terfering effect of Bi cannot be eliminated in such a way but according to our experiences the interfering effect of lOO/pg of Bi may be neglected for 5/pg of T1 (Table 111). Precision of the Method. Table 111 shows the accuracy of our method both for 1-10 ppm and for 0.1-1.0 ppm ranges. Even in the lower range, the relative standard deviation doesn’t exceed 20% which is general at the determination of such a small amount of impurities.
ACKNOWLEDGMENT
The authors are indebted to Prof. Dr. T. Millner and Dr. L. Bartha for encouraging and helpful advice in the course of this work. RECEIVED for review April 27, 1970. Accepted August 10, 1970.
ElectroI yte SoIutions J. V. Leyendekkers and Michael Whitfield Dicision of Fisheries and Oceanography, CSIRO,Cronuffa,Sydney, 2230, Australia Two liquid ion exchange electrodes (Orion Calcium 9220 and Chloride 92-17) were used to monitor changes in the activity coefficient of calcium chloride in the two aqueous systems CaCI2-MgCl2 and CaCI2-SrCl2 over the ionic strength range 0.1-6 molal at 25 OC. On the basis of comparison with isopiestic data, it is considered that useful estimates of Harned’s coefficients can be made with these electrodes provided the selectivity characteristics are not too unfavorable and depending on the complexity of the ionic interactions. Experimental selectivity isotherms are presented for both systems. Parameters are derived, on the basis of simple ion exchange theory and regular solution theory, which facilitate interpolation over the experimental range. A THOROUGH AND VERY USEFUL review of ion-selective electrodes has recently become available (I). Among the important points discussed are two which are relevant here. The first concerns the need for more thermodynamic studies made on a basis of comparison with other, unrelated, methods. For example, how useful are electrodes of the liquid-ion exchange type for measuring activities in mixed electrolyte solutions as compared with isopiestic studies? The second point concerns electrode selectivity. This is of general interest, as many natural systems contain appreciable concentrations of different counterions. Even in the analysis of single electrolyte solutions, complexing reagents and buffers can introduce significant levels of interfering ions. In short, the behavior of the electrode in a mixed electrolyte solution is often the main concern. Since the electrode measures activities, data on the activity coefficients of the electrolytes in mixed solutions will be needed. Data for around 30 twoelectrolyte systems are available ( 2 , 3), and a number of theories and empirical relationships (2-6) enable reasonable (1) “Ion-Selective Electrodes,” Richard A. Durst, Ed., Nat. Bur. Stand. (US.) Spec. Pubi., 314, 474 pp (1969). (2) H. S. Harned and R. A. Robinson, “Multicomponent Elec-
trolyte Solutions,” in The International Encyclopedia of Physical Chemistry and Chemical Physics, Topic 15, Vol. 2, Pergamon Press, London, 1968. (3) R. M. Rush, Oak Ridge National Laboratory Report ORNG 4402, UC-4-Chemistry, 1969. (4) G. Scatchard, J . Amer. Chem. Soc., 90, 3124 (1968). ( 5 ) J. Leyendekkers, J. Phys. Chem., 74, 2225 (1970). (6) Y . C . Wu, R. M. Rush, and G. Scatchard, ibid., 13, 2047 (1969). 322
ANALYTICAL CHEMISTRY, VOL. 43, NO. 3, MARCH 1971
estimates of activity coefficients in multicomponent systems to be made; however, many more experimental data are needed. In this paper, results are given of measurements of activity coefficients of calcium chloride in the presence of strontium(I1) or magnesium(I1) over the ionic-strength range 0.1-6 molal. Isopiestic data are available for these two systems and these are used as a guide to the usefulness of the electrodes for such measurements. The selectivity characteristics of the Orion calcium activity electrode in these systems are also given. This electrode has an acidic organophosphorus exchanger, and from the considerable amount of study on this type of extractant over the past few years (7), it has been shown that in general the cation exchange reactions are more complicated than those of exchange resins. However, even though the actual mechanism of the reaction is unknown, the distribution can always be described in terms of a convenient chemical reaction through which meaningful and useful information can be obtained concerning the system studied, We adopted this attitude in a previous paper (8) where, by means of a simple theory, the composition of an exchange site was derived. This enabled an estimate to be made of the parameters A and B in the equation log (amcs*+/an.wz+)[(l - y)/yl = A - B(2 Y - 1) (1) where a represents the activity of the ion in the aqueous phase, M the interfering counterion (of charge z ) , y the mole fraction of exchange sites occupied by calcium(II), and rn and n are integers related to the exchange reaction [A and B were formerly represented by log K, and B’, respectively (S)]. This facilitated interpolation so that the ionic strength range 0-6 molal was completely covered. The same procedure is adopted here, with some limited extrapolation to extend the range to an ionic strength of 7.5 molal. EXPERIMENTAL
The same equipment and technique were used as described previously (8). The additional reagents used were AR grade (7) Y . Marcus and A. S. Kertes, “Ion Exchange and Solvent Extraction of Metal Complexes,” Wiley-Interscience, London,
1969.
(8) Michael Whitfield and J. V. Leyendekkers, ANAL.CHEM., 42, 444 (1970).
01
1
,
I
1
0.2
0 3
0 4
os
CMClzI
Figure 1. Titration curves for ionic strength 3 molal
2 = 3 IogY*:L'ac12f const -logy, slope = -9a12for linear section (Equations 2 and 3) 2 is dimensionless, 0.01 unit represents a change in AEof -2 mV Upper curve: CaCI2-MgCl2 Lower curve: CaC12-SrC12
strontium chloride (SrC12 6H20) and magnesium metal which was neutralized with hydrochloric acid. Stock solutions were prepared and standardized by titrating with DCTA, using calcium chloride (prepared from Mallinckrodt calcium carbonate) as the primary standard and the Orion divalent electrode (92-32) to monitor the end point. The stock solutions were diluted on a weight basis. Temperature was maintained at 25.0 0.1 "C and pH at 7.5-8. 3
*
RESULTS AND DISCUSSION Activity Coefficients. The analysis of the results was carried out as described previously (8). The Harned slopes (a1*) were determined on the basis of the relationships
+
AEIS - log [Cal[C112 = 3 log y+csc12
(2)
where AE is the difference in potential between the calcium and chloride electrodes, the square brackets denote concentrations, S represents 2.303RT/2F and has the value 29.58 mV at 25", y is defined above, and
-
~
Values of a I 2Estimated from Activity Measurements (Reciprocal ionic strength units) Ionic strength This work Mean Isopiestic CaC12-MgC12 0 . 1 -0.047 -0.038 -0,029 0.3 -0.017 -0,022 -0.028 0.7 -0,012 -0.010 -0.009 -0.0096 (9) 3 -0.0107 -0.0100 -0.0093 -0.0111 6 -0,0117 -0.0114 -0.0110 CaC12-SrC12 0.06 0.028 0.022 0.017
constant - l o g y
log y i ~ , L = ~ 1constant 2
Table I.
~
S
C
I
,
0.3
0.003
0.0035
0.0055 (10)
0.004
3
0.0060
0.0054
0.0055
6
0.0047 0.0047
0.0044
0.0061
0,0040
(3)
where ZXICI,is the ionic strength of either strontium or magnesium chloride. Provided y is close to 1.00 and Harned's rule is obeyed, the lefthand side of Equation 2 us. [MCh] should give EL linear plot of slope -9a12. Figure 1 shows representative plots for the two systems. Linearity holds only for 10% (strontium chloride) to 20% (magnesium chloride) of the curve, after which the term (log y ) becomes significant. This represents a maximum change in A E of only 1-3 mV as the ion interactions in the alkaline earth systems are small. Nevertheless, it was possible
to estimate C Y H from a least-squares fit of the points along the linear section (Table I). On the basis of comparison with the isopiestic data (9, IO) the results at the higher concentrations are surprisingly good in view of the severe limitations imposed by the nonspecificity of the electrode for calcium. The assump(9) R. A. Robinson and V. E. Bower, J. Res. Nat. Bur. Statid., A , 70, 305 (1966). (10) Souheng Wu, University of Kansas, Ph.D. Thesis, 66-6062 University Microfilms, Inc., Ann Arbor, Michigan, 1965. ANALYTICAL CHEMISTRY, VOL. 43, NO. 3, MARCH 1971
323
i
2of/ IO Figure 2. Selectivity isotherms for the Orion calcium activity electrode at constant ionic Symbols strength at 25 'C and pH -7.5. represent experimental points ( A ) Aqueous CaCIQ-MgC12; 0 0.1 molal; 0 0.7 molal; A 3 molal; A 6 molal ( B ) Aqueous CaClrS,CI2; X 0.3 molal; 0 3
molal; 0 6 molal. - curves derived from Equation 1 and Table I1
1
I
1
I
4 0
50
bO
10
I
0
2 0
10
30
"lo I.S.
,
80
1
90
I 100
CaCiz
A I 00
90
-
a0
-
70
-
bO
-
I
'"1 I///'
-'
40
I\/ I /
/
I
/
~~
~
4 0
50 "/o
1s
b 0
70
80
90
100
cac I~
B tion that the electrode behavior is Nernstian [ref. (8)andimplied in our use of Equation 21 seems justified, even though deviations could be expected at these concentrations (11). The low values of cyl2 for the system with strontium chloride at 6 molal may be due to such deviations. As far as we know, there are no comparative data at the lower concentrations so that no firm conclusions can be made regarding the accuracy of these data. (11) Rima Huston and J. N. Butler, ANAL.CHEM., 41, 200 (1969). 324
ANALYTICAL CHEMISTRY, VOL. 43, NO. 3, MARCH 1971
The overall results suggest that the electrodes should be useful for estimating changes in activity coefficients in twoelectrolyte systems provided the selectivity does not exceed that shown here. Results for the NaC1-CaCh system at an ionic strength of 3.15~1,using the Orion (98-20-02) exchanger, showed that the selectivity for sodium was too high at this ionic strength for an accurate estimate of cy21 to be made (12). (12) "Ion-Selective Electrodes," Richard A. Durst, Ed., Nat. Bur. Stand. Spec. Publ., 314, 184 (1969).
1
I
9 0
1 m
I t
I
0
0
0
I
1
OllV'Y
9 0
-1
0
0
Allh11>313S
Even so, a tangent drawn to the first part of the curve gives a value of around -0.027 compared with -0.0165 for the isopiestic method, which shows that at least a reasonable estimate can be made in a case as unfavorable as this. However, it is difficult to generalize, since apart from selectivity, the magnitude and complexity of the ion interactions will influence the results. For instance, if Harned's rule is not
obeyed, this would greatly complicate the analysis. On the other hand, if the alpha coefficients are relatively large and Harned's rule is obeyed, the accuracy should be higher. Selectivity. The values of y were estimated from the deviations of the smoothed curve from the Harned slope (Figure 1). The experimental isotherms could then be constructed (Figure 2) and the selectivity ratio K' (Figure 3) ANALYTICAL CHEMISTRY, VOL. 43, NO. 3, MARCH 1971
325
1210-
08
-
Ob
-
0 4 02
-
0-
R-Ob
-
-
-08-I 0 -I.?
-
-I 4 -
-I b -
Figure 4. Linear fit of the data on the basis of three divalent ions occupying an exchange site, rn = n = 3 in Equation 1
R represents 3 log
+
-14
-
1
-01
I
-02
I
I
1
I
I
I
1
-01
0
01
01
03
0 4
OS
-
Ob
07
I
,
OB
09
10
2 y -I
A
(aca2+/-
axZ+) log(1 - y)/y - 3 1% (YCa+/YMZ+) ( A ) Mz+represents Mgzt
( B ) MZt represents Sr*+
2r-1
B
calculated from the relationship
at a given ionic strength. The activities were used in all other cases (e.g., Equation 4 and Table 11). Values of the mean molal activity coefficients for the single electrolyte solutions were obtained from references (13) and (14). The
O n the basis of regular solution theory and simple ion exchange concepts (8), the results indicate that three divalent ions (M2+) occupy a n exchange site so that rn = n = 3 in (13) Roger Parsons, “Handbook of Electrochemical Constants,” Equation 1. The fit is illustrated in Figure 4, where the Butterworths, London, 1959. concentrations have been used for simplicity. This will not (14) H. S. Harned and B. B. Owen, “The Physical Chemistry of affect the closeness of the fit very much since log yCa*+/Yhf*+ Electrolytic Solutions,” 3rd ed., ACS Monograph 137, Reinhold (equivalent to log Y * . C ~ C I ? / ~ + N C I ~ is ) approximately constant Publishing Corp., New York, N. Y . , 1958. 326
ANALYTICAL CHEMISTRY, VOL. 43, NO. 3, MARCH 1971
Table 11.
Ionic strength 0.1 0.3 0.7 1 .cP
The experimental points fit the predicted curves reasonably well, although not over the whole composition range. Deviations are greatest for the low ionic strengths where the smoothed and experimental values of the parameters A and B differed by 2 - 3 x . The change of selectivity with solution composition is similar to that found for the calcium chloridesodium chloride systems (8) and, as expected from the manufacturer’s handbook, the interference from the alkaline earth cations (listed as K’gr = 0.017, K ’ u ~= 0.014) is greater at lower concentrations than that of the sodium ion (K”& = O.OOl), the magnesium effect being less than that of strontium. The effect of ionic strength is relatively much smaller, however, so that sodium interference exceeds that of the alkaline earths at higher concentrations. In the simple chloride systems studied so far it does not appear that the anion plays a major role in the ion exchange process; however, other anions (e.g., the nitrate ion) may be extracted with the cation at the higher concentrations so that mixed complexes are formed in the organic phase. In extreme cases, the extraction may occur viu a solvation mechanism characteristic of non-ionic phosphorylated extractants (7). It is not known, therefore, whether the selectivity data given here will apply generally to alkaline earth salts at high ionic strengths.
Selectivity Parameters of Equation 1 (with m = n = 3) CaC12-MgC12 CaClrSrClz -A
-B
-A
-B
5.60
4.50
3.5P 3.52
4.206 4.02 3.46 3.02 2.66 2.42 2.25 2.20
3 . w
3.40” 3.50 3.60” 3.66 3.76 3.66 3.44 3.20 2.80 2.10
4.99 4.78 2.0“ 4.19 3. O 3.71 4.0” 3.24 5.0” 2.78 6.0 2.37 7.0” 2.09 a Interpolated values. * Smoothed value.
3.47 3.36 3.14 2.66 2.10 1.24
0.08
Harned coefficients of references (9) and (10) were used as far as possible, otherwise our experimental values or estimates (interpolations) were used. The variations of A and B with ionic strength were obtained by drawing smooth curves through the experimental data in Table 11. Values of these parameters at other ionic strengths could then be estimated and the corresponding isotherms and selectivity curves constructed (Figures 2 and 3).
RECEIVED for review September 14, 1970. Accepted December 7,1970.
Effects of Structure of Peptide Stationary Phases on Gas Chromatographic Separations of Amino Acid Enantiomers J. A. Corbin,’ J. E. Rhoad,* and L. B. Rogers Department of Chemistry, Purdue University, Lafayette, Ind. 47907
The gas chromatographic behavior is described for a series of systematically-substituted, optically-active, peptide derivatives used as stationary phases. Results with pairs of phases containing one racemic center indicated that the major part of the separation occurred at the amide end of the dipeptide. Nevertheless, CY values were sensitive to changes in the structure of the side group at the ester end. A steady decrease in CY occurred as the bulkiness of the peptide side groups decreased. A tripeptide derivative produced CY values nearly as large as the corresponding dipeptide and showed a small, but significant, increase in chromatographic temperature stability. However, underivatized solid peptides used as stationary phases did not show enantiomer separations. An increase in the bulkiness of the solute ester group produced an increase in CY on all phases studied. An increase in bulkiness at the solute CY carbon atom produced a consistent, but not straightforward, effect. THE COMPOUND N-TFA(trifluoroacety1)-L-valyl-L-valine cyclohexyl ester (vv) has been shown to be an excellent stationary phase for the separation of the enantiomers of a wide variety Present address, Celanese Fibers Co., Charlotte, N. C. 28201
* Present address, Department of Chemistry, Vanderbilt University Nashville, Tenn. 37203
of amino-acid derivatives (1-3). Recently, a second dipepcyclohexyl ester, tide phase, N-TFA-~-phenylalanine-t-leucine was synthesized which also gave good enantiomer separations and could be used at higher operating temperatures (4). The particular ease of separability associated with the dipeptide phase was attributed to the possibility of formation of three hydrogen bonds in a diastereomeric bridged association complex between the solute and solvent. While there have been detailed studies of solute behavior on the above stationary phases, there has been no study of systematic changes in the peptide stationary phase itself. The main purpose of this study was, therefore, to investigate those factors which were most important in leading to a n enantiomer separation with peptide phases so that, hopefully, new and better stationary phases, suitable for specific applications, could be prepared. First, a study was done to (1) E. Gil-Av, B. Feibush, and R. Charles-Sigler in “Gas Chromatography 1966,” A. B. Littlewood, M., The Institute of Petroleum, London, England, 1967, p 227.
(2) E. Gil-Av and B. Feibush, Tetruhedrorr Lett., 35, 3345 (1967). (3) S.Nakaparksin,P. Birrell, E. Gil-Av,and J. Oro, J . Clironintogr. Sci.. 8, 177 (1970). (4) W. A. Koenig, W. Parr. H. A. Lichtenstein, E. Bayer. and J. Oro, ibid., p 183. ANALYTICAL CHEMISTRY, VOL. 43, NO. 3, MARCH 1971
327
