450
GENERAL-ANALYTICAL METHOD@
[51]
"02"" column containing only 3 and 4 carbon atoms, which implies that two of the three oxygens are located very close to each other in the molecule. The molecular ion (m/e 314, see Fig. 13) would normally be in the lower right-hand corner of the map, but is absent in Fig. 32 because the computer was asked to omit ions of low intensity due to space limitations. However, this situation is easily recognized-in the element map, since the ion of highest mass (m/e 296, M-H20) contains fewer oxygens than an ion of lower mass, m/e 229. In all columns, the carbonto-hydrogen ratio is high, indicating the presence of a saturated aliphatic chain. Similar considerations apply to the remainder of the spectrum, following the interpretation based on the low-resolution spectrum, discussed on p. 404. The handling of high-resolution data by computer techniques was developed primarily out of need, but in addition leads to further possibilities for the expeditious handling and treatment of the data involved. Commercial systems have recently become available for the complete data handling and computerization of high-resolution data, using intermediate magnetic tape storage or direct on-line small computers. Although .such sophisticated systems are very expensive and highly specialized, the impetus for their development has been the considerable potential value of these techniques in structural problems in organic and biochemistry. Acknowledgment The work from the author's laboratory described in this article was supported by the Robert A. Welch Foundation (Q-125) and the National Institutes of Health ( G M 13901). Figures containing mass spectra were plotted at the C o m m o n Research Computer Facility of the Texas Medical Center, supported by the United States Public Health Service through Grant F R 00254.
[51 ] P r e p a r a t i v e G a s C h r o m a t o g r a p h y
By R. S. HENLY and D. J. ROYER The high separating power of gas chromatography (GC) on an analytical scale has led to interest in its use as a purification or preparative tool. From the reported uses of this technique in the literature, it might be said that preparative GC is all things to all people. The term "preparative GC" has denoted the purification of milligram and submilligram quantities of materials on analytical size columns on the one extreme and gram quantities of materials and larger on columns 1-4
[51]
PREPARATIVE GAS CHROMATOGRAPHY
451
inches in diameter on the other extreme. The small-scale preparative GC is essentially analytical GC with the addition of a sample collection system at the column outlet. In this case, where only milligram quantities or less of product are desired, the preparative GC exhibits one of the attributes of analytical GC; i.e., very small quantities of materials can be conveniently handled. 1 However, the term "preparative GC" should, and usually does, denote the use of large-diameter columns to purify relatively large quantities of materials. In referring to the magnitudes of column diameters and material quantities, let it suffice to say that these are larger than those used in analytical GC. The term "scaled-up GC" is more descriptive of the process, but the term "preparative GC" is probably here to stay. The fact that the range of sample sizes (micrograms to milligrams) for analytical GC cannot be increased significantly on analytical size columns without an appreciable loss of resolution led to the use of larger diameter columns and vaporizers for preparative GC. Many examples of the study and use of preparative GC have been published, but most of the reported results deal with relatively volatile, stable materials, e.g., low molecular weight hydrocarbons and fluorocarbons. The proportion of published applications of preparative GC for purification of lipids has been small and much of the published work describes purification of small amounts of material in analytical size columns. A review of the literature on preparative GC for lipid application has been given by Henly. 2 Rose, Royer, and Henly 3-5 more recently presented the results of an extensive experimental study of design variables to optimize the preparative separation of the C8 to C2o saturated straigh~chain fatty acid methyl esters on 1- and 2-inch diameter columns. It is not the purpose of this chapter to present a detailed thesis on preparative GC, but to emphasize the difference between analytical and preparative GC. The factors which affect the separation process (sample vaporization, column design, and operating variables) are discussed. Detection and collection devices are not considered here. Comparison of Small- and Large-Diameter Columns Until 1964 it was generally accepted that increasing column diameter led to extreme loss in column efficiency and resolution. However, it has ' N. Pelick, R. S. Henly, R. F. Sweeny, and M. Miller, J. Am. Oil Chemists' 8oc. 40, 419 (1963). 2R. S. Henly, J. Am. Oil Chemists' Soc. 42, 673 (1965). SA. Rose, D. J. Royer, and R. S. Henly, Separation Sci. 2, 211 (1967). 4A. Rose, D. J. Royer, and R. S. Henly, Separation Sci. 2, 229 (1967). JA. Rose, D. J. Royer, and R. S. Henly, Separation Sci. 2, 257 (1967).
452
G E N E R A L ANALYTICAL METHODS
[51]
been shown 2,e, T that much of this loss in resolution is due to overloading the columns with large samples; i.e., the sample sizes were being scaledup to a much greater degree than the column size. Just as resolution is lost on an analytical column if the sample size is too large, resolution is lost on a preparative size column if the sample is too large, Henly 2 presented experimental chromatograms which show that a ¼ inch diameter column and a 5~ inch diameter column give similar results with a C~JC16 saturated methyl ester mixture when run under overloaded and nonoverloaded conditions. It is a natural technique to overload a preparative column as long as the desired separation can be obtained. However, the same technique can be done on a smaller scale with an analytical column. It is incorrect to compare the resolution or plate efficiency obtained on an overloaded large-diameter preparative column to that obtained on a nonoverloaded small-diameter analytical column. Two such columns should be compared with the same sample sizes per unit column cross-sectional area. Thus a 1/~l sample on a ~ inch column is equivalent to only a 16 ~l sample on a 1 inch column, a 64 #I sample on a 2 inch column, etc. Going in the reverse direction, a 1 ml sample on a 1 inch column i~ equivalent to a 62.5 ~l sample on a 1~ inch column. One generally does not use 60 ~l samples in analytical GC and expect good resolution. However, the tendency in preparative chromatography too often has been to inject the 1 ml sample into a 1 inch preparative column and then report that the preparative column is not nearly as good as an analytical column. Column Overloading The high resolution obtained in analytical GC is due to separation characteristics under conditions of high dilution of the sample. The vapor volume of the sample is small compared to the volume of the column, and liquid phase concentrations are small, approaching infinite dilution. The sample entering the column occupies a very small proportion of column volume, and in most cases solute partition isotherms are linear or at least approach linearity at the low concentrations. The assumption of linear partition isotherms is the basis of the analytical GC theory and much of the preparative GC theory2, 9 The two major factors in column overloading are excessive sample vapor volumes entering the column and excessive liquid phase solute ' R. S. Henly, A. Rose, and R. F. Sweeny, Anal. Chem. 36, 744 (1964). ' D. T. Sawyer, and H. Purnell, Anal. Chem. 36, 457 (1964). ' S. M. Gordon, G. J. Krige, and V. Pretormus, J. Gas Chromatog. 2, 241, 246, 285 (1964); ibid. 3, 87 (1965). ' S . M. Gordon and V. Pretorius, J. Gas Chromatog. 2, 196 (1964).
[51]
PREPARATIVE GAS CHROMATOGRAPHY
453
concentrations.2, e,~ Both these factors can cause solute peaks to broaden excessively. The effect of excessive sample vapor volumes has been treated mathematically for the assumption of linear partition isotherms.1°,11 Broad symmetrical peaks (but non-Gaussian in shape) are obtained if the sample inlet profile is rectangular. Broad trailing peaks are obtained if the sample inlet profile is trailing, However, this factor is important only at conditions resulting in small retention volumes or elution times. If the ratios of the sample vapor volume to solute retention volumes are small, the effect of sample volume on peak widths will be negligible. This phenomenon is explained simply and schematically by Henly. ~ Let it
8
8
Elution lime in minutes of|el injection of sample
Fro. 1. Peak behavior and widening characteristics with increasing sample size. 3 foot X 1 inch, i.d., 20% SE-30 column at 215 °. C~JCI, methyl ester injections. [A. Rose, D. J. Royer, and R. S. Henly, Separation 8ci. 2, 211 (1967).] ~J. J. van Deemter, F. J. Zuiderweg, and A. Klinkenberg, Ohem. Eng. 8ci. 5, 271 (1958). 11p. E. Porter, C. H. Deal, and F. H. Sta-oss, J. Am. Crhem. 3oc. '/8, 2999 (1958).
454
GENERAL
ANALYTICAL
[51]
METHODS
suffice here to say that some insight into this can be obtained, if the extreme detrimental case of sample still entering the column inlet as solute starts to emerge from the column outlet is considered. If retention volumes are increased to minimize the effect of largesample vapor volumes, high solute liquid phase concentrations are obtained with large samples. Nonlinear partition isotherms at high concentrations cause broad skewed peaks. 6 Also interaction among solutes can occur; i.e., the elution characteristics of any one solute is affected by the presence of the other solutes. 8 These interactions can be either detrimental or beneficial to separation. The nature and degree of peak distortion, broadening, and interaction at high concentrations depend upon the solute-stationary phase system. Figures 1 and 2 show these concentration effects experimentally for a binary C14/C18 methyl ester mixture on SE-30 silicone rubber and ethylene glycol succinate (EGS) columns. The
4000/~1 o
°
/ I
_.,J__:, .... ~...:.,,~ ¢
a 18
Elutlon
time
in
,
i 22
,
i 26
,
, 30
minutes ofte¢ injection of
i
, 34
,
, 38
,
, 42
,
, 46
sample
Fzo. 2. Peak behavior and widening characteristics with increasing sample size. 3 foot × 1 inch i.d., 17% EGS column at 143.°. C,/C~, methyl ester injections. [A. Rose, D. J. Royer, and R. S. Henly, Separation Sc/. 2, 211 (1967).] columns were 3 feet long X 1 inch i.d. Note that the peak distortion, broadening, and interaction are much more extreme with the EGS column than with the SE-30 column. The strong peak interaction on the EGS column is beneficial in that the C~4 displaces the C~e instead of overlapping it at large sample sizes. This displacement effect does not go on indefinitely. At the conditions used in Fig. 2, the peaks start to overlap with sample sizes greater than 4 ml. Another factor which enters into the operation of large-diameter columns with large sample sizes is transient temperature changes in the column due to passage of solute. *,1~,13 This is due to heats of absorption and desorption of the solute and the fact that heat transfer between the "J. Peters and C. B. Euston, Anal. Chem. 37, 657 (1965). "R. P. W. Scott, Anal. Chem. 35, 481 (1963).
IS1]
455
P R E P A R A T I V E GAS CHROMATOGRAPHY
column wall and the center of the packed bed is not instantaneous. An experimental example of this is shown in Fig. 3 for 0.1 ml of methyl myristate near the outlet of a 3 foot X 1 inch i.d. column. The magnitude of the temperature change in the bed center above and below the steady state temperature here was approximately 2 ° . As the solute enters the section and is absorbed, heat is evolved and is not dissipated fast enough to prevent a temperature rise. However, some of this heat is lost to the surroundings and as the solute evaporates it regains this lost heat by cooling the packing below the steady state temperature. After the solute has completely evaporated from the section, the packing regains its steady state temperature. The degree of this temperature change
~
":~ o.
ii o
/Total concentration , f profile (detector)
~,...~-Center "L Temperature ~ Wall J profile (thermistors)
,./i !.
,J
I
2
I
4
I
I
,o
I
,,
Elutmn time in minutes after injection of sample
Fza. 3. Temperature variations produced in the column bed by passage of a 0.1
ml methyl myristate solute band at the column exit. 3 foot X 1 inch i.d. 22% EGS column at 155°. [A. Rose, D. J. Royer, and R. S. Henly, ,Separation 8ci. 2, 229 (1967)]. varies from the bed center to the column wall, is greatest at the column inlet and decreases as solute passes down the column, and increases with sample size. 4 Although extensive studies have been made on these transient temperature changes' under various conditions which gave poor to optimum separations of saturated methyl esters on 3 foot X 1 inch i.d. columns, a it has not yet been possible to describe in detail the overall effect of the temperature changes on the elution and the separation process. It may well be that to a certain extent these transient temperature changes are beneficial.
456
GENERAL ANALYTICAL METHODS
[51]
Column Design There has been strong conviction in the field of preparative GO that the loss of resolution on large-diameter columns is due to radial solute band concentration gradients which in turn are caused by any one or all of: (1) radial carrie~ gas velocity gradient, (2) radial temperature gradients, and (3) radial nonuniform stationary phase gradients. Items 1 and 3 are thought to be caused by nonuniform packing distribution or density. Item 2 is thought to be due to the poor heat transfer characteristics of column packings and the increased length of path for heat transfer from the wall to the bed center in large diameter columns. These factors are reviewed briefly by Henly 2 and discussed in detail by Giddings. I~ Attempts to minimize these gradients in large-diameter columns have led to the use of disk and doughnut type baffles by Abcor, Inc., 15,1e sintered metal disks called flow homogenizers (developed by Continental Oil Corp.) by Hewlett-Packard,15 Annular tube hi-wall columns by Nester Faust, 15 long (up to 250 feet), small diameter (8/~ inch) columns by Varian Aerograph, 15 a number of parallel, small diameter columns by Beckman Instruments, 17 and internal fins parallel to the column axis and extending inward toward the center.18 Extensive work with straight 3 foot X 1 inch i.d. columns without baffles of any type indicates that radial solute concentration gradients are negligible if the column is packed well and temperature control is good.~ The column inlet consisted of a cone which expanded from the 3~ inch diameter vaporizer connector to the column diameter. Columns heated with uniformly wrapped nichrome resistance wire and insulation were compared with columns heated in a well-designed forced-air circulation oven? It was found that oven columns gave much better results, in terms of maximum separable sample size and production rate (maximum separable sample size divided by total elution time) for binary mixtures of saturated methyl esters, than did the nichrome-wrapped columns. Results were also compared with the inlet cone: (1) packed with column packing, (2) filled with glass wool, (3) maintained at the column temperature, and (4) maintained 60 ° above the column temperature. 8 With both methods of column heating it was found that best results were obtained with the inlet cone packed with column packing and maintained 14j. C. Giddings,Y. Gas Chromatog. 1(1), 12 (1963). WAnonymous, Ohem. Eng. Ne~ss 43(26), 46 (1965). mAnonymons, Ohem. Eng. News 44(21), 52 (1966). ITT. Johns, M. R. Burnell, and D. W. Carle, in "Gas Chromatography" (H. J. Noebels, R. F. Wall, and N. Brenner, eds.), p. 207. AcademicPress, New York, 1961, uj. L. Wright, $. Gas Chzomatog. 1(11), 10 (1963).
[51]
PREPARATIVE GAS CHROMATOGRAPHY TABLE
457
I
EFFECTS OF C O N E PACKING AND INLET G A S T E M P E R A T U R E ON SOLUTE SEPARATION a' b
Inlet cone (°C)
Outlet cone (°C)
Approx. maximum separable sample size (ml)
160, unpacked c 16~ packed~ 225, unpacked 225, packed 160, packed 225, packed 225, packed
160, unpacked 16~ unpacked 160, unpacked 160, unpacked 160, packed 160, packed 225, packed
0.30 0.30 0.50 1.00 0.60 1.00 1.I0
Evenly nichrome-wrapped column, 17% EGS, l-inch i.d. × 3-loot-long column, column temp. 160 °, injection 50/50 by weight C,/C~e, fatty acid methyl ester mixture. b A. Rose, D. J. Royer, and R. S. Henly, Separation ~ci. 2, 211 (1967). Packed, filledwith column packing; unpacked, filledwith glass wool and screen for support.
60 ° above the column temperature. These results are summarized in Tables I and II. The temperature distribution in the oven columns was much more uniform than in the niehrome-wrapped columns s as shown in Figs. 4 and 5. Operation of the inlet cone at 60 ° above the column tern182
l't4
Central bed ear wall
®
Inlet cone above column temp.
"~ ISE a
-
m
150
142
,34
"%'~'~ Near wall 1 ~Centrol bed .~ Inlet cone at column temp.
4
2'0
2'4
Awiol column position from inlet (in.)
Fro. 4. Column bed temperature variations at steady state in a 3 foot X 1 inch i.d. EGS column heated with uniformly wrapped nichrome wire. [A. Rose, D. J. Royer, and R. S. Henly, ,Separation Sd. 2, 229 (1967).]
458
GENERAL ANALYTICAL METHODS
[51]
TABLE II EFFECTS OF INLET CONE PACKINGAND TEMPERATURE F O R EGS AND SE30 OVEN-HEATEDCOLUMNS a' b 20% EGS columnc
20% SE30 columnJ
Maximum Maximum Maximum Maximum separable production separable production sample size rate sample size rate (ml) (ml/hr) (ml) (ml/hr)
Inlet cone Unpacked, at column temp. Unpacked, 60° above column temp. Packed, at column temp. Packed, 60° above column temp.
0.250 0.400 1.20 3.50
0.968 1.74 2.75 3.48
0.80 2.20 2.20 3.50
3.99 9.42 10.0 12.8
Oven-heated columns 1-inch i.d. X 3 feet; injection 50/50 by weight C~,/CIe fatty acid methyl ester mixture. b A. Rose, D. J. Royer, and R. S. Henlyl Separation Eci. 2, 211 (1967). =Column and outlet-cone temperature 160°. Column and outlet-cone temperature 216°.
=
perature gave improved results, even though most of the stationary phase was lost from the cone and the inlet section of the column as shown in Fig. 6. Studies made on methods of packing these columns showed t h a t simultaneous vibration of the vertical columns in both the vertical and horizontal planes while the packing was slowly poured in produced 182
[74
166
P
m~ 150
f ~ ^
lnlet cone above column temp. //--Inlet cone at column temp.
o Centrol bed t, Near-wall bed
142~" Axial column position from inlet (in)
Fro. 5. Column bed temperature variation at steady state in a 3 foot X 1 inch i.d. EGS column heated in a forced-air circulation oven. [A. Rose, D. J. Royer, and R. S. Henly, Separation Sd. 2, 229 (1967).]
[51]
459
PREPARATIV]~ GAS CHROMATOGRAPHY
uniform distribution of the particles. 19 Results of studies on radial solute concentration profiles in the 3 foot X 1 inch i.d. columns 4 showed that carrier gas rioT, as determined with injections of nonabsorbed gases, is 5-10% faster at the column wall than at the center of the bed. However, solute radial concentration distribution for the methyl esters did not appear to be affected by the carrier gas flow distribution. In nichrome wire-wrapped columns, with uneven temperature distribution, radial solute concentration gradients for small samples were negligible. For large overloading samples, radial solute concentration gradients were observed at the back of the peaks, with solutes eluting faster at the bed Inlet cone
24
~
20
6o
16
o o 5
12
g
s
o
4
u~ L~
0
bJ
A Wall bed position o Central bed inlet cone
O
,
,
,
,
4
8
12
16
•
20
,
,
24
28
.
32
,
36
Distance from column inlet (in,)
FIG. 6. Stationary phase concentration variation with distance into the column. Original column packing was 22 wt. % EGS. [A. Rose, D. J. Royer, and R. S. Henly, 8eparalion 8ci. 2, 229 (1967).] center than at the wall. Also peak resolution was better at the bed center than at the wall. In oven-heated columns, radial solute concentration gradients were negligible for both small nonoverloading and large overloading samples. These results all indicate that radial and axial temperature distribution in a column is an important factor in preparative GC. The improved results claimed by the use of various types of bafl]es in a column" may be their effect on temperature distribution rather than gas flow distribution. Vaporizer Design and Sample Injection The design of the vaporizer and the method of sample injection becomes very important when large overloading samples are injected into 19D. J. Royer, Ph.D. Thesis, The" Pennsylvania State University, 1966.
460
GENERAL ANALYTICAL METHODS
[51]
large-diameter columns. Extensive studies on these factors 3 compared a rapid injection of large samples into a 4 inch long X 1 ¼ inch i.d. cylindrical vaporizer cavity packed with ~ inch steel shot with a slow, rotary-sprayed injection on the wall of the cavity emptied of the steel shot. The results of these comparisons are shown in Table I I I . The latter design and technique gave a 20-25% improvement in results on a 1 inch i.d. column. However, the degree of improvement on a 2 inch i.d. column was much greater. This vaporizer design and injection method was required to obtain results on the 2 inch i.d. column which were equivalent TABLE III COMPARISON OF INJECTION AND VAPORIgATION TECHNIQUE FOR 1- AND 2-INCH I.D. E G S OVEN COLUMNs, b
Maximum separable sample size Column diameter Vaporizer (inches) cavity 1 1 1 2 2 2
Filled Empty Empty Filled Empty Empty
Maximum production rate
2-inch i.d. 2-inch i.d. Achieved scale-up ~ Achieved scale-upt Injection technique (ml) (ml) (ml/hr) (ml/hr) Normal Normal Turned-sprayed Normal Normal Turned-sprayed
3.00 2.60 4.00 3.20 4.50 13.0
-----16.0
4.44 4.52 5.00 11.3 8.20 19.5
---20.0
20-22 % EGS column at 160°; column length 3 feet; carrier gas flow rates: 1-inch i.d. column 1.54 liters/minute; 2-inch i.d. column 6.16 liters/minute. A. Rose, D. J. Royer, R. S. Henly, Separation ~ci. 2, 211 (1967). t Scale-up based on column crose-sectional area and data of the third entry. to predicted scale-up results from the 1-inch i.d. column. I t was found t h a t there is a range of optimum injection times for each sample size. Too rapid or too slow sample injection rates are detrimental. The optimum injection rate increases with decreasing sample sizes and with decreasing solute elution times. The slow injection technique apparently decreases the high liquid phase solute concentrations. Packing distribution and radial carrier gas velocity and solute concentration gradients were not investigated in a 2 inch i.d. column. However, the fact t h a t predicted scaled-up results were obtained on the 2 inch i.d. column indicates t h a t radial solute concentration distribution is similar in both 1 inch and 2 inch i.d. columns. Whether this holds true as column diameter is further increased still needs to be determined experimentally.
[51]
PBEPARATIVE G A S CHROMATOGRAPHY
461
Operating Conditions Extensive studies have been made on the effect of operating conditions on production rates (sample size divided by total elution time) for the separation of binary mixtures of adjacent even carbon saturated methyl esters2 It was found that the production rates increased with increasing carrier gas flow rate and went through a maximum with increasing column temperature. Figure 7 shows production rate as a function of 5i'
41 J~
3~ t9
3z v o~ o
3C
"o o
E .~_
22 18
o
14
6
Iuu
~':.'0
260
300
Column temperature ("C)
Fro. 7. Maximum production rate studies, C,JC,e injection, 3 foot × 1 inch i.d. SE-30 column. [A. Rose, D. J. Royer, and R. S. Henly, Separalion Sci. 2, 257 (1967) .] column temperature and carrier gas flow rate for the C1,/Cle binary mixture on a 3 foot X 1 inch i.d. SE-30 column. Similar results were obtained for all binary mixtures of adjacent even-carbon saturated methyl esters in the range of C8 to C~o. The optimum column temperature increased and the maximum production rate decreased with increasing carbon number. Similar results were obtained on both SE-30 and EGS columns. However, over all, production rates were lower on EGS columns due to more extreme peak broadening causing longer elution times. Figures 8 and 9 show production rate as a function of methyl ester carbon
462
GENERAL ANALYTICALMETHODS
[51]
number and gas flow rate on SE-30 and EGS, respectively. On the SE-30 column, the optimum temperatures for the C8/Clo and C18/C2o mixtures were 180 ° and 270 °, respectively. On the EGS column these temperatures were 150 ° and 210 ° . Still higher production rates were obtained at carrier gas flow rates of 4.0 liters/minute, but it was found that columns deteriorated rapidly at this flow rate at the higher range of column temperatures. This indicates that gas flow rates cannot be increased indefinitely for preparative GC columns. 140
120 ..c
\
~ I00 •
g
so
~
6o
E Ex o
\
(liler/rain )
,~O '\ ' ~ ~ k ~
2.86
20
0
I
I
8
I
I
12
I
I
I
16
I
20
I
I 24
Binary mid-carbon number
Fio. 8. Maximum production rate variation with methyl ester carbon number and carrier gas flow rate for a 3 foot X 1 inch i.d. SE-30 column. [A. Rose, D. J. Royer, and It. S. Henly, Separation 8ci. 2, 257 (1967).] E v a l u a t i o n of Columns
Many preparative GC columns are evaluated by the size of the sample that can be injected and a separation obtained. However, for many purposes production rate is the important factor and the maximum production rate does not necessarily occur at the operating conditions which allow the largest sample size. This fact is shown in Fig. 10. The maximum separable size and the elution time both decreased with increasing column temperature, but in such a manner that the produc-
[51]
PREPARATIVE GAS CHROMATOGRAPHY
463
40
35 =g
Carrier gas
30
o
25
,~
2.00
20
E= E x o
15 I0
5 0
0
I
I
I
I
I
I
I
4
8
12
16
20
24
28
Binory mid-carbon number FIa. 9. M a x i m u m p r o d u c t i o n rate v a r i a t i o n with m e t h y l ester carbon n u m b e r and
carrier gas flow rate for a 3 foot X 1 inch i.d. EGS column. [A. Rose, D. J. Royer, R. S, Henly, Separation 8ci. 2, 257 (1967).]
2 7 o ~ 6
~0
~5
26
® o
i
zz
4
~ ~
50~ 40 .~_ E
30 .~
E 3i
14 K E ~0 .{
o
o
g 20 =
g~
L) .~00
240 Column
temperature
280
6
o
(°C)
FIG. 10. Maximum production rate, sample elution time, and maximum separable sample size as a function of column temperature. 3 foot X 1 inch i.d. SE-30 column. Carrier gas flow rates 1.54 liters/minute. [A. Rose, D. J. Royer, R. S. Henly, 8eparatlon 8ci. 2, 257 (1967).]
464
O~NEP~L ANALYTICAL METHODS
[51]
lion rate passed through a maximum with increasing temperature. Figure 11 shows some of the chromatograms from which the data in Fig. 10 were obtained. Theoretical plates as a function of sample size are often reported and the plates reported normally decrease with increasing sample size. Theoretical plates calculated from the solute peaks are valid only at small sample sizes. The widening of peaks under overloaded conditions is not due to a decrease in the plate equivalence of the column. 2, e The columns used for the methyl ester work reviewed above showed efficiencies of 200-300 plates per foot with small samples, s This is in the range of °C 1.0 ml 0.7 ml/hr
| li r252°c ~) •.,=
| || I-" 3.Oml ~ I~ 34.0ml/hr II
,
l /1 j, /
2
~,
r215°C Column temperature F 5.5 ml Sample injection .,~LI65ml/hr Max" production rotelt.s )
~
~'
Ib'
&
i~,
~,
113
2'0
Elution time in minutes oiler injeclion of sample FIe. 11. M a x i m u m separable sample ~ize peak characteristics at varying column
temperatures. C,,/C_n, injections on a 3 foot X 1 inch i.d. SE-30 column. Carrier gas flow rate 1.54 liters/minute. [A. Rose, D. J. Royer, R. S. Henly, Separation Sci. 2, 257 (1967).]
many packed analytical columns, although packed analytical columns containing 400-500 plates per foot can be made. However, the operating conditions which resulted in the maximum production rates with overloaded columns were far from those giving the maximum plates per foot for small samples, a9 This is not to say that the true plate efficiency of a column is not important. If one considers the above methyl ester work, under nonoverloaded column conditions, separation factors were high and the total number of theoretical plates available were far in excess of the minimum required for separation of the binary mixtures. With small samples, this results in a large distance between peaks and allows room for overloading the column. If a column is equivalent to the minimum number of plates required for the separation of a small sample (peaks just barely separated), it can be expected that little, if any, overloading can be tolerated. Increasing the plate equivalent of the column would increase the amount of tolerable overloading.