Ion Exchange Method for the Determination of Fluoride in Potable

Zygmunt Marczenko , Maria Balcerzak. 2000,189-197. Fluoride. W John Williams. 1979,335-373. Photometrische Bestimmung von Fluoriden im Trinkwasser mit...
0 downloads 0 Views 419KB Size
fects on the electrodm could lead t o erroneous results. To minimize this potential source of error, electrode polarity was reversed midway in the sparking step and also in the washing step of each run. dlthough solubility of carbon dioxide in 96y0 sulfuric acid is low at room temperature (0.8 cc., s.t.p./ml. acid) ( 6 ) , and exchange oj carbon dioxide with 96y0 sulfuric aci1-l is very slow a t room temperature (Itl is 2 days at 134" C.) (8),it was felt that any possible error arising from the dichromate step could best be eliminated or standardized by using a fresh sample of dichromate solution for each exchange. One precaution which the authors suggest is that all !,topcocks in the apparatus be of a t least 4-mm.

bore. The authors had already constructed their vacuum line and apparatus with some Z-mm.-bore stopcocks and found that p u m p d o w time was unduly long. I n processing a series of sulfur dioxide samples the average time per sample was about 50 minutes. This could probably be reduced significantly and still maintain the desired 10-~-mm. vacuum if larger bore stopcocks were used throughout. Finally, it is suggested that adequate protection of the eyes be used while operating the spark bulb for long periods of time. ACKNOWLEDGMENT

The authors are indebted to Henry Taube on whose mass spectrometer all measurements were made.

LITERATURE CITED

(1) Giifillan, E. 'i. Jr., , Polanyi, M.,Z. Phvsik. Chem. A166. 255 (1933). (2) Grigg, E., Lauder; I., Trans.' Farad. SOC.46, 1033 (1950). (3) Halperin, J., Taube, H., J . .4nz. Chem. ~SOC.74,375 (1952). ( 4 ) Hoering, T. C., Kennedy, J. IT.,Ibid., 79, 56 (1957). ( 5 ) Hriston, J. I,., J . Phys. Chem. 63,

389 (1953). (6) . , Mrrkham. A. E., Kobe, K. A,. J . Am. Chem. Soc. 63. 1165 (1941). (7) Sakata, S.,'J. Chem. SOC.Japan 64, 635 (1943). (8) Spittlcr, T. M., Ph.D. thesir, Loyola UniversitJ-, Chicago, Ill., (1961).

RECEIVE^ for review June 14, 1963. Accepted Xovember 13, 1963. Work supported by National Science Foundation Grant S o . G-4484 and a Cooperative Graduate Fellowship.

Ion-Exchange Method for Determination of Fluoride i n Potable Waters FOYMAE S. KELSO, JOHN M. MATTHEWS, and HARRY P. KRAMER Public Health Service, Training Program,

b The adsorption of fluoride by resin, with subsequent stripping, eliminates the necessity of distilling potable water samples to remove interferences prior to spectrophotometric analysis, The accuracy and precision obtained b y the ion-exchange method are comparable to those obtained b y standard distillation proc(2dures and the saving in time and equipment is considera ble.

T

ODAY more than 2000 cities and to\\-ns in the United States (4), serving more than 40,000,000 people, add fluorides to their public water supplies as a caries preventive. For this purpose, the ('oncentration of fluoride is normally maintained a t approximately 0.8 to 1.2 mg. per liter. However, mater supplies in many parts of the country vontain naturally occurring fluorides in v,trying quantities, and industrial disch2 rges often contribute additional fluorides to the normal content. For years distillation has been standard procedure for eliminating interferences in the determination of fluoride in water samples (6). Distillation is a prolonged procedure requiring the close attention of the analyst throughout the process. For laboratories routinely performing water anal) ses, this timeattention factor has bectome a significant coqt consideration. This report details a simple, accurate,

R. A.

Tuft Sanitary Engineering Center, Cincinnati 26, Ohio

rapid, and economical ion-exchange method for removing interferences from waters containing low concentrations of fluorides. With this technique, a laboratory analyst can perform 20 to 30 water analyses a day with a precision and accuracy of zt0.1 mg. per liter while performing other laboratory activities. Talvitie and Brewer ( 5 ) developed an ion-exchange procedure for separation of fluoride from urine. Their method employs formation of a berylliumfluoride complex assumed to be (BeF4-2), as does the procedure discussed here. Feigl and Schaeffer (3) reported on this complex in 1951. EXPERIMENTAL

Beckman Model B spectrophotometer, or equivalent, with 1-cm. matched cells. Magnetic stirrer t o accommodate a 1500-ml. beaker. Sedgwick Rafter cone, 500-ml. capacity, with bolting cloth disks. Chromatographic columns, 10 mm. i.d. X 300 mm., with Squibb separatory funnels (reservoirs) and 14/35 standard taper male joint. Corning Glass Works Order X-731864. Eight-liter carboy for washing resin and a siphon assembly. Reagents. Anionic resin (Dowex 2 x 4 ) chloride form, 50-100 mesh. Ethanol, U.S.P. Hydrochloric acid, 3M. Sodium acetate, 1M. Beryllium acetate stock solution, 0.05M. Apparatus.

Ammonium (ethy1enedinitrilo)tetraacetic acid (EDTA) solution, 0.13M. Mixed S P A D S S solution, sodium 2( p - sulfophenylazo) - 1,8 - dihydroxy-

naphthalene-3,6-disulfonate. SPADSS reference solution.

To purify the resin, wash in the followiiig manner and decant after each wash. Place 450 grams of resin in a 1500-ml. beaker, wish with two 300-ml. portions of 957, ethanol, two 300-ml. portions of distilled water, and five 300-ml. portions of 3111 hydrochloric acid, then transfer the resin to a Sedgwick Rafter cone and wash with 8 liters of distilled water by means of a siphon or similar arrangement. This removes excess chlorides and restores p H to neutral. Transfer the resin to a 1500-ml. beaker, add 350 ml. of 1-11 sodium acetate solution, and stir with a magnetic stirrer for 15 minutes. Diqcard the supernatant solution. Prepare a 1 to 1 slurry of the acetate-resin and dirtilled water and store in a closed polyethylene container. Prepare beryllium stock solution by pipetting 57 ml. of glacial acetic acid into 500 ml. of distilled water, dissolving 2.6 grams of reagent grade beryllium carbonate, (BeO)5.COs.5H20 (Fisher Scientific), and diluting to 1 liter with distilled water. This solution is 0.05-11 with respect to beryllium and 1.0-If with respect to acetic acid. ( S o t e toxicity of beryllium, and exercise care.) Dilute the beryllium stock solution 1 to 10 with distilled water to provide the beryllium eluting solution used to remove the fluoride from the acetate resin. Use purified white quartz sand, 60VOL. 36, NO. 3, MARCH 1964

577

Table

I.

NO

interference Mean Av. 70recovery Std. dev.

0.52 104 0.023

Mean Av. % recovery Std. dev.

1.02 102 0.021

Precision and Accuracy of Fluoride Values in the Presence of Interferences

(For distilled water containing 0.50 and 1.00 mg./l. fluoride) 20 mg./l. 1000 m.g./l. 2 . 0 mg./l. 1000 mg./l. phosphate 400 mg./J. chloride hexametasulfate (P04+3) CaC03 (el-) phosphate (SO,-2) 0.50 Mg./L. of Fluoride Added 0.50 0.51 0.52 0.49 0.53 104 98 106 100 102 0.014 0.023 0.034 0.021 0.057 1.00 Mg./L. of Fluoride Added 1.06 1.01 0.96 1.08 0.98 106 101 108 96 98 0.043 0.073 0.060 0.035 0.067

to laO-mesh, to protect the surface of the resin from disturbance and to prevent drying at top of column. Purify the sand by digestion a t 100" C. for 1 hour with 250 ml. of 20% sodium hpdroxide, discard the supernatant solution, and wash with 250 ml. of 1 to 3 hydrochloric acid. R a s h in a 2-liter flask with distilled water until all traces of chloride are removed. Use ammonium (ethylenedinitrilo)tetraacetic acid solution when aluminum is present. To prepare the 0.1331 KH4EDTA solution, dissolve 50 grams of Sa2C10H1408N2 2H20 in 70 ml. of concentrated xH4OH and dilute to 1 liter with distilled water. The mixed SPAIDKS solution is a mixture of 2-(p-sulfophenylazo)-1,8-dihydroxynaphthalene - 3,6 - disulfonate (Eastman 7309) and zirconium-acid solutions. Prepare i t by dissolving 0.958 gram of the above dye in about 250 ml. of distilled water and diluting to 500 ml. Prepare the zirconium-acid solution by dissolving 0.133 gram of zirconium oxychloride octahydrate in about 50 ml. of distilled water, adding 350 ml. of concentrated HC1, and diluting to 500 ml. An equal mixture of these solutions will produce the mixed SPADSS solution (1, 2 ) . The reference solution is used for setting the spectrophotometer in the proper range to read the standards and samples. Prepare i t by diluting 10 ml. of the SPADYS solution to 100 ml. with distilled water, diluting 7 ml. of concentrated HC1 to 10 ml. with distilled water, and then mixing the two solutions (1, 2 ) . Procedure. While the magnetic mixer keeps t h e purified resin slurry i n suspension, pipet 25 ml. into each reservoir. After t h e resin has settled i n t h e columns, rinse all remaining resin from t h e reservoirs. Add purified sand t o each column to form a t o p layer 1 cm. thick. Wash t h e reservoirs again with distilled water t o remove sand particles from sides, permitting wash water t o drain before addition of sample or standard. Prepare a standard curve for the range 0.00 to 0.06 mg. of fluoride by diluting appropriate amounts of fluoride to 50 ml. with distilled water, and passing each standard through ion-exchange columns. Prepare individual columns for each sample, If no aluminum is present, pipet 50 ml., or a n aliquot of sample

578

e

ANALYTICAL CHEMISTRY

diluted to 50 ml., into the reservoirs. The flow rates of the columns are selfregulated a t about 3 ml. per minute. After the samples and standards have been pipetted into individual columns and the columns allowed to drain, wash the columns with 100 ml. of distilled water, and allow to drain. Discard all sample and wash waters. Remove fluoride adsorbed on the resin by addition of 100 ml. of the beryllium eluting solution. Carefully collect eluent in 250-ml. glass-stoppered graduated cylinders. Pipet 20.0 nil. of mixed S P d D N S solution (SPADSS zirconium-acid in a 1 to 1 ratio) (1) into each flask, mix well, and read in 1-cm. cells a t 570 mp against the reference solution ( I , 2). Treat the standards exactly as the samples, both of which contain the beryllium eluting solution. If aluminum is present in excess of 0.5 mg. per liter, add 1 ml. of the KH4EDTA solution, followed by 2 ml. of 0.5A' S a O H to samples and standards prior to passing both samples and standards through ion exchange columns. These additions will adjust the p H to 11.5 to 12.0, which is the range necessary for complete complexation of the aluminum fluoride complexes. Then pass samples and standards through the columns and follow the procedure detailed above, but allow 30 minutes for full development of color in the presence of aluminum.

+

-

DISCUSSION OF RESULTS

Evaluation of Effect of Individual Interferences (Table I). Reported concentrations of interfering substances causing an error of 0.1 mg. per liter at 1.0 mg. per liter of fluoride ( I ) , employing the SPADNS-zirconium lake procedure (2) are:

Substance Alkalinity (CaCOa) Aluminum (A1 Chloride (Cl-) Iron (Fe + 3 ) Hexametaphosphate [(YaPOdd Phosphate Sulfate

Concentration, mg./l. 5000 0.1 7000 10

1.0 16.0 200

Type of error

+-

+

++

0 . 5 mg./l.

chlorine (CL)

5 . 0 mg./l. aluminum (~1+3)

0.52 104 0.022

0.50 100 0.014

1.09 109

0.99 99 0.031

0.036

The concentrations of interfering substances selected for this study were significantly higher than those usually found in potable waters. The interfering substances shown in Table I mere added to individual portions of distilled water containing 0.50 and 1.0 mg. per liter of fluoride, respectively. I n Table I, at the 0.5-mg. per liter fluoride level, the standard deviation ranged from 0.014 to 0.057 mg. per liter. Accuracy obtained in the presence of various interfering substances is indicated by the average percentage recovery. Pretreatment of sample was practiced only in the aluminum series, as discussed in the procedure. All other interferences (in solution containing the fluoride) were added directly to the An resin without pretreatment. aluminum concentration of 0.5 mg per liter produced significantly low results unless masked by means of complexation with E D T d . I n other fluoride methods, a delay in evaluating the samples on the spectrophotometer appears to eliminate the negative aluminum interference, but a delay of 70 hours did not produce this effect with the subject method. Pretreatment of samples and standards eliminated this interference up to a concentration of 5 mg. per liter of aluminum. Evaluation of Effect of Combined Interferences. All the interfering substances evaluated individually in Table I, in the same concentrations, are combined in this sample. Standard deviations of 0.038 and 0.073 were obtained for the two concentrations of fluoride in the seven replicate analyses performed for each concentration. Mean values of 0.55 and 1.01 mg. per liter, respectively, were obtained for the replicate analyses. Dosed Tap Water. The dosedwater sample (Cincinnati t a p water) may be considered t o be more representative of municipal water supplies t h a n the previous samples. The mean of the replicate analyses, for both fluoride concentrations, varied only 0.01 mg. per liter from the dosed concentrations. The tap water con-

tained 58 mg. per liter of chloride, a total hardness of 154 per liter, 156 mg. per liter of sulfate, less than 0.1 mg. per liter of aluminum and total phosphate, and 0.020 mg. per liter of naturally occurring fluoride The average percentage recovery a t the fluoride level of 0 19 mg. per liter was 105% with a standard deviation of =kO0.O35mg. per liter. The precision obtained for the 0.87 mg. per liter concentration was +0.034 mg. per liter with an average recoveiy of 101%. Effects of Different Lots of Resin (Table 11). T o investigate the possibility of resin effect upon the method, five different lots mere obtained and eight replicate determinations were made on each lot. The standard deviations mere all less than those shown with the samples studied, except for lot C. Here the value of 0.090 exceeds the highest standard deviation value (0.072) found for the samples. This was not investigated further. Lack of data concerning regeneration and re-use of the resin became apparent during development of the procedure. For the samples analyzed, freshly prepared resin could be re-used three times, a t a dosage of 1.0 m g . per liter of fluoride, before it had to be regenerated. The resin in each c:olumn was regenerated by adding 2!5 ml. of 3M HCl, followed by 25 ml. of dutilled water and 25 ml. of 1.0121 NaC2H302solution, and finally rinsing with 50 ml. of distilled water. Each solution was allowed to drain to waste, prior to addition of the next solution. Our drka indicate that the regenerated resin columns may again be used three times before discarding.

-~

____

Table II.

Concn. of fluoride added, mg./l.a

Comparison of Data Using Different Lots of Resin

Resin lot no.

0.50 0.50

A

1.00 1.00

B

No. of replicates analyzed 8 8 8 8 8

B

1 .oo

A

C

Mean,

mg./l. F 0.51 0.52 1.00 1.04 1.03

Av. % recovery

Std. dev.

102 103 100 104 103

0.040 0.027 0.047 0.055 0,090

Sample with all previously discussed interferences added at concentrations indicated in Table I. a

Table 111.

Series of Standard Curves

O.OOmg.F Mean Std. dev. Coeff. of variation,

0.351 0,009 2.6

7’

Additional research regarding extension of the life of the needed. Resin columns may served over extended periods of keeping the resin covered with water. Standard Curves (Table

further resin is be pretime by distilled

111). For

the 15 replicate sets of fluoride standards prepared during this study, good reproducibility was obtained, as indicated in Table 111. A11 standards were passed through ion-exchange columns and treated exactly as the samples. ACKNOWLEDGMENT

We acknowledge the interest and suggestions of S . A. Talvitie, Public Health Service, Salt Lake City, Utah, and of Ervin Bellack, Division of

Absorbance 0.02nig.F O.O1mg.F 0.313 0,010 3.2

0.270 0.006 2.2

0.06mg.F 0.232 0.007 3.0

Dental Public Health, Public Health Service, Bethesda, Md. LITERATURE CITED

(1) American Public Health Association,

American Water Works Association, and Water Pollution Control Federation. “Standard Methods for the Examination of Water and Wastewater,” 11th ed., 1960. (2) Bellack, E., Schouboe, P. J., ANAL. CHEM.30, 2032 (1958). (3) Feigl, Fritz, Schaeffer, A., Ibid., 23, (19.51 ).. _ _ ~ _

(4) J.A W L . Water Works Assoc. 559 (1962). (5) Talvitie, N. A., Brewer, L. A,, Am. Ind. Hyg. Assoc. J. 21, 287-95 (August 1960). (6) U. S . Public Health Service Cin-

cinnati, Ohio, “Water Fluoride,”’ Analytical Reference Service Report, Robert A. Taft Sanitary Engineering Center, 1961.

RECEIVED for review September 26, 1963. Accepted December 30, 1963.

..

Rapid and Precise Method for DetermI nI ng Surface A,reas JASPARD H. ATKINS Research Laboratories, Cabot Corp., Cambridge, Mass.

b A rapid and precise method for the determination of surface areas, developed by refinement of the apparatus and analytical procedures described by Nelsen and Eggertsen, involves adsorption of nitrogen from a stream of helium (2nd nitrogen a t liquid nitrogen temperatures and measurement of the nitrogen desorbed from the sample upon removal of the liquid nitrogen. Relative standard deviations for a series of singlepoint analyses on pt?lletized carbon blacks range from +o.25yOfor a low volatile black to k1.35% for one of high volatility. Use of a correction factor, which is a function

of relative pec.. areas, narrows the range of relative standard deviations To get good to *0.25 to *1.01%. agreement with multipoint volumetric BET determinations, it is necessary to analyze a sample in the apparatus a t two different relative pressures, so as to determine the intercept on the BET plot. A trained operator can do 25 single-point or 15 two-point determinations in an 8-hour work day.

T

most widely accepted method for determining surface area involves measuring the amount of gas adsorbed on a solid surface a t a temperaHE

ture close to the boiling point c- the gas. Kitrogen is most commonly used as the adsorbate. If the adsorption is measured a t several gas pressures, the Brunauer-Emmett-Teller (BET) (3) equation can be used to calculate the amount of adsorbate required to form a monolayer. This value multiplied by the proper factors for area covered per unit amount of nitrogen gives the surface area. Usually the amount of gas adsorbed is determined by measuring pressure differences in a calibrated glass vacuum apparatus. The new method, developed originally by Nelsen and Eggertsen ( d ) , is also based on gas adsorption and use of the VOL. 36, NO. 3, MARCH 1964

579