V O L U M E 23, NO. 6, J U N E 1951 in a patient with myelogenous leukemia before, during, and after :tdministration over B period of 25 days (Figure 4). During this whole period the serum arsenic levels, although showing a tendency to increase after arsenic administration, ranged from 3.5 to 7.0 micrograms % ' R hich is within the normal range. The 24-hour urinary arsenic increased from 5.6 micrograms before medication to 500 during medication, and was still elevated (30 mirrogininq) 10 days after the lnst administration of areenic. LITERATURE CITED
(1) Boltz, D. F., and Mellon, RI. G., ANAL.('HEY., 19,873(1947). (2) Case, 0.P., Ibid., 20,902 (1948). (3) Chaney, A. Id., and Magnuson, H. J., IXD. Evc. CHEW,. l s a r , . ED.,
12,691 (1940).
~ 4 )Chaney, A4. L.,
and Magnuson, H. J.. private communication. 15) Gorgy, S., Rakestraw, N. W.. and Fox, D. L., J.M n i i n c Research 6&is
Foundation). 7,22 f194R).
919
IND. E x . CHEM.,BSAL.ED., 13,915 (1941). (7) Hunter, F. T.,Kip, A. F., and Irvine, J. W., Jr., J. Pho.rmacoZ. Ezptl. Therap., 76,207,221 (1942). (8) hfaechling, E. H., and Flinn, F. B., J . Lab. Clin. Med., 15, 779 (1930). (9) hfagnuson, H.J., and IVatson, E. B., IND. ENQ.CHEX.,A s . \ i . . ED., 16,339(1944). (10) hiaren, T. H., Ibid., 18,521 (1946). (11) Morris, H. J., and Calvery, H. O., Ibid., 9,447 (1937). (12) Reyes, E.A, Rev. quim.farm. (Santiago, Chile),4,2 (1947). (13)Satterlee, H. S.,and Blodgett, G., IND.ENO.CHEM.,A N . ~ I XI)., .. 16,400(1944). (14) Sultaaberger,J.A.,Ibid., 15,408(1943). (6) Hubbard, D. bl.,
RECEIVED September
15, 1950. Presented before t h e Southern Califoriliu Section of the American Association of Clinical Chemists, Loa Angeles. Calif., M s y 1950. Reviewed by the Veterans Administration and published with the spproval of the chief medical director. T h e statements and conclusions of the arithors s r e t h e result of their own s t u d y and do not. new.. sarily rrflcct the opinion or policy of the Vetersns Administration.
Microdetermination of Fluorine in Solid Halocarbons H. R . RICKARD, FRANCES L. BALL,
AND
W. W. HARRIS
h--45 Plant, Carbide and Carbon Cheniicals Dieision, Union Carbide wid Carbon Corp., Oak Ridge, Tenn.
During investigation of the structures of some laboratory-synthesized halocarbons, the need arose for rapid microanalytical determination of fluorine in .solid compounds which contained fluorine, carbon, chlorine, and nitrogen, and, particularly, for a method that could be used in the presence of nitrogen. A method for the microdetermination of fluorine in solid organic compounds containing fluorine, chlorine, bromine, and nitrogen in addition to carbon has been developed. By decomposing the sample at 1100" C. in a stream of moist
T
H E synthesis of high boiling halocarbons required pro-
cedures for ultimate determination of carbon, chlorine, fluorine, nitrogen, sulfur, and hydrogen. Of the available proredurcs, those for fluorine were the least satisfactory with rei p w t to accuracy, precision, and time required for a singlc determination. This paper describes a rapid method for the dcterniination of fluorine in solid halocarbons. There are two aspects of the problem of fluorine determination in halocarbons4ecomposition of the sample, and collection and determination of the liberated fluorine. PREVIOUS WORK
Sample Decomposition. The methods of fluorocarbon deromposition have been reviewed by Elving and Ligett ( 4 ) .
Halocarbons containing fluorine can be quantitatively decomposed by prolonged digestion with sodium and liquid ammonia a t room temperature in sealed glass reaction tubes (13) or at 400"C. in an evacuated tube in much shorter periods of time ( 4 ) . Kiniball and Tufts ( I f ) found the sodium and liquid ammonia treatment insufficient for many halocarbons and recommended a bomb fusion with potassium a t 550" C. Teston and hfcKenna ( 2 0 )used a direct combustion technique in which the sample was burned in a stream of oxygen a t 1000" C. in a quartz combustion tube packed with platinum and crushed quartz. Fluorine reacted with the quartz to form silicon tetrafluoride which was sorbed on activated alumina a t 175' C. The exact conditions necessary for quantitative sorption of silicon tetrafluoride on alumina have not been definitely established. Miller and McBee (14)obtained quantitative sorption a t 650" C. Horton and Kirslis (9) studied the reaction of silicon tetrafluoride on alumina a t various temperatures and found that quantitative -orption requires careful preparation of the alumina, Schumb and Radimer (18) described a Combustion twhnique,
oxygen anddetermining the fluorine colorinietricallj with a standard ferric ion-salicylic acid solution, 200 to 1500 micrograms of fluowine can be determined with an average deviation of from 1 to 2Yo a precision of 3.9Yo of the fluorine present on a 95Yo confidence interval. .4 determination of fluorine can be made in 30 minutes in most instances. The use of quartz, platinized quartz, or platinuni combustion tubes in the pyrohydrolysis of halocarbons caused no significant differences in the amounts of fluorine that were recovered.
appliciible to volatile organic compounds, depending upon psi ohydrolysis of the sample in a stream of moist oxygen in a platinum combustion tube a t 1100" C . Pvlilner (16) modified this procedure to permit analyses of volatile and nonvolatile eompoundc Gaseous halocarbons have been quantitatively decomposed using stcam by Cline and Kestbrook ( 3 ) . Fluorine Determination. ?Jethods for determining fluoridc have been reviewed recently by Rinck (16).
Fluorine may be determined gravimetricdly rn lead chlorofluoride ( I S ) ,volumetrically using thorium nitrate (21 ), condnctometrically ( 8 ) , amperometrically ( l a ) , or colorimetrically using the bleaching effect of fluoride ion on metallo-organic complexes w c h as ferric ion-salicylir acid (6-7, 19). The moderakly high solubility of lead chlorofluoride makes t L T use of this compound unsatisfactory for microdetermination. The thorium nitrate titration using alizarin red S as an indicato~ is too insensitive for microgram amounts of fluoride. The fluorescent titration of Horton (10) can be used in the low microgranl region, but requires pure morin which was not available a t the. time of this study. Langer (12) has titrated fluoride amperometrically using thorium nitrate in concentrations as low as 5 micrograms per ml. Colorimetric procedures employing various iron complexes arc extremely sensitive to fluoride when a spectrophotometer i. used to measure color density-for example, the method described by Greenspan and Stein ( 7 , 1 7 ) is usable in the fluoride concentration range of 2 to 50 micrograms per ml. Fluoride bleaches the wine-colored ferric ion-salicylic acid solution, Thib hleaching action is caused by the preferential formation of ,901uble fluoride complexes with ferric ions which no longer show the color reactions of ferric ions alone. EXPERILMENTAL
The experience in this laboratory with sodium or potassium fusions in h m h q hac bpcn th:it not all fluorocarfinns ran he coni-
920
ANALYTICAL CHEMISTRY
pletely decomposed without damaging the bombs and necessihas t,&tingfrequent replacement. ~~~i~~ with sodium resulted in explosions. Of the procedures reviepved above, the method of decomposition used by Cline and Westbrook ( 3 ) in combination with the Greenspan and Stein ( 7 ) method of determination appeared most applicable to the compounds encountered.
Figure 1. Pyrohydrolysis Assembly
Reagents. Sodium fluoride solution, 100 micrograms per ml. Ferric ammonium sulfate, 0.007 Jf solution. Salicylic acid, 0.01 Jf solution. Hydrochloric acid, pH 3.1. Apparatus. pyrohydrolysis train as shown in Figure 1. The mercury bubbler assembly and a filling funnel were connected to the steam generator, which consist,ed of a round-bottomed 1-liter flask. h 500-ml. steam trap was placed b e t n w n the steam generator and the combustion tube. The com1,uation tube !vas approximately 60 cm. long and 10 mm. in inside diameter and was made of quartz, platinized quartz, or platinum. The portion of the tube which was within the combustion furnace was packed with platinum stars. The combustion furnace consisted of a steel shell lined with firebrick and ash s t o s , through which four 44 X 1 cm. globar rods were placed. h 220-volt power source gave an efficient operating temperature of 1100" The fluoride, which was absorbed in a borosilicate glass rec.eivc,r joined to the combustion tube by a borosilicate glass de]ivc,ry tip, \yas determined Lvith a Beckman l\Iode] D U spectraphotometer using 1-cm. Corex cells. 811 weighings were made a i t h an Ainsworth keyboard microbalance. Colorimetric Procedure. Fluorine was determined by measuririg the decrease in absorbancy of a standard ferric ion-salicylic acid solution a t 530 mp. This solution was made by adding 80 ml. of the ferric ammonium sulfate solution t o 90 ml. of the salicylic acid solution and diluting t o 1 liter rvith the hydrochloric acid solution. The final pH was 3.1 =t0.2 and this stock solution was alloived to stand in a dark bottle for 24 hours before use. Greenspan and Stein ( 7 ) have shown that the absorbancy of the ferric ion-salicylic acid solution is a maximum a t p H 2.9 to 3.2 and there is a very rapid decrease in absorbancy on either side of this range. In practice it has been found that maintaining the pH of thP reference solution and of the sample within this range is the most critical aspect of the determination. A$
c.
The absorbancy of the stock solution compared x i t h that of distilled water a t 530 mp was between 0.820 and 0.880 and remained constant for a t least a week. Figure 2 shows characteristic absorption curves of a ferric ion-salicylic acid solution for various fluoride concentrations measured against distilled water. Each new stock solution requires a calibration curve, as it was found difficult to prepare different solutions with the same absorbancy. The slopes of the curves are the same, however. This may be due to variations in the purity of ferric ammonium sulfate lots. I n contrast to most spectrophotometric methods, the sample solution had a lower absorbancy than did the reference solution. Therefore, the fluoride solution was set to zero absorbancy on the spectrophotometer and the absorbancy of the reference solution was measured against it. The system dpes not follow a linear relationship between absoT.banc\ and fluoride concentration.
Analytical Procedure. Organic samples weighing from 1 t o 10 mg. were weighed int,o a platinum micro combustion boat, tube which Was then placed in the 7.5 cm. (3 inches) from the furnace entrance. The receiver containing 25 of stoclc solution ,vas then mounted on the system, an ice bath was placed around the receiver, and then steam was permitted to flow through the combustion tube. The rate of steam flox was maintained a t approximately 1 gram per minute by the mercury bubbler, but, could be adjusted by varying the amount of mercury in the bubbler. Heat from a burner was then cautiously applied 5 cm. (2 inches) in front of the sample and as the sample hydrolyzed the burner was moved toward the furnace entrance. The burning cycle was repeated to ensure complete sample decomposition. The receiver \vas removed from the system and the solution cooled to room temperature. It \vas then transferred to a 100-nil. volumetric flask and diluted to volume with the pH 3.1 hydrochloric acid solution. The absorbancy of this solution was considered to be zero and the absorbancy of the reagcnt blank (15 ml. of stock solution in a 25-ml. volumetric flask, made to volume lvith the p H 3.1 hydrochloric acid) was measured against the sample solution a t 530 mp. Fluoride ion concentration was obtained from a calibration curve. The ferric ion concentrations in t,he reagent blank and the sample solution differ. The ratios chosen need not be identical, so long as a calibration curve is used, and it was found that the ratio used here made the colorimetric determination more sensitive, The time for one determination about 20 minutes, but varied dependirlg upon the ease of sample decomposition. Presentation of Data. Data for the fluorine determination in seven solid halocarbons, prepared by the Fluorocarbon Section of this laboratory, are presented in ~ ~ 1, b l ~
Table I. ComPound CloHisNOzFz
Determination of Fluorine in Solid Samples Tube
Platinum Platinized quartz Quartz
(CF2CFC1)n
Platinized quartz 'Iatinum
CIH,,SOCIF
Platinized quartz
Quartz
C1eH2iNozFz P1atinizedquartz C ~ N ~ S O Z F Z Platinized quartz c ~ H , ~ o ~ F Platinized ~ quartz
iyo, of Detns. 10 12 10
F l u~orine, % Thee. Found 17.1 17.3 17.3 17.3 17.3 17.2
AX.. Dev., % i0.3 10.3 -0.3 iO.5
9 9 9
49.1 49.1 49.1
49.1 49.6 48.9
12,3 21.3
10.2
3 3
12.' 21.5 27.3
28.3
10.1
27.3 10.1 10.1
27 8 a 10.0 10.3
10.2 10.4
10.5 i.0 5
10.4
CaHloOsB4 3
C8H7NoC1F Platinized quartz 3 a Clark hIicroanalytica1 Laboratory, Urbana, Ill.
10.3
T o test the effect of different combustion tube materials, determinations were made on two different compounds using a quartz tube, a platinized quartz tube, and a platinum tube. Statistical analysis of variances and covariances ( 1 ) of the data presented in Table I showed no significant difference between any of the combustion tube materials in the fluoride analysis. Accordingly, the data for each compound were considered t o have been obtained from the same combustion tube and calculations to establish the precision of the method were based upon all the determinations for each compound taken as one set. The data obtained are presented in Table 11. DISCUSSION
Pyrohydrolysis of Sample. The data in Table I show that quartz, platinized quartz, or platinum combustion tubes may be used without causing significant difference in fluoride recovery However, it is desirable to use a quartz tube partially platinized
V O L U M E 23, N O . 6, J U N E 1 9 5 1 Table 11.
92 1
Precision of Replicate Determinations Precision (95% Confidence Interval), 7 c Single detns. Group mean!
Compound No. of Detns. C~H~SNO~FZ 32 (CF2CFCl)n 28
0.68
0.12 0 31
1 61
0 A -0.25
-0.50
’!-,I
Interferences. Elements which act as reducing agents after passing through the system may reduce ferric ions to ferrous and act as interferences. Sulfur acts probably as a reducing agent, causing an interference. Phosphorus compounds 4 ere not investigated, but phosphate is known to complex the ferric ion. K h e n chlorine, bromine, and nitrogen Tvere present in the organic compounds no interference was observed. Alkali and alkaline earth fluorides formed by hydrolysis of metallo-organic compounds are not completely decomposed by pyroh) drolysis (8,17, p. 239) and, therefore, will interfere n ith this method n ithout modification. The modification consists in bedding the sample in the combustion boat with uranourmic oxide (U808)(3). Other colored complexes t h a t are bleached by fluoride, such as ferric-feeron ( 5 ) , ferric-thiocyanate (6). and titanic acid (19) may, in some cases, be substituted for ferric ion-salicylic acid \There specific interferences are presrnt. ACKNOWLEDGMEXT
I
-075
m
The authors are indebted to Olive K a r r a m for the statistical treatment of the data, J. D. Gibson for the research samples, and R. H. Lafferty, Jr., and 1,. H. Rogers for their revien. of €he manuscript.
4 CI
2
-1.00
m a 0
9
-1.25
LITERATURE CITED
(1) Brownlee, K. -L, “Industrial Experimentation,” 2nd rev. Ameri-
can ed., Brooklyn, Chemical Publishing Co., 1948.
-1.50
(2) Cline, \T. D., and Warf, J. C., Iowa State College, R e p t .
CC-2723,1945. (3) Cline, W. D., and Westbrook, J. A , , Carbide and Carbon Chemicals Corp., K-25 Plant, Oak Ridge, Tenn., R e p t . K-262,1948. (4) Elvine. P. J., and Ligett. lv. B., IND.ESG. CHEM.,ANAL.ED.,
-1.75
14, 449-63 (1941).
- 2.00 380
I 400
I
I
I
I
440 480 520 560 W A V E L E N G T H , MILLIMICRONS
I 600
Figure 2. Characteristic Spectral Absorption Curves of Ferric Ion-Salicylic Acid Solution Various fluoride concentrations measured against di9iilled water A . 1 X 10-4 mole of fluoride per liter B . 3 X 10-4 mole of fluoride per liter C. 6 X 10-4 mole of fluoride per liter
so that the preburn may be observed as samples must be vaporized slowly t o ensure complete deconiposition during passage through the combustion tube. Some samples may hydrolyze more readily than others; h e w e each must be burned slowly A4carbonaceous refiidue in the receiver is considered evidence. of incomplete combustion. To ensure complete sample decomposition, any residue remaining in the platinum boat after the first burn was heated to a red heat again, but the ferric ion-salicylic acid solution was kept below boiling temperature to prevent its decomposition. Liquid halocarbons are difficult to control during the vaporization step in the heating cycle, because they distill through the systr’m too rapidly.
Fahey, J. J., I b i d . , 11, 362-3 (1939). Foster, 11.D., J . d i n . Chem. Soc., 54, 4464-5 (1932). Greenspan, J., and Stein, S.J., Kellex Corp., R e p t . KZ-891, 1944. Harms, J., and Jander, G., 2 . Elektrochcm., 42,315-19 (1936). (9) Horton, A. D., and Kirslis. 9. S., Carbide and Carbon Chemicals Corp., K-25 Plant, Oak Ridge, Tenn., Re@. K-372,1949. Ph.D. thesis, University of hIichigan, 1949. ., and Tufts. L. E.. -&xu..CHEM.,19, 150-3
(5) (6) (7) (8)
(1947). (12) Langer. ,I.,IND.Esc. (‘HEM., As.AI.. ED., 1 (13) hliller, J. F., Hunt, H., and JIcBee, E. T
148-9 (1947). (14) Miller. J. F.. and McBee, E. T., Purdue University, R e p t . A-1515, 25-9, 1944. (15) RIilner. 0. I., AS.^.. ( ‘ H E M . , 22,815-17 (1950). (16) Rinck, E., Rirli. SOC. ch1’7!2., F r a n c c , 1948, 305-24. (17) Rodden, C‘. J., “Analytical C‘heniistry of the Manhattan Project,’’p. 241, Kew York. AlcGraw-Hill Book Co., 1950. (18) Schumh. W.C., and Radimer, K. J., A x . 4 ~ .C H m r . , 20, 871-3 (1948). (19) Steiger, G., J . .-lnz. Chrm. Soc., 30,219-25 (1908). (20) Teston, R. O’D., and AIcKenna, E’. E.. . h - . i x . . CHmr., 19, 192-6 (1947). (21) Killard, H. H., and \Tinter, 0. R., ISD.ENG.CHEM.,AN.\I.. ED., 5, 7-10 (1933). RECEIVED July 3, 19.50. Work performed at the K-25Plant of Carbide and Carbon Chemicals Division, Union Carbide and Carbon Corp., Oak Ridge, Tenn., under contract with t h e Atomic Energy Commission.