DISCUSSION
Differential electrical migration in special background solutions has provided simple, rapid procedures for the isolation of particular ions from complex mixtures. There are, however, circumstances when the elongation of the zone of one ion, especially uranium, gold, or platinum, may cause it t o overlap the separate, well-defined zone of another. Under these circumstances, the method of separation must be considered in the light of the possible presence of the contaminating substance. If contaminating substances are shown to be absent in mixtures to be examined, these migration methods are especially effective for the separation of particular metal cations that would otherwise be contaminated. As in so many differential migration methods, the behavior of individual ions by electrochromatography was subject to considerable variation. Readily reproducible separations were obtained only when the specified concentrations and pH values n-ere maintained. Even under these conditions there were many noticeable variations, particularly the migration distance and the size of the individual zones. If the electrochromatographic method were utilized with pairs of ions, com-
plete and absolute separations could be made, as demonstrated with silver and thallium ions (5). But if one considers 24 ionic species, as in these separatory investigations, there would be 276 pairs to be separated, 2024 triplets, etc. The exploratory work has shown that many combinations of two or a few kinds of ions would be separable under the conditions employed. Nevertheless systematic variation of all the possible pairs, triplets, quadruplets, etc., would require a prodigious amount of work. The separations by electrochromatography bear little or no relation to those obtained by the conventional chemical methods of analysis. From the objectiye of this investigation, these electrical migration methods are not directly applicable to the separation of the elements of the common analytical groups. There are, however, many areas where the electrochromatographic methods may render valuable service t o the analyst-the separation of small quantities of ions that may be detected a t low concentration, as potassium to be estimated by flame photometry, cesium in fallout to be estimated by radioactivity, uranium in soils to be estimated by color reactions, etc. I n many of these areas of investigation, the electrochromatopraphic
methods have several advantages over chemical methods of separation. They permit the complete separation of minute quantities of the desired ions. The separated species are present in a form suitable for detection and estimation, and there is no loss of the species by preliminary fractionation. ACKNOWLEDGMENT
J. F. Binder and J. J. Hines contributed preliminary, exploratory experiments on which some of these results are based. LITERATURE CITED
(1) Engelke, J. L., Strain, H. Ti., \Vood,. S. E.. ANAL.CHEM.26.1864 (1954).
Ibid., 33,527 1 (6) Wood, S. 1 26,1869 (195.1,.
RECEIVED for revieK September 26, 1962. -4ccepted November 28, 1962. Based on work performed under the auspices of the U. S. Atomic Energy Commission.
Chromathermography, the Application of Moving Thermal Gradients to Gas Liquid Partition Chromatography R. WAYNE OHLINE' and DONALD D. DeFORD Department of Chemistry, Northwesfern Universify, Evanston, 111.
b The application of moving thermal gradients to gas chromatography (chromathermography) yields a separatory method which has some unique properties. Among these are a substantial decrease in the time necessary for a given analysis and a particular suitability to trace analysis. A quantitative comparison of isothermal gas chromatography (IGC) and gradient gas chromatography (GGC) is given.
I
1951, Zhukhovitskir and coworkers published a paper (6) relating to the application of moving thermal gradients to gas solid partition chromatography. The basic features of the technique were as follows : A linear column held the adsorbent, and a furnace placed a t the head of the column produced a temperature N
gradient. The highest temperature was a t the head of the column, and the temperature decreased to some lower value some distance down the column. The sample was injected at the head of the column into a carrier gas stream, and the furnace (and hence the temperature gradient) moved down the column. Tudge (5) has recently extended and noted some corrections in the Soviet work on chromathermography for gas solid partition. -4 complete bibliography is included in his paper. With an arrangement of this type, the most volatile or least strongly adsorbed components in the sample move rapidly down the column until their rate of movement is slowed down because of the decreased volatility in the low temperature region of the gradient. The more strongly adsorbed compo-
nents (least volatile) tend t o be held in the region of higher temperature. Thus a separation is effected. As the furnace moves down the column, the components are desorbed and eluted one by one. If it is assumed that the distribution isotherms and experimental conditions are such that each solute moves less rapidly than the furnace if placed in the coldest region of the temperature gradient, but more rapidly if placed in the hottest region, it is clear that a steady state must eventually be reached in which each solute moves a t the same rate as the furnace. A second feature of gradient gas chromatography (GGC) is that the gradient produces a compressing action Present address, Department of Chemistry, New Mexico Institute of Mining and Technology,Socorro, N. M. VOL. 35, NO. 2, FEBRUARY 1963
227
D
Figure 1.
on the band. The trailing edge of a band is always a t a higher temperature than the leading edge. Hence, the fraction of solute in the gas phase always tends to be greater a t the tailing edge than the leading edge. As a consequence of this, a compressive action results. Counteracting the compressive action of the gradient are the factors which normally cause band spread in chromatography. Here again, it is clear that a steady state will eventually be reached in which the compressive action of the gradient exactly balances the forces which normally tend to expand the band. R e have extended the work of the Soviet workers into the realm of gas liquid partition, paying particular attention to the steady-state band width. EXPERIMENTAL
Figure 1 shows the essentials of the apparatus. To keep the furnace centered with respect to the column, the entire assembly was mounted on an optical bench. Support C held the column stationary with respect t o the bench. The furnace, D, was held by two similar supports, E and F . E held the hot end of the furnace, and F the cold end. A latch on F could be attached t o the endless chain, G, a t will, and a microswitch a t the end of the bench automatically shut off the driving mechanism when it made contact with F . The furnace moved in the direction indicated a t D. Carrier gas entered the column a t A through a l/le-inch brass tube. The packed portion of the column began approximately a t P, which was slightly below the highest temperature in the gradient when the furnace was a t its starting position. A thermocouple, placed in the column at K , was used to obtain the temperature as the furnace passed over the end of the column. A small-bore exit tube, T, led to a hydrogen flame ionization detector. The sprocket wheel, H , was connected to a Zero-Max synchronous motor equipped with a Zero-Max gearhead 228
ANALYTICAL CHEMISTRY
Basic features of column and heater mounting
through a 30 to 1 reducing gear. With this arrangement, the speed of the furnace could be varied continuously from about 0.05 t o 2.8 cm. per second. The furnace consisted of an aluminum bar 15 inches long and 19/leinches in diameter. The column passed through a hole l / 4 inch in diameter, which was drilled through the center of the bar. Fitted on each end of the heater were two large rubber stoppers. Brass cylinders 3l/2 inches in diameter were fitted over these rubber stoppers to form two heat-exchanging compartments. The temperature gradient was usually obtained by running tap water through the leading heat exchanger, and steam through the trailing heat exchanger. In obtaining lower gradients, thermostated water was pumped through the heat exchanger compartments. The highest temperature gradient attainable was 8.5" per em. The lowest gradient used was 1.0" per cm., but lower gradients could probably be obtained if they were desired. The gradients produced by this device were linear, particularly in the central region. Heat transfer between furnace and column was no problem. Various experiments showed that the temperature of the wall of the column followed that of the furnace up to the highest furnace velocities used (1.5 cm. per second), The chromatograms were obtained by feeding the signal of the thermocouple (at K in Figure 1) into the X response of a Moseley X-Y recorder, and feeding the signal from the flame detector into the Y response of the recorder. When the column was operated with the gradient, four sigma degrees were determined by the base line intercept method, and this value u-as divided by 4 and the gradient (degrees per centimeter) to obtain sigma in centimeters. Chromatograms recorded on the X - Y recorder had essentially the same appearance as those obtained from ordinary IGC, except that all the eluted bands had about the same base width. Except for the study of sigma as a function of the gradient, experimental results presented here mere all obtained with a single column. The column was a l/c-inch o.d., 0.014-inch wall brass tube packed with 100- t o
120-mesh firebrick (20 grams of diisodecyl phthalate per 100 grams of firebrick). The length of the packed portion was 56 em. The weight of the column packing was 6.96 grams. The dead volume of the column was 12.0 cc., and A , / & was 0.101. Diisodecyl phthalate was used as solvent in all the experiments, because the work of Porter, Deal, and Stross (4) indicated that this solvent would yield linear isotherms with the normal paraffins used as solutes. Phillips research grade hydrocarbons were used throughout this work. For isothermal operation of a given column, the furnace was removed from the bench, and a brass jacket, constructed with Swagelok fittings, n-as placed over the column. Water from a constant temperature circulator was then pumped through this jacket. The temperature of the water was constant to &0.05" C. Although most of the data presented here are limited to the column described above with n-heptane as solute, the work n-as extended to other columns and solutes with complete verification of the equations derived below. THEORY
Characteristic Temperature. The basic theory of GGC can best be understood in terms of isothermal operation. I n any chromatographic column, the speed of a solute through a volume element of the column is equal to the product of the mobile phase velocity and the fraction of the solute present in the mobile phase. Expressed mathematically, this is
where Vband V , are the velocity of the band and gas, respectively, C, and C, are the masses of solute per unit volume of the mobile and static phases, and A , and A , are the cross-sectional areas of the mobile and static phases. In addition CJC, may be expressed by the relation
'E
40
' TX I 0
5
Figure 2. Graphical demonstration of chromathermographic characteristic temperature equation
where Q is the molar heat of vaporization, R and T have the usual significance, and a is a constant. Combination of Equations 1 and 2 yields
Equation 3 is verified by the fact that plots Of log [ ( V Q- v b ) / v b ] US. I/T are linear with slope &,/2.3Ras shown in Figure 2. The case is now considered where the column is run under gradient conditions. When this is done, each band may be made to move at the same velocity as the furnace T'j-that is, T/'b = Thus, for gradient conditions, the following relation holds:
v,.
where T,, the characteristic temperature, is the temperature around which a band is centered. It is of interest to note the manner in which the members of a homologous series are eluted in GGC. Writing Equation 4 explicitly in terms of T,one obtains Q
Thus, T, varies directly as Q. If a does not change too much from member to member of a homologous series, the consecutive members of the series will be approximately equally spaced in a linear temperature gradient. This approximate linear spacing was observed for the normal hydrocarbons used in this work. The spacings in temperature were approximately the same as the difference in boiling points. Thus in Figure 2, if ( V , - V,)/V, is set a t 10.0, and n-octane through n-hexane remain in the gradient, the reciprocal temperatures around which the octane, heptane, and hexane bands would be centered are, respectively, 287,309, and 340 X (' K.)-I (75', 51 ', and 21 O C.). The preceding treatment neglects the effects of both the temperature and pressure gradients on the gas velocity. The gas velocity a t any point in the temperature gradient, for the case of no pressure drop, is given by the relation
where ( V g ) T , , , is the velocity a t temperature T,, at which the velocity is measured. The dashed line marked "Charles' law correction" in Figure 2 was calculated for the case where the gas velocity is measured at 333 x (' K.)-l. When this correction is ap-
plied, the characteristic temperatures are now 71.0,' 48.5", and 22.0' C., respectively, and not 75", 51', and 21 ' C. (as was calculated for the case of constant gas velocity). In contrast to the case of isothermal gas chromatography, no average pressure correction is applicable in GGC. The gas velocity varies along the length of the column because of the pressure gradient. Therefore, from Equation 5 , T, will also vary (since all quantities except V oremain constant). However, one is interested only in the characteristic temperature a t the instant of elution, and this is given by the velocity a t the outlet pressure. Except for the case of high pressure drop (where the gas velocity changes very rapidly near the end of the column) the band position will continually adjust itself according to the gas velocity at that point in the column, and will be eluted a t a temperature corresponding to the gas velocity a t the exit end of the column. I n general, IGC data are representative of the average column parameters, but the data from GGC depend solely on the parameters a t the exit end of the column. I n this work, special care was taken to obtain uniform columns, and low pressure drops were used (PindPoutlot
