Analyst, February, 1971, Vol. 96, pp. 117-122 117 The Determination of Fluorine in Rock Materials by y-Activation and Radiochemical Separation BY J. S. HISLOP, A. G. PRATCHETT AND D. R. WILLIAMS (Analytical Sciences Division, Atomic Energy Research Establishment, Harwell, Berks.) A technique is described for the determination of fluorine in rock materials involving irradiation in a source of high energy y-photons to induce the 1:F (y,n) 18F reaction. Fluorine-18 is then separated from the radioactive matnx by distillation and its activity measured either in the distillate or as precipitated calcium fluoride and compared with that of irradiated calcium fluoride standards. The technique has been applied to the analysis of standard rock materials G1 (638 p.p.m.), W1 (221 p.p.m.), T1 (476 p.p.m.) and to Apollo 11 lunar fines (76 p.p.m.).Precautions are taken to eliminate interferences. The results obtained for the standard rocks are in good agree- ment with those of conventional methods but disagree with other activation results. The limit of detection of the method is 0.002 pg. SEVERAL methods have been developed for the determination of fluorine in rocks. These may be conveniently classified into chemical and activation techniques. Chemical methods, particularly those involving the spectrophotometric measurement of the coloured complex of fluorine with zirconium - Eriochrome cyanine R1s2 can be extremely sensitive but are susceptible to contamination errors unless rigorous precautions are taken. Activation methods on the other hand are much less susceptible to contamination errors and are independent of reagent blanks.Radiometric methods for the determination of fluorine have been reviewed by F~reman.~ Jeffery and Bakes4 have proposed a non-destructive technique involving fast-neutron activation in which the 19F (n,cc) 16N reaction is used, but because of the high limit of detection, viz., 0.5 per cent. of calcium fluoride, this technique is only applicable to fluorine ores and concentrates. Thermal-neutron activation to give fluorine-20 has also been used by Mapper at A.E.R.E., Hanvell, but the short half-life (11-2 s) of this nuclide and interference from the 28Si (n,p) 28A1 reaction restricts its application. On the other hand, activation by y-photons permits the determination of fluorine to be made via the 19F (7,n) l8F reaction.Fluorine-18 is a pure p+-emitter and its non-destructive determination in a rock matrix, except in a few exceptional cases, for example, in the absence of titanium-45 (/?+, half-life 3.1 hours), is likely to be di&cult. This technique, by using a Betatron, has recently been reported by Kosta and Siunecko5 for the non-destructive analysis for fluorine in a variety of other matrices. Car- penter6 claims that this technique is applicable to samples of minerals from marine and terrestrial sediments. Fluorine-18 produced by the 19F (n,2n) 18F reaction has been used to determine fluorine non-destructively in fluoro-organic materials.' The half-life of fluorine-18 (110 minutes) is sufficiently long to enable it to be separated chemically from the irradiated matrix.Wilkniss, Skinner and Cheek have used this technique following y-activation for the determination of fluorine in synthetic rain water8 by distillation of fluorine-18, and in sea waterg by adsorption of fluorine-18 on calcium sulphate. The distillation technique has also been reported by Reed10 and Reed and Jovanovicll for the determination of fluorine in meteorites and rock material. This technique has, however, produced significantly higher results for the analysis of standard rock materials than those reported by more conventional methods. The work reported in this paper is based on radio- chemical separation of fluorine-18 from y-irradiated rock matrices by using a modification of Willard and Winter's distillation technique, and efforts have been made to eliminate several possible sources of error in the technique reported by Reed,lo particularly with regard to interferences and standards. 0 SAC; Crown Copyright Reserved.118 INTERFERENCES- Fluorine-18 is not produced solely from fluorine and it is necessary to consider alternative reactions by which it can be formed.Several reactions may occur during y-activation, mainly 20Ne (y,d) and (y,np), 21Ne (y,t) and (y,nd), 22Ne (y,tn) and 23Na (r,na), together with a considerable number involving charged-particle activation, which include l60 (a,d) and (3He,p) and 1 8 0 (p,n). Because of the extremely low concentration of neon present in rock material, interference from this element is likely to be small. Similarly the flux of charged particles during y-irradiation is also likely to be low.Nevertheless, complete dismissal of charged- particle interference may not be justified. Wilknissl2 has investigated the production of low levels of fluorine-18 activity from y-irradiation of pure water and has shown that it may be produced by the l80 (p,n) 18F reaction, the protons resulting from the l60 (y,p) lSN reaction. As rock matrices are complex this or similar reactions may occur, but their effect would be expected to be reduced the lower the energy of y-radiation used. Similarly fluorine-18 may be produced by both the 19F (y,n) 18F and 19F (n,2n) 18F reactions, and thus the possibility of interference exists should a higher fast-neutron flux be present in the sample than in the standard resulting from the presence of an element with high y,n cross-section in the former but not in the latter.The relative specific activities of the two reactions under the experimental conditions used, however, suggest that this interference will be small. The most significant source of fluorine-18 interference in rock materials is that from 23Na (y,ncc), bearing in mind the high sodium concentration in many rocks. There are two methods of overcoming this interference : (a) irradiation can be carried out with y-radiation with maximum energy below 20.9 MeV, the threshold for the 23Na (y,na) reaction, and above 10.4 MeV, the threshold for the 19F (y,n) reaction; (b) alternatively, irradiation can be per- formed at an energy higher than the threshold energy for the 23Na (y,ncc) reaction, but with the inclusion of sodium standards from which a correction can be made for sodium inter- ference when the sodium concentration of the sample is known.With the Harwell 45-MeV electron linear accelerator it is extremely difficult to operate reliably at an energy as low as 20 MeV for long periods of time, consequently a conveniently low energy (23 to 25 MeV) was used and the sodium correction procedure adopted. Certain samples, particularly those with high sodium content, were irradiated by using the 17-MeV electron accelerator (Linac) at Harwell’s Wantage Research Laboratory, which eliminated sodium interference in these analyses. HISLOP et al. : DETERMINATION OF FLUORINE IN ROCK [Analyst, Vol. 96 EXPERIMENTAL IRRADIATION- About 50 to 100 mg of material were packaged in an aluminium sample container in double aluminium foil cups, 6 mm in diameter, which were sandwiched between two standards of Optran pure grade calcium fluoride and two standards of AnalaR sodium sulphate.The total volume occupied by the standards and sample was a right cylinder 6 mm in diameter and 0.5 cm long. The samples were irradiated for 30 minutes id the brehmsstrahlung produced by bombarding an air-cooled &inch thick tungsten target with electrons of maximum energy of about 23 to 25 MeV by using the Harwell Linac, or electrons of maximum energy 17 MeV by using the Wantage Linac. The Harwell irradiation facility has been described e1~ewhere.l~ The Wantage Linac, manufactured by Vickers, is a double-section, single klystron machine with energy variable between 5 and 17 MeV.Unlike that of the Harwell Linac the y-irradia- tion facility is not a permanent feature of the machine and the tungsten converter and sample container were placed in the dead-ahead electron beam position prior to each irradiation. To ensure homogeneous distribution of flux across the diameter of the samples the sample container was rotated about the axis through the centres of the samples during irradiation. Beam currents, measured on the tungsten converter, of 5 to 10 pA were obtained with the Harwell Linac and 50 to 60pA with the Wantage Linac. Samples were cooled by air in both instances. Specific activities of fluorine-18 were comparable with the two machines, the lower energy of the Wantage machine, and hence lower activation cross-section, being compensated for by the higher beam currents available.February, 19711 MATERIALS BY Y-ACTIVATION AND RADIOCHEMICAL SEPARATJON 119 CHEMICAL SEPARATION- After irradiation the sample was weighed into a platinum basin, 50 mg of Optran calcium fluoride carrier were added and mixed thoroughly with the sample, and the mixture fused for 2 minutes with 0.7 g of sodium hydroxide.The melt was cooled and extracted with 14 ml of distilled water. The solution was neutralised with 3 ml of 35 per cent. perchloric acid and transferred with 13 ml of water to a single-necked 50-ml distillation flask containing 25 ml of 35 per cent. perchloric acid, 1 ml of silver perchlorate (prepared as in reference 14) and some glass beads.Distillation of fluorine as fluorosilicic acid (H2SiF6) was then carried out by using a modification of Kubota's method.15 This technique, which involves the distillation of the fluorosilicic acid in an apparatus containing a heated (145' C ) outer jacket, in which the dilute acid is used as the source of steam, permits high, reproducible recoveries of fluorine in a small volume of distillate to be achieved. Perchloric acid was used in preference to sulphuric or phosphoric acids to simplify yield determinations. A 25-ml volume of distillate was collected in about 1 hour, which was either counted directly in a 50 mm diameter polycarbonate container or the fluoride precipitated as calcium fluoride. Yield determination was carried out in one of two ways: (i) if fluorine was counted in the 25 ml of distillate the yield was determined when the activity had decayed, by using the method of Popov and Knudson,16 which involved the precipitation of fluorine with excess of lanthanum nitrate and determination of the excess of lanthanum with cupferron ; and (ii) initially, problems were encountered in obtaining a reliable method of precipitating fluorine in a form that could be used directly for counting and for yield determination. That eventually used consisted of co-precipitation of calcium fluoride with calcium carbonate by a method similar to that used by Berze1ius.l' The pH of the 25 ml of distillate was adjusted to 8 by addition of 2 N sodium carbonate and 1 ml added in excess.To the heated solution were then added 10 ml of 0.5 N calcium chloride and, on further heating for 2 to 3 minutes, calcium fluoride - calcium carbonate was precipitated.The precipitate was cooled, centrifuged and washed with 10ml of hot water. It was then transferred to a platinum basin with ethanol and ignited in an oven at 700" C for 2 minutes. The residue was crushed and calcium carbonate decomposed with 10 ml of 10 per cent. acetic acid. After evaporating the solution to dryness the residue was re-heated on a hot-plate for 15 minutes. The residue was transferred to a filter tube with 10 ml of water and washed with a further 20 ml of water. The filter-paper and contents were then ignited at dull red heat and the residue transferred to a weighed counting tray. The total time required for distillation and preparation of precipitated calcium fluoride sources was about 3 hours.Experiments with radioactive calcium fluoride showed that yields from distillation were in excess of 85 per cent. and, for distillation PLUS precipitation, in excess of 70 per cent. A similar tracer experiment resulted in no significant loss of fluorine-18 during fusion. A further tracer experiment also showed that yields obtained by the lanthanum fluoride - cupferron technique were in excellent agreement with those for the radiochemical technique. SOURCE PREPARATION- Calcium fluoride standards used when the fluorine-18 from the samples was counted as 25 ml of distillate were dissolved in boric acid - nitric acid solution, a suitable dilution being made up in 25 ml with water and counted in similar geometry to the sample.When samples were counted as precipitated calcium fluoride sources the standards consisted of weighed aliquots (about 250 mg) of a solution of fluorine-18 prepared from the calcium fluoride stan- dards evaporated to dryness on a counting tray with an infrared lamp. To provide maximum P+-annihilation, and to obtain a reproducible source geometry, both fluorine standards and the sample, precipitated as calcium fluoride, were counted on aluminium trays sandwiched between two discs of copper sheet (100 mg cm-2). Amounts of about 30 mg of sodium sulphate irradia- ted to monitor 23Na (y,na) 18F interference were either diluted and counted in 25 ml of water or counted directly on a counting tray sandwiched between copper sheets as described.DETECTION OF ACTIVITY- All distillates or samples precipitated as calcium fluoride were examined with a Laben 512-channel 3 x 3-inch NaI (Tl) y-ray spectrometer to determine whether activities other than p+-annihilation radiation were present; in no instance was this found to be the case. The activity of all samples was then followed for at least four half-lives of fluorine-18 by120 HISLOP et al. : DETERMINATION OF FLUORINE IN ROCK [Analyst, Vol. 96 using a 100-channel 3 x 3-inch NaI (Tl) y-ray spectrometer incorporating a multi-position sample changer. The area under the 0-51-MeV annihilation full energy peak was then calcu- lated by Covell’s method,ls and decay curves obtained to verify the purity of the calcium fluoride sources. The peak area values were also analysed by a least squares fitting technique, which enabled accurate determinations of fluorine-18 activity at the end of irradiation, and hence the fluorine concentration of the samples, to be made.In addition, the errors resulting from counting statistics of the samples and standards were calculated. SENSITIVITY- Specific activities of fluorine-18 (at end of irradiation) obtained from irradiation of calcium fluoride standards were of the order of 10 to 100 counts s-l pg-l of fluorine with the irradiation and counting conditions already described. The background count on the detector used was about 5 counts s-l in the region of 0.5 MeV. Assuming that samples can be counted for a 2-hour period, 2 hours after irradiation the detection limit for fluorine (giving a fluorine count equivalent to 30 of the background) is 0.02 to 0.002 pg under the present conditions.RESULTS The calculated fluorine concentrations together with the calculated counting errors for individual analyses of three standard rocks and Apollo 11 lunar fines are given in Table I. G1, a granite, and W1, a diabase, are samples of standard rock material issued by the U.S. Geological Survey. T1 is a tonalite issued by the Geological Survey of Tanganyika. A sum- mary of the results calculated for each material is included. TABLE I FLUORINE CONCENTRATIONS OF ROCK MATERIALS Irradiation, Calculated interference Corrected fluorine concentration, p.p.m. MeV Sample from sodium, per cent.* (with counting errors) 23 to 25 w 1 13.7 214 f 9 t 23 to 25 W1 16.1 227 f 9 t 23 to 25 w 1 3 233 & 3t 23 to 25 G1 9.5 651 f 141.17 G1 - 620 f 7$ 17 G1 - 617 f 14: 17 G1 - 648 f lo$ 17 G1 - 655 f 13: 17 T1 - 496 f 18: 17 T1 - 488 f 9$ 17 T1 - 474 f 8$ 17 w 1 - 227 f 2$ 17 w 1 - 203 3 5f 17 T1 - 477 f 5: 23 to 25 Lunar fines 1.1 77 f I t 23 to 25 Lunar fines Nil 75 f 2: Summavy- Mean chlorine Number of Coefficient of variation, Sample concentration 8 determinations per cent. G1 638 18 5 3 w 1 22 1 12 5 6 T1 484 10 4 2 Apollo 11 fines 76 - 2 - * Based on fluorine-18 activity induced in sodium sulphate and sodium concentration obtained 7 Counted as 25 ml of distillate. : Counted as calcium fluoride. from literature. DISCUSSION Certain aspects of this work require further comment. The results obtained at higher irradiation energies and corrected for sodium interference are not significantly different from those obtained at lower energy in the absence of inter- ference, and confirm the validity of the interference correction.This lack of dependence on irradiation energy also indicates that charged-particle reaction interferences are not significant.February, 197 11 MATERIALS BY 7-ACTIVATION AND RADIOCHEMICAL SEPARATION TABLE I1 SUMMARY OF LITERATURE RESULTS FOR DETERMINATION OF FLUORINE IN STANDARD ROCKS24 9" 121 Author Huang and Johns2 . . Evans and Sergeant1 . . Shapirole. . .. .. Ingamells20 . . .. Jeffery21 . . .. .. Jeffery81 . . .. .. Shima22 . . .. .. Peek and Smithzs . . Reed and Jovanovicll . . Reedlo . . .. .. Present work G1 705 f 92 (3) 622 -f 8 (5) 700 720 (3) 629 (4) 627 (4) 720 & 70 600 (3) 831 & 66 935 f 509 1153 f 90 1080 f 116 1138 f 52 1023 1050 638 f 18(5) W l T1 208 f 5 (4) 228 f 8 (6) 200 - 390 f 10 (2) 455 & 5 (2) 305 (2) - 290 (4) - 290 (4) - 190 f 30 - 220 (3) - 279 f 28 - 329 426 f 35 - 552 f 24 - - - 221 f 12(5) 484 f lO(4) Techniques Spectrophotometric Spectrophotometric Spectrophotometric Chemical Photometric Titrimetric Colorimetric Spectrophotometric y-Activation y -Activation y- Activation Numbers in brackets indicate number of determinations.The precision of each determination calculated on the basis of counting statistics alone is significantly less than the standard deviation calculated from several replicate analyses, which is in agreement with our experience of y-activation analysis and may largely be attri- buted to uncertainties in the technique used for the calculation of the relative y-photon flux in the irradiation position of the sample compared with that of the standards.Because of this uncertainty, the accuracy of individual analyses is estimated to be not better than & 10 per cent. A summary of the values recorded in the literature for the fluorine concentrations of standard rocks is given in Table 11. Several features can be observed: the present results are in general agreement with most of the published values, and are in particularly good agreement with those of Evans and Sergeant1 and Peek and Smith'3; and there are significant differences between the results of the present work and those of Reedlo and Reed and Jovano- vicll who also used a y-activation technique.Reed and Jovanovicll have commented on the differences between the two sets of results that they reported for G1 and W1 and attribute them to the fact that two different aliquots were used. This heterogeneous sampling may also be the reason why their results are higher than those of other workers. The fact that the results of the present work are in general agreement with those obtained with a range of other techniques, in which, presumably, a range of aliquots was used, indicates that y-activation does not necessarily produce high results for fluorine concentration in rock materials. We are indebted to the staff of the Harwell Linac, and to Mr. R. W. Marriott of the Wantage Linac, for their assistance and co-operation during the irradiation of samples.1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. REFERENCES Evans, W. H., and Sergeant, G. A., Analyst, 1967, 92, 690. Huang, W. H., and Johns, W. D., Geochim. Cosmochim. Acta, 1967, 31, 597. Foreman, J. K., Analyst, 1969, 94, 425. Jeffery, P. G., and Bakes, J. M., Ibid., 1967, 92, 151. Kosta, L., and Siunecko, J . , Analyt. Chem., 1970, 42, 831. Carpenter, R., Diss. Abstr. (B), 1969, 29, 2796. England, E. A. M., Hornsby, J. B., Jones, W. T., and Terrey, D. R., Analytica Chim. Acta, 1968, Wilkniss, P. E., Skinner, I(. J., and Cheek, C. H., Radiochim. Acta, 1968, 10, 76. Wilkniss, P. E., Ibid., 1969, 11, 138. Reed, G. W., Geochim. Cosmochim. Acta, 1964, 28, 1729. Reed, G. W., and Jovanovic, S., Earth Plan. Sci. Lett., 1969, 6, 316. Wilkniss, P. E., I n t . J. Appl. Radiat. Isotopes, 1967, 18, 809. Hislop, J. S., and Wood, D. A., U.K. Atomic Energy Authority Research Reflort, A.E.R.E. R.6165, 40, 365. H.M. Stationery Office, London, 1969.122 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. HISLOP, PRATCHETT AND WILLIAMS “Standard Methods of Analysis,” United Steel Co. Ltd.. Fifth Edition, Percy Lund, Humphries Kubota, H., Microchem. J., 1967, 12, 525. Popov, A. I., and Knudson, G. E., Analyt. Chem., 1954,26, 892. Kodama, K., “Methods of Quantitative Inorganic Analysis,” First Edition, Interscience Publishers, Covell, D. F., Analyt. Chem., 1959, 31, 1785. Shapiro, L., Prof. Pap. U.S. Geol. Surv., 575-D, 1967, 233. Ingamells, C. O., Talanta, 1962, 9, 607. Jeffery, P. G., Geochim. Cosmochim. Acta, 1962, 26, 1355. Shima. M., Sci. Pap. Inst. Phys. Chem. Res., Tokyo, 1963, 57, 150. Peek, L. C., and Smith, V. C., Talanta, 1964, 11, 1343. Fleischer, M., Geochim. Cosmochim. Acta, 1966, 29, 1263. -, Ibid., 1969, 33, 65. & Co. Ltd., London, 1961, p. 191. a division of John Wiley & Sons Inc., New York and London, 1963, p. 448. Received August loth, 1970 Accepted September 24th, 1970