ANALYST, JANUARY 1991, VOL. 116 45 Masking of Zirconium(1v) in the Determination of Fluoride With an Ion-selective Electrode: Application to Zirconium(1v) Fluoride-based Glasses Akio Yuchi, Jun-ichi Baba, Hiroko Wada and Genkichi Nakagawa Department of Applied Chemistry, Nagoya Institute of Technology, Gokiso, Showa, Nagoya 466, Japan The performance of six chelating reagents [ethylenediamine-N,N,N’,N’-tetraacetic acid (EDTA); trans-1,2- cy c I o h exa n ed i a m i n e- N, N, N ’, N ’-tetra acetic N ’- ( 2- h y d r ox yet h y I ) et h y I e n ed i a m i n e- N, N, N ’- triacetic acid (HEDTA); trie”yIen“etraamine-N,N,N’,N’~N’’’,N’’’-hexaacetic acid (TTHA); diethylenetriamine- N,N,N’,N”,N’ipentaacetic acid (DTPA); and citrate] has been studied for masking zirconium(1v) in the determination of fluoride with an ion-selective electrode.Citrate was not suitable because it produced a prolonged electrode response. Of the aminopolycarboxylates, DTPA has a much greater masking ability than the others. Using DTPA at pH 5-6, fluoride was successfully determined at a concentration of 1 X 10-5 rnol dm-3 in the presence of up to 4 x 10-6 rnol dm-3 zirconium(1v). The proposed method was applied to the analysis of a number of zirconium(1v) fluoride compounds and ZrF4-based glasses after fusion with sodium carbonate. Keywords: Fluoride determination; ion-selective electrode; fluoride glass; zirconium masking; dieth ylenetriamine-N, N, N ’ , N”, N”-pentaacetic acid a c i d ( C DTA) ; In the last decade, heavy metal fluoride glasses, which gener- ally contain zirconium(1v) , hafnium(1v) or thorium(1v) , have attracted much attention in relation to their potential use in long-distance fibre optics.192 In accordance with progress in this area, the determination of fluoride in the presence of these metal ions has increasingly been required; e.g. , chemical durability testing is essential to assess the utility of each fluoride glass, as the transparency of the glass in the middle infrared region (0.2-8 pm) easily deteriorates as a result of attack from environmental water. To monitor the dissolution rate of a fluoride glass in an aqueous solution, a recent review3 recommends following the appearance of the dissolution products over a period of time, by analysis of the soaking solution, instead of following the loss in the mass of the glass.Metal components have been determined by various spectro- scopic methods, and fluoride by potentiometry with a fluoride ion-selective electrode.&g Zirconium(1v) , hafnium(1v) or thorium(rv) contained in these glasses has an extremely high affinity for fluoridel”15 and thus is expected to interfere seriously with the determination of fluoride. To eliminate the interference, commercially available TISAB (total ionic strength adjustment buffer solution), which contains citrate or truns-l,2-cyclohexanediamine-N, N , N’, ”-tetraacetic acid (CDTA) as a masking reagent for metal ions,16,17 has been used. These reagents are effective for masking common metal ions such as Al”, Fe3+ and Ca2+ but are not always suitable for metal ions in higher oxidation states.For example, Simmons and Simmons8 have observed that all the fluoride ions were not in the free form when CDTA was used as a masking reagent for zirconium(1v). No further work appears to have been undertaken since that report. In previous papers,l8-20 Yuchi and co-workers have studied the reaction of the fluoro complexes of trivalent metal ions with various masking reagents and found that mixed ligand complexes were generally present with a masking reagent and fluoride. Reagents forming less stable mixed ligand complexes are more efficient for the masking of a metal ion in the determination of fluoride by potentiometry using a fluoride ion-selective electrode. It has been found that ethylene- diamine-N, N , N‘, ”-tetraacetic acid (EDTA) complexes of tetravalent metal ions also form stable mixed ligand com- plexes with fluoride.21 In the present paper, six chelating reagents [EDTA, CDTA, N’-(2-hydroxyethyl)ethylenediamine-N, N , N‘-tri- acetic acid (HEDTA), diethylenetriamine-N,N,N’,N”,N’- pentaacetic acid (DTPA), triethylenetetraamine-N, N , N‘, - N’, N”, N”-hexaacetic acid (TTHA) and citrate] have been examined for the determination of fluoride in the presence of zirconium(1v).Using DTPA as a masking reagent, fluoride was successfully determined in some zirconium fluoride compounds and ZrF4-based glasses. As fluoride in these materials is prone to be replaced by oxygen-containing species such as OH- and 0 2 - ions, these data will be complementary to the results for the determination of oxygen in fluoride glasses by charged particle activation analysis.22 Experimental Reagents All the reagents used were of analytical-reagent grade.Potassium nitrate was recrystallized twice. Carbonate-free potassium hydroxide solution was prepared as described elsewhere .23 Potassium fluoride was dried in a platinum crucible for 24 h at 110°C. Fluoride solutions were stored in polyethylene containers. Zirconium(1v) stock solution was prepared by dissolving zirconium(1v) oxide nitrate, ZTO(NO~)~, in a 4 mol dm-3 nitric acid solution, which prevents the formation of polymeric hydrolysed species .24 The concentration of zirconium(1v) was determined by titration with EDTA in 1 rnol dm-3 HN03 at 90°C using Xylenol Orange. The zirconium tetrafluoride (Morita Kagaku Kogyo) and potassium hexafluorozirconate (Kanto Chemicals) used as samples were of technical grade.Measurement The equipment used was the same as that described previ- ously. 1 ~ 1 All the potentiometric measurements were perfor- med at 25 “C and at an ionic strength of 0.1 rnol dm-3 KN03. The effects of pH and the concentrations of fluoride and zirconium(1v) on the electrode response were studied by utilizing various masking reagents; a series of solutions containing 2.5 x 10-6-2.5 x 10-4 rnol dm-3 zirconium(Iv), 1.25 X 10-5-1.25 x 10-3 mol dm-3 fluoride ion and 5 X 1 0 - h l x 10-1 rnol dm-3 masking reagent were titrated with 0.1 rnol dm-3 potassium hydroxide. After each addition the46 ANALYST, JANUARY 1991, VOL. 116 100 80 - u L 0 cc 60 40 3 4 5 6 7 A n A " 0 0 0 0 0 o u A P,,,, PH 3 4 5 6 7 3 4 5 6 7 Fig. 1 R or R'(%) versus pH.Masking reagent: (a) none; ( 6 ) EDTA; and 1.25 X R: 0, A and 0; and R': 0. cF:cZr = 5. cL: cZr = and 0, 0, 1.25 X pF and pH were measured with a fluoride ion-selective electrode and a fluoride-resistant glass electrode. The recovery of fluoride, R = [F-]/cF (cF = total fluoride concentration), was calculated from the pF values. The recovery, taking the protonation of fluoride into account, R', was also calculated by using the relevant constants and both pF and pH values's ( K is the formation constant in each instance). R' = ([F-] + [HF] + ~[HF~-])/CF = ([F-] -k KHF[H+][F-] + 2KHF2KHF[H+][F-]2)/CF Recommended Procedure A 0.05 g portion of the sample to be analysed is placed in a platinum crucible and covered with 1 g of Na2C03.The crucible is heated at 900 "C for 15 min. The cooled melt is digested with 60-70 cm3 of 1 X rnol dm-3 DTPA. After dissolution, 20 cm3 of 1 rnol dm-3 HN03 are added, and the solution is diluted to 250 cm3. After a further 500-fold dilution the solution is analysed for fluoride. Results and Discussion Zirconium(1v) seriously interferes with the determination of fluoride as shown in Fig. l(a). Although R increases with an increase in pH or with dilution of the sample, it does not reach 100% in the pH range suited to the use of the fluoride ion-selective electrodes. Addition of a masking reagent generally improves the recovery. The effects of pH and the concentrations of fluoride, zirconium(1v) and masking reagent were studied for each system (Figs.1-3). Effects of pH and Fluoride Concentration In the presence of a masking reagent, R also increases with pH, steeply in an acidic medium and gradually in a neutral medium [Fig. l(b) and (c)]. The increase in R at pH <5 is due to the deprotonation of HF. As the formation of HF is negligible above pH 5 , a plateau or an inflection point appears in the graph of R versus pH depending on the concentrations of fluoride and zirconium(1v). At higher concentrations of fluoride and zirconium(rv), the formation of ZrF, is mainly responsible for the interference in a slightly acidic medium, whereas at lower concentrations of fluoride and zirconium(1v) the interference is caused by ZrLF. [As the fully deprotonated ligands (L) used in this study have different electric charges, the net charges on the zirconium complexes are different to each other and therefore have been omitted for simplicity.] ZrF, + L' ZrLF + ( n - 1)F- (1) ZrLF ZrL + F- (2) (c) DTPA.cF/mol dm-3: 0, 1.25 x 10-3; A , - 3 : 80 1 ," 6ot; I , , I 3 4 5 5 7 PH Fig. 2 Comparison of aminopolycarboxylates as masking reagents for ZrIV. Masking reagent: 0, DTPA; 0, TTHA; A, HEDTA; 0, CDTA; and ., EDTA. cF = 1.25 X 10-5 rnol dm-3. cZr = 2.5 x 10-6 rnol dm-3. cL = 5 x 10-6 mol dm-3 Both equilibria, particularly that given by equation (l), shift to the right by simple dilution of the sample. Hence, the sample should be diluted as much as possible within the dynamic range of the fluoride ion-selective electrodes. Such an effect has also been utilized to eliminate interference from alumi- nium .20,25 The increase in R with pH found for the EDTA system [Fig.l(b)] at pH >6 is ascribed to the replacement of fluoride in the mixed ligand complexes by hydroxyl ions to form ZrL(0H) or Zr2b(OH)2. ZrLF + OH- e ZrL(0H) + F- lr 1/2[ Zr2L2 ( OH121 (3) The R versus pH curves obtained agree well with those calculated using the relevant stability constants.21J6-28 Comparison of Masking Reagents Fig. 2 shows the masking abilities of aminopolycarboxylates for 0.25 x 10-5 rnol dm-3 zirconium(1v) at a total fluoride concentration of 1.25 x 10-5 mol dm-3. Satisfactory recovery was obtained only with DTPA at pH >5. As R exceeds 80% for a sample with cF : cZr = 5 in the neutral pH region (Fig. 2), the average number of fluoride ions bound to zirconium is less than 1.A higher concentration of an aminopolycarboxylate did not give a higher recovery. Hence, the equilibrium [equation (l)] is completely shifted to the right. In such solutions, the following relationships hold: (4) CF = [ZrLF] + [F-] cZr = [ZrL] + [ZrLF] ( 5 ) KFZrLF = [ZrLF]/[ZrL][F-] (6)ANALYST, JANUARY 1991, VOL. 116 47 100 I s 80 Q 60 I 0 0 I . 1 I I 1 3 4 5 6 7 PH Fig. 3 CF = 1.25 X 0, 0.1; a, 0.01; and 0, 0.001 Effect of citrate concentration on the recovery of fluoride. rnol dm-3. cZr = 2.5 x 10-6 mol dm-3. cLlmol dm-3: 0 ' 1 I 1 I 4 5 6 -Log (cz,/mol dm-3) Fig. 4 Calculated recovery of fluoride versus -log cZr when masking zirconium with A, DTPA and B, EDTA. CF = 1.25 x lo-' mol dm-3 For a solution containing known concentrations of fluoride and zirconium(iv), the recovery of fluoride can be calculated from these equations.The formation constant of the mixed ligand complex, I(FZrLF, for DTPA was found to be 103.80, whereas that for EDTA has been previously reported2I to be 104.52. Recovery of 1.25 x 10-5 rnol dm-3 fluoride in the presence of various concentrations of zirconium(1v) was calculated and is shown in Fig. 4. When DTPA was used as a masking reagent, the tolerable amounts of zirconium(1v) for 99 and 98% recovery of fluoride are 2 X 10-6 and 4 X 10-6 mol dm-3, respectively. These values correspond to 16 and 32% as the molar ratio of zirconium(1v) to total fluoride and are sufficient for the analysis of fluoride glasses, because the molar ratios of these glasses are generally lower than 25%.Using EDTA, on the other hand, the tolerable amounts of zirconium(1v) are 4 X 10-7 and 8 x rnol dm-3 corresponding to only 3 and 6%, respectively. The potentially octadentate ligand, DTPA, may be the correct size to form a stable and coordination-saturated complex with zirconium(1v) similar to bis(nitrilotriacetate)zirconium,zg and the resultant complex has a much lower affinity for fluoride ion. For citrate (Fig. 3), a slight increase in recovery with an increase in the concentration of citrate from 0.001 to 0.1 rnol dm-3 indicates a different reaction schenx. Even 0.1 rnol dm-3 citrate solution, however, has a -nasking ability inferior to DTPA. Moreover, a higher concentration of citrate results in a prolonged response time of the fluoride ion- selective electrodes, which has been pointed out in relation to the masking of alurninium.30,31 Pre-treatment Dissolution of zirconium fluoride compounds is not easy; e.g., 0.004 g of finely powdered ZrF, suspended in 100 cm3 of a 0.5 x 10-3 rnol dm-3 DTPA solution stirred continuously required 7 h at pH 6-7 and 3 h at pH 3 to dissolve.Fusion with sodium carbonate was examined as a general method of pre-treatment for the dissolution of zirconium fluoride com- Table 1 Determination of fluoride in samples containing zirconium Fluoride (%) Sample Measured Calculated ZrF4 43.5,43.6,44.0 45.5 ZB glass* 35.1,35.1,35.2 37.3 K2ZrF6 37.6,37.9,38.3 40.2 ZBLAN glass? 40.7,41.2,41.7 39.1 * ZrF4: BaF2 = 2 : 1. t ZrF4 : BaF2 : LaF3 : AIF3 : NaF = 53 : 20 : 4 : 3 : 20.pounds. After fusion, even a fluoride glass sample could be dissolved in a DTPA solution. As fusion for 15, 30 or 60 min produced the same results, 15 min proved to be sufficient. It was necessary to treat the cooled melt with a DTPA solution before neutralization with nitric acid, in order to avoid prolonging the dissolution time. For samples containing relatively large amounts of zirconium, small amounts of a white precipitate, zirconium hydroxide or hydrated zirconium oxide, were formed during the neutralization of the DTPA solution with nitric acid. As potassium hexafluorozirconate was soluble in water, fluoride could be determined without fusion. When the same sample was pre-treated as described above, a white precipitate was formed but the analytical results were in good agreement with each other; thus the amount of fluoride in the precipitate is negligible. The addition of nitric acid results in a pH of about 6, which gradually increases with time owing to the evolution of C02.Neither the precipitation of zirconium compounds nor the increase in pH interferes with the subsequent determination of fluoride. The solutions thus obtained can be kept for at least 1 week without any deterioration. Determination of Fluoride in Samples Containing Zirconium Fluoride in commercially available zirconium fluoride com- pounds of technical grade and in fluoride glasses was determined using the proposed procedure. Diethylene- triamine-N, N , N ' , A'", N"-pentaacetic acid was effective for masking Ba2+, La3+ and relatively small amounts of AP+.The results in Table 1 show a satisfactory reproducibility. Purities of commercial ZrF, and K2ZrF6 were 96.2 and 94.3%, respectively. The infrared absorption band at 1640 cm-1 suggests the presence of strongly adsorbed water molecules .22 Prior separation of fluoride by conventional steam distil- lation was virtually impossible, although a recent paper32 describes a modified method, which is effective in the presence of zirconium(1v). As demonstrated above, poten- tiometry with a fluoride ion-selective electrode by using DTPA as a masking reagent is a more convenient and less time consuming method for the determination of fluoride in samples containing zirconium. We thank Professor Y. Kawamoto and Asahi Glass Company Ltd. for providing the samples.This study was supported by a Grant-in-Aid for Scientific Research (No. 02640444) from the Ministry of Education, Science and Culture, Japan. References 1 2 3 4 5 Comyns, A. E., ed., Fluoride Glasses, Wiley, New York, 1989. Almeida, R . M., Halide Glasses for Infrared Fiberoptics, NATO AS1 Series E-123, Martinus Nijhoff, Dordrecht, 1985. Moynihan, C. T., and Loehr, S. R., Muter. Sci. Forum, 1988, 32133, 243. Simmons, C. J., Sutter. H . , Simmons, J . H., and Tran, D. C., Muter. Rex Bull., 1982, 17, 1203. Doremus, R . H . , Murphy, D., Bansal, N. P., Lanford, W. A., and Burman, C., J . Muter. Sci., 1985, 20, 4445.48 ANALYST, JANUARY 1991, VOL. 116 6 Tregoat, D., Liepmann, M. J., Fonteneau, G., Lucas, J., and Mackenzie, J. D., J. Non-Cryst. Solids, 1986, 83, 282.7 Ravaine, D., and Perera, G., J. Am. Ceram. SOC., 1986,69,852. 8 Simmons, C. J., and Simmons, J. H., J. Am. Ceram. SOC., 1986, 69, 661. 9 Simmons, C. J., J. Am. Ceram. SOC., 1987, 70, 295. 10 Noren, B., Acta Chem. Scand., 1967, 21, 2457. 11 Noren, B., Acta Chem. Scand., 1969, 23, 379. 12 Noren, B., Acta Chem. Scand., 1967, 21, 2435. 13 Noren, B., Acta Chem. Scand., 1967, 21, 2449. 14 Baumann, E. W., J. Inorg. Nucl. Chem., 1970, 32,3823. 15 Klotz, P., Mukherji, A., Feldberg, S., and Newman, L., Inorg. Chem., 1971, 10, 740. 16 Frant, M. S., and Ross, J . W., Anal. Chem., 1968,40, 1169. 17 Harwood, J. E., Water Res., 1969, 3, 273. 18 Yuchi, A., Hotta, H., Wada, H., and Nakagawa, G., Bull. Chem. SOC. Jpn., 1987, 60, 1379. 19 Yuchi, A., Ueda, K., Wada, H., and Nakagawa, G., Anal. Chim. Acta, 1986, 186, 313. 20 Yuchi, A., Yanai, N., Wada, H., and Nakagawa, G., Analyst, 1988, 113, 1405. 21 Yuchi, A., Ban, T., Wada, H., and Nakagawa, G., Inorg. Chem., 1990, 29, 136. 22 Mitachi, S., and Tick, P. A., Mater. Sci. Forum, 1988, 32/33, 197. 23 Powell, J. E., and Hiller, M. A., J. Chem. Educ., 1957,34,330. 24 Noren, B., Acta Chem. Scand., 1973, 27, 1369. 25 Shiraishi, N., Murata, Y., Nakagawa, G., and Kodama, K., Anal. Lett., 1973, 6 , 893. 26 Intorre, B. I., and Martell, A. E., J. Am. Chem. SOC., 1960,82, 358. 27 Intorre, B. I., and Martell, A. E., Inorg. Chem., 1964, 3, 81. 28 Bottari, E., and Anderegg, G., Helv. Chim. Acta, 1967, 50, 2349. 29 Hoard, J. L., Willstadter, E., and Silverton, J. V., J. Am. Chem. SOC., 1965, 87, 1610. 30 Edmond, C. R., Anal. Chem., 1969,41, 1327. 31 Ingram, B. L., Anal. Chem., 1970,42, 1825. 32 Reig, F. B., Moreno, A. C., Pardillo, M. B., Martinez, V. P., and Adelantado, J. V. G., Mikrochim. Acta, Part 111, 1989,49. Paper 01031 09G Received July 1 Oth, 1990 Accepted September 1 Oth, 1990