Retention Model for Anionic, Neutral, and Cationic Analytes in

rewrite eq 7 as The terms within parentheses, (ads) and (exch), represent the contributions of adsorption and ion exchange, respectively. ..... Go...
1 downloads 0 Views 162KB Size
Anal. Chem. 1996, 68, 4494-4500

Retention Model for Anionic, Neutral, and Cationic Analytes in Reversed-Phase Ion Interaction Chromatography C. Sarzanini,* M. C. Bruzzoniti, G. Sacchero, and E. Mentasti

Department of Analytical Chemistry, University of Turin, Via P. Giuria 5, 10125 Turin, Italy

A model for the prediction of retention data for 1- and 2-charged analytes in ion interaction chromatography, previously developed by our group, has been extended to account for the retention behavior of cationic and neutral analytes. By using a single mathematical procedure, it is possible to develop applicative equations to describe the simultaneous effect of the concentration of ion interaction reagent, organic modifier, and counterion on the capacity factor k′ of analytes with different charges (neutral, negative, and positive). The modeling considers the retention process as a result of several adsorption and ion exchange equilibria, whose constants can be calculated by an iterative method of nonlinear regression. The applicability of the equation developed has been tested on different classes of organic (aromatic sulfonates and carboxylic acids) and inorganic analytes. The experimental design has been planned in order to describe the chromatographic behavior as a function of the mobile phase components: cetyltrimethylammonium chloride, sodium chloride, and methanol. Ion interaction chromatography1 is a chromatographic technique suitable for the separation of hydrophilic ionic solutes that show little or no retention on lipophilic stationary phases when typical reversed-phase eluents are used. The chromatographic process involved has been described by a variety of names which imply some sort of mechanism for the process itself. Therefore, in the current literature, it is common to refer to this particular chromatographic technique with alternative names2 such as dynamic ion exchange chromatography, ion pair chromatography,3 etc. The theoretical description of the mechanism involved is not an easy task, but the approach to the theory is often based on ion pair, dynamic ion exchange, or ion interaction mechanism, the last including thermodynamics and a stoichiometric basis. A comprehensive description of the advantages and drawbacks of the main models proposed for both qualitative and quantitative descriptions of analyte retention behavior in ion interaction chromatography has been discussed elsewhere.2-4 (1) Bidlingmeyer, B. A.; Deming, S. N.; Price, W. P., Jr.; Sachok, B.; Petrusek, M. J. Chromatogr. 1979, 186, 419-434. (2) Haddad, P. R.; Jackson, P. E. Ion ChromatographysPrinciples and Applications; Journal of Chromatography Library 46; Elsevier: Amsterdam, 1990; Chapter 6, pp 165-193. (3) Poole, C. F.; Poole, S. K. Chromatography Today; Elsevier Science Publishers: Amsterdam, 1991; Chapter 4. (4) Bruzzoniti, M. C.; Mentasti, E.; Sacchero, G.; Sarzanini, C. J. Chromatogr. A 1996, 728, 55-65.

4494 Analytical Chemistry, Vol. 68, No. 24, December 15, 1996

Retention and selectivity in ion interaction chromatography are influenced by several experimental variables, including type, hydrophobicity, and concentration of ion interaction reagent, ionic strength, concentration of organic modifier, and sorptive properties of the stationary phase.3 The effect of increasing hydrophobicity and concentration of ion interaction reagent (IIR) is to increase its concentration on the hydrophobic stationary phase surface and, therefore, to increase the retention of analytes.5 The ionic strength modification may affect the ion-paired complexes formation through the competition of the added ions with analytes.6 Moreover, salting-in or salting-out effects could be originated, modifying the solubility (i.e., retention) of solutes and ion-paired species.3 The organic modifier concentration affects the hydrophobic interactions between IIR, ion-paired species, and the stationary phase, so its increase leads to reduced retention of analytes.7 The separation of different charged analytes in ion interaction chromatography has been considered by several authors.1,5,8-12 Attention has been focused mainly on an eluent containing octylamine1 or, more generally, an alkylammonium cation as ion interaction reagent. The analytes include both organic and metal species.11,12 In no case has the chromatographic behavior of different charged analytes yet been described in order to predict retention changes of charged solutes when the three most significative parameters (concentration of ion interaction reagent IIR, percentage of organic modifier, and concentration of salt) are varied simultaneously. Moreover, in the current literature, in no case has the chromatographic behavior of cationic solutes been described in order to predict its retention in the presence of a pairing ion of the same charge as a function of all the components of the mobile phase. In fact, the modeling for cationic analytes generally only considers the retention behavior as a function of one parameter.13-16 (5) Bartha, A.; Billiet, H. A. H.; De Galan, L.; Vigh, G. J. Chromatogr. 1984, 291, 91-102. (6) Zhang, Y. K.; Zou, H. F.; Hong, M. F.; Lu, P. C. Chromatographia 1991, 32, 538-542. (7) Zou, H.; Zhang, Y.; Lu, P. J. Chromatogr. 1991, 545, 59-69. (8) Jandera, P.; Chura´cek, J.; Taraba, B. J. Chromatogr. 1983, 262, 121-140. (9) Bartha, A.; Vigh, G. J. Chromatogr. 1983, 265, 171-182. (10) Hansen, S. H.; Helboe, P.; Lund, U. J. Chromatogr. 1983, 270, 77-85. (11) Sacchero, G.; Abollino, O.; Porta, V.; Sarzanini, C.; Mentasti, E. Chromatographia 1991, 31, 539-543. (12) Sarzanini, C.; Sacchero, G.; Aceto, M.; Abollino, O.; Mentasti, E. J. Chromatogr. 1993, 640, 127-134. (13) Bartha, A.; Ståhlberg, J. J. Chromatogr. A 1994, 668, 255-284. (14) Bartha, A.; Vigh, G.; Ståhlberg, J. J. Chromatogr. 1990, 506, 85-96. (15) Ståhlberg, J.; Fura¨ngen, A. Chromatographia 1987, 24, 783-789. (16) Bartha, A.; Ståhlberg, J. J. Chromatogr. 1990, 535, 181-187. S0003-2700(96)00536-7 CCC: $12.00

© 1996 American Chemical Society

In a previous paper,4 we developed an equation to describe the retention mechanism in ion interaction chromatography. The species obtained by interaction of reagent Q+ with X2- were considered, where Q+ is the ion interaction reagent and X2- a double-negative charge analyte. The retention model was derived by taking into account the simultaneous effects of ion interaction reagent, organic modifier, and counterion (C-) present in the mobile phase. The retention mechanism was considered the result of the contribution of adsorption and ion exchange equilibria between pairs of ions. The aim of this work is to derive, by a single mathematical procedure, an equation to account for the retention behavior of anionic, neutral, and cationic solutes in the presence of a cationic IIR. The equation was applied to analytes of different charge and structure (inorganic, carboxylic, aromatic) with a methanol/water eluent containing cetyltrimethylammonium and chloride as ion interaction reagent and counterion, respectively. The capability of the cation Na+ to act as pairing ion in the retention of anionic analytes has been evaluated. EXPERIMENTAL SECTION Instrumentation. The chromatographic apparatus was a Varian Model 9010 pump (Varian, Walnut Creek, CA), equipped with a Rheodyne injector and a Model 332 UV-visible variable wavelength detector (Kontron Instruments, Milan, Italy). The chromatograms were recorded with an Axxiom Chromatography Model 727 data station (Axxiom Chromatography, Calabasas, CA). The chromatographic column was a Merck LiChrospher 100 CH8, 10 µm (250 mm × 40 mm i.d.). An Orion digital pH meter (Orion, Cambridge, MA) was used for pH measurements. All chromatograms were obtained at room temperature (20 °C). The flow rate was 1.0 mL/min. UV absorbance detection was performed at 210 nm. Retention times were the mean of triplicate injections. Column dead volume, determined from the unretained peak of water, was 2.8 mL for the chromatographic conditions chosen. Reagents and Solutions. Eluents were prepared by dissolving analytical reagent grade chemicals in high-purity water, obtained using a Milli-Q system (Millipore, Bedford, MA) and filtering through a 0.45 µm membrane filter (Millipore HAWP 04700), and degassed under vacuum before use. Eluent components were methanol for chromatography, sodium chloride (Merck, Darmstadt, Germany), and cetyltrimethylammonium chloride (CTACl) (Fluka, Buchs, Switzerland). Before the organic modifier was mixed the eluent pH was adjusted with ammonia to 7.5. Concentrated stock solutions of the analytes were prepared by dissolving in water/methanol (40:60 v/v): benzene, benzyldimethylhexadecylammonium chloride, naphthalene, naphthalene1-sulfonic acid sodium salt, naphthalene-1,5-disulfonic acid disodium salt, naphthalene-2-sulfonic acid sodium salt, propionic acid sodium salt, sodium nitrate, benzenesulfonic acid, benzene-1,3disulfonic acid disodium salt, oxalic acid dihydrate, and toluene4-sulfonic acid sodium salt (Fluka). Benzene, benzyldimethylhexadecylammonium, propionate, and oxalate solutions were 1.0 × 10-2 M, and all the others were 1.0 × 10-4 M. Procedures. Columns and tubings were cleaned daily with a methanol/water (80:20, v/v) solution for 30 min and with pure methanol for 10 min at a flow rate of 1.0 mL/min. The

reproducibility of the chromatographic system was verified daily through the injection of a test solution containing different structured analytes at the same eluent composition. In addition, the performance of the column was periodically checked by verifying the reproducibility of data previously obtained for different chromatographic conditions. Eluents were prepared daily with high-purity water and contained cetyltrimethylammonium chloride, sodium chloride, and methanol as required (see below). Standard solutions were obtained by diluting proper quantities of the stock solutions in a 40-60% (v/v) water/methanol mixture. Multivariable Regression. Calculations and graphic elaborations were performed using a 486 PC with Sigma-Plot for Windows (Jandell Scientific Software). The multivariable nonlinear regression analysis is based on the Marquardt-Levenberg algorithm and allows the determination of the equation parameters by iterative calculations. THEORETICAL SECTION Anionic Analytes. (a) Adsorption Equilibria. As previously discussed,4 ion pair reagent Q+, added to the mobile phase, is adsorbed as charged and neutral ion pairs (QX- and Q2X, respectively) with analyte X2- onto the stationary phase. The resulting equilibria and their constants are K1

As + [Q+] + [X2-] \ y z (QX-)

K1 )

K2

As + 2[Q+] + [X2-] y\z (Q2X-)

(QX-) (1) As[Q+][X2-]

K2 )

(Q2X) As[Q+]2[X2-] (2)

where As is the number of free adsorption sites on the lipophilic stationary phase, according to Xianren and Baeyens,17 while the parentheses and the square brackets refer to stationary and mobile phases, respectively. In absence of the ion pair reagent, the anionic analyte can be adsorbed onto the stationary phase and interact with other positively charged species of the eluent (e.g., Na+ ion). In this case, the sodium ion formally acts as a pairing ion for the analyte, and the following adsorption equilibria take place: K1A

As + [Na+] + [X2-] y\z (NaX-) K1A )

(NaX-) (3) As[Na+][X2-]

K2A

As + 2[Na+] + [X2-] \ y z (Na2X-) K2A )

(Na2X) (4) As[Na+]2[X2-]

(b) Ion Exchange Equilibria. Every adsorbed ion pair can exchange its anion with the other ones present in the mobile phase. For a general ion C-, the involved equilibria and their constants are (17) Xianren, Q.; Baeyens, W. J. Chromatogr. 1988, 456, 267-285.

Analytical Chemistry, Vol. 68, No. 24, December 15, 1996

4495

2] Kc1 ) (QC)[X (QX )[C-] (5)

Kc1

(QX-) + [C-] \ y z (QC) + [X2-]

Kc2

y z (Q2C2) + [X2-] (Q2X) + 2[C-] \ Kc2 )

(Q2C2)[X2-] (6) (Q2X)[C-]2

It has been verified by mathematical computation that the formation of NaC and Na2C2 in the stationary phase is negligible, so the relative ion exchange equilibria have not been considered. Introducing the adsorption capacity of column k0, defined as the amount of adsorbed species and free sites still available,17 eq 7 can be derived:

k0 ) As + (Q2X) + (QX ) + (NaX ) + (Na2X) + -

-

work.4 The equation describes the retention behavior of doubly charged anionic analytes in ion interaction chromatography in the presence of a cationic IIR and an anionic counterion in an organic/ water mixture. The equation evidences and includes the dominant contributions to the retention mechanism of analytes, that is, adsorption (ads) and ion exchange (exch) equilibria between pairs of ions. The retention behavior for a singly charged analyte is also described by eq 10 (K2 and Kc2 ) 0). Neutral Analytes. When the analyte has no charge, it can only interact with the stationary phase by the partition equilibrium K1A

As + [X] \ y z (X)

k0K2[Q+]2

(Q2X) [X2-]

)

(8)

1 + [X2-] (ads) + (exch)

The terms within parentheses, (ads) and (exch), represent the contributions of adsorption and ion exchange, respectively. Their expressions are

(ads) ) K2[Q+]2 + K1[Q+] + K1A[Na+] + K2A[Na+]2 + 2

-

(Q2X) + (QX-) + (NaX-) + (Na2X) 2-

[X ]

(ads) 2-

1 + [X ](ads) + (exch)

b ecφ

(12)

K1* )

(XC) (13) As[X+][C-]

The pairing ion is involved in the ion exchange equilibrium, Kc1*

(XC) + [Q+] \ y z (QC) + [X+] Kc1* ) (QC)[X+] (14) (XC)[Q ] +

(10)

where b includes the product k0Φ. The values of b and c (c < 0) are constant for a given ion pair reagent-organic modifier combination and for each solute; φ is the concentration of the organic modifier. Equation 10 is in accordance with the expression of k′ introduced by the electrostatic theory,6,15,19 as stated in a previous (18) Zou, H. F.; Zhang, Y. K.; Hong, M. F.; Lu, P. C. Chromatographia 1993, 35, 390-394. (19) Ståhlberg, J. J. Chromatogr. 1986, 356, 231-245.

4496

b ecφ 1 + K1A[X]

The capacity factor is independent of the concentrations of ion interaction reagent and counterion, as will be shown. The parameters that affect retention behavior are the analyte and organic modifier concentrations. The dependence predicted is an inverse proportionality between k′ and this parameter, in agreement with the theories developed in reversed-phase chromatography.18-21 Cationic Analytes. The adsorption of the cation in the stationary phase with a counterion of opposite charge takes place according to the following equilibrium:

(9)

(Φ is the phase ratio, Kd is the distribution coefficient). Substituting (QX-), (NaX-), and (Na2X) obtained by the previous equilibria, it is possible to write an equation as a function of (Q2X)/[X2-]. Introducing ratio as expressed by eq 8 and the contribution of organic modifier,4,7,18 k′ can be written as

k′X2- )

K1A

K1*

The capacity factor k′ for the solute ion X2- can be defined17 by eq 9:

k′ ) ΦΚd ) Φ

k′X )

y z (XC) As + [X+] + [C-] \

- 2

(exch) ) K1Kc1[Q ][C ] + K2Kc2[Q ] [C ] +

(11)

With the same mathematical procedure already shown, the following expression for the capacity factor can easily be obtained:

(QC) + (Q2C2) (7) Obtaining As, (QX-), (NaX-), (Na2X), (QC) and (Q2C2) from the equilibria written above, it is possible to rewrite eq 7 as

K1A ) A(X) s[X]

Analytical Chemistry, Vol. 68, No. 24, December 15, 1996

Following the mathematical procedure used for the anionic solutes, eq 15 is obtained:

k′X+ )

Kc1*[C-]b ecφ 1 + K1*[C-][X+] + K1*Kc1*[C-][Q+]

(15)

Among the analytes with the same charge as the ion interaction reagent, only the lipophilic ones can be retained. In this case, Q+ acts as counterion in competition with analyte X+, changing the roles of the ions (Q+ and C-) involved. RESULTS AND DISCUSSION The retention model derived has been checked for analytes with different charge and different lipophilic structure. Monova(20) Bellot, J. C.; Condoret, J. S. J. Chromatogr. 1993, 657, 305-326. (21) Guan, H.; Guiochon, G. J. Chromatogr. 1994, 687, 179-200. (22) Guan, H.; Guiochon, G. J. Chromatogr. 1994, 687, 201-212. (23) Lehmann, U.; Augenstein, M.; Neidhart, B. J. Liq. Chromatogr. 1994, 17, 3285-3306.

Table 1. Solute-Specific Constants (K1, K2, K1A, K2A, Kc1, Kc2, b, c) of the Analytes Studied analyte

K1

benzenesulfonate toluene-4-sulfonate naphthalene-1-sulfonate naphthalene-2-sulfonate nitrate propionate benzene-1,3- disulfonate naphthalene-1,5-disulfonate oxalate benzene naphthalene BDHA

9.8 × 105 1.1 × 106 3.3 × 105 1.2 × 106 1.9 × 105 4.5 × 104 2.8 × 104 5.9 × 104 3.0 × 104

a

5.7 × 103 a

K2

1.3 × 109 3.4 × 108 4.5 × 10-5

K1A 2.3 × 10-5 2.6 × 10-5 3.5 × 10-7 1.2 × 10-5 9.5 × 10-7 18 1.8 × 10-5 2.6 × 102 3.0 × 10-5 25 2.3 × 106

K2A

Kc1

4.7 × 10-4 5.0 × 10-5 4.8 × 10-4

6.7 × 10-3 5.3 × 10-3 2.2 × 10-2 6.6 × 10-3 5.0 × 10-2 0.50 33 4.0 0.44

Kc2

0.16 0.48 4.4 × 103

18b

b

c

5.7 22 7.1 × 102 1.8 × 102 5.4 4.5 2.8 × 104 2.6 × 104 10 9.7 43 13

-13 -15 -18 -17 -11 -8.5 -20 -20 -9.6 -7.6 -11 -6.4

For cationic solutes, the constant was defined as K1*. b For cationic solutes, the constant was defined as Kc1*.

lent and divalent anionic solutes, neutral species, and cationic analyte were studied (Table 1). The cationic analyte considered was benzyldimethylhexadecylammonium; in fact, the lipophilic structure of this solute offers the possibility to study its behavior in ion interaction chromatography. The experimental design has been planned in order to describe the chromatographic behavior in a multidimensional space: capacity factor (k′) versus concentration of cetyltrimethylammonium (ion interaction reagent, CTA+), chloride (counterion), and methanol (organic modifier). The concentrations of Cl-, CTA+, and CH3OH of the eluent have been varied in a relatively wide range, giving 21 retention data for benzene-1,3-disulfonate and naphthalene-1,5-disulfonate, 23 for nitrate, benzenesulfonate, toluene4-sulfonate, naphthalene-1-sulfonate, naphthalene-2-sulfonate, and oxalate, and 24 for propionate, benzyldimethylhexadecylammonium, benzene, and naphthalene. Preliminary experiments on the solubility of the analytes in the mobile phases made it possible to choose as analytical ranges of concentrations 0.43-5 mM for cetyltrimethylammonium chloride, 0-80 mM for sodium chloride, and 55-65% for methanol. Equations 10, 12, and 15 were verified by multivariable nonlinear regression. The concentrations of the chloride ion, used in the calculations, have been expressed as total chloride concentrations ([NaCl] + [CTACl]). The analyte concentrations used in the calculations were 1.0 × 10-2 M for benzene, propionate, oxalate, and benzyldimethylhexadecylammonium and 1.0 × 10-4 M for all the other analytes. These amounts, in fact, were actually injected into the separator column during the experimental work. According to previous statements on the effect of analyte concentration on k′, this parameter was kept constant in the mathematical computation.2,4 On the basis of the equations tested by means of the experimental data, adsorption constants, ion exchange constants, and parameters b and c were determined, for all the 12 analytes studied, by iterative calculations. Table 1 shows the results of this process for the solutes investigated. Among the adsorption constants of disulfonates, K1 is lower than K2 by about 4-5 orders of magnitude, in agreement with the greater interactions of uncharged ion pairs with the lipophilic stationary phase in the retention mechanism. The interactions between the analytes and Na+ as pairing ion are weak, as shown by the low values of the constants K1A and K2A compared with K1 and K2. It means that equilibria 1 and 2 are dominant compared with equilibria 3 and 4 in the retention mechanism. At all events,

the related equilibria offer an explanation for the retention of the analytes in the column when no cetyltrimethylammonium chloride is added to the mobile phase. As a consequence, it becomes possible to include chromatographic data obtained at [CTA+] ) 0 in the iterative calculations. The slight difference between the ion exchange constants, on the other hand, suggests that the charged ion pair QX- provides a more significative contribution to the counterion effect. A comparison between the adsorption constants of the neutral ion pair shows that those of the 1-charged analytes are smaller than those of 2-solutes having the same structure (compare K1 of benzenesulfonate (naphthalenesulfonate) with K2 of benzene-1,3disulfonate (naphthalene-1,5-disulfonate)). These data confirm that, for analytes with the same aromatic structure but different charge, both electrostatic and hydrophobic interactions between solute ions and ion interaction reagent in the stationary phase are important in the whole chromatographic mechanism. The low value of K1A shows that the retention of the 1-analytes is predominantly due to equilibrium 1 and not equilibrium 3, as already observed for divalent solute ions. Concerning the values for the ion exchange constants of the species studied, it should be noted that the constants obtained for 2-analytes are higher when compared with those for the 1-solutes. This means that the counterion has a greater influence on the disulfonates than on the monosulfonate ions. The values of b and c are characteristic of the solute considered, as previously stated.4 Nevertheless, it is advisable to remember that a larger absolute value of c means that there is a stronger effect on the retention of the solute due to the organic modifier percentage. Predictions can be made for the capacity factors of each analyte at different eluent compositions using the values of the constants obtained. Tables 2-4 show the individual deviations between the measured (k′meas) and predicted (calculated, k′calc) capacity factors for the analytes with benzene and naphthalene structure. The errors are given as percent values of the relative differences, according to the expression 100|k′meas - k′calc/k′meas; it is possible to see that the errors are homogeneously distributed at the different eluent conditions studied, meaning that a comparable prediction power is maintained in the analytical ranges of the variables studied, covering the usual working conditions. In table 5, the average errors for the analytes studied are shown. It can be noted that the predicted k′ values are in accordance with the ones obtained experimentally. In fact, the mean value of the errors Analytical Chemistry, Vol. 68, No. 24, December 15, 1996

4497

Table 2. Comparison of Observed and Predicted Capacity Factors for Benzenesulfonate and Benzene-1,3-disulfonatea benzenesulfonate

benzene-1,3-disulfonate

CTA+ (M)

eluent (methanol %)

Cl- (M)

k′meas

k′calc

error (%)

k′meas

k′calc

error (%)

0.0050 0.0020 0.0010 0.0004 0.0050 0.0020 0.0010 0.0005 0.0050 0.0020 0.0010 0.0010 0.0004 0.0050 0.0020 0.0004 0.0050 0.0020 0.0010 0.0004 0.0050 0.0020 0.0010 0.0005

65.0 65.0 65.0 65.0 60.0 60.0 60.0 60.0 55.0 55.0 55.0 55.0 55.0 55.0 55.0 55.0 65.0 65.0 65.0 65.0 60.0 60.0 60.0 60.0

0.0250 0.0220 0.0210 0.0204 0.0250 0.0220 0.0210 0.0205 0.0250 0.0220 0.0210 0.0810 0.0204 0.0850 0.0820 0.0804 0.0850 0.0820 0.0810 0.0804 0.0850 0.0820 0.0810 0.0805

2.29 1.36 1.25 0.54 4.39 2.75 1.71 1.29 8.64 5.25 3.43

2.31 1.44 0.86 0.42 4.46 2.78 1.67 0.92 8.61 5.36 3.22

1.0 5.9 30.9 22.1 1.5 1.0 2.7 28.1 0.3 2.2 6.0

17.92 6.79 6.04 1.54 43.64 19.36 8.79 5.82

16.11 7.50 3.91 1.68 43.92 20.44 10.67 5.34

10.1 10.5 35.2 9.1 0.6 5.6 21.4 8.3

2.29 5.29 2.43 1.07 1.29 0.96 0.57 0.29 2.50 1.64 1.00 0.75

1.59 4.65 3.50 1.37 1.24 0.94 0.65 0.37 2.40 1.81 1.26 0.78

30.4 12.1 44.2 28.2 3.2 2.7 14.6 28.9 3.8 10.3 26.4 4.6

30.32 6.64 12.11 33.96 15.46

29.08 7.75 12.75 33.83 14.94

4.1 16.7 5.3 0.4 3.4

4.25 2.21 1.07 0.43 11.86 4.93 2.46 1.25

4.55 2.01 1.04 0.47 12.41 5.48 2.84 1.46

7.1 9.2 2.7 10.1 4.6 11.2 15.4 16.4

a

Missing data are due to experimental conditions involving high k′ (analyte not eluted in 120 min).

Table 3. Comparison of Observed and Predicted Capacity Factors for Naphthalene-1-sulfonate (N-1-S) and Naphthalene-1,5-disulfonate (N-1,5-S)a N-1-S

N-1,5-S

CTA+ (M)

eluent (methanol %)

Cl- (M)

k′meas

k′calc

error (%)

k′meas

k′calc

error (%)

0.0050 0.0020 0.0010 0.0004 0.0050 0.0020 0.0010 0.0005 0.0050 0.0020 0.0010 0.0004 0.0050 0.0020 0.0010 0.0050 0.0020 0.0010 0.0004 0.0050 0.0020 0.0010 0.0005

65.0 65.0 65.0 65.0 60.0 60.0 60.0 60.0 55.0 55.0 55.0 55.0 55.0 55.0 55.0 65.0 65.0 65.0 65.0 60.0 60.0 60.0 60.0

0.0250 0.0220 0.0210 0.0204 0.0250 0.0220 0.0210 0.0205 0.0250 0.0220 0.0210 0.0204 0.0850 0.0820 0.0810 0.0850 0.0820 0.0810 0.0804 0.0850 0.0820 0.0810 0.0805

5.82 3.36 3.18 1.36 14.36 8.14 4.96 3.82 36.14 21.14 11.86 7.93 16.50 13.82 8.36 3.64 2.75 1.36 0.50 8.50 5.71 2.93 2.43

6.14 3.68 2.15 1.02 14.80 8.87 5.18 2.82 35.68 21.37 12.50 6.04 17.38 13.11 9.13 2.99 2.26 1.57 0.88 7.21 5.44 3.79 2.35

5.5 9.6 32.3 25.0 3.1 8.9 4.4 26.3 1.3 1.1 5.4 23.8 5.3 5.2 9.2 17.9 18.0 15.8 76.4 15.2 4.8 29.3 3.4

18.29 8.32 6.39 1.71 51.14 24.00 10.25 6.57

18.87 9.11 4.84 2.19 50.57 24.40 12.97 6.74

3.2 9.4 24.3 27.6 1.1 1.7 26.5 2.6

35.43 16.46 33.68 17.82 10.57 4.96 3.00 1.71 0.84 15.89 6.93 3.29 1.75

34.74 16.04 34.94 17.69 10.32 4.87 2.46 1.44 0.85 13.04 6.60 3.85 2.44

1.9 2.6 3.7 0.7 2.4 1.9 17.9 16.1 1.8 17.9 4.7 17.2 39.6

a

Missing data are due to experimental conditions involving high k′ (analyte not eluted in 120 min).

for the 12 analytes studied does not exceed 12%. This value represents a good result compared with the experimental error of about 5%. Figures 1-3 show the retention surfaces, for the analytes with benzene and naphthalene structure studied, calculated according to the equations developed in this work. Figure 1 shows the variation of k′ of benzenesulfonate and benzene-1,3-disulfonate with different NaCl and CTA+ concentrations at a constant percentage of methanol (65%), as predicted by 4498 Analytical Chemistry, Vol. 68, No. 24, December 15, 1996

eq 10. The curves for singly and doubly charged benzenesulfonate show a different and increasing slope with the increase of cetyltrimethylammonium concentrations, as predicted by the ion interaction mechanism. The k′ values of these anionic compounds decrease when the concentration of chloride, acting as a competing ion, increases. The different degree of decrease is predicted by the equations developed. Figure 2 shows the application of eqs 12 and 15 for benzene and benzyldimethylhexadecylammonium, respectively, at the same

Table 4. Comparison of Observed and Predicted Capacity Factors for Benzene and Benzyldimethylhexadecylammonium (BDHA) benzene

BDHA

CTA+ (M)

eluent (methanol %)

Cl- (M)

k′meas

k′calc

error (%)

k′meas

k′calc

error (%)

0.0050 0.0020 0.0010 0.0004 0.0050 0.0020 0.0010 0.0005 0.0050 0.0020 0.0010 0.0004 0.0050 0.0020 0.0010 0.0004 0.0050 0.0020 0.0010 0.0004 0.0050 0.0020 0.0010 0.0005

65.0 65.0 65.0 65.0 60.0 60.0 60.0 60.0 55.0 55.0 55.0 55.0 55.0 55.0 55.0 55.0 65.0 65.0 65.0 65.0 60.0 60.0 60.0 60.0

0.0250 0.0220 0.0210 0.0204 0.0250 0.0220 0.0210 0.0205 0.0250 0.0220 0.0210 0.0204 0.0850 0.0820 0.0810 0.0804 0.0850 0.0820 0.0810 0.0804 0.0850 0.0820 0.0810 0.0805

1.39 1.36 1.43 1.32 1.93 1.96 2.04 2.00 3.07 2.96 2.93 2.93 3.04 3.04 3.11 2.89 1.54 1.46 1.43 1.43 2.36 2.04 1.93 1.89

1.40 1.40 1.40 1.40 2.05 2.05 2.05 2.05 2.99 2.99 2.99 2.99 2.99 2.99 2.99 2.99 1.40 1.40 1.40 1.40 2.05 2.05 2.05 2.05

0.6 3.2 2.0 6.0 6.0 4.1 0.5 2.3 2.8 0.7 2.0 2.0 1.6 1.6 3.9 3.2 9.2 4.3 2.0 2.0 13.2 0.5 6.0 8.0

2.25 3.89 4.89 9.50 2.82 4.43 7.36 10.75 3.57 6.07 8.71 15.25 5.39 8.66 12.71 20.43 3.04 4.50 6.43 9.75 4.07 7.25 9.64 11.82

1.94 3.83 5.67 7.79 2.67 5.28 7.81 10.26 3.68 7.26 10.74 14.67 3.86 8.16 12.97 19.37 2.04 4.31 6.85 10.22 2.80 5.93 9.42 13.37

13.7 1.5 15.9 17.9 5.3 19.2 6.1 4.6 3.0 19.6 23.3 3.8 28.4 5.8 2.0 5.2 32.9 4.3 6.5 4.9 31.1 18.2 2.3 13.0

Table 5. Average Errors (Percentage) between Measured And Calculated Capacity Factors analyte

mean of error (%)

analyte

mean of error (%)

benzenesulfonate toluene-4-sulfonate naphthalene-1-sulfonate naphthalene-2-sulfonate nitrate propionate

13.5 12.3 15.1 12.7 15.4 11.9

benzene-1,3-disulfonate naphthalene-1,5-disulfonate oxalate benzene naphthalene BDHA

9.9 10.7 20.4 3.7 5.5 12.0

Figure 1. Calculated retention surfaces (eq 10) for benzenesulfonates: benzenesulfonate (BS) and benzene-1,3-disulfonate (B1,3-S). Experimental conditions: eluent, methanol/water (65:35 v/v), NH3 up to pH 7.5, NaCl and CTACl as shown.

mobile phase compositions. The retention behavior proves very different as a function of analyte charge. The k′ values of the benzyldimethylhexadecylammonium cation decrease when the concentration of the ion interaction reagent increases, according to eq 14. For the same reason, the k′ values of benzene, which

Figure 2. Calculated retention surfaces for benzene (B, eq 12) and benzyldimethylhexadecylammonium (BDHA, eq 15). Experimental conditions as in Figure 1.

has no electrostatic interaction, prove unaffected by ion interaction reagent concentration (eq 12). Moreover, the k′ values of benzene are not affected by sodium chloride concentration, due to the uncharged nature of the analyte, while ionic strength does not play a significant role in the liquid-liquid partition. On the other hand, the k′ values of the hydrophobic cation species increase Analytical Chemistry, Vol. 68, No. 24, December 15, 1996

4499

Figure 3. sulfonates: disulfonate water, NH3 shown.

Calculated retention surfaces (eq 10) for naphthalenenaphthalene-1-sulfonate (N-1-S) and naphthalene-1,5(N-1,5-S). Experimental conditions: eluent, methanol/ up to pH 7.5, 80 mM NaCl, CTACl and methanol as

with the concentration of chloride ions. In this case, the chloride ion acts as a pairing ion (eq 13). Figure 3 shows the application of the retention model as a function of methanol percentage and [CTA+] at a constant sodium chloride concentration (80 mM) for 1- and 2-naphthalenesulfonates. As predicted by the k′ expression, greater concentrations of methanol enhance eluent strength, thus leading to lower capacity factors. CONCLUSIONS The validity of the model developed, for both prediction and interpretation of retention behavior for different structures and differently charged solutes (0, 1+, 1-, and 2-), is also supported by the good agreement between measured and calculated capacity factors for the analytes studied (Figure 4). The slope of the log k′calc vs log k′meas plot is 0.982. The correlation coefficient calculated for the 276 data pairs is 0.979. This study confirms the applicability of the theoretical retention model developed for data prediction in ion interaction chromatography as a function of mobile phase components (ion pair reagent, counterion, and organic modifier concentrations). The parameters investigated in this study are the most significative factors affecting the retention behavior in ion interaction chro-

4500

Analytical Chemistry, Vol. 68, No. 24, December 15, 1996

Figure 4. Relationship between measured and calculated capacity factors for the 12 analytes studied, corresponding to 276 data pairs. b, Benzene-1,3-disulfonate; 9, naphthalene-1,5-disulfonate; 2, oxalate; 1, benzyldimethylhexadecylammonium; [, benzene; >, naphthalene; O, benzenesulfonate; 0, toluene-4-sulfonate; 4, naphthalene1-sulfonate; 3, naphthalene-2-sulfonate; ], propionate; ", nitrate.

matography. The effect of other parameters, such as pH, can be considered separately. In fact, pH affects the molar fraction of weakly ionized compounds (both analytes and buffers), changing their actual concentrations without changing the basic structure of the modeling. ACKNOWLEDGMENT Financial support from Ministero dell’Universita` e della Ricerca Scientifica e Tecnologica (MURST, Rome) and from the Italian National Research Council (CNR, Rome), is gratefully acknowledged. Received for review June 3, 1996. Accepted October 2, 1996.X AC960536W X

Abstract published in Advance ACS Abstracts, November 15, 1996.