960
ANALYTICAL CHEMISTRY
Table 111. Effect of Counting Conditions and Distance on Relative Back- Scatter Contribution
cussed along with techniques for counting. The method will accommodate from 20 y t o 120 mg. of calcium in the sample. ACKNOWLEDGMENT
-4
B C D E
1.00
1.00b 1.07b
1.18
1.145
1.00
1 .oo
The authors are indebted t o Joan Gurian of the Biostatistics Group for assisting in the mathematical analysis of the data.
1.10
1.13 1.20b a Counting conditions illustrated in Figure 5. b Averages from measurements of 8 samples. Deviation from mean ratio did not exceed &2.0%. Other figures are averages from 4 determinations.
LITERATURE CITED
(1) Anthony, D. S., personal communication, 1948. (2) Aten, A. H. W., Jr., Nucleonics, 6, 68-74 (1950). (3) Bernstein, W., and Ballentine, R., Rev. Sci. Instr., 20, 347-9 (1949).
Comar, C. L., Hansard, S. L., Hood, S. L., Plumlee, M. P., and Barrentine, B. F., Nucleonics, 8, 19-31 (1951). (5) Ellis, G. H., IND.ENQ.CHEM.,ANAL.ED.,10, 112 (1938). (6) Furman, N. H., editor, “Scott’s Standard Methods of Chemical Analysis,” 5th ed., Vol. I, pp. 205-17, New York, D. Van Nostrand Co., 1939. (7) Glendenin, L. E., and Solomon, A. K., Science, 112, 623-6 (4)
per sq. em., with an absorption coefficient of 0.107, which is almost identical to that obtained by Glendenin and Solomon and by Comar. As found by Comar and by Nervik and Stevenson, neither the distance of the sample from the Geiger-Muller tube nor the thickness of the Geiger-Muller tube RindopT influenced the shape of the self-absorption curve significantly. SUMMARY
In the determination of calcium-40 and radioactive calcium-45 in the same sample, the element, precipitated a s the oxalate, is mounted for beta-particle counting as a part of the over-all process of calcium determination. Total calcium is measured by titration with cerate and the result is used to evaluate the self-absorption correction to be applied to the observed beta activity. The self-absorption and self-scattering curves are presented and dis-
(1950). (8) Graf, W. L., Comar, C. L., and Whitney, I. B., Nucleonics, 9, 22-7 (1951). (9) Ma&lin,P.T’Feldman, L., Lidofsky, L., and Wu, C. S., Phyn. Rev., 77, 137 (1950). (10) Nervik, W. E.. and Stevenson, P. C., Nucleonics, 10, 18-22 (1952). (11) Norris, W. P., and MacLeish, A., Biological and Medical Divi-
sions. Arnonne National Leboratorv. Quarterlv ReDort AHL4333,’ l l G 1 6 (May, June, July 1949). (12) Sacks, J., ANAL.CHEM.,21,876-7 (1949). (13) Smith, G . F., and Fly, W. H., Ibid., 21,1233-7 (1949). I
-
RECEIVED for review November 20, 1952. Accepted February 20, 1953
Fluorometric Determination of Small Amounts of Fluoride W. ALLAN POWELL’ AND J. H. SAYLOR D u k e University, D u r h a m , N . C . This work was initiated in an effort to obtain a better method for the determination of microgram quantities of fluoride. The methods finally developed depend upon the fact that the intensity of fluorescenceof the compounds formed on reaction of aluminum chloride with the dihydroxyazo dyes, Eriochrome Red B and Superchrome Garnet Y, is decreased on addition of fluoride. The methods were tested by analyzing solutions of pure sodium fluoride (0.2 to 100 micrograms of fluoride) both before and after a Willard-Winter distillation. The precision obtained is good and ions which are often present after distillation, such as sulfate and phosphate, do not interfere seriously.
A
FEW quantitative fluorometric methods for fluoride determination have been described. A method developed by
OkaE (7) for the determination of large amounts of fluoride depends upon the reaction of fluoride ion with aluminum ion, the fluorescence of the aluminum-morin complex serving for endpoint determination. Bournstyn ( I )has suggested that a method based on the fluorescence of aluminum-Pontachrome Blue Black R should be feasible. Willard and Horton (19) have published the results of a study of fluorescence indicators for the thorium nitrate titration and have developed a method (14) based upon the use of quercetin. About the time that the experimental portion of the present work was completed, they (16)published methods for the fluorometric determination of traces of fluoride which use the aluminum complex of either morin or 8-quinolinol. I n the present paper, methods are described for the fluorometric determination of traces of fluoride with a precision equal to or better than previous methods and which are less subject to anion interference. 1 Present sddresa, Department of Chemistry, University of Richmond, Richmond, Va.
APPARATUS AND REAGENTS
A Lumetron fluorometer was employed for all the work. A
Corning 5860 primary filter was used for the isolation of the desired excitation band and a Corning 3389 secondary filter for the fluorescence band. Twenty-five-milliliter rectangular cells were used in the fluorometer. A Beckman pH meter was employed for all pH measurements. Reagent grade chemicals were used for all the solutions, with the exception of the organic compounds, in which case commercial grade reagents were used a s received or were purified before use. Sodium fluoride solutions were made with pure sodium fluoride prepared according t o the procedure of Parker and Vosburgh (8), which is a slight modification of that of Reynolds and Hill (IO). PRELIMINARY SURVEY
Most of the methods previously described for the fluorometric detection or determination of aluminum were studied: those using morin ( l a ) , quercetin (Z), 8-quinolinol (a), Eriochrome Blue Black B (9),which is an isomer of Pontachrome Blue Black R ( I I ) , and Eriochrome Red B (9). In addition, Superchrome Garnet Y , which had not been used previously for either aluminum or fluoride, was studied.
961
V O L U M E 25, NO. 6 J U N E 1 9 5 3
8 K W
g
t
tf
60'
'
'
I
,
4.4
4.0
t
I
,
/
.
5.2
4.8
I
,
5.6
6.0
PH
6
l
4.0
,
l
4.4
,
l
4.8
,
l
5.2
1
5.6
,
ter the initial reading but did not change appreciably at the pH optimum for intensity, which indicated that the rate of reaction decreased as the pH decreased. Almost the same reading R ould be obtained over a range of several tenths of a pH unit, if sufficient time were allowed for equilibrium to be attained at the low pH values. The pH effect in the low region of pH results from the fact that the dyes are weak acids. A t pH values above that required to give maximum fluorescence, the low readings are probably caused by the increase in the concentration of various aluminum-hydroxide complexes. Figure 2 shows that the sensitivity to fluoride is greater a t low pH values. I n spite of this fact, the p H to give maximum intensity of fluorescence was selected for the final methods, as the sensitivity is sufficient a t this pH and the use of a smaller pH would increase the time of development of the complex and probably the interference from anions. These curves also indicate the error to be expected as a result of variation in pH. For 25 micrograms of fluoride, under the conditions employed, an error of 1 microgram would result from a variation in pH of 0.03 for Superchrome Garnet Y and 0.08 for Eriochrome Red B. The intensity of fluorescence decreases slightly with increase in the buffer concentration; however, the sensitivity is not affected appreciably if the solution used to set the instrument and the fluoride solution have the same buffer concentration.
I
6.0
PH
Figure 1.
Effect of pH on Intensity of Fluorescence of Aluminum-Dye Systems Upper.
Aluminum-Superchrome Garnet Y
PH
0. Readings taken after 2 hours
0 . Readings taken after 20 hours
Lower. Aluminum-Eriochrome Red B 0. Readings taken after 5 hours 0 . Readings taken after 18 hours
4.83 4.67
It should be possible to use all of the above reagents for the determination of fluoride; however, Eriochrome Red B (sodium salt of 4-p-sulfo-2-naphthol-~-azo-1-phenyl-3-methyl-5-hydroxypyrazol) and Superchrome Garnet Y (sodium salt of 5-sulfo-2-hydroxybenzeneaxoresorcinol) were selected for detailed study for several reasons, among which are the relatively small anion interference and small pH effect on the intensity of fluorescence. STUDY OF ALUIMI~WM-ERIOCHROME RED B AND ALUMIIYUMSUPERCHROME GARNET Y SYSTEMS
Sources of Dyes and Concentrations Used. Superchrome Garnet Y (Colour Index 168), furnished by E. I. du Pont de Kemours & Co. and the National Aniline Co., was used as a 0.1%solution in water and Eriochrome Red B (Colour Index 652), furnished by Geigy and Go. and Du Pont, was used as either a 0.1 or 0.05% solution in 95% ethyl alcohol. For some studies solutions were prepared from dye purified as follows: A saturated aqueous solution of the dye was extracted with ether, and enough sodium chloride was added to the aqueous solution to precipitate a small proportion of the dissolved dye. The dye was then filtered and dried. Method of Measurement. The solution of a series which had the maximum intensity was employed to set the instrument, using the smallest reduction plate permissible. A blank solution consisting of all reagents except aluminum chloride and fluoride ion was used t o set the instrument on zero. Unless otherwise indicated, studies of the effects of variables were made with an aluminum concentration of 50 micrograms in 50 ml. of final solution. Effect of pH and Buffer Concentration. 4 n acetic acidsodium acetate buffer was used for all the work. The curves shown in Figure 1, which are plots of intensity of fluorescence versus pH, show a maximum a t pH 4.8. The intensities of solutions with low values of pH increased considerably af-
4.50
\\ A
I
20
t
\ \
I4
\I ,4.43
L
d
4.82
\
I
2I0
I
40 I
I
60 I
I
80 I ,%4.08, 100
120
MICROGRAMS OF FWORIDE
Figure 2. Effect of pH on Calibration Curves 0. Aluminum-Superchrome Garnet Y 0 . Aluminum-Eriochrome Red B
Effect of Concentration and Purity of Dye. In Figure 3 the intensity of fluorescence a t pH 4.8 is plotted against moles of dye per mole of aluminum. These data were obtained by adding varying quantities of purified dye to a fixed quantity of aluminum. The curves indicate that the molar ratio of aluminum to dye for the fluorescing species is 1 to 1 for both aluminum-Eriochrome Red B and aluminum-Superchrome Garnet Y. Addition of dye in excess of this ratio causes a decrease in the intensity, probably as a result of absorption of ultraviolet light and emitted light by the dye. The amount of unpurified dye required to give the maximum intensity for a given amount of aluminum was found to depend to a great extent upon the source of the dye. However, if quantities
ANALYTICAL CHEMISTRY
962
of impure dye equivalent to that of pure dye were used, the same sensitivity for fluoride was obtained with dyes from different sources. A volume of solution slightly larger than required to give the maximum intensity was selected for use in the methods developed. Solutions of both dyes were found to give somewhat lower intensity readings after standing for 6 months. However, the effect of aging on the sensitivity for fluoride during this period was observed to be small for Superchrome Garnet Y , if the same dye solution mas used for both the fluoride solution and that used to set the instrument. The change was somewhat larger with Eriochrome Red B, particularly for very small amounts of fluoride.
'OT
Eriochrome Red B under similar conditions, a decrease of 7.0 divisions was observed in 1 hour, of which 4.5 \\as caused by the increase in temperature. However, most of this decrease is a temporary one, and the effect is not serious if readings are made in less than 1 minute. Standard Solutions of Aluminum-Dye. Because of the relatively long time required for reaction of aluminurn a ith the dye a t room temperature and because, as shown in Figure 1, the pH effect on intensity is much smaller after long periods of time, it seemed advantageous to prepare a solution containing all reagents and to allow this solution to reach equilibrium before addition t o the solution containing fluoride. The small rate of change in intensity of fluorescence after reaching a maximum indicated that this should be possible. The following procedure was adopted for the preparation of standard aluminum-dye solutions.
To prepare 1 liter of Eriochrome Red B standard solution, mix 60 ml. of 2 AI acetic acid, 40 ml. of 2 .)I sodium acetate, 40 ml. of aluminum chloride solution (50 micrograms of aluminum per ml.), and 440 ml. of 95% ethyl alcohol in a 1-liter volumetric flask. .4dd GO ml. of 0.1 70 solution of dye in 9570 ethyl alcohol (prepared from Geigy dye as received, and filtered), and make up to 1 liter with distilled water. For Superchrome Garnet Y mix 50 ml. each of 2 JI acetic acid and 2 21 sodium acetate. 40 ml. of aluminum chloride solution (50 micrograms of aluminum per ml.), 48 ml. of 0.1% aqueous solution of Sational Aniline dye or 33.2 ml. of 0.1% aqueous solution of Du Pont dye, and mater to make 1 liter.
"/
w
Quantitative experiments showed that the standard solutionq should be alloaed to stand 1 day a t room temperature before u3ing. The time required for the reaction of fluoride with the standard aluminum-dye solutions at room temperature was determined for various amounts of fluoride and different volumes of standard solution. The rate of reaction is greater with Superchrome Garnet k'than with Eriochrome Red B. At 31-32" C., using sufficient standard aluminum-Superchrome Garnet Y solution to
LL
tl I
0.4 Figure 3.
,
I
,
I
,
I
,
I
,
I
0.8 1.2 1.6 2.0 2.4 MOLES DYE 1 MOLE ALUMINUM
,
I
28
,
I
3.2
' IOOS
Variation of Intensity of Fluorescence with Molar Ratio of Dye to Aluminum 0. Superchrome Garnet Y
.
Eriochrome Red B
Effect of Other Variables. The intensity of fluorescence increases linearly with increase in concentration of ethyl alcohol and decrease in temperature. Neither factor affects the sensitivity to fluoride appreciably if the solution containing fluoride and that used to set the instrument contain the same alcohol concentration and are a t the same temperature. The fluorescence with both dyes is sufficiently intense without alcohol; however, alcohol is needed with Eriochrome Red B to prevent precipitation of the dye a t the pH used. An increase in temperature increases the rate of reaction of the dyes with aluminum. At room temperature, using the conditions previously selected, the time required for attainment of maximum intensity was 90 minutes for Superchrome Garnet Y and 4 hours for Eriochrome Red B. A similar solution containing Superchrome Garnet Y, heated a t 60°C. for 5 minutes and cooled, had reached the maximum intensity by the time the solution could be read. A solution containing Eriochrome Red B, and heated to 82" C., developed to maximum intensity in the time required to heat and cool. Exposure to ultraviolet light causes a decrease in the intensity of fluorescence when either of the dyes is used. Using Superchrome Garnet Y with a concentration of 50 micrograms of aluminum per 50 ml., a decrease of 4.9 divisions was observed in 1 hour; 2.0 divisions of the decrease were due to an increase in temperature as a result of heating in the instrument. With
302010-
%
"
10 ' 1 20 " ' 310 ' 40 1 " 50 ' 1 60 ' MICROGRAMS OF FLUORIDE
Figure 4.
70
Calibration Curves
Superchrome Garnet Y with 50 y of AI per 50 m l . A. Superchrome Garnet Y with 25 y of A1 per 50 ml. 0 . E+oohrome Red B with 50 y of A1 per 50 m l . A. Enochrome Red B with 25 y of AI per 50 ml.
0.
V O L U M E 25, NO, 6, J U N E 1 9 5 3
963 Table I.
F - Added, y
Precision of Methods Using Pure Sodium Fluoride Solutions Eriochrome Red B, Superchrome Garnet Y, 5 y of A I + + + 5 y of A l + + + Sens., y/div. F - found, y Sens., y/div. F - found, Y 5 y of A1 in 50 Ml. of Solution 0.06 0.14 .. .... 0.19 0 24 + ++
0.20
401
i.06
Mean dev. 1 0 . 0 8 0.10 1.96 1.95 2.05 2.07 Mean dev. 1 0 . 0 5 0.12 2.79 3.16 3.05 3.03 Mean dev. 10.11 0.22 4.79 5.11 5.20 4.98 hlean dev. 1 0 . 1 4
2.00
MICROGRAMS OF FLUORIDE
Figure 5. Calibration Curves 0. Superchrome Garnet Y with 5 y of A1 per 50 ml. 0 . Eriochrome Red B with 5 y of AI per 50 ml.
give 25 micrograms of aluminum in 50 ml. of final solution, the maximum time required for attainment of minimum reading for the fluoride concentrations studied was 1 hour. K i t h Eriochrome Red B under similar conditions but a t 26" C., the maximum time required was 3 hours. The time required for Eriochrome Red B can be reduced to approximately 1 hour by heating to 50' C., for 10 minutes. Calibration Curves. The curves shown in Figures 4 and 5 for several ranges of fluoride concentration were obtained by using different volumes of standard solution. A greater decrease in reading for a given amount of fluoride occurs with a low aluminum concentration. For a given amount of aluminum the sensitivity is greater for low fluoride concentrations. No appreciable change in the standard curves for Superchrome Garnet Y was found when the same standard aluminum-dye solution was used for 2 weeks. Although a small change was found with Eriochrome Red B during this period, no appreciable error would be introduced by this aging effect if the same solution were used for the standard curve and for the sample. PROCEDURE FOR FLUORIDE DETERMINATION
3.00
5.00
0.27
0.29
0.45
i.07 1 0 04 1 87 2 10 2.00 2 07 3~0.08 3.15 3.08 2.90 2.93 fO.10 5.09 4.94 4.94 5.00 10.04
Eriochrome Red B Superchrome Garnet Y 25 y of A l + + + 50 y of .41+++ 25 y of A1+++ 50 y of A l + + '
PPPnSens., found, Sens. found, Sens., found, Sens., found, y y yldiv. y/di