AnaZyst, October, 1968, Vol. 93, fq5. 643-652 643 An Evaluation of Some Methods for the Determination of Fluoride in Potable Waters and Other Aqueous Solutions BY N. T. CROSBY, A. L. DENNIS AND J. G. STEVENS (Laboratory of the Government Chemist, Ministry of Technology, Cornwall House, Stamford Street, London, S.E. 1) Five spectrophotometric procedures for the determination of fluoride in water have been evaluated with respect to reproducibility, sensitivity, range, stability of coloured products and of reagents, specificity and effect of temperature. The thorium nitrate titration is briefly discussed, and the use of the Orion fluoride-ion electrode for pF measurement has also been investigated. Various samples of water containing natural or added fluoride have been analysed by four of the spectrophotometric methods, and the results compared with those obtained by titration and with the electrode.The electrode is shown to be less susceptible than the colorimetric methods to interference from other ions in solution, and it gives theoretical recoveries of fluoride added to several drinking water supplies. FLUORINE is extensively distributed throughout nature, and has been detected in such diverse substances as water, rocks and minerals, fossils, teeth, foodstuffs and many biological specimens. Interest in the fluorine content of these materials arises chiefly from the toxic effects of prolonged ingestion of small amounts of fluorine (fluorosis), together with the detrimental results of sub-optimum levels in the diet. There is, therefore, a need for a method for the determination of trace amounts of fluorine, which is applicable to a wide variety of substances.Ideally, the method should also be simple and rapid, as many analyses may have to be carried out on a routine basis. Methods of analysis vary with the nature of the starting material and the particular interfering elements to be removed. Thus urine,l the chief medium for the excretion of fluorine from the body, is first evaporated to dryness in the presence of alkali, the residue ashed to destroy organic matter, then the fluoride is distilled to separate it from interfering ions and finally determined in the distillate. On the other hand, samples of water2 are frequently analysed directly for fluoride, although more accurate values can be obtained after prior distillation.The distillation procedure is time consuming, potentially hazardous and can only be used by a skilled analyst, as it requires careful control to obtain reliable results. Recovery of fluoride depends on the design of the still, the concentration and nature of the acid used, temperature control and rate of distillation and the volume of distillate collected. Modern techniques are based on the studies and recommendations of Willard and Winter,3 and are satisfactory as the recoveries of fluoride fall regularly in the range 95 to 100 per cent. An alternative apparatus, which incorporates a constant-temperature jacket of refluxing sym- tetrachloroethane, has been described by Samachson, Slovik and S ~ b e l . ~ This method has also been used successfully by Glover and Phillips5 in their studies on fossils.Some attention has been paid to diffusion,6 as an alternative to steam-distillation, to effect the fluoride separation. This technique is hardly any less time consuming except, possibly, when many determinations have to be made concurrently. Most colorimetric methods proposed for the final determination of fluoride depend on the bleaching action of fluoride ions on a particular organometallic dye c ~ m p l e x . ~ ~ * ~ ~ s l o ~ l l The optimum reaction conditions have been investigated by many workers over the years, but the methods are subject to interference by other ions, which can also form stable complexes with either the metal or the fluoride ions present. A new principle was established in 1959 when Belcher, Leonard and West12 reported the reaction between the red cerium(II1) chelate of alizarin complexone and fluoride ions.The resulting complex is blue, and contains fluorine 0 SAC; Crown Copyright reserved.644 [Analyst, Vol. 93 and cerium - alizarin complexone in a 1 : 1 molar ratio.13 This is the first positive colour- development reaction of the fluoride ion and, as a result of these investigations, a new colorimetric procedure was ev01ved.l~ This method, as modified by Greenhalgh and Riley,= has much greater sensitivity than conventional “bleaching” techniques but, unfortunately, is still subject to interference from several ions. A further advance16 is the recent development of an ion-specific electrode for fluoride which, it is claimed, is unaffected by a large excess of the common interfering ions and exhibits a Nernstian response over a wide concentration range.This electrode has been used by Raby and Sunderlandl’ to determine fluoride in tungsten, following a simple fusion. Fluoride was then determined directly, without the need for any separation step. From this point of view, the electrode appears to be suitable for the determination of fluorides in water and other aqueous solutions. In this work, the four most frequently used reagents for the determination of fluoride have been investigated. The spectrophotometric procedures are critically evaluated, with particular regard to linearity and range of calibration graph, reproducibility, sensitivity and limit of detection, stability of colour produced and of reagents, effect of temperature and interfering ions.The titration method is included as it is the official method of the Society for Analytical Chemistry. Finally, the use of the Orion ion-specific electrode with several aqueous solutions is reported, and the results compared with those obtained by using colorimetric procedures. CROSBY, DENNIS AND STEVENS: AN EVALUATION OF SOME METHODS Methods examined are given below. Method Reagent 1 Alizarin red S . . ,. .. .. 2 Alizarin red S .. .. .. .. 3 Eriochrome cyanine R (Solochrome) . . 4 Eriochrome cyanine R (Solochrome) . . 5 SPADNS .. . . .. . . 7 6 Alizarin complexone . . . I .. Procedure Spectrophotometric . . .. .. Titration .. .. .. .. Spectrophotometric (Cooke, Dixon and Sawyer’s method) Spectrophotometric (Megregian’s method) .. .. Spectrophotometric . . .. .. Spectrophotometric . . .. .. Ion-specific electrode . . .. .. Reference 7 8 9 10 11 15 18 EXPERIMENTAL APPARATUS- Optical density measurements were made in 10, 20 or 40-mm glass cells with a Unicam SP500 or SP600 spectrophotometer, de-mineralised water or the recommended reference solution being used in the compensating cell. An E.E.L. Spectra (Evans Electroselenium Ltd.) filter absorptiometer was also used; pF readings were obtained with the Orion, Model 94-09, electrode, with an E.I.L. (Electronic Instruments Ltd.) calomel reference electrode on a Pye “Dynacap” expanded-scale pH meter. REAGENTS- Analytical-reagent grade chemicals were used when possible. The dyestuff reagents commercially available were of laboratory-reagent quality, and frequently contained considerable and varying amounts of inorganic and possibly other impurities.Samples were obtained from more than one manufacturer, but were purified only when specified in the method. A fluoride stock solution was prepared by dissolving 0.221 g of dry sodium fluoride in de-mineralised water and making the solution up to 1 litre. This solution was stored in a polythene container. A fluoride working solution was prepared by diluting 100 ml of the fluoride stock solution to 1 litre. This solution was also stored in a polythene container and found to be stable for at least 3 months. (1 ml of solution = 10 pg of fluoride.) TECHNIQUE- Each method was assessed by at least two experienced analysts working independently.In the main, the work was carried out in two different rooms within the same laboratory, where the temperature varied over the range 15” to 25” C . The instructions laid downOctober, 19681 645 originally for each method were closely followed, but the interpretation of these instructions varied between analysts in some instances. For example, one analyst added the reagents from burettes while another used pipettes. In one study the colour development was carried out in calibrated flasks, while in another graduated Nessler cylinders were used. When alternative procedures were possible in a method, both were examined. No attempt was made to modify the concentrations of the reagents to produce colour intensities nearer to the optimum sensitivity range of the spectrophotometer.In a preliminary study of each method, a calibration graph was prepared and the wavelength of maximum absorption checked. Thus, familiarity with the manipulative techniques of the methods was achieved before attempting to assess the precision of each method. RESULTS WITH SPECTROPHOTOMETRIC METHODS LINEARITY OF STANDARD GRAPH- A standard graph was prepared for each method after the reagents had aged for at least 4 hours. Additional checks were made after 1 day and again at longer intervals to determine the stability of the reagents. Methods 1, 5 and 6 gave a graph that was linear to within +2 per cent. over the range stated in Table I, which also gives the path length FOR THE DETERMINATION OF FLUORIDE IN POTABLE WATERS TABLE I LINEAR RANGE OF FLUORIDE STANDARD Linear range 1.3. 4. 5. 6. of the f I Method Fluoride, pg Fluoride, p.p.m. Alizarin (photometric) . . 0 to 250 0 to 2.5 0 to 2.0 Eriochrome cyanine R . . 0 to 15 0 to 0-8 0 to 200 (Cooke, Dixon and Sawyer) (Megregian) SPADNS .. .. . . 0 to 70 0 to 1.4 Alizarin complexone . . 0 t o 25 0 to 1.25 0 to 0-6 0 to 0.3 Eriochrome cyanine R . . 0 to 60 0 to 1-2 0 t o 12 0 to 6 GRAPHS Path length of cell used, Amax., cm nm 1 535 2 2 540 1 527.5 1 570 1 620 2 4 cell used for the optical density measurements and the wavelength of maximum absorption found experimentally. Methods 3 and 4 showed a more marked deviation from linearity. Method 3 gave the least satisfactory standard graph, as all three analysts found that it was difficult to achieve reproducible results from day to day.Table I1 shows the range of optical densities obtained by one analyst, together with the mean value and the calculated standard deviation for each point on the graph. It can be seen that the variation was greatest for the 15-pg sample and least for the 25-pg sample. Application of the statistical F-test to the ratio of the variances determined for the 15 and 25-pg samples shows that the variation observed at the 15-pg level is not significantly different from that found at the other levels. However, the shape of the graph obtained from the mean values in Table I1 was a little closer to linearity than that published by the original authors. TABLE I1 STANDARD GRAPH WITH ERIOCHROME CYANINE R Method No. 3 Fluoride, P.lg 0 5 10 15 20 25 30 Optical density range 0.895 to 0.955 0-720 to 0.801 0.543 to 0.613 0.377 to 0.440 0.246 to 0.299 0.153 to 0.203 0.076 to 0.133 Mean optical density 0.927 0.759 0-590 0.419 0.283 0.179 0.118 Standard deviation 0.0180 0.0187 0.01 74 0.0203 0.0171 0.0135 0.0191 Degrees of freedom 11 11 11 11 11 11 11 In view of the observed variation with this reagent, it was decided to investigate the procedure described by Megregian,lo by using the same reagents but at different concentra- tions.Further work in which Megregian's reagents were used was reported by Sarma.lg646 CROSBY, DENNIS AND STEVENS: AN EVALUATION OF SOME METHODS [Analyst, Vol. 93 Megregian's results were confirmed by the present authors, but the reproducibility of methods 3 and 4 over the full range of the calibration graph appeared to be about the same.With all the methods it was generally necessary to prepare a new standard graph for each batch of reagents prepared. However, the difference between successive standard graphs, even between those prepared by different analysts, in method 6 was very small. A check determination at a single concentration of fluoride is, therefore, probably sufficient when using this method for most applications, except when the highest degree of accuracy is required. The standard graph obtained after 3 hours in method 1 differed significantly from those produced after the reagents had aged for 1 day, or more. It is considered, therefore, that the recommended ageing time of 1 hour for the reagents in method 1 is insufficient. STABILITY OF COLOURED PRODUCTS- Table I11 illustrates the change in optical densities with increasing time of colour development for each method.Generally, the colour fades with time, but the rate of change is sufficiently slow to permit accurate determinations of fluoride content without too rigid a time limit. In method 1, however, the change is more marked, and it is recommended in TABLE I11 CHANGES IN OPTICAL DENSITY WITH DEVELOPMENT TIME Optical densities after development for Fluoride 0 Method taken min. 1. Alizarin 1.0p.p.m. - (photometric) Nil* - cyanine R (Cooke, Dixon and Sawyer) cyanine R Nil* 1.16 (Megregian) 3. Eriochrome 10 pg 0.285 4. Eriochrome 1.0 p.p.m. 0-561 5. SPADNS 0-8 p.p.m. 0.215 Nil* 0.372 6. Alizarin com- 5 pg - plexone 10 15 20 30 45 min. min.min. min. min. 0.288 0.417 - - - - - - - - - 0.275 - 0.276 - 0.561 - - 0.559 - 1.19 - - 1.19 - - - - - - 0.231 - 0.230 0.230 - 60 min. 0.296 0.426 0.278 0.559 1.18 0.207 0.372 0.226 75 90 min. min. 0.304 0.313 0.433 0.438 - 0.277 - 0.560 - 1.16 - - - 0.222 120 240 min. min. 0.324 0.336 0.446 0.451 - - 0.560 - 1.18 - 0.202 0.202 0.362 0.362 * Blank determination. the method that the optical densities be measured 60 2 minutes after addition of the reagents to the sample. However, it can be seen in Table I11 that the difference in optical density between the sample and the blank changes only slowly with time. TABLE IV STABILITY OF REAGENTS ON AGEING Fluoride Method taken (photometric) 1.0 p.p.m. 2.5 p.p.m. cyanine R (Cooke, Dixon and Sawyer) cyanine R 1 p.p.m.(Megregian) 5. SPADNS 1 p.p.m. 6. Alizarin complexone 6 pg 1. Alizarin Nil* 3. Eriochrome 10 CLg 4. Eriochrome Nil* (in daylight) (in the dark) Optical densities after ageing for 1 day 0.42 0.28 0.10 0-55 1-15 0.58 0- 176 0.278 0.278 1 week 0.41 0.27 0.09 0-55 1-18 0-56 0,176 12 days 3 weeks 4 weeks 5 weeks - 0.40 - 0-40 - 0.27 - 0-27 - 0-08 - 0.09 - 0.58 - 0.54 - 1-12 - 1.16 - 0.6 1 - 0.57 - 0-182 - 0.175 0.263 - 0.251 - 0.278 - 0.280 - * Blank determination.October, 19681 STABILITY OF REAGENTS- It is convenient in routine analysis to be able to use a reagent that is stable for several weeks when prepared. As can be seen in Table IV, there is no evidence to suggest that instability of any of the reagents studied would give rise to serious errors over a period of a few weeks. The alizarin complexone combined reagent deteriorates slightly when kept on the open bench, but it is more stable when stored in the dark.REPRODUCIBILITY AND SENSITIVITY- Most methods for the determination of fluoride can only be used over a narrow concen- tration range (see Table I). In practice, measurements are made at or about the 1 p.p.m. level and, accordingly, the reproducibility of each method was assessed at this value alone. To enable the limit of detection to be determined, the reproducibility of the blank was also determined. FOR THE DETERMINATION OF FLUORIDE IN POTABLE WATERS 647 TABLE V REPRODUCIBILITY AT A CONCENTRATION OF 1 P.P.M. Mean optical Standard Method Analyst density deviation 1. Alizarin (photometric) A 0.158 0.0014 B 0.280 0.001 1 C 0.217 0.00 1 1 3.Eriochrome cyanine R A 0.273 0-013* (Cooke, Dixon and B 0.61 0-043t Sawyer) C 0.283 0.017* (Megregian) 4. Eriochrome cyanine R A 0-575 0.026 5. SPADNS A 0.195 0.007 C 0-184 0.003 6. Alizarin complexone A 0.240 0,006 B 0.220 0.007 * 20 pg of fluoride. t 10 pg of fluoride, with an E.E.L. Spectra. OF FLUORIDE Degrees of freedom 20 20 20 30 20 20 20 Standard deviation1 sensitivity 1.0 0.8 0.8 0.9 3.0 1.1 2.4 30 1.8 20 0.8 20 0.24 20 - Optical densities of 20 to 30 replicate samples, each containing 1.0 p.p.m. of fluoride, were recorded over a period of 1 to 2 weeks. The standard deviations were then calculated in accordance with accepted practice and are shown in Table V. It can also be seen that methods 3 and 4 are considerably less reproducible than the remaining methods.However, as the sensitivities of the methods differ so widely (Table VI), it was felt that the magnitude of the standard deviation was not, on its own, a fair measure of the relative reproducibility of each method. Sensitivities were calculated from the slope of the standard graph, within TABLE VI SENSITIVITY AND LIMIT OF DETECTION Sensitivity A I =l Change in From standard optical density graph over the range 0.9 to 1.1 p.p.m. (optical density per Method cm per pg of fluoride) of fluoride 1. Alizarin 0.001 0.015 3. Eriochrome cyanine R 0.015 0-055* (photometric) Sawyer) (Megregian) (Cooke, Dixon and 0.120t 4. Eriochrome cyanine R 0.01 1 0.11 6. Alizarin complexone 0.025 0.100 5. SPADNS 0.004 0-040 * 10-ml sample.t 20-ml sample. Limit of detection I A 3 Standard Degrees deviation of Fluoride, of blank freedom p.p.m. 0.0055 20 0.02 0.0188 11 0.08 0.017 6 0.10 0.0055 8 0.02 0.0021 10 0.02648 CROSBY, DENNIS AND STEVENS: AN EVALUATION OF SOME METHODS [AutaEyst, Vol. 93 the linear range, allowance being made for the path length of the cell used. It is considered that the value obtained by dividing the standard deviation by the sensitivity (Table V) represents a more accurate estimate of the true reproducibility of each method. The differences observed between methods 1, 3, 4 and 5 are probably not significant, but the results suggest that, on this basis, method 6 (alizarin complexone) is considerably more reproducible than the others at a level of 1 p.p.m.of fluoride. In addition to the theoretical sensitivity calculated from the standard graph, Table VI gives the differences in optical densities between solutions containing 0.90 and 1-10 p.p.m. of fluoride that were actually observed under the conditions recommended for each method. These levels represent the permitted limits of dosing that should be maintained in the artificial fluoridation of potable waters. Methods 3, 4 and 6 are considerably more sensitive than methods 1 and 5 and would, therefore, more readily detect dosing fluctuations at this concentration. The limit of detection is an important criterion when the fluoride content is very small or the amount of sample available for analysis is restricted. Following Wilson20 and Roos,~~ it is usual to define the limit of detection in terms of the reproducibility of the sample and of the blank.The values obtained by using the expression derived by Roos21 are shown in Table VI, but it should be remembered that other factors, such as the pre-treatment of the sample, distillation technique or interfering ions, may introduce more uncertainty in the detection of fluoride at very low levels. EFFECT OF INTERFERING IONS- The effects of those ions most likely to be present in waters were studied individually, and the results are presented in Table VII. Aluminium and metaphosphate produce the most noticeable effects while, in most instances, sulphate, chloride and calcium hardness give rise to smaller inaccuracies. The effect of colour was not fully investigated, but it is unlikely to be a serious problem with most waters.Sample 2040, a river water (Table VIII), was deep yellow, with a Hazen figure of 125, and it contained only a trace of natural fluoride. Methods 1, 5 and 6 gave slightly high results with this sample, but the recovery of added fluoride was in all instances not less than 90 per cent. 0.1 ' Aluminium 0-25 0- 50 0.1 * Iron 1.0 2.0 Calcium 100' hardness 250 500 loo* TABLE VII EFFECT OF INTERFERING IONS AT A CONCENTRATION OF 1 P.P.M. OF FLUORIDE Fluoride found by method t A1(N0,),.9H20 + FeC1,.6H2O t CaCl, 500 0.5 * Phosphate 1.5 1 Alizarin (photo- metric) 0.94 0.86 0.73 1.00 1.00 1.00 1.00 1.00 1.00 1.00 0-98 0-96 1.00 1.03 1.11 1-06 1.00 1.00 1.1 1 1.21 1.42 0.98 0.98 0-98 (NaPO,) ,,.Na,O 3 Erio- chrome cyanine R 0.93 0.78 0-72 0.95 1.00 1.00 0.86 0.86 0.90 1.11 1.16 1.17 1.08 1.10 1.25 1.00 1.03 1-08 1.07 1-12 1-14 1.04 1.04 1.04 5 SPADNS 0.96 0.88 0.79 1.00 0.99 1.00 1.00 1-05 1.06 0.99 1.05 0.98 1.02 1-05 1-07 1.00 0-96 0-96 1.08 1-14 1.20 1.03 1.02 0.99 6 Alizarin complexone 0.90 0-86 0.75 0.99 1.03 1-12 0-93 0.86 0.79 0-96 0.96 0.96 1.00 0.95 0.95 0.95 0.93 0.90 0.94 0-95 0.94 0.99 1.03 1-03 i Ion- specific electrode 0.98 0.91 0-87 1-00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1-00 1.00 0.99 0.98 1.00 1.00 0-96 1.00 1.00 1.00 1.00 1-00 1.00October, 19681 649 Public supplies of potable waters contain small residual amounts of free chlorine added for disinfection purposes, which are normally removed before analysis by reduction with sodium arsenite solution; a slight excess of the sodium arsenite solution does not disturb the reaction with fluoride.TABLE VIII FOR THE DETERMINATION OF FLUORIDE IN POTABLE WATERS DETERMINATION OF FLUORIDE IN VARIOUS WATERS Fluoride, p.p.m., found by method i Sample Ladora- tory number Description 9 Ground waters, Derby- shire 12 27 51 2021 Borehole waters, London 2022 River waters- 17 R. Ouse 2040 R. Irthing (after fluoridation) (after fluoridation) (after fluoridation) 8 R. Thames a t Windsor Fluoridated waters- 656 Anglesey 657 658 66 1 663 705 636 Watford 637 638 639 Eluates from dental cements- 16 19 9 25 Alizarin metric) 1.39 0.59 0.40 0.10 1.89 1.73 0.27 1.13 0.10 1.00 Absent 1.02 0.92 0.20 0.96 0.80 0-82 0.94 0.90 1.09 0.97 1.05 (photo- 13.5 6.0 19.5 2.6 2 Alizarin (titra- tion) 1.4 0.5 0.7 0.1 2-5 1.6 - - - - - - 0-9 0-3 1-1 0.7 0.7 0.8 1.0 0.9 0.9 1.0 Method 6 after dis- tillation 19.8 12.0 6.0 3.5 - .. 3 Erio- chrome (Cooke, Dixon and Sawyer) 1.40 0-60 0-50 0.10 2-40 2.30 0.50 1.20 1.00 0.29 1-21 1.02 0.36 1-08 0-96 0.94 0.99 1.06 1.14 1.09 1-20 <0*1 17.0 9.0 20.0 2.2 5 SPADNS 1.48 0.61 0.50 0.15 2-12 2.00 0.34 1.12 0.14 1.05 0.23 1-07 0.93 0.14 0-98 0.85 0.88 1.01 1.10 1.18 1.18 1-19 11.0 5.5 20.0 2.5 6 Alizarin com- plexone 1-45 0.63 0-44 0.14 2.00 1-87 0.21 1.15 0.12 1.01 0-28 1.09 1.07 0.21 1-06 0.83 0.98 0-93 1-06 1.11 1.11 1.12 16.3 7.3 2.0 2.0 7 Ion- specific electrode 1.50 0.62 0.45 0-14 2.00 1-85 0.21 1.19 1.05 0.17 1.15 0.96 0.20 0.93 0.79 0.81 0.91 1-06 1.06 1.08 1-08 <0.1 20.0 12.5 6.0 3.6 In methods 3 and 4, interferences by aluminium can be reduced by addition of excess of sodium hydroxide solution.Aluminium is then converted into the meta-aluminate state, when it is not able to complex fluoride ions in solution. On addition of reagent, reaction between fluoride and zirconium - dye occurs before hydrolysis of meta-aluminate to the ionic form, Al3+. Sulphate ions, when present in excessive amounts, can be precipitated as described in method 3, or the error can be corrected by using a nomograph, as in method 4. However, the only satisfactory methods for removing all of the interfering ions are distillation or diffusion. The present authors have no experience of diffusion, but distillation has proved satisfactory on a wide variety of samples, giving recoveries of fluorine of better than 95 per cent.EFFECT OF TEMPERATURE- Temperature has only a small effect on the colour developed in most methods, and the inaccuracies should be negligible under normal laboratory conditions. The Alizarin red S650 CROSBY, DENNIS AND STEVENS: AN EVALUATION OF SOME METHODS [ArcaZyst, Vol. 93 and Eriochrome cyanine R reagents are the most sensitive to temperature changes and, for accurate work, the temperatures of samples and standards should be the same (k2' C). The change in optical density obtained within the range 15" to 25" C is shown in Table IX. TABLE IX EFFECT OF TEMPERATURE ON OPTICAL DENSITY Optical density at Method Solution 150 c 20" c 25' C Alizarin (photometric) Blank 0.436 0.430 0-428 1-0 p.p.m. of fluoride 0-300 0.299 0.299 Difference 0.136 0.131 0.129 Eriochrome cyanine R 10 pg of fluoride 0.56 0.55 0.54 SPADNS 1.0 p.p.m.of fluoride 0.179 0.179 0.179 Alizarin complexone 5 pg of fluoride 0.218 0.215 0.216 against reference against reference against reference TITRATION METHOD In the titration procedure8 (method 2), thorium nitrate titrant is added to the test solution until a faint permanent pink colour is observed in the presence of the indicator, Alizarin red S. The same volume of titrant is then added to a comparison solution, which becomes more pink than the test solution. The final titration is carried out with a standard fluoride solution that bleaches the comparison solution progressively to matching-point. Previous experience with this method had shown that it was subject to considerable personal error because the end-point is rather indistinct, but consistent results can, however, be obtained with it after much practice in observing the end-point change.In the original report22 of the Analytical Methods Sub-committee, great stress was laid on the importance of experimental technique on the accuracy of the results obtained. A sample of evaporated milk was analysed by six members, all experienced in the method, and the results ranged from 5.8 to 6.8 p.p.m. of fluoride. What proportion of this error arose from the titration itself, as distinct from the preliminary treatment of the sample, is not stated, but clearly the percentage error in the titration will be greater at low levels of fluorine and will vary between analysts. The titration is normally carried out in 100-ml Nessler cylinders.The optical density of the comparison solution, at a wavelength of 520nm, decreases linearly with volume of fluoride titrant added. However, the change in optical density was found to be only 0.008 (4-cm cell) per ml of 10 p.p.m. fluoride titrant, and there is no change in Amax. While a 100-ml Nessler cylinder has a colour depth of about 3+ times that of a 4-cm cell, the human eye is less sensitive than a spectrophotometer, and it is not surprising that many analysts find it difficult to obtain consistent results. ORION ION-SPECIFIC ELECTRODE Frant and RosP have described a fluoride-sensitive electrode in which the potential developed across a lanthanum fluoride crystal is dependent on the ratio of the fluoride activities on either side of the crystal.As the internal fluoride activity is constant for all practical purposes, the potential developed depends only on the value of the fluoride activity in the external solution. In use, the electrode forms a cell with an external reference electrode, normally the calomel electrode. The activity of fluoride ions in a test solution will depend on the total ionic strength of the solution. Thus, if the electrode is calibrated with standard fluoride solutions, errors in the concentration of fluoride will result when the unknown solutions contain appreciable amounts of ions other than fluoride. This effect is minimised by diluting the sample with an equal volume of a total ionic strength adjustment buffer, which can be purchased from the manufacturer of the electrode.This buffer also largely eliminates errors from pH changes and complexing agents such as aluminium. The buffer used in the present study was of an improved composition and contained phosphate, citrate and EDTA, and was recommended by workers at Electronic Instruments Ltd. It prevents serious interference from most ions present in water supplies, as can be seen from Table VII. Experiments in these laboratories have shown that an improved tolerance to aluminium, without loss ofOctober, 19681 651 sensitivity, can be obtained by increasing the strength of citrate in the buffer, although it it is not certain whether this would affect the life of the electrode. The potential developed with the electrode is reproducible to + 1 mV, and the response time was about that of a conventional pH electrode assembly.A waiting time of about 5 minutes is advisable when the concentration of fluoride in the sample is below 0.3 p.p.m. of fluoride, or when succeeding samples differ widely in fluoride content. Vigorous agitation of the sample solution reduces the time taken to attain equilibrium. Changes in temperature affect the response of the electrode, as the values of both the slope and the intercept factors in the Nernst equation will change. Additionally, the solubility equilibrium of the saturated calomel electrode may be altered. Some correction of these effects can be built into the pH meter used with the electrode, but for precise work both standard and unknown solutions should be at the same temperature. COMPARATIVE ANALYSIS OF WATERS- Six of the methods were used to determine the fluoride content of a series of ground waters containing natural fluoride, together with several artificially fluoridated waters from the study areas at Anglesey and Watford.The results have been collected together in Table VIII. Three river waters were also examined to determine the natural fluoride content ; 10 ml of a 10 p.p.m. standard fluoride solution were then added to 90 ml of each river water, and the determination was repeated to check the percentage recovery of fluoride. The results calculated from the figures given in Table VIII are shown below. FOR THE DETERMINATION OF FLUORIDE IN POTABLE WATERS Percentage recovery of fluoride by method 1 3 5 6 i R. Ouse . . .. 91 83 86 97 100 R. Irthing . . . . 92 96 93 91 100 R. Thames .. . . 102 96 88 87 100 Method 7, the ion-specific electrode, is clearly the most satisfactory from this standpoint. This approach was extended to samples of water from several other public supplies, and in every instance the recovery of fluoride calculated from results obtained with the electrode agreed with theory. Further evidence of the reliability of the electrode can be seen in the results of the analysis of eluates of dental silicate cements that contain relatively large amounts of phosphate and aluminium, in addition to fluoride. The close agreement between method 6, after distillation to remove interferences, and method 7 (ion-specific electrode) can be seen from the results given in Table VIII.DISCUSSION The number of methods available to the analyst for the determination of fluorine is considerable and continually increasing. Simple comparator method^,^ 923 924 while of value for the routine testing of water supplies at the works, are far less sensitive than the photo- metric methods used in the laboratory. Further, these visual techniques are frequently associated with a large personal error and so were excluded from the present survey.* How- ever, many of the conclusions listed under the appropriate reagent used with an instrumental procedure are likely to apply equally to the visual technique. All of the instrumental methods have their own peculiar advantages and disadvantages, which have to be weighed one against the other in the light of the particular situation and requirements of the individual analysts.In general, little attention is paid to the purity of the dye reagents, and this may be partly responsible for some of the difficulties encountered in the use of these methods. The alizarin complexone procedure has many advantages over the bleaching methods, and is particularly suitable for samples containing only very small amounts of fluorine. However, the fluoride electrode surpasses all the colorimetric methods with regard to speed, accuracy and convenience, and is recommended for the routine monitor- ing of fluoridated water supplies. The indcations are that it will also prove satisfactory for the determination of fluorine in a wide range of materials, following a simple pre-treatment to bring the fluorine into solution.This paper is published with the permission of the Government Chemist, Ministry of Technology. * It is planned to examine these techniques a t a later date.CROSBY, DENNIS AND STEVENS REFERENCES Adams, D. F., Koppe, R. K., and Mayhew, D. J., Analyt. Chem., 1957, 29, 1108. “The Conduct of the Fluoridation Studies in the United Kingdom and the Results Achieved Willard, H. H., and Winter, 0. B., Ind. Engng Chem. Analyt. Edn, 1933, 5, 7. Samachson, J., Slovik, N., and Sobel, A. E., Analyt. Chem., 1957, 29, 1888. Glover, M. J., and Phillips, G. F., J . Appl. Chem., 1965, 15, 570. Hall, R. J., Analyst, 1963, 88, 76. “Standard Methods for the Examination of Water and Wastewater,” American Public Health Association, New York, Twelfth Edition, 1965, p. 142. “Official, Standardised and Recommended Methods of Analysis,” Society for Analytical Chemistry, . W. Heffer & Sons Ltd., Cambridge, 1963, p. 123. Cooke, J. R., Dixon, E. J., and Sawyer, R., Proc. Soc. Wat. Treat. Exam., 1965, 14, 145. Megregian, S., Analyt. Chem., 1954, 26, 1161. “Standard Methods for the Examination of Water and Wastewater,” American Public Health Belcher, R., Leonard, M. A., and West, T. S., Talanta, 1959, 2, 92. Leonard, M. A., and West, T. S., J . Chem. Soc., 1960, 4477. Belcher, R., Leonard, M. A., and West, T. S., Ibid., 1959, 3577. Greenhalgh, R., and Riley, J. P., Analytica Chim. Acta, 1961, 25, 179. Frant, M. S., and Ross, J. W., Science, N . Y . , 1966, 154, 1553. Raby, B. A., and Sunderland, W. E., Analyt. Chem., 1967, 39, 1304. “Instruction Manual for Fluoride Electrode,” Orion Research Incorporated, Cambridge, Massachussetts, First Edition, 1966. Sarma, P. L., Analyt. Chem., 1964, 36, 1684. Wilson, A. L., Analyst, 1961, 86, 72. Roos, J. B., Ibid., 1962, 87, 832. Report of the Analytical Methods Sub-committee, Ibid., 1944, 69, 243. Palin, A. T., J . Amer. Wat. Wks Ass., 1967, 59, 255. Lim, C. K., Analyst, 1962, 87, 197. after Five Years,’’ H.M. Stationery Office, London, 1962, p. 21. Association, New York, Twelfth Edition, 1965, p. 144. Received March 15th. 1968 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24.