Anal. Chem. 1980, 60,34-37
34
experimental values of KO(* agree very well, for the most part, with theoretical values based on thermodynamic constants derived by the method presented. Agreement with literature values for the constants is an essential requirement for the establishment of the method. While it is clear that thermodynamic constants cannot be calculated from ion mobility spectra with accuracies comparable to the established MS techniques, the reasonable agreements listed in Table I lend strong support to the method presented here.
ACKNOWLEDGMENT L. J. Hart designed and implemented the computer interfaces and programs without which this work would have been much more difficult.
LITERATURE CITED Castleman, A. W., Jr.: Keesee, R. G. Chem. Rev. 1986, 86, 589-618. Blyth, D. A. I n Proceedings of the International Symposlum on Protection Against Chemical Warfare Agents, Stockholm, 1983; pp 65-69. Carrico, J. P.; Davis, A. W.; Campbell, D. N.; Roehl, J. E.;.Slma, G. R.; Spangler, G. E.; Vora, K. A,; White, R. J. Am. Lab. (Fairfield, Conn.) 1986, 18, 152-163.
(4) Mason. E. A. I n Pklsma ChromtOaretW: Carr, T. W.. Ed.; Plenum: New York, 1984; Chapter 2. (5) Carroll, D. I.; DzMIc, I.; Stlllwell, R. N.; Hornlng, E. C. Anal. Chem.
im
47 1956-1959 .__ ._.
(6) Parent, D: C.; Bowers, M. T. Chem. Phys. 1981, 6 0 , 257-275. (7) Lubman, David M. Anal. Chem. 1984, 56, 1298-1302. (8) Slegel, M. W. I n Plasma Chromatcgraphy;Carr, T. W., Ed.: Plenum; New York, 1984; Chapter 3. (9) Casselman, A. A.; Gibson, N. C. C.; Bannard, R. A. B. J. Chromatogr. 1973, 78, 317-322. (IO) Harden, C. S., Chemlcal Research Development and Engineering Center, Maryland, unpublished work, 1985. (11) Davidson, W. R.; Sunner, J.; Kebarle, P. J. Am. Chem. SOC. 1979, 101, 1675-1680. (12) Moet-Ner (Mautner), Michael; Sieck, L. Wayne J. Am. Chem. SOC. 1983, 705, 2956-2961. (13) Lau, Y. K.; Saluja, P. P. S.;Kebarle, P. J. Am. Chem. SOC. 1980, 102, 7429-7433. (14) Lubman, David M.; Kronick, Me1 N. Anal. Chem. 1983, 5 5 , 1486-1492. (15) Kolaitis. Leonidas; Lubman, David M. Anal. Chem. 1986, 5 8 , 1993-2001.
RECEIVED for review February 5, 1987. Accepted August 31, 1987* This work was presented in part at the Canadian Chemical Conference, Saskatoon, SK, 1986.
Isotopic Analysis of Lithium as Thermal Dilithium Fluoride Ions L. W. Green,* J. J. Leppinen, and N. L. Elliot
General Chemistry Branch, Chalk River Nuclear Laboratories, Chalk River, Ontario, Canada KOJ 1JO
A U Isotopic analysrS method based on measurement of Up+ lons by thermal lonlzation mass spectrometry was developed. The Li,F+ ions were volatlzed from sample fllaments by radiant heat from the center filament; the optimal mole ratio for fllamenl loading was 211 LI/F. The M o p e fractionation rate with this molecular specles was low and was negligible over the normal analysis perlod. For appllcatlon of the method to samples wlth complex mafrlxes, a two-column lonsxchange process was developed for separatlon of LI from the sample matrix and conversion to LlOH for mlxlng wlth HF. The preclslon of the entlre procedure, InclUcAng the chromatography, was 0.2 % relatlve standard deviation, and the accuracy was wkhln the same range provided m a s and column blas factors were corrected for.
The isotopes of lithium have many practical uses in the biomedical ( I ) , geological (2), and nuclear industries (3). For example, in the nuclear industry, 6Li is used as a tritium breeder in fusion reactor blankets because of its high cross section for the n,a reaction. The natural abundance of the mass six isotope is only -7.5%, thus for tritium generation enrichment of X i is desirable. Single-stage enrichment fadors of up to 4 % have been obtained by chemical exchange techniques ( 4 ) and currently several of these are being studied in our laboratories. For all of the above uses of Li, precise measurements of the 617 ratio are required. A number of laboratories (5-8) have reported isotopic analysis techniques that achieved reproducibilities between 0.1% and 0.7% RSD; in some cases mass spectrometers specifically designed for Li isotopes were used ( 5 , 6 ) . Recommended filament loading forms were the iodide,
chloride, and nitrate salts of Li, and atomic ions (Li') were generated by interaction of the sample vapors with an ionizing filament, except in the procedure of Brown et al. (6) in which electron impact ionization was used. For the technique used with large magnetic sector instruments designed for heavier elements, very rigid control of sample deposition and analysis parameters was required, especially for the ionizing filament temperature (8). Control of isotope fractionation is the most important factor in Li isotopic analysis, because of the low atomic mass and high relative mass difference between the isotopes. To obtain reproducible results, the volatilization rate of each ionic and molecular species must be constant or controllable during the analysis and reproducible from one loading to the next (9). However, determination of the volatilization rates of all of the important species is not possible with thermal ionization, because the neutral species are invisible. The practical approach has been establishment of reproducible volatilization rates for the atomic positive ions followed by further optimization of procedures until reproducible results were obtained (5-8). An alternative approach is to use a molecular species that greatly reduces the fraction of lithium vaporized as atomic ions or neutrals. Theoretical and experimental data have shown that the ratio of molecular to atomic species volatilized is a major factor in the fractionation rate, expecially for triple filament sources in which the evaporation and ionization processes are spatially separated (9-1 1). Furthermore, estimated decreases in fractionation rate caused by increases in proportion of molecular species volatilized were largest for lithium, which was the lightest of the elements modeled by Kanno (9). Of the compounds of lithium that would be of practical use for thermal ionization techniques, the fluoride compounds are
0003-2700/S8/0360-0034$01.50/0 @ 1987 American Chernlcal Society
ANALYTICAL CHEMISTRY, VOL. 60, NO. 1, JANUARY 1, 1988
among the most stable; for example, the heat of formation of LiF(s) is -146.6 kcal/mole, whereas that of LiI(s) is -64.6 kcal/mol (12). This paper reportes on use of lithium fluoride as the filament loading form for Li, and dilithium fluoride, Li2F+,as the analyte ion. An isotopic analysis method is described that can be used with large, single-detector, magnetic sector mass spectrometers; the type most isotope ratio laboratories are equipped with. Included in the paper is a chemical preparation procedure for application of the method to samples with complex matrixes.
EXPERIMENTAL SECTION Reagents and Materials. A cation exchange column was prepared by packing a 4-cm-i.d. polypropylene column with AG5OW-X8, 100-200 mesh resin (Bio-Rad Laboratories, Richmond, CA) to a height of 5 cm. To remove contaminants and convert the column to the H+ form, it was washed with 8 mL of deionized water, 10 mL of 1M HC1, and 20 mL of deionized water. An anion exchange column was prepared by packing a similar column to a height of 3 cm with AG1 X8, 50-100 mesh resin (Bio-Rad Laboratories, Richmond, CA). This exchanger was converted to the OH- form by repeated washings with 2 M NaOH (Aristar Grade, BDH Chemicals, Ltd., Poole, England) and deionized water. The HCl used for column washings and elutions was subboiling distilled (Seastar Chemicals, Sidney, BC), and the methanol was HPLC grade (Fisher Scientific, Fairlawn,NJ). Subboiling distilled HF was prepared locally. A lithium isotopic standard, L-SVEC (U.S.National Bureau of Standards,Washington,DC), was used for calibration purposes. Highly enriched 8Li and 'Li hydroxides (Oak,Ridge National Laboratories,Oak Ridge, TN) were used for preparation of a series of standards ranging from 7% to 95% 6Li. These were used in isotope exchange experiments and for testing analytical procedures. Apparatus. The mass spectrometer was a Nuclide (Nuclide Corp., State College, PA) 90° magneticsector instrument equipped with Cathodeon type 553 triple filament assemblies, a single Faraday cup detector, a Vacumetrics (Vacumetrics,Ventura, CA) ETP AEM 1000 electron multiplier, and an IBM PC-based automation system (13). The original center filament was replaced with Rhenium Alloys (Rhenium Alloys, Inc., Elyria, OH) zonerefined Re ribbon, which was outgassed at 1850 OC and 2 X 10" Torr for at least 3 h before use. Filament temperatures were measured with a Pyro Micro-Optical pyrometer (The Pyrometer Instrument Co., Northvale, NJ) that had been calibrated against a tungsten ribbon filament lamp. Reported temperatures are observed temperatures, uncorrected for window effects or emissivities. Procedure. A 25-100-pL aliquot, that contained 1 X lo-' mol Li, was loaded onto the cation exchange column and washed with 8 mL of HzO, 10 mL of methanol (to remove organics), and a further 8 mL of HzO. Li was separated from other cationic species by passage of 10 mL of 0.1 M HC1 and was eluted with 10 mL of 1M HCl. The last 9 mL of eluate were collected, reduced to a volume of 1 h) of lower fractionation, as observed by others (8). However, overall precisions of isotope ratios were consistently poorer than required (0.5% RSD) with each of these loading forms. The poor precisions were attributed to changes in the fractionation rates from one loading to the next, electronic instabilities associated with large steps in magnetic field or high voltage between masses 6 and 7, and slight drifting of the center filament temperature. In attempts to reduce fractionation effects and other low mass problems such as peak step size, various molecular species were considered as analyte ions. Efforts to thermally generate LiI+, Li21+,or Li212+ions from LiI deposits were unsuccessful, although these species have been observed in electron impact spectra (14,15).Lithium hydroxide deposits yielded Li20+ions, but the signals were too weak for quantitative work. Much stronger Li20+signals were obtained from LiN03deposits, however, there were interferences at masses 28 and 29, so the use of Li20+was not pursued. Lithium fluoride deposits yielded well-defined peaks for Li2F+,and after optimization of filament loading and analysis conditions, spectra such as that shown in Figure 1 were obtained. Optimal signals were obtained for -2-pg deposits of Li with a Li/F mole ratio of 2/1. There was evidence that LiF precipitated from the LiF solutions if left to stand several days, so samples were prepared freshly. Chemical Blank and Interferences. Lithium is a ubiquitous contaminant, and thus all containers had to be rigorously cleaned by soaking in 4 M HN03and all reagents were of the highest purity available (e.g. subboiling distilled). During evaporations, solutions or filaments were covered to avoid ambient atmospheric contamination. Ion exchange columns were repeatedly washed with the appropriate reagents and distilled, deionized water prior to use. These precautions maintained the Li chemical blank at negligible levels relative to the microgram quantities loaded. Large amounts of reagents, even if free of Li, had to be avoided because they suppressed the Li2F+signal, either by chemical competition for Li+ or P or by physical interference
38
ANALYTICAL CHEMISTRY, VOL. 60, NO. 1, JANUARY 1, 1988
Table 11. Li 6/7 Ratio in Enriched LiOH
anal no
6/7
atomic ration 20.697 20.603 20.651 20.670
AVERAGE
20.660 0.067 21.11 21.12
f S
bias corrected
ORNLb
Average of at least 24 ABBA cycles for separate loadings. See ref 17. 0,1386 20
26
36
44 t(mln)
Ffawe 2. Li,F+ 32/33 ratio vs time for 6.5% 'LI lithium at 1630 OC; each point represents a mean of six ABBA cycles.
measured 6/7 atomic ration
1
0.081 83 0.081 86 0.082 05 0.082 06 0.082 15 0.082 08
2 3
4 5 6
0.082 01 0.000 13 0.16%
f S
RSD knownb % bias
0.083 656 1.973
Each result is a mean of at least 24 ABBA cycles for a separate loading. bSee ref 16. on the filament. Also, some reagentsgenerated a large number of small but significant peaks in the mass 25-40 range. Filament Blank. All three filaments of the triple filament assemblies had to be outgassed at 1850 O C for at least 3 h to suppress the filament blank to negligible levels. The source was baked after each sample by heating a source-mounted triple filament assembly as described above, to remove Li species adsorbed onto the electrodes. In addition, the source electrodes were frequently removed and cleaned. These latter steps were necessary because, as the procedure involved volatilization of LizF+by radiant heat from the center filament, it was susceptible to memory effects caused by radiant heat volatilization of Li species on the electrodes. Experience showed that memory effect errors could be as high as 10% if the source was not baked between samples. Precision a n d Accuracy. Isotopic data showed that the fractionation rate with the Li2F+species was very low, in fact the degree of fractionation over the normal analysis period, 20-45 min, was negligible relative to the scatter of the data (Figure 2). During the warmup period, 0-20 min, the signal was not sufficiently stable to identify a consistent fractionation pattern. At times >45 min, the fractionation rate remained very low until the sample was exhausted. This control of the fractionation rate was attributed in part to the high proportion of molecular species volatilized due to the use of lithium fluoride as the loading form. Other processes may have also contributed: dissociative/associative processes on the filament to yield high molecular weight polymeric species as the volatilized species, dissociative/associativeprocesses in the vapor phase, and ionization processes that yielded stable LizF+ion beams (10,111. A detailed study of the species in the various phases would be required to determine, with certainty, the
-
analysis no 1 2
Table I. Li 6/7 Ratio in NBS L-SVEC anal no
Table 111. Repiicate Isotopic Analysis of Li through a Two-Column Ion-Exchange Procedure
3 4 5
6 f s
RSD direct (no columns) % bias
6/7
atomic ration
0.9384 0.9418 0.9393 0.9408 0.9421 0.9433 0.9410 0.0018 0.2% 0.9463 f 0.0018 -0.56
Average of at least 24 ABBA cycles for each sample run separately through the columns.
fractionation controlling processes. Results for six replicate analyses, including sample reloading, of a Li isotopic standard showed good precision (0.16% RSD)and a mas bias factor of 1.97% (Table I). This precision is identical to that achieved by the atomic ion technique in a similar series of analyses, also with a large single-detector mass spectrometer (8). The molecular ion technique was much less sensitive to ionizing filament temperature, which varied by h30 "C from loading to loading. A comparison of isotope ratio results for a highly enriched LiOH material with results obtained at Oak Ridge National Laboratories (ORNL) showed good agreement after correction for the masa bias factor (Table 11). The relatively large mass bias was attributed in part to the mass bias of the electron multiplier detector; the bias factor is close to that expected (1.6%)from the ratio of the velocities of isoenergetic Li2F+ ions. Ionization and dissociative processes of vaporized molecular species near the ionizing filament likely also contributed to the mass bias (11). Sample Preparation. Samples from fusion breeder blanket testa and isotope enrichment experiments contained reagents that severely suppressed the Li2F+signals and caused interferences in the mass spectra. In addition, many samples required treatment with strong reagents to liberate Li+, and this introduced more interferences. Removal of such interferences required removal of the reagents, and this was accomplished by ion exchange chromatography. Initial work with a cation exchanger and an HC1 eluent showed effective removal of the reagents and satisfactory signals for Li2F+,after addition of H F to the sample fraction. However, the HC1 matrix generated many interferences in the Li2F+mass region. Consequently, a second column, packed with an anion exchanger in the OH- form, was used to remove C1- and yield Li as LiOH. Water was used as the eluant to avoid addition of more reagents. Fraction collection was arranged to recover
Anal. Chem. 1988, 60, 37-39
almost all of the Li, to minimize isotope fractionation effects. Results, Table 111, for replicate analyses of a 50% eLi-enriched material, processed through the two column procedure for each analysis, showed good precision (0.2% RSD) and a small bias of 0.56% relative to results for the same material analysed directly (no columns). Correction for the column bias factor was not required for many applications, since it was reproducible and since only relative changes in isotope ratios were of interest. Otherwise the above correction factor was used. The bias may have been caused by isotope fractionation on the ion exchange columns, although chance of this effect was minimized by recovery of >98% of the Li. Applications. Aqueous solutions of lithium trifluoracetate were equilibrated with a solution of cryptand 221 in chloroform and, in a separate experiment, monobenzyl-15-crown-5 in hexanol. Lithium in the organic phases was released by back extraction into water. Isotope ratio measurements were reproducible within 0.2% (1RSD), and 6Li enrichments between l and 4% were observed for the organic phases. At one stage in the work ultrapure H N 0 3 (Ultrex, J. T. Baker Chemical Co, Phillipsburg, NJ) was used to wet-ash the organic phase and release Li+, but results showed significant contamination with natural Li. Thus, in general, concentrated reagents should be avoided and Li dissolution or liberation steps should be tailored to each type of sample. The Li isotopic analysis method can be readily extended to Li concentration determination by use of the isotope dilution technique. The high precision achieved with a common single-detector mass spectrometer, and the simple steps for purification of samples, make the method attractive for any Li determinations that require high precision.
37
ACKNOWLEDGMENT The authors thank L. Plante for preparation of the isotope exchange samples and some of the standards. Registry No. t i , 14258-72-1;'Li, 13982-05-3;LiF, 7789-24-4; Li2F+,50927-99-6.
LITERATURE CITED Lloyd, J. R.; Field, F. H. Blamed. Mass Spechorn. 1981, 8 , 19-24. Svec, H. J.; Anderson, A. R. Geochim. Cosmochim. Acta 1985. 2 9 , 633-641. Hastings, I. J. Canadian Fusion Fuels Technology Project Report CFFTP-B-66003 (AECL 9201); Chalk River Nuclear Laboratories, Chalk River, Ontario, 1986. Jepson, B. E.; Cairns, G. A. Monsanto Research Corp. Report MLM 2622, Mound Facility, Ohio, 1979. Svec, H. J.; Anderson, A. R. J . Sci. Instrum. 1986, 43, 134-137. Brown, H. L.; Blk. C.; Anbar, M. Int. J . Mass Spechom. Ion Phys. 1977, 25, 167-181. Michiels, E.; De Bievre, P. Int. J . Mass Spectrom. Ion Phys. 1983, 48, 369-372. Mlchiels, E.; De Bievre, P. Int. J . Mass Spectrom. Ion Phys. 1983, 49. 265-274. Kanno, H. J. Chem. SOC.Jpn. 1971, 44, 1808-1812. Moore, L. J.; Heald, E. F.; Fllliben, J. J. A&. Mass Spectrom. 1978, 7A, 448-474. Habfast, K. Int. J . Mass Spectrom. Ion Phys. 1983, 51. 165-189. CRC Handbook of Chemisby and Physics; Weast, R. C., Ed.; CRC Press: Boca Raton. FL, 1985. Green, L. W.; Barsczewskl, J. S.; Elliot, N. L. Int. J . Mass Spectrom Ion Processes 1985, 87. 253-265. Friedman, L. J . Chem. Phys. 1955, 23, 477-482. Matsumoto, K.; Kiba, N.; Takenchi, T. Talenta 1975, 2 2 , 695-697. Flesch, G. D.; Anderson, A. R.; Svec, H. J. Int. J . Mass Spec. Ion PhyS. 1973, 12, 265-272. Kent, R., Oak Ridge National Laboratory, Oak Ridge, TN, private communication, 1986.
RECEIVED for review January 12,1987. Resubmitted May 18, 1987. Accepted September 15, 1987.
Isotopic Fractionation of Gallium on an Ion-Exchange Column Lawrence A. Machlan and John W. Gramlich* Inorganic Analytical Research Division, Center for Analytical Chemistry, National Bureau of Standards, Gaithersburg, Maryland 20899
slgnmcant isotopic fractlonatlon of galllum has been observed during elution through an anlon-exchange resln In the thiocyanate form. Two samples of NBS Standard Reference Materlal 994 (certified @@Ga/"Ga= 1.506 76 f 0.000 39) were passed through the lon-exchange columns, and fractbns were analyzed for lsotoplc composltlon and galllum content by uskrg thermal lonlzstkn isotope dMlon mass Spectrometry. The results show a depletion of the ilght isotope In early fractions from the columns, wlth a steady increase In the llght Isotope throughout the elution. A materials balance of the product of the lsotoplc compositlon and galllum content of each fractlon Is In agreement wlth the lsotoplc composltlon of the starting material.
Isotope separation resulting from ion-exchange chromatography has been reported in the literature for nearly 50 years. Taylor and Urey (I),in 1938, found that by passing solutions of potassium salts through zeolites, the 39K/41K ratio could be altered by as much as *lo% from the normal
abundance ratio. Similar studies with lithium showed variations in the 6Li/7Liratio of up to 60% between the leading and trailing edges of the elution (1). Early work in this area was limited in accuracy due to the difficulty in controlling isotopic fractionation in the ion source, an inherent problem in the thermal ionization mass spectrometricprocess. In 1978, Russell and Papanastassiou (2) reported 40Ca/44Caisotopic variations of over 1 % between the first and last fractions from an ion-exchange separation. These authors employed a double spike technique to correct for variations in the mass spectrometric fractionation between analyses. Most reports of isotopic fractionation during ion-exchange chromatography have been observed with light elements, where the mass difference between isotopes is large relative to the average atomic weight of the element. Recently Fujii et al. have published observations of isotopic fractionation of copper, which approaches the gallium mass range, using electromigration through a cation-exchange membrane (3). Isotopic fractionation of gallium by physical processes has been previously observed. Variations of more than 10% have been reported when a continuous electrical current is passed
This article not subject to US. Copyright. Published 1987 by the American Chemical Society
