Determination of Fluoride in Water By the Aluminum-Hemotoxylin Method 3IARGARET J. PRICE AND OSMAN J. WALKER Department of Chemistry, University of Alberta, Edmonton, Alberta, Canada There is need for a rapid photometric method, accurate to 0.1 p.p.m., for the determination of fluoride in natural and fluoridated water. The factors that make use of the bleaching effect of fluoride on the aluminum-hematoxylin lake as a photoelectric colorimetric method for fluoride in water have been studied. This method allows the determination of 0 to 1.5 p.p.m. of fluoride ion with an accuracy of 0.05 p.p.m. if recrystallized hematoxylin, which has been oxidized w-ith hydrogen peroxide, is used at a pH of 4.6. Satisfactory color development occurs within 4 hours. With higher concentrations of fluoride smaller samples are taken and diluted to the proper volume with distilled water. Sulfates and, to a lesser extent, phosphates interfere in this method, but have less effect than on other colorimetric and photometric methods. This method can be carried out accurately and quickly on water samples, if sulfates and phosphates are determined and corrections applied when necessary.
T
HE highly colored lakes formed by reaction of aluminum and
zirconyl salts with hematoxylin have been known for some time. The zirconyl-hematoxylin lake is bleached by small amounts of fluoride, and this effect has been used in developing methods for determining small amounts of fluoride in water by such investigators as Gad and Saumann ( I ) , Jendrassik and Papp (4),and Jendrassik and Dippold (3). The corresponding aluminum-hematoxylin lake is the basis of a method for determining small amounts of aluminum ( 2 ) . Okuno (b, 6) has described a colorimetric method for small amounts of fluoride in water based on the bleaching of the latter lake by fluoride. I n it he uses 10 to 50 ml. of one of three solutions containing different amounts of aluminum salt and hematoxvlin, depending on the amount of fluoride present. He claims it is accurate for 0 to 0.40 mg. of fluoride. The present authors were unable to duplicate either the results of Jendrassik and coi~orkerswith the zirconyl lake or those of Okuno with the aluminum lake. Much of the difficulty in working with these lakes is due to the instability of the hematoxylin, which seems to undergo oxidation resulting in modification of the color and intensity of the metallic lake. Many modifications were made in the zirconium-hematoxylin method, but none gave satisfactory results. However, a method for determining small amounts of fluoride in water based on the bleaching of the aluminum-hematoxylin lake has been developed which differs widely from the one proposed by Okuno.
The preparation of hematoxylin solution presented considerable difficulty to previous workers, as it seemed to oxidize over a period of time and thus the composition of the lake was not a constant. Jendrassik and Dippold (3) after making u p the hematoxylin solution passed a stream of air through it for 20 minutes. In the experience of the present authors there Kas much uncertainty in this operation, so an oxidized form was obtained by using a definite quantity of hydrogen peroxide. Better results were also obtained by recrystallizing hematoxylin, Eastman Kodak Practical, three times from water to n hich a little sodium bisulfite had been added ( 7 ) . The method of making the hematoxylin solution also had to be altered to the folloning procedure. Recrystallized hematoxylin, 0.1 gram, is added to a solution of 55 ml. of 0.009 S hydrochloric acid, previously heated t o about 60” C. in a 250-ml. flask on a water bath; the solution is shaken and heating continued until dissolved. While still warm, 0.02 S sodium hydroxideis added, until the solution just turns red, then 3 ml. of 376 hydrogen peroxide is added, followed by 100 ml. of 1% acetic acid. The solution is allowed to stand 1 hour before using. A saturated solution of sodium bicarbonate and a 1 to 2 acetic acid solution are also needed. The reagent or “indicator’, is made by diluting 100 ml. of the standard aluminum salt solution with 850 ml. of distilled water, adding 25 ml. of saturated sodium bicarbonate solution with stirring, and adding 15 ml. of the oxidized hematoxylin solution with mixing. After standing 1 hour, 10 ml. of 1 to 2 acetic acid are added and the solution is shaken and allowed to stand for 48 hours before using. A fresh indicator solution should be prepared each week. The lake is purple in color.
ALUMINUM-HEBIATOXY LIN METHOD FOR FLUORIDE
The bleaching action of fluoride is due to the formation of the very stable aluminofluoride ion. Aluminum-hematoxylin lake purple
+ 6F- + hematoxylin + AIFe--yellow
EQUIPMENT
A Beckman B spectrophotometer with a 2-em. absorption cell was used for measuring the absorption spectra of hematoxylin and of the aluminum-hematoxvlin lake. Measurements of fluoride content were carried out in a Lumetron 402-E photoelectric colorimeter using a 15-em. light absorption cell. REAGENTS AND MATERIAL
Ten milliliters of a solution containing 2.21 grams of C.P. sodium fluoride per liter were diluted to 1 liter with distilled water to give a stock solution containing 0.01 mg. of fluoride per ml. The standard solution of aluminum salt was prepared from C.P. aluminum sulfate and contained 0.05 mg. of aluminum per ml.
EXPERIMEh-TAL
Absorption Spectra of Hematoxylin and Purple Complex. Figure 1 contains the absorption spectrum of hematoxylin, A, as well as that of the purple lake, B, each based on three sets of measurements. The absorption curves are obtained over the visible part of the spectrum to serve as a means for choosing the proper color filter to be used in the photoelectric colorimeter. Measurements with the photoelectric colorimeter are carried out with a color filter that transmits light in a narrow band as close as possible to the maximum absorbancy of the colored compound measured. Where tRo colored constituents are present, the absorbancy of the second compound should be small in this region. The maximum absorbancy of the aluminum-hematoxylin lake occurs a t about 550 mp, while the absorbancy due to hematoxylin is small a t this wave length. For this reason, best results in determining fluoride are obtained by using the 550 mp color filter in the photoelectric colorimeter.
1593
1594
ANALYTICAL CHEMISTRY
Calibration Curves. The calibi ation cuives for the colorinieter are based on the procedure developed for analyzing water, which makes use of 100 nil. of sample.
90,
I
Knoxns containing 0 to 0.14 mg. of fluoride (0 to 1.4 p.p.m.) are made up from the standard sodium fluoride solution and diluted to 100 ml. with distilled water; 10 ml. of the indicator are added to each of these. The p H value of the solution is now 4.6. The solution is allowed to stand for 4 hours, transferred to the 15-cm. absorption cell, and read in the photoelectric colorimeter using the 550 mp color filter. Distilled water is used as the blank (transniittancy = !OO%). The per cent transmittancy as determined is plotted against parts per million of fluoride ion.
A
HEMATOXYLIN
B
THE A L U M I N U M HEMATOXYL'N L A K E
I
70
1 I
01
0
02
0 4
06
08
10
l Z
14
FLUORIDE, P. P. M.
Figure 2. Calibration Curve for Determination of Fluoride by Photoelectric Colorimeter
Table 11. F - Added, P.P.31. 0.0
a
350
400
450
500
550
600
650
'
WAVE LENGTH, Mp
Figure 1. Absorption Spectra of Hematoxylin and Aluminum-Hematoxylin Lake
Table I. F - Added, P.P.M. 0.0 0.2 0.4 0.6 0.8 1.0
1.4 a
Effect of Time on Behavior of Indicator
% Transmittanoy 2 6 9 26 days days' days days 17.6 9.1 9.5 9.6 1 3 . 4 1 2 . 9 13.2 22.4 2 4 . 9 2 2 . 8 2 1 . 9 27.6 4 2 . 0 3 5 . 9 35.3 40.0 5 5 . 9 4 9 . 6 4 6 . 3 45.0 69.3 6 1 . 0 5 5 . 8 51.5 8 4 . 5 8 6 . 5 8 5 . 3 85.0
F - Found, P.P.M. 2 daya
0.0
0.2 0.4 0.7 0.9 1.1 1.4
6 9 2 days" days 0.0 0.0 0.2 0.2 0.4 0.4 0.6 0.6 0.8 0.8 1.0 0.9 1.4 1.4
6 days 0.3 0.4 0.5 0.7 0.7 0.9 1.4
Taken as standard.
The results shonn in Figure 2 are the averages of three sets of readings with an average deviation of 1.5%. Effect of Time. Time seems to have some effect on the behavior of an indicator. .4 series of determinations was carried out using portions of the same indicator that had been made up for 2, 6, 9, and 26 days. The results, each the average of four values, are shown in Table I. The per cent transmittancy obtained for different amounts of fluoride did not differ much for the first 9 days, but when the indicator had been kept for 26 days, the per cent transmittancy, a t similar fluoride content, differed widely from those obtained previously.
Change in Transmittancy of Solutions on Standing % Transmittancy 2 hours 4 hours"
0.2 0.4 0.6 0.8 1.0 1.4 Taken as standard.
F - Found. P.P.M. 2 hours 4 hoursa 0.0 0.0 0.2 0.2 0.4 0.4 0.6 0.6 0.8 0.8 1.0 1.0 1.4 1.4
After the reagents have been added to the sample the extent to which the lake is bleached by fluoride depends on the time that elapses before the measurement is made. I n Table I1 are presented data taken 2 and 4 hours after mixing; each set of figures is the average of three values. There is not a great deal of difference, and either time of standing may be used. As experience has indicated that better duplication can be obtained if the longer time is selected, the 4-hour period is recommended. Effect of pH. The pH must be closely controlled. At a p H above 6 (6.3, 7.3, and 9.1) the purple color, intense at first and changing to orange on standing, is not bleached by fluoride. Aluminum hydroxide was precipitated a t a pH of 5.0 to 6.0. Below a pH of 4.2 the color of the indicator is too faint and there is very little change in color on the addition of fluoride. Satisfactory color development and bleaching took place a t a pH of 4.6. Several buffers -sere investigated. Ammonium carbonate (5, 6) led to an unstable indicator giving results impossible to check; even a t a pH of 4.6 the indicator turned yellow on standing. Indicators containing sodium succinate failed to develop color. It was an easy matter to arrive a t a pII of 4.6 with sodium bicarbonate and the color development was satisfactory. I t was therefore wed in all determinations reported. Interfering Ions. Colorimetric and photometric methods for fluoride ion are affected by phosphates and sulfates that are found in many natural waters. In some waters the sulfate ion content may exceed 1000 p.p,m., while phosphate ion content seldom exceeds 5 p.p.m. I n this laboratory phosphate in water is determined by a photoelectric method, not yet published, involving the fo~mationof niolybdenum blue, while sulfate content
1595
V O L U M E 24, NO. 10, O C T O B E R 1 9 5 2 Effect of Sulfate on Determination of Fluoride
Table 111. F - present,
100
P.P.M. 0.0 0.2
Sulfate Added, P.P.M. 500 800 1000
2000
0.0 0.2
0.4 0.6
0.4 0.6 0.8 1.0
0.8 1.0
COVCLUSIONS
...
1.2 1.4
1.5
1.4
Table IV. ~
200
- p ~ P.P.M. 0.0 0.2
Effect of Phosphate on Determination of Fluoride ~0.2 ~ 0.1 0.3
0.4
0.4
0.6
0.6
0.8 1.0 1.2 1.4
0.8 1.0 1.2
1.3
ference is equivalent to a correction of 0.1 p.p.m. of fluoride Rith as little as 0.2 p.p.m. of phosphate, and rises to 0.4 p.p.m. for a phosphate content of 8 p.p.m. With higher fluoride concentrations, 0.4 to 1.4 p.p.m., the correction is negligible up to at least 8 p.p.m. of phosphate Thus phosphate interferes less than in other colorimetric and photometric methods for determining fluoride in water.
~
Phosphate Added, P.P.U. , 1.0 3.0 L O Fluoride Found, P.P.M. 0.2 0.2 0.4 0 1 0.3 0.3 0.4 0.5 0.4 0.4 0.4 o i 0.6 0.6 0 6 0 6 0.8 0.8 0.8 0.8 0.9 0.9 i.0 0.8 1.2 1.2 i i 1.2 1.4 1.3 1 4 1.3
0.5 ~ t
8.0 0.5 0.4 0.5 0.6 0.8 1 .0 1.1 1.3
i i determined b y titrating the xvater with a standard solution of barium chloride using the sodium salt of tetrahydrosj-quinone (THQ) as the indicator (8). It is a wise precaution to determine t11e.e ions in samples of water before measuring the fluoride ('ontent. Effect of Sulfates. Sulfates interfere x i t h most visugl and photometric methods for determining fluoride. Table I11 shows the results obt,airied for fluoride samples with sulfate up to 2000 p.p.m. when 0 to 1.4 p.p,ni. of fluoride was present. There is some interference a t a sulfate content of 200 p.p.ni. .4t 1000 p.p.m. the interference is such that a correction of only 0.2 p ~ p . m .of fluoride is necessar, This is somewhat less than found by Walker and Finlay ( 9 ) r the zirconyl-sodium alizarin sulfonate method. Effect of Phosphates. The interference by phosphate follows an irregular pattern, as can be seen from Table IF'. Each value is the average of three determinations, none of which differed from it by more than 0.1 p.p.ni. K i t h no fluoride present the iriter-
Fluoride can be determined in water satisfactorily by a photometric method involving the bleaching effect of fluoride ion on the aluminum-hematoxylin lake, provided that the hematoxylin i3 partly oxidized previousl~and the pH is closely regulated. Sulfates and phosphates interfere to a smaller extent than in other photometric and colorimetric methods. The method can be used satisfactorily in the examination of water supplies that naturally contain fluorides as well as of those that have been fluoridated. The limit of fluoride content that can be measured is 1.5 p.p ni on a 100-ml. sample, but a i t h greater concentrations smaller sample3 of water are taken nnd diluted to 100 ml. Kith distilled r a t e r befoie the ieagentq ale added. 4ChYOWLEDG?IEhT
Funds for the carrying out of this project JTere supplied by the Associate Committee on Dental Research of the Sational Research Council of Canada and are gratefully acknon ledged. LITER-%TURECITED
(1) Gad, G., and Saumann. K., Gas-'u. Wasserjack, 81, 193-5 (19391. (2) Houghton, G. L-., AizaZg,zt, 68,208-11 (1943). (3) Jendrassik, A,, and Dippold, d.,M a g y a r C'henz. FolyLiirut. 54, 19-26 (1948). (4) Jendrassik, A., and Papp, S., Ibid., 49, 137-46 (1943). (5) Okuno, H., J . Chem. SOC.J a p a n , 63,23-6 (1942). ( 6 ) Okuno, H., J . Faculty Sci.,Hokkaido I m p . Uniu., 3 (3j, 94-126 (1942). (7) Perkins,'K. H., and Yares, J., J . Chem. Soc., 81, 235-8 (1902). (8) Sheen, R. T., and liahler, H. L., IND. ESG.CHEM.,A s a ~ED., . 8, 127-30 (1936). (9) Walker, 0. J . , and Finlay, G., C a n . J . Research. B18, 151-9 (1940). RECEIVED for revieiv Fc:>rmry 4 ,195.2. Accepted Aug\iqt 13, i i C j ?
Spectrophotometric Method for Determination of the Fluoride Ion H. E. BUMSTED AND J. C. WELLS Division of Industrial Hygiene, Indiana State Board of Health. Indianapolis, Ind.
D
URIIiG the past ten years there has been a rapid increase in the industrial use of hydrofluoric acid and its salts The use of hydrofluoric acid in the petroleum industry and the fluoride fluves in soldering and welding has become rather common. The recent rapid advances in fluoro-organic chemistry and the fluoridation of public water supplies are expanding the industrial usage of fluorine and its compounds With this increase in usage, more industrial workers are exposed to fluoride compounds. I t is the duty of the industrial hygienist to determine whether these exposures constitute occupational health hazards. Consequently, the number of fluoride determinations in atniosphei ic and hiological samples has greatly increased. DEVELOPMENT OF REAGENT
I n a study of the present methods used for the determination of emall amounts of fluoride ion, it became apparent that some
instrumental method of analysis would be desirable to eliminate as far as possible, the variations of different analysts reading either Nessler tubes or the end points of titrations. After some preliminary investigation, i t was decided that the spectrophotometer offered the best approach t o this problem. The colorimetric reagents investigated &-ere iron thiocyanate ( d ) , titanium-hydrogen peroxide ( 5 , 6, 9), and zirconium-alizarin sulfonate ( 1 , 3, 7 , 8, IO). I t was felt that the zirconium-alizarin sulfonate method offei ed the best possibilities for the colorimetric determination of small amounts of fluoride ion. During thp investigation of the various published zirconium-alizarin sulfonate reagents, i t was observed that as the ratio of alizarin sulfonate to zirconium iTas increased, the final color developed in the solution was intensified. I n selecting the composition of the reagent the following factor3 were considered: