Semimicrodetermination of Fluorine in Volatile Organic Compounds

Semimicrodetermination of Fluorine in Volatile Organic Compounds. W. C. Schumb, and K. J. Radimer. Anal. Chem. , 1948, 20 (9), pp 871–874. DOI: 10.1...
15 downloads 0 Views 495KB Size
Semimicrodetermination of Fluorine in Volatile Organic Compounds WALTER C. SCHURIB AND KENNETH J. RADIIIER', Massachusetts Institute of Technology; Cambridge, Mass..

acid, and h j drogen peroxide. The bleaching effect of the fluoride on the brown color of the acidic peroxidized titanium solution is measured in a colorimeter of the split-path type. The optimum ratio of the number of equivalents of oxygen to the equivalents of oxidizable carbon and hydrogen in the gas passed through the platinum tube w-as at least 40 to 1, whereas a slight excess of hydrogen (as water) over fluorine sufficed. The effect of temperature on the solution at the time of the colorimetric measurement was appreciable.

A simple, rapid semimicromethod for determination of fluorine in volatile organic compounds has been developed. There is no interference from hydrogen in the sample, which is passed at a constant rate with moist oxygen through a platinum tube at 1100' C. The resultant hydrogen fluoride is absorbed in water in an absorber constructed of saran tubing, the chlorine and carbon dioxide being largely carried off. To the hydrofluoric acid solution phenolphthalein is added, and the solution is neutralized and added to a solution of titanium sulfate, sulfuric

HE identification of fluorocarbons may be accomplished by either physical or chemical means. Many physical methods depend upon the separation of the pure compound, or an azeotrope of it, with a Podbielniak column (5,6,d l ) , followed by a determination of the melting point, boiling point, refractive index, or liquid density. Gas density and melting point lowering may &Is0 be used t o characterize fluorocarbons. Physical methods applicable to impure samples are measurements of infrared absorption and specific inductive capacity. The use of chemical means of analyzing fluorocarbons involves the choice of methods, first, for decomposit,ion of the sample, and secondly, for determination of fluoride in the decomposed sample. For determining fluoride in the decomposed sample, colorimetric procedures recommend themselves from the standpoint of simplicity and speed. Such colorimetric methods for fluoride determination may be classified into three groups: those utilizing the tendency of zirconium to form complex fluorides in solutions; the SteigerMerwin procedure, in which the broLvn color of the ion present in acidic perosidizcd titanium solutions is bleached by fluoride ; and the determination based on the bleaching effect of fluoride on ferric thiocyanate. A review of the literature indicated that there were no rapid accurate semimicromethods for the determination of organic fluoride with which hydrogen does not interfere and which do not involve the use of expensive and specialized equipment or an excessive amount of manipulation. It was felt that the development of such a method would be very desirable. Although the item of expense has not, been eliminated in the procedure described, the amount of manipulation has been reduced and a technique developed that permits analysis of volatile organic fluorine compounds in a flowing system, a distinct advantage in industrial oDerations. T ~methods , ~ of decompos~t~on ere tried: d e c o m p o s ~ ~ of ~on the sample in the presence of oxygen and water vapor in a hot platinum tube and decomposition by passage through a hot silica tube. The silica tube decomposition method x a s tried first with fluorochloro compounds, using nitrogen as t,he carrier gas, but was abandoned because of the interference of silicate and fluosilicate in the direct analysis of the absorbed decomposition products. p i l e use of as the carrier gas in place-of nitrogen offers some promise for the silica tube method and further study of decomposition under these circumstances is contemplated.) 1

Present address, Indiana University, Bloomington, Ind.

871

THERMAL DECOMPOSITION IN PLATINUM TUBE

A nickel tube packed with scrap'platinum was eaten througb by the reaction products of trichlorotrifluoroethane, water, and oxygen a t 1100' C., and therefore a platinum tube was employed, as shown in Figure 1. Various types of absorbers were tried before the saran tubing absorber shown was decided upon as the most efficient for absorption of hydrogen fluoride. Sonaq stopcock grease was used throughout to avoid loss of sample by absorption in the lubricant. The flow of wet oxygen through the tube was 3 ml. per second. Six inches (15 em.) of the foot-long 0.25-inch platinum tube were heated to 1100" C., the highest temperature obtainable with a simple Xichrome-wound resistance furnace (52). Under these conditions it was found that, whereas the number of equivalents of hydrogen (as water) needed to be only slightly greater than the number of equivalents of fluorine, the number of equivalents of oxygen must be about forty times the number of equivalents of oxidizable carbon and hydrogen t o obtain satisfactory results. From this it can be shown that only if the sample being analyzed contained carbon tetrafluoride, hexafluoroethane, octafluoropropane, or CFIX 11 ould it be advantageous to use ammonium hydroxide instead of water for saturation of the oxygen, as suggested by Bockemuller (4). Passage of the products of decomposition through water absorbs any hydrogen fluoride, while carbon dioxide and any chlorine present largely pass through the acidic solution. The inch-long flames about an inch above the absorber serve to eliminate frothing in the absorption solution. T o the solution thus obtained, phenolphthalein was added; the solution was neutralized and added to a solution of titanium sulfate, hydrogen peroxide, and sulfuric acid. The bleaching effect of the fluoride was determined with a KlettSummerson colorimeter-a single-path instrument being inadequate for the purpose-using a Corning KO.554 filter and a calibration curve prepared with known amounts of sodium fluoride. PREPARATION OF SOLUTIONS

Titanium Sulfate-Sulfuric Acid. About 2.5 grams of titanium powder (obtained from Metal Hydrides, Inc., Beverly, Mass.) were dissolved with gentle heating in 300 ml. of a mixture of equal volumes of concentrated sulfuric acid and distilled water. The blue solution was added slowly t o 2100 ml. of the sulfuric acid solution and shaken; oxygen was admitted occasionally, until the blue titanous sulfate oxidized. The entire solution wLLS drawn through a fritted-glass filter. Sodium Fluoride. One milliliter of 4794 reagent hydrofluoric acid was added to 2'1 grams Of 'Odium fluoride and loo ml. Of distilled water in a platinum dish. The solution was evaporated slowly until the volume about 25 ml., then cooled. The crystallized sodium fluoride was removed and ignited in a covered platinum crucible for an hour. (Sodium fluoriae n-as fused in the platinum crucible prior to its use for this ignition until a white melt was obtained.) About 0.86 gram of the product, carefully weighed. was dissolved in 1 liter of distilled water. A n aliauot waGtitrated with acid, using phenolphthalein as indicator, and'the equivalent weight oE alkali was deducted from the sodium fluoride

872

ANALYTICAL CHEMISTRY

P

-9

Figure 1.

Apparatus for Determination of Fluorine in Volatile Organic Substances

,Oxygen capillary flowmeter Oxygen a t constant pressure Water bubbler Capillaries Thermocouple leadn Nickel tube silver-soldered t o housekeeper seal and platinum tube G. Wet cloth wick H. Electric furnace a t 1100’ C.

A. 6. C. D. E. F.

J. K. L. M.

Saran tubing absorber wired t o bent iron strip Small cool gas flames Flexible small-diameter glass tubing Semimicro nample holder

P. Semimicro sample N* Isad hammer Q. Nitrogen capillary flowmeter R . Nitrogen a t constant pressure

sample weight. (Although it is not known whether the sodium fluoride hydrolyzed slightly to yield sodium hydroxide on ignition, or whether the end product was sodium carbonate, deduction of a weight corresponding to half sodium hydroxide and half sodium carbonate, left a maximum uncertainty with regard to the amount of sodium fluoride of 0.11% of the concentration itself.) CALIBRATION OF COLORIMETER

Using 10 ml. of the titanium sulfate-sulfuric acid solution and 5 ml. of 30% hydrogen peroxide, solutions containing known amounts of sodium fluoride were prepared and diluted to 100 ml. A 0.1’ thermometer and the colorimeter tube were rinsed several times with the solution to be measured. Another colorimeter tube was filled with distilled water and used to balance the instrument to avoid errors due to insufficient rinsing of the colorimeter tubes. The fluoride solution with the thermometer in it was placed in the colorimeter, the temperature observed, the thermometer raised, and the galvanometer balanced immediately. The heating effect of the colorimeter itself raised the standard solution’s temperature slowly. Measurements were made for each of the standard fluoride solutions and the results plotted as shown in Figure 2, by using 40 X 50 em. coordinate paper graduated in millimeters. The temperature of the solution in the colorimeter is critical, as is shown by the displacement of the curves in the figure. PREPARATION OF NITROGEN FLOW PLOT

In order to achieve the optimum rate of introduction of the sample being ana1;yZed into the oxygen stream, a Plot of N Q Wainst the boiling Point of the compound was Prepared, where Q is the maximum permissible flow of nitrogen through the sample and N is the maximum possible number of equivalents of oxidizable carbon and hydrogen per mole of the lowest boiling constituent of the sample. This plot was developed, assuming that the compounds obey Ti-outon’s rule, Hm/T = 22, where T is the boiling point in degrees Kelvin, and USing the modified form of the Clapeyron equation, 2.3 log ( p / p ’ )

H,,, ( T

R

-

T’)

TT’

where p is the vapor Pressure a t the absolute temperature, T. T’ was taken as 300” K., as high a r w m temperature as

s.

Macro gas sample weighing bulb T. Nitrogen flow U. Fine capillary V . Mercury w. Macro liquid sample weighing tube X. Regulated nitrogen flow

might ordinarily be expected; p is 1 atmosphere. p ‘ was calculated for each boiling point. If moles of nitrogen having a volume and a pressure p are saturated by a tile liquid of vapor pressure p ’ and the total pressure is kept equal to p , giving the saturated nitrogen a v r , so that the partial pressure Of nitrogen is P’’, then, allowing n’ to be the number of moles of volatile liquid vaporized, it may be shown that: n’ = n p ’ / ( p

S

W

- P’)

Hence the flow of carbon plus hydrogen into the oxygen stream is N Q p ’ / ( l - p ’ ) equivalents per second. Equating this to 1/40 of the number of equivalents of oxygen per second in the flow of 3 ml. per second a t 300” K., iVQ was calculated and the results were plotted (Figure 3). Experiments have show7.n that the use of water for saturation of the oxygen will supply adequate hydrogen for all but samples containing carbon tetrafluoride, hexafluoroethane, octafluore propane, and CFaX,where x’ contains nothing that n,ill consume oxygen. If any of these substances are contained in the sample, the rate of introduction of sample must be reduced to obtain enough hydrogen for conversion of fluorine to hydrogen fluoride. ANALYTICAL PROCEDURE

A sample containing a t most about 15 mg. of fluorine was weighed into the conventional type of microsample tube, made from wafer-thin Pyrex tubing drawn to 2-mm. diameter from 25-mm. tubing. With a small amount of experience adjustment of

V O L U M E 20, NO. 9, S E P T E M B E R 1 9 4 8

873

sample weight was unnecessary. The sample was placed in the sample tube holder which was detached from the system (stopcocks being closed), and the sample tube was broken by allowing the glass-encased lead weight to shatter it. The nitrogen flow was adjusted by consulting the plot of boiling point against nitrogen Bow and the sample tube holder was again attached to the system. Oxygen flow was checked to ensure a flow of 3 ml. per second, the furnace temperature was checked at 1100" C., and the sample tube holder's stopcocks were opened, allowing the sample to be carried over. In less than 15 minutes the samples were completely vaporized. After 30 minutes, the sample tube holder was blackened with a cool flame from a torch. After 50 minutes the solutions of the reaction products were removed, phenolphthalein was added, and the solutions were neutralized in a plastic beaker (unattacked by fluoride) with 0.25 N sodium hydroxide and 0.05 N hydrochloric acid. They were then added t o 10 ml. of the titanium sulfate-sulfuric acid solution and 5 ml. of 30Oj,hydrogen peroxide, diluted to 100 ml., and compared as were the sodium fluoride solutions above. The calibration plot gave the fluoride content of the solution 1 hour from the time the weighed sample was received. Analytical results are shown in Table I.

Table I. Compound CsHsCFi CsHaF ClFls

Semimicro Experimental Results

No. of Detns.

Calcd. 39.01 19.77 78.34 75.98 30.42

2 2 2 2 2

CsFia CnClaFi

Per Cent Fluorine Found (av.) 38.70 19.59 78.33 76.34 30.44

f 0.44 f 0.02

i= 0 . 0 8 i 0.05 f 0.07

Table 11. JIacro Experimental Results No. of Detns.

Compound

Calod. 39.01 19.77 78.34 75.98 30.42 40.12 31.43 0.

Per Cent Fluorine Found (av.) 38.92 f 0 . 5 6 19.84 f 0.09 76.35 i 0 . 7 5 75.30 f 0 . 1 5 30.68 f 0.51 39.52 i 0 . 0 8 31.23 rt 0 . 6 3 0.015 i 0.013

DISCUSSION

Developed to give a constant indication of the amount of fluorocarbon in a flowing system, the method outlined allows for the analysis of the sample while it is being collected, a feature that is absent in other methods. Inasmuch as the development of the method was done largely with macrosamples of trichlorotrifluoroethane, and no considerable amount of attack of the platinum tube by chlorine was noted even with the large number of macroscale chlorofluorocarbon samples employed, it is seen that the attack of platinum on the semimicro scale is negligible. The larger liquid and gas samples Tyere weighed in the devices shown in Figure 1; the gases were displaced into the nitrogen stream with a constant flow of mercury, secured by means of a constant head above a fine capillary. Results of the macro work are shown in Table 11. (The large errors in these preliminary results are now known to be due a t least in part to the variation

I-

2o

COLORIMETER CALIBRATION FOR FLUORIDE DETERMINATION L E F T TO RIGHT, 2(2