Gas Chromatographic Chara,cteristics in the Separation of Protium and Deuterium Forms of Trimethylsilyl Sugars N. C. Saha’ and Charles C. Sweeley’ Department of Biochemistry and Nutrition, Graduate School of Public Health, University of Pittsburgh, Pittsburgh, Pa. 15213 Separation of the protium and deuterium forms of trimethylsilyl derivatives of carbohydrates by gasliquid chromatography has been investigated in the context of present theories of peak separation and of band broadening factors. Disengagement of peak centers has been related to differences in vapor pressure and differential solubility in the liquid phase for isotopic-substituted and parent species. The gas chromatographic method for determining gaseous interdiffusion coefficients has been extended for use with mixed solutes in a dilute solution. Changes in anomeric configuration have a more striking effect on gaseous diffusion coefficient than the effect caused by isotopic substitution in sugars. About 80% of the theoretical plate height is contributed by the gas phase. Isotopic substitution i s also reflected by some differences in the solute-solvent interaction forces, especially the binary liquid diffusion coefficients.
IT IS WELL DOCUMENTED that isotope fractionation can occur in a wide range of processes used for the separation or purification of compounds. In gas chromatography, for example, there are many instances of partial or complete separation of simple molecular mixtures, consisting of multiple isotopic species, into monoisotopic fractions; among the types of compounds which have been studied are oxygen isotopes ( I , 2), hydrogen isotopes ( 3 , 4 ) ,deuterio methanes and ethanes (3, tritiated butanes (6) and deuterio cyclohexanes, perdeuterobenzene and deuterio toluenes (7). Partial separations have also been achieved with isotopic mixtures of substances of much higher molecular weights, such as steroids and carbohydrates (7). It is apparent from these studies that isotope fractionation occurring during chromatography is not a phenomenon restricted to any particular type of isotopic atom, nor to compounds of very low molecular weight. In some cases mass differences are not wholly responsible per se for the observed separations, and additional secondary factors resulting from isotopic substitution must be involved. At the present time it is difficult to rationalize the collective observations on isotope fractionation with any single mechanism. In a recent review on the chromatographic fractionation of isotopes, Klein said “it will be the systematic correlation with known parameters that will eventually unravel the 1 Present address, Fertilizer Corporation of India: Sindri, Bihar, India. 2 Present address, Department of Biochemistry, Michigan State University, East Lansing, Mich. 48823. Inquiries should be sent to CCS at this address.
-
(1) W. Bocola, F. Bruner, and G. P. Cartoni, Nature, 209, 200 (1966). (2) F. Bruner, G. P. Cartoni, and A. Liberti, ANAL.CHEM., 38,298 (1966). (3) P. L. Gant and K. Yang, Science, 129,1548(1959). (4) W. J. Haubach and D. White, J . Chirn. Pliys., 60, 97 (1963). (5) W. A. Van Hook and J. T. Phillips, J . Phys. Chern., 70, 1515 (1966). 16) W. E. Falconer and R. J. Cvetanovic, ANAL.CHEM.,36, 1135 (1964). (7) R. Bentley, N. C. Saha, and C. C. Sweeley, ANAL.CHEM., 37, 1118 (1965). I
,
1628
ANALYTICAL CHEMISTRY
mechanisms of isotope effects exerted in these systems” (8). Toward this end, we have undertaken the present investigation on the measurement of various factors responsible for peak separation and peak broadening in gas chromatography of protium and deuterium forms of carbohydrates. EXPERIMENTAL
An F & M Model 810 gas chromatograph equipped with a flame ionization detector was used in this study. The preparation of columns and poly-0-trimethylsilyl (TMS) derivatives of the carbohydrates was described in a previous paper (7). The TMS derivatives of a and /3 anomers of a protium or deuterium sugar were dissolved in n-hexane and aliquots containing 0.5 to 1.0 pg were injected into the chromatograph with a 1-p1 Hamilton syringe (2.5-inch needle) for determinations of diffusion data. For calculating retention and partition data, injections of from 1 to 2 pg of derivatives were made. The temperatures of the injection port, column, and detector were 250, 175, and 200 “C, respectively. The zone velocity was calculated from the air retention time by the method of Peterson and Hirsch (9). A mixture of methyl esters of capric, lauric, and myristic acids in n-hexane was injected and the air retention time ( t a ) was calculated from Equation 1, ta
=
fRt ’ fRi fR3
+
fR1
- tRz2 - 2tRz
(1)
wherein f&, tRz, and tRl are the retention times of the methyl esters of myrjstate, laurate, and caprate, respectively. This method was checked against and found to agree well with the direct method of Hilmi (IO), which consists of saturating the carrier gas with a volatile compound such as butane or pentane and determining the time to negative deflection after injection of air. Standard Free Energy Change. The increment in standard free energy difference in helium for protium and deuterium TMS sugars, A(AGO), was determined with 20-feet X l/8inch 0.d. packed copper columns of 5 . 3 z SE-30 and 5 . 0 z Carbowax 20M, using Equation 2. In this equation R is the gas constant, T is absolute temperature, and the separation is the ratio of factor a H , D A(AGo) = -RT In (2) the corrected retention times (measured from the air retention time) of protium and deuterium forms of a TMS sugar. Specific Retention Volume. The specific retention volume V , of a solute was determined from Equation 3, which takes into account the corrections due to dead volume and column pressure drop. In Equation 3, f 2 = 3 ( p 2 - 1)/2(p3 - 1);
(3) p is the ratio of inlet to outlet pressure, F, is the corrected
carrier gas flow rate measured at the end of the column with a soap bubble flow meter, tR is the retention time of solute, w z is the weight of stationary liquid phase in the column, and T is the absolute column temperature. (8) P. D. Klein, “Advances in Chromatography,” J. C. Giddings, R. A. Keller, Eds., Marcel Dekker, New York, 1966, pp 3-65. (9) M. L. Peterson and J. Hirsch, J . LipidRes., 1,132 (1959). (10) A. K. Hilmi,J. Chromafog.,17,407 (1965).
0.056
-
0.052
0.054
0.048
0.050
0.044 0.042
0.040 0.038
0.036 0.024
' 1 0.020
00024
8
I
0.016
n
0.01 2
0.008
1
0.004k
-1 700
400
1000
vo q
1300
I
700
I
I
I
1000 1300 1600 vo
'
D,,
Figure 2. Graphic presentation for determination of
CFI
p-Heptadeuteromannose TMS
o
Figure 1. Graphic presentation for determination of
I
400
ci
0-HeptadeuteromannoseTMS
Partition Coefficient. The partition coefficient K was calculated from Equation 4, in which
columns and packed column alone. A high chart speed of 6 or 12 inches per minute was used to minimize the error in calculating band width. Retention times and band widths were determined as averages from measurements on 3 to 5 peaks. The interdiffusion coefficient D o at the column pressure was then calculated from Equation 6 ( I ] ) , in which co is the linear carrier gas velocity measured from the flow rate at the column outlet, and r is the
(4) pFI is the density of the stationary liquid phase (0.851 g/cc for SE-30) at the column temperature T OK. Binary Gas Diffusion Coefficient. T o standardize the gas chromatographic method, Do values of benzene and perdeuterobenzene were determined with a short copper tube (2 feet X '/8 inch packed with 2.5% SE-30 on Gas Chrom S, 100- to 120-mesh) connected in series with a long, empty copper diffusion tube (150 feet X '/8 inch). Values of Do for the deuterium and protium forms of TMS sugars were determined with a short packed column (1.5 feet X 3/1f, inch containing 2.5% SE-30 on Gas Chrom S, 100- to 120-mesh) connected to a copper diffusion tube (100 feet X 1/4 inch). The diffusion tubes were washed thoroughly with benzene, chloroform, and finally dry n-hexane, after which they were dried in a current of nitrogen for about 2 hours. The plate height H in the actual diffusion tube was calculated from separate band widths (distances on the base line between tangents drawn at the inflexion points) in the short packed columns and the combined columns, using Equation 5. In this expresssion L is the length
H
--L
wc2- wp2 - tp)z
16 ( t c
(5)
of the diffusion column, W is the band width, t is the retention time of a solute, and the subscripts c and p stand for combined
radius of the diffusion tube as determined from the weight of water filling an accurately measured length of the tube. Values for D, were corrected to atmospheric pressure by Formula 7, where p o is the Po Dgo = 760
(7)
column outlet pressure in mm Hg. Liquid Mass Transfer Term. The method for experimental determination of the liquid phase mass transfer term, CI,was similar to that developed by Saha and Giddings (12). No corrections were made for instrumental dead volume and time constants of detector and recorder in the present investigation, however, since these factors were judged to be insignificant. The TMS derivatives of the protium and deuterium forms of a and p mannose were selected for determination of C F I ,using a copper tube (10 feet X 'is inch) packed with 2.5 % SE-30 on Gas Chrom S, 100- to 120mesh. Inlet pressures used were 15 to 50 psi (gauge) fox nitrogen and 25 to 80 psi (gauge) for helium. Recorder (11) J. C . Giddings and S. L. Seager, J. Clzem. Phys., 33, 1579 (1960). (12) N. C. Saha and J. C . Giddings, ANAL.CHEM., 37,822 (1965). VOL 40, NO. 11, SEPTEMBER 1968
1629
0.056
0.058
0.052
0.054
k
k
A
0.050
0.048
0.046
0.044
0.042
0.040 01038
0.036
0.020
s
P
N
i -
rc
\
0
a"
0.0 I 6
0.020 0.0 I 6
0.01 2
0,008
0.012
Y-
0.004
0.008
700
400
0.004
1000 1300 1600 vo I D g o
I
I
1
I
400
700
1000
1300
Figure 4. Graphic presentation for determination of cl &Mannose TMS
vo /Dgo Figure 3. Graphic presentation for determination of a-Mannose TMS
CI
chart speeds were selected so as to provide band widths at half peak heights of at least 2 cm. The plate height, H , was calculated from average measurements of 3 to 5 peaks, and c1 was then determined from Equation 8, in which the pressure correction termfi = 9(p4 - 1 X p 2 - l)/8(p3,- l ) z , and uo and Do, are velocity and gas diffusion coefficient at the column outlet pressure.
A(H/h)and A(f2Doo/f)are the increments in the corresponding values (measured from Figures 1 to 4)of helium over nitrogen. Values of CZ,averaged over the whole range of u,/Do,,for the mannoses were calculated from Figures 1, 2, 3, and 4 and Equation 8. Precautions were taken to minimize or eliminate errors in the measurements of gas chromatographic data. The extent of overlap in average deviations was generally sufficient with individual sets of data for hydrogen and deuterium forms, however, to indicate nonsignificance of the results. On the other hand, the results for all the pairs of labeled and unlabeled anomers of glucose and mannose showed a very high degree of consistency. It is significant, for example, that the average value of H was always greater for the deuterium form of these sugar derivatives.
RESULTS AND DISCUSSION Systematic study of the factors responsible for isotope fractionation during chromatographic processes has received very little attention. Klein made a careful examination of many 1630
ANALYTICAL CHEMISTRY
examples of this phenomenon and concluded that no single mechanism would be invoked to account for all of the systems (8). However, quantitative information about the relative chromatographic behavior of a parent and isotope-labeled species of molecule is available in very few cases. In the case of partition chromatography of TMS sugars on a gas chromatographic column, we considered that some information about the separation of protium and deuterium forms might be obtained by studies of the equilibrium process of disengagement of peak centers, and the nonequilibrium processes of band broadening. Biegeleisen (13) and Van Hook and Phillips ( 5 ) have reported on theoretical investigations on the origin of vapor pressure and activity coefficient and their relation to peak separation in isotopic and parent species. We are not aware of any work on the various factors responsible for band broadening in these mixtures. Peak Separation Parameters. Disengagement of two peak centers in gas-liquid chromatography is due to vapor pressure differences and/or differential solubilities in the stationary liquid phase. Mathematically, the corrected retention volume, VB,can be related to vapor pressure p o and activity coefficient y as shown in Equation 9, where H and D subI n -V =R IDn - + l nY- X VR,H
YD
PI€'
(9)
PDO
scripts are for protium and deuterium sugars, respectively (13). That the fractionation of TMS derivatives of protium and deuterium sugars is not due entirely to differences in vapor pressure is clear, since the differences in standard free energy changes for the two forms [A(ACO)shown in Table I] are not identical in two liquid phases of different polarity, SE-30 and (13) J. Biegeleisen, J. Chem. Phys., 34, 1485 (1961).
Table I. Comparison of Standard Free Energy Difference of Protium and Deuterium Carbohydrates 5.3 SE-30 5 Carbowax-20M Sugar aH,D A(AGo) cal/mole aH,D A(AGO) cal/molea-Mannose TMS 1.014 f 0.003 -12.3 f 2.4 1.020 It 0.004 -17.5 4 2.7 - 8.4 & 2.1 1.010 f 0.003 @-MannoseTMS 1.005 i: 0.003 - 4.3 It 1 . 9 -32.1 dz 2 . 5 a-Glucose 1.004 f 0.003 - 3 . 5 It 2 . 0 1.037 f 0.004 @-GlucoseTMS 1.016 f 0,002 -13.9 f 1 . 4 1.039 f 0.004 -34.4 z!z 4.0
Carbowax 20M. The more polar Carbowax 20M gives values of A(AGo) which are much larger, implying that in polar liquids the polar regions of TMS sugars play an important role in peak separation. In Table I1 are shown the partition coefficients and specific retention volumes for parent and isotopic-labeled forms of glucose and mannose with SE-30 as liquid phase. The contribution to partition coefficient by the solid support is not likely to be of any significance because the support was carefully deactivated by acid-washing and silanizing. This factor is further minimized by use of a very nonpolar phase, SE-30, The observed difference in the partition coefficients, AK, of the p-anomeric forms of glucose is considerably greater than those calculated for the other anomeric pairs shown in Table 11. This agrees with and partially explains the known relative ease of separation of the protium and deuterium forms of @-glucoseas compared with the other pairs. It is evident from the results in Table II that differences in vapor pressure and in solubility in the liquid phase both contribute to the peak separation of protium and deuterium forms of TMS sugars. As an example of isotope effect on liquid phase selectivity Van Hook and Phillips ( 5 ) calculated that the ratios of activity coefficients of benzene and perdeuterobenzene vary from 1.02 in squalane to 0.99 in silicone oil. These authors claim that differences in vapor pressure and activity coefficient are associated with the covalent bonds involving isotopic atoms (C-D bonds in the present study) according to Equation 10, wherein A is the first-order quantum correction factor, B is the contribution from zero point energy, and T is the column temperature.
Band Broadening Factors. Band width also plays a role in determining separation efficiency, particularly in difficult separations such as the protium-deuterium pairs of TMS sugars. Any improvement in the sharpness of the bands of two adjacent peaks without significantly altering their At value will give an approximately proportional increase in separation efficiency. It is obviously essential to optimize all column parameters and operating conditions in order to obtain bands as narrow as the system will allow. Even under these circumstances, in very long, narrow-bore packed columns, it was observed earlier (7) that the band widths of TMS P-glucose and TMS P-glucose-d7 were considerably different. The deuterium form had the lower retention time but the greater band width. One of the main objectives of this work was to define the relative contributions of the band spreading factors, especially those originating from solutesolvent interactions, and t o determine the changes in these factors brought about by isotopic substitution in the molecules. Interdiffusion coefficients in the gas phase, D,,and the stationary liquid phase, D2,are the only two basic properties which enter into the plate height equation and which
Table 11. Comparison of Specific Retention Volume and Partition Coefficients of Protium and Deuterium Carbohydrates Sugar v, cc/g K AK a-D,-mannose TMS 461.6 =!z 2.8 644.6 f 3.9 8.8 a-H7-mannoseTMS 467.9 f 2.7 653.4 f 3.8 (3-D7-mannoseTMS 698.0 f 3 . 4 974.8 f 4.7 4.8 979.5 f 4.9 P-H,-mannoseTMS 701.4 f 3.5 a-D,-glucoseTMS 642.3 i 3.5 897.0 f 4.9 3.8 a-H,-glucose TMS 645.1 f 3 . 5 900.9 f 4.9 1330.0 It 5.9 21.4 P-D;-glucoseTMS 952.4 f 4.2 @-H,-glucoseTMS 967.7 z!= 4.2 1351.4 f 5.9
therefore may be associated with an isotope effect in band broadening processes. Binary Gas Diffusion Coefficient. Giddings and coworkers ( I l , 14) have shown that a gas chromatographic method for the determination of gaseous diffusion coefficients is superior to conventional methods with respect to simplicity, accuracy, and speed. One disadvantage of their method, however, is that the solute must be injected neat and in highly pure form into the diffusion tube. The trimethylsilyl derivatives of sugars are seldom obtained in absolute anomeric purity; furthermore, these derivatives are viscous liquids which cannot conveniently be injected into a diffusion tube without the aid of a solvent as diluent. To avoid these difficulties, we have modified the gas chromatographic method so that dilute solutions of impure solutes can be studied. By coupling the long empty diffusion tube with a short packed column which separates solvent and impurities from solute, D o values can be obtained with relative ease. Simultaneous measurements can also be made on more than one component in a single gas chromatographic analysis. The use of a short packed column between injection port and diffusion tube has additional advantages. It provides a more steady flow profile as evidenced by somewhat better than 0.5% reproducibility in retention times, thus eliminating the need for manostats on the carrier gas line, and it eliminates errors due to injector geometry, instrumental dead volume, and time constants of the electrometer and recorder. Pressure drops in the diffusion tubes were less than 2 cm of mercury. The method was standardized with benzene alone and in a mixture with methyl laurate. It was convenient to use a narrow bore diffusion tube and positive root in Equation 6 for the large D,values of isotopic benzenes and to use relatively large bore diffusion tubes and negative root in Equation 6 for the small D,values of trimethylsilyl derivatives of protium and deuterium forms of sugars, Values of H in the diffusion tube alone and of D,,for carbohydrates and benzene isotopes, using helium and nitrogen (14) J. C. Giddings and S. L. Seager, Znd. Eng. Chem. Fundamentals, l, 277 (1962). VOL. 40, NO. 1 1 , SEPTEMBER 1968
1631
Table 111. Comparison of Binary Gas Diffusion Coefficients of Isotopic TMS Carbohydrates in Helium and Nitrogen at 175 "C In helium6 In nitrogenb Compound H , cm D,,, cm2/sec H , cm D,,, cmz/sec a-Di-rnannoseTMS 0,40751 0,0041(6) 0.0480f 0.0005 (6) 0.958f 0.010(9) 0.0192f 0.0002(9) a-H,-mannose TMS 0.4131f 0.0087(6) 0.0473 f 0.0011(6) 0.946f 0.014(6) 0.0194f 0.0003(6) p-D,-mannose TMS 0.5183 f 0.0137(9) 0.0373f 0.0010(9) 1,191f 0.021 (6) 0.0154f 0.0003(6) p-Hj-mannoseTMS 0.5724 f 0.0117(9) 0.0337f 0.0007(9) 1.199 k 0.008(3) 0.0153& 0.0001 (3) a-Di-glucoseTMS 0.4183 & 0.0195(6) 0.0468=k O.OO23(6) 1.177f 0.011(8) 0.0156& 0.0002(8) a-Hj-glucoseTMS 0.4365i 0.0140(6) 0.0447f 0.0015(6) 1.190f 0.021(6) 0.0154i 0.0003(6) P-D,-glucose TMS 0.6252i 0.0644(4) 0.0311 f 0.0033 (6) 1.332i 0.036(6) 0.0138f O.OOO4(6) p-Hi-glucose TMS 0.6681 f 0.0089(6) 0.0288f 0.0004(6) 1.562f 0.023 (6) 0.0117i 0.0002(6) Per deuter obenzenea 0.2171 0.706 ... ... Benzene" 0,2022 0.659 0,0599 0.177 a Determined from 2-foot X lis-inch 0.d. packed and 150-foot X l/s-inch 0.d. empty diffusion tube. b Number of individual determinations in parentheses.
-
~~
Table IV. Comparison of Liquid Phase Mass Transfer and Interdiffusion Coefficients of Protium and Deuterium Mannose Ratio of D Iof protium Sugar ct x 104 A(l - R) to deuterium carbohydrates a-Hj-mannoseTMS 7.61.0.4 0.0461 i 0.0009 a-D,-mannose TMS 8.4f 0.4 0.0470f O.OOO9 1.084f 0.054 p-H,-mannose TMS 9.1 f 0 . 5 0.0312-I: 0.0005 p-D7-mannoseTMS. 10.2f 0 . 5 0.0323 f 0.0007 1.083 f 0.054
as carrier gases, are given in Table 111. A marked difference in H was observed with anomeric and isotopic pairs in the sugars. Differences in diffusion coefficients were much smaller for protium and deuterium pairs of sugars than for anomeric pairs of the same protium or deuterium sugar. It can be shown from the theories of binary gas diffusion coefficients that the contribution of mass difference to separation of isotopic molecules is insignificant compared with the contribution by differences in D, [for a review, see Fuller and Giddings ( I S ) ] . The ratio ( 1 / M A 1/MB)1'2in helium for mannose or glucose is 1.00004; that is, the protium form will have 1.00004 times greater D,,value than that of the deutero form (this is opposite from experimental results) had there been no change in the denominator of Equation 11. In all of the cases studied, the D,values found for deuterium forms was greater than those for the corresponding protium forms (see Table 111). The fact that the increment in D, for substitution with deuterium varied with the structure of the sugar suggests that D , depends somewhat on molecular configuration in these trimethylsilyl ethers of sugars. It follows from present theories of binary gas diffusion coefficients that the collision diameter (separation of molecular centers upon collision between unlike molecules), determined by intermolecular forces of unlike molecules, is different for anomeric and isotopic pairs. Recently, Fuller, Schettler, and Giddings (16) proposed a modified theory for gas diffusion coefficients based on contributions from atomic and other structural features. The gaseous diffusion coefficient of A in B(DAB)in Equation 10 is related to temperature (T OK), pressure (p atm.), the molecular
+
weights of A and B ( M Aand MB),the atomic and special structural volume increments of solute A ( YA),and the atomic vol(15) E. N. Fuller and J. C. Giddings, J. Gas Chromatogr., 3, 222 (1965). (16) E.N.Fuller, P. D. Schettler, and J. C. Giddings, Znd. Eng. Chem., 58, 18 (1966).
1632
ANALYTICAL CHEMISTRY
ume increment of the carrier gas B ( VB). The precise nature of the term VA is not completely clear. For example, it is difficult to decide what special structural feature should be included in Va to explain known differences in diffusion coefficients of simple isomeric compounds such as hydrocarbons. Although use of this theory cannot yet be made for a relation of structural features in complex compounds such as trimethylsilyl sugars to gaseous diffusion coefficients, some broad generalizations can perhaps be made. Perdeuterobenzene has a diffusion coefficient of 0.709 at 448 OK in helium. Using values for atomic and structural volume increments calculated from C = 16.5 and aromatic ring = -20.2, as given by Fuller and Giddings (1-9, it is found that the volume increment for deuterium is negative (-0.9). In the TMS sugars, even this value for a deuterium atom cannot account entirely for the observed increment in the D, value of the isotopic species compared with that of the parent carbohydrate molecule. Perhaps isotopic substitution also changes the contribution to the V A term by the aromatic ring in perdeuterobenzene and the cyclic pyranose ring in the sugars. Liquid Phase Diffusion Coefficient. Absolute values of the binary liquid diffusion coefficient, D l , were not determined from the experimental c l values of packed columns, because the effective liquid film thickness term could not be calculated reliably from pore size distribution data of the supports (12). Instead, ratios of D L for protium and deuterium pairs were calculated from the relation below, in which R is the ratio of the retention times of air and solute (2). The results are given in Table IV. Values of D Ifor the protium forms of
TMS sugars appeared to be slightly greater than those of the deuterium forms. That is, the deuterium-substituted compounds diffused more slowly in the liquid films of the stationary phase. Contribution to Plate Height by Gas and Liquid Phase Terms. Plate height, corrected for pressure, can be expressed the !, total contribution by the gas phase, and HI, in terms of %
the total contribution by the liquid phase. In Table V are shown the relative contributions of the gas phase and liquid phase to the observed plate height at an arbitrary carrier gas velocity of 6.68 cmjsec. Effects attributed to the gas phase
H
=
Bg+ HL
or or
(1 3)
contribute about 80% of total plate height with the protium forms and about 857% with deuterium forms of sugars. Column efficiency is dependent primarily on effects in the gas phase, therefore, and improved column performance in the fractionation of isotopic-substituted compounds of the types studied here will be obtained by careful control of the band broadening factors in the gas phase. Giddings (17) has made a thorough theoretical study of the nature of the gas phase term, H,, and Saha and Giddings (18) have shown experimentally how various column parameters and support characteristics affect the Ho term. These authors show that major contributions to ,U,are related to the physical nature of the support, especially pore size distribution, pore geometry, and particle size distribution, and by the uniformity of packing in the column. A further reduction in mesh size range and, perhaps, a decrease in the column diameter would effect ~~~
Table V. Relative Contributions to Plate Height by Gas Phase and Liquid Phase V = 6.68 cmisec, helium carrier
Sugar a-D7-mannoseTMS a-H7-mannoseTMS p-D,-mannose TMS P-H7-mannoseTMS
Zcm 0.0408 0.0391 0.0416 0.0397
-
-
~1,cm 0.0056 0.0051 0.0068 0.0061
~,,cm 0.0329 0.0340 0.0326 0.0336
HJH 0.81
0.87 0.78 0.85
greater isotope fractionation efficiency by reducing band broadening in the gas phase. In our previous studies on separation of protium and deuterium forms of the TMS sugars on 50-foot packed columns of small diameter (9, we found that band width in the deuterium form was greater than that in the protium form by 11%. The reverse was found in this study, with short packed columns; that is, band widths were somewhat greater in the protium forms. The two forms were injected together on the long columns, and singly on the short columns, but the reversal of band width differences cannot be attributed to this factor. It is very difficult to account for this phenomenon at the present time because it might be related to almost any of the terms in the plate height equation.
~
(17) J. C. Giddings, ANAL.CHEM., 34,1186(1962). 37,830 (1965). (18) N. C. Saha and J. C. Giddings, ANAL.CHEM.,
RECEIVED for review May 2, 1966. Resubmitted May 29, 1968. Accepted June 10,1968.
Determination of Normal Paraffins in Petroleum Heavy Distillates by Urea Adduction and Gas Chromatography J. R. Marquart, G . B. Dellow, and E. R. Freitas Shell Development Co., Erneryuille, Gal$ A simple and efficient method has been developed for the determination of normal paraffins in heavy gas oils. The method consists of a urea adduction step for isolating the normal paraffins, which in turn, are analyzed with good resolution by gas-liquid chromatography. The final temperature of adduction determines the lower carbon number limit of applicability of the method, C1, at 25 OC and Clzat 0 O C . The method of analysis has been shown to be in excellent agreement with the composition of known calibration blends. DIRECT ANALYSIS of normal paraffins in petroleum distillates in the heavy gas oil range by gas-liquid chromatography (GLC) is generally unsatisfactory because of resolution problems encountered in such complex mixtures of hydrocarbon types. A number of methods using SA-Type Molecular Sieves or urea adduction to simplify the chromatogram have been proposed. Each of these, however, has some shortcoming, either in complexity of operation or inability to handle oils containing high molecular weight paraffins. The following are typical illustrations: 1. Subtractive GLC methods (1, 2 ) involve duplicate (1) N. Brenner and V. J. Coates, Nufure, 181, 1401 (1958). (2) B. T. Whittam, ibid.,182, 391 (1958).
chromatograms of the oil with and without a precolumn of 5A Type Molecular Sieves. Comparison of the chromatog r a m reveals the contribution of n-paraffins to the overall spectrum. The method is fairly satisfactory for oils in the kerosine range. For higher molecular weight oils, however, temperature programmed GLC is necessary and duplication of chromatograms is generally poor. 2. Subtraction and re-elution GLC methods (3, 4 ) involve chromatography of an oil while removing the n-paraffins on a molecular sieve precolumn and subsequent elution of the sieve bed at elevated temperatures to obtain a chromatogram of the adsorbed n-paraffins. This method is also limited to the kerosine range, because higher molecular weight paraffins cannot be removed readily from the sieves by heating. 3. Molecular sieve recovery methods (5,6) involve removal of the n-paraffins on molecular sieves and subsequent recovery of the n-paraffins by destruction of the sieves. These methods have been used for heavy distillates, but are often laborious (3) F. T. Eggertsen and S. Groenning, ANAL. CHEM.,33, 1147 (1961). (4) G. C. Blytas and D. L. Peterson, ibid., 39, 1434 (1967). (5) J. V. Brunnock, ibid., 38, 1648 (1966). (6) J. V. Mortimer and L. A. Luke, Anal. Chim. Acta, 38, 119 (1967). VOL. 40, NO. 1 1 , SEPTEMBER 1968
e
1633