mobilities and orderly sequence in tetrahydrofuran contrast sharply with the low mobilities and disturbed sequence observed in dioxane. In many of the solvents, especially with a nonacidic background electrolyte, sorption and limited solubility of the migrating ions may have contributed to the low mobility and anomalous sequences. The changes in migration sequence and the mobility with variation of solvent give no clear concept of the migrating species. It was not possible to correlate either the migration se-
quence or the observed mobilities with the functional groups, electronic configuration, or dielectric constant of the solvents. N o general statement can be made, therefore, to guide in predicting migration sequence or mobilities of alkali metal ions in organic solvents.
( 3 ) Sato, T. R., Kisieleski, W. E., Korris, W. P., Strain, H. H., ANAL. CHEX25, 438 (1953). ( 4 ) Sato, T. R., Norris, W. P., Strain, H. H., Zbid., 26,267 (1954). ( 5 ) Z b i d , 27, 521 (1955). ( 6 ) Schier., 0.., Anaew. Chem. 6 8 , 63 (1956). .
I
LITERATURE CITED
(7) Seiler, H., Artz, K., Erlenmeyer, H., Helv. Chim. Acta 39, 783 (1956). (8) Strain, H. H., ANAL.CHEM.32, 3 R
( 1 ) Evans, G. H., Strain, H. H., ANAL. CHEM.28,1560 (1956). ( 2 ) Harasawa, S., Sakamoto, A., J . Chem. SOC.Japan, Pure Chem. Sect. 73, 614 (1952).
RECEIVED for review October 5, 1959. Accepted January 22, 1960. Work performed under auspices pf United States Atomic Energy Commission.
(1960).
Optimization of Resolution-Time Ratio with Packed Chromatographic Columns R. J. LOYD, B. 0.AYERS, and F. W. KARASEK Research and Development Deparfmenf, Phillips Pefroleum Co., Barflesville, Okla.
b The effects of the quantity of partitioning agent and the choice of carrier gas on the resolution-time ratio obtained with packed chromatographic columns were investigated. Use of columns containing a low proportion of partitioning agent and a carrier gas of low viscosity and high diffusivity gives a ninefold improvement in the time required toobtain a chromatographic separation.
S
obtained with a given chromatographic column (1,S, 14). During these studies a small-volume sampling valve was constructed to minimize the influence of sample addition. This valve has a maximum sample volume of 0.05 ml. and is designed to minimize dead volumes in the valve and fittings. It introduces the sample into the column in a cylindrical slug with a total width not greater than twice theoretical with only slight tailing. This work has led to the development of a pneumatic diaphragm valve with similar characteristics (10).
Chromatographic columns cannot be optimized independent of sample addition and detection systems. The quantity of sample and the manner in which it is added to a c,hromatographic column strongly influence the shape of the chromatographic peaks and hence the degree of separation which can'be
Two types of detectors are used: a small-volume thermal conductivity detector and a microhydrogen flame ionization detector similar to that of McWilliam and Dewar (13). The time constant for the ionization detector is 8 milliseconds, while that of the thermal conductivity cell is 160 milliseconds. To assure no detector limitations, the thermal conductivity cell was used in this study only to determine capacity factors. A Sanborn Model 150 recorder is used. Columns are fabricated from stainless steel tubing '/a inch in outside diameter having a wall thickness of 0.02 inch. The support material is Johns-Manville Chromosorb 100/140 mesh, used as obtained from the supplier without additional screening or chemical treatment. Eastman Kodak bis[2 - (2 - methoxyethoxy) ethyl] ether is used as partitioning agent. Packing for the columns is prepared by thoroughly mixing the proper weight of partitioning agent with dry solid support. Reproducibility of packing from column to column is obtained by carefully following a standardized packing procedure. The sample valve, chromatographic column, and thermal conductivity de-
TUDIES on
the effect of various parameters, such as size distribution of solid support, per cent of liquid partitioning agent, and type of carrier gas on the efficiency of chromatographic columns (1, a, 12,15),have been concerned primarily with obtaining the maximum number of theoretical plates from a chromatographic column and secondarily with the time required to obtain a separation. It does not follow necessarily that a column having the larger number of theoretical plates will provide a required resolution between two sample components in a minimum time. The present study has the specific aim of optimizing the resolution-time ratio to permit chromatographic separations to be made more rapidly. INSTRUMENTATION
698
ANALYTICAL CHEMISTRY
SECONDS
Figure 1. Chromatogram from a 30% ether column with nitrogen carrier gas Chromatogram A lsobutane n-Butane 1-Butene f isobutylene frons-2-Butene cis-2-Butene 6. 1,3-Butadiene
1. 2. 3. 4. 5.
tector are enclosed in a thermostated air bath regulated to 25' 0.1" C. The hydrogen flame detector is placed external to the air bath with a minimum length of 1/16-inch tubing connecting the detector to the column. A concentric tube arrangement is used to add hydrogen to the detector when nitrogen or helium carrier gases are used. A reproducibility of =kO.l% is realized. THEORY
The function to be optimized for high speed operation of chromatographic columns is the ratio of the resolution of two sample components, i and j, to the retention time of the ith component, &/ti. Resolution is defined as
where ti and t i are the retention times for components i a n d j measured to peak maxima from the time of sample injec-
Y
E
C
n
0 06
A
0 04
IO
0
20
30
20
IO
30
50
40
70
60
60
4' COLUMN AZHYOROGEN I: NITROGEN
I 80
20 40 60 O U T L E T FLOW-CCIMINUTE
Figure 3. Efficiency of 30% ether column with various carrier gases
SECONDS
0
A
Resolving power ( 8 )may be defined in terms of the retention time of ti single component, i, and the residence time of a nonabsorbed gas as ( 7 )
100
90
SECONDS
Figure 2. Chromatograms from 10% ether columns with various carrier gases 1. 2. 3.
lsobutane n-Butane 1-Butene
4. 5. 6.
+ Isobutylene
'
frons-2-Butene cis-2-Butene 1,3-Butadiene
4"IZ'-HYDROGEH Az4' -HYDROGEN
The relationship to be optimized is then
a t a minimum acceptable value of Ri = u/(fi-I), where a = 0.7 to 0.8. Residence time for a nonabsorbed gas is defined (11) by Equation 9.
1
0 06 I 0
+ w
I n Equation 9, V , is the volume in the column occupied by the gas phase, F is the flow rate measured a t the outlet pressure, Po and Pi/P, is the ratio of inlet to outlet pressures. Combining Equations 8 and 9 gives
8
0 04
I
I
002 I
1
1
I
I
I
tion and At is the distance intercepted along the base line by tangents through the points of inflection of the peaks (4). While this is rigorously correct, it cannot be related in a simple manner to column parameters. A criterion for resolution is needed only in the case of difficultly separable peaks, where t j > ti-t,. If N i s NI, N being the number of theoretical plates calculated from the chromatogram as recommended by the nomenclature committee a t the 1956 London symposium (6),then A h and A t j are nearly equal. Resolution may then be defined (9) with good approximation by rewriting Equation 1 as
The retention time for the ith sample component is related to the residence time, tu, of a nonabsorbed gas by ti
=
(1
+ Kr)t,
(3)
where (4)
I n Equation 4, Ci is the capacity of the liquid phase for the i t h sample component and C is the capacity of the mobile or gas phase for the sample component (6). Equation 2 can be rewritten as NU¶ Ki Rij
=
(Pij
- 1) 1 + K i
X 4
(5)
where Bij = K j / K i
(6)
The important contributions of the capacity ratio, Ki, and the relative solubility, &, to the resolution obtainable with a chromatographic column are shown by Equation 5. It is possible to have a column with a large number of plates, yet with a low resolution, should Ki and j%j be low.
The quantities on the right side of Equation 10 are interrelated and cannot be varied independently. However, it is of interest t o examine the effect of these quantities on Ri/ti and to determine how they vary with experimental parameters such as the quantity of liquid phase and the type of carrier gas. The type and quantity of partitioning agent are related to the number of theoretical plates and the carrier gas flow rate by the van Deempter equation (3). This equation predicts, and it has been experimentally verified ( I D , that a reduction in the quantity of liquid phase results in an increase in the number of theoretical plates and in the optimum carrier gas flow rate. Both these changes contribute to an increase in the value of Ri/ti. The factorKi/(l Ki)' is also a function of the quantity of partitioning agent and has a maximum value a t K i equal to 1. The effect of the type of carrier gas on the quantities of Equation 10 is more uncertain. According to the van
+
VOL. 32, NO. 6, MAY 1960
699
12
A HYDROGEN I:NITROGEN HELIUM
0
6 50
100
I50 ELUTION TIME-SECONDS
200
250
300
Figure 5. Resolving power vs. elutiontime for 1 ,g-butadiene for 30% ether column with various carrier gases
Deempter equation, an increase in the difFusivity of the carrier gas leads to an increase in the optimum flow rate but a decrease in the number of plates. The van Deempter equation, with an added term for nonequilibrium due to gaseous diffusion (7), predicts that the minimum plate height will be independent of carrier gas dxusivity when liquid diffusion effects are negligible. Under these conditions, optimum carrier gas velocity will be proportional to the first power of the diffusivity of the carrier gas rather than the one-half power as predicted by the unmodified equation. Use of a carrier gas of low viscosity is indicated by the pressure term of Equation 10, which increases as the viscosity decreases. The quantity of liquid phase and the type of carrier gas are thus important experimental parameters in the optimization of R&. The effects of these
Table 1. Nominal Composition of Phillips Hydrocarbon Mixture 37
Concentration, Sample.Component Mole % Isobutane 3 %Butane 14 Isobutylene (2-methylpropene) 7 1-Butene 17 &-%Butene 22 trans-2-Butene 19 1,a-Butadiene 18
Table 11.
Run No. 1 2
3 4 5 6 7
700
4 *I2'-HYDRDGEN A=4'-HYDROGEN I-4'-NITRDGEN 0*4'-HELIUH
L 0
I
1 50
ANALYTICAL CHEMISTRY
I
I
2 00
250
Figure 6. Resolving power vs. elution time for 1,3-butadiene for 10% ether columns with various carrier gases
parameters are investigated in this study. EXPERIMENTAL RESULTS
Phillips hydrocarbon mixture 37 is used as the reference sample to obtain the data in this study. The nominal composition of this mixture is given in Table I, Four-foot lengths of column give data in the desired range and this length is adopted as standard. Variations in plate height with particle size, homogeneity of packing, and column diameter were studied experimentally and found to be in excellent agreement with the results of Scott (16) and of Bohemen and Purnell ( I ) . On the basis of these experiments, columns of l/s-inch diameter and 100/140 mesh solid support were chosen for the remainder of the studies. Experimental and calculated results obtained from studies of the effect of quantity of liquid phase and carrier gas type on the resolution-time ratio are listed in Table 11, which also identifies
Data for 1,3-Butadiene a t Optimum Carrier Gas Flow Rates
CarFl rier cc./ t{, Column Gas Pi/P, Min. Seo. 4fOOt, Nz 1.73 14.2 227.3 He 2.46 33.0 116.6 30% HZ 1.87 35.5 90.9 Kz 1.81 14.7 94.5 4fOOt, He 3.33 58.8 33.6 10% H1 2.40 62.2 25.2 12f00t, HZ 4.33 75.5 96.8 10%
I 100 IS0 CLUTION TIME-SECONDS
Ki N 7.31 2,390 7.31 1,970 7.31 2,100 2.21 3,830 2.21 3,310 2.21 3,950 2.43 11,300
chroRdti matogram 0.0474 A 0.0838 0.111 10.6 0.112 B 9.91 0,295 10.8 0.430 C 18.8 0.195 D
Ri 10.8 9.76 10.1
corresponding chromatograms of Figures 1 and 2. Changes in plate height for 1,3-butsdiene as a function of carrier gas flow rate were determined with nitrogen, helium, and hydrogen carrier gases for the columns indicated in Table 11. Results for the 30% column are shown in Figure 3 and those for the 10% column in Figure 4. Under the optimum conditions of Figure 4! one can decrease analysis time by factors of 2 to 3 without significantly affecting the plate height. One can also note the reproducibility of these parameters between the 4- and 12-foot columns. Values of the resolving power, Ri, as calculated from Equation 7 for each column and for each carrier gas are shown in Figures 5 and 6. DISCUSSION
Effect of Decreasing Quantity of Liquid Phase. When the liquid phase is reduced from 30 to lo%, optimum flow rates and pressure drops are the same for both columns when operated with nitrogen carrier. Equal resolution is obtained with the two columns. This arises from the compensating effects of K J ( l + K i ) and N I P in Equation 7. The retention time for butadiene, however, decreases by a factor of 2.4 as predicted by Equation 3. Hence, the value of Rijti is larger for the 10% column. When operated with hydrogen carrier, an additional increase in Rdti is obtained by reducing the quantity of liquid substrate. This results from an increase in the last two terms ,of Equation 10 in going from 30 to 10% liquid. The difference in optimum plate
heights for hydrogen and nitrogen carrier (Figure 3) indicates that diffusion in the liquid phase contributes significantly to plate height when 30% liquid is used. The equal minimum plate heights obtained with hydrogen and nitrogen on the 10% column (Figure 4) indicate that this effect has become negligible. Effect of Use of Different Carrier Gases. The data obtained on the 4foot 30% column with nitrogen and hydrogen carriers show t h a t the number of plates, and hence the resolution, is slightly smaller with hydrogen. HoLTever, the increased optimum carrier gas flow rate obtained with this gas reduces the retention time for butadiene by a factor of 2.3. The effect of the carrier gas is more pronounced for the 10% column. Again equivalent resolution is obtained with either carrier, but the time required with hydrogen is only one fourth that required for nitrogen. Performance with hydrogen is superior to that with helium on both the 3oY0and 10% columns. I n each case, the smaller number of plates and higher pressure drop obtained with helium carrier make the value of Ri/ti for helium smaller than for hydrogen. If the column used to obtain run 5 were lengthened to provide the resolution obtained in run 6, the helium column would require 75% more time than the hydrogen column.
Eff ect of Column Length. Figures 5 and 6 show that the values of Ri/ti increase with decreasing retention time (increasing carrier gas flow rate). This indicates t h a t for a required resolution, Ri, a larger value of & / t i may be obtained, in some cases, by operating longer columns a t flow rates well above optimum. Thus the 10% column which will give an optimum value of Ri/tt with a resolution of 10.8 will lie somewhere between the 4-fOOt and 12-foot columns and will operate a t a flow rate greater than 62 ml. per minute. The resolution obtained with the 12foot column, a t optimum flow rate, is 4 3 times that obtained with the 4-foot column, as predicted by Equation 8. The value of Ri/ti decreases in agreement with Equation 10. Comparison of chromatograms B and D is of interest. The time required to obtain the separation is the same in both cases, but the resolution is almost twice as good with the 12-foot column and hydrogen carrier. The combined effects of the quantity of liquid phase and the type of carrier gas are shown in Table 11. The same resolution is obtained from the 4-fOOt 30% column with nitrogen carrier and the 4-foot 10% column with hydrogen carrier, while the ratio of the times required is 9 to 1.
LITERATURE CITED
(1) Bohemen, J., F e l l , J. H., “Gas
Chromatogra hy, D. H. Desty, ed., p. 6, Academic ires,, New York, 1958. (2) Cheshire, J. D., Scott, R. P. W., J. Znst. Petrol. 44, 74 (1958). (3) Deem ter, J. J. van, Zuiderweg, F. J., Klinkenterg, A., Chem. Eng. Sci. 5 , 271 (1956). (4) Desty, D. H., ed., “Gas Chromatography,” p. xi, Academic Press, Xew York, 1958. (5) Desty, D. H., ed., “Vapor Phase Chromatography,” p. xi, Academic Press, Kew York, 1957. ( 6 ) Golay, M. J. E., ASAL. CHEM.29, 928 (1957). (7) Gola;, hf. J. E., “Gas Chromatography, D. H. Desty, ed., p. 36, Academic Press, Kew York, 1958. (8) Golay, M. J. E., Nature 182, 1146 f 1958).
(9) Jones, W. L., Kieselbach, R., ANAL. CHEM.30, 1590 (1958). (10) Karasek, F. W., Ayers, B. O., Instrument Society of America, 5th National Symposium on Instrumental Methods of Analysis, Houston, Tex., I!lay 1959. (11) Keulemans, A. I. bl., Gas Chromatography,” p. 135, Reinhold, New York, 1957. (12) Keulemans, A. I. M., Kwantes, A., “Vapor Phase Chromatography,” D. H. Desty, ed., p. 15, Academic Press, New York. 1957. (13) McWfiliam, I. G., Dewar, R. A., “Gas Chromatography,” D. H. Dest , ed., p, 142, Academic Press, New Yorz, 1958. (14) Porter, P. E., Deal, C. H., Jr., Stross, F. H., J.Am. Chem. SOC.78,2999 (1956). (15) Scott, R. P. W., “Gas Chromatography, D. H. Dest ed., p. 189, Academic Press, New &rk, 1958. RECEIVED for review October 2, 1959. Accepted February 11, 1960.
Molecular Sieve Adsorption Application to Hydrocarbon Type Ana lysis JOHN G. O’CONNOR and MATTHEW S. NORRIS
Gulf Research & Development Co., Pittsburgh, Pa.
b A method is described for the determination of normal hydrocarbons in 100’ to 650” F. petroleum distillates using powdered Molecular Sieves, Type 5-A, without prior fractionation of the distillates into narrow boiling cuts. The accuracy, calculated as a standard deviation, is to =tO.8% of the normal hydrocarbon content of the sample. The procedure consists of weighing the sample into an adsorption column containing the sieves and eluting the nonadsorbed hydrocarbons with isopentane. The excess eluent is removed by vacuum evaporation at room temperature, and the adsorbed normal hydrocarbon content
determined by weighing. Also described is a procedure for recovering the adsorbed normal hydrocarbons from the sieves by extraction with n-pentane. The mechanism of recovery appears to be a diffusion-controlled process, with the rates of desorption varying inversely with the molecular weight.
P
knowledge of the normal hydrocarbon content of wide boiling range hydrocarbon distillates is important to the petroleum industry. This information aids in the evaluation of various processes, such as the hydroRECISE
isomerization of gasolines or the production of kerosines of low freezing point. Molecular Sieve adsorption has found a convenient application for these purposes. Barrer (I,.%’) reported the use of naturally occurring zeolitic minerals for the adsorption of normal paraffins and investigated conditions for adsorption, the size of the molecules adsorbed, and the relative rates of adsorption on chabazite. When synthetic zeolites became available commercially, Nelson, Grimes, and Heinrich (7) and Schwartz and Brasseaux (9) published methods for measuring the normal hydrocarbon content of petroleum distillates. Both VOL. 32, NO. 6, M A Y 1960
701
