Microdetermination of Fluorine JOSEPH SAMACHSON, NORMAN SLOVIK, and ALBERT E. SOBEL Department of Biochemistry, Jewish Hospital of Brooklyn, Brooklyn 38, N. Y.
b The apparatus described permits the distillation of microgram quantities of fluoride in a small volume of distillate with low blanks. A modified alizarin reagent was developed for the spectrophotometric determination of the distilled fluoride. Chloride interference is removed by addition of silver oxide to the distillation chamber. The method permits the determination of fluorine in different portions of small bones and teeth and may b e extended to the analyses of other substances such as blood, where high chloride content might otherwise interfere with the determination of small amounts of fluorine.
A
was needed for the determination of small amounts of fluorine in sm’all samples of bones and teeth. While work was in progress on such a procedure, the method of Singer and Armstrong (9) was published. This is based on the diffusion of hydrogen fluoride and applied not to bone itself, but to bone ash. To avoid the extra step of ashing with its possible loss of fluoride, the method of Willard and Winter (IO) was selected for modification. This method is based on the distillation of fluoride as hydrofluosilicic acid from a solution of perchloric acid containing silica or glass beads, followed by titration of the distillate with tkorium nitrate, using sodium alizarin sulfonate as indicator. I n the titration a colorless thorium fluoride complex is formed. The end point is indicated by the formation of a red thorium-alizarin sulfonate lake. Willard and Winter’s method has been modified by Gilkey, Rohs, and Hansen (4) and by Huckabay, Welch, and Metler (6) to give better and simpler control of distillation temperature and thus permit more accurate determinations. The method described by Huckaby, Welch, and Metler, however, gives a blank of 4 y and is not suitable for the determination of 1 to 10 y of fluoride. Extreme care must be exercised in order to avoid high blanks and to obtain significant results in the determination of microgram quantities of fluoride (8). The micrometkd of Armstrong ( I ) , an earlier adaptation of the Willard-Winter method, is tedious and time-consuming. METHOD
1888
ANALYTICAL CHEMISTRY
Moreover, in titrating small amounts of fluoride as suggested by these authors. it is difficult to determine the end point with sufficient precision. Much depends on the individual analyst’s color perception (2, 7’). Bien (S), using an improved indicator, found variations from 6.0 to 6.9 y per ml. in the fluoride equivalence of his standard thorium nitrate solutions. Icken and Blank ( 7 ) substituted a spectrophotometric method which allowed the reproducible estimation of fluoride in concentrations as low as 0.05 y per ml. Unfortunately, their color reagent is unstable and begins to settle out after a few minutes. If an additional determination is to be made 0.5 hour after the color reagent has been used, a new reagent and new standards must be prepared. After the color reagent has been added to the fluoride solution, a 2-hour waiting period is required for the color to attain stability. The improvements desired were of two kinds: adaptation to micro purposes to permit the distillation of all the fluoride in a small volume, at the same time keeping blanks low, and improvement of the color reagent. I n addition it was advisable to develop a method of removing chloride, both because it might be necessary to include the fluoride method as part of a scheme of analysis in which hydrochloric acid was used, and because it might be desirable to extend the analysis to other substances such as blood, which has a high enough chloride content to interfere. SIDE
VIEW
The method developed in this laboratory used a modified form of the apparatus of Huckabay, Welch, and Metler (6). KO silica or glass beads were added, the silica for the distillation being supplied by the apparatus itself. Sulfuric acid was substituted for perchloric. The fluoride was then determined spectrophotometrically as in the method of Icken and Blank, using a color reagent based on the half-neutralized buffer of Hoskins and Ferris (6) made up in aqueous solution ( I ) instead of in alcohol. EQUIPMENT
The apparatus (Figure 1) differs from the apparatus of Huckabay, Welch, and Metler in several respects. The inner chamber is reduced in size, as is the entire apparatus. The inner chamber narrows at the bottom. This permits the steam to bubble up through more of the liquid, and thus removes fluoride more efficiently. The chamber widens a t the top to prevent liquid from being carried over by foaming or bumping. Because the steam is passed through more slowly, the apparatus is more easily preheated and the steam inlet spiral is reduced to a single turn. The glass-to-glass joints between the distilling apparatus and the condensers are replaced by rubber tube connections in order to avoid strain and breakage during shipment and use and to permit flexibility. Steam n-as generated from doubleFRONT
VIEW
CONDENSER
I IOOMM.
1 SYM-
Figure 1 .
Apparatus for determination of fluorine
distilled water in a round-bottomed flask containing an exit tube and an escape tube controlled by a screw valve. The rate of steam formation was controlled with a heating mantle. A three-necked flask was used as a trap between the rate generator and the apparatus so that, in case of accidental backflow, no sulfuric acid would contaminate the water in the generator. The third neck mas also supplied with a bypass screw valve. The liquid used in the reflux chamber n-as sym-tetrachloroethane ( 4 ) with a boiling point of 146" C. This was found to react slowly when heated in moist air and it was protected by a soda limecalcium chloride tube inserted in a paraffined cork a t the top of the reflux condenser. Enough was added to form a layer 3 to 4 cm. deep a t the bottom of the reflux chamber. If too much was used, the apparatus required too much time for heating and cooling. The apparatus was heated with a 350-watt electric heater. The vapor was led into a narrow, 11ater-cooled condenser which consisted of l/r-inch tubing surrounded by an outer chamber 9 inches long. The distillate was collected in a 10-ml. volumetric flask if fluoride in amounts of 10 y or less was suspected, or in a 25-ml. flask if a larger amount was helieved present. Rubber tubing could be used for connections without giving high blanks. d rubber stopper was also used to replace a badly fitted ground-glass stopper of the inner chamber.
REAGENTS
Alizarin sodium monosulfonate, as Alizarin Red S, certified for staining bone (Allied Chemical and Dye Corp.), dissolved in double-distilled water to give a 0.2y0 stock solution. Thorium nitrate tetrahydrate, analytical reagent grade, dissolved in distilled water to give a 0.25% stock solution. Monochloroacetate buffer, containing 18.9 grams of analytical reagent grade monochloroacetic acid and 4.0 grams of sodium hydroxide dissolved in 100 nil. of distilled water. Fluoride Standards. A stock solution is prepared to contain 1 mg. of fluorine per ml. by dissolving 221 mg. of analytical reagent grade sodium fluoride in 100 ml. of distilled water. Another solution containing 10 y of fluorine per ml. is prepared by adding 1 ml. of the concentrated solution to 100 ml. of distilled water. Color Reagent. First 24 ml. of the chloroacetate buffer and 15 ml. of the 0.2% alizarin monosulfonate solution are mixed, then 6 ml. of thorium nitrate solution and 5 ml. of water are added. This order of addition must be preserved. The buffer and alizarin monosulfonate solutions need not be perfectly fresh. Once prepared, the color reagent fades slowly. It may still be used after several weeks, although fresher reagent is desirable.
PROCEDURE
Distillation. If the apparatus is being used for the first time or if new rubber has been used, steam is passed in rapidly and about 50 ml. of distillate is collected. The apparatus is then cooled. It is dried by rinsing with acetone and applying suction from a water pump to the steam inlet sidearm. A fragment of bone or a small sample of powder is dropped directly into the dried distillation chamber. Liquid samples may be pipetted so that they drop past the side arm. If large amounts of chloride are present in the sample, 100 mg. of silver oxide is dropped in. If the liquid sample is 1.0 ml. or more, 0.5 ml. of C . P . concentrated sulfuric acid is then added. When there is less liquid in the sample (or none), 1.0 ml. of diluted acid (1 to 1) may be added. The total volume should be no more than 2.0 ml.; otherwise bubbles of steam may form suddenly before the proper distilling temperature is reached and cause bumping. The heater is turned on, and as the tetrachloroethane begins to boil, steam is passed through slowly. If no more than 10 y of fluoride i5 expected, approximately 8.5 ml. of distillate should be collected in a 10-ml. volumetric flask over a period of 10 to 20 minutes. Because of the slow rate of passage of steam, visible fluctuations may occur, but even if the steam flow stops entirely for a moment or two, no harm is done so long as the generator is not allowed to cool and suck back the sample. If the steam flow is too rapid, the distillation of fluoride is less efficient and more distillate must be collected. At the same time, there is a greater danger that droplets of acid may be carried over and give high results. If larger amounts of fluoride are expected, a slightly more rapid steam flow may be permitted and more distillate collected (21 ml. in a 25-ml. volumetric flask in about 0.5 hour). I n either case a pair of blanks should also be run with a series of determinations. I n case much chloride is present and silver oxide has been added, there will be a silver chloride precipitate which does not dissolve in the sulfuric acid. This does not hinder the distillation. After the distillation the same sulfuric acid may be used for the next determination if the sample contained little organic matter. Otherwise, the acid is diluted with distilled water and removed by suction. The chamber is rinsed with distilled water and then with acetone, both of which are removed similarly. Even when silver oxide was used and silver sulfate crystallized out after cooling, the chamber could be easily cleaned. Determination of Fluoride in Distillate. The color reagent, prepared as indicated, is added t o the distillate in the receiving flask (1 ml. to a 10ml. flask, 2.5 or 3.0 ml. to a 25-ml. flask), and the volume is made up t o
the mark rrith distilled water. At the same time a series of standards is prepared containing 1 to 10 y of fluorine when diluting up to 10 ml. or 1 to 25 y of fluorine when diluting up to 25 ml. A bleached blank containing excess fluorine is prepared by adding 100 y of fluorine (0.1-ml. stock standard) to the 10-ml. volumetric flask and 250 7 of fluorine (0.25-ml. stock standard) to the 25-ml. volumetric flask. The flask contents are well mixed and the absorbance is determined, as suggested by Icken and Blank (71, a t a wave length of 525 mp on a Beckman Model D U spectrophotometer, using 10-mm. silica cells and a slit width of 0.18 mm. The instrument is set for 100% transmittance with the bleached standard. X-STANDARDS WITH FRESHLY PREPARED L A K E
r'
30
w
m 4
.20 ,180 I601
,
.
i
.
I
2 4 6 1012 MICROGRAMS OF F L U O R I N E PER I O M L .
0
Figure 2. Absorbance at 525 mp on Beckman DU spectrophotometer, 1 cm. path, against micrograms of fluorine per 10 ml. of solution
-
I n making up the color reagent, calibrated pipets are unnecessary. The same pipets must be used, however, in adding the color reagent to the series of standards and to all unknowns, as a slight error in the amount of reagent added is reflected in a large reading error. The color reagent itself is best prepared before or during the final distillation of a series. After it is added to the fluoride solutions, the readings may be made either a t once or after a delay of several hours. There is a slow change on standing. I n the case shown in Figure 2, there was no change in 1 day. The change is inappreciable during the time required for a series of readings. The reagent may be made more concentrated (by omitting the added water or by using more concentrated buffer) or more dilute, as convenient. When prepared in the manner indicated, it is a clear, deep cherry red, and was still usable, despite some fading, after 1 month. RESULTS AND DISCUSSION
Some of the results obtained with sodium fluoride solutions are listed in VOL. 29, NO. 12, DECEMBER 1957
1889
Table I.
Fluoride Determination in Sodium Fluoride Solutions
(Values expressed in micrograms of fluorine)
Present 0.1
0.5
Fluorine
Found5
Fluorine Found with 7.3 Mg. of HCl
Fluorine Found with 7.3 Mg. of HCl 100 Mg.
+
of Ag,Oa
0 . 1 f 0.2 0.7 0.1 1.1 f0.1 5.0
*
1.0 5.0 15* 5.0 0 . 1 10.0 9.9 19.3 f 0.6 20.0 19.4 & 0 . 7 50.0 51.3 i 1 . 3 a Standard deviations given are averages of four or five values. * Only a single value given because results are obviously in error as a result of hydrochloric acid addition. Despite buffering the pH dropped from the usual value of 2.75 to 2.35.
Tables I and 11. Blanks were small, averaging about 0.2 y, and so variable, from 0.0 to 0.5 y, that they were ordinarily not subtracted from the values found. Only in the cases of the 0.5and 1.0-7 samples did the blanks appear to be relativdv significant, but these cases are a t the limit of applicability of the method. When distillatian was too rapid, blanks were high and titration of the distillate with 0.05.V potassium hydroxide showed them to be inore acid than usual This indicated tnat some sulfuric acid was coming eve*, presumably in the form of droplets. One drop (0.03 or 0.04 ml.) of the potassium hydroxide corresponded to 0.0001 ml. of 20N sulfuric acid, approximately the concentration of acid in the distillation chamber. Droplets of this size or less could not always be prevented from coming over, even by slow distillation. Attempts to filter them out of the vapor with glass wool and by other methods were ineffectual. The data in Table I1 confirm the findings of Armstrong ( 1 ) that phosphate does not affect the results. The 1.5 mg. of dibasic sodium phosphate indicated in Table I represents 0.31 mg. of
Table II.
*
phosphorus which is of the same order as the amount of phosphorus in a 5-mg. sample of rachitic bone. Recoveries from bone and dentin samples (Table 111) likewise indicated that phosphate did not interfere. Even 100 mg. of reagent grade tricalcium phosphate did not affect the results. When 200-mg. samples were used, however, slightly higher fluoride values were obtained, along with high blanks. These samples were much larger than those for which the apparatus was intended, and there was bumping and splattering. In addition, when such large samples were used, one tenth of the sulfuric acid was converted to phosphoric acid. In all probability, traces of phosphoric acid -were carried over with the sulfuric acid droplets. Experiments on the addition of neutral phosphate and sulfate to the fluoride standards showed that phosphate had a much greater bleaching effect, of the order of 100 times the sulfate effect. In contrast to the tricalcium phosphate of Armstrong (I), \yhich contained 0.63% fluorine, the analytical reagent grade phosphate used in these experiments contained approximately 0.0015% fluorine. This fluorine was
Fluoride Determination in Sodium Fluoride Solutions in Presence of Phosphate
(Values expressed in micrograms of fluorine) Phosphate Fluorine Present, Present Mg. Fluorine Founda 0.0 1.5, Na2HPOa 0 . 2 f 0.1 5.0 1.5, Ka2HPOa 5.3 & 0.4 10.0 1.5, NazHPOa 9.8 f 0 . 4 b 20.0 100, tricalcium 19.7 & 0 . 3 phosphate 0.0 200, tricalcium 1.O in 10 nil. phosphate 0.0 200, tricalcium 2.9 i 0 . 9 in 25 ml. phosphate 20.0 200, tricalcium 21.6 i 0 . 5 phosphate Average of four or five values in most cases. Average of 11 values. 1890
ANALYTICAL CHEMISTRY
Fluorine Found with 7.3 Mg. of HC1 100 Mg. of
+
agzo
2 1 . 2 =k 0 . 4
removed by a preliminary distillation in the apparatus itself, and successive distillates were tested for fluoride before recoveries were run on sodium fluoride solutions. Any chloride present in the sample evolved in the distillate as hydrochloric acid. The amounts of chloride from small bone samples did not affect the pH of the buffered solution or the final readings, and could, therefore, be disregarded. But when 0.1 ml. of 2N hydrochloric acid was added to the sample, the distillate had a pH of 2.35 with the buffered color reagent, instead of the 2.75 of control samples without added acid. The spectrophotometric reading corresponded to a fluoride value of 15 y instead of the proper 5 y. The hydrochloric acid was easily prevented from coming over, however, without affecting the fluoride by the addition of 100 mg. of silver oxide directly to the distillation chamber along m-ith the sample. The beneficial effect of the silver oxide was shown by lack of acidity of the distillates when titrated with dilute potassium hydroxide, by the proper pH values found when the buffered color reagent was added, and by the correctness of the values obtained for fluoride (Tables I and 11). The 0.1 ml. of 2-1hydrochloric acid used was equivalent to 7100 y of chlorine, a great excess compared to the amount of fluorine. When concentrations of potassium chloride of the order of 0.01N were added to the fluoride standards, the readings n-ere unaffected. Higher concentrations (0.1N and up), however, gave higher readings. If a great deal of chloride is present in the sample, it is obviously of advantage to remove it by adding silver oxide to the distillation chamber. The color reagent used here differs from that of Icken and Blank (7) in the means employed to obtain the required pH. Icken and Blank added to their reagent a small amount of hydrochloric acid. When their distillate was neutralized with dilute potassium hydroxide and the color reagent added, the final pH mas approximately 2.75. The present color reagent attained the same final pH, without neutralization. by means of the chloroacetate buffer. In spite of differences in appearance of the two reagents, absorption curves were similar. The absorprion curve of a standard containing no fluoride, in an instrument set for 100% transmittance with a bleached standard containing excess fluoride, showed a maximum a t 525 mp, as found by Icken and Blank, and this same wave length was adopted for measurements. In contrast to the color reagent of Icken and Blank, however, the present color reagent was clear and remained so indefinitely. It could be used several weeks after preparation, but whenever
~~~
used, new standards had to be made up. In addition, possibly because the particles of color lake were smaller than those of the Icken and Blank reagent, bleaching by fluoride took place more rapidly and readings could be made a t once, or after a delay of several hours, whichever was more convenient. If the color reagent and standards were prepared beforehand, it was possible to complete an entire fluoride determination in from 0.5 to 1 hour. An additional advantage of the buffered color reagent was that standards prepared from it gave a straight-line curve in the range from 1 to 10 y per ml. when absorbance was plotted against concentration (Figure 2). When the Icken and Blank reagent was used, the type of curve reported by these authors was usually found, although occasionally a straight-line segment would be obtained up to about 5 y of fluorine per 10 ml. The data of Table I11 are typical of the results obtained when the method was applied to bone and tooth samples. The values for cartilage do not indicate the presence of any fluoride. The values for the dentin samples indicate barely detectable amounts. LITERATURE CITED
(1) Armstrong, W. D., IND.ENG.CHEM., ANAL.ED.8, 384 (1936). (2) Armstrong, W. D., Univ. of Minnesota Medical School, Minneapolis 14, Minn., personal communication.
(3) Bien, S. M., J. Dental Research 22, 123 (1943). (4) Gilkey, W. K., Rohs, H. L., Hansen, H. V., IND.ENG.CHEM.,ANAL. ED.8,150 (1936). (5) Hoskins. W. M.. Ferris. C. A.. Zbid., 8 , 6 (1936).
~~
Table 111.
Analysis of Bone and Tooth Samples for Fluorine
Sample Cattle bone
Sample Wt., Mg. 20.40 11.60 25.00 25.00 50.00 50.00 27.68 15.23 25.00 25.00 50.00 50.00
Rat bone (diaphysis)a Rat bone ( ~ a r t i l a g e ) ~
2.89 2.86 2.81 2.02
Fluorine, y Added Found 4.3 2.3 5.7 5.2 10.5 10.8 3.0 8.6 4.0 7.4 5.0 10.4 5.0 10.7 10.0 21.8 10.0 20.8 7.5 9.0 0.3 0.3
% F in Sample 0.021 0.019 0.023 0.021 0.021 0.022
F Recovered, Y
3.1 4.4 4.9 5.2 11.1
10.1 0.26 0.31
