ANALYST, AUGUST 1991, VOL. 116 807 Improvement on the Microdiff usion Technique for the Determination of Ionic and Ionizable Fluoride in Cows' Milk Jacobus F. van Staden Department of Chemistry, Faculty of Science, University of Pretoria, Pretoria 0002, South Africa Sophia D. Janse van Rensburg Center for Stomatological Research, Faculty of Dentistry, University of Pretoria, Pretoria 0002, South Africa ~~~ ~~~ Different microdiffusion techniques for the determination of ionic and ionizable fluoride in milk have been evaluated for measurement with a fluoride-selective sensor. This work culminated in a modified version of the hexamethyldisiloxane-acid diffusion technique where an increased amount of perchloric acid and sodium hydroxide seemed to be a prerequisite for accurate and precise results.The configuration of the fluoride-selective electrode was also modified by using an adapted microanalytical procedure to determine fluoride in small volumes (50 p1 each) and to avoid contamination between successive milk samples. The resultant procedure is free of interferences and is capable of measuring fluoride in milk at concentrations from 0.02 to 10.00 mg dm-3 with improved accuracy and precision compared with earlier work. Keywords: Milk; fluoride determination; microdiffusion; fluoride-selective electrode The effect of maternal fluoride intake on the fluoride content of breast milk forms an important part of the ultimate fluoride intake of infants. According to the literature,' there is no change in the fluoride content of milk, regardless of the fluoride intake of the mother.Despite this, the documented val~es2~3 for the fluoride content of cow's milk still span a wide range between 0.04 and 0.8 mg dm-3. With a fluoride content of 0.8 mg dm-3, milk can be an important source of fluoride during the mineralization period of teeth in children.2 However fluorosis may occur when additional fluoride is taken and the individual is overdosed. It would therefore be an advantage to have a reliable method available to detect small changes in the concentration of fluoride in milk. According to Dabeka et al. ,4 the variation in the documented values for the fluoride content in milk could be due to the variety of analytical methods that are used. Although the response time of the lanthanum trifluoride ion-selective electrode is relatively fast, giving rapid and reliable fluoride measurements, the pre-treatment of small volumes of biological samples with very low fluoride contents is not easy to perform.A variety of procedures for the determination of fluoride in biological fluids have been described, ranging from direct potentiometry,2,5 hydrolysis and pyrohydrolysis,5 adsorption on calcium phosphate,6 different ashing procedures2-7 and different microdiffusion techniques using hydrochloric,7~8 sulphuric9 and perchloric acids10 for the release of ionic fluoride. Despite the develop- ment of relatively simple, accurate and reliable analytical techniques for the determination of ionic fluoride, the values reported from different laboratories continue to span a wide range and many differences of opinion still exist.This study was initiated because of the controversy sur- rounding the actual concentrations of ionic and ionizable fluoride in milk, measured using the various reported methods, and because of the potential for fluoride in milk to act as an anticariogenic agent. It is possible that the differences in the values obtained for fluoride can be attributed to the pre-treatment of milk samples and an investigation to evaluate different microdiffusion procedures for the determination of free and ionizable fluoride in milk was undertaken. Experimental Solutions All solutions were prepared from analytical-reagent grade chemicals unless otherwise specified, dissolved in doubly distilled, de-ionized water and stored in polyethylene con- tainers. Sodium hydroxide solution, 0.5 rnol dm-3.Dissolve 20.0 g of NaOH carefully in 600 cm3 of distilled water, cool to room temperature and dilute quantitatively to lo00 cm3 with distilled water. A 0.05 mol dm-3 solution of NaOH was prepared by suitable dilution with distilled water. Acetic acid solution, 0.2 mol dm-3. Dissolve 11.6 cm3 of acetic acid (99.5 per cent by mass; relative density = 1.05; and concentration = 17.4 mol dm-3) in 500 cm3 of distilled water and dilute to lo00 cm3 with distilled water. H2S04 solution, 1.5 mol dm-3. Dissolve 84 cm3 of H2S04 (96 per cent by mass; relative density = 1.84; and concentra- tion = 18.00 mol dm-3) carefully in 800 cm3 of distilled water, cool to room temperature and dilute to 1000 cm3 with distilled water.Perchloric acid solution, 6.0 mol dm-3. Add 516 cm3 of 70 per cent by mass HC104 (relative density = 1.66; and concentration = 11.6 mol dm-3) to approximately 600 cm3 of distilled water and dilute to lo00 cm3 with distilled water. A 4.0 mol dm-3 perchloric acid solution is prepared by suitable dilution of the 6.0 rnol dm-3 perchloric acid with distilled water. HCl sohtion, 0.5 rnol dm-3. Add 44.5 cm3 of HCI (35 per cent by mass; relative density = 1.18; and concentration = 11.3 mol dm-3) to approximately 600 cm3 of distilled water and dilute quantitatively to lo00 cm3 with distilled water. Sulphuric (or perchloric) acid saturated with hexamethyl- disiloxane (HMDS). Saturate a 1.5 mol dm-3 H2S04 solution with HMDS by adding 10 cm3 of HMDS to 500 cm3 of 1.5 rnol dm-3 H2SO4 in a separating funnel and shaking it vigorously for 5 min.Prepare an HMDS-saturated HC104 solution by mixing 10 cm3 of HMDS with 500 cm3 of 6.0 mol dm-3 HCI04 (or 500 cm3 of 4.0 rnol dm-3 HC104) in a separating funnel and shaking it vigorously for 5 min. Use these solutions within 24 h. Standard calibration fluoride solutions. Prepare standard calibration fluoride solutions (containing 0.1,0.3,0.5,1.0,2.5 and 5.0 mg dm-3 of F-) by suitable dilution of the stock standard NaF solution containing 100 mg dm-3 of fluoride (Type S3596 from Radiometer, Copenhagen). Add a fixed aliquot of a background ionic solution containing total ionic strength adjustment buffer (TISAB) (TISAB 11, Orion Cat. No. 940909), 0.5 mol dm-3 HCl and 0.5 mol dm-3 NaOH to each fluoride standard solution, and dilute with distilled water.This gives standard calibration fluoride solutions with the same ionic background as is used for sample measurement.808 ANALYST, AUGUST 1991, VOL. 116 Standard fluoride solutions for recovery experiments. Pre- pare standard fluoride solutions (containing 0.02, 0.05, 0.1, 0.2, 0.3, 0.5, 1.0, 5.0 and 10.0 mg dm-3 of F-) by suitable dilution of the stock standard NaF solution containing 100 mg dm-3 of fluoride (Type S3596 from Radiometer) with distilled water for the recovery experiments, in order to evaluate the accuracy and precision of the methods. The standard fluoride solutions are then treated in the same way as the milk samples. The main purpose of these standard solutions is for evaluation of the sample pre-treatment procedures of different diffusion techniques, as discussed later.Sample Collection Milk was collected from the cow’s teat in polyethylene bottles, the udder being pre-washed with distilled water. The fluoride contents of the cows’ drinking water were varied in order to establish a variation in the fluoride contents of the milk from the different cows. All samples were kept refrigerated and analysed within 12 h. Sample Pre-treatment Milk samples were treated using the microdiffusion technique reported by Whitfords and a modified microdiffusion tech- nique described by Spak et al.10 This was done in order to isolate and concentrate the fluoride in the samples. The following modified diffusion sample pre-treatment procedure was used.Pipette 3 cm3 of milk samples into the bottom parts of non-wettable plastic Petri-dishes. Spot the lids of the Petri-dishes carefully with 200 1-11 of 0.5 mol dm-3 NaOH solution, using a micropipetting system (Oxford instamatic solid displacement, Lancer, St. Louis, USA). Seal the lids tightly to the bottom parts of the dishes with vaseline. Add 6 cm3 of the HDMS-saturated 6.0 rnol dm-3 HC104 diffusion reagent through a small hole in the lid of the diffusion dish. Seal the small hole tightly with vaseline directly after addition of the acid. Diffuse the samples for 24 h at room temperature (22°C) in the sealed Petri-dishes. The ionic and ionizable fluoride are thus released from the sample and adsorbed by the NaOH solution on the lids. After diffusion, dry the lids in a desiccator.Dissolve the dried spots by adding 200 1-11 of 0.5 rnol dm-3 HC1 followed by 200 pl of TISAB buffer. The final volume is 400 pl and the pH of the solution is 5.1. Sample Measurement Following sample pre-treatment the final sample solutions (at pH 5.1) are analysed using a microanalytical procedure involving modificationsll-14 of the fluoride-selective elec- trode. The advantages are that not only are the interferences removed, but the fluoride contents of the samples being concentrated are above the detection limit of the fluoride- selective electrode. The amount of sample obtained from the sample pre- treatment is sufficient to allow three successive fluoride analyses of 50 1-11 each to be carried out. In this method the fluoride-selective electrode is filled with a fluoride electrode internal solution (0.1 rnol dm-3 KCl, 1 x 10-3 rnol dm-3 NaF saturated with AgCl and 1% agar gel) via a fine needle into a small hole just above the internal side of the membrane, taking care to avoid the formation of air bubbles.The hole is sealed with bees wax. The fluoride-selective electrode is inverted with the active membrane surface positioned vertically upwards. A liquid solution of KCI-agar gel forms a bridge between the reference electrode and the fluoride-selective electrode.13 A 50 pl aliquot of the pre- treated sample solution is pipetted onto the membrane of the electrode. The full sensing surface of the electrode was used with no adaptor positioned on the active membrane surface. The main reason for not using an adaptor is that the presence of a sleeve is a possible source of contamination.13 Fluoride measurements were performed using a Radiometer F1052F fluoride-selective electrode in conjunction with a Radiometer K4040 calomel reference electrode.The potentials were measured at room temperature with a Radiometer (Model Ion 85) ion analyser. The voltage was recorded after 2 min which was the time at which maximum equilibration at the electrode was achieved. The electrode measurements were linear over the whole concentration range except at the low concentration of 0.02 mg dm-3, where the calibration graph deviates from linearity . Recovery Experiments Recovery experiments were performed on standard solutions of fluoride and on milk samples. Standard Solutions The recovery of standard solutions containing fluoride at concentrations of 0.02, 0.05, 0.1, 0.2, 0.3, 0.5, 1.0, 5.0 and 10.0 mg dm-3 was evaluated by subjecting them to the same sample pre-treatment as the milk samples, followed by sample measurement.The main purpose of this operation was to evaluate the recovery of ionic fluoride using the different microdiffusion techniques described, compared with pure ionic fluoride standards at the primary starting-point , before proceeding to recovery experiments on milk samples for further evaluation. Milk Samples The recovery of ionic fluoride (0.02,0.1,0.2,0.5 and 1.0 pg of fluoride) added to milk samples was also evaluated following the same sample treatment and with the same aim. Results and Discussion Many a~thors~38-10 considered that the use of the HMDS-acid diffusion technique for the determination of ionic fluoride in biological fluids gave very satisfactory ,9 accurate and preciselo results.The development of the initial diffusion technique by TaveslsJ6 marked an important breakthrough for the determi- nation of ionic fluoride in biological samples, because it isolated the ionic fluoride in samples from interferences and concentrated the fluoride in the original sample to a smaller volume for measurement.9 This has the advantage that the analysed solution contains an ionic fluoride concentration above the limit of sensitivity of the fluoride-selective elec- trode, which improves the accuracy and precision.9 In addition to the normal requirements for good analysis (physical operations), it seems that the efficiency of the microdiffusion technique depends on the amount of sample used, the type, concentration and volume of acid added, and the concentration and volume of NaOH spotted on the lids of the Petri-dishes.The efficiency of the technique was evaluated in a series of recovery experiments. The results, using standard solutions with the procedure described by Whitford9 with 24 h diffusion, revealed a recovery of between 68 and 76% for fluoride concentrations ranging from 0.02 to 0.5 mg dm-3. This decreased sharply to 30% for the standard solution containing 1.0 mg dm-3 of F- and about 14% for the solutions containing 5.0 and 10.0 mg dm-3 of F-. The results also showed that the recovery from the same standard fluoride solution (0.02 mg dm-3) varied considerably. This clearly indicated that the dissociation of the HMDS molecule as described by Whitfords in the presence of 1.5 rnol dm-3 H2S04 is not quantitative and repeatable.It was also observed from results obtained when H2S04 was replaced by HCI that the results are again not quantitative and repeatable. Although 50 pl of 0.05 rnol dm-3 NaOH is only sufficient to trap about 0.047 mg of F- theoretically, the same percentage of recovery (68-76%) indicated that up to 0.5 mg of F- couldANALYST, AUGUST 1991, VOL. 116 809 Table 1 Recovery of ionic fluoride from standard solutions with the modified HMDS-acid diffusion technique Standard fluoride Fluoride mg dm-3 mgdm-3 Recovery*( %) solution used/ recovered/ 0.020 0.050 0.100 0.200 0.300 0.500 1 .000 5 .OOO 10.000 0.019 0.048 0.101 0.195 0.303 0.476 0.879 4.300 8.200 95.0 f 5.7 96.0 f 5.7 101.0 k 5.8 97.5 f 5.6 101.0 f 5.6 95.2 k 5.5 87.9 f 5.2 86.0 f 5.0 82.0 f 4.9 * Average of 14 tests in each instance, with the appropriate standard deviation. be trapped experimentally and that the lack of NaOH solution is not the main cause and did not contribute significantly to the low precision and recovery obtained.The NaOH solution however played a significant role for concentrations above 0.5 mg dm-3 of F-. The evaluation of the recovery of ionic fluoride with standard solutions according to the procedure described by Spak et al.10 (50 yl of 0.5 rnol dm-3 NaOH, 2 cm3 of 4.0 rnol dm-3 HC104 and 24 h diffusion) showed a recovery of between 82-85% for fluoride concentrations in the range 0.02-0.5 mg dm-3, which is an improvement on the 68-76% obtained with Whitford’s method.9 The recovery decreased to about 66% for solutions containing 0.5-5.0 mg dm-3 of F- and to 50% for those containing 10.0 mg dm-3 of fluoride.It is clear from these results that Spak’s method10 showed an over-all improvement on the reagents used by Whitford.9 Two factors contributed to the improvement. The HC104 solution seemed to be more efficient in dissociating the HMDS molecule, which contributed significantly to the release of more fluoride. The recovery was also enhanced by increasing the concentration of NaOH on the lid from 0.05 mol dm-3 used by Whitford9 to 0.5 rnol dm-3 as used by Spak et al. 10 The precision also improved, giving a coefficient of variation of 19.5% for 14 tests.However, it was clear that the amount of HC104 was still insufficient for the dissociation reaction to proceed to the extent where all the ionic fluoride was released. Our results also showed that the amount of NaOH solution was insufficient for trapping all the released fluoride. With these two factors in mind, the HMDS-acid diffusion technique was modified by increasing the amount and concentration of HC104 to 6 cm3 and 6.0 rnol dm-3, (twice the volume of the milk sample used) and the NaOH solution to 200 yl (instead of 50 PI) and 0.5 rnol dm-3. The recovery results obtained for fluoride standard solutions with concen- trations from 0.02 to 10.0 mg dm-3 using this method are given in Table 1. It followed from the results that a significantly increased amount of fluoride is transferred to the NaOH, 95.0-101% for solutions containing 0.02-0.20 mg dm-3 of F-.The results also indicate an improvement in recovery for the fluoride solutions with higher concentrations ranging between 0.5 and 10.0 mg dm-3 of F-. The precision also improved giving a coefficient of variation of 6.0% for the 14 tests. The recovery of different amounts of fluoride added to actual milk samples is shown in Table 2, which shows an improvement in the accuracy in fluoride transfer and entrap- ment with the proposed modified diffusion technique. The coefficient of variation was 7.4%. It is also clear from the percentage recovery obtained, that the added fluoride is not covalently bound to such an extent by the protein and Ca2+ ion in milk that it cannot be released sufficiently by the proposed modified diffusion technique.Table 2 Recovery of ionic fluoride from spiked milk samples with the modified HMDS-acid diffusion technique [F-] of milk Fluoride Recovery of used*/mg dm-3 added/pg fluoride added? (%) 0.300 0.037 0.065 0.075 0.067 0.175 0.097 0.067 0.067 0.075 0.067 0.055 1 .o 1 .o 1 .o 1 .o 0.5 0.5 0.1 0.1 0.1 0.1 0.02 0.02 100.0 k 6.8 86.0 k 5.9 100.0 f 6.3 91.0 k 6.5 88.0 f 5.7 88.0 f 5.8 97.0 f 6.6 94.0 & 6.6 99.0 * 6.7 99.0 k 6.7 100.0 k 5.6 100.0 & 5.6 * Mean result as determined by microdiffusion and direct poten- -t Average of 14 tests in each instance with the appropriate standard tiometry (accuracy and precision taken into consideration). deviation.Table 3 Comparison of the results for milk samples obtained with the proposed modified HMDS-acid diffusion technique and direct potentiometry HMDS method [F-]/mg dm-3 0.156 0.241 0.122 0.092 0.061 0.079 0.094 0.098 0.134 0.120 0.079 0.141 0.109 0.060 Direct potentiometry [F-]/mg dm-3 0.162 0.214 0.126 0.094 0.078 0.080 0.074 0.092 0.106 0.104 0.078 0.108 0.090 0.066 The fluoride-selective electrode tends to become unstable, owing to the deposition of organic material from the milk samples on the sensor over a period of time in direct potentiometry.2.5 The time involved depends on the amount and nature of protein in milk. In the present study, after between 2 and 9 determinations the membrane surface had to be cleaned before further analysis, Direct potentiometry is therefore avoided in the determination of fluoride in milk on a routine basis, however, for 1-2 determinations at a time, before cleaning of the membrane surface, it acts as a good comparative method.In the present study the membrane surface was cleaned between individual determinations and the electrode calibrated. The accuracy of the proposed modified HMDS diffusion technique was confirmed by comparing the results obtained with direct potentiometry (Table 3). The results indicate an acceptable level of agree- ment between the two methods. The proposed modified HMDS diffusion technique however offers the following advantages over direct potentiometry. The fluoride contents of the samples are concentrated by the diffusion technique to a level above the detection limit of the fluoride-selective electrode, which is not possible with direct potentiometry; an advantage which increases the accuracy and precision. The active membrane surface of the LaF3 crystal is not contami- nated or blocked, as any interferences are removed by the diffusion technique and therefore contamination between successive samples is avoided.It is also not necessary to prepare different calibration standards for samples with different matrices as the same background ions are present in810 ANALYST, AUGUST 1991, VOL. 116 the diffused standards and samples. The percentage recovery of the added fluoride is higher with the proposed diffusion technique. It can be concluded that the proposed procedure is free of interferences and can be used to measure fluoride in milk in the concentration range 0.02-10.00 mg dm-3 with improved accuracy and precision compared with earlier work.The accuracy in the concentration range 0.022-0.50 mg d ~ n - ~ of F- is superior to that in the 1.00-10.00 mg dm-3 range. This project was supported financially by the Medical Research Council of South Africa, the Foundation for Research Development (FRD), Pretoria and the University of Pretoria. References 1 Ekstrand, J., Boreus, L. O., and de Chateau, P., Br. Med. J. 1981,283,761. 2 Backer Dirks, O., Jongeling-Eijndhoven, J. M. P. A., Flissel- baalje, T. D., and Gedalia, I., Caries Res., 1974, 8, 181. 3 Larsen, M. J . , Senderovitz, F., Kirkegaard, E., Poulsen, S., and Fejerskov, O., J. Dent. Res., 1988, 67, 822. 4 Dabeka, R. W., Karpinsky, K. F., Mckenzie, A. D., and Bajdik, C. D., Food Chem. Toxicol., 1986,24,913. 5 Duff, E. J., Caries Res., 1981, 15, 406. 6 Venkateswarlu, P., Singer, L., and Armstrong, W. D., Anal. Biochem., 1971,42, 350. 7 Esala, S., Vuori, E., and Helle, A., Br. J. Nutr., 1982,48, 201. 8 Fry, B. W., and Taves, D. R., J. Lab. Clin. Med., 1970, 75, 1020. 9 Whitford, G. M., in Monographs in Oral Science, ed. Myers, H. M., Karger, Basle, Switzerland, 1989, pp. 1-30. 10 Spak, C. J., Hardell, L. I., and de Chateau, P., Acta Paediatr. Scand., 1983, 72, 699. 11 Venkateswarlu, P., Clin. Chim. Acta, 1975, 59, 277. 12 Hallsworth, A. S., Weatherell, J. A., and Deutsch, D., Anal. Chem., 1976,48, 1660. 13 Vogel, G. L., Chow, L. C., and Brown, W. E., Caries Res., 1983, 17,23. 14 Retief, D. H., Summerlin, D. J., Harris, B. E., and Bradley, E. L., Caries Res., 1985, 19, 248. 15 Taves, D. R., Talanta, 1968, 15, 969. 16 Taves, D. R., Talanta, 1968, 15, 1015. Paper 0105331 G Received November 27th, 1990 Accepted April 11 th, 1991