employed. However, this system did not effect the separation of l-propanethiol and methyl ethyl sulfide. Tests on known synthetic laboratory mixtures and commercial mixtures containing these thiols, sulfides, and isopropanol indicated t h a t the areas under the chromatograms are proportional to the weight per cent present with a n accuracy of the amount present. of about =t4yo Column -4, di-n-butyl phthalate, exhibited complete resolution of all thiols and sulfides in this investigation but did not effect resolution of isopropyl alcohol and 1-propanethiol. This column may be used with the same accuracy as the tandem column C-D for the measurement of isopropyl alcohol in mixtures not containing l-propanethiol. Column El, DC 200 silicone fluid, performance closely paralleled that of column .A with no resolution of 1-propanethiol and isopropyl alcohol peaks. Column E, tritolyl phosphate, did show some resolution of the isopropyl alcohol and the 1-propanethiol peaks but no quantitative separation of these two compounds. Ryce and Bryce ( 7 ) report excellent resolution of methyl and ethyl alcohol from their corresponding thiols on a TTP column with programmed temperature. Separation of methyl and ethyl alcohol from mivtures of these thiols and sulfides was effected on the columns used in this investigation also. Other columns used in this investigation were firebricks n ith Viscasil 60,000
Figure 1. Observed relative retention times of thiols on column C-D vs. columns A, B, and E Column A, Column E,
A.
Column B, 0 .
0
I
0
and oxybis-2-ethylbenaoate. These colUmnS showed poorer separation of thiols and sulfides under conditions of reasonable flow rates and temperatures. ACKNOWLEDGMENT
The authors thank W.W.Hanneman of the California Research Corp. for his technical assistance in this and related basic chromatographic techniques.
,
1
, , I
I
,
I
,
/
(2) Coleman, H. J., Thompson, C. J., i$'ard, C. C., Rail, H. T., CHBM. 30, 1592 (1958). (3) Desty, D. H,, xature 179,241 (1957). (4) Desty, D. H., Whyman, B. H. F., A N A L . CHEM. 29,320 (1957). ( 5 ) Karchmer, J. H., Ibid., 31, 1377 (1909). (6). Liberti, A,, Cartoni, G. P., Chim. e. znd*
399
821
(7) Ryce, 9. A., Bryce, W. A,, ASAL.
cHEY. 29, 925 (1957).
LITERATURE CITED
(8) Spencer, C. F., Baumann, F., Johnson, J. F., Ibid., 30, 1473 (19%). (9) Sunner, S., Karrman, K. J., Sunden, V., Mikrochim. Acta 1956, 1144.
(1) amberg, C. H., Can. J . Chena. 36, 590
RECEIVED for review August 8, 1961.
(1958).
Accepted Xovember 39, 1961.
Liquid-Liquid Partition Chromatography of Steroids Systematic Approach Relating Column to Paper Chromatography Using the R, Function PETER KABASAKALIAN and JOSEPH M. TALMAGE Chemical Research and Developmenf Division, Schering Corp., Bloomfield, N.
b A direct extension of R p data for steroids from paper to partition column chromatography has been made. The practical quantitative range of a series of Zaffaroni-type solvent systems has been described using the generalized R,), function previously reported.
L
partition column chromatography has been used estensively by steroid chemists to separate and isolate milligram to gram quantities of unknown compounds after all the simple methods of separation have IQUID-LIQUID
J,
failed. It would be most helpful to know the solvent system required for the separation and the volume in which the compound would be eluted from such a column. The first requisite can be fulfilled by the use of paper chromatography (4,6), while the second requisite necessitates the use of some function which would interrelate the solute mobilities in paper and column chromatography. Consden, Gordon, and Martin (S), who have considered paper chromatography to be simply a form of liquidliquid partition chromatography in which the filter paper acts as the inert
support of a stationary aqueous phase, have defined the quantity ( R F ) pfor paper chromatography as the ratio, v/V, of the distance traveled b y the leading edge of the solute band to the distance traveled by the solvent (Figure 1). The quantity ( R F ) , for column chromatography (6) is defined as the ratio, v/V, of the rate of movement of the maximum concentration of a solute band down the column to the movement of the eluting solvent in the packed column. The measurement of the movement ( V em.) of the developing solvent in descending paper chromatography is started a t the top of the paper VOL. 34, N O . 2, FEBRUARY 1962
273
PPPER
while the corresponding measurement (V cc.) in column chromatography is started at the bottom of the column, since the column is already filled with ( u cc.) mobile phase. The free-column volume (6) ( u cc.) must pass down the column before any solute can be eluted from it. Partition column chromatography is analogous to descending paper chromatography with the solvent front allowed to run off the paper until the solute has migrated to the edge of the paper. This analogy is represented in Figure 1. I n paper chromatography V is a constant and v varies from solute to solute while in column chromatography the inverse is true. For practical reasons, in column chromatography the use of (l/Rp), is preferred over the use of (RF)c; since each solute must move v cc. before it is eluted from the column, the number of free-column (6) or column holdup volumes (V/v) of solvent necessary to achieve this movement is equal to ( l / R F ) c . According to the Martin-Synge (’7) theory of liquid-liquid partition chromatography, the RF value depends on the relative amounts of the two phases in contact, K (ratio of the volumes of mobile and stationary phase), and on the partition coefficient of the solute, CY (ratio of the concentrations in the stationary and mobile phase), as a = K [ ( l / R F ) - 11
(1)
Csing the R,Mfunction defined by BateSmith and Westall ( I ) , as RM =
log [(1/RF)
-
11
(2)
Equation 3 defining (R.M),for partition column chromatography can be derived from the ( R M ) ,for paper chromatography and Equations 1 and 2. (R.+f)c=
(R.M)P
+ log (KPIKJ
EXPERIMENTAL
The columns (70 cm. long X 1.5 cm. i.d.) were packed by gravity flow (8) with 20 grams of Supercel suspended in 200 ml. of the mobile phase (equilibrated with the stationary phase), and 10 ml. of stationary phase (equilibrated with the mobile phase). Xitrogen pressure ( 5 t o 10 p s i . ) was applied to compact the columns. The flow rate during development under a constant liquid head (no pressure applied) was about 2 ml. per minute (8). The solvent systems previously reported (5) were used (see Table 11). The column holdup volume, v, was determined by passing a dye such as Calco Oil Red (American Cyanamid Co., Bound Brook, N. J.), which has a small CY, through the column and determining the volume at which it first appears in the eluate. A 50-mg. steroid sample was introduced into the column with 2 t o 5 ml. of the mobile phase. ANALYTICAL CHEMISTRY
m.
m!. FREE-COLUMN VOLUME
I/ u 1 ( I I R p lc
V/v
Figure 1. Symbolic comparison of paper and column partition chromatography
The column eluents were analyzed by passage through a Phoenix Differential Recording Refractometer (Phoenix Precision Instrument Co., Philadelphia,
Pa.).
Table I.
Paper chromatography was performed as previously reported ( 5 ) . -4plot of (h!,+f)c us. ( R M )of~the calibrating compounds for each of the six solvent systems was made from a-hich the
Calibration Data for
K,/K,
(3)
The subscripts, p and c, refer to paper and column, respectively.
274
COLUMN
Compound
Determination
Solvent System
( R F P)
(~IRF)~
1 1
0.10
5.0 3.6
2 acetate 2 1lp,l7aJ21-Trihydroxypregn-4-ene-3,20-dione 21-acetate 17~,21-Dihydroxypregna-1,4diene-3,11,20-trione 212 acetate 2 17a,21-Dihydroxypregn-4-ene-3,11,20-trione 21-acetate 9~-Fluoro-11~,17~~,21-trihydroxypregna-1,4-diene-3,203 dione 21-acetate 3 lia,2l-Dihydroxypregn-4-ene-3,20-dione 17~,21-Dihydroxypregna-lJ4-diene-3,11,20-trione 213 acetate 3 17~,21-Dihydroxypregn-4-ene-3,11,20-trione 21-acetate 3 17~,21-Dihydroxypregn-4-ene-3,20-dione 21-acetate 3 21-Hydroxypregn-4-ene-3,2O-dione 4 21-Hydroxypregn-4-ene-3,2O-dione 4 21-Hydroxypregn-4ene-3,20-dione acetate 4 Pregn-4-ene-3,20-dione 5 17&Hydroxypregn-4ene-3,2O-dione acetate 5 3p-Hydroxypregna-5,16-dien-2O-one 5 16a,17~-Oxidopregn-4ene-3,20-dione 5 Pregn-4-ene-3,20-dione 5 3,%Hydroxypregna-5,16-dien-20-one acetate 5 58-Pregnane-3,20-dione 6 58-Pregnane-3,20-dione 6 3p-Hydroxy-5a,22aY,25D-spirost-9( ll)-en-lZone acetate 6 5a,22~~,25D-Spirostan-3@-01 6 Stigmasta-4,22-dien-3-one
0.12 0.24
10.0
0.41
2.5 2.4
llpJl7a,21-Trihydroxypregna-1,4-diene-3,2O-dione 1Ip, 17~~,2l-Trihydroxypregn-4-ene-3,20-dione llp,l7a,21-Trihydroxypregna-1,4-diene-3,2O-dione 21-
0.16
0.47 0.17
3.9
6.7
0.20
4.5
0.32
2.3 2.3 1.9 1.3 11.2
0.36
0.45 0.69
0.10 0.33 0.57
0.17
0.18 0.21
2.9 1.7 4.3 5.4
3.5
0.31
2.8
0.12 0.23
7.3 4.7 3.0
0.55 0.42
1.4 2.1
0.30 0.70
1.3
intercept, log ( K D / K C was ) , determined. The useful chromatographic [R,wIc range for each solvent system was calculated using tlie generalized [R.v], ( 5 ) in Equation 3 and R, limits corresponding to 1.4 and 10.0 (l/RF)cunits. The (l//RF)cunits for steroids used in the evaluation study were calculated using Equation 4 and compared with their experimentally determined values. (~/RF), (K,IKc)[(l/RF), 5
-
11
+1
(4)
RESULTS
The compounds used in the calibration study are listed in Table I together with their experimentally determined (RF),and ( ~ / R P values. ), The inter, as a result cepts, log ( K p / K c )obtained of plotting ( R M )1;s. ~ ( R . b , ) F values are listed in Table 11, with their useful [ R M ]ranges. , Table I11 contains the predicted and experimental (l/RF)c values of the steroids used in the evaluation study. DISCUSSION
The plot of (R.v)~us. ( R M )of~ the calibrating steroids gave a slope of unity indicating that paper and column partition chromatography in the six solvent aystems studied are essentially similar. The intercept, log ( K p / K c ) , yielded K F / K , values ranging from 0.47 for the chloroform-formamide system to 1.11 for the benzene-formamide system. As a first approlimation (1/RF),can be assumed to be equal to (l/RF), if the experimentally determined K,/Kc value for a particular solvent system is not available. The use of Equation 4 with the experimentally determined K p / K c value indeed yields a predicted ( ~ / R Fvery ) , close to the experimentally determined one (Table 111). The precision of the predicted (I/ RF),was limited to the precision of the ( R F ) , which was 0.05 R F unit while the experimental ( l / R F \ ,values were good to 0.1 (l/RF)cunit. The alternate direct prediction of
Table 11.
Useful
[Rnl] Ranges for Column Steroidal Solvent Systems
Log Solvent System ( K,/Kc) (KdKC) 0.47 1 Chloroform-formamideb -0.329 0.040 1.11 2 Benzene-formamide 0.79 3 Toluene-propylene glycol -0.103 1.00 -0.001 4 Ligroin-propylene glycol -0.147 0.71 5 Heptane-methyl Cellosolve -0.060 0.87 6 Heptane-phenyl Cellosolve a Based on 1.4 to 10.0 ( ~ / R Frange. ), * Fisher Scientific Co., Fair Lawn, N. J., stabilized.
[RM]Range” +1.88 to $0.56 $1.46 to +0.14 +0.85 to -0.47 -0.52 to -1.58 -0.82 to -2.14 -1.27 to -2 59
( 1 / R F ) c Values of Steroids Used in Evaluation Study Solvent Exptl. System Compound (~/RF), Sa-Fluoro-16a-methyl-1 lp, 17a121-trihydroxy1 0.19 3.0 3.3 pregna-l,4diene-3,2O-dione 16a-Methyl-1la1l7a,21-trihydroxypregna-l,40.21 2.8 1 2.8 diene-3,aO-dione 17a,20p,21-Trihydroxypregna-l,4diene-3, 110.30 2.1 1 1.9 dione 16a-Methyl-1 1aJl7ac,21-trihydroxypregna-l,42 0.13 8.4 9.7 diene-3,20-dione 21-acetate Sa-Fluoro-16a-methyl-1 la, 17a,21-trihydroxy0.21 5.2 2 5.7 pregna- 1,Pdiene-3,20-dione 2 1-acetate Sa-Bromo-l6a-methyl-11pll7a,21-trihydroxy4.4 0.28 3.9 2 pregna-lJ4-diene-3,20-dione21-acetate 0.12 6.8 3 7.6 Estra-1,3,5( lO)-triene-3,17p-diol 0.56 1.6 3 1.6 17a-Methyl-17p-hydroxyandrost-4-en-3-one 16a-Methyl-9p, 1lp-oxido-17~,21-dihydroxy0.65 1.4 1.3 pregna- 1,4diene-3,20-dione 21-acetate 3 4 0.24 4.2 4.3 17a-Methyl-17p-hydroxyandrost-4en-3-one 16a-Methyl-9p, 1lp-oxido- 17a,2 l-dihydroxy4 2.8 pregna-l,Pdiene-3,20-dione 21-acetate 2.7 5 16a-Methyl-3p-hydroxy-5a-pregnan-20-one 1.5 5 5~~,22a,25D-Spirostan-3p-o1 4.8 6 16~-Methyl-3/3-hydroxy-5a-pregnan-20-one Table 111.
16~-Methyl-17a,20a-ouido-5~-pregnane-3pI2O~6 diol diacetate
( l / R F ) , from the distribution coefficient, CY, and the ratio, K,, of volumes of mobile and stationary phase (Equation l), is often very difficult. First, it assumes the availability of a very pure sample of solute for the CY determination. Secondly, the determination of the ratio of volumes of mobile and stationary phase is no simple task ( 2 ) . LITERATURE CITED
(1) Bate-Smith, E. C., Westall, R. G., Biochim. et Biophys. Acta 4, 427 ( 1950). (2) Butt, W. R., Morris, P., hIorris, C. J.
0.51
1.8
1.9
0. R., TT7illiams,D. C., Biochern. J . 49, 434 11951’). ~. (3) Consden, R., Gordon, A. H., Martin, A. J. P., Ibid., 38, 224 (1944). (4) Decker, P., il’aturuiissenshuften 45, 464 (1958). ( 5 ) Kabasakalian, P., Basch, h., ANAL. CHEY.32.458 (1960). (6) Laitinen, H. A,,“Chemical Analysis,” pp. 493, 528, McGraw-Hill, New York, 1960. (7) yartin, A. J. P., Sgnge, R. L. ?”I., Bzochem. J. 35, 1358 (1941). (8) Trenner, N. R;: “Column Partition Chromatography, Merck and Co., Rahway, W.J., 1957. RECEIVEDfor review August 11, 1961. Accepted November 14, 1961. ~
VOL. 34, NO. 2, FEBRUARY 1962
275