ard mixture may prove to be worth while, since at that point, if enough conditions are held constant, deviations from ideal behavior may be expected to be the same for the two mixtures. For quantitative analysis, the probe is mainly of interest where a reservoir cannot be used, and only when the properties of the components are so nearly alike as to prevent a chromatographic separation.
ACKNOWLEDGMENT The author thanks Vernon H. Dibeler for reading the manuscript* RECEIVED for review September 11, 1967. Accepted D e cember 18,1967.
Accurate Mass Spectrometric Determilnation of Low Concentrations of Carbon Dioxide in Nitrogen Ernest E. Hughes and William D. Dorko Institute for Materials Research, Division of Analytical Chemistry, National Bureau of Standards, Washington, D . C . 20234 A method for the rapid and accurate determination of low concentrations of carbon dioxide in nitrogen has been developed. The method is based on highpressure mass spectrometry in which the mass 44 to mass 28 ratio is compared to the same ratio in a carefully calibrated standard. Concentrations from 180 to 380 ppm were determined with an accuracy of better than 1%. The accuracy of the method depends on calibration of the mass spectrometer with a series of carefully analyzed standards. Any system which can be chemically converted to a mixture of carbon dioxide and nitrogen can be analyzed by this method with equivalent accuracy.
INCREASED INTEREST in possible long-term variations in atmospheric composition has led to a close scrutiny of the carbon dioxide content of air in various parts of the world. Accurate intercomparison of data from different observers and from different locations requires a carefully characterized reference gas whose carbon dioxide content approximates that of the atmosphere. In the past, most standards of this type were produced and distributed by the Scripps Institute of Oceanography under the direction of C. D. Keeling. The investigation described below was undertaken in anticipation of future demands which would seriously tax the facilities of the former supplier, The production of a standard gas mixture requires that there be available a method of analysis of the required accuracy which is suitable for the routine monitoring of the concentration of carbon dioxide in a large number of samples. The method described below is a result of a continuing investigation of the application of high-pressure mass spectrometry to gas analysis problems. In general, mass spectrometers for gas analysis have an inlet system in which the gas to be analyzed is confined at a known pressure and from which it is allowed to flow through a molecular leak into the ionization region. At inlet pressures below 0.5 mm, the measured current due to a particular ion is directly proportional to the absolute inlet pressure of the gas producing the ion. However, deviation from this direct proportionality begins to occur as inlet pressures are increased. The quantitative relationship is not the same for all gases so that empirical calibration data are required. This is well illustrated by the recent work of Suttle, Emerson, and Burfield (I) in which varia(1) E. T. Suttle, D. E. Emerson, and D. Burfield, ANAL.CHEM., 38, 51-3 (1966).
750
ANALYTICAL CHEMISTRY
tions of the sensitivity (ion yield per unit of inlet pressure) with increasing inlet pressure are given for a number of trace constituents in helium,. Some earlier work at the National Bureau of Standards in this laboratory ( 2 ) had indicated the same effects in mixtures containing low concentrations of benzene and butane in air. Further, it was observed that the sensitivity of the minor constituent tended to increase relative to that of the major. Inlet pressures in the region where sensitivity data for the individual components can be used reliably in calculating the concentration in mixtures are too low to yield ion currents which can be measured with sufficient accuracy to determine atmospheric carbon dioxide concentrations (0.03 mol %) with a reliability of better than about 3 0 z . Thus it was evident from the beginning that if mass spectrometry was to be the method of analysis, the inlet pressures would have to be much higher than for normal operation. The conditions necessary for the degree of accuracy required would have to be determined using gas mixtures with compositions known to better than 1 of the carbon dioxide present. EXPERIMENTAL Instrumentation. The mass spectrometer used was a Consolidated Electrodynamic Corp. Model 21-103C. Several modifications were found necessary or desirable for analyses of low concentration mixtures. The ion pump normally used on the exhaust pumping system was supplemented with a two-stage mercury diffusion pump connected in parallel. The input resistor of the low sensitivity side of the amplifier was changed to allow a difference between low and high sensitivity of 1 to 50 rather than 1 to 10. A strip chart recorder was connected to the amplifier output to facilitate rapid data readout. For the experiments involving the three primary calibration mixtures, a McLeod gauge was attached to the inlet system to provide pressure measurements to 1 %. For other low Concentration mixtures prepared later, pressures were not measured with great accuracy but were duplicated from run to run by expansion from a small volume at a pressure duplicated to about 3 %. Samples of the low concentrations mixtures were admitted to the inlet system of the mass spectrometer at a pressure of about 1 mm. At this pressure and at an ionizing current of about 40 PA, both the ion current at mass 28 (actually t E / e = 28), read with the amplifier set at low sensitivity, and that at (2) J. K. Taylor, Ed., National Bureau of Standards Technical Note 403, U. S. Government Printing Office, Washington, D. C.,
Sept. 1966.
*----
!I
1
CYLINDER
CONNECTIONS
TO VACUUM PUMP
Figure 2. Manifold showing arrangement of valves and gauges
rc
Figure 1. Typical mass spectrum showing method of reading peak heights B is the ion current caused by m/e rent caused by m/e = 44
= 28
and A
+ B + C is the cur-
mass 44, read with the amplifier at high sensitivity, were of sufficient magnitude to allow them to be read to better than 0.5%. In practice, the instrument was focused to begin a scan just below mass 28. This mass was scanned about 5 times before the instrument was refocused to start a scan just below mass 44, which was also scanned about 5 times. When calculating the peak heights, the base line was extended from the low sensitivity scan to a particular mass 44 peak in the high sensitivity scan (Figure 1). The peak height of mass 28 was extrapolated to the same mass 44 peak. The peak heights of both mass 44 and 28 were read on the center line of this peak. This assured that both peaks were read after the same time interval had elapsed, thus eliminating problems caused by pressure drop in the inlet system. Diffusive separation after prolonged flow through the orifice into the ionization region was observed with consequent enhancement of the mass 44 peak. This effect was minimized by following a strict time schedule so that the ion currents were all measured at approximately the same time after the sample began to flow. It was found that the rate of change of the ratio of 44 to 28 was about 0.2 % per minute. In most cases, the elapsed time between introduction of sample and recording of ion currents was about one minute so that the error involved, while unavoidable, was insignificant. The time involved in a triplicate measurement and calculation of the mass 44 to 28 ratio is about 20 minutes. If a number of samples are to be analyzed, the average time required for each analysis including calibration seldom exceeds 30 minutes. This has considerable advantage over gravimetric methods for determination of carbon dioxide concentrations of 0.03 mol which may require as much as 4 hours for each determination. Gas Mixtures. Gas mixtures were prepared from analyzed samples of carbon dioxide and nitrogen using the simple gas mixing manifold shown in Figure 2. The gauges for use above one atmosphere were calibrated over their entire range with a dead weight piston gauge. The gauges could be read to better than 0.2% and the piston gauge was accurate to at least 0.1%. The vacuum gauge was calibrated using a mercury manometer read with a cathetometer. It was not used in any subsequent operations but was included
in the manifold to provide capability for measurements at pressures below atmospheric. When preparing gas mixtures by measurement of the pressures of the individual components, it is advisable to use pressures as high as practical with subsequent larger volumes of gas. In this way, errors caused by gases inadvertently trapped in valves and tubing or absorbed in the system become quite insignificant. The best way to avoid these problems in practice is to dilute stepwise starting with a mixture of relatively high concentration. The mixture representing each dilution is carefully analyzed and no errors are introduced over those that would accompany a single step mixing procedure. Gas mixtures were prepared in clean, dry, evacuated 1A gas cylinders. The first component or mixture was added after which the system was closed and allowed to stand for a period of time until temperature equilibrium, as indicated by constant pressure and constant cylinder skin temperature, was attained. The second component or diluent was then added to the cylinder after thorough evacuation of all gauges and tubing in the manifold. The final pressure was then read after temperature and pressure had equilibrated. Care was taken that no gas was lost through back expansion into the manifold during manipulation of valves. The gas mixtures in the cylinders were mixed by inclining the cylinders and heating the bases with an infrared lamp for several hours. The cylinders were then allowed to stand for at least a day to assure thorough mixing. The concentration of carbon dioxide was calculated using the following expression: Pcoz - - x - Tz zcoz Mole per cent Carbon Dioxide = T1 x 100 (1) Pcoz Tz + 5 2 ZCOZ TI ' z N 2 where Ti is the temperature of the cylinder after addition and equilibration of the carbon dioxide and Tz is the ternperature after addition and equilibration of the nitrogen. TI and T? should never differ by more than a few degrees so the simple correction is quite adequate for this system. The pressure of carbon dioxide (Pcoz) is either measured directly or, in the case of an intermediate mixture, is calculated from the observed pressure and the previously determined concentration of carbon dioxide. The compressibility factors, 2,are obtained from the literature (3), and are (3) J. Hilsenrath, C . W. Beckett, W. S. Benedict, L. Fano, H. J. Hoge, J. F. Masi, R. L. Nuttall, Y.S.Touloukian, and H. W. Woolley, National Bureau of Standards Circular 564, U. S. Government Printing Office, Washington, D. C., Nov. 1955. VOL. 40, NO.
4, APRIL 1968
751
SOCKET FOR CUtP
METER
F,LOW
MAGNESIUM PERCHLORATE
3-WAY STOPCOCK
MIGNESIUM PERCHLORITE
U \
SAMPLf CYLlNDtR
I
WET TfST METER
DRYING TUBE
I
Figure 3. Absorption train those applicable a t the temperature and pressure for the individual species. Gravimetric Analysis. The gravimetric analysis depends on adsorption of the carbon dioxide using Ascarite. The adsorption train is shown in Figure 3. The adsorption bulb shown in Figure 4 is of a type similar to that designed by Martin Shepherd of the National Bureau of Standards but never published by him. The three-way stopcock a t the entrance to the train allows the flow rate to be adjusted before beginning the analysis and facilitates periodic adjustment if needed. The mercury bubbler prevents build-up of pressures in the system and allows the system ahead of the adsorption bulbs to be flushed before the analysis. The saturator assures that the gas entering the wet test meter is saturated. The U-tubes filled with magnesium perchlorate prevent diffusion of water vapor into the system from the saturator and also remove traces of water from the gas entering the adsorption train. The wet test meter has a range of 1 liter per revolution and was calibrated by the Fluid Meter Section of the National Bureau of Standards. Adsorption bulbs are connected with ball-joints lubricated with small quantities of a hydrocarbon lubricant. Nitrometer tubing is used for all other joints. The cleanliness of the train is thoroughly checked by passing quantities of nitrogen through the bulbs and observing that no weight change occurs in the first bulb. In practice, the cylinder whose content is to be analyzed is connected with a short length of nitrometer tubing to the inlet of the train. The flow rate is adjusted, after which the stopcock is turned to allow the gas to flow through the train. Barometric pressure at the beginning and end of the gas flow is observed. The temperature of the water in the wet test meter is recorded at frequent intervals during gas flow. Adsorption bulbs number 1 and 2 are weighed against num-
Table I. Determination of Carbon Dioxide in Nitrogen, Mixture No. 1 Analytical Cot in mole per cent method 5 . 5 2 i 0.07a((rz= 5)b Mass spectrometry Orsat analysis 5 . 5 i0 . 2 a ( I 2 = 5)b 5.54 i 0 . 0 8 n ( r ~ = 6)b Gravimetry 5 . 6 0 i O.O@ Calculated value a 9 5 x confidence limits for the true average. b n = number of measurements. c Estimate of systematic error involved in mixing.
SOCKlT'i09 CLAMP
Figure 4. A . Absorption tube B. Rotatable cap showing orientation of side arms
ber 3 used as a counterpoise as in the procedure outlined by Pennington (4). The concentration of carbon dioxide is given by the expression: wt Mole per cent COz =
coz
44,Ol VN2 +44.01 V ,
WtCOz
x
100
Saturation pressure of water in millimeters of mercury at the average temperature of the wet test meter P = Measured atmospheric pressure in millimeters Vm = Molar volume of the gas a t average temperature of the wet test meter and measured atmospheric pressure.
P,
=
All the usual precautions associated with the weighing of adsorption bulbs must be carefully observed. In addition it is necessary to reproducibly wipe the ball-joints immediately after removal of the bulbs from the train. This is done by first removing the excess grease with an adsorbant tissue followed by wiping with tissue soaked in a highly volatile solvent such as dichloromethane. Any traces of grease in the tubing at the joints is carefully removed with a pipe cleaner soaked in clean solvent. RESULTS
Analysis of Mixtures. An initial mixture of carbon dioxide of approximately 5 mol % was prepared from analyzed samples of nitrogen and carbon dioxide. This mixture was analyzed by mass spectrometry, by absorption using an Orsat type gas analyzer, and by gravimetric analysis. The results together with the value calculated from the measured mixing pressures are shown in Table I. This region of concentration is ideal both for accurate mass spectrometric and for gravimetric analyses. Some interaction has been reported between carbon dioxide and nitrogen a t ele(4) W. A. Pennington, ANAL.CHEM., 21,766-9 (1949).
752
0
ANALYTICAL CHEMISTRY
(2)
vated pressures (5). This interaction, unless corrected for, would result in significant error in the calculation of concentration from measured pressures at this concentration of carbon dioxide. Unfortunately, the data of Haney and Bliss do not cover the range of pressures used for compounding the first mixture so only a crude approximation of the correction can be made. This approximation amounts to 1.25% and reduces the calculated value from 5.67 to 5.60 mol %. Because of the uncertainty in this value and because of the reliability of the gravimetric and mass spectrometric procedures, the concentration of this mixture is considered to be the average of the latter two methods or 5.53 0.08 mol %. Mixture No. 1 was then used to prepare a second mixture with a calculated composition of 0.629 mol %. In this concentration range, gravimetry is the method capable of the highest degree of analytical accuracy. The results are shown in Table 11. Molecular interaction has no significant effect at this concentration of carbon dioxide and the only deviation from ideality to be considered in the mixing data is the compressibility of nitrogen. The gravimetric procedure is satisfactory at this concentration but because of the greater volumes of gases to be measured and because of the increased effect of unavoidable variations in ambient temperature and pressure during a determination, it is considered inherently less accurate than values determined from the measured pressures. However, the substantial agreement between all three results indicates that no significant error was made in compounding the mixtures. The concentration of carbon dioxide is therefore assumed to be that obtained by calculation using the measured temperature and pressure of mixing. Mixtures number 3, 4, and 5, respectively, were prepared from the 0.629 mol % mixture. Their concentations, determined both by gravimetric analysis and by calculation, are shown in Table 111. These three mixtures were also analyzed by mass spectrometry as previously described and the ratio of mass 44 to mass 28 was calculated. The procedure was repeated on the succeeding day. The results obtained on both days are shown in Table IV. The values are simply the ratio of the measured peak heights in arbitrary units, with no correction applied for difference in the sensitivities of the two molecular species nor for the fact that the ion currents were measured under different amplification. The fact that the final concentration depends only on a comparison of the ratio in a mixture compared to that of a standard, requires that neither the individual sensitivity nor the amplification factor be known. A plot of the ratio of the mass 44 to mass 28 against the concentration of carbon dioxide calculated from the mixing data shows a strong straight line relationship. The linearity indicates that the ratio of the ion current at (44) to that of (28) is directly proportional to the concentration of carbon dixoide. The inlet pressures for all samples reported in Table IV varied from 0.86 mm to 1.04 mm. The variations of the measured ratios with pressure are random and show no relation to the pressure in the range of pressures measured. The line, if extended, would pass through the origin. A small but constant background current at mass 44, probably caused by hydrocarbon residues, is always present and, if a correction is not made, a straight line still results but it is displaced
*:
( 5 ) R. E. D. Haney and H. Bliss, Znd. Eng. Chem., 36, 985 (1944).
Table 11. Determination of Carbon Dioxide in Nitrogen, Mixture No. 2 Analytical method
C02in mole per cent
Gravimetry Calculated from measured pressure Mass spectrometry
0.634 i 0.005" (n = 9)b 0.629 i 0.006~ 0.63 f O.Ol"(n
=
3)b
95% confidence limit for the true average. = number of measurements. c Estimate of the systematic error involved in mixing, a
*n
Table 111. Concentration of Carbon Dioxide in Mole Per Cent Mixture Gravimetry
3 4
0.0235 i 0.0004" (n = 6)b 0.0311 f 0.0009" ( n = 5)* 0.0344 i 0.0007" (n = lo)*
5 a
Calculated from measured pressures
No.
0.0235 f 0.0002~ 0.0313 i 0.0003c 0.0346 f 0.0003~
95 % confidencelimits for the true average. = number of measurements. Estimate of the systematic error involved in mixing.
*n c
Table IV. Ratio of
(44)for Three Mixtures
-
(28)
Mixture Date 3/24
No. 3
No. 4
No.?
1.31 1.32 1.29 1.30 1.31
1.96 1.96 1.95 1.94 1.95
1.97 1.97
Average
1.306
1.76 1.75 1.77 1.76 1.77 1.74 1.76 1.77 1.760
3/25
1.31 1.32 1.32
1.77 1.77 1.76
1.317
1.767
Average
1.952
1.95
1.95 1.960
Table V. Concentration of Carbon Dioxide in Mole Per Cent Calculated from Mixture No. Gravimetry measured pressures 6 7 8
9
0.0185 i 0.0005" 0.0284 f 0.0004" 0.0312 =t0.0003" 0.0382 Z!Z 0.0003"
0.0188 i 0.0002* 0.0276 f 0.0003b 0.0332 i 0.0003b 0.0384 i 0.0004*
95% confidence limit for the average.
* Estimate of error involved in mixing.
slightly in a positive direction on the ordinate. The magnitude of the background current at mass 44 was found to be about 1% of the total ion current at concentrations of 0.03 mol carbon dioxide. APPLICATION
With three accurately analyzed standards available, such as those described above, it should be possible to compound mixtures of nitrogen and carbon dioxide in the range covered by the standards and to analyze them quickly and accurately. To verify this, four additional mixtures, numbers 6, 7, 8, and VOL 40, NO. 4, APRIL 1968
753
2 +I9
SERIES
I s
SERIES SERIES
2 S A
A
i-
i 0 -
A
I
6
3
7
0
4
5
9
MIXTURE NUMBER
9, respectively, were prepared using the same technique. The concentrations as determined both by calculation and by gravimetry are shown in Table V. A large discrepancy exists in the case of the results for sample number 8. The four mixtures were analyzed mass spectrometrically by the same procedure described for mixtures 3,4, and 5 . At the same time, mixtures 3,4, and 5 were also analyzed. When all of the values are plotted (Figure 5), it is quite obvious that an error has been made in compounding mixture number 8 and that the most nearly correct value is given by gravimetry. The value of the carbon dioxide content of mixture number 8 corresponding to the measured ratio of mass 44 to m a s 28 is 0.0308 mol % carbon dioxide, The gravimetric results for this mixture confirm this value and it is assumed to be correct and that some gross error was made in the pressure measurements. Another lesser discrepancy evident in Table V is resolved by the mass spectrometric analysis of mixture 7. The agreement between the measured ratio of mass 44 to mass 28 confirms the calculated concentration and indicates a high value for the gravimetric result.
ACCURACY The accuracy with which an unknown mixture can be analyzed by measurement of the mass 44 to mass 28 ratio and comparison with standards is, of course, dependent on both the error in the ratio measurement and the error in compounding and characterizing the standards. Consider first the accuracy with which a mixture can be characterized from a measurement of the pressures of the individual substances. The limits shown on the values “Calculated from Measured Pressures,” appearing in the tables are considered a conservative estimate of the error-Le,, 1%. In fact, it is possible to read any of the pressure
Table VI. Ratio of Mass 44 to Mass 28 Measured on Three Different Days ~ i coz ~ in- Series, standard deviation of the measurements ture mol No. 7Z 1 2 3 6 0.0188 0.990f0.006 1.000zt 0.012 0.980 zt 0.009 3 0.0235 1.221 i 0.009 1.252 f 0.008 1.208 i.0.008 1.427 zt 0.012 7 0.0276 1.451 f 0.009 8 0.0308 1.614 f 0.007 1.635 =k 0.018 1.609 i.0.010 4 0.0313 1.635 f 0.009 1.666 f 0.009 5 0.0346 1.807 i 0.020 1.841 f 0.014 1.787 f 0.008 9 0.0384 2.001 & O o . 0 0 4 2.008 f 0.016 2 . W & 0.021
754
ANALYTICAL CHEMISTRY
Figure 6. Graphical representation of results gauges used to about 1 part in 500 which is also the precision with which they have been calibrated. Temperatures of the gases can be measured to about 0.5 degree or approximately an accuracy of 1 part in 600 on the absolute scale. If care is taken to assure that the actual gas temperature and the cylinder skin temperature are the same, the error in temperature measurement should be less than that of the pressure measurement. Compressibility factors (3) are reported to at least 1 part in 10,000. The reliability of the data as indicated by the spread in the values used to compile the tables is in all cases better than 0.1%. Regardless of how the above errors of each of these variables are combined we can be reasonably sure that all systematic errors will not result in an inaccuracy of more than 1% of the calculated concentration. The accuracy with which the ratio of mass 44 to mass 28 can be measured is dependent on the accuracy with which the ion currents can be measured and also on the stability of the instrument. The peak heights can be estimated reproducibly to about 0.2%. It was found, however, that the reproducibility of successive runs was seldom as good as this. The standard deviation of the average of 10 determinations of the mass 44 to mass 28 ratio was 0.5%. The deviations are most likely the result of instrument instability although no specific cause was found for this. It was noticed that the indicated ionizing current was unstable at times. This was most often observed at the beginning of a series of measurements and is probably caused by small leakage currents associated with relatively high pressure in the ionizing region. In spite of this, rather good results are obtained. Table VI contains results obtained for three series of measurements obtained on three different days approximately two weeks apart. Each value of the ratio represents the average of 3 to 5 determinations on different samples of the gas. In only a few cases is the standard deviation as great as 1%. Figure 6 is a graphical representation of the data in Table VI. A line was fitted for each series by the method of least squares and the deviation of each point from the line was calculated in per cent carbon dioxide. The points represent the deviation with respect to total carbon dioxide and with the exception of one point lie well within f1% of the true value. The one point greater than 1% actually had a deviation less than some other points but because of the low total carbon dioxide the percentage value was greater. The use of three standards is all that is necessary to establish the slope with sufficient accuracy to allow a determination of an unknown with an accuracy of better than 1%. For some
purposes it is possible to establish the slope with sufficient accuracy using only one mixture. For example, a series of 10 determinations of a sample gave a value of 1.660 with a standard deviation of only 0.5%. Because the line must pass through the origin, the slope can thus be established with sufficient reliability for most purposes. If the concentration of the mixture used for the single calibrating point is known with the requisite accuracy, the averaging effect of three standards is not necessary.
nitrogen (mixture No. 5). The carbon dioxide concentration was found to be as follows:
OTHER APPLICATIONS
Similar accuracy could not be attained on direct analysis of the mixture even if inlet pressure were similarly high. The base peak of methane, mje = 16, would be obscured both by the tailing of the m]e = 14 because of N+ and by mle = 16 because of oxygen,
Any system which can be chemically converted to a mixture of carbon dioxide and nitrogen can be analyzed by this method with equivalent accuracy. An illustration of this is the analysis of nitrogen containing low concentration of methane and other hydrocarbons. A mixture of 0.0207 mol methane in nitrogen was prepared by dilution of a mixture of higher methane content. Enough oxygen was added to the mixture to oxidize the methane. The methane was then oxidized by passage over hot copper oxide and the resulting mixture was analyzed by measuring the mass 44 to 28 ratio. The concentration of carbon dioxide was determined by calculation after determining the slope of the ratio to concentration line using a single known concentration of carbon dioxide in
Average
0.0204 0.0206 0.0206 0.0208 0.0207 0.0206 mol % carbon dioxide
RECEIVED for review October 19, 1967. Accepted January 2, 1968. Certain commercial materials and equipment are identified in this paper in order to adequately specify the experimental procedure. In no case does such identification imply recommendation or endorsement by the National Bureau of Standards, nor does it imply that the material or equipment identified is necessarily the best available for the purpose.
Gel Filtration Behavior of Inorganic Salts P. A. Neddermeyer' and L. B. Rogers Department of Chemistry, Purdue University, Lufayette, Ind. 47907
Concentration profiles for small samples of inorganic salts eluted with water from Sephadex G-10 and 6-25 columns were badly skewed, having diffuse front and sharp back edges. The peak volumes increased with sample volume and sample concentration. These behaviors are opposite to those normally found in chromatography and result from a Donnan salt-exclusion effect that arises between the ionic solutes and a small number of fixed negative charges within the gel matrix. Neutralization of the negative sites with acid diminished the salt-exclusion. The effect was eliminated completely and elution behaviors became normal when a moderate concentration of electrolyte was present in the eluent. In spite of the Donnan saltexclusion effect, the inorganic salts eluted in the order of decreasing formula weight as expected from a gel filtration mechanism.
GEL FKTRATION CHROMATOGRAPHY has become a useful method of separating and characterizing molecules on a basis of their molecular dimensions ( I , 2). The theoretical basis for the technique, though still in its infancy, has been dealt with successfully in a number of articles (3-6). Separations in 1 Present address: Research Laboratories, Eastman Kodak Co., Rochester, N. Y. 14650
-
(1) R. L. Pecsok and D. Saunders, Sep. Sci., 1,613 (1966). (2) H. Determan, Angew. Chem. Intern. Ed. Engl., 3, 608 (1964). (3) T. C. Laurent and J. Killander, J . Chromatog., 14, 317 (1964). (4) D. M. W. Anderson and J. F. Stoddard, Anal. Chim. Acta, 34, 401 (1966). (5) G. K. Ackers, Biochem., 3, 723 (1964). (6) J. C. Moore, J . Polymer Sci., A2, 835 (1964).
gel filtration result from the preferential diffusion of small solute molecules into the porous gel structure with the exclusion of large molecules. Those solutes are characterized by the equation
The elution volume, V , , is equal to the sum of the void (interstitial) volume, Vo,and a fraction of volume, V,, which is related to the solvent volume imbibed by the gel beads. Kd is normally derived from the above equation and is similar to a distribution coefficient. It represents the fraction of the imbibed volume available for solute penetration and can take on values between zero (representing complete exclusion from the gel interior) and unity (representing complete penetration). When only the gel filtration mechanism is operative, Kd does not exceed unity, but if it does, adsorption is clearly indicated. However, when adsorption, ion exchange and other interfering processes do not cause a peak to occur beyond a total of (Vo Vt), the interference is often very difficult to discern. Gelotte (7) was one of the first to report on the secondary interactions encountered in gel filtration chromatography with the gel bed material Sephadex. He noted that aromatic and heterocyclic substances tended to adsorb on Sephadex. At low electrolyte concentrations, basic substances also adsorbed while acidic substances were excluded from the gel interiors. Wilk, Rochlitz, and Bende (8) obtained separations of a num-
+
(7) B. Gelotte, J. Chromalog., 3, 330 (1960). (8) M. Wilk, J. Rochlitz, and H. Bende, Zbid., 24, 414 (1966). VOL. 40, NO. 4, APRIL 1968
755
