420
Anal. Chem. 1900, 52,428-430
Liquid Chromatographic Resolution of the Unsaturated Sesquiterpene Alcohol Isomers Using Silica Gel-Binary Solvent Systems Shoji Hara," Akiko Ohsawa, Jiro Endo, Yutaka Sashida, and Hideji Itokawa
Tokyo College of Pharmacy, Horinouchi, Hachioji, Tokyo 192-03, Japan
For optimization of related separation procedures, the retention behaviors of four unsaturated sesquiterpene alcohol isomers, cy-, (3-, y-eudesmol and hinesol in silica gel liquid chromatography were systematically investigated. Positional Isomers differing In the location of a double bond were separated by an argentation column. A high efficiency preparative silica gel column was useful also for the resolution of these closely related isomers. Pure samples were fractionated from the essential oils of plants and their molecular structures were identified and confirmed by spectrometric measurement.
CY-, p-,y-Eudesmol and hinesol have been identified in the essential oils from the bark of Magnolia obovata Thumn, and Atractylodes lancea DC. (I, 2 ) . These compounds are classified as unsaturated sesquiterpene alcohols and are very closely related in their structures. The structural formulas are illustrated in Figure 1. a-, p-, and y-Eudesmols are isomers differing in the position of a double bond and are characterized as endo trialkyl, exo dialkyl, and endo type tetraalkyl substituted alkenes respectively. Hinesol is also characterized as a n endo form trialkyl substituted alkene. Because of the structural similarity of these compounds, their retention behaviors in gas-liquid, thin-layer, and liquid column chromatography are also very similar ( I , 2 ) . Consequently, procedures for the determination and the preparative isolation of these compounds have proved to be tedious and timeconsuming (3). Such structurally similar unsaturated alcohols are distributed widely in natural products giving need for easier determination procedures. T h e optimization of the chromatographic system used in the separation of eudesmol isomers should be a suitable index for systematization of separation procedures for a variety of natural organic compounds. T h e analytical and preparative resolutions of these compounds were thus examined in binary solvent-silica gel liquid chromatography which has been the most commonly utilized in the field of natural product chemistry.
EXPERIMENTAL Samples. Two fractions containing a mixture of CY-, d-, and y-eudesmols (fraction 1)and a mixture of hinesol and 6-eudesmol (fraction 2) obtained by procedures given in the literature from the bark of Magnolia obovata Thumn ( 1 )and Atractglodes lancea DC. ( 2 ) ,respectively, were employed as sample solutes and were dissolved in the mobile phase to form a solution with a concentration of 10% (w/v). Columns. Packing material was irregularly shaped silica gel having a pore size of 70 8, and a particle diameter of 10 Fm, Wakogel LCH-10 (Wako Pure Chemicals Co., Osaka). A silica gel slurry in chloroform/carbon tetrachloride/dioxan (2:1:2 v/v) was packed into a glass tube, 4 mm i.d. X 30 cm, CIG column system ( 4 ) (Kusano Scientific Co., Tokyo) and a stainless steel tube, 8 mm i.d. X 50 cm (Chemco Co., Osaka) using a constant pressure pump model DSTV-122G (Haskei Engineering and Supply Co., Burbank, Calif.) under a pressure of 1000 psi for the glass column and 4000 to 8000 psi for the stainless steel column. 0003-2700/80/0352-0428$01 .OO/O
The number of theoretical plates for the glass column and the wide bore stainless steel column were 5000 per 30 cm and 15000 per 50 cm, respectively, using diethyl phthalate as the solute and n-hexane-ethyl acetate (9:l v/v) as the solvent. The argentation column was made by the following procedure ( 5 ) . A solution of 0.8 g of silver nitrate in 200 mL of methanol was cycled through the column for 3 h at the flow rate of 1.5 mL/min, impregnating the surface of the silica gel with silver nitrate. Chromatography and NMR Spectra. An injector model 7120 (Rheodyne, Berkeley, Calif.) and a pump model KP-9H (Kusano Scientific Co.) were linked to a model R 401 differential refractometer (Waters Associates, Milford, Mass.). Flow rates were 1 mL/min for analytical and 5 mL/min for preparative uses. Sample sizes in each injection were 0.1 to 0.2 mg for analytical and 20 mg for preparative purposes. Dead time was measured by using cyclohexane as a sample. Capacity ratio k'was obtained by the formula: k' = t,/t,, where t , is adjusted retention time and t , is dead time. Separation factor a was calculated by the formula: a = k',/k\. Chromatography was performed at ambient temperatures (ca. 12 "C) and solvents were equilibrated with ambient moisture prior t o use. NMR spectra were obtained in CDC13 solution with a JEOL-PS-100, 100-MHz spectrometer (JEOLCO, Tokyo).
RESULTS A N D DISCUSSION Crude extracts, fraction 1 containing a-, p-,and y-eudesmols ( I ) and fraction 2 containing hinesol and 6-eudesmol (2) were employed as sample mixtures for silica gel liquid chromatography. The retention behaviors of the solute compounds were systematically examined by preparing the binary solvent systems. n-Hexane (0) was selected as a diluent and an 0 + ethyl acetate (B2)system was first applied in the analysis. The codes for solvents used in this paper were suggested by one of the authors (Hara) in previous reports ( 4 , 6). T h e identification of the peaks on chromatograms was accomplished by comparing directly with the injection of a pure sample obtained from the preparative separation, the procedure of which will be described in the latter part of this article. By varying the solvent composition, t h e retentions of the constituents were determined and are shown in Table Ia. Because the solutes have a tertiary hydroxyl function as their active adsorption group, their adsorptivity was assumed to be weaker than that of compounds having primary or secondary hydroxyl groups and comparable to the adsorptivity of compounds with a carbonyl group. In accordance with this assumption, a comparison of the present results and those obtained previously (6) using steroid samples indicated that the actual adsorptivity of these compounds is similar to that of keto compounds. y-Eudesmol was separated from a and B isomers in fraction 1 and hinesol was separated from 0-eudesmol in fraction 2; however, the separation of cy and p isomers was unsuccessful. In order to examine the selectivity of the mobile phase leading to the resolution of the a,@ mixture, other basic and acidic solvents such as diethyl ether (Bl),acetone (B3),tetrahydrofuran (B4),and 2-propanol (AB) were adopted as the stronger components of the solvent systems. A solvent system containing 8% (mol/mol) ethyl S 1980 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 52, NO. 3, MARCH 1980
429
Table I. Retention Behavior of 0 - , 3 - , and 7-Eudesmol and Hinesol on the Silica Gel Colurnns Using Binary Solvent Systems Containing n-Hexane as Diluent a. Silica gel glass column fraction 1 so Iv en t
stronger solvent EtOAc ( B 2 )
separation factor a , c y - , 0-17-e
capacity ratio k ' y -ea a-,$-ea 0.89 1.00 1.58 1.68 2.10 2.26 4.66 4.06 7.81 7.16 6.67 5.86 1.95 1.76 3.58 3.86 5.71 5.32 3.97 3.39 5.84 6.37 9.00 8.46
composition, mol/mol% 20 15 12 8 4.5 13 10 5 3 0.68 0.34 0.30
fraction 2 capacity ratio k ' ~
~__
separation factor a,
ha
P -e
d-elh
3.53 7.15 5.80 1.61 3.35
4.53 7.68 7.13
1.28 1.07 1.23
1.95
1.21
4.16
1.24
3.16
3.97
1.26
7.53
9.40
1.25
1.12
1.06 1.08 1.15 1.09 1.14 1.11
1.08 1.07 1.17 1.09 1.06
Argentation silica gel glass column
b.
solvent composition,
capacity ratio k '
separation factor
a
~ _ _ _ _ _ _ _
stronger solvent
EtOAc (B:) Me,CO ( B , ) 2-PrOH ( A B )
1.19 3.76 6.58 5.85 7.77
20 10
3 0.4 0.3
1.30 4.26 7.24 6.56 8.48
1.63 5.68 8.42 7.65 10.45
1.09 1.13 1.10
1.12 1.09
1.25
1.33 1.16 1.17 1.23
c. Silica gel high efficiency-preparative column
e-,
stronger solvent
solvent composition mol/mol%
Et OAc (B, ) Me,CO ( B , )
4 2.6
capacity ratio k '
separation factor
Y -e
a -e
fi -e
9.72 7.94
10.59 8.74
10.86 8.94
0
I7
1.09 1.10
il
3 :u
1.03 1.02
3 - , y - e = a-,8 - , y-eudesmol; h = hinesol.
/-Eudt
JOO
H ttiesol
Figure 1. Structural formulas and NMR data of eudesmol isomers and
hinesol acetate (B,) in a diluent of n-hexane afforded a capacity ratio of approximately 4 for these solute compounds. This system was selected as a standard and equi-eluotropic binary solvent systems were extensively designed according to a procedure proposed by one of the authors (Hara) ( 7 ) . The concentrations of the stronger solvents in equi-eluotropic binary systems were calculated as follows: 13% for B , , 3% for B3, and 0.3% for AB in n-hexane (0).A system containing tetrahydrofuran (0 + B4) was also prepared and the concentrations of these solvent systems were finally adjusted according to sample compounds. Capacity ratios and separation factors are shown in Table Ia. Although various solvent systems were examined, the separation factor was similar in all cases and no mobile
phase suitable for separating a and p isomers was found. The preferable selectivity of the stronger component was not observed. For resolving such a difficult mixture, two directions of the investigation seemed to be useful. One was to increase the selectivity of the interaction between the solute and the stationary phase. The other was to increase the column efficiency itself. In order to differentiate and to separate positional isomers differing in location of a double bond, argentation silica gel columns have been rather widely utilized. Silver nitrate in methanol was equilibrated in a silica gel column by using Heftmann and co-workers' procedure ( 5 ) . The retention behavior of fraction 1was examined on the silver nitrate-impregnated silica gel column. It was found that a and isomers were separated with a separation factor of 1.16 to 1.33 by using 0 + BP, 0 + B,, and 0 + AB binary solvent systems (Table Ib). The adsorption sequence of y < a < @ observed on this column was opposite to the order of steric hindrance at the double bond and was the same as the sequence observed on non-impregnated silica gel. Capacity ratios for this column were larger than those of the silica gel column using the same solvent systems (Figure 2). These facts suggested that the double bond is associated with the adsorption phenomenon together with a hydroxyl group which is considered to be the main active site in the solute molecule. A mechanism in which the silver ion adsorbed on the silica gel surface enhances the activity of the adsorbent by charge-transfer r-complex association can be assumed. By means of an impregnation procedure providing an increase in system selectivity, a mixture of a-, p-, and y-eudesmol was fractionated; however, it is difficult to remove silver nitrate
ANALYTICAL CHEMISTRY, VOL. 52, NO. 3, MARCH 1980
430
Y
0,i
the analytical column is limited; consequently, the efficiency of the column is often decreased by overloading, especially when the system is linked to a refractive index detector which provides low sensitivity as a detection system. T o prevent reduction of efficiency, a large high-efficiency column was prepared by the slurry-packing procedure using constant high pressure. When the retention was adjusted so as to be greater than in the analytical procedure, the separation of a and /3 isomers was accomplished with a separation factor of 1.03 by using n-hexane-ethyl acetate (0 + BJ or n-hexane-acetone (0 B3) binary systems. Capacity ratios are shown in Table IC and Figure 2. In order to isolate and t o identify the molecular structure of the constituents, the sample mixture fraction 1 was fractionated by repeatedly injecting quantities of approximately 20 mg. Fraction 2 was also preparatively separated t o obtain pure hinesol and 0-eudesmol samples. T h e chromatograms are shown in Figure 3. NMR spectra of specimens in deuterated chloroform showed that the chemical shifts and the s p i n s p i n coupling constants of the compounds were identical with those described in the literature (I,2) and supported their proposed molecular structures. These values are indicated in Figure 1. Although both the structures of the four unsaturated sesquiterpene alcohol isomers and their retention behaviors in chromatography were very similar, by employing an argentation silica gel column, these solutes, differing only in the position of their double bond were successfully separated. I t should be noted that fractionation was also achieved using high-efficiency preparative silica gel column having a large loading capacity.
+
Flgure 2. Comparison of column selectivity and efficiency for the separation of eudesmol isomers ?-r.
I'
LITERATURE CITED
-1
t
_--I
I
20
30
IO
,I,
IU
Figure 3. Chromatograms of eudesmol isomers (0-, 6-,y e )and hinesol (h) in the preparative fractionation. Silica gel column. Solvent: (a) = 2.6% (mol/mol).(b) = 5% (mol/mol)acetone in n-hexane
from the silica gel surface and to recover the original column material. Therefore, attention was turned in the other direction of improving the efficiency of the silica gel column itself. It has been pointed out t h a t the loading capacity of
M. Fujita, H. Itokawa, and Y. Sash&, YakugakuZasshi, 93, 415 (1973). I. Yoshioka and T. Kirnura, Chem. Pharm. Bull., 17, 856 (1969). F. J. McQuillin and J. D. Parrack, J. Chem. SOC., 1956, 2973. S. Hara, J. Chromatogr., 137, 41 (1977). E. Heftmann. G. A. Saunders, and W. F. Haddon. J. Chromatogr.. 156, 71 (19781. (6) S. Hara, Y. Fujii, M. Hirasawa, and S. Miyarnoto, J. Chromatogr., 149, 143 (1978). , (7) S. Hara, M. Hirasawa, S.Miyarnoto, and A. Ohsawa, J. Chromatogr., 169, 117 (1979).
(1) (2) (3) (4) (5)
.
-
,
\
RECEIVED for review July 18, 1979. Accepted December 10, 1979.