10-3M standard test solution to give a potential difference of from 10 to 100 mV from the pure test solution. Selectivity coefficients were calculated from the Eisenman equation:
AE
[ + Ki
= (slope) log
1
__
where at and aA are the activities of the interfering anion and electrode anion in the test solution, respectively. When the concentration of the interfering anion used was large enough to change the activity of the test anion, appropriate corrections were applied. For these cases A E was given by :
A E = AEmeaaursd4- (slope) ( A h a A ) where Alog aA is the difference in the logarithm of the activity of the electrode anion in the two solutions. The interferences, expressed as selectivity coefficients, of several anions are shown in Table I1 for both the coated wire electrodes and the liquid-membrane electrodes. It is interesting to note that the coated wire electrodes give greater selectivity-i.e., lower Ki values in almost every case. This work demonstrates the general validity of the design and construction of so-called “liquid membrane” ion-selective
electrodes without an internal reference solution. Further refinement in fabrication should permit the development of electrodes sufficiently small to be used in intracellular measurements. The functioning of these electrodes requires that either the polymer-metal interface maintain a constant potential difference, or that this potential difference respond in a definite and reproducible manner to that of the test solutionpolymer interface. Although the first alternative might seem a more attractive hypothesis, the potential of a platinum wire in an electrolyte solution not containing a redox couple will drift significantly in relatively short time periods. Study of the effect of various electrode design parameters such as the nature of the polymer matrix (viscosity, dielectric constant, chemical type), nature of the solvent used for the ion association complex and for the polymer, nature of the metal (or metal/insoluble metal salt combination) wire is now under way in this Laboratory. RECEIVED for review September 13, 1971. Accepted October 21, 1971. ‘The authors acknowledge the assistance of a U.S. Public Health Service grant.
Separation of Fluoride and Chloride by Solvent Extraction Using Triphenylantimony(V) Derivatives Henry Chermette, Claude Martelet, Denise Sandino, and Jean Tousset institut de Physique Nuclkaire, Uniuersitk Claud Bernared de Lyon; (Institut National de Physique Nuclkaire et de Physique des Particules) 43, Bd du 11 novembre 1918, 69- Villeurbanne, France RECENTLY, a number of organometallic salts have been studied, of which many react with halides and are extractable using organic solvents (1-3). But such a halide separation is not selective, as many anions (nitrate, sulfate, all halides, perchlorate, permanganate, perrhenate, chromate, arsenate. . . ) are coextracted or form insoluble salts. Selective halide extraction requires halide complexes whose stability constants of formation are different. For instance, tetraphenylstibonium salts react with fluoride, giving tetraphenylstibonium fluoride, which is well known to be extractable by chloroform or carbon tetrachloride (4-7). But in this case, the separation is not complete and many ions interfere (8). Triphenylantimony (TPA) dihalides are stable complexes, very soluble in benzene or carbon tetrachloride, and have different stability constants (9). Thus, it seems that TPA dihalides formation is an interesting procedure for halide separations. In fact, if halide extractions are performed with an excess of TPA, the extracted salt is TPA hydroxyhalide, (1) R. Bock and H. J. Semmler, Z . Anal. Chem., 230, 161 (1967). (2) S. Tribalat, Anal. Chim. Acta, 4, 228 (1950). (3) M. F. Ali and G. S . Harris, Inorg. Nucl. Chem. Lett., 5, 701 (1969). (4) R. Bock and E. Grallath, 2.Anal. Chem., 222, 283 (1966). (5) L. H. Bowen and R. T. Rood, J. Inorg. Nucl. Chem., 28, 1985 (1966). (6) C. Martelet, Thesis, Lyon, 1970. (7) I. A. Carmichael and J. E. Whitley, The Analyst, 95, 393 (1970). (8) D. Aulagnier, H. Chermette, C . Martelet, J-P. Schlienger, and J. Tousset, Bull. SOC.Chim. France, 1971, 1135. (9) H. Chermette, Thesis, Lyon, 1971.
if the pH value is not too low. As it is suitable to have an excess of TPA for a complete halide separation, it is necessary to examine the TPA hydroxyhalide stability constants which have been determined recently’ (9, 10); their values are reported on Table I. The partition constants, which we have defined as the ratio of the solubilities of one chemical species in the organic and aqueous phase, respectively, are reported in Table I1 for TPA hydroxyhalide in benzene/water and carbon tetrachloride/ water systems. The high values obtained for all TPA hydroxyhalides, associated with the high value of the TPA hydroxyfluoride stability constant, indicate that fluoride can be easily separated by solvent extraction from various and complex media. Of course, the separability of fluoride by TPA hydroxyfluoride extraction depends on what other species are present which complex fluoride; however, it is shown in this paper that good separations have been obtained after adding, if possible, reagents masking the species complexing fluoride. The values obtained for TPA hydroxychloride show that chloride should be extracted easily from various media, but the selectivity is not so good as for fluoride and sometimes a double extraction is required. Hydroxyhalides may be obtained in a solvent extraction, either by halide exchanges or by direct synthesis in a reactional medium mixture with triphenylstibine in an oxidant medium. (10) H. Chermette, C . Martelet, D. Sandino, and J. Tousset, J . Inorg. Nucl. Chem., in press. ANALYTICAL CHEMISTRY, VOL. 44, NO. 4, APRIL 1972
857
Table I. TPA Hydroxyhalide Stability Constants X log,, KXOH + F
14.5 11.9 11.o
c1 I
0.8
0.8 0.8
Table 11. Partition Constants Solvent Species (C6Hb)zSbFOH (CsH5)aSbCIOH (CaH5)aSbIOH
CCla
CeH6 7,000
5500 5 500
10,ooo 9,000
+ 2X(CsHdsSb f HzOz 4- Y- e (C6H5)aSbYOH + OH-
(CsH5)3SbXz 4- Y-
4- OH-
% (C6H5)3SbYOH
(1) (2)
Both procedures give good results but the halide exchange is quicker than the second process. EXPERIMENTAL
Apparatus. All extractions were performed in 150-ml polypropylene separatory funnels having polytetrafluoroethylene stopcorks. Radioactivity measurements of fluorine18 and chlorine-34 were made by counting the two gammas of the p + annihilation, with a coincidence counting system connected to two 3-in. x 3-in. NaI(T1) well type scintillation detectors at 180 “C.
Other gamma radioactivity measurements were made with a modular counting system connected to one NaI(T1) detector. Measurements of pH were done with a digital pH meter Tacussel ISIS-4000. Reagents. The radioactive isotopes fluorine-18 and chlorine-34 were produced by irradiation of pure water or pure sulfur by, respectively, alpha particles of 54 MeV or deutons of 27 MeV, in the Synchrocyclotron of Lyon. Chlorine46 and iodine-131 were obtained from the Commissariat 31 1’Energie Atomique (France). TPA diiodide was prepared by halogenation of triphenylstibine, by the procedure of Michaelis and Reese (11). Triphenylantimony dichloride and triphenylstibine were commercial products (Alpha Inorganics and Fluka, respectively) and were used without further purification since chloride and antimony analysis agreed well with theoretical values, and the melting point agreed with recent literature values (9, 12). Metallic salts and other reagents were of “pure for analysis” grade of purity. The water used throughout the study was of great purity (resistivity higher than 2 MQ cm). Procedure. All extractions were performed using 10to 20-1111 volumes of aqueous phase and 20 ml of organic phase, at 25 “C. In the aqueous phase were the halide ions to be separated, metallic and other ions, pH buffer ions (sodium phosphates for fluoride extraction, sulfuric acid for chloride extraction), and, if necessary, 1,2-diaminocyclohexanetetraacetic acid (DCTA) sodium salt solution. Hydrogen peroxide was (11) A. Michaelis and A. Reese, Justus Liebigs Ann. Chem., 233, 39 (1886). (12) G. 0. Doak, G. G. Long and L. D. Freedman, J. Organometaf. Chem., 4, 82 (1965).
PH
Figure 1. Effect of pH value-fluoride 1. 2. 3. 4. 5.
858
(lo1
N) extraction
TPA dichloride (CCL), 2 X 102M; 2 X 10-aM 2 x 10-4~ 2 x 10+M
2 x 1WOM
ANALYTICAL CHEMISTRY, VOL. 44, NO. 4, APRIL 1972
PH
-
Figure 2. Extraction of fluoride (10-aiV) and chloride( 10-8N) by TPA diiodide (2 X lO-3M, CCW added for triphenylstibine oxidation reaction. The organic solvent contained the triphenylstibine or the TPA dihalide (dichloride for fluoride extraction, and TPA diiodide for chloride extraction). Radioactive isotopes 1*F,34Cl,36Cl,1311 were used to determine the distribution ratio; distribution studies were performed at various total concentrations of halides, various interfering ion concentrations, and various pH values at equilibrium. The ionic strength was usually unity in aqueous phase. The samples were agitated on a mechanical shaker for 10 min for halide exchange reactions, if DCTA was not present in the solution; in other cases (DCTA present, or triphenylstibine oxidation reaction), 20 minutes were required for reaching equilibrium. Following agitation, the solutions were allowed to stand until complete phase separation had occurred (a few minutes). The phases were then physically separated and 10- or 5-ml samples were taken for gamma counting in beakers with identical geometry.
RESULTS AND DISCUSSION Kinetics of Extraction. The halide extraction ratio, i.e., the percentage of halide (in all chemical states) found in the organic phase, has been measured in definite shaking conditions. The halide exchange reactions are quicker than the triphenylstibine oxydation with hydrogen peroxide in a halide medium. For instance, a 99% extraction ratio of fluoride needed only three minutes shaking in the first case (with TPA dichloride) whereas 20 minutes were required for the same result using the second way (triphenylstibine hydrogen peroxide). Influence of pH Value (at Equilibrium). The pH is an important parameter of the extraction, as it measures the hydroxide ion concentration. Because of the great stability of TPA dihydroxide, too high a pH value can affect the halide
+
extraction ratio, as is shown on Figure 1 for fluoride extraction. It is then necessary to control the pH value, and a pH buffer is often required. Influence of Some Interfering Compounds on TPA Halide Extraction. Fluoride and chloride extraction ratios may be disturbed if compounds or ions interfering with the system are present in the solution. It is possible to classify these interfering compounds in four groups: The first class contains reducing agents, like sulfide or thiosulfate ions, which give triphenylstibine by reaction with the TPA group. Triphenylstibine alone does not give extractable halide complexes, so that a preventive oxidation of such ions is required; it has been shown that an excess of oxidant does not interfere with TPA dihalide extraction. The second class consists of ions which form complexes with the TPA group and are therefore competitive with halides ; generally, their TPA complex stability constants are lower than TPA halide constants (9) but a great concentration of such ions will affect the halide concentration ratio. For this reason, the use of pH buffers containing competitive ions like acetate or citrate, isprohibited. However, some pH buffer ions do not interfere at all with halides ; thus, phosphate or sulfate ions have been used in high concentrations for stabilizing the pH values. The third group of the classification contains some organic compounds giving association salts with hydrohalic acids ; in this case the halide ions are extracted by TPA group with a lower separation ratio. For instance, it is not possible to separate fluoride or chloride from media containing pyridine or pyrocatechol at concentrations exceeding lO-'M, whatever the pH value is. Consequently, high concentrations of similar organic compounds must be avoided. The fourth class consists of halide complexing ions. Some metallic cations give, with halide, complexes of great stability ANALYTICAL CHEMISTRY, VOL. 44, NO. 4, APRIL 1972
859
and it becomes difficult to separate the halides from the solution. For example, cations like aluminium, zirconium, uranyl, and ferric iron are well known fluoride reactants. It is then necessary to add a metal complexing reactant like the well known EDTA or, better, the 1,2-diaminocyclohexanetetraacetic acid (DCTA). In this way, equilibrium is displaced and fluoride ions can be separated from aqueous solutions. We have studied fluoride separation from aqueous media containing various metallic ions [aluminium, zirconium(IV), ferric ion, uranyl] and have obtained a complete separation after addition of DCTA and phosphate buffer to work at a pH value of about 6. The results obtained with an aluminium medium have been presented in a recent paper (13). We have also tested that various cations and anions do not interfere. The separability of fluoride does not depend on what anions are present, like phosphate, nitrate, sulfate, cyanide, and other halides (all at concentrations higher than lM), permanganate, chromate, and perchlorate (all at concentrations of 5 X lO-ZM), etc. In the same way, we have noticed that alkaline cations at concentrations higher than 1 M and some metallic cations (Ni2+, Crs+, Mn2+, Cd2+, Ag+, all 5 x 10-zM) produce no decrease in fluoride extraction ratio.
Fluoride can be easily separated from chloride when operating at a pH value high enough for chloride not to be extracted, as shown on Figure 2. Similar results have been obtained for chloride separation: generally the extraction ratio value is about 90% at pH 1.5, so that for a better separation a double extraction is required, as some anions are partly coextracted (bromide, iodide). Moreover, bromide should not be present at concentrations exceeding 10-zN, nor should iodide, cyanide, and nitrate at concentration higher than lo-”. Silver should not be present at all. [At pH 1.5, it is not complexed by EDTA or DCTA (Id)]. Chloride has been separated from fluoride medium after addition of aluminium(II1) ions, giving a complex the stability of which is more than sufficient for chloride separation to be achieved: with TPA diiodide 2 X 10-3M, we obtained a chloride extraction ratio of 90%, and the fluoride (5 x lO-*M) was not extracted at all (less than 1 %).
(13) H. Chermette, C. Martelet, D. Sandino, and J. Tousset, C . R. Acad. Sci., 273, C-543 (1971).
(14) A. Ringbom, “Complexation in Analytical Chemistry,” John Wiley and Sons, New York, N.Y., 1963.
ACKNOWLEDGMENT
The authors thank G . Baudin for his interest in the planning of this work and for his helpful advice and suggestions.
RECEIVED for review July 29, 1971. Accepted October 21, 1971.
Improvements in the Determination of Sulfur Hexafluoride for Use as a Meteorological Tracer P. G . Simmonds, G. R. Shoemake,’ and J. E. Lovelock International Science Consultants, 5200 Palm Drive, La Canada, Calif, 91011
H. C . Lord2 Environmental Data Corporation, 608 Fig Avenue, Monrovia, Calif, 91016
SULFUR HEXAFLUORIDE @Fa) has been shown to be a particularly useful tracer for meteorological studies of moving air As a gas, it is inert, odorless, nontoxic, is masses (I-@. both chemically and thermally stable under normal conditions, and may be conveniently dispensed from cylinders at moderate pressures. Furthermore, it may be detected after Present address, Texas Engineering and Science Consultants, Inc., 1345 Blalock, Houston, Texas 77055. 2 To whom all correspondence should be addressed. (1) G . F. Collins, F. E. Bartlett, A. Turk, S. M. Edmonds, and H. L. Mark, J. AirPollut. C o w . Ass., 15, 109 (1965).
(2) B. E. Saltzman, A. I. Coleman, and C. A. Clemmons, ANAL.
CHEM., 38,753 (1966). (3) C. A. Clemmons, A. I. Coleman, and B. E. Saltzman, Paper presented at the 152nd National Meeting of the American Chemical Society, Division of Water, Air, and Waste Chemistry, New York, N.Y., Sept 1966. (4) A. Turk, S. M. Edmonds, and H. L. Mark, Enciron. Sci. Technol., 44, 2 (1968). ( 5 ) L. A. Nieineyer, and R. A. McCormick, J . Air Polfur. Contr. Ass., 18,403 (1968). (6) C. A. Clemmons, A. I. Coleman, and B. E. Saltzman, Enciron. Sci. Technol., 2, 551 (1968). 860
ANALYTICAL CHEMISTRY, VOL. 44, NO. 4, APRIL 1972
extreme dilution in air by a gas chromatographic procedure with an ultrasensitive electron absorption detector. A sensitivity of 10-5 ppm was reported in one study without concentration of the sample (2). Since oxygen also has an appreciable affinity for thermal electrons (7), it is not possible to measure dilute concentrations of SF6 in air directly, unless a method is provided for selective removal of the oxygen. A convenient solution to this problem, and the technique of choice is to separate the oxygen and SF6 chromatographically prior to their detection. Various solid adsorbents such as alumina and silica gel have been used successfully for this purpose, although in every separation reported oxygen is eluted before the SFs. For very low concentrations of SF6, this presents a practical difficulty in that the trace of SF6is superimposed on the tail of a very large oxygen peak. To improve the separation by using a longer column, slowing the analysis, or injecting smaller samples has the undesirable effect of a reduction in detectivity. The response to oxygen is considerably diminished when helium is used as carrier gas, without a corresponding re(7) J. E. Lovelock, ANAL,CHEM., 35,474 (1963).