V O L U M E 2 5 , N O . 1, J A N U A R Y 1 9 5 3
121
COO]to 10’ C. Bdd 6%. CUPferron dropwise with stirring until a 100% or greater escess of the reagent is added. Centrifuge and decant. Neutralize the clear solution with 307, sodium hydroxide and add a IO-ml. excess. .4dd 1ml. of acetone and electroplate for 10 minutes at 7 to 8 volts on a copper-plated SargentSlomin gauze electrode. Wash the electrode with water and o 019 o 018 o 001 5 fi 0 017 dissolve the zinc in 20 ml. of 0 019 1 to 1 sulfuric acid. Bend a 0 018 strip of pure aluminum into a triangle and place it in the solution. Boil for 10 minutes, Filter with a small Diece of aluminum on the filt& paper. Neutralize the filtrate with 307, sodium hydroxide and add a 10-ml. excess. If the solution is not clear, centrifuge and decant. Dilute to 45 ml., add 7 drops of acetone, and transfer to the electrolytic cell. Plate the zinc on the disk cathode for 1.5 minutes a t 4 volts. Determine the specific activity.
Table 11. .4nalysis of .4luminum .4lloys
Sample S B S 86-0
Material AI alloy (casting), 8% Cu,1% Xfn, 0.04% Fe
XBS 85-a
AI alloy (wrought), 2 5% c u , 0 66% bln, 0 2% Fe
SBS Value, 0 Zn 1 50
0 019
Sample Weight, Grams 1.0037 1.0112 1.0009 1.0104
io
5664 1 1 0888 1 1 9084 10 0042
Standard Deviation of Single hleasureSpecific 1 ment, % Activity G.’ Zinc Found, Abso- Relaof Deposit, Counts/ Mg./Counts/ Mean lute tive hlin. value value value Min./Mg. 705 14.18 X IO-’ 1,53 1.50 0.03 2.0 719 13.90 1.53 758 13.06 1.47 742 13.48 1.49
5435 5747 4902 6061
I 84 1 74 3 04 I 65
x
lytic solution. Rinse the cell with water, disassemble, and mash the disk cathode consecutively with water, alcohol, and ether. Dry and weigh as before. Determine the activity of the deposited zinc with the cathode disk in the counter changer and an aluminum absorber (225 mg. per sq. cm.) between the disk and the TGC-3 Geiger tube. -4fter making coincidence and background corrections on the observed activity, calculate the specific activity of the deposit. The specific activity obtained for each concentration of untagged zinc is shown in Table I and Figure 2. The working curve may be readily corrected for any time interval between its preparation and use by making a decay correction for one point on the line, using the known half life of zinc-65, and drawing a straight line through this corrected point and the point where the graph originally crossed the y-axis. Because of the 250-day half life of the zinc, such a correction need be made only once every few days. Determination of Aluminum Alloy Sample. Dissolve the sample (about 1 gram for an alloy containing 1 to Zwo zinc) in 25 ml. of l5Y0 sodium hydroxide by warming on the hot plate for 5 to 10 minutes. Add 30 ml. of water, 30 ml. of 1 to 1 sulfuric acid, and 5 ml. of 37, hydrogen peroxide. Heat to boiling. Decant the clear solution, leaving about 5 ml., which should contain all of the still undissolved residue. Bdd a few drops of concentrated nitric acid to the residue and heat. After the residue has dissolved, combine t.his R ith the decanted solution. -4dd 10 ml. of the radioactive zinc solution and niis thoroughly.
10-4
The results obtained for zinc in two representative aluminum alloys are shown in Table 11. The maximum relative deviation of an observed value from the mean is 2% for the sample containing 1.5% zinc and 5.6% for the sample containing 0.02% zinc. LITERATURE CITED
(1) Beilstein, F., and Jawein, L., Ber., 12, 446 (1879); 2. anal. Chem., 18, 588 (1879). (2) Cohen, A , , Helv. Chim. Acta, 25, 325 (1942). (3) Ihid., 26, 75 (1943). (4) Nickolls, L. C., and Caskin, J. G. K., Analyst, 59, 391 (1934). (5) Osborn, G. H., J . SOC.Chem. Ind. (London), 62, 58 (1943). (6) Smith, E. F., “Electro-.4nalysis,” 6th ed., p. 127, Philadelphia, P. Blakiston’s Son & Co., 1918. (7) Torrance, S., Analyst, 62, 719 (1937). 18) Ibid.. 63. 488 (1938). i9) Winches&, R:, and Yntema, L. F., TND. ENG.CHEM.,ANAL. ED.,9, 254 (1937). RECEIVED for review August 20, 1952.
Sccepted October 4, 1952.
Determination of Purity of Steroids Solubility Analysis WILLIAM TARPLEY AND MILTOX YUDIS Physical Organic C h e m i s t r y D e p a r t m e n t , Scientific Research Division, Schering Corp., Bloomfield, .V. J .
P
ROBLEMS of purity determination in the steroid field have
received little attention in the past, because of the limited quantities of these natural products generally available, and the lack of suitable purity criteria. Steroids are frequently difficult to crystallize, form stable solvates and other complexes, and are often heat labile. Furthermore, ultimate analysis cannot differentiate the many position isomers. Comparison of physical constants with those of known compounds requires frequently unavailable pure reference standards. ;iccordingly, the application of various methods of purification and purity analysis have been under investigation in these laboratories for the pafit several years. Among the classical methods for determining purity are the freezing and melting curves, which have been used very successfully in studies of petroleum products (14). However, such procedures are reliable only for conipounde which are completely
stable under the conditions of study. Melting point methods are frequently inapplicable in the determination of purity of steroids, as certain of these substances are heat labile and give decomposition ranges rather than equilibrium melting points (20). Polymorphic modifications occur frequently in the steroid series (8) and for this reason, phase transitions near the region of the melting point tend to obscure the melting observation. The countercurrent distribution method described b.i- Craig ( 4 ) has found considerable use in purity determination. Although the time required to reach equilibrium for this liquidliquid system is probably less than that for the liquid-solid equilibrium of solubility analysis, the countercurrent method of analysis is limited in that considerable experimentation is required in order to find a solvent system with sufficient selectivity. One of the primary tests for the demonstration of impurity is the successful separation of components by physical means. The
122 Suitable criteria were sought for the estimation of the purity of certain steroids for which the usual melting point determinations are unsuitable because of decomposition, polymorphism, solvation, etc. The significant factors leading to the satisfactory application of solubility analysis to steroid samples have been considered. Examples draw-n from laboratory problems include measurements on cortisone acetate, methylandrostendiol, pregnenolone acetate, pregnenolone, desoxycholic acid, the steroid lactone of Miescher and Fischer, dihydrocortisone acetate, and 4-bromodihydrocortisone acetate. These examples include the preparation of small samples of pure materials for use as standards, the use of solubility analysis to follow purification processes, and the identification of minor components. The solubility method of analysis has been extensively explored and found suitable for quantitatively determining the amount of impurity in steroid samples. The technique has been extended to the identification of unknown steroid impurities, and small quantities of pure steroids have been prepared for reference purposes.
effectiveness of chromatographic separation therefore has suggested ( 3 ) its use as a criterion of purity. This technique has been invaluable in this laboratory for the qualitative detection of impurities, but chromatography also requires extensive preliminary studies of both solvent and absorbent, and quantitative estimation of impurities was found to be inadequate. I n the light of the limitations of the above methods for determining the purity of steroids, the solubility criteria of purity first described by Northrop and Kunitz ( I S ) suggested a useful method for the analysis of steroids. This technique had also been used for estimating the purity of other substances of biological interest (5, 6, 12, I S ) . The solubility method of analysis is applicable to all species of molecules; knowledge of the nature of the impurity is not required. It is based upon the fundamental thermodynamic principles of heterogeneous equilibria. Solubility analysis measures not only the amount of impurity but also the number of components in the system, and the solubility of the major component and of each of the impurities. I n addition, if there is impurity present, a separation is effected so that, in the course of the analysis, very pure material can be obtained from an initially impure sample. Comparison of the properties of the material in solution with the purest material may give some insight to the chemical nature of the impurity. Furthermore, all of the sample can be recovered, only simple equipment is required, and little applied time is necessary. This analysis has been applied to manv different steroids in this laboratory over the past 2 years, largely as a research tool in the preparation of purified steroids and the correlation of purity with physical and biological properties. I n addition, this analytical method has been of value from a synthetic point of view in indicating those cases where the starting material has been impure. With this information, further purification of the 8tarting material has often resulted in increased over-all yields. The theoretical development of the solubility method of analysis has been thoroughly discussed bp Xorthrop and Kunitz ( I S ) , Herriott (91, nnd Webb (18). The solubility method of analysis is based on the principle that a solution saturated with a pure component will be of constant composition regardless of the amount of excess solid phase. If, however, the solid contains two or more components, the amount of materidl in solution will be a linear function of the total amount of material added to the system until saturation with respect to hoth major and minor components is attained.
ANALYTICAL CHEMISTRY The slope of a graph of solute concentration plotted against the total amount of solid per unit volume added to the system thus yields a measure of the solute purity. Any general method of measuring the concentration of solute is satisfactory. Sorthrop and Kunitz ( I S ) , in their work vvith proteins, measured the amino nitrogen concentration of the solution. Other observers have used optical rotation ( 7 ) , refra2tive index ( 2 1 ) , and difference in vapor pressure (IO) as a measure of the concentration of the solution. Furthermore, as Xorthrop and Kunitz (13) have pointed out, the solubilities of the various components of a mixture can be calculated from the same phase diagram. The solid phase which is in equilibrium with a solution not yet saturated with the minor components will consist of the “pure” major component, while the solution system which is saturated with respect to all components will be the most impure sample (tailings). In this way, in the present investigation. small quantities of “pure” steroids were prepared and their physical constants compared with the tailings fraction. I n the case of steroids, the solubility method of analysis fails to give a quantitative measure of the amount of impurity in only two theoretical cases: if the mixture consists of a solid solution, and if the mixture consists of two or more components present in a ratio equal to the ratio of their solubilities. If the solid phases exist in the form of a solid solution, two types of phase diagram may result. I n the unlikely event that components which make up the solid solution have identical solubilities, a curve exactly similar to that of a pure substance will result, and no indication of impurity is seen. However, if the components have different solubilities, the phase diagram of this system may appear as a curve with continuously changing slope and inconsistent values of the extrapolated solubilities may be observed for samples of various degrees of purity. No quantitative knowledge of the amount of impurity can be obtained from this curve, but the very nature of the curve indicates that the sample is not a pure substance. The problem of solid solutions is most difficult and careful observation must be made to ascertain if one faces this complication.
5
14/20
35 ML.
FRITTED-GLASS FILTER (FINE)
Figure 1. Positive Pressure Filter All ground-glass joints
If the components are present in direct proportion to their solubilities, the stoichiometry of the system demonstrates that no indication of impurity will be obtained. T o obviate such false readings, the analysis may be conducted in another solvent system, in which the solubility ratio will probably not be the same. If the second analysis again shows constant solubility, the sample in all probability is pure. An additional test to distinguish a multicomponent mixture from a solid solution may be obtained by repeating the solubility analysis on that solid which first separates out of solution in the initial equilibrium system. If a rounded curve aimilar to that originally found is produced, then the original material is a solid. solution.
123
V O L U M E 25, NO. 1, J A N U A R Y 1 9 5 3 A further limitation expected for this method of analysis might arise from nonideality of the solution. The effect would be manifested by either a positive or negative error, depending upon whether the impurity increases the concentration of the major component by solubilization ( I ) , or decreases it through salting out phenomena. Xonideality can be eliminated or minimized by xorking in very dilut,e solution. A mixture of two or more crystalline polymorphic forms of a pure compound will, if equilibrium has been obtained. give the phase diagram of a single component, IIowever, the rate a t which equilibrium is reached might vary widely for the two forms. Conip:trison of the crystalline properties of the original material with the solid obtained from the soluhility analysis would be likely to show differences, as the latter now should consist wholly of that crystalline form which is stable under the experiment'al conditions. EXPERIMENTAL
The solubility method involves mixing different amounts of a solid sample with known volumes of solvent until equilihrium is reached. The equilibration is generally carried out in sealed glass ampoules. The solid phase is then separated from t'he solution and the concentration of solute is determined. This concentration is plotted against the total amount of sample added per unit volume. Choice of a suitable solvent is of primary importance in the analysis. I t appears that, for steroids, best results were obtained when the sample had a solubility between 0.1 and 1%. SoluhilitJ- greater than this requires ewessive amounts of sample and may lead to solutions which are 110x1ideal. For ease of measurement and evaporatioq the boiling point of the solvent should t)e betw\.ren 65'' and 100" C. High boiling solvents require long periods of time for aliquots of solution to tje evaporated to constant weight. Solvents xhich are known to form complexes should be avoided. and care must he taken to dissociat,e any such complexes in the original sample of material under investigation. I n general, precise data can be obtained with from six to eight equilibration ampoules in an experiment designed to consume a minimum amount of sample. On]>- a rough estimate of the solubility need be known beforehand. One equilibration ampoule should contain just enough sample to permit complete solution a t the temperature of the study. The amount of material found in solution should be the same as that originally added per unit volume. Thus, the precisioii of the experimental techniques may be checked. The second equilibration ampoule was planned to contain a small amount, of solid phase out of solution (5 to 10% of the total sample added). Two or three additional equilibration ampoules were set up Jyith progressively more solid phase present at. equilibrium (from 20 to 5070 of the total sample added). Emphasis must hr given to this region of the phase diagram. since very small amounts of impurity can he detected only if there is a sufficient number of points following the abrupt change in the phase diagram corresponding to saturction with the major component. This is particularly true if the impurity is considerably less soluble than the major component and if the impurity is present in sizable amounts. Three to four additional equilibration ampoules were prepared with increasing amounts of sample, the last planned to yirld approximately 90% of the total sample out of solution. Procedure. The requisite amount of finely ground thoioughly dried sample was transferred into a soft-glass ampoule (20-ml. capacity) by means of a small long-stemmed funnel, care heing taken that no solid adhered to t,he neck. The amount of compound added was obtained to f O . l mg. by the increase in ampoule weight. A clean glass bead was added, and the neck of the ampoule was carefully heated, and drann out to a diameter which would barely permit the passage of a long 21-gage hypodermic needle. Thirteen milliliters of solvent were added by means of the hypodermic needle attached through a ground joint to an accurate buret. -4fter the ampoules had been chilled in an ice
bath for 2 to 3 minutes, they were sealed in such a way as to leave no narrow pocket in which small amounts of solid could , lodge. The ampoules were agitated by slow end-over-end mixing in a constant temperature water bath. For the yajority of the analyses, the solubilities were measured a t 25 =k 0.02" C., although a bath temperature controlled to =k0.lo C. has proved to be satisfactory. That equilibrium had been reached was confirmed by approaching the solubility equilibrium from both the supersaturated and undersaturated sides. This was accomplished by heating one of the intermediate equilibration ampoules in a warm water bath (37" C.) for 24 hours prior to equilibration a t 25' C. In general, 2 weeks were sufficient for most steroids and equilihrium had been attained after 3 weeks In every case.
i Z
I
=z
2.4-
23
SOLVENT: BENZENE I
-
f2.23 -I2.0 -
0
-
FR.
WEIGHT OF
m
SAMPLE MG,/ML SOLVENT
Figure 2.
Cortisone Acetate
The simplest and most generally applicable method of determining the amount of so1ut.e in solution is to measure the dry weight of an appropriate aliquot. Accordingly, two successive aliquots ( 5 ml.) were removed by pipet into separate residue flasks and evaporated to dryness in a vacuum oven (40' C.). Removing these aliquots without also drawing solid particles into the pipet has been one of the biggest problems in the application of the technique t'o steroid samples. Many of the suitable solvents have nearly the same density as do t'he steroid crystals, while others are so dense that the crystals float. Originally the ampoules were allowed t,o stand in an upright position in the constant temperature bath for 24 hours prior to pipetting off the clear supernatant solution. A glass u-001 plug over the end of the pipet was used to filter out the solid particles, but was not suitable when the particles were very small. I n order to overcome this difficulty, a positive pressure filtering device, shown in Figure 1, has been const,ructed for sampling the suspensions. This filter, which can be immersed in the constant temperature bath, had the additional advantage of completely separating solid from solution. Thus, all of the solution could be used for sampling and the solid phase was freed from adhering solution for further examination. The equilibration flasks were removed from the watser bath, FThich was in a 25" =I= 2' C. room. They were quickly wiped dry, opened, and poured into the filter. Positive pressure was applied by means of nitrogen previously bubbled through the same solvent. The filter was then removed from the bath and the aliquots were pipetted into tared residue flasks. [For much of t8he authors' work, flat-bottomed glass ampoules with the necks removed have been found suitable. However, a small flask with a ground-glass top containing a capillary 1 to 2 mm. in diameter is recommended (19). Longer time is required for evaporation, but there is less chance for bumping and spattering.] The aliquot volume may be determined either by the pipet, or more precisely through weighing the solution. The contents of the residue flasks were evaporated to dryness under vacuum (at less than 40" C.) while a small volume of dry nitrogen was passed continuously through the oven. The residue flasks were stored in a desiccator for 1 hour, weighed, dried again, and reweighed. This process was repeated until constant weight was obtained. While the weights have been obtained on a microbalance (to f 0 . 0 0 5 mg.), an analytical balance may be used if larger aliquots of the solution are taken. Curves \\-ere obtained in which the amount of solute per milliliter of solution 7%-asplotted against the amount of solid added ini-
ANALYTICAL CHEMISTRY
124 *
tially per milliliter of solvent. The amount of impurity was calculated from the slope of the curve immediately following the point a t which an insoluble phase first appeared. For a pure material, this slope will be zero. For an impure substance containing one contaminant the weight fraction of the contaminant is equal to the slope of the curve. The solubility of the major component is given by the Y intercept of this slope, while the solubility of the minor component is given by the intercept of the constant solubility curve minus the Y intercept of the slope.
9 0 MG. SAMPLE SOLVENT: METHYL ALCOHOL
5
&
0
I-
3
OCH3
cn
z -
0
k'6I
w
( 3 -
% I
SOLVENT: ETHYL ALCOHOL
-.I
I
I
Y
..
.
.,
-
1.0%IMPURITY
(3 1.61
X
I
L_
FR.1
w WEIGHT OF SAMPLE Figure 3.
MG./ML. SOLVENT
Cortisone -4cetate
Multicomponent systems produce phase diagrams with several points of inflection (cf. Figure 2), each corresponding to a component. For such systems the total amount of impurity is still given by the first slope, while the weight fraction of the individual impurities may be computed from the difference in slopes on either side of each point of inflection. RESULTS AND DISCUSSION
Cortisone Acetate. In a study of the physical properties of cortisone acetate, a number of samples a t different stages of purification were examined by solubility analysis. Benzene was found to be the solvent of choice, because of the small solubility of cortisone acetate. The results of the solubility analysis for this series of samples are plotted in Figure 2. The phase diagrams clearly sho\v that impurity is present in all the samples, diminishing considerably in quantity as the final purification stages are reached. The second fraction appeared to have only one minor component (4.5%), while the crude material had a t least two impurities (5.5 and 9,2%), and possibly more, as saturation with respect to all components was not reached. I n the case of the purest ample of this series, containing 0.8% impurity, the solution did not become completely saturated. This appears to be the consequence of the very small amount of minor component in this sample. So little was present that a very great excess of sample must be added per unit volume of solvent to achieve complete saturation and the flat portion' of the phase diagram. The solubility of pure cortisone acetate at 2.5" C. in benzene, obtained by extrapolation of the total impurity slope of each of the three samples to the Y axis, averaged 1.33 mg. pcr ml. (1.36, 1.32, and 1.30 mg. per ml., respectively). The utility of the solubility method in confirming the identity of the impurity may be demonstrated by calculating its solubility. In the crude sample the major impurity had a solubility in benzene of 0.5 mg. per ml., while that measured for free cortisone, the most likely impurity based on the previous history of the samples, was 0.48 mg. per ml. Comparison by means of infrared, ultraviolet, and other methods (16) between the solid phase which first separates from solution prior to complete saturation (pure cortisone acetate) and
I 14 L1
40
20
60
100
80
WEIGHT OF SAMPLE MG,/ML. SOLVENT Figure 5. Methylandrostenediol the completely saturated solution, confirmed the presence of free cortisone. To establish with certainty the purity of the best sample, it was analyzed in ethyl alcohol (Figure 3). The amount of impurity determined in this solvent was 1.0% as compared to 0.8% in benzene. The experiments described above have all required a total sample of approximately 600 mg. However, in the earlier investigations of cortisone acetate when only small quantities were available, a solubility analysis a-as carried out using 0.2-ml. aliquots after equilibration. Only 90 mg. of material were required, all of which a a s recovered a t the end of the experiment. The results shown in Figure 4 indicate the sample to be approxi-
SOLVENT: TOLUENE x x
x
FROM FROM
UNDERSATURATION SUPERSATURATION Y n
HO
I
I
I
I
I
I
1
I
2
4
6
8
IO
12
14
16
WEIGHT OF SAMPLE MG/ML. SOLVENT Figure 6. Methylandrostenediol
V O L U M E 25, NO. 1, J A N U A R Y 1 9 5 3
125
F 10.0-
U
g
3 A 9.5-
f t-
/&o" a%i M PUR ITY
9.0-
I
1
-
(3
g
8.5-
I
I
I
I
I
I
mately 98% pure. Improved handling techniques and a less powerful solvent, such as benzene, now permit similar solubility analyses to be carried out with only 30 to 40 mg. of material. Methylandrostenediol [A5-17 Methylandrostene-3( p), 17( p)-diol]. I n a study of the biological properties of methylandrostenediol. samples of known purity were required. Samples known to be impure were measured by the solubility method of analysis. Figure 5 shows the results of these preliminary investigations, which are of interest in that the minimum amount of impurity which could be present was estimated from only two point$. The sample was measured in two different solvents (chloroform
and acetone) and surprisingly good agreement was obtained between the results. While estimates based on two-point studies are valuable, they indicate minimum values for impurity. The second point in each curve might actually lie on the constant solubility line. thus possibly leading to an estimate of purity which may be too high. After purification of this material by recrystallization from ethyl acetate, methanol, and toluene, carefully controllt~d solubility analyses were carried out. I n one case, toluene was used as the solvent. As indicated on Figure 6, equilibrium was proved by approach from supersaturat'ion as well as from undersaturation. The slope calculated by the least squares method indicated 1.5% impurity. Detailed analysis of the phase diagram indicated two impurities present, one to the extent of 0.9% and the other to about 0.6%. The solubility of the former appeared to be 0.7 mg. per ml. as estimated from the diagram. This very low solubility of the impurity suggests that it is a highly polar contaminant. Confirmation of the amount of impurity found in the toluene system was obtained by analyzing the same methylandrostenediol sample in ethyl acetate (Figure 7 ) . Three per cent of impurity was present, as determined by the second analysis. The agreement in the results ohtained from different solvents strongly suggests that this particular sample is at least 97% pure.
x
x
FROM UNDERSATURATION FROM SUPERSATURATION
x
SOLVENT ACETONE x
4 % IMPURITY (FR.II)
f9 t
ACETONE
SOLVENT:
10% I M P U R I T Y (CRUDE)
c
,.
W
BENZEN.E
2 % I M P U%R I T Y (FR.1)
-::-
- 6
s 'I100
SOLVENT:
c9 w
I
I
I5 0
(RECRYSB
I
I
I
250
200
300
OF SAMPLE MG./ML. SOLVENT
WEIGHT
Figure 8.
Pregnenolone Acetate
i I \
W'
5
SOLVENT : ACETONE
z
1, cn
z
II
F t 3
,
,
100
200
,
300
,
400
WEIGHT OF SAMPLE MG./ML. SOLVENT Figure 9.
Pregnenolone Acetate
Prepared from diosgenin
I
IO
I
I
20
I
I
30
I
I
40
WEIGHT OF SAMPLE MG/ML. SOLVENT'
ETHYL ASETAT
I% I M P U R I T Y
I $)
,I 500
Figure 10. Pregnenolone
A6-Pregnen-3(p)-ol-20-one Acetate. I n a study of the polymorphism of pregnenolone acetate, it was necessary to establish the purity of the material isolated from different sources. While an initial crude sample prepared through the oxidation of cholesterol acetate dibromide was only 90% pure, recrystallization from isopropyl alcohol and from benzene-ethyl ether increased the purity to about 99% as determined in ethyl acetate (Figure 8). Furthermore, pregnenolone acetate prepared from diosgenin yielded similar results (about 99% pure, measured in acetone, Figure 9). The anomalous melting behavior of samplrs from the different sources was attributed to the presence of varying proportions of a t least two polymorphic forms (8) A5-Pregnen-3(p)-ol-20-one. In an early study, pregricnoloile crvstallized from acetone was analvzrd by the solubilitv method using acetone as the solvent. The results shown in Figure 10 indicated approximately 4% of an impurity. A later analysis of pregnenolone demonstrated the improved techniques. A less effective solvent (benzene) was used, permitting larger aliquots without requiring exccssively large samples and the solutions were removed from the solid phase by the positive pressure filtering device. The phase diagram shown in Figure 10 indicated that the sample was 98y0 pure and that solubility equilibrium had been reached. 3p-Acetoxy-20-hydroxy-5-cholenicAcid Lactone. I t had been
ANALYTICAL CHEMISTRY
126
ol)served in this laboratory in 1949 that the infrared spectrum Table I. Estimation of Amount of Dihydrocortisone of the lactone first isolated from the oxidation products of choAcetate in Unknown Mixture It.stxol acetate dibromide by Miescher and Fischer (11) was not Added, Mg./311. Found in Solution, consistent with the &lactone structure generally aczrpted (2, 1 7 ) Unhnown - Mg./Ml, Residue hut suggested a ?-lactone structure (16). I n order to e s t a i h h Anipoule Reference mixture No. 1 No. 2 Average the authenticity of the sample, and to characterize it further hv .1 23 6 10.1 26.648 26.523 26.385 B 23.2 .. 18.808 18.983 18.893 physical methods, a solubility analysis was carried out in benzene. AIixture dissolved, mq. 7.690 Analysis of the phasr diagram (Figure 11) shows that approxiMixture undismately 3y0 impuritv n as present. Furthermore, some pure inasolved, mg. e . 400 terial v a s obtained by filtering off the solid which separated from the equilibration ampoules, not yet saturated with respect to the inipuritv. Larger quantities were prepared by repeating a t the same solid to solvent ratio using 100-ml. ampoules. Dihydrocortisone Acetate [ Pregnane-li( ~~),21-diol-3,11,20-trione 21 acetate]. I n connection with some early studies on dihgdrocortisone acetate, solubility analysis was carried out using ethyl acetate as a solvent (Figure 12). rlt least txr-o impurities \\ere identified as one of the impurities and it was shown that a t least found, amounting to approximately 1.3 and 2.6%. I n addition 24y0 of the residue consisted of this substance. The presence of to determining the purity of the sample, sufficient pure mateiial dihydrocortisone acetate was confirmed by isolation and characmas obtained for use as a reference sample to establish melting terized by other physical constants. point, infrared spectrum, and optical rotation standards. PRECISION
.jG l E
2
SOLVENT:
2 3.0-
BENZENE
It is estimated that when properly planned and conducted purity analyses are precise to *0.5%,. Certain of the early parts of this investigation and those phase diagrams n-ith fewer than six points undoubtedlj- have larger errors.
Z
!-5I
02.5
m
CH3c
