.A line of “best fit” was calculated from these potential - concentration values and was extrapolated to the potential line for contaminated distilled water. 300 > 7 200- 100- - I I 1 1 1 1 1 1 1 I I I I OOOOI 0.00 I 0.0 1 010 I .o I00 101 0 The concentration at the intersection of these lines was taken as the first approximation to the sodium content of the contaminated distilled water. All the solutions had been prepared identically, and therefore their total concentrations would be higher than the known sodium additions by an amount equal to that in the contaminated distilled water (0.005 p.p.m.). All the points were corrected on this basis, and re-plotted against potential. The best straight line was calculated and is shown graphically in Fig.4. The intersections of this line with the lines representing potential readings of distilled and glass-contaminated distilled water then gave second approximations to the sodium concentrations of these waters. 0.000 I 000 I 0.0 I 0.10 I *O I 0 0 100.0 Sodium, p.p.m. Fig. 4. Plot of previous results corrected for the effect of contamination of samples: A, “contaminated” pure water; B, pure water: 0, corrected points; x , original points162 used to prepare the standard solutions (see Fig. 5). much lower sodium content than the distilled water previously used. HAWTHORN AND RAY: DETERMINATION OF LOW LEVELS OF SODIUM [Autatyst, Vol. 93 A similar procedure was adopted with the results obtained when de-ionised water was It can be seen that this water had a Fig.5. Calibration graph obtained with water treated by powdered ion-exchange resins : A, water treated by ion-exchange resins; B, “con- taminated” de-ionised water: 0, observed results; x , corrected results PRECISION TESTS- Starting with a 25 p.p.m. standard solution, distilled water was “spiked” to give increases of 0.00025, 0.0025, 0.025, 0.25 and 2.5 p.p.m. of sodium, as described previously. The potentials of this range of solutions were measured on ten different occasions over a period of 7 days, with a fresh batch of solutions for each series of measurements. A batch of measurements was always made starting with the lowest and finishing with the highest concentration level. Previously, it had been noticed that on isolated occasions at low sodium concentrations, some interference associated with the operation of the pump was obtained on the recorder and, as a result, the trace appeared as a band 8 mV wide.As earlier work had established that identical results could be obtained with either static or flowing solutions, Batch No. 1 2 3 4 5 6 7 8 9 10 11 12 Mean . . Standard deviation PH 10.95 11-00 10.95 11-00 10.95 10.95 10.90 11.00 10.85 10.95 10.90 10.95 10.95 Tem- pera- ture, “C 27.8 28.6 28.5 29-8 28.7 30.0 29.2 28.5 28.5 29.4 28-3 29.3 28.89 0.65 TABLE I OBSERVED RESULTS OF PRECISION TESTS Potential, mV A f > Con- tamin- Additions of sodium to distilled water, p.p.m. Stock ated of sodium water water + 0.00025 + 0.0025 + 0.025 + 0.25 + 2.5 + 25.0 311 305 304 289 257 201 145 84 314 308 304 294 253 198 138 78 309 303 299 296 248 194 137 77 309 303 299 287 251 197 139 79 311 303 299 289 254 199 141 81 308 302 300 290 255 201 143 82 310 302 297 288 253 198 139 80 313 305 298 290 256 202 142 83 314 307 300 293 259 199 144 81 306 299 293 281 249 195 136 76 306 299 294 285 252 195 136 77 315 306 298 290 255 201 141 81 310.5 303.6 298.8 288.5 253-5 198.3 140.1 79.9 3.1 2.8 3.3 3.6 3.2 2.3 3.1 2.5 distilled distilled I A \March, 19681 163 it was decided to take readings under static conditions to avoid the possibility of interference from the pump during the precision tests; accordingly, the following procedure was adopted.When a steady trace was obtained after the introduction of each fresh solution, the pump was stopped for 30 seconds and the potential noted; two further readings were obtained during &minute stops after runs of 2Q minutes. The three readings of potentials were always found to be the same.The pH of the solution leaving the cell was continuously recorded, and the temperature inside the cabinet was measured when each batch of samples was analysed. IN WATER BY USING A SODIUM-ION GLASS RESPONSIVE ELECTRODE TABLE I1 RESULTS OF PRECISION TESTS CORRECTED, ASSUMING A CONSTANT POTENTIAL AT +0.25 P.P.M. OF SODIUM Con- tamin- Addition of sodium to distilled water, p.p.m. Stock ated of sodium water water + 0.00025 + 0.0025 + 0.025 + 0.25 + 2.5 + 25.0 distilled distilled I h I Mean potential, mV . . 312.1 305.3 300.5 290-2 255.2 200 141.8 81.6 Standard deviation . . 2.5 2.6 3-0 2-7 1.5 0 1.6 1.0 Equivalent concentration of standard deviation, p.p.m.of sodium . . 0.0003 0.0004 0.0005 0.006 0*0015 0 0.1 1.0 Table I shows the results obtained during these precision tests. The results given in Table I1 have been derived by correcting the observed values of potential to a constant- potential reading (200 mV) at the 0.25 p.p.m. level. This is equivalent to standardising the instrument with a 0.25 p.p.m. solution before the analysis of each batch of solutions. Slopes of lines of best fit were calculated from these values of potential by using all of the points, after making an allowance for the level of sodium present in the contaminated distilled water, and the 0.25, 2.5 and 25 p.p.m. points. E, values were calculated from the slopes given by all of the points.The theoretical slope at the temperature of the cabinet was calculated from the Nernst equation. The results obtained are given in Table 111, in which details are given of the levels of sodium estimated to be present in contaminated distilled and distilled waters, respectively, as calculated from intercepts of the appropriate lines. TABLE I11 VALUES OF CALCULATED SLOPES AND ESTIMATED LEVEL OF SODIUM IN DISTILLED WATER Sodium, p.p.m. I A \ Con- Con- taminated taminated distilled distilled water, water, 1st 2nd Distilled approxi- approxi- water mation mation Mean . . 0.0032 0.0043 0.0040 Standard dekation 0.0003 0.0005 0*0004 mV per 10-fold change in concentration Slope calculated from all of the corrected points 58.71 0.5 1 Slope calculated from the 0.25,2.5 and 25.0 p.p.m.Theor- points etical Eo only slope + mV 59.25 59.53 91.7 0.50 0.13 2.4 RESPONSE- A similar series of tests was carried out to determine the response time of the system to various changes in sodium concentration in the 0.005 to 25 p.p.m. range. A 90 per cent. response was achieved in less than 3 minutes, and steady-state conditions were reached in less than 5 minutes. Essentially, these times are those required to displace and rinse the previous solution from the system. The electrode appeared to respond instantaneously to changes in concentration within the range studied.164 PREPARATION OF CALIBRATION GRAPHS- In the preparation of the calibration graphs shown in Figs. 3, 4 and 5, the assumption was made that the electrode gives a response of constant slope down to the level of sodium present in the water used to prepare the “spiked” solutions.The two main factors likely to cause departure from linearity are changes in activity coefficient with concentration of sodium ions and the influence of hydrogen or other ions on the selectivity of the glass electrode for sodium ions at low concentrations. Changes in activity coefficient at low sodium levels would be expected to be extremely small when a constant ionic background was used, and under the experimental conditions used, there- fore, the ratio of hydrogen (or possibly ammonium) to sodium-ion activity is the major factor that could affect the response of the electrode. E.I.L.3 state that interference occurs when the ratio aH+/aNa+ exceeds a critical value (about and that the response of the electrode to changes in sodium will decrease; consequently the slope of the calibration graph at low sodium levels would decrease.Readings taken from a calibration graph con- structed on the basis of linear response at low levels of sodium could, therefore, give high values with respect to the true sodium concentrations, which, for power station applications, is preferable to readings biased low. The method of extrapolation shows that the maximum bias cannot be greater than the estimated level of sodium in the “contaminated” water. During this work, water from powdered ion-exchange resins gave the lowest level of sodium in the “contaminated” water (0.0018 p.p.m.), and demonstrated that bias must be within this level under these conditions.The true value of bias might be even lower, but a much purer water than was available during these investigations would be required to establish this point. The calibration graphs show clearly that at low levels of sodium slight changes in sodium concentration (0.00025 p.p.m.) can be detected. PRECISION TESTS- The results given in Table I show that a high level of precision could be achieved, even under unfavourable operational conditions. For example, the variations in potential observed over several days included errors arising from (a) electrometer drift, (b) change of slope with temperature, (c) slight day-to-day variations in the sodium contents of the stock, “contami- nated” and “spiked” solutions, and (d) changes in ammonia solution concentration and pH caused by variations in the rate of evolution of ammonia with temperature (this is not an equilibrium condition).In practice, the monitor now supplied by E.I.L. can be calibrated either manually or automatically with a standard solution of sodium, and a correction will be applied auto- matically to compensate for any electrometer drift. In addition, a fixed theoretical slope- to-temperature relationship is assumed and changes in temperature will be compensated for automatically. Variations caused by (a) and (b) will therefore be negligible. A relatively strong solution will be used (0.1 to 0.2 p.p.m. of sodium) for standardising the instrument so that variations from (c) will be minimised. The instrument will be thermostatically controlled, hence variations caused by (d) would not be expected to arise.Results in Table I1 were obtained by applying a correction based on an assumed constant potential of 200 mV at the 0.25 p.p.m. sodium level. From Table I11 it can be seen that the mean slope calculated from the 0.25, 2.5 and 25 p.p.m. points was only slightly lower than the mean theoretical slope. The mean slope calculated from all of the points, however, was somewhat lower, even after both corrections had been applied, being 97.8 per cent. of the theoretical. This indicates that there is a slight decrease in response at low sodium levels. An instrument calibrated on the basis of a theoretical slope would therefore tend to give values biased slightly high with respect to values obtained from a practically derived slope.From the results obtained when water treated by powdered ion-exchange resin was used the maximum bias possible from the basic assumption of linear response can be fixed. CONCLUSIONS The sodium-sensitive glass electrode responds linearly to changes of concentration over the range 0.025 to 25 p.p.m. with solutions buffered from pH 10-8 to 11.0, and the slope of potential against concentration approximates to that calculated from the Nernst equation. HAWTHORN AND RAY: DETERMINATION OF LOW LEVELS OF SODIUM [Alzalyst, Vol. 93 D I s c u s s I o NMarch, 19681 IN WATER BY USING A SODIUM-ION GLASS RESPONSIVE ELECTRODE 165 After corrections have been applied below this range for the sodium present in the “pure” water used to prepare “spiked” solutions of sodium, the linear calibration graph can be extended and used down to the level of sodium in the pure water (say 0.001 p.p.m.).The maximum bias introduced by using this method would be +O.OOlS p.p.m. of sodium. At low sodium levels, the electrode is capable of detecting slight changes in sodium concentration. For example, at the estimated 0.005 p.p.m. level a change in concentration of 0*00025 p.p.m. could be detected, corresponding to an observed potential difference of 5 mV. By using the commercial E.I.L. instrument, which operates on a fixed theoretical slope - temperature relationship, it should be possible to monitor sodium levels in the range of 0.005 to 25 p.p.m. to a high level of precision. At levels below 0.025 p.p.m. possible errors may arise from the assumption, which is made when calibrating the instrument, that a constant theoretical slope of potential - concentration applies throughout this concentration stage. Results would be biased slightly high relative to the true concentrations. This paper is published by permission of the Central Electricity Generating Board. The authors thank Electronic Instruments Ltd. for providing equipment used in this work and for helpful discussions on experimental techniques involved in the use of the glass electrode, and acknowledge helpful discussions and advice in the preparation of this paper from Dr. J. Brown, Senior Chemist, C.E.G.B., Midlands Region, Scientific Services Department. REFERENCES 1. 2. 3. Eisenman, G., Bates, R., Mattock, G., Friedman, S. M., “The Glass Electrode,” Interscience Mattock, G., Analyst, 1962, 87, 930. “Glass Electrodes Reponsive to Sodium Ions,” E.I.L. Publication, T.D.S. Elect. 14 Issue, August Received August loth, 1967 Publishers, a division of John Wiley & Sons Inc., New York and London, 1966, p. 213. lst, 1964.