( 4 ) Beukenkanip, J., Rieman, \T7., A x . 4 ~ CHEX . 2 2 , 5 8 2 (1950).
111,
( 5 ) Bird, P . G., U. S. Patent 2,244,325
(1941).
( 6 ) Cale)., E. R., Foulk, C. K., J . -4m. Chenz. SOC.51, 1661 (1929). (i) 13 Malo, T.'. ( t o lIonsanto Chemical Co.), U.8. Patent 2,515,949(19.50). (8) Emblem. H G., C'loherty, D. J. (to
Rolls-Ru\
w-
Limited) Can. Patent
544,022 ( l % o r) (9) H,urd, C. B.. Raj-mond, C. L., Miller, P b., J . Phys. Chem. 38,663 (1934).
S. (to E. I. du Pont de Kernours B: Co.), U. S. Patents 2,395,880,2,408,654,2,408,656 (1946). (11) Kirk, J. S., Iler, R. K. (to E. I. du Pont de Semours B: Co.), Ibid., 2,408,655 (1946). (12) llarshall, ill. D. (to llonsanto Chemical Co.), Zbid., 2,515,960(1950). (13) Xational Aluminate Corp., Brit. Patent, 611,914 (1948). (14) Robinson, J. IT., Jr. ( t o E. I. du Pont de Nemours & Co.), U. S. Patent 2,392,767(1916).
(10) Kirk, J.
(15) Rule, J. RI. (to E. I. du Pont de Xemours B: Co.), Ibid., 2,577,484
(1950). (16) White, J. F. (to Monsanto Chemical Co.), Ibid., 2,285,477(1942). (17) Wolter, F. J. (to E. I. du Pont de Tiemours 8- Co.), Ibid., 2,601,352 (1950).
RECEIVEDfor review June 6, 1950. Accepted November 5, 1950.
Rapid Determination of Fluoride in Silica-Alumina Catalyst by Steam Hydrolysis LEON W. GAMBLE, WALTER E. PRICE,l and WILLIAM H. JONES Esso Research laboratories, Louisiana Division, Esso Standard Oil Co., Baton Rouge, l a . b The steam hydrolysis or pyrohydrolysis method of Warf, Cline, and Tevebaugh has been adapted to the rapid determination of fluorine in silica-alumina catalyst and other silicabearing materials. Quartz or Vycor has been substituted for the more expensive platinum apparatus of other workers. The sample i s hydrolyzed b y steam in a horizontal quartz tube electrically heated to 760" C. and the liberated hydrogen fluoride i s collected in the condensate. The fluoride solution i s then titrated with standard thorium nitrate in the conventional manner, using sodium alizarin sulfonate as a color indicator. Analyses b y four laboratories show that the accuracy of the steaming method is equivalent to that of the conventional chemical method-to about 1% of the average value-and its precision i s better. These data were reproducible to within about 6 to 9% o f the average value a t the 95% confidence level.
D
the past few years, the determination of fluoride in silicaalumina catalysts and other aluminosilicates lias been of interest to the petroleum ( I ) and other industries. Simple and rapid analysis, required for control purposes, is niade possible by the steam hydrolysis or pyrohydrolysis method of Karf, Cline, Tevebaugh, et a2. (2-4, 6, 9, I O ) , which employs the principle of decomposition of the fluoride with superheated steam t o form the oxide of the metal and volatile hydrogen fluoride. This is a n improvement over the classical chemical method of Willard and Kinter (11). which involves distillation after direct acid decomposition. or 1
URISG
Ileceased.
acid decomposition follou iiig a fusion with sodium carbonate. I n this procedure, large amounts of alumina and silica retard the volatilization of fluorine. This interference has been eliminated by Hoffman and Lundell's tn o-step precipitation (5) and more recently by Cliu and Shafer's one-step precipitation ( I ) . Hoffman and Lundell decompose the sample b y carbonate fusion. The bulk of the silica and alumina is then precipitated by ammonium carbonate and the traces are removed b y ammoniacal zinc oxide. I n a simplified procedure, Chu and Shafter fuse the sample v-ith sodium peroxide, and after dissolution in nater, precipitate the silicates and aluminates with zinc sulfate. The fluoride is distilled arid titrated in the usual manner. Fusion is necessary if the alumina-silica ratio is greater than 1 to 3. Many fluorides are amenable to the siiiiple method of steam hydrolysis. This requires only 20 to 40 minutes, in contrast with 10 to 12 hours b y the usual chemical procedure. However, because the pyrohydrolytic method is not universally applicable, the qualitative composition of the material being analyzed must be known. The method should not be used with uriknonn materials. The limitations of the steaming method are discussed below under details of the procedure. I n case of doubt, the usual chemical method should be resorted to. Briefly, the pyrohydrolytic method consists of steaming the fluoride sample in a horizontal quartz tube electrically heated to '760" C. The fluoride is hydrolyzed by the superheated steam and the liberated HF (or more accurately fluosilicic acid) passes over into the condensate. The fluoride solution is then titrated with standard thorium nitrate in the conventional manner, with sodium alizarip sulfonate
Table
I.
Residual Fluorine Steaming
after
7cFluorine Snmule
Removed hv
S
steaming 0 5ii
1
Residual after steaming
Table II. Blank Fluorine Determinations on Pure Inorganic Compounds
Compound +12(
so,
)3
h a?SO*
ILP04 P?06
Appa1r.nt ut. 5F 0 100, 0 09.5 0 010, 0 030 0 0.3 4 0
as a color indicator. Earlier workers used quartz-platinum (3, I O ) . nickelstainless steel ( I O ) , or all-nickel ( 3 ) ;but the technique described in this paper employs a n all-quartz or Vycor system. This system is trouble-free and economical to construct. The platinum or nickel apparatus may be preferred for nonsiliceous materials. DESCRIPTION OF APPARATUS
The apparatus is shonii in Figure 1. The reaction tube is a quartz combustion tube 1 inch in diameter and 30 inches long, nrapped with about 12 feet of No. 26 Kichronie wire and then insulated with thick asbestos. The unit is heated by 110-volt current controlled by a rheostat. The steam is preheated in a superheater constructed from a modified borosilicate glass spray trap also wrapped with Tichrome wire and heated electrically. The exit steam is condensed in a nater-jacketed conVOL. 32, NO. 2, FEBRUARY 1960
* 189
D
';!:
WB
g o yt' 0
1
-d+
4-7
-
J1 % . ~---=
DETAIL VIEW OF STEAM INLET O l 0
I
~-
20
_-___J
40 60 80 T I M E , MINUTES
100
Figure 2. Removal of fluorine from silica-alumina catalyst by steaming Figure 1. Apparatus for fluorine determination by steamhydrolysis A. B. C. D.
E.
F.
Steam inlet Steam superheater Steam inlet to quartz tube Quartz tube Asbestos insulation
G.
H. 1.
Wt. 7 0 F 0.2-0.5 0.5-1.0 1-5
denser, and the distillate is caught in a series of small borosilicate glass beakers. The sample buat is the usual porcelain type. The temperature is measured by a thermocouple.
Grams of Sample 2.5-3.0 1.5-2.5 0.5-1
The steam rate is adjusted to yield a condensate of approximately 4 to 5 CC. per minute for about 40 minutes. The condensate, collected in a small beaker, is kept slightly alkaline to p H paper by the dropwise addition of 2% sodium hydroxide. For more rapid analyses of samples containing less than 1% fluorine, the effluent maybe collected a t short
EXPERIMENTAL
The sample is weighed into a porcelain boat and inserted into the quartz tube in which the temperature has been adjusted to 760' i 40" C. The sample size is varied according to the fluorine content:
Table 111.
Steamed In Porcelain boot A Porcelain boat H Porceloinboot X Fluid unit
0.7-mm. i.d. 1 8 / 7 spherical joint Steam inlet to quortz tube Asbestos insulation
(Silica-alumina catalyst) Metal Fluoride CaF2
yc Fluorine on Cat. 2 19
Steaming Time, Min. at 1400' F. 20 40
60
80 100 120 SaF
.41FI
Table IV.
0.51
20
0.41
20
40
40
TTt. yo Fluorine, Removed by Steaming
On cat. 0 35 0 53 0 69 0.80 0.88 0.97 0.47
0.51 0.39 0.42
On total fluorine 16
24 32
37 40
44 92
100 95 100
Analysis of Other Inorganic Materials Containing Fluorine
Sample N.B.S. opal glass (No. 19) (pop-dered)
Wt. Yo Fluorine Chemical Steaming method method 5.64 1.26 (75 min.) 5.60
accepted Value 5.72
5.68
Mineral cryolite 3 S a F . AlF, (through 200 mesh) Mineral topaz ill(OH, F),SO, (through 200 mesh)
1%
ANALYTICAL CHfMlSTRY
54.0 53.5
22 (40 min.) 26 (80 min.)
54.4
14.1
0 . 0 (40 min.) 0.0
Variable
14.2
H10, MI./Min.
Temp., C.
6
540 650 760 480
3 3 3
20
6 6 2
'
intervals and titrated in increments in order to establish the optimum time. The condensate is titrated in the conventional manner with 0.05N thorium nitrate, using sodium alizarin sulfonate as an indicator. The thorium solution is standardized against a solution of sodium fluoride containing 1 mg. of fluorine per ml. T o establish optimum conditions for the steaming period, the following points were investigated: 1. Temperature range 2. Steam partial pressure
3. Residual fluorine after steaming 4. Effect of chemical combination of fluorine
Analysis of Catalysts Containing Metal Fluorides
.-ldded
cat. Wt., G.
10 to 20
Temperature and Steam Partial Pressure. I n a study of the effect of temperature a t a given steam partial pressure, several tests n-ere carried out on catalysts a t a steam rate of 4 to 6 ml. per minute. Three temperatures, 540°, 650", and 760' C., were used and the fluorine removed after yarious time intervals was determined. T h e results, plotted in Figure 2, show that practically all the fluorine can be removed in 20 minutes a t 650' C., or higher, from a catalyst containing about 1% fluorine. PoFell and Menis (Y),using a stream of moist oxygen for pyrolysis, found that with tungstic oxide as an accelerator fluoride is completely recovered from sodium fluoride at 650" C. in 30 minutes. Thus, silica-alumina catalyst is an accelerator for the pyrohydrolysis of fluoride and is equivalent to tungstic oxide or uranium oxide. Tungstic oxide is approximately 16 times as expensive as silica-alumina catalyst. At 540" C., however, 80 minutes are necessary for the complete removal of fluorine. To study the effect of steam partial pressure and gas velocity, a limited
I
1
CAT. t 0.41% A I F 3
O O t
20
TIME
0
Figure 3. Removal of fluorine from catalyst by steam in a fluidized bed
. X
0.5
1 .8
A
1.0
0
1.0 #
2.0
HeO, MI./Min. 0.5
20-gram catalyst charge N2, Vel., 1./Min Ft./Sec. 0 0.2
O
c.
1.2
0.8 0.4 0.8
480 480 480 480
0
0.8
480
amount of data was obtained in fluidized solids bed using steam-nitrogen mixture. The data shown in Figure 3 indicate the following:
Residual Fluorine after Steaming.
Several samples were analyzed after steaming. T h e d a t a in Table I sholv t h a t approximately 0.05y0 fluorine remains on t h e catalyst after steaming. The error introduced by the incomplete removal of fluorine in the analysis of samples containing 0.5% or more would be relatively small; at concentrations below 0.5%, it would be appreciable.
1
120
100
40 60 80 TIME, MINUTES
Figure 4. Removal of fluorine from silica-alumina catalysts containing metal fluorides 0.5- to 3-gram sample, 6 ml. HzO/min., 760' C., porcelain boats
Table V.
Accuracy and Precision of Fluorine Analyses
Lab. 1
Average Synthetic Average minus synthetic Standard deviation 2
Average Synthetic Average minus synthetic Standard deviation
yo Fluorine Sample A Sample B Chemical Steaming Chemical Steaming 0.50 0.52 0.53 0.517
0.50 0.48 0.52 0.500
0.68 0.72 0.70 0.700
0.506 $0.011 -0.006 &0.013 xt0.017 0.475 0,473 0.510 0.511 0.467 0.496 0.484 0.493
*0:01i 0.890 0.885 0.888 0.888
0.506 -0.013 3zO.021 10.015 0.49 0.51 0.52 0.51 0.47 0.51
1 0 :006 0.85 0.83 0.81
0.82 0.77 0.81 0.800
*o:
022 0,850 0.851 0.839 0.847
-0,022 3
Average
0.493
k0: 006 0.82 0.78 0.81 0.81 0.805
0.510
0.830
iO:bl6 0.96 0.97 1.01 0.980
161615
Average
0,506 -0.013 $0.004 f0.021 fO.OOO 0.48 0.47 0.50 0.50 0.54 0.50 0.507 0.490
Synthetic value Average minus synthetic Standard deviation
$0.001 k0.025
+o:
=kO:016
Synthetic Average minus synthetic Standard deviation 4
Blanks with Nonfluoride Materials.
Certain anions such a s Po4 and so4 m a y c a w e varying errors in blank titrations. For this reason blank steaming trials were made on pure sulfate a n d phosphate salts ( 5 grams) t o determine if a n y interfering anion is carried over by steam hydrolysis. T h e apparent fluorine titrated is shomn in Table 11. The sulfates and phosphates do not hydrolyze, but free phosphoric acid volatilizes and seriously interferes. A number of anions interfere with the thorium nitrate titration method (11).
20
Temp.,
.
The rate of fluorine removal increases with steam partial pressure up to about 50 mole 70, above which additional steam partial pressure is without effect. The rate of fluorine removal also increases with velocity of fluidizing gas and temperature. The removal of fluorine from a fluidized solids bed at 480" C. agrees very well with steaming in a porcelain boat a t 540" C. Comparable data for a porcelain boat a t 480" C. were not obtained.
CAT. t 2.19 % C a F 2
j
, MINUTES
1
i
1
I
0.506 -0.016 f0.014
022
0.94 0.91 0.91 0.920
Statistical Evaluation of All Analyses Average Synthetic Average minus synthetic Standard deviation 9570 confidence limits
0.500
0.498
0.849
0.506 -0.008 10.023 +0.016 & O . 046 +O .032
*o:i15 f O ,230
0.840
-0,006
VOL. 32,
10:049 fO ,098
NO. 2, FEBRUARY 1960
191
Analyses of Catalysts Containing Metal Fluorides. Samples of silicaalumina catalyst impregnated and/or mixed with different metal fluorides have been analyzed b y t h e steaming method for different steaming times a t 760" C. (Table 111). These d a t a show t h a t both sodium and aluminum fluorides release fluorine rapidly and quantitatively. However, calcium fluoride releases its fluorine very slon ly, for one half being evolved in 2 hours. Analysis of Other Inorganic MaTKO terials Containing Fluoride. minerals, topaz and cryolite, containing fluorine, and a sample of opal glass were analyzed b y both t h e chemical a n d the steaming methods. T h e d a t a in Table I V reveal t h a t the fluorine is not so readily released from these materials as from aluminum and sodium fluorides. T h e fluorine retention is similar t o t h a t of calcium fluoride. Fluorides in general are divided into two classes: rapidly hydrolyzable and slowly hydrolyzable. However, U308 acts as an accelerator and slowly hydrolyzable fluorides may be analyzed more rapidly by use of this compound (10). Recently Silverman and Bowen
(8) were able to pyrohydrolyze cryolite
at 1200" C. using an all-nickel reactor. A 3 to 1 ratio of alumina to cryolite was necessary for complete fluoride recovery. The steaming method is a rapid and valuable tool for the analysis of some materials, but the longer chemical method should be used for unknown or unusual samples. ACCURACY A N D PRECISION
I n Table V are given data from a program in which four laboratories tested both chemical and steaming methods on a synthetic sample and on an unknown sample. The t-wo methods are equally accurate but the steaming method is more precise. ACKNOWLEDGMENT
The authors gratefully acknowledge the assistance of E. J. Kewchurch, J. S. McIlhenny, and Dorothy Webb, and thank the following organizations, for participation in the accuracy and precision tests of Table V: Technical and Research Division, Humble Oil 8: Refining Co., and Products Research and
Process Research Divisions, Esso Research and Engineering Co. LITERATURE CITED
(1) Chu, C.-C., Shafer, J. L., -1s.4~.CHEU. 27, 1429-31 (1955). ( 2 ) Cline, IT. D., Tevebaugh, R; D., Warf, J. C., Method Abstract, Analytical Chemistry of the Manhattan Project," ed. by C. J. Rodden, National Kuclear Energy Series, Vol. VIII-I, p. 239, McGraw-Hill, Ken. York, 1950. (3) Gahler, 8.R., Porter, Galen, ASAL. CHEX 29,296-8 (195i). ( 4 ) Haff, L. V., Butler, C. P., Bisso, J. I).,
Ibid., 30,984 (1958). (5) Hoffman, J. I., Lundell, G. E. F., Bur. Standards J . Research 3, 581 (1929). (6) Lee, J. E., Edgerton, J. H., Kelley, 11,T., A S A L . CHEJI. 28, 1441 (1986). ( 7 ) Powell, R. H., Menis, O., Ibid.,30, 1546 (1958). (8) Silverman, H. P., B o ~ e n F. , J., Ibid., 31, 1960 (1959). (9) Susano, C. D., Khite, J. C., Lee, J. E., Jr., Ibid.,27, 453 (1955). (10) Warf, J. C., Cline, TI-. D., Tevebaugh, R. D., Ibid., 26,312 (1954). ( l l ) W i l l a r d , H. H., Kinter, 0. B., IND.EXG. CHEX., ASAL. ED. 5 , 7 (1933). RECEIVEDfor review May 27, 1959. Acceoted Kovember 2. 1959. Division of Ana&tical Chemistri, 131st AIeeting, .1CP, Miami, Fla., April 1957,
Analytical Assay of Diosgenin STEPHEN KAUFMANN, J. C. MEDINA, and CLAUD10 ZAPATA Syntex, S. A., Mexico City, Mexico
b
Fractional sublimation is applied to the assay of crude diosgenin. The results obtained by this technique are evaluated by comparison with those obtained using two other well-known analytical methods: column chromatography and countercurrent distribution. A good agreement was obtained between results, indicating a similar accuracy for the three methods. An estimate of the reproducibility of the sublimation method gives statistical control limits within f0.34%. Because of its accuracy, precision, and simplicity, the fractional sublimation technique seems promising for use in the ndustrial quality control of diosgenin.
D
is the starting material for many steroid hormones. The yields and quality of the steroids derived from diosgenin depend to a considerable extent on the purity of the raw material. Thus, analytical procedures for assaying technical grade diosgenin are important in the production control of the steroid industry based on this sapogenin. 192
IOSGESIN
ANALYTICAL CHEMISTRY
A seemingly simple spectrophotometric method (6) for the identification and estimation of steroidal sapogenins is based on the absorption curves of the sulfuric acid chromogens originally reported by Diaz, Zaffaroni, Rosenkranz, and Djerassi ( 1 ) . This method was unsatisfactory because the impurities in technical grade diosgenin react with sulfuric acid to produce a color which interferes in the determination. Also this procedure (6) is time-consuning and difficult to apply to crude materials. Sublimation has long been recognized as a valuable technique for the isolation of pure organic compounds. I n some instances it has definite advantages over other purification methods ( 4 ) . The excellent general reviery of the topic by Tipsoii ( 5 ) provides some striking examples of its application. V h e n applied to sapogenins, sublimation of glycosides easily permits the isolation of the aglycon, as the sublimate (9). This technique for assaying technical grade diosgenin has been explored in this laboratory and compared n-ith the chromatographic and countercurrent distribution methods. The results are described below.
EXPERIMENTAL
Chromatography. Columns filled with activated alumina oxide are used for adsorption; diosgenin is eluted n i t h several portions of benzene, benzene-ether (8:2, 6:4, 2 : 8 ) , and ether and collected in fractions, nhich are evaporated to dryness. The fractions containing only diosgenin (melting point over 200" C.) are dissolved in chloroform, mixed together, evaporated t o dryness, washed with hexane, and dried to constant weight. This weight is considered as pure diosgenin for calculations. Countercurrent Distribution. A continuous extraction Craig apparatus (manufactured by E d m u n d Buhler, Tubingen, catalog Xo. 30-01-01) consisting of 35 glass tubes is used. The solvent system is composed of 99% heptane and methanol (1 to l ) , heptane being the upper moving phase. The sample is submitted to 45 transfers. Under these conditions, diosgenin has a partition coefficient of 0.667, from which a theoretical distribution curve is calculated. .A comparison of this curve n i t h the experimental curves obtained for each determination shows that all the diosgenin, almost pure, is contained in tubes 10 to 27, whereas the impurities are retained in the other tubes. Thus,
