possible to explain our results qualitatively from knowledge of the differences in pore structure of the two materials. Mercury porosimetry data (30) for Porasil beads indicate a relatively narrow distribution of pore sizes. The slopes of the volume-pressure curves are steep, suggesting that once mercury has penetrated a pore opening it takes a very small increase in pressure to force it all the way into the pore network. On the other hand, mercury porosimetry data for Styrageltype particles (31) indicate a much broader range of pore-size distribution. Scanning electron micrographs furnished by Godwin (32) and Moore (31) indicate very great differences in pore structure between Styragel and Porasil. While the range of sizes of pore openings is narrower in Porasil, these openings are frequently followed by enlargements down the pore channel within the particle. In addition, large cavities are encountered within Porasil beads, with many interconnected pore channels leading into them. Styragel particles have a wider range of pore sizes but a structure that does not exhibit as many gross irregularities as in Porasil. Basically, the deep internal pore structure of the glass beads is much more accessible from the pores near the surface of the bead than in the case of the poly(styrene-divinylbenzene) beads. Therefore, once a molecule enters a pore in Porasil, it has more freedom to continue diffusing toward the center of the particle than in Styragel. Such diffusion greatly increases the observed dispersion over that due to mobile-phase effects. We feel that at normal GPC flow rates, because of such differences in pore structure, dispersion in these porous glass bead columns is primarily controlled by permeation (mass-transfer) effects, while in the poly(styrene-diviny1benzene)-packed columns dispersion is primarily controlled by mobile-phase effects oc-
curring outside the gel. These results demonstrate the importance of understanding both the origin of mobile-phase dispersion and the influence of pore structure on dispersion in order to tailor such size-separation techniques for maximum efficiency. While the broadening associated with Porasil is greater than that with Styragel, porous glass packings may offer significant practical advantages such as structural rigidity, chemical inertness, and compatibility with selected solvents, in comparison to other materials. The direct comparison of our dispersion data obtained o n Styragel and Porasil columns is difficult, since the particle size of the Porasil packing used was approximately two to three times as large as that of the Styragel. The larger particle size leads to greater interparticle diffusion distances and therefore greater dispersion. In addition, pore depth is much greater in the Porasil system, leading to greater broadening effects due to permeation. We expect that the level of dispersion with a Porasil system can be significantly reduced by going to smaller gel particle sizes and lower flow rates. ACKNOWLEDGMENT
We thank G . W. Johnson and J. C. Moore for many helpful discussions, and R. W. Godwin and J. C. Moore for allowing us to examine unpublished scanning electron micrographs.
(30) A. J. de Vries, M. LePage, R. Beau, and C. L. Guillemin, ANAL. CHEM.,39, 935 (1967). (31) J. C. Moore, Dow Chemical Co., Freeport, Tex., private communication, 1969. (32) R . W. Godwin, Celanese Fibers Co., Charlotte, N. C., private communication, 1969.
RECEIVED for review September 26, 1969. Accepted November 17, 1969. Presented in part at an ACS Division of Petroleum Chemistry Symposium on Gel Permeation Chromatography, Houston, Tex., February 1970. Work supported by The Dow Chemical Co. and Waters Associates. One of us (R.N.K.) thanks Hercules, Inc., for fellowship support. Work performed in Rensselaer’s Materials Research Laboratory, a facility supported by the National Aeronautics and Space Administration.
Application of the Trichloroacetyl Isocyanate Reaction with Terpene Alcohols to Quantitative Fractionation of Essential Oils P. A. Hedin, R. C. Gueldner, and A. C.Thompson Etitomology Research Dirision, Agricultural Research Seroice, Utiited States Department of Agriculture, State College, Miss. 39762
The quantitative reaction of trichloroacetyl isocyanate (TCAIC) with alcohols was used to fractionate essential oils. The TCAIC esters were extracted from the oil with dilute alkali in aqueous methanol with concomitant hydrolysis to carbamates. When the alcohols were regenerated by refluxing with 10% KOH in 80% aqueous methanol the recovery was essentially quantitative except in the case of the tertiary sesquiterpenoids. The method was evaluated with 15 alcohols, a phenol, a sterol, a synthetic mixture, and cotton bud essential oil.
IN THE COURSE of our investigations of the essential oil of the cotton plant (Gossypium hirsutum L. var. Deltapine Smoothleaf) for constituents attractive to the boll weevil, Anthonornus grandis Boheman, we found several tertiary alcohols that were
difficult to separate from other oxygenated classes by column chromatography and were dehydrated during GLC. Several procedures that involve masking or reaction with the hydroxyl function have been reported. Hefendehl ( I ) added boric acid to the stationary phase of gas-liquid chromatographic (GLC) columns to achieve selective removal of Primary and secondary alcohols; Grechukhina and Nesmelov (2) converted primary and secondary aliphatic alcohols to the corresponding alkyl nitrites; and Larkhan and Pagington (3) (1)
F. w.Hefendehl, Nuturwissenshuftetz,
51,
138 (1964).
(2) F. N. Grechukhina and V. V. Nesrnelov, Zh. Prikl. Khim., 39,
2574 (1966). w. Larkhan and (1967).
(3) T,
J.
s. Pagington, J . C/lromatogr., 28, 422
ANALYTICAL CHEMISTRY, VOL. 42, NO. 3, MARCH 1970
403
subjected the chloroacetate derivatives to GLC. Additional procedures were reviewed by Guenther et a/. (4). Also, Goodlett (5) used trichloroacetyl isocyanate (TCAIC) t o esterify hydroxy compounds for characterization of alcohols by nuclear magnetic resonance, and his procedure has since been used by us routinely for assignment of the hydroxyl proton resonance of isolated terpene alcohols. In a preliminary study, TCAIC was added to the cotton essential oil, and the reaction mixture was analyzed without further treatment by GLC. A reduction in the number of peaks was observed; however, removal of the TCAIC esters from the oil with subsequent regeneration t o the original alcohols promised additional valuable information. This report describes the development of a quantitative procedure for the solvent extraction of terpene alcohol TCAIC esters from the essential oil of the cotton plant, concomitant hydrolysis t o carbamates and the conditions for quantitative regeneration (saponification) of the alcohols. The carbamate intermediate was characterized, and the development of appropriate G L C and thin layer chromatographic (TLC) systems for analysis of the products of the reaction is discussed. EXPERIMENTAL
Apparatus and Reagents. The trichloroacetyl isocyanate was reagent 3937 from Eastman Organic Chemicals and was used without further purification. Melting points are uncorrected values obtained with a Fisher-Johns block. C, H , N analyses were performed by Galbraith Laboratories, Inc., Knoxville, Tenn. Infrared spectra (in cc14 and KBr pellets) were obtained with a Beckman IR-5A recording spectrometer. PMR spectra (in cc14 and D-6 DMSO) were determined on a Varian A-60 recording spectrometer. G L C separations were performed on an Aerograph A-95-P-3 FID instrument by using the following columns and conditions: A, a 0.0032- X 1.220-113 stainless-steel column packed with 30% Carbowax 4000 on 60/80 mesh Chromosorb P treated with HMDS. Carrier gas flow NS at 48 ml/min, column temperature 120 OC, injector 170 "C, detector 190 OC; B, a 0.0032- X 6.145-m stainless-steel column packed with 10% SE-30 on 60/80 mesh Chromosorb W treated with HMDS. Carrier gas flow N I at 50 mlimin, column temperature 175 "C, injector 185 O C , detector 190 O C . The thinlayer chromatographic plates were 20 X 20 cm glass, 250 p depth, developed in the ascending fashion in solvent vapor saturated chambers to 15 cm on a support of 9 parts of Silica gel (plain) (Warner-Chilcott Laboratories, Richmond, Calif.) and 1 part of calcium sulfate (anhydrous powder). Irrigants were 100 % 2-chloropropane and several specified mixtures of methylal and pentane. Color development was achieved by heating at 100 O C for 3-5 minutes after spraying with 3 vanillin in 0.5 % concd. H 2 S 0 4absolute methanol. The alcohols were procured commercially and purified by preparative chromatography. Procedures. TCAIC was added in slightly excess molar quantities to terpene alcohols in spectral grade cch. The mixture was allowed to stand for 5-10 minutes and extracted 5 times with 2 % KOH in 80% meth?n01-20% water. The concentration of K O H in the combined extract was adjusted to 10% by the addition of pellets. Primary, secondary, and tertiary alcohols were refluxed for 2, 6, and 48 hours, respectively. After the reaction mixtures cooled, they were extracted 5 times with CCL,, and the combined CC14 extracts were adjusted to a suitable volume for G L C determination of percentage recovery. The standard was the original, untreated sample similarly diluted. In a typical analysis, 0.1
ml of the alcohol in 5 ml of CCL4 was mixed with 0.2 ml TCAIC. For the G L C analysis, 1.0 pl of a 2 5 4 CC14 solution containing 0.1 ml of the alcohol was tested, and the results were compared with the standard by calculating peak areas and peak heights at the same retention time. Procedural modifications were required with eugenol, which could not be back-extracted into CC14 from the alkaline phase until after acidification, and with 1-hexadecanol carbamate and @-sitosterolcarbamate, which remained in the cc14 phase but were successfully saponified after removal of the solvent and subsequent suspension in the described alkaline media. The formation of the TCAIC ester, the conversion to the carbamate intermediate, and the progress of the saponification were monitored by TLC. Also, physical data for three selected alcohol carbamates were collected to demonstrate the formation of this intermediate. Ethyl Carbamate. Absolute ethanol (1 ml) in 25 ml c c l 4 and 2 ml TCAIC were mixed at room temperature, and the reaction mixture was extracted repeatedly with 2% K O H in 80% aqueous methanol. The combined extract was concentrated and subjected to preparative TLC. The white, visible band at R, 0.3 (silica gel-2-chloropropane) was eluted and recrystallized from benzene (yield 2.0 grams), m.p. 48-50 "C. PMR analysis (D-6 DMSO) showed ppm (6) 1.33 t (3, 7.5); 4.24 q (2, 8.0); 5.68 s. br (2) and was identical with that of an authentic sample (Fisher Scientific Co. U-26). The I R spectrum in CC14 included v,"-.~ 1064, 1320, 1365, 1400, 1580, 1690 (C=O), and 3250 (-NHJ. /-Menthyl Carbamate. /-Menthol (1 gram), 25 ml CCl,, and 2 ml TCAIC were mixed at room temperature. The reaction mixture was extracted repeatedly with 2% KOH in 80% aqueous methanol. The combined extract was concentrated to remove most of the methanol and subjected to preparative TLC. The white visible band at Rf 0.3-0.5 silica gel-2 chloropropane) was eluted and recrystallized from benzene (yield 1.8 grams), m.p. 166-7 "C. A m / . Calcd for Cl1H21O2N: C, 66.29; H, 10.62; N, 7.02. Found: C, 66.26; H , 10.65; N , 7.06. PMR analysis (in D-6 DMSO) showed ppm (6) 1.09 q (9); 0.70-2.40 (9); 4.68 s, br (1); and 6.75 s (2). The I R spectrum (KBr pellet) included v,,,~, 778, 826, 905, 1032 s, br, 1319 s, br, 1585-1660 (C=O), 2820, and 3130 ("2). Eugenyl Carbamate. Commercial eugenol (1 ml), 25 ml CC14, and 2 ml TCAIC were mixed at room temperature. The reaction mixture was extracted repeatedly with 2 K O H in 80% aqueous methanol. The combined extract was concentrated and subjected to preparative TLC. The white visible band at R, 0.5-0.7 (silica gel-methylal) was eluted and recrystallized from benzene, m.p. 115-6 "C. Anal. Calcd for CllH1303N: C, 63.75; H, 6.32, N, 6.76. Found: C, 63.72; H, 6.30; N, 6.79. PMR analysis (in D-6 DMSO) showed the following: 3.41 d(2,6.5), 3.82 s(3), 5.05 s, br (l), 5.25 s, br (l), 6.00 m, br (l), 6.72-7.14 m (5) [7.00 s(2) amide], The IR spectrum (KBr pellet) included v,,,, 918, 980, 1030, 1113, 1145, 1200, 1265, 1370,1490,1590,1690(C=O), and 3260 (-"I). Analysis of Mixtures. Commercial samples of B-pinene, citronellal, /-methone, myrcenyl acetate, citronellol, linalool, and /-menthol were purified by preparative GLC. A test mixture was prepared with 0.1 gram of each diluted to 5 ml in CC14. For analysis, 1.0 ml of the CCl, solution was treated with TCAIC as previously described and analyzed by GLC. Cotton bud essential oil (6) was fractionated on a 2.0- X 20-cm water-jacketed column. The support was silica gel coated with Carbowax 20M (0.75%, w/w) (6). The hydrocarbons were removed by elution with pentane, and the OXY-
(4) E. Guenther, G. Gilbertson, and R. T. Koenig, ANAL.CHEM., 41, 40 R (1969). ( 5 ) V. W. Goodlett, ibid.,37,431 (1965). 404
ANALYTICAL CHEMISTRY, VOL. 42, NO. 3, MARCH 1970
~-
(6) J. P. Minyard, J. H. Tumlinson, A. C. Thompson, and P. A. Hedin, J . Agr. Food C/7em.,15, 517 (1967).
Table I. Chromatographic and Hydrolysis Data for Selected Terpene Alcohols and Related Compounds TLC-SGGb GLCa Compound Alcohol Ester Carbamate Recovery,c SE-30 C4000 Geraniol 1234 1805 0.33 0.69 0.18 96 Nerol 1221 1770 0.36 0.72 0.19 96 Citronellol 1124 1765 0.30 0.63 0.21 98 Borneol 1177 1732 0.30 0.57 0.12 99 I-Menthol 1173 1671 0.31 0.57 0.14 95 a-Terpineol 1201 1662 0.23 0.64 0.14 90 4-Terpinenol 1160 1652 0.43 0.70 0.19 91 Linalool 1052 1558 0.32 0.50 0.13 93 t r a m , trails-
Farnesol 1745 2342 0.27 0.49 0.12 Nerolidol 1540 2049 0.37 0.77 0.14 a-Santalol 1660 2312 0.33 0.50 0.17 Cedrol 1616 2105 0.38 0.67 0.18 fi-Bisabolol 1666 2148 0.61 0.86 0.29 1-Decanol 1280 1762 0.28 0.64 0.21 1-Hexadecanol 1710 2372 0.32 0.65 0.18 Eugenol 1344 2135 0.52 0.36 0.10 d fi-Sitosterol ... 0.25 0.43 0.15 Ik (Kovgts indices) values for conditions described in Experimental-e.g., pentadecane = 1500, hexadecane = 1600 (7). R J in silica gel G-2-chloropropane system. Recovery of alcohol after esterification and hydrolysis. Not analyzed.
96 92 100 80
75 92 101 94
...
Q
genated terpenes were subsequently eluted with methylalpentane ( 5 :95). The oxygenated fraction was treated in the same manner as the test mixture, and each fraction was investigated by GLC. RESULTS AND DISCUSSION
TCAIC was observed t o react rapidly and quantitatively with 8 monoterpene alcohols and 5 sesquiterpene alcohols. All chromatographic and recovery data are summarized in Table I. The reaction also occurred with a sterol, two aliphatic alcohols, and a phenol, though in this last instance, it proceeded more slowly and could be monitored by repetitive P M R sweepings of the hydroxyl proton shift. Conditions for successful gas chromatography of the esters could not be developed, probably because of the large mass and the content of chlorine. However, their TLC R, values and vanillin reagent colors were similar t o the related acetate esters. Peak positions of the N H proton were observed for several esters a t 9-10 ppm, as reported by Goodlett (5). I n initial efforts t o devise a procedure for separating the TCAIC esters from other components expected t o be present in the essential oil, chromatographic techniques were emphasized. TLC indicated that the esters were partially converted t o components of slightly higher polarity than the original alcohols and that alkaline conditions favored this conversion. Because elemental and infrared analysis of the intermediate showed no chlorine, the loss of trichloroacetic acid (K+ salt) was presumed. Further investigation with ethanol, (-menthol, and eugenol demonstrated that in each instance the respective ester was converted to the carbamate quantitatively by a n excess of dilute alkali in aqueous methanol. I n the process, the carbamates were quantitatively extracted into the aqueousalcoholic phase from CC14, but the hydrocarbons, esters, carbonyls, and several oxygen ring compounds remained almost exclusively in the less polar phase. TLC R, values for the carbamates were generally 50-75 % of the free alcohol and of similar hue after diagnostic color development. P M R .~
(7) E. sz. Koviits, 2. A n d . Clzem., 181,351 (1961).
shifts of the amide proton were generally located between 5.7 and 7.0 ppm. Hydrolysis of the carbamates was best achieved by refluxing with 10% KOH in 80% aqueous methanol. Hydrolysis of the primary and secondary carbamates was complete in 2 and 6 hours, respectively; hydrolysis of the Clo tertiary carbamates was complete in 8-24 hours; hydrolysis of the CIStertiary carbamates reached a n optimum of about 75% in 1-2 days. With the carbamate of cedrol, a saturated tricyclic alcohol, 80 % recovery was achieved with saturated K O H in 50% aqueous methanol, but two unsaturated tertiary alcohol carbamates (nerolidol and P-bisabolol) were extensively rearranged and dehydrated at this higher refluxing temperature. A special case, a phenol (eugenol), was amenable to the sequential procedures but could not be recovered from the alkaline medium until after acidification. Also, 1-hexadecanol carbamate remained in the CCl, phase but was quantitatively hydrolyzed after solvent replacement. @-Sitosterol was quantitatively converted to the TCAIC ester (TLC inspection) which was converted to the carbamate after treatment with alkali but remained in the CC1, phase; saponification for 6 hours by the usual procedure regenerated the free alcohol. The applicability of the procedure to the fractionation and analysis of essential oils was demonstrated by the results obtained with a test mixture and with a fraction of cotton bud essential oil from which the hydrocarbons had been removed. The test mixture contained P-pinene, citronellal, linalool, I-menthone, myrcenyl acetate, /-menthol, and citronellol. Figure 1 gives the G L C profile analysis. The alcohols (peaks 3, 6, and 7) disappeared after treatment with TCAIC (Figure 1B). When the Ccl4 reaction mixture was treated with alkaline aqueous methanol, P-pinene, citronellal, I-menthone and myrcenyl acetate were retained in the CCl4phase (Figure IC). When the contents of the alkaline aqueous methanol layer were saponified, the alcohols were regenerated (Figure 1 0 ) . Myrcenyl acetate was observed t o isomerize on prolonged exposure to TCAIC; the apparent product, which possessed a shorter retention time, was cis-l,1,5-trimethyl4,6-heptadien-l-ol acetate. This isomerization could not be
ANALYTICAL CHEMISTRY, VOL. 42, NO. 3, MARCH 1970
405
CCL,
1
a
OR~GINAL MIXTURE TCAIC
+
LAYER
0.REGENERATED ALCOHOLS
Figure 1. GLC profiles of a synthetic mixture before and after reaction with TCAIC 20 ft X 1/8 in. Clooo at 175 "C. 1 = pinene, 2 = citronellal, 3 = linalool, 4 = I-menthone, 5 = myrcenyl acetate, 6 = Imenthol, 7 = citronellol
I l U t MIN I r
Figure 3. GLC profiles of the untreated oil fraction from Figure 2 (CCI4 layer) and the regenerated alcohols 20 ft X 1/8 in. Cloo0at 175 "C
1
1
G
1
'I
b. ' : '
22-
? ,,
2 , I , . >I ---I
B
3)
,b
3
llMt MINUltS
Figure 2. GLC profiles of the hydrocarbon-free cotton bud essential oil before and after reaction with TCAIC 20 ft X 1/8 in. C ~ OatO 175 O "C
induced by trichloroacetic acid and was not observed with any of several other terpene esters. Addition of TCAIC to the oxygenated fraction of cotton bud oil caused a number of peaks to diminish or disappear. Figure 2 gives the G L C profile of the oil before and after treatment. Figure 3 gives the G L C profiles of the CCl, layer (nonalcohols) and the regenerated alcohols. It is not the purpose of this report to identify all the alcohols, but several are known, and identification of the others is currently in process. The C5 and C6 alcohols include peaks through number 6. Only small quantities of Cloalcohols are present in cotton bud oil, and they are found submerged in peaks 14-24. The CISalcohols are present in peaks 28-35, P-bisabolol is the major component (8). Some anomalies should be explained. The addition of TCAIC to the cotton bud oil (Figure 2) caused 3 peaks (19, 20, and 21) t o increase greatly in intensity. They were
also produced by a purified sample of P-bisabolol treated with TCAIC and then analyzed by GLC. Also, t h e same 3 peaks could be produced by refluxing fi-bisabolol with Woelm's Alumina in xylene. Peak 19 had been previously identified by the alumina dehydration procedure as 7,8-dihydrocurcumene, and peak 20 identified as a mixture of 8,ll-dihydrocurcumene and y-bisabolene (8). Peak 21 has not been identified but it possesses the expected retention time for y-curcumene. Dehydration is minimized if the excess TCAIC is reacted with methanol before analysis (Table I, Figure 3). Cotton bud essential oil contains several oxides, one of which has been identified (P-caryophyllene oxide, peak 27a) (9). P-Caryophyllene oxide was shown by Nigam and Levi (10) t o be degraded as a function of temperature and acidity during GLC. TCAIC was also shown t o enhance degradation of a synthetic sample of P-caryophyllene oxide in this study, but this effect was minimized if the excess reagent was reacted with methanol before analysis. I n summary, TCAIC reacted quantitatively with every alcohol investigated. Some dehydration of tertiary alcohols occurred, one tertiary ester isomerized, and the rearrangement of one epoxide was enhanced, but these effects can be minimized by neutralization of the excess reagent before analysis. Thus, the TCAIC esters can be converted t o carbamates with alkali, and separated from most o r all of the other oxygenated oil by solvent partitioning, so the original alcohols can be recovered in high or quantitative yields by saponification.
(8) J. P. Minyard, A. C. Thompson, and P. A. Hedin, J . Org. Chem., 33, 909 (1968).
406
ANALYTICAL CHEMISTRY, VOL. 42, NO. 3, MARCH 1970
RECEIVED for review October 23, 1969. Accepted January 7, 1970. Paper presented at 158th ACS National Meeting, Division of Agricultural and Food Chemistry, New York, N. Y . ,Sept. 7-12, 1969. Mention of a proprietary product in this paper does not constitute an endorsement of this product by the U. S. Department of Agriculture. (9) J. P. Minyard, D. D. Hardee, R. C . Gueldner, A. C. Thompson, and P. A. Hedin, J . A g r . Food Chem., 17, 1093 (1969). (10) I. C. Nigam and L.Levi, J . Org. Chem., 29, 2803 (1964).