548
A N A L Y T I CA L CHfM I STRY ACKNOWLEDGMENT
The authors are grateful to A. K. Klein, Food and Drug Administration, for checking the method and offering valuable suggestions. LITERATURE CITED A\epli, 0 . T., Munter, P. A, and Gall, J. 610 (1948).
F., ANAL.CHEM.,2 0 ,
Baernstein, H. D., IND.ENG.CHru., ANAL.ED.,15, 251 (1943). Bost, R. W., and Nicholson, F., Ibid., 7 , 190 (1935). Daasch, L. W., Ax.4~.CHEM.,19, 779 (1947). Daasch, L. W., and Smith. D. C., U. 9. Naval Research Inst., Rept. P-3033 (Feb. 6 , 1947). Davidow, B., and Woodard, G., J . Assoc. Ofic.Agr. Chemists, 3 2 , 7 5 1 (1949).
Drayt. G., ~ ~ K A LCHEM., . 20, 737 (1948). Fairing, J. D., and Phillips, TV. F., Division of Agricultural . SOC., Boston, and Food Chemistry, 119th Meeting - 4 ~CHEX. Afass., 1951.
Furman, D. P., and Hoskins, W.AI., J . Econ. Entomol., 41, 106 (194s).
Goldenson. J., and Sass, S., .%SAL. CHEY..1 9 , 3 2 0 (1947). Horsley, H. L., Ibid., 19, 508 (1947). Janovsky, J. V., Ber., 2 4 971 (1891). Lfathews, F. E., J . Cheni. iSoc., 61, 103 (1892).
(14) JIeunier, J., Compt. rend., 114, 75 (1892). (15) Xakarima, Jl.,Inagaki, K., and Tati, T., Botyu-Ragaku, 16, 107 (1951); Englishresume, p. 110. (16) Nakazima, hf., and Oiwa, T., Ibid., 15, 114 (1950); English resume. p. 116. (17) Pearce, S. J., Schrenk, H. H., and Tant, TV. P., S.Bur. Mines, Rept. Inreat. 3302 (1936). (18) Ramsey, L. L., and Patterson, IT. I . , J . Assoc. Ofic. d g r . Chemists, 29, 337 (1946). (19) Riemschneider, R., and Ottman. G., 2. .\-utialfuo,.sch., 5b, 307 (1950). (20) Schrenk, H. H., Pearce, S.J., and Tant, I-. P., L-. S. Bur. Mines, Repts. Inwst., 3287 (1938, revised 1937). (21) Schrenk, H. H., Yant, JT. P., and Pearce, S. J., Ibid., 3293 (1935). ( 2 2 ) Trenner', K. R., Walker, R . IT-., and Buhs, R. P., . 4 ~ . 4 CHEM., ~. 21. 285 11849). (23) Yant, W.P., Pearce, S. J., and Schrenk, H. H.. C. S.Bur. Mines, Repts. Inccst., 3323 (1936). (24) l a n t , W. P., Schrenk, H. H., and JIautz, P. H., Ibid., 3282 (1935).
r.
I
~~~
j
~I
RECEIVED for review July 6 , 1931. Accepted October 30, 1951. Presented before the Division of Agricultural and Food Chemistry and Analytical Chemistry, Symposium on Methods of Analysis for Micro Quantities of Pesticides, a t the 119th Meeting of the A X E R I C ACIIEMICAL ~ SOCIETY, Boston, Mass.
Spectrophotometric Determination of Inorganic Fluoride A. D. HORTON, P. F. THORIASON, AND F. J. MILLER Analytical Chemistry Dirision, Oak Ridge National Laboratory, Oak Ridge, T e n n .
'The discovery that fluoride ion diminishes the color produced when thoron reagent, 1-(0-arsonophenylazo)-2-naphthol-3,6-disulfonic acid, is added to thorium nitrate solution. created further interest in thoron as a possible colorimetric reagent for the determination of microgram quantities of fluoride. Fluoride ion has been estimated spectrophotometrically in the range of 0 to 50 micrograms with an overall accuracy of zt4Yo and a mean standard deviation
R" 1
GENTLY, Willard and Horton (7') made a thorough study of indicators for the titration of fluorides with thorium solu-
Lions. They developed an excellent photofluorometric titration method wing quercetin as a fluorescent indicator and thorium nitrate as a titrant. This method is useful in the range l t o 40 mg. of fluoride A colorimetric method By Nonnier et al. (3), based on the diminution of the color obtained with ferric ion and 5-sulfosalicylic acid, has been used t o determine 0.2 to 1.0 mg. of fluoride. More recently, Thrun (6) developed a colorimetric procedure for determining fluoride in waters, that is dependent upon the change of color when fluoride ion is added to a n aluminum lake of eriochromecyanine. This procedure covers a range of 0 1 t o 6 p.p.ni. I n the course of developing the colorimetric procedure for thorium, Thomason et al. ( 5 ) ,found that the presence of fluoride ion caused a diminution of the red color of thorium nitrate solutions containing thoron reagent, l-(o-arsonophenylazo)-2-naphthol3,Gdisulfonic acid, which was synthesized by this laboratory according t o Kuznetsov (1). This fact resulted in an attempt to develop a spectrophotometric procedure for fluoride. The feasibility of the method was demonstrated by Thomason and Miller (4),who added varying quantities of a standard sodium fluoride solution t o known volumes of standard thorium nitrate aolution, acidified with hydrochloric acid, and added thoron reagent, Comparison in the Beckman spectrophotometer with a reference containing only thoron resgent and hydrochloric acid
of 0.83 microgram. Separation of fluoride from interfering substances has been improved by an automatically controlled steam-distillation apparatus which allows distillation to take place unattended. The total time required for the analysis is about 40 minutes, which includes 15 minutes for development of color. The procedure providea a fairly simple means for determining micro quantities of fluoride, and does not require a skilled analyst.
shoxed that the diminution of color was directly proportional to the fluoride ion present. The standard calibration curve (Figure 1) does not fol~owBeer's law, but its shape indicates that it is us-
I
0
I
10
I
20 F- ION
(a)
I
I
30
40
Figure 1. Standard Calibration Curve
!
V O L U M E 2 4 , NO. 3, M A R C H 1 9 5 2
549
The heater for the steam generator is the Cenco Hotcone, ll5-volt, 350-watt electric heater. The JVheelco Model 224 T C B automatic temperature controller is w e d to regulate the flask temperature. DISCUSSIOS AND RESULTS
The original Willard-Winta apparatus (8) contains a distillation flask fitted with a two-hole rubber stopper. A thermometer passes through one hole and a glaw capillary tube through the other. The temperature is controlled within the desired distillation range by admitting water through the capillary. The flask is heated by a gas burner, and the upper portion of the flask is protected by an asbestos mat placed about the lower one third of the flask.
CYLIN3?1CI-
CENCO HOTCONE
I KG Figure 2.
Better temperature control is obtained by using a therinostaticallv controlled electric heater with an iron-constantan thermocouple immersed in mineral oil in the thermocouple well of the flask. GENERATOR Steam is used in place of water to minimize teiriperature fluctuation. A rubber tube u-ith an acljustable pinchclanip is at,tached to the stopper in the s t a i n generator to control steam f o x (Figure 2). The original development of the procedure was carried out, using standard sodium fluoride solution without distillation. The total volume for producing the optimum curve was 10 ml. However, when the tlistillat~ionprocedure was required t o remove interfering ions, it \,-as necessary t o collect a total volume of 12 1n1.in order to recover all the fluoride. The t'ot'al volume of additional reagents required was nearly 2 d . ; therefore the total volume was adjusted to 15 nil. The disti]lat,ion rate affects considerably. I , re-~ ~ sults are obtained when the rate of distillation-Le., the rate of steam flow a t the specified temperature-is too high, because the amount of condensate produced is far greater t,hitn that required to carry the fluosilicic acid irlto the receiver. consequently, t,he receiver fills with steani condensntg before the dwired pi,oduct has completely dietilled over. The collection of 12 ml. of distillate in approsiinately 12 niinutes W:LS found to be the Inost sat rate, The optiniuin distillation temperature range was found to be 125" to 135' c. Beloiv 125' C., recovery is incomplete; above 135' C., t,here is danger of impurities distilling over with t,he fluorides, Ai]unlinunlor phosphorus an)- ion producing a color in acid solution, such as uranium, would interfere without distillation. .Uuniinuni complexes fluoride, causing the color to appear darker than norinai, and indicating less fluoride than is actually present. ~ ~ phosphorus ~ complexes~ thorium, ~ thereby ~ producing a color lighter than nomal, and indicating more fluoride than is actually present. Several varieties of pure silica arid silica-containing materials were t,ried to find t,he best one for increasing the reaction surface for optimum production of fluosilicic acid. It is formed aecording to the following react'ion:
1
STEAM
Diagram of Apparatus
able in the range 0 t o 50 micrograms of fluoride. The tendency of thg curve to flatten out above 50 micrograms precludes its use above this amount. PROCEDL'RE
The sarllple containing between 0 t o 50 inicrograllls of fluoride is placed in the distillation Aaik (Figure 2) containing a few quartz than 2 ml., it should chips. (If the volume of the samp\e is be evaporated until it is 2 ml. 01' less.) The condenser is sealed into the flask with 70% perchloric acid as a lubricant. The condenser tip is placed in a 15-ml. roluinetric centrifuge cone containing a sodium hydroxide pellet in approximately 1 m]. of distilled ,vatera B~ means of a pipet caiibratecl in divisions of o,ol ml., 0.75 m]. of 70% perchloric acid is added through the side arm of the distilling flask. The flask is heated to 125' C. by a thermostatically controlled electric heater (Figure 2). Steam from the electrically heated generator is introduced when the flask teniperature reaches 1250 c., and distillation is carried out betneen 1 2 j 0 and 135" C. until 12 Inl. of distillate have been collected. The di.stillate is acidified n i t h 0.35 rnl. of concentrated hydrochloric acid, which gives a solution of approximately pH 2. A few crystals of potassium iodide are added, and the distillate is alloLved to stand 3 minutes, ~h~~ o,l ml. of 85% hydrazine hydrate, 1.0 ml. of standard thoriuln nitrate solution, and 0.375 nil. of 0.2% solution of thoron reagent are added. The volume is made up to 15 ml. with distilled m t e r , and the color is allowed t o develop for 15 minutes. (The color is reasonably stable for 3 ~h~ is co,,lparecl ill the ~~~k~~~ ?\lode, DG quartz spectrophotometer at s45 nip n.ith a reference containing only 0.375 ml. of 0.2% thoron reagent and 0.35 ml. of concentrated hydrochloric acid diluted to a volume of 15 ml. with distilled water. REAGENTS ASD APPARATUS
Reagent grade or C.P. chemicals were used in all experiments. IIydrazine hydrate, 85% (Eastman Kodak or A. D. Rlackay) Hydrochloric acid, C.P. grade, specific gravity 1.188 to 1.192 Perchloric acid, 70% Potassium iodide Sodium hydroxide pellet5 Sodium fluoride Thorium nitrate solution (120 micrograms of T h + + + +per ml.) (Lindsay Light and Chemical C o . ) 'I'horon reagent, 0.2%, l-(o-arsenophenvlazo)-2 naphthol-3,6&sulfonic acid ( 1 ) Quartz chips for increasing reaction surface area were obtained from broken pieces of sand-finish, pure quartz tubing, The Beckman hlodel DU quartz spectrophotometer with cells of 1-cm. path length \vas used for all optical density readings. A A spectral band width of 2.0 ni,u was used to obtain the standard calibration curve. The distillation apparatus is shown in Figure 2. 811 glass apparatus except the steam generator and the centrifuge cone was constructed by the laboratory glass blower. The flask heater is a n aluminum-jacketed electric heater with a 40-mm. inside diameter and a 96-mm. outside diameter, It is 66 mm. high. It contains 20 feet of KO.26 Nichronie wire and is rated a t 250 watts and IlOvolts.
6HF
+ Si02 -+-H&Fs + 2H2@
Besides providing a large reaction surface, the silica is needed in a form that s i l l prevent bumping of the solution in the distillation flask. Builders' sand, glass chips, glass capillary tubing, broken glass frits, glass wool, porcelain chips, and silica gel were found to be unsatisfactoryfrom the standpoint of either bumping or low red t s . ~ 0 -results i ~ are generally attributed to the lack of rough, area-producing surfaces on the chips, and/or impuritiFs present in the substances. S ~ sand R is satisfactory for complete recovery of the sample, as well as Preventing bumping. However, it is objectionable bec a u ~ eit sticks to ground-glass surfaces, making tight seals impos-
~
~
ANALYTICAL CHEMISTRY
550
sible without extra cleaning. Quartz chips possessed all the desirable qualities of sea sand and none of the objectionable characteristics.
Table I. Determination of Fluoride by Distillation of Standard Sodium Fluoride Solutions Deviations
EFFECT OF PERCHLORIC ACID
I n nearly all cams, when standard sodium fluoride solutions were distilled, the results were higher than those obtained from duplicate samples assayed without distillation. This was a t first attributed t o the formation of chlorine dioxide from the decomposition of perchloric acid during distillation. Hom ever, Mellor ( 2 ) showed that chlorine dioxide reacts with water to form chloous and chloric acids:
2C102
Sample NO.
1
+ 3HC1 +2C12 + 2H20 HC103 + 5HC1 +3C12 + 3H20 HCIO,
Potassium bromide was first used t o remove free chlorine. Excess bromine thus formed was removed by 0.1 ml. of 1% phenol solution. However, as there is a tendency for bromine and phenol to form a precipitate of bromophenol, this combination was abandoned in favor of potassium iodide for free chlorine removal and 0.1 ml. of 85% hydrazine hydrate to reduce excess iodine. Even though all free chlorine was removed, the results continued t o be higher than those from undistilled duplicates, and it became necessary t o plot a standard calibration curve from the values obtained from a series of distillations of standard sodium fluoride solution. The higher results obtained from the distilled samples were finally attributed t o the action of undecomposed perchloric acid on thoron reagent.
It is estimated that 0.5 ml. of perchloric acid distills over during each determination. This .amount was added to an undistilled standard containing 10 micrograms of fluoride and a few crystals of potassium iodide were added. The solution Tras allowed to stand 3 minutes and 0.1 ml. of 85% hydrazine hydrate, 1.0 ml. of standard thorium nitrate solution, and 0.373 ml. of 0.2% thoron reagent were added. The optical density was compared with the value obtained] for the same quantity, from the standard calibration curve for distilled sodium fluoride solutions. These values agreed within the stated accuracy of this procedure. Tests were made t o determine the minimum practical amount of perchloric acid, which would produce complete recovery of fluoride and also minimize its effect on thoron reagent. This amount was determined t o be 0.75 ml. The results obtained from samples containing aluminum, phosphorus, and uranium are shown in Table I.
15.0
Aluminum’ Phosphorus
45.0
3
Sone
30 0
6
As the distillate is collected in sodium hydroxide, the resulting products are a mixture of alkali hypochlorite, chlorite, chlorate, and chloride. JTThen the solution is acidified] chlorous and chloric acids are formed and react with excess hydrochloric acid t o form chlorine and water.
Y
2
4 5
+ H,O +HCIO, + HClOi
Interfeqing Ions Aluminum
Fluoride Known, Found,
Phosphorus Vranium Uranium
50.0 20.0 35.0
From known,
Y
15.8 13.8 15.8 15.0 47.0 48.0 289 32 3 31.1 53.0 19.9 35 0 36.0 34.2
Y
+0.8
-1.2 +0.8 0.0 +Z.O
+3.0 -1 1 +Z 3 $1.1 ‘
+3.0 -0.1 -0.0 fl.O -0.8
% 5.3 8.0 5.3 0.0 4.4 6.7 3 7 7 7 3.7 6.0 0.5 0.0 2.9 2.3
Standard, Y
0.82 0.50 1 41
,..
... 0.74
SUMMARY
Fluoride ion was determined spectrophotometrically in t h e range 0 t o 50 micrograms of fluoride with a mean standard deviation of 0.85 microgram and an over-all accuracy of +4.0(%., T h e modified Willard-Winter still (8) alloivs the distillation of fluoride ion as fluosilicic acid, without impurities, provided the temperature does not exceed 135” C. The color is stable for a t least 24 hours for samples that do not require distillation. I n the presence of the amount of perchloric acid used in the distillation, the color does not change appreciably in 3 hours. The standard calibration curve does not follow Beer’s law. Its most sensitive portion is in the range 0 to 20 micrograms, where an average increase of 1 microgram produces a n average decrease of 0.083 in optical density. ACKNOWLEDGMENT
The distillation apparatus was adapted from a macro fluoride still designed by H. L. Hemphill of this laboratory. LITERATURE CITED (1) Kuanetsov, V. I.,J . Gen. Chem. (U.S.S.R.), 14, 914-19 (1944). (2) Mellor, J. TV., “Comprehensive Treatise on Inorganic and Theoretical Chemistry,” Vol. 11, London, Lorigmans, Green d: Co., 1922. (3) Monnier, D., Rusconi, Y., and Wenger, P., Help. Chim. Acta, 29, 521-5 (1946). (4)
Thomason, P. F., and Miller, F. J., Oak Ridge National Laboratory, ORNL-686,p. 13, Quarterly Report for January-March
1950. (5) Thomason, P. F., Perry, 51. A , , and Byerly, W. M., ANAL. CHEM., 2 1 , 1 2 3 9 (1939). (6) Thrun, IT. E., Ibid., 22, 918-20 (1950). (7) Willard, H. H., and Horton, C. 9., I b i d . , 22, 1190-4 (1950). (8) Willard, H. H., and Tinter, 0. B., ISD. ENG.CHEM.,AKAL.ED., 5,7-10 (1933). RECEIVED for review July 25, 1961. Accepted October 18, 1951.