I n many instances of electron absorption the final product may be the ionradical itself, and its stability may play a cardinal role in the extent of electron absorption. Where cleavage to even more stable products can occur, as was probably the case with tert-butyl bromide and tert-butyl iodide, affinities are more enhanced. Trapping and identification of the products formed in the detector would shed more light on these reactions. PULSCR METHOD.When sensitivities u ere determined by the puleer technique certain differences a ere noted when compared to the results obtained by the d.c. method. From Table VI it can be seen that tert-butyl chloride is now seven times more sensitive than sec-butyl chloride, n hile tert-butyl bromide is 230 times as sensitive as sec-butyl bromide. Sensitivity of the pulser technique depends more heavily on the formation of stable products. This would explain the enhanced affinity of the tertiary halides including the chloride. The rffect of initial electron attack on over-all sensitivity of the system therefore must be small when using pulsed voltage. This is borne out by the inability of n-butyl chloride and isobutyl chloride to give an effective response under the conditions of the experiment. Undoubtedly all four chlorides are attacked to a similar extent by electrons
but only the sec- and tert-isomers furnish the driving force for effective capture by virtue of the more stable radicals they can form. Our knowledge of these mechanisms is still limited and study of more varied types of structurally related compounds is required to support more firmly some of the proposed concepts. This study can act as a model for predicting the relative sensitivities of more complex molecules, regardless of functional group. An extension of present knomledge to compounds containing other electronegative heteroatoms, particularly oxygen, can now be made easily. I n the field of biochemistry a similar study of steroids mould appear to hold great promise in the elucidation of certain structural details. These compounds contain mainly hydroxyl and keto groups but the number of structural variations surrounding these moieties is indeed large. RIoreover, the electron capture detector should have very high sensitivity for some highly oxygenated steroids making this system very useful in this respect as well. Finally, a comparison of the affinity for electrons of certain steroids with their biochemical activity may be possible as has been achieved with Krebs cycle members (4). The present work also indicates that it Kould be desirable to tailor-make certain derivatives that would impart high
electron affinity to compounds which ordinarily have little or no capacity for free electrons thereby greatly augmenting their detection by this technique. The use of polyhalogenated derivatives could conceivably yield sensitivities even higher than the 3.0 X 10-l6 mole per second herein reported for tert-butyl iodide. The use of such high affinity compounds coupled with optimum operating conditions would then make this system the most sensitive mode of detection for organic vapors to date. LITERATURE CITED
(1) Clark, S. J., Division of dgricultural and Food Chemistry, .4CS, 140th
Meeting, Chicago, Ill., September 1961. (2) Goodwin, E. S., Goulden, R., Reynolds, J. G., Analyst 86, 697 (1961). (3) Lovelock, J. E , ANAL. CHEM.33, 162 (1961). (4) Lovelock, J. E., Nature 189, 729 (1961). (5) Lovelock, J. E., Lipsky, S. R., J. Am. Chem. SOC.82, 431 (1960). (6) Lovelock, J. E., Zlatkis, A., ANAL. CHEM.33. 1958 11961). ( 7 ) Lovelock, J. E., Zlatkis, 8., Becker, R. S., Nature 193,540 (1962). RECEIVEDfor review March 5 , 1962. Accepted April 16, 1962. Division of Analytical Chemistry, 141st Meeting, ACS, Washington, D. C., March 1962. Work supported by grants from the National Heart Institute (H-3558) of the U. S. Public Health Service, the National Dairy Council and the Nutrition Foundation.
New Peak-Shift Technique for Gas-Liquid Chromatography Preparation of Derivatives on the Column M. W. ANDERS and G. J. MANNERING Oeparfmenf o f Pharmacology and Toxicology, University o f Wisconsin, Madison 6, Wis.
b A peak-shift technique for gasliquid chromatography is described whereby derivatives are formed directly on the column by following the injection of the parent compound with an injection of either acetic or propionic anhydride. Acetates and propionates can be formed simultaneously by injecting a mixture of the anhydrides. In many cases, derivatives are formed with alcoholic or phenolic hydroxyl groups and with primary and secondary amines. The method is applicable particularly to the characterization of alkaloids and steroids. Trifluoracetates and trimethylsilyi ethers have also been formed on the column.
730
ANALYTICAL CHEMISTRY
studying the metabolic fate of codeine in man, an attempt was made to apply gas-liquid chromatography to the quantitative estimation of codeine and its metabolites, norcodeine and morphine, excreted in urine. Codeine and norcodeine could not be separated under conditions similar t'o those described by Lloyd et al. (5). HoFYever, when a mixture of the two alkaloids in ethyl acetate was warmed with equal volumes of acetic anhydride and pyridine and t'hen injected, excellent resolution was observed. The rather improbable idea then presented itself that one might achieve direct esterification of these compounds by immediately follo\\-ing HILE
their injection onto the column with an injection of acetic anhydride. First results employing morphine were sufficiently gratifying to encourage the extension of the study to include a number of other esterifiable compounds, particularly alkaloids and steroids, the latter being of special interest because the gas chromatographic behavior of the acetates of certain of these compounds has already received attention (6, '7). I t was also found that propionic acid esters could be formed on the column in a similar manner and, indeed, that simultaneous formation of acetates and propionates occurs when a mixture of acetic and propionic anhydrides is injected.
~~
Table
Compound Amphetamine Pseudoephedrine Ephedrine Lupinine 1-Naphthol p-Phen ylphenol
I.
Column Temp., O C. 125 125 125
pl. of 0.57, Solution 0.5
150 150 215 215 215 215 215 215 215 215
0.5 1.0
N-Methyl-3-methoxymorphinan
3-Methoxymorphinan
.Y-Methyl-3-hydroxymorphinan
3-Hydroxymorphinan N-Allyl-3-hydroxymorp hinan Xorcodeine Codeine Morphine Dihydromorphinone ~V-~Methyl-2,3-dihydroxyniorphinan Thebaine Androsterone 17-Methyl-androstene-3,17-d~01
Estrone 17p-Estradiol Testosterone Methyltestosterone Progesterone Cholestane Sarsasapogenin Diosgenin Tigogenin Digitoxigenin a
Peak-Shift after Treatment with 5
150
215
215 215 235 235 235 235 235 235 235 235 250
250 250 275
2.0 4.0 1.o
1 .o 0.8
1.5 1.2 3.0 1.5
1 .o 6.0 5.0 5,O 2.0 1 .o
2.0
1.5
2.0
1.0 1 .o
2.0
3.0 2 0 3.0 3.0 5.0
PI. of Acetic Anhydride
With Acetic Anhydride Without Acetic Area, sq. cm., Anhydride of unreacted Retention time, min. Retention Time, parent (area sq. cm.) of Min. (area, sq. cm.j compound shifted peaks 2.8(7.0) 0 18.7(4.8) 8.7(9.8) 0.6 55.5(5.2) 9,2(21,0) 0 63.0(10.8) 4.8( 11. O ) 9.0 8.6(0.2j 6.8(6.7) 3.4 9.1(2,0) 16.3(12.1) LI 2.7 23.3(6.0) 7.0(6 . O ) 6.0 7.3(9.5) 7.4 9.0(9.7) 10.5( 1 , 6 ) 7.7 9.1(11.8) 0 30.8(1 . 7 ) , 3 3 . q 5 . 5 ) 13.0(14.7) 3.0 15.4(6.8) 0 14.1(8.3) 41,2(5.8) 14,3(18.2) 16.6 20.910.3) 16,5(15.8) 0 20.3(22.5),~ 29.'2(11.5) 17.2(2.2) 0 25.0(2.3) 18.7(5.6) 0.7 24.2(8.5) 20.2(15.3) 15.3 9.2(9.7) 7.6 11.8(0.8) 10.9(13.3) 15.7(0.7) 13.9 11.7(13.2) 15.3( 10.9) 2.3 12.3(13.1) 0.8 16.4(9.6) 12.4(10.4) 8.6 16.8(0.6) 13.q7.0) 7.5 19.9(15.4) 15.8 24.3(19.9) 22.0 32.7(25.9) 19.0 42.0(3.5) 11.0 47.7(5 . O ) 34.7(1 8 , 4 ) 47.0(4.4) 11.2 35.0(1 7 , 8 ) 23.7(21,7) 28.0(0.6) 16.9 (1
(1
No derivative formed.
Langer and Pantages (2) have used the term "peak-shift technique" to denote the method whereby volatile compounds are converted t o volatile derivatives and subsequently examined b y gas-liquid chromatography. Others (4, 6, '7) have prepared derivatives outside the column and examined them later by gas-liquid chromatography. APPARATUS A N D PROCEDURE
A Barber-Colman Model 10 gas chromatograph equipped with a Sr90 argon ionization detector was used. The column was a U-shaped borosilicate glass tube, 7 mm. i.d. and 6 feet long filled with a solid support of acid- and base-washed Gas-Chrom S, 80- to 100mesh, treated with dimethyldichlorosilane and coated with a liquid phase of the methyl silicone polymer, SE-30, which was applied by the solution Two hundred fifty technique (1). milliliters of a 2% solution of SE-30 in toluene (w./v.) vias used for 25 grams of Gas-Chrom S. The column m-as conditioned overnight a t 300" C. The detector was operated at 1000 volts and the relative gain was 10-7 ampere. Column temperatures were 125", 150", 215", 235", 250", or 278" C. and the corresponding flash heater temperatures and argon flow rates were 198", 214", 260", 276", 285", and 302" C. and 300, 250, 185, 173, 160, and 150 ml. per minute, respectively. The inlet pressure was always 20 p.s.i.g. and the outlet pressure was atmospheric. The samples mere 0.5% solutions (w./v.)
of each compound in methanol, ethyl acetate, or tetrahydrofuran. Unless otherwise noted, the injection of the anhydride followed t h a t of the sample within 5 seconds. Areas were measured with 4 planimeter. RESULTS A N D DISCUSSION
The peak-shifts effected by the injection of 5 p l . of acetic anhydride are given in Table I. As might be expected, the degree of esterification varied from compound to compound. Where no esterification was possible, as with thebaine, progesterone, and cholestane, no peak-shift was noted. Amphetamine, ephedrine, 3-hydroxymorphinan, and morphine reacted completely to form derivatives, while methyltestosterone was not altered. However, methyltestosterone possesses a tertiary hydroxyl group which would require more vigorous treatment for acetylation. The reactivity of the other compounds tested varied widely between these two extremes. However, the degree of reactivity of a given compound was sufficiently constant from trial to trial as to offer another parameter for identification. The retention times of the derivatives were always greater than those of the compounds from which they were derived. k e a s of the shifted peaks often varied considerably with respect to the areas of the corresponding parent compounds. This effect was particularly marked with
primary and secondary amines where the areas of the derivative peaks were much less than those of the unacetylated compounds. I n contrast, the acetylation of morphine and estrone resulted in increased areas. Such variations in molar response have been noted previously when the argon ionization detector was employed ( 5 ) . While three derivatives are frequently possible when two esterifiable groups are present, in no instance were more than two derivative peaks observed. However, the possibility must not be ignored that two derivatives could be represented by a single peak. I n certain cases when two hydroxyl groups were present, e.g., digitoxigenin and N methyl-2,3-dihydroxymorphinan,only a single peak was obtained. When two derivatives were formed, the areas of their peaks usually varied greatly. No attempts were made to characterize the derivatives except in the case of morphine where the retention time of the first peak corresponded to t h a t obtained with authentic 6-acetylmorphine and the second peak was identical with that obtained with diacetylmorphine. The possibility exists that the first peak obtained with morphine might represent a mixture of 3-acetyl- and 6-acetylmorphine but this seems unlikely in view of the instability of the 3-acetyl derivatives. However, the usefulness of this technique does not depend upon the isolation and VOL 34, NO. 7, JUNE 1962
731
identification of the derivatives but upon bhe relationships of t'he retention times of t,hese derivatives. The speed with which the injection of acetic anhydride followed that of the sample was not critical within seconds. Acetylation still occurred after intervals between injections of 2, 5 , and 10 minutes. However, as the intervals increased, the retention times of t'he shifted peaks decreased as was expect'ed, since after esterification the products had shorter distances to travel. Derivative peaks were also formed n-hen the anhydride and compound w r e mixed previous to injection, but problems of mixing and dilut'ion, which niay be considerable when only small volumes are available, are minimized n-hen separate injections arr employed. Pyridine did not alter the peak pattern rvhen injected with acetic arihydride. Favorable conditions for esterificat,ion ivould seem to exist on the column nithout the aid of a catalyst. The success achieved in carrying out esterification reactions on the column may be att'ributed to the fact that' the reaction products are being removed cont'inuously from the reactants, thus overcoming unfarorable equilibrium conditions. The high temperatures employed, along with the inert environment provided by argon, might also be expected to favor the reaction. The effect of t'he quant'ity of the injected anhydride was st'udied. Complete conversion of morphine occurred with as little as 0.2 pl. of acetic anhydride and about 50% conversion was obtained with 0.05 pl. Larger qiiant'ities of the anhydride effected quantitative changes in the acetylation p t t e r n . When 50 pl. of acetic anhydride was used, t'he area corresponding to the diacetyl derivatiw incrtlased Jvith a conconiitant decrease in tlir area of the rnonoacetyl peak.
Table II.
Conipound hlorphine
10.~
MOHPHINE
10-7
MORPHINE bCETlC
I
10-7MORPHINE
I
O
c.
+
PROPIONIC
ANHYDRIDE
MORPHINE C bCETIC ANHYDRIDE PROPIONIC 4 I
-MINUTES
Figure 1. Peak-shift of morphine after treatment with acetic and propionic anhydride Top chromatogram:
30 pg. of morphine
Second chromatogram: 30 pg. of morphine foilowed b y 5 pl. of acetic anhydride Third chromatogram:
Bottom chromatogram:
pl. of 0.5%
Solution
30 pg. of morphine followed b y 5 pl. of propionic anhydride 30 pg. of morphine followed b y a mixture of 5 pl. of acetic anhydride 'and 5 PI. of propionic anhydride
PI. of Acetic Anhydride or 50 PI. of Propionic Anhydride
With Acetic Anhydride Area, sq. Kithout Anhydride em. of Retention Retention unreacted time, min. Time, Min. parent (area, sq. em.) (area, sq. cm.) compound of shifted peaks
215
6
16.6(19.2)
Androsterone 17-Methylandrostene3,1 7-dioI Estrone I 7P-Estradiol
235
1
9.9(5.1)
2.1
235 235 235
1 2 2
10.9(6.6) 11.7(12.7) 12,3(11.6)
2.7 0 0
Testosterone Methyltestosterone Cholesterol Tigogenin Sarasapogenin Diosgenin
235 235 250 250 250 250
13,2(9. O )
Iligit osigpnin
2i5
1 1 8 3 2 3 5
3.0 8.7 4.4 8.9 9.6 7.8 3.8
732
ANALYTICAL CHEMISTRY
f
ANHYDRIDE
I I
Peak-Shift after Treatment with 50
Column Temp.,
+
bNHYDRIDE
By comparing a number of compounds which were reacted with 5 and 50 p1 of acetic anhydride (Tables I and 11): the use of the larger quantity of acrt>-lating agent nearly always gave larger yields of derivatives. It is conceivablr that even greater esterification might, b t realized by employing still larger quantities of anhydride. JTith either acetic or propionic anhydride, surprisingly little tailing of the anhydride occurred even when 50-pl. volumes were employed. The column suffered no apparent deleterious effects as a result of these repeated injections and the balance current varied little from day to day. I n Table I1 i t may be seen that propionic anhydride is as effective an esterifying agent as acetic anhydride and in some cases it is even more effective. When a mixture of acetic and propionic anhydrides is injected, both acetates and propionates are formed simultaneously. The pattern of peaks that resulted when morphine was subjected to the mixture of anhydrides is shown in Figure 1. When til-o hydroxyl groups occur on the same molecule, as is the case with morphine, a mixed ester may result. This is the most probable explanation for the extra peak seen a t 39 minutes in the bottom chromatogram. Preliminary experiments indicate that trifluoracetates can be formed on the colunin. Khen the injection of morphine, .Y-methyl-3-hydroxymorphinan, or estrone was followed by 5 p l . of trifluoracetic anhydride nith the detector operated a t 1'750 volts, complete conversion was observed. As would he expected from the results obtained by others (6), the derivatives had retention times shorter than those of t,he parent compounds. Attempts to form the trimethylsilyl ethers ( 2 , 4 ) by injecting hexamethyldisilazane, with or without triniethylchlorosilane as a catalyst, were not successful under the
14.2(8.5)
24.2(20,4) 33.0(43.8) 33.0(30.5) 33.5(27.5) 24.7(23.1)
0
With Propionic____ Anhydride Area, sq. em. of Retention unreacted time, nun. parent (area, sq. cm.) compound of shifted peaks
20.3(11.5) 29.5(15.8) 12.6(2.8)
0
15.6(3.3) 15.2(1 7 . 3 ) 16.0(5 . 0 ) 21.5(5.5) 17.7(3,5) 20.8(trace) 33.5( 1 7 . 2 ) 46.2(36.0) 42.9( 19.3) 46,l(lG 5) 30.4( 11.5)
1.6 0 0
1.6
1.7 8.7 1.4 3.7 4.0 2.1 1.9
26.3(5.4) 51.0(20 2 ) 15.7(5.9) 20.3(4.2) 20.2(14 8) 21.8(2.6) 37,5(6,2) 2 2 , 8 (5 . 3 ) 27.0( trace) 43.3( 15.8) 60.3(43.7) 53.7(25.5) 59.5(25.5) 37.7(14.8)
conditions employed for forming acetates and propionates. However, if a mixture of hexamethyldisilazane, triniethylchlorosilane, and morphine was allowed to remain in the flash heater section of the column while the gas flow mas shut off and 5 minutes later the gas flow was quickly raised to 185 ml. per minute, about 50y0 conversion of the morphine was observed. T h a t this was a true peak-shift rather than the result of a decomposition product of morphine arising from sustained exposure to the high flash heater temperature was shon-n by the failure of morphine itself to decompose when i t was allowed to remain in the flash
(2) Langer, S. H., Pantages, P , S u t u r e 191, 141 (1961). (3) Lloyd, H. 1,Fales, H 31 Highet, P. F , S’andenHeuvel, W.J. A , Wildman, IT. C., J . d m Chem. SOC.8 2 , 3791
heater for the same period of time. This derivative had a retention time longer than that of morphine. It is highly probable that other reagents ndl be found which may be employed to form derivatives directly on the column.
~
iiq(in) \ - - - - I
(4) Luukkainen, T., VandenHeuvel, W. J. A., Haahti, E. 0. h.,Horning, E. C., Biochim. et Biophys. Acta 5 2 , 599 (1961). ( 5 j Sweeley, C. C., Chang, T. L., ANAL. CHEM.33, 1860 (1961). (6) VandenHeuvel, W. J. A., Sjovall, J., Horning. E. C., Biochim. et B i o.~ h-w . Acta 48, 556 (1961). (7) Wotiz, H. H., Martin, H. F., J . Biol. Chem. 236, 1312 (1961).
ACKNOWLEDGMENT
The authors gratefully acknowledge the technical assistance of Richard Tu’. Wolfe. LITERATURE CITED
(1) Horning, E. C., Moscatelli, E. A., Sweeley, C. C., Chem. & Ind. (London) 1959,
751.
RECEIVED for review January 22, 1962. Accepted April 3, 1962.
Silica Gel Structure and the Chromatographic Process Surface Energy a n d Activation Procedures PETER D. KLEIN Division of Biological and Medical Research, Argonne National laboratory, Argonne, 111.
b The effect of deactivation and activation treatments on chromatographic performance has been studied in three silica gels of known structure. These gels are representative of three different body structures ranging from high surface area, small pore diameter to low surface area and large pore diameter. Chromatographic performance was evaluated by the ratio of elution volume to bandwidth of the adsorbate sample, cholesterol acetate; for each gel it was degraded b y activation procedures to an extent which depended on the gel structure. Performance values and water contents of the treated adsorbents illustrate the contributions of body structure to the chromatographic process. The basis of this contribution has been discussed in terms of the surface chemistry and local geometry of the silica surface.
T
HE previous paper in this series (8) introduced the topic of the structure of the adsorbent and its influence on the adsorption and differential migration of sterol acetates. Khile thc data presented there and in the current \Tork have a special pertinence to the isolation and separation of these particular compounds, the findings also appear to have application t o chromatographic separations in general since they focus attention on the role of the adsorbent structure. The present paper deals with the surface energy of the adsorbent and the
roles of activation and deactivation in the achievement of optimum column performance. Although this subject has been treated b y a number of investigators, the structures of the silica gels used in their studies were either not reported (3, 11, 12) or were exclusively of the high surface area, small pore type ( 2 ) . This study deals with three silica gels of different known structures in columns in which cholesterol acetate was the migrating compound. The effectiveness of the gel treatment was measured by the ratio of the retention volume to the bandwidth of the adsorbate. This criterion n-as selected because of its direct applicability t o the prediction of separation efficiencies. THEORETICAL
I t is appropriate t o examine the basis of the surface energy and its distribution in a given silica gel. The surface of such a gel does not possess a uniform energy; a n adsorbate molecule will be held more tightly in one region of the surface than in another. This heterogeneity results from two features of the surface; the types of chemical groups on the surface and the local geometry of neighboring particles comprising the gel. Among the chemical groups present it is possible t o distinguish three, and perhaps four types. The first is the silanol
I
(-Si-OH)
I
group which is predominant
in the usual commercial gel. During the gelation process and subsequent drying of the gel, a certain proportion of these groups will undergo dehydraI I tion to form siloxane (-Si-0-Si-) I
I
linkages between neighboring silicon atoms. Other silanol groups persist in a third form, as the hydrated silanol (-Si-OH 1
,
OH), H even after the drying
I process. It is also possible that some of the siloxane linkages may still hold a water molecule. This possibility is extremely unlikely if the ambient relative humidity is less than 50%. Rehydration of siloxane linkages to silanol groups does not occur at an appreciable rate in atmospheres below saturation with water vapor. Each of the three groups knonn to be present on the surface of the gel possesses a different average energy; for example, Brunauer, Kantro. and Weise ( I ) reported that the energy of the siloxane surface in amorphous silica is 259 ?=I 3 ergs per sq. em. (as determined by the heats of solution in nitric-hydrofluoric acid miutures) whereas the silanol surface was calculated to have an energy of 129 + 8 ergs per sq. em. These t n o values are so widely separated that they ostensibly support the belief that physically adsorbed water can be completely removed from the surface before chemisorbed water is rcbmoved VOL. 34, NO. 7, JUNE 1962
733