Determination of Fluoride by Thermometric Titration W. L. Everson and Evelyn M. Ramirez Shell Development Co., Emerycille, Calif. Thermometric titration is useful for the determination of fluoride ion; the technique is simple, and repeatability and accuracy are reasonably good. I t also provides a convenient means for studying the general course and nature of metal-fluoride reactions. A number of cations were found to be suitable titrants; thorium, cerium, aluminum, and calcium are discussed. By selection of titrant and pH range, fluoride can be titrated in the presence of moderate to considerable amounts of sulfate, borate, phosphate, or silicate.
PURKAYASTHA ( I ) used thermometric titration to study formation of beryllium fluoride complexes. Priestly, Sebborn, and Selman (2), in connection with development and applications of an automatic titrator, titrated fluoride and halides with a mixed thorium nitrate-silver nitrate titrant. We found (1961) that fluoride ion could be determined by thermometric titration with calcium ion, and started a general survey to determine which elements might be useful as thermometric titrants. Thermometric titration is quite useful for observing the general course of metal-fluoride reactions as well as for analysis. Endothermic reactions are immediately apparent. Delayed precipitation, successive formation of complexes, unfavorable equilibria, slow reaction, and effects of competing ions and interferences can usually be inferred from the titration curve. A considerable number of cations gave evidence of reaction with fluoride ion. Four of them-thorium, cerium, aluminum, and calcium-are sufficient to characterize most of the useful types of behavior observed, and are discussed in some detail. The behavior of a number of other elements is noted briefly. EXPERIMENTAL
Titrations were made in 75-ml evacuated glass vessels with close-fitting Teflon lids ; solutions were stirred magnetically before and during titration. The general procedure was to add, in order, sodium fluoride solution, interfering ions (if interferences were being studied), pH-control reagents, and water to give a total volume of 50 ml. Titrant (0.2was added at delivery rates of 0.1 to 0.5 ml/minute by means of a dual syringe drive (Harvard Apparatus Co., Inc., Dover, Mass.). Temperature was indicated by a 10,000-ohm thermistor in a modified Wheatstone Bridge circuit; bridge output was displayed on a 1-mv strip chart recorder (Texas Instrument Co.), generally at a chart speed of 1.5 or 3.0 inches/minute and temperature sensitivity of 0.1 C for full-scale deflection. For titration in alcoholic media, part or all of the water was replaced by an equivolume mixture of isopropyl alcohol (IPA) and water, previously allowed to cool to room temperature. To eliminate heat-of-dilution effects, the differential titration technique of Tyson, McCurdy and Bricker (3) was used.
l.Ow
(1) B. C. Purkayastha, J. Ind. Chern. SOC.,24,257 (1947). (2) T. Priestly, W. S. Sebborn, and R. F. W. Selman, Analyst, 88, 797 (1963). (3) B. Tyson, W. H. McCurdy Jr., and C. E. Bricker, ANAL.CHEW, 33, 1640 (1961).
B-
D
E
1
Start of 'itra t ion
Figure 1. Typical titration curves obtained in thermometric titration of fluoride ion A . Idealized form of TT curve (Th) B. Slight ( a ) to extensive (b) rounding at equivalence point; slightly
unfavorable equilibrium or solubility, or slow reaction at end point (Ce, Al, Ca) C. Slight curvature, rising segment following the break. Two or more reactions with some overlapping-e.g., formation of two complexes with steps not cleanly separated (Zr) D. Initially linear, followed by rounding. Reaction takes place in more than one step, with initial step(s) going to completion but unfavorable equilibrium in later steps (Be) E. Delayed precipitation : ( a ) simple retardation, (b) initial complex-forming reaction indicated, followed by considerable heat release at start of precipitation (Li, Sc) GENERAL RESULTS
Figure 1 shows typical titration curves obtained, together with their usual interpretations. Of the four selected titrants, thorium gave the sharpest breaks (type A ) ; for the others, the breaks were slightly to moderately rounded (type B-a to B-6). Thorium, although having the best temperature sensitivity and giving the sharpest breaks in the absence of interferences, is rather sensitive to the presence of other ions. Cerium has poorer temperature sensitivity but appreciably greater tolerance for most anions. Aluminum has high tolerance for sulfate and phosphate. Calcium has a high tolerance for borate and, following treatment with zinc acetate, can be used in the presence of considerable silicate and phosphate (the precipitate of zinc silicate-phosphate need not be filtered off). In strongly alkaline solution (pH 11 to 12), titration with calcium gives an approximate value for fluoride in the presence of moderate amounts of aluminum, which is converted to aluminate at this alkalinity. Behavior of these cations is discussed later. Thorium. Titrations were made at about pH 2.5 (0.1M monochloroacetic acid). In the absence of interferences, the VOL. 39, NO. 14, DECEMBER 1967
1771
theoretical ratio of four is followed very closely. However, thorium is quite sensitive to some anions; this is partly a consequence of making titrations at high acidity, where ionization of HF is repressed and some types of interference are enhanced, and partly a characteristic of thorium chemistry. The effectsof various ions on the titration curve are shown in Figure 2 and Table I. Relatively small amounts of chloride and nitrate cause appreciable rounding of the break and slightly low results. Sulfate reacts endothermically with thorium, forming a weak complex; its effect is to compete with fluoride toward the end of the titration, depressing the curve and making the break considerably less sharp. Phosphate also reacts endothermically with thorium, forming a precipitate, and has a similar adverse effect. This may account for the descending slope after the break, although in general a slight descent merely represents trivial phenomena such as nonadiabaticity, cooler titrant, or (in differential titration) slight differences between the titration and reference cells. The effect of silicate can be ascribed to the formation of fluorosilicate : 6 F f HSi03-
4-5H'
=
Sips-' f 3H20
This reaction is exothermic; and while thorium reverses the reaction, by removing fluoride ion from solution, the net reaction is endothermic. The result is a curve like C-1 if both anions are present, or like C-2 if enough silicate is present to convert the fluoride to fluorosilicate. As hydrogen ion shifts the equilibrium to the right, silicate will cause increasing interference with decreasing pH ; thus silicate interference is more troublesome with thorium as titrant than with other titrants used at lower acidities. The effect of borate is similar; the assumed reaction is
+ + H+ = BF(OH)2 f HzO
H ~ B O S F-
plus possible further reaction, e.g. BF(0H)z
+ F- + H-
=
BFzOH f HzO
As with silicate, hydrogen ion shifts equilibrium to the right; this is consistent with the well-known retarding effect of boron in Willard-Winter distillation of fluorosilicic acid. Our experience with various titrants is similar; borate interference decreases markedly as pH increases, and is negligible above pH 5. Aluminum does not affect curve shape, but interferes seriously, This may involve reaction rate as well as stability of the aluminum fluoride complex; the interference is greater than might be expected from consideration of the stability constants alone (log Kl for Th = 8.7, for A1 = 6.1). Other cations which form stable fluoride complexes, e.g. iron (log K1 = 5.3) can be expected to interfere more or less seriously. Cerium. Cerium(1V) and Cerium(II1) both react exothermically with fluoride ion, but as Ce(1V) is more prone to hydrolysis and has no apparent advantages, the trivalent salts are preferred. Most titrations were made at pH 3.5, obtained with a chloroacetate buffer (1M HX, 1M NaX; 5 ml added per titration) but precision and curve shape appear to be equally good over the pH 2 to 5 range. Interference from borate and silicate should be less at the higher
PH. Results are generally similar to those with thorium, and effects of interferences on curve shape are qualitatively the same. Table I1 shows results in the presence of various ions. As compared to thorium, the greater tolerance for most ions perhaps outweighs the disadvantage of lower temperature sensitivity. Lanthanum gives results essentially equivalent to 1772
0
ANALYTICAL CHEMISTRY
S t a r t of
2
T
I
0.01"C
\
S t a r t of T i t r a tion
F
I
15 rng 'F
Figure 2. Interference of various common ions in titration of 15 rng F- with thorium nitrate A . No interference B. 10 mg Fe +a c. 5 mg HSiOa-, C-2 = 10 mg D. 2.5 mg AIfS E. 15 mg S04-2 F. 2 mg B (as HaBOa)
those with cerium. Presumably, the other lanthanides are also usable as titrants; however, Kury ( 4 ) states that fluorides from LaF, through EuF3 have a hexagonal crystal structure, while GdF3, TbFa, and DyFs are orthorhombic and considerably more soluble. Aluminum. In aqueous solution, in the absence of ions other than those of the reactants, titration of fluoride ion with aluminum ion shows very little heat output and no detectable break. With increasing IPA concentration, heat output increases and curve shape improves; some semblance of a break appears at about 15% v and at 30z v the break, corresponding to AlFel3, is well defined. In the absence of alcohol, addition of an indifferent salt (e.g., NaC1) or an acid-base buffer has a similar effect. In the presence of both IPA, and indifferent ions the effects reinforce each other and good titration curves are obtained at 2 0 x v IPA. Figure 3a shows the effect of increasing alcohol concentration. It is interesting to note that a quite different curve (Figure 3b) is obtained in the reverse titration of aluminum ion with 1M potassium fluoride. Apparently AlF6-3 forms immediately when fluoride is in excess, while with aluminum initially in excess, complex formation proceeds stepwise. (4) .T. W. Kury, Ph.D. Thesis, U. of Calif., 1953, USAEC-UCRL2271.
Fluoride taken, mg 15.0 15.0 8.0 8.0 8.0 8.0 8.0
8.0 15.0 15.0
Table I. Effect of Some Common Ions on Accuracy of Thorium Titration Ion added, Fluoride Fluoride Ion added,
amount, mg c1-, 200 Nos-, S04-2,
soh2, sod-’, HPOd-’, HPOd-’, HP04-2, &Boa-, HzBOs-,
200 6 15 130 3 6 9 7 14
14.2 13.8 8.0 8.1 7.2 8.4 8.8 9.3 14.8 15.1
14.9 14.9
HPOd-’, HP04-’,
3 9
14.2 11.0
14.9
~1+3,
5
9.9
...
...
taken, mg
found, mg
I
15.0 15.0
amount, mg HSiOa-, 2.5 HSiOa-, 7.5
15.0
~ 1 + 3
15.0 15.0 15.0 15.0 15.0
APS, ~ 1 + 3 ,
Fefs, Fef3, Fef3,
...
0.5 2.5
11.9 2.1
1
14.2 13.5 13.2
io 5
10
... ...
14.9 14.9
HSi03-, 5 HSiOa-, 25
15.0
14.9 14.9
Fe+3, Fef3,
14.9 14.1
1
10
24.7 24.7
HzBOa-’
Fluoride taken, mg 25.0 25.0 25.0
24.6 22.8
1
24.7 24.7 24.7 I a IPA concentration reduced to 10% v to avoid precipitation. Break badly rounded. b IPA and buffer omitted. Break somewhat rounded. c Precipitate (NaaFeF6?) formed on adding NaCl solution. NaCl solution omitted. 5 25
Fe+a Fefa Fe+3
0.5
3.5 3.5
Table IV. Effects of Some Common Ions on Titration with Calcium in Weakly Acid Solution Fluoride Fluoride Ion added, Ion added,
amount, mg Cl900 c13600 c16400”
16.4
...
24.7
24.7 24.7
0.0
... ...
...
Table III. Effects of Some Common Ions on Titration with Aluminum Ion added, Fluoride Fluoride Ion added, amount, mg found, mg taken, mg amount, mg 24.7 HP04-’ 5 24.7 c1600 24.7 HP04-2 50 12.7 c15400“ 24.7 HPOd-’ 100 24.4 NOa700 24.7 HSi031 24.3 NOS3300h HSi03- 2 24.7 25.0 HSi03- 5 24.7 Sod-2 550
Fluoride taken, mg 24.7 13.0
Fluoride found, mg 12.5 13.2
found, mg 24.8 24.6 25.0
25.0
NOa-
200
24.8
25.0
SOa-2
100
24.8
taken, mg 25.0 25.0
I
amount, mg HSiOs- 5 HSiOa- 25
Fluoride found, mg 24.4 24.3, 24.6 24.3 23.0 21.1 18.6 23.2 13.7b 18.Y
Fluoride found, mg 23.5 16.6
50.0 50.0
Al+S
0.5
~ 1 + 3
5
29.5 6.6
24.7 24.7 24.7 24.7
Fe”8 FeCS Fei3 Fe-S
1 2 5 10
23.4 23.6 31.7 23. lo
24.7 HBOa-’ 5 24.0 24.7 HBOa-2 25 24.2 24.7 HP04-’ 20 24.4b a IPA reduced to 10 v to prevent precipitation of NaCl. Endo-exo curve. Fluoride based on exo portion. Endo portion equivalent to 22 mg HP04-’. Neutralized with 0.5 NaOH until alkaline to phenolphthalein, 1 ml of NaAc-HAC buffer added, then IPA, etc. pH reading after titration, 5.1. ~
VOL. 39, NO. 14, DECEMBER 1967
1773
Table V. Effectiveness of Zinc Acetate Treatment in Inactivating Silicate, Phosphate, and Iron a 30%~
5%
Fluoride added, mg
Interference added, mg
ZnAcz added, ml
24.9 24.9 24.9 24.9 24.9 24.9 10.0
None HSi03 25 HSi03 50 HSi03 50 HSi03 100 HSi03 100 HSi03 25 HSi03 50 HSi03 100 HP04-' 42 HPOd-' 85 HPOd-' 85 HPOd-' 85 HPOd-' 42 HPOd-' 85 HPOd-' 30' HSiOa- 50) Fe+3 35
2 to 10 24.3-25.1' 4 25.5 5 26.4 6 25.8 8 25.8,25.8 80 28.1 4 10.2 5c 11.0 6 11.0 3 24.7 4 30.5 5 25.1 6 24.4 3 10.1 5 10.0 8 25.2 5 25.8 5 8.4
10.0 10.0
24.9 24.9 24.9 24.9 10 0 10.0
24.9 . .. 24.9 24.9
~ 1 + 3
5
Breaks Correspond t o AIF,---
Fluoride, found, mg
pH reading after titration 7.5-7.7 7.4 7.3 7.3 7.1
15%~
Start of Titration
-?e==
7.8 9.e 7.7 7.3
N e u t r a i Fluoride Solution Titrated with AI+"-
...
Effect of I P A
7.4 7.5 7.8 7.4 7.3 .
I
.
7.6 0.02oc
Procedure: Add sample, ZnAcs solution, make just acid to phenolphthalein. Add 3 ml 10% NHhAc solution, dilute to 2530 ml and add an equal volume of 1 :1 IPA. Reference vessel contains 20 ml water, 25 ml 1 :1 IPA, 5 ml 10% triethanolamine solution. b NH4Acsolution varied from 2 to 5 ml without significant effect. c Solution made just alkaline to phenolphthalein after adding NH4Ac solution. a
L
i
""4
/
AIttf
Start of Titration
0
e
Solution Titrated with Fluoride i n 5 0 % I~P A
Figure 3. Reaction of fluoride with aluminum ion The following conditions gave good results in the titration of 10 to 50 mg of fluoride: Mode, differential; salt, 10 ml of 10% NaCl; buffer, 5 ml 1M HAC-liM NaAc; solvent, 25 ml 1:1 IPA added, diluted to 50 ml; titrant strength, 0.25M; delivery rate, 0.5 ml/min; chart speed, 3 inch/min; and sensitivity, 0.1 O C span. The F/Al ratio deviates slightly from the expected 6 :1. If the titrant is standardized with 25 mg of fluoride, results are precise but 1.5 to 2.0% low at 50 mg and slightly high at 10 mg. Below 10 mg, results are erratic unless IPA concentration is increased to 30% v, in which case results are high by perhaps 5 % at the 5-mg level. The deviation is tentatively ascribed to the rather low value for the last-stage equilibrium constant (log K s = 0.5). However, repeatability is good and corrections can be made for the deviation. Acidity has little effect on fluoride recovery over the range pH 3.0 to 5.5, but there is some tendency to impairment of sharpness of the break beyond either end of this range. The above buffer gives a pH reading of 4.5 to 5.0. The chief merit of aluminum as titrant is its insensitivity to several anions which cause difficulty with other titrants. As it forms a soluble complex rather than a precipitate, there are no errors from coprecipitation. Considerable amounts of sulfate and phosphate can be tolerated; borate and silicate cause much smaller relative errors than with thorium and cerium, and the curves are essentially normal in shape. Results in the presence of various ions are given in Table 111. The rounding at very high chloride and nitrate concentrations probably represents reduction in IPA concentration as well as salt effects. Sulfate was added as the potassium salt; a precipitate, probably K3AlF6,formed during titration. No precipitate forms at comparable concentrations of sodium ion. 1774 *
ANALYTICAL CHEMISTRY
With phosphate present, aluminum phosphate precipitates after titration of fluoride is complete, but the heat of this reaction is very low and hardly shows on the titration curve, Although the curve shape is normal in the presence of small amounts of silicate, this anion interferes seriously. With large amounts, the curve rises after the break, Calcium. In aqueous solution, calcium ion gives a fair titration curve with a somewhat rounded break. As with aluminum, titration in alcoholic solution improves both temperature sensitivity and curve shape. Above 30% v IPA, problems may be encountered with solubility and heat-ofdilution effects; the best range appears to be 2 3 - 2 5 x v. Differential titration can be used to cancel out heats of dilution, or titrations may be made at a fixed alcohol concentration, with the titrant made up to have the same alcohol concentration. There appears to be a slight variation in fluoride recovery with titrant addition rate; curve shape and precision were slightly better at 0.5 ml/minute than at 0.2 ml/ minute. The optimum range, under the conditions used, was 10-50 mg. Unlike the other titrants, calcium can be used over a wide pH range. Curves are fairly satisfactory at acidity as high as pH 3.3. At pH above 4.5, there is no interference from borate. In the range from 4.5 to 5.5, small amounts of silicate, phosphate, and iron can be tolerated. In weakly alkaline solution (pH 7-8) moderate amounts of silicate and phosphate can be rendered noninterfering by treatment with zinc acetate; the precipitate need not be filtered off before titration. In strongly alkaline solution (pH 11-12), aluminum is completely converted to aluminate ; although coprecipitation is trouble-
some, an approximate value for fluoride can be obtained in its presence. The effectsof a number of common ions in weakly acid solution (pH 4.5) are shown in Table IV. Most of the curves were normal. Chloride had little effect within the limits of its solubility. The nitrate curve showed slight rounding and the sulfate curve appreciable rounding at the breaks, In the presence of phosphate, calcium phosphate precipitates first, endothermically; calcium fluoride then precipitates exothermically. It may be possible to determine both ions simultaneously, but it is anticipated that coprecipitation might be troublesome. Table V shows the effectiveness of zinc acetate treatment in eliminating interference of silicate, phosphate, and iron. Results with silicate are slightly high; increasing the amount of zinc acetate does not appear to reduce the error. Possibly a little silicate, which reacts exothermically with calcium ion in this pH range, remains unprecipitated by the zinc treatment. Aluminum interferes seriously. In a strongly alkaline solution containing only hydroxide and fluoride as anions, the titration curve is normal in shape but results are high by 10% or more. Apparently, there is some coprecipitation of calcium hydroxide. If aluminate ion is also present, a curve with an “inside-out’’ break is obtained, in which the transition is from precipitation of CaFz to the more strongly exothermic precipitation of calcium aluminate, This break is not very sharp. Triethanolamine represses the precipitation of calcium aluminate, giving a normal curve, but is not effective in preventing coprecipitation of calcium hydroxide. Yet another complication is that aluminate appears to have a solubilizing effect on calcium fluoride, so that apparent recovery varies both with the amount of fluoride and the amount of aluminum present. Table VI shows typical results obtained under these conditions; it is apparent that only approximate values are obtained for fluoride. The above behavior indicates that calcium aluminate is more soluble than calcium fluoride. However, it appears that aluminum (as aluminate) can be titrated quantitatively with calcium ion by omitting the triethanolamine. In a single test 20.0 mg of aluminum was taken and 20.7 mg found; the curve was normal, with a somewhat rounded break. The indicated stoichiometry was two atoms of calcium per atom of aluminum. Behavior of Other Elements. In the course of this work, a large number of cations were tested as titrants. In addition to those discussed above, exothermic reactions were obtained with Li, Be, Sr, and Ba, Sc and La, Ti and Zr, Fe(III), Sn(II), Sn(IV), Sb(III), Sb(V), Pb, and Bi; Mg reacts endothermically. Vanadium(V) showed some reaction at pH 1-3, but temperature sensitivity was low; vanadium(1V) gave no evidence of reaction. Cr(II1) showed a slightly endothermic reaction ; Mn(I1) and Moo1-* gave no evidence of reaction. Other elements and valences were not tested because it was judged from their general chemistry and position in the periodic table that either they would not react or would be unlikely to offer any advantage as titrants over the elements studied. Not all of those which showed reaction are satisfactory as titrants. Lithium showed delayed precipitation (Curve type E-a, Figure 1, but badly rounded) even in 80% v IPA. SrFz and BaFz are more soluble than CaFz; the curves are more rounded, and also temperature sensitivity is poorer. MgFz has undesirably high solubility, although the endothermic nature of the reaction is of some interest. Scandium behaves similarly to aluminum, forming an ScF6+ complex; it did not appear to have any advantage over aluminum. Lan-
Table VI, Titration of Fluoride with Calcium in Presence of Aluminate AluAlu-
Fluoride taken, mg 49.8 24.9 10.0
49.8 24.9
minum minum added, Fluoride, Fluoride added, Fluoride mg found, mg taken, mg mg found, mg 0 56.8 49.8 50 51.0 0 29.1 24.9 50 22.9 0 14.4 10.0 50 10
55.6
...
...
.,.
10 26.7 10.0 10 9.7 a Badly distorted curve with no definite break.
.
I
.
thanum, and presumably yttrium and the remaining lanthanides, behave much like cerium. Titanium forms a TiF6-2 complex and gives a curve with a single, well defined break; there is negligible interference from sulfate. However, it is necessary to work in strongly acid solution, and temperature sensitivity is low. Iron(II1) is inferior to aluminum; tin and antimony did not give well defined breaks in either valence. Lead, in a chloride-containing solution, showed delayed precipitation (like Li); “overshoot” was also observed, indicating slow reaction after precipitation started. Bismuth (as the Bi14- complex) gave peculiar V-shaped curves in differential titration, with the formation of a brick-red precipitate which was not analyzed but was suspected to be Bi12F. Zirconium showed the highest temperature sensitivity of any metal tested; it also forms the most stable fluoride complex of all metals (log IC = 9.8). However, it undergoes complex reactions and does not appear to be a useful titrant. The curves generally show two (sometimes three) well defined breaks, but these do not occur at definite stoichiometric ratios; the F/Zr ratio for the principal break varies continuously from over five at pH 4-5 to less than two in 0.11.OM HCI. A number of schemes were tried to correct this situation, but without success. Hafnium probably shows essentially the same behavior. Some work was done with organometallic compounds. Ferricinium chloride reacts with fluoride, but the reaction appears to be slow and/or the equilibrium unfavorable. Titanium and zirconium form cyclopentadienyl derivatives, (CjHJ2MClZ; the titanium compound showed reaction and gave a fairly well defined curve, but could only be dissolved to the extent of 0.02144 (in 75 % ethanol-25Z toluene). The zirconium compound dissolved in water, but apparently hydrolyzed, either concurrently or rapidly thereafter. The compound [Sb(CsH&]2S04,readily water-soluble, can be used as a titrant, but the solubility of the fluoride is undesirably high; the titration curve continues upward after the break. Tin forms a family of compounds with the formula R3SnX; the chlorides are appreciably soluble in alcohol and the fluorides much less so. However, where the chloride is readily soluble, the fluoride is likely to be too soluble to be of interest, and where the fluoride is highly insoluble, it is very difficult to find a suitable solvent for the titrant and titration solvent. Of six compounds examined, tributyltin chloride had the best balance of properties; reasonably good titration curves were obtained using differential titration in 40-50 IPA over a fairly narrow pH range (4.5-6.0). Apart from the 1:1 combining ratio, it appeared to have no advantage and some drawbacks as compared to other titrants. The reverse possibility, of using fluoride as a titrant for alkyltin halides, might be of interest; the general field of organometallic compounds is indicated as an interesting field for further study. VOL. 39, NO. 14, DECEMBER 1967
m
1775
DISCUSSION
It is no particular feat to obtain accurate results for fluoride in the absence of interfering ions. While most interferences can be eliminated by a Willard-Winter distillation, this step (despite its merits) is time-consuming and better avoided if possible, particularly if fluoride-retarding elements such as boron and aluminum are present, The chief merit of thermometric titration is that it provides a possibility for avoiding interference from sulfate, phosphate, borate, and silicate, and thus may eliminate the need for distillation in specific cases. It is not well suited for titrating the dilute solutions obtained in distillation. In 1954, Elving, Horton, and Willard (5) listed some 1400 references dealing with the determination of fluoride; Horton’s (6) recent revision and expansion of this work adds some 650 more, Nevertheless, he noted that “The determination
of ionic or elemental fluoride is still very difficult despite the continual appearance of new methods and the variations of older methods described in the scientific literature.” The present work has little effect on this situation, but it does represent an area not previously explored and it opens up some possibilities for handling certain special situations more expeditiously. RECEIVED for review July 3, 1967. Accepted August 31, 1967. (5) “Fluorine Chemistry,” Vol. 2, Chap. 3, P. J. Elving, C. A. Horton, and H. W. Willard, Eds., Academic Press, New York,
1954. (6) C. A. Horton, Sec. A, Vol. 7, in “Treatise on Analytical Chemistry,’’ I. M. Kolthoff and P. J. Elving, Eds.
entiometric Titration of Fluoride with etraphenylantimony Sulfate James B. Orenberg and Michael D. Morris Department of Chemistry, The Pennsylcania State University, University Park, Pa. 16802 Aqueous solutions of fluoride ion may be titrated with tetra phenyla nt i mo ny su Ifate , [(CsH&S bJZSO4.The relative error is +I.% over the range 1 x lO-3M to 5 X 10-2M fluoride. Fluoride i s extracted into chloroform as the ion pair (CfH5)4SbF. The aqueous phase fluoride activity is monitored potentiometrically with a fluoride sensitive electrode. Equal initial volumes of aqueous phase (pH 4-5) and extractant are used. The aqueous sample is made O.1f i n sodium sulfate to inhibit emulsification and facilitate phase separation. Thirty seconds of intimate phase contact (stirring) and 15 seconds for phase separation are required between additions of titrant. Phosphate, arsenate, arsenite, and sulfate do not interfere. Nitrate and perchlorate interfere, but are readily removed by the addition of tetraphenylarsonium sulfate; nitrate forms an extractable ion-pair with the reagent, while perchlorate forms a very insoluble salt. Sulfite and nitrite interferences are removed by oxidation with hydrogen peroxide to sulfate, and the nitrite to removable nitrate. The halides and thiocyanate interfere and are removed by silver nitrate precipitation followed by extraction of the nitrate with tetraphenylarsonium sulfate as above. Conditional partition coefficientsfor the fluoride extraction have been measured and used to demonstrate agreement between theoretical titration curves and experiment.
THESEVERAL METHODS commonly employed for fluoride titration are not completely satisfactory. The popular thorium nitrate titration ( 1 ) has several difficulties. The stoichiometry frequently deviates from the theoretical ThF4 and numerous ions interfere. The most serious interferences are phosphate, arsenate, arsenite, sulfate, and sulfite, all of which precipitate the titrant thorium(1V). Lanthanum acetate has been suggested as a conductimetric titrant (2), but fluoride must be (1) C. A. Horton, “Treatise on Analytical Chemistry,” I. M. Kolthoff, P. J. Elving, and E. B. Sandell, Eds., Part 11, Vol. 7, Interscience. New York, 1961, p. 259. (2) Ibid., p. 264.
1776
a
ANALYTICAL CHEMISTRY
separated from interferences by a prior Willard and Winter distillation. Direct (3) and back (3) titrations based on the precipitation of lead chlorofluoride, PbClF, have been recommended. Indirect titrations based on precipitations of calcium fluoride and titration. of excess calcium have also been recommended ( 4 ) . Phosphate and sulfate interfere with methods based on lead chlorofluoride or calcium fluoride precipitation. Many of the difficulties of standard fluoride titrations are overcome by the use of tetraphenylantimony sulfate, [(C6Hs)4Sb]SOa, as an extractive titrant for fluoride. Tetraphenylantimony salts were first employed as analytical reagents by Willard and Perkins ( 5 ) , who used the bromide as a precipitant for perchlorate and several other inorganic anions. Affsprung and May (6) used tetraphenylantirnony sulfate to precipitate carboxylates. Morris ( 7 ) has used this reagent io titrate perchlorate, using amperometric end point detection. Moffert, Simmler, and Potratz (8) first proposed tetraphenylantimony cation as an extractant For fluoride. These authors noted that while about 9 7 x of the fluoride could be removed from the aqueous phase in three extractions, 10-20% of the other halides present were co-extracted. Bowen and Rood ( 9 ) have used extraction of tetraphenylantimony fluoride into carbon tetrachloride to obtain carrier-free F l 8 . These authors have made a detailed study of the extraction system. They note that extraction is most efficient at pH 3. In more acid solution, formation of undissociated HF decreases the (3) Ibid., p. 265. (4) Ibid., p. 266. (5) H. H. Willard and L. R. Perkins, ANAL.CHEM., 25, 1634 (1953). (61 H. E. Affmrune: and H. E. May, Ibid., 32,1164 (1960). (7j M. D. M&ris,>bid., 37,977 (565). (81 K. D. Moffert. J. R. Simmler. and H. A. Potratz, Ibid., 28, \
,
I356 (1956). (9) L. H. Bowen and R. T. Rood, J. Inorg. N L I CChem., ~. 28, 1985
(1966).