‘(2) Boltz, D. F., Mellon, M. G. IND. ENG.CHEV.,ANAL.ED. 19, 873 (1947). (3) Chaney, A. L., Magnuson, H. J., Zbid., 12,69! (1940). (4) Denigks, G., Compt. rend. 171, 802 (1920). (5) Jacobs, M. B., Sagler, J., IND. ENG. CHEM.,ANAL.ED.14, 442 (1942). (6) Jay, R. R., Dickson, L. R., Petroleum Processing 9, 374 (1954). ( 7 ) Kingsley, G. R., Schaffert, R. R., ANAL. CHEM.23,914 (1951).
(8) Magnuson, H. J., Watson, E. B., IND.ENG. CHEM.,ANALED. 16, 339 (1944). (9) Maranowski, N. C., Snyder, R. E., Clark, R. O., ANAL. CHEM. 29, 353 (1957). (10) Morris, H. J., Calvery, H. O., IND. EKG.CHEM.,ANAL. ED. 9 , 447 (1937). (11) Moser, L., Ehrlick, J., Ber. 55, 437 (1922). (12) Sandell, E. B., “Colorimetric Deter-
mination of Traces of Metals,” 2nd ed. Interscience, New York, 1950. (13) Shipman, G. F., Milner, 0. I., ANAL. CHEM.30, 210 (1958). (14) Truo E., Meyer, A. H., IND. ENG. CHEM.,~ N A L ED. . 1, 136 (1929) (15) Woods, J. T., Mellon, M. G., Zbid., 13,760 (1941). (16) Zinzadze, C., Ibid., 7, 230 (1935).
RECEIVED for review November 8, 1957. Accepted April 25, 1958.
Separation of Fluoride from Inorganic Compounds by Pyrolysis R. H. POWELL and OSCAR MENIS Analytical Chemistry Division, Oak Ridge National laboratory, Oak Ridge, Tenn. An effective pyrolytic method has been developed for the separation of micro and macro quantities of fluoride from inorganic materials. The pyrolysis is carried out under a stream of moist oxygen in a fused silica reactor tube. The fluoride, which is volatilized, i s absorbed in a small volume of dilute sodium hydroxide, and then determined either by an acidimetric or spectrophotometric titration. Excessive dilution of the released fluoride by condensate, which occurs if pyrohydrolytic or Willard and Winter distillation techniques are used, is avoided; this is advantageous when microgram quantities of fluoride are to be determined. For the more difficultly decomposed fluorides,-e.g., alkali and alkaline earth fluorides-a reactive oxide such as uranium oxide or tungstic oxide is added to the sample to accelerate release of the fluoride. Optimum operating conditions are established whereby milligram quantities of fluoride can b e separated in 20 minutes or less and subsequently determined with a coefficient of variation of 0.1 For trace amounts of fluoride (-10 y) the coefficient of variation is of the order of 5%.
70.
N
for a rapid, simple, effective method for the separation of microgram quantities of fluoride from compounds of thorium and uranium, and their aqueous slurries, prior to the determination of the fluoride, led to the development of a pyrolytic method which has distinct advantages over earlier methods. Fluoride was usually separated from interfering substances as silicon tetrafluoride, prior to the development of the Willard and Winter distillation method (9) in 1933, in which fluoride was volatilized as hydrofluosilicic acid. EED
1546
ANALYTICAL CHEMISTRY
The next significant advance was the introduction of the pyrohydrolytic method of Warf, Cline, and Tevebaugh (8), whereby fluoride salts were hydrolyzed in a stream of superheated steam, the volatile fluoride mas recovered in the condensate, and then determined by one of several methods (4, 6). Modifications of Warf’s apparatus have since appeared in the literature. Susano, White, and Lee, (7) substituted a nickel apparatus for the platinum tube. Lee, Edgerton, and Kelley (3) constructed a special apparatus for the pyrohydrolysis of semimicro and micro quantities of fluoride compounds in radioactive materials. Hibbits ( 2 ) applied pyrohydrolysis to the removal of fluoride from uranyl fluoride. Recently, Gillies, Keen, Lister, and Rees (1) described a silica apparatus for the pyrohydrolytic determination of fluoride and uranium in uranium tetrafluoride. All earlier work involved the hydrolysis of fluoride salts a t a high temperature by superheated steam, for lvhich Karf adopted the term pyrohydrolysis. The proposed method uses moist oxygen rather than steam in the reactor to purge the system during the high temperature pyrolysis of either readily or difficultly decomposed fluoride compounds. I n the pyrolysis of the latter, the release of fluoride is accelerated by the addition of a reactive oxide to the sample; uranium oxide, which is widely used in the pyrohydrolytic method, is a satisfactory accelerator. Of other oxides evaluated, tungstic oxide was found to be more effective than uranium oxide a t a lower temperature. I n addition, a simplified apparatus was designed in which the reactor tube, fabricated of fused silica with a n air-cooled delivery tube, replaced the water-cooled condenser. The effect of variables, such as reactor tem-
perature, f l o ~rate, moisture content of the purge gas, and materials used as reactants, on the separation of micro and macro quantities of fluoride was studied, and optimum operating conditions were established. I n general, the pyrolytic method is more rapid than either the pyrohydrolytic or Willard and TJ7inter distillation procedures. The fluoride is recovered in a small volume of solution rather than in a large volume of condensate. This is advantageous if microgram quantities of fluoride are to be determined, because high sensitivity can be attained without concentrating the solution prior to determination of the fluoride. Also, the pyrolytic method is applicable to a wider variety of materials than the hydrofluosilicic acid distillation method of Willard and Winter, the limitations of ivhich are summarized by Simons (6). Aluminum, zirconium, or gelatinous silica, which retard or prevent volatilization of the fluoride in the distillation procedure, offer no difficulty in the pyrolytic separation. APPARATUS
The apparatus (Figure I) consists of the following components. Gas cylinder n-ith flowmeter and regulator. Oxygen is supplied from the cylinder, while the flow of the gas is measured by means of a calibrated flowmeter. Water tower. Oxygen is passed through the scrubbing water tower, containing water at room temperature, to saturate the gas with moisture. Reactor tube, fused silica, 24 inches in length and 11/4 inches in outer diameter. The reactor tube is connected to the gas supply by a 29/42 standard taper joint, lubricated with powdered graphite. Delivery tube, fused silica, 10 inches in length and 3/s inch in outer diameter. This tube is fused a t right angles t o the exit end of the reactor tube.
add the washings to the flask. Determine the fluoride content of the trap solution by alkalimetric or spectrophotometric titration, whichever is apphcable. EXPERIMENTAL
.
Figure 1. A. B. C. D. E.
F.
Micro and macro pyrolytic apparatus
Regulator Flow indicator W a t e r tower Joint, T/S 2 9 / 4 2 , fused silica Pyrolytic reactor tube, fused silica Split type furnace, 12"; 11/4(' bore
Hevi-Duty, split combustion furnace. This furnace is used to heat the reactor tube; the power supply is controlled by a variable transformer (Variac, Model v-20). A calibrated pyrometer with a Chromel-illume1 thermocouple is used to measure the temperature of the furnace; the thermocouple is a t the external wall of the reactor tube mid-point between the ends of the tube furnace. Combustion boats, Alundum, 31/2 inches long X inch wide X jlleinch high, or platinum, 3112 inches long x inch wide x 3/* inch high, REAGENTS
Reagent grade chemicals and demineralized water were used in the preparation of all reagents and in the procedure. Sodium fluoride, NaF. This reagent was dried a t 110" C. for 1 hour. Tungstic oxide, W03. Tungsten(V1) acid, purified, anhydride. Uranium oxide, U308. The oxide was prepared by igniting precipitated ammonium diuranate a t 900" C. for 1 hour. Standard mixture of sodium fluoride and uranium oxide, 100 y of fluoride per gram. A 12.7-mg. portion of sodium fluoride was mixed with 57.6 grams of uranium oxide by grinding the mixture in a mortar, after which the mixture was further blended by rotating it for 16 hours in a bottle attached to a V-shaped tumbler. During this period, portions of the mixture were removed, again ground in a mortar, and returned to the bottle for further blending. Standard 0.0020-V sodium hydroxide. The solution was standardized with potassium hydrogen phthalate in the prescribed manner. PROCEDURE
Preparation of Sample. SOLID SAMPLES.Transfer 3 grams of finely powdered tungstic oxide to a 4-inch diameter, octagonal Mullite mortar. Add a known weight of a solid sample (0.1 to 2.0 grams), depending on the fluoride concentration, to the mortar, and mix by grinding. Transfer the dry
G.
H. 1. 1. K.
Receiving flask Mognetic stirrer Buret, goose-neck Variable transformer Thermocouple and pyrometer
mixture quantitatively to a platinum or Alundum boat of suitable size ( 3 l / 2 to 6 inches, depending on the volume of the dry mixture). LIQUIDSAMPLES OR SLURRY. Transfer a 2-ml. test portion of a liquid sample, or a suitable weight of slurry, to a 3112 inch platinum boat and place on a clean 7 x 9 inch porcelain tray. Add 3 drops of phenolphthalein solution to the boat. Add sufficient 1M sodium hydroxide to the boat to make the solution alkaline to phenolphthalein (avoid a large excess of sodium hydroxide). Evaporate to dryness under an infrared lamp. Keep the solution alkaline during the evaporation by addition of 1147 sodium hydroxide as required. Add 3 grams of finely powdered tungstic oxide to the residue in the boat; pack Kith a microspatula. Pyrolytic Separation. Assemble the pyrolysis apparatus as shown in Figure 1. Start the flow of oxygen, saturated with water vapor a t room temperature, and regulate the flow rate a t 4 to 6 cubic feet (120 t o 180 liters) per hour. Turn on the variable voltage transformer which controls the temperature of the pyrolysis apparatus. Heat the fused silica reactor tube to a specified temperature according to the material being pyrolyzed and the reactant that is present: no reactant (see Table I); tungstic oxide 825" C.; and uranium oxide 1000" C. Adjust the setting of the variable voltage transformer to maintain the selected temperature. Add a suitable volume of standard sodium hydroxide solution to an Erlenmeyer or volumetric flask; place the flask under the delivery tube from the pyrolysis apparatus so that the tip of the tube is immersed in the solution to a depth of approximately 0.5 inch. Add 4 drops of phenolphthalein solution to the receiving vessel. Remove the plug from the fused silica reactor tube, and insert the platinum (or Alundum) boat containing the prepared sample in the center of the furnace. Quickly replace the plug and continue the pyrolysis for 15 minutes, maintaining the selected temperature. Remove the flask containing the sodium hydroxide solution, Wash down the tip of the delivery tube;
Structural Material of Reactor. After experimentation, fused silica was selected as the material for t h e reactor tube. Warf ( 8 ) , in his method of pyrohydrolysis, utilized a reactor tube of platinum and fused silica. Later, Susano (7') and his coivorkers substituted nickel as the structural material. The nickel apparatus has been found to be durable, compact, economical, and easy to manipulate; and in the initial tests, a nickel reactor tube mas utilized. It was found, hon-ever, that fluoride was being retained on the inner surfaces of the nickel equipment, LT hen oxygen was substituted for steam. I n an effort to eliminate this retention, an Alundum reactor tube was substituted, Although fluoride was not retained on the Alundum surfaces, this substitution was unsatisfactory, since Teflon plugs, machined to fit a t both the entrance and exit of the tube, were difficult to fabricate, and did not fit tightly during use. A fused silica reactor and delivery tube was substituted for the Alundum apparatus, and tests were made to evaluate the applicability of the reactor to the separation of fluoride by pyrolysis. After this study m s completed, it was learned that Gillies et al. (1) also used an all-silica apparatus, with a generated steam and water-cooled condenser, in carrying out the usual pyrohydrolytic procedure. The results of tests in the simplified reactor apparatus revealed that fluoride can be recovered quantitatively in the absence of steam, but in a continuous flow of oxygen. Rate of Recovery of Fluoride. The rate of recovery is dependent on such factors as temperature of the reactor tube and delivery tube, rate of flow of oxygen, moisture content of the gas, and the reactive oxide present. TEMPERATURE.The rate of recovery of fluoride is enhanced as the temperature of both the reaction site and the exit of the reactor is increased. When the exit of the reactor tube is 8 em. away from the furnace and consequently a t a lower temperature, the rate of recovery is considerably slower. If the temperature of the exit is increased by moving the delivery tube to within 1 cm. of the furnace, it may reach as high as 150" C., and the rate of recovery is increased. RATE OF FLOWOF OXYGEN. The effect of the rate of f l o ~of oxygen on the rate of recovery of fluoride, vas such that, with a flow of oxygen maintained a t 60 liters per hour, the recovery rate was reduced; but, with oxygen a t VOL. 30, NO. 9, SEPTEMBER 1958
1547
Table 1.
Compound ThFr . zHZO AlF3.3l/z HzO ZrFl. rHzO UOzFz, z H ~ O UF,
Pyrolysis of Inorganic Fluoride Compounds (2 meq. of fluoride)
Pyrolysis Temp., O C. 800
900 650
800 800 700
Npz3'F4
Time, Minutes 20 20 10 10
Temp. Range, C. Min. Max. 290 800 600 900 son 350 __600 900 600 900 300 1000 ~~~
20 20
Table 11. Precision of the Pyrolysis of a Standard Mixture of Sodium Fluoride and Uranium Oxide
Temperature, C. Oxygen, flow rate, liters/hr. hloisture, g./L Time, minutes Fluoride, y Added Found 10.1 9.6 9.6 10.0 10.5 10.8 11.0 9.8 11 2 10.5
1020 120 to 180 0.01-0.02 5
Difference -0.5 -0.4 +0.3
% 95.0 96.0 103.0 -1.2 89.0 -0.7 94.0 Average 95.0 Coefficient of variation, 70 5
I
CONDITIONS:
45 6
FLUORIDE, AS NaF, mg, OXYGEN: FLOW RATE, fl3/hr. MOISTURE, p g / L I
0'
I
40 I
I
20
40
TIME, MINUTES
Table 111. Pyrolysis of Sodium Fluoride with Tungstic Oxide or Uranium Oxide
(Precision and accuracy in determination of macro quantities of fluoride) Reactive oxide, W03 or U308,
3
g.
XaF, g. 0 167 to 0.173 120 t o 180 Oxygen: F l o rate, ~ l./hr. Moisture, g./L 0 01-0 02 Pyrolysis time, min. 20 Reactive PresRecovery, c Oxide ent Found /O WO? 78 4 78.2 99.8 76.7 76 6 99 9 77.9 78.0 loo. i 75 8 75 6 99.8 Average 99.9 S = 0.14 Coefficient of variation, V , 70 0 . 1 US08 75.9 75.5 99.4 78.1 77.7 99.4 75.8 75.4 99.4 77.5 77.2 99.6 76.1 75.6 99.3 Average 99.4 s = 0.11 Coefficient of variation, V , 700 . 1 120 liters per hour, fluoride was collected a t a much faster rate. K i t h a flow rate of 80 liters per hour, 15 minutes was required to recover 50% of 2 meq. of fluoride, while with a flow rate of 100 liters per hour of oxygen, this was accomplished in less than 1 minute. MOISTUREIN OXYGEN.When oxygen \vas substituted for steam, the pyrolysis of fluoride was conducted with a minimum of moisture present. This substitution was made to aroid diluting 1548
Figure 2. Recovery of fluoride from delivery tube during pyrolysis with dried oxygen
ANALYTICAL CHEMISTRY
I
(0
20 TIME MINUTES
I 30
Figure 3. Rate of recovery of fluoride as a function of the reactive oxide and temperature
the final solution before the estimation of microgram quantities of fluoride. Because ordinary compressed oxygen contains traces of moisture, tests viere made in which this moisture content was reduced to 10 y of mater per liter, after which other tests were made in which the oxygen was saturated with moisture (20 mg. of water per liter) at atmospheric pressure. With a flow of relatively dry oxygen, a period of 4 minutes was required to recover 100% of the fluoride from a sample containing 300 y of fluoride, while, with saturated moisture, 1 minute was required. A small quantity of moisture accelerates the recovery of
fluoride, and this was elucidated by a n experiment in which a larger amount of sodium fluoride was pyrolyzed in a stream of oxygen dried by passing the gas through drying tubes containing silica and magnesium perchlorate. The interior of the delivery tube was rinsed a t intervals with water by applying a slight vacuum a t the entrance of the tube. With rinsing, large fractions of the fluoride retained on the walls of the delivery tube n'ere recovered. The data are presented in Figure 2. The increase in rate of recovery of fluoride when oxygen saturated with moisture a t room temperature is used, can be attributed t o the release of fluoride from the walls of the delivery tube by the small amounts of water in the gas stream. ACCELERATORS. Karf (8) described the role uranium oxide plays as an accelerator in releasing difficultly hydrolyzable fluoride through its reaction I\ ith the other products of pyrohydrolysis to form a diuranate. I n experiments with uranium oxide as an accelerator, both formation of the uranyl fluoride as an intermediate and a diuranate has been found. These compounds were identified from x-ray diffraction data. Figure 3 s h o m the rate of recovery of fluoride, from sodium fluoride, a t different temperatures, when uranium oxide or tungstic oxide is used as the accelerator. A rapid rate of recovery is achieved with tungstic oxide as the accelerator a t 825" C., comparable with that obtained with uranium oxide a t 1000" C. At 650" C., complete recovery of fluoride is attained with tungstic oxide within 30 minutes; at this temperature 1% ith uranium oxide as the accelerator, less than 75% of the fluoride is collected in 30 minutes; nor is recovery complete even after prolonged periods of pyrolysis. Only 92% of the fluoride was recovered at the end of 31/, hours of continuous pyrolysis. When no acceleration is necessary to release the fluoride from fluoride-bearing materials, lower temperatures may be utilized in the pyrolysis of these compounds Rith moist oxygen. Data for the pyrolysis of several of these materials are given in Table I. I n this table, the temperature at nhich the initial release of fluoride was noted is given as the minimum temperature, 11-hile the maximum temperature is the highest temperature a t which the pyrolysis of these materials can be quantitatively carried out. At temperatures above the maximum, fluoride was lost. Precision of Method. The precision and t h e accuracy of this method for determination of microgram quantities of fluoride were established by pyrolyzing aliquots of a mixture of sodium fluoride and uranium oxide of known composition. From the data
in Table 11, it can be noted that 10 y of fluoride can be separated from sodium fluoride by pyrolysis with a coefficient of variation of 5%. The precision and accuracy for the determination of macro quantities of fluoride, 4 meq. of sodium fluoride, with uranium oxide or tungstic oxide as the reactive oxide xere evaluated (Table 111). When tungstic oxide is used as the reactant, the coefficient of variation is 0.1% and no bias is observed. When uranium oxide is present a s the reactant, the coefficient of variation is 0.1% and a negative bias is apparent. This bias was also noted by Warf et al. Interferences. The present study was restricted mainly t o t h e separation of fluoride b y pyrolysis from materials which contained no other volatile acid-forming components. After liberation by pyrolysis, fluoride was determined either by an alkalimetric titration or spectrophotomet-
rically with alizarin red-S as the chromogenic reagent ( 5 ) . Neither orthophosphates nor silicates, when present in the pyrolyzed sample, interfered. Sulfates, chlorides, or nitrates will interfere with the acidimetric titration of fluoride. I n the presence of chloride, nitrate, or small amounts of sulfate, the fluoride can be determined by spectrophotometric titration with thorium nitrate, using alizarin r e d 3 or thoron as the indicator. ACKNOWLEDGMENT
The authors acknowledge the assistance of H. P. House and hl. A. Marler in the preparation of the manuscript, and of N. 11.Ferguson for some of the fluoride determinations, LITERATURE CITED
(1) Gillies, G. &I., Keen, N. J., Lister,
B. A. J., Rees, D., Atomic Energy
Research Establishment C/M 225. G. Brit., Oct. 26, 1954. (2) Hibbits,-_J. O., AXAL. CHEM. 29, 1760 (193,). (3) Lee, J. E., Jr., Edgerton, J. H., Kelley, 31.T., I b z d , 28, 1441 (1956). (4) Mellon, ?VI. G., Boltz, F. F., Ibzd., 28, 785 (1956). ( 5 ) Kicholas, M. L., Kindt, B. H., Ibid., 22, 785 (1950). (6) Simons, J. H., “Fluorine Chemistry,” Vol. 2, pp. 83-9, Academic Press, New York, 1954. (7) Susano, C. D., White, J. C., Le?, J. E., A N A L . CHE\f. 27, 453 (1955). (8) Warf, J. C., Cline, W. D., Tevebaugh, R. D., Ibzd., 26, 342 (1954). (9) Willard, H. H., Winter, 0. B., IXD.ENG. CHEII., ANAL. ED. 5 , 7 (1933). RECEIVEDfor review October 14, 1957. Accepted March 24, 1958. Division of Analytical Chemistry, 132nd meeting, ACS, New York, 5 . Y., Sept. 1957. Work carried out under Contract KO.W-7405eng-26 a t Oak Ridge Sational Laboratory, operated by Union Carbide Kuclear Co., a division of Union Carbide and Carbon Gorp., for the U. S. iitomic Energy Commimion.
Determination of the Alkalinity and Borate Concentration of Sea Water JAMES A. GAST and THOMAS G. THOMPSON Department of Oceanography, University of Washington, Seattle, Wash.
b The mannitol method for determining borates in sea water has been improved to reduce its experimental error, thereby giving greater reliability to the variations observable in the boron-chlorinity ratio of sea water. The procedure allows simultaneous determination of the alkalinity cf the water. One of the causes of deviation from a constant ratio is the presence of polyhydroxy organic compounds dissolved in sea water, which react with the borate ions to give low results with the mannitol method. To determine the total borate it is essential to oxidize the organic matter. A procedure using potassium permanganate as the oxidant is given. The difference between th8 determinable borate in an oxidized sample (total borate) and in an unoxidized sample (free borate) gives an estimation of the concentration of some of the organic matter dissolved in the sea water. The borate-boron chlorinity ratio, calculated from free borate boron, indicates certain conditions within a water mass.
S
of the chemical composition of sea water have shon-n that the major constituents occur in constant TUDIES
ratios to each other and that sea water thus varies only in the degree of dilution. This is evidenced by the fact that, with constant temperature, the physical properties of sea waters are a function of the chlorinity or salinity. Changes in the concentration of the major constituents are so small that they cannot be measured because of the limitations of analytical methods. On the other hand, easily observed changes in the concentration of many of the trace elements or their compounds vary considerably from place t o place, a t different depths, and a t different seasons of the year. Between the two extremes of inajor and trace constituents, sea water contains substances generally designated as minor constituents, some of which show a constant ratio with the chlorinity. The borate ion is one bf the minor constituents. The averaged values of different investigators indicate that sea water having a chlorinity of 19.000/oo (parts per thousand) \Till contain borate in a concentration equivalent t o 0.4237 nig. atom of boron per kg. of sea m t e r , and the ratio of these t n o values is 0.0223. I n 1933, Moberg and Harding ( 7 ) analyzed some 50 samples of nater
secured from various depths along thc California coast, near the Hawaiian Islands, and near the Dry Tortugas off the coast of Florida. They obtained a value of 0.0221 for the ratio. I n the same year Buch (1) reported a ratio of 0.0223 for samples of widely varying chlorinities on waters taken from the North Sea and some of its arms. Buch also quoted a ratio of 0.0231 obtained b y Wattenburg on Atlantic water. Rakestraw and l l a h n cke (8)in 1935 reported borate-boron chlorinity ratios for t n o areas of the n estern Atlantic-on waters collectcd between Cape Cod and Bermuda, thcy obtained a ratio of 0.0222; for the Gulf of Maine, 0.0236. I n 1938, Igelsrud. Thompson, and Zn-icker (4) published a n average ratio of 0.0223 for waters taken a t various depths in the northeast Pacific; the waters of Haro Strait and the San Juan ilrchipelago in the State of Kashington; Hecate Strait, Britipli Columbia; Dixon Entrance, Pearcc Canal, and Portland Canal, Alaska. The 377 samples analyzed shoned a n average deviation of 0.0008 from the ratio, with 127 of the samples having a ratio greater than the average deviation; the average maximum deviation for the six areas was 0.0027. VOL. 30, NO. 9, SEPTEMBER 1958
1549
