- Email: [email protected]
Polarity parameters of the Symmetry C18 and Chromolith Performance RP-18 monolithic chromatographic columns
Journal of Chromatography A, 1107 (2006) 96–103
Polarity parameters of the Symmetry C18 and Chromolith Performance RP-18 monolithic chromatographic columns Pere Izquierdo, Mart´ı Ros´es, Elisabeth Bosch ∗ Departament de Qu´ımica Anal´ıtica, Universitat de Barcelona, Diagonal 647, 08028 Barcelona, Spain Received 17 August 2005; received in revised form 18 November 2005; accepted 5 December 2005 Available online 27 December 2005
Abstract A set of 12 compounds of different chemical nature has been established to characterise RPLC columns on the basis of a polarity retention model previously developed: log k = (log k)0 + p(PmN − PsN ). This model allows the calculation of the retention factor (k) of any non-ionized compound using one parameter which describes the polarity of the solute (p), another one for the polarity of the mobile phase (PmN ) and two more parameters for the characterisation of the stationary phase ((log k)0 and PsN ). The selected set of compounds allows the determination of (log k)0 and PsN of stationary phases and it has been used to characterise two commercial columns (Symmetry C18 from Waters and Chromolith Performance RP-18 monolithic from Merck). Column parameters, together with those of the mobile phase permit successful transfer of retention data between chromatographic systems. Prediction of retention of a variety of non-ionized analytes has been also successfully achieved using the column descriptors and p values of solutes from a previously established p data base. © 2005 Elsevier B.V. All rights reserved. Keywords: Column polarity; Retention prediction; Polarity model; Column characterisation
1. Introduction In RPLC, an accurate prediction of the analyte retention factor before the experimental analysis allows a significant simplification of optimization procedures for the experimental conditions. Several models have been proposed to explain the chromatographic behaviour of the solutes in RPLC. Some of these approaches are based only on solute-mobile phase interactions, where descriptors such as the organic modifier volume fraction in the mobile phase (ϕ) [1] or mobile phase polarity expressed through Dimroth and Reichardt parameter (ETN ) [2,3] were related with the logarithm of the retention factor. Although acceptable predictions could be achieved using these models, accurate predictions could be done only in a specific range of mobile phase compositions. Further, it is quite obvious that chromatographic retention is too complex to be explained considering solely solute-mobile phase interactions. More complete models try to explain the chromatographic behaviour using mobile phase-solute-stationary phase interac-
tions. One approach was introduced by Abraham and coworkers [4–7] and extended by Carr and coworkers [8–10]. They developed a solvation equation, based in a linear free energy relationship, which accounts for the interactions of the solute with the mobile and stationary phases, due to hydrophobicity, polarity, polarizability, hydrogen-bond acidity and hydrogen-bond basicity. Snyder and coworkers [11–14] proposed an empirical linear-free-energy equation of form similar to Abraham’s equation to characterise the column selectivity in terms of five retention-related parameters that accounts for column hydrophobicity, steric selectivity, hydrogen-bond acidity, hydrogen-bond basicity and cation exchange behaviour. Our model (Eq. (1)) was directly derived from the Abraham–Carr equations [15,16] and predicts the retention factor of any non-ionized solute from polarity parameters of the N ) and stationary ((log k) and P N ) phases and the mobile (Pm 0 s polarity of the chromatographed solute (p) [16–21], according to N log k = (log k)0 + p(Pm − PsN )
(1)
or ∗
Corresponding author. Tel.: +34 934021284; fax: +34 934021233. E-mail address: [email protected] (E. Bosch).
0021-9673/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.chroma.2005.12.019
N log k = q + pPm
(2)
P. Izquierdo et al. / J. Chromatogr. A 1107 (2006) 96–103
2. Experimental
where q=
97
(log k)0 − pPsN
(3)
(log k)0 and PsN are related to the stationary phase, although they also depend on the organic modifier used in the mobile phase. PsN includes hydrophobicity, dipolarity and hydrogen-bond ability of the stationary phase and it is a generic retentivity descriptor of the column. (log k)0 is the intercept of the correlation and it accounts for the retention in the hypothetical case where mobile and stationary phases show the same polarity. It mainly depends N is a polarity parameter easily calculaon the phase ratio. Pm ble from the volume fraction of organic modifier in the mobile phase (ϕ) by means of Eqs. (4) and (5) for acetonitrile–water and methanol–water mobile phases, respectively [16,18,20,21]. N Pm = 1.00 −
2.13ϕ 1 + 1.42ϕ
(4)
N = 1.00 − Pm
1.33ϕ 1 + 0.47ϕ
(5)
The polarity of the solute, p, mainly depends on the compound, but the organic modifier used in the mobile phase and the stationary phase itself produce small changes of its values as well. A wide database, which contains p values for more than 250 compounds, and referred to a Waters Spherisorb ODS2 (100 mm × 5 mm) column and acetonitrile–water mixtures as mobile phases, was built [18,19]. The database p value for a solute is linearly related to p of the same solute in any different chromatographic system (column and/or mobile phase) according to psystem,2 = a psystem,1 + b
(6)
where psystem,1 is the p value in the database of any solute and psystem,2 is the p value of the same solute in any other chromatographic system. Further, a and b are constants that depend only on the related chromatographic systems. Then, a proper column characterisation allows accurate prediction of retention and retention data transfer between columns and mobile phases, as has been widely demonstrated [16,18]. The aim of this paper is to propose a fast, simple and reliable procedure to characterise reversed-phase chromatographic columns, that is, to obtain (log k)0 and PsN parameters of a column for acetonitrile/water and methanol/water mobile phases. Thus, on the basis of the p database, a representative characterisation set of 12 compounds has been selected. The column parameters for the own chromatographic system used to build the database have been recalculated using the selected set. Results agree with those derived from the retention of more than 150 compounds and 600 experimental retention measurements. Therefore, the usefulness of the selected set has been validated in this way, and the selected compounds have been used for the characterisation of two commercial C18 columns with different stationary phases, one particulate and the other one monolithic. Good results for both retention prediction and retention data transfer between chromatographic systems, have been achieved using the derived parameters for both columns and acetonitrile and methanol as mobile phases organic modifiers.
2.1. Apparatus pH measurements were taken with a combined Ross electrode Orion 8102 (glass electrode with a 3.0 mol L−1 KCl solutions as a salt bridge) in a Crison micropH 2002 potentiometer with a precision of ±0.1 mV (±0.002 pH units). A Shimadzhu (Kyoto, Japan) HPLC system equipped with two LC-10ADvp dual pumps, a SIL-10ADvp autoinjector fixed to 10 l, a SPD-10AVvp UV detector, a CTO-10ASvp column oven at 25 ± 0.1 ◦ C and a SCL-10Avp system controller was employed. The detection wavelengths were 200 nm for KBr, 254 nm for benzene derivatives and 282 nm for phenols. 2.2. Columns Retention data were measured on a Waters Symmetry C18, 5 m, 150 mm × 4.6 mm column and a Merck Chromolith Performance RP-18 monolithic 100 mm × 4.6 mm column. Characteristics of the studied columns are given in Table 1. 2.3. Chemicals Methanol and acetonitrile were HPLC grade from Merck and water was purified by the Milli-Q plus system from Millipore. Other chemicals were reagent grade or better and were from Aldrich, Carlo-Erba, Flucka, Merck, Normapur or Sigma. Solutions of 100 g mL−1 of each substance were prepared in methanol. 2.4. Procedure Each compound was chromatographed on both columns at 25 ◦ C, using a variety of methanol–water and acetonitrile–water mobile phases (10, 20, 30, 40, 50, 60, 70, and 80% in volume of organic modifier). Most mobile phases were prepared with aqueous 0.1 M acetic acid (pH 2.7) and the organic modiTable 1 Chromatographic features of the chromatographic columns studied Features
Chromolith Performance
Symmetry C18
Silica type Structure type Particle size (m) Total porosity (%) ˚ Median pore diameter (A) Macropore size (m) ˚ Mesopore size (A) Surface area (m2 g−1 ) Pore volume (ml g−1 ) Total carbon Endcapped Column length (mm) Column diameter (mm) pH stability range Temperature stability (◦ C) Pressure stability (bar) Flow rate (ml min−1 )
High purity Monolithic – >80 – 2 130 300 1 18 Yes 100 4.6 2–7.5 max. 45 max. 200 2
High purity Particulate 5 65 86 – – 346 – 19.57 Yes 150 4.6 2–8 – – 1
98
P. Izquierdo et al. / J. Chromatogr. A 1107 (2006) 96–103
fier, although mobile phases based in aqueous phosphate buffer (pH 7.3) were employed for aniline derivatives. A solution of 100 g mL−1 of KBr was used as void volume marker. The flow rate was 2 mL min−1 on the Chromolith Performance column and 1 mL min−1 on the Symmetry C18 column. A 10 L aliquot of each sample was injected. All the measurements were taken in triplicate. 2.5. Data treatment Starting values were obtained from retention data of a group of compounds which was chromatographed on each column using acetonitrile–water and methanol–water with different contents of organic modifier as mobile phases. N , which is According to Eq. (2), log k is linearly related to Pm easily calculable from Eqs. (4) or (5). Then, Eq. (2) allowed the computation of q and p parameters for each solute and organic modifier. Paired values, q and p, of the totality of compounds were fitted to Eq. (3), which provided a first approach to the column parameters, (log k)0 and PsN . To refine the (log k)0 and PsN values and also the descriptor p of each compound, an iterative process was carried out by comparison of predicted and experimental log k values. Initial (log k)0 and PsN were those calculated by Eq. (3) and initial p for each compound were those computed by Eq. (2). In a particular step, predicted log k values are those calculated by Eq. (1) from the
Table 2 Characterisation set and their p values from the database [8] Substances
p (ACN–water)
p (MeOH–Water)
1,2-Dihydroxybenzene Benzamide 3-Methylphenol Propiophenone Methyl benzoate 4-Nitrotoluene Butyrophenone Naphthalene Propylbenzene Heptanophenone Butylbenzene Chrysene
1.79 1.69 2.82 3.70 3.59 3.95 4.19 4.86 5.56 5.78 6.13 7.33
1.98 2.10 2.98 3.88 3.89 4.08 4.36 5.12 5.71 6.06 6.32 8.46
refined parameters of the preceding step. Compounds with a p residual value higher than three times the standard deviation were withdrawn in the following step. Progressively, the (log k)0 and PsN of the column and p for each solute were improved together and the new values were substituted in Eq. (1) to continue with the iterative process. Step by step, in subsequent iterations the sum of residual squares value (δ2 ) decreased until it reached a minimum and those polarity parameters in the minimum were accepted as the optimal. To implement the iterative process, the Solver tool of Microsoft Excel was used.
Fig. 1. Prediction of retention factors of 152 compounds on Spherisorb ODS-2 column. Column parameters determined through the 12 compounds set and (a) acetonitrile–water and (b) methanol–water mobile phases. Column parameters determined through the 152 compounds set and (c) acetonitrile–water and (d) methanol–water mobile phases.
P. Izquierdo et al. / J. Chromatogr. A 1107 (2006) 96–103
3. Results and discussion 3.1. Proposal of a characterisation set of compounds As it was previously demonstrated, when the chromatographic system characterisation parameters were obtained using a large number of compounds, the polarity model, Eq. (1), provided very good retention predictions [16–18]. Thus, this procedure guarantees accurate values for both column ((log k)0 and PsN ) and compound (p) parameters. On the basis of literature data [22–31] and our own measurements [6,15,16], we reported a wide p database, for more than 250 compounds. This database is related to a specific column, a Waters Spherisorb ODS-2, and acetonitrile and methanol as the organic modifiers of the mobile phase. These chromatographic systems were used by Smith and Burr [22–27] to determine retention factors of compounds with different chemical structure and the reported data were the source for a preliminary p database of 152 compounds. Some compounds of this data set were also chromatographed by other authors [6,15,16,28–31] on a variety of stationary phases, that allowed us to get their p values related to each chromatographic system and, also, to correlate them with p values of the original database. On the basis of these correlations the p database was extended to more than 250 compounds [18,19]. The Spherisorb ODS-2 column was characterised according to Eq. (1), using the 152 compounds chromatographed by Smith and Burr. The column parameters (log k)0 and PsN , after
99
a refining process, were (−1.04, −0.03) for acetonitrile–water and (−1.26, −0.08) for methanol–water mixtures, respectively, and they allowed to predict successfully the retention of many other compounds on this column [18]. However, in practice, it would be desirable to get column descriptors of similar quality by means of a faster characterisation procedure, which involves a smaller number of experiments. This means the proposal and validation of a small set of representative compounds, able to carry out the characterisation process and to get properly the parameters of any column with any mobile phase. Then, a complete statistical study was made with groups of 10 compounds randomly selected among the 152 compounds belonging to the initial database and satisfactory results were obtained with most of them. Thus, derived values for (log k)0 and PsN were (−1.01, −0.03) for acetonitrile–water and (−1.17, −0.06) for methanol–water mixtures, respectively [18]. In this work, a set of 12 compounds has been selected from the whole database, composed by more than 250 compounds. This group includes solutes of diverse polarity, homogenously distributed through all the available p range, and contains compounds that show different functional groups. It includes a significant fraction of highly hydrophobic compounds (p > 5) to ensure correct predictions for this kind of solutes, and this is a major difference with respect to the previous studied characterisation sets mentioned above. Further, these compounds are commonly used in laboratory practice. The selected compounds and their p database values for acetonitrile–water and methanol–water mobile phases are presented in Table 2. In order to check the
Table 3 Initial solute parameters obtained by Eq. (2) Compound
Acetonitrile–water
Methanol–water
q
p
r2
n
q
p
r2
n
Symmetry column 1,2-Dihydroxybenzene Benzamide 3-Methylphenol Propiophenone Methyl benzoate 4-Nitrotoluene Butyrophenone Naphthalene Propylbenzene Heptanophenone Butylbenzene Chrysene
−1.01 −1.29 −1.03 −0.82 −0.83 −0.86 −0.86 −0.82 −0.73 −0.70 −0.57 −0.51
1.88 2.23 3.13 3.53 3.46 3.96 4.23 4.82 5.34 5.67 5.51 6.23
0.917 0.982 0.999 0.999 0.999 0.999 0.999 0.999 0.999 1 1 1
6 6 6 5 5 5 4 3 3 2 2 2
−1.05 −1.15 −0.87 −0.90 −0.87 −0.79 −0.89 −0.66 −0.61 −0.84 −0.60 −0.85
2.05 2.21 2.91 3.62 3.60 3.65 4.26 4.72 5.40 6.39 6.15 9.35
0.989 0.981 0.995 0.999 0.999 0.999 0.999 0.999 1 1 1 1
8 8 8 6 6 6 5 4 3 3 2 2
Chromolith column 1,2-Dihydroxybenzene Benzamide 3-Methylphenol Propiophenone Methyl benzoate 4-Nitrotoluene Butyrophenone Naphthalene Propylbenzene Heptanophenone Butylbenzene Chrysene
−1.40 −1.52 −1.41 −1.24 −1.25 −1.16 −1.28 −1.24 −1.17 −1.24 −1.12 −1.14
1.86 1.94 3.00 3.45 3.39 3.57 4.14 4.69 5.24 5.90 5.71 6.58
0.986 0.939 0.999 0.999 0.999 0.996 0.999 0.999 0.999 0.999 0.999 0.999
6 6 6 6 6 6 5 5 4 3 3 3
−1.56 −1.67 −1.38 −1.37 −1.30 −1.24 −1.36 −1.11 −1.06 −1.35 −1.08 −0.96
2.08 2.36 3.00 3.68 3.58 3.59 4.29 4.61 5.24 6.51 6.08 7.19
0.972 0.972 0.998 0.999 0.999 0.999 0.999 0.999 0.999 0.999 1 0.999
8 8 8 8 8 7 7 6 5 4 4 3
100
P. Izquierdo et al. / J. Chromatogr. A 1107 (2006) 96–103
Fig. 2. Correlations between the solute polarity parameters, p, from the database and the data obtained in each working chromatographic system on Symmetry C18 column in (a) acetonitrile–water and (b) methanol–water mobile phases and on Chromolith RP-18 column in (c) acetonitrile–water and (d) methanol–water.
capabilities of this set to obtain accurate (log k)0 and PsN parameters, the Spherisorb ODS-2 column has been characterised again using the selected compounds and the data treatment described in Section 2.5. The results obtained have been: (−1.02, −0.03) for acetonitrile–water and (−1.13, −0.05) for methanol–water mixtures, respectively, which are consistent with those already calculated and given in the previous paragraph. Therefore, the suitability of the proposed 12 compound set to characterise this column with a high accuracy has been demonstrated.
Fig. 1 shows the agreement between calculated and experimental retention values of the 152 compounds eluted on the Spherisorb ODS-2 column when the column parameters ((log k)0 and PsN ) were obtained from the selected 12 compounds and from the 152 compounds of Smith and Burr. Solute parameters, p, have been taken from the database. As previously observed, the standard deviation values are slightly greater for methanol–water than for acetonitrile–water mobile phases.
Fig. 3. Correlation between solute polarity parameters, from acetonitrile–water and methanol–water mobile phases when (a) Symmetry C18 and (b) Chromolith RP-18 were used.
P. Izquierdo et al. / J. Chromatogr. A 1107 (2006) 96–103 Table 4 Initial characterisation parameters Parameters
(log k)0 PsN r2 SD n
Symmetry C18
Chromolith RP-18
ACN–water
MeOH–water
ACN–water
MeOH–water
−1.39 −0.13 0.791 0.10 12
−1.03 −0.04 0.276 0.15 12
−1.52 −0.06 0.629 0.08 12
−1.73 −0.10 0.616 0.14 12
3.2. Characterisation of two C18 columns using the proposed set of compounds Two different C18 RPLC columns, Waters Symmetry C18 with a particulate packing, and Merck Chromolith Performance RP-18 with a monolithic packing, have been characterised by means of the established set of 12 compounds. The different steps of the characterisation process are shown in this section. The data treatment is described in Section 2.5. Initially, the retention of each compound, log k, is plotted N , calculated through against the polarity of the mobile phase, Pm Eqs. (4) or (5). As shown in Table 3, good linear correlations are obtained for all the compounds of the group, with a regression coefficient (r2 ) usually higher than 0.99. Also, q and p parameters for each compound corresponding to the intercept and the slope of Eq. (2) are shown. As expected, the values of p are slightly dependent on both, the mobile and the stationary phase and they are homogenously distributed through the entire p range (1.86–6.58 for acetonitrile–water, and 2.05–9.35 for methanol–water mobile phases). Normally, p values are larger for methanol than for acetonitrile. Thus, a specific change in the composition of the mobile phase would produce greater variations in methanol–water chromatographic retention than in acetonitrile–water mixtures. According to Eq. (3), the linear regression between q and p provided initial (log k)0 and PsN values. The results obtained Table 5 Parameters of the column and polarity parameter of each substance after the refining procedure Parameters
(log k)0 PsN p(1,2-Dihydroxybenzene) p(Benzamide) p(3-Methylphenol) p(Propiophenone) p(Methyl benzoate) p(4-Nitrotoluene) p(Butyrophenone) p(Naphthalene) p(Propylbenzene) p(Heptanophenone) p(Butylbenzene) p(Chrysene)
Symmetry C18
Chromolith RP-18
ACN– water
MeOH– water
ACN– water
MeOH– water
−1.40 −0.14 2.13 1.88 3.04 3.66 3.60 3.93 4.14 4.65 5.21 5.49 5.64 6.28
−1.31 −0.12 2.07 2.06 3.03 3.59 3.63 3.80 4.10 4.92 5.54 5.73 6.10 7.70
−1.49 −0.06 1.84 1.65 2.85 3.54 3.46 3.83 4.09 4.66 5.28 5.69 5.81 6.52
−1.76 −0.12 2.05 2.07 3.05 3.63 3.63 3.76 4.14 4.80 5.41 5.77 6.01 7.11
101
for each chromatographic system are shown in Table 4. It is noticeable that the regression coefficients (r2 ) are very different from 1 because the slope of these regression lines is close to 0. However, the standard deviations obtained are quite good. Values given in Table 4 are used as initial values in the iterative refining procedure. The initial p values for each chromatographic system are those shown in Table 3. The columns and solutes parameters after the refining process are given in Table 5. These values have been chosen as the optimal characterisation values and they provide, by means of Eq. (1), specific equations for each studied chromatographic system. The comparison of PsN parameters for the two solvated microparticulate columns shows that Spherisorb is more polar than Symmetry and Chromolith column shows an intermediate polarity. 3.3. Transfer of p values between columns The feasibility of transferring retention data from a given column to another column filled with a different stationary Table 6 List of 40 solutes used to test the accuracy of the predictions Butyrophenone Heptanophenone Propiophenone Valerophenone Phenol 2-Bromophenol 2,4-Dichlorophenol 2,4,6-Trichlorophenol Pentachlorophenol 3-Methylphenol 3-Nitrophenol Aniline N-Ethylaniline 2-Nitroanilne Benzaldehyde Benzamide 1,2-Dihydroxybenzene Benzene Biphenyl Chlorobenzene Propylbenzene Butylbenzene Hexachlorobenzene Methoxybenzene 1,4-Dimethylbenzene n-Hexylbenzene Methyl benzoate Methyl 4-hydroxybenzoate n-Propyl 4-hydroxybenzoate 4-Nitrotoluene Toluene Benzoic acid 3-Methylbenzoic acid 1-Naphthoic acid Thymol Chrysene Naphthalene Anthracene Pyrene 2-Naphthol
102
P. Izquierdo et al. / J. Chromatogr. A 1107 (2006) 96–103
phase was already shown in previous work [8]. Thus, p values of a series of compounds from two different columns are linearly related by means of Eq. (6). This equation can be easily employed to transfer data from the Spherisorb ODS-2 column (system1) to any of the columns characterised here (system2). For this purpose, the p values of the 12 solutes (Table 5) have been plotted against the corresponding p values from the database (Table 2). Good correlations, which show statistical parameters slightly better for acetonitrile mobile phases, have been obtained in all instances. The correlations for each system are shown in Fig. 2. 3.4. Transfer of p parameter between solvent systems Similarly to Section 3.3, the feasibility of retention data transfer between mobile phases with different organic modifier has been tested. The characterisation set is employed to establish, through Eq. (6), the linear relationships between p values obtained in the same column but different organic modifier in the mobile phase (Table 5). The plots are shown in Fig. 3. Fair statistical parameters are obtained on both columns. 3.5. Prediction of retention of several solutes In this section, the accuracy of the model is tested by comparison between predicted and experimental retention factors.
For this purpose, solute polarity parameters, p, for 40 compounds listed in Table 6 have been taken from the database. These p values have been transferred to the appropriate chromatographic system employing the equations shown in Fig. 2. The appropriate column parameters given in Table 5 have been N substituted in Eq. (1) for each chromatographic system. The Pm values for each specific mobile phase have been calculated by means of Eqs. (4) or (5). Finally, predicted data have been plotted against experimental data. The predictions were done in 10–60% acetonitrile–water and 10–80% methanol–water mobile phases. The results are shown in Fig. 4. Generally, predictions are better in mobile phases that contain acetonitrile than those with methanol, but the model shows good results in methanol too. Further, predictions of similar quality could be obtained on both columns although slightly better precision is achieved on the Symmetry C18 than on the Chromolith RP-18 column. This is clearly noticeable when mobile phases with a small amount of organic solvent (10%) are used on the Chromolith column. In these instances, the results obtained were not as accurate as those for other mobile phase compositions. Therefore, an improvement of the statistical parameters associated with the Chromolith column was attempted when the predictions for the 10% organic modifier containing mobile phase have been withdrawn. A r2 = 0.975, SD = 0.098 for a population of n = 180 points (Fig. 4c) in acetonitrile–water and r2 = 0.963, SD = 0.106 when n = 233 points (Fig. 4d) in methanol–water
Fig. 4. Predictions in the different studied chromatographic systems: (a) Symmetry C18 in acetonitrile–water mobile phases, (b) Symmetry C18 in methanol–water mobile phases, (c) Chromolith Performance RP-18 in acetonitrile–water mobile phases, (d) Chromolith Performance RP-18 in methanol–water mobile phases. Legend: () 80%, () 70%, () 60%, (×) 50%, (*) 40%, (䊉) 30%, (+) 20%, (−) 10% of organic modifier in the mobile phase.
P. Izquierdo et al. / J. Chromatogr. A 1107 (2006) 96–103
have been obtained. On the other hand, the results obtained on the Symmetry column when low concentrations of organic modifier are employed are equivalent to those obtained for the rest of mobile phase compositions. 4. Conclusions On the basis of the model to predict retention from the polarity of stationary and mobile phases and the polarity of the own analyte, a representative characterisation set of 12 compounds that allows the successful characterisation of RPLC columns is proposed. The solutes included in the characterisation set are homogenously distributed through the available polarity range and have different chemical nature. Column parameters for the Spherisorb ODS-2 column determined from the proposed set agree with those estimated from a much bigger number of compounds. This result validates the selected set of compounds to evaluate the polarity of RPLC columns. Two C18 columns with different silica support structure, particulate and monolithic (Waters Symmetry C18 and Merck Chromolith Performance RP-18) have been characterised using the proposed set and acetonitrile and methanol as organic modifiers of the mobile phase. Moreover, the linear equations that allow the data transfer between the studied chromatographic systems (columns and solvents) have been determined, too. The results obtained are used to predict the retention factors of 40 different substances on both working columns and a variety of acetonitrile–water and methanol–water mobile phases and they are consistent with the experimental results. Predictions using mobile phases with acetonitrile are more accurate than those with methanol. Otherwise, an increase in dispersion was observed when low proportions of organic modifier are present in the mobile phase for the Chromolith column. Acknowledgements We thank financial support from the Ministerio de Ciencia y Tecnolog´ıa of the Spanish Government and the Fondo Europeo de Desarrollo Regional of the European Union (Project CTQ2004-00633/BQU) and from the Catalan Government (Grant 2001SGR 00055).
103
References ´ [1] C.F. Poole, The Essence of Chromatography, Elsevier, Amsterdam, 2003. [2] C. Reichardt, Solvents and Solvent Effects in Organic Chemistry, third ed., Wiley, Weinheim, 2003. [3] B.P. Johnson, M.G. Khaledi, J.G. Dorsey, Anal. Chem. 58 (1986) 2354. [4] P.C. Sadek, P.W. Carr, R.M. Doherty, M.J. Kamlet, R.W. Taft, M.H. Abraham, Anal. Chem. 57 (1985) 2971. [5] M.H. Abraham, Chem. Soc. Rev. 23 (1993) 660. [6] M.H. Abraham, M. Ros´es, J. Phys. Org. Chem. 7 (1994) 672. [7] M.H. Abraham, M. Ros´es, C.F. Poole, S.K. Poole, J. Phys. Org. Chem. 10 (1997) 358. [8] W.J. Cheong, P.W. Carr, Anal. Chem. 60 (1988) 820. [9] W.J. Cheong, P.W. Carr, Anal. Chem. 61 (1989) 1524. [10] M. Reta, P.W. Carr, P.C. Sadek, S.C. Rutan, Anal. Chem. 71 (1999) 3484. [11] N.S. Wilson, M.D. Nelson, J.W. Dolan, L.R. Snyder, R.G. Wolcott, P.W. Carr, J. Chromatogr. A 961 (2002) 171. [12] N.S. Wilson, M.D. Nelson, J.W. Dolan, L.R. Snyder, P.W. Carr, J. Chromatogr. A 961 (2002) 195. [13] N.S. Wilson, J.W. Dolan, L.R. Snyder, P.W. Carr, L.C. Sander, J. Chromatogr. A 961 (2002) 217. [14] J.J. Gilroy, J.W. Dolan, L.R. Snyder, J. Chromatogr. A 1000 (2003) 757. [15] M. Ros´es, E. Bosch, Anal. Chim. Acta 274 (1993) 147. [16] E. Bosch, P. Bou, M. Ros´es, Anal. Chim. Acta 299 (1994) 219. [17] M. Ros´es, P. Bou, E. Bosch, K. Siigur, Org. React. 29 (1995) 51. ´ [18] J.R. Torres-Lapasi´o, M.C. Garc´ıa-Alvarez-Coque, M. Ros´es, E. Bosch, J. Chromatogr. A 955 (2002) 19. ´ [19] R. Bosque, J. Sales, E. Bosch, M. Ros´es, M.C. Garc´ıa-Alvarez-Coque, J.M. Torres-Lapasi´o, J. Chem. Inf. Comput. Sci. 43 (2003) 1240. ´ [20] J.R. Torres-Lapasi´o, M.C. Garc´ıa-Alvarez-Coque, M. Ros´es, E. Bosch, A.M. Zissimos, M.H. Abraham, Anal. Chim. Acta 515 (2004) 209. ´ [21] J.R. Torres-Lapasi´o, M.C. Garc´ıa-Alvarez-Coque, E. Bosch, M. Ros´es, J. Chromatogr. A 1089 (2005) 170. [22] R.M. Smith, C.M. Burr, J. Chromatogr. 475 (1989) 57. [23] R.M. Smith, C.M. Burr, J. Chromatogr. 475 (1989) 75. [24] R.M. Smith, C.M. Burr, J. Chromatogr. 481 (1989) 71. [25] R.M. Smith, C.M. Burr, J. Chromatogr. 481 (1989) 85. [26] R.M. Smith, C.M. Burr, J. Chromatogr. 485 (1989) 325. [27] R.M. Smith, C.M. Burr, J. Chromatogr. 550 (1991) 335. [28] T. Hanai, J. Hubert, J. Chromatogr. 302 (1984) 89. [29] T. Hanai, J. Hubert, J. Chromatogr. 290 (1984) 197. [30] T. Hanai, J. Hubert, J. High Resolut. Chromatogr. Chromatogr. Commun. 6 (1983) 20. [31] A. Kaibara, M. Hirose, T. Nak, Chromatographia 29 (1990) 551.
Polarity parameters of the Symmetry C18 and Chromolith Performance RP-18 monolithic chromatographic columns Pere Izquierdo, Mart´ı Ros´es, Elisabeth Bosch ∗ Departament de Qu´ımica Anal´ıtica, Universitat de Barcelona, Diagonal 647, 08028 Barcelona, Spain Received 17 August 2005; received in revised form 18 November 2005; accepted 5 December 2005 Available online 27 December 2005
Abstract A set of 12 compounds of different chemical nature has been established to characterise RPLC columns on the basis of a polarity retention model previously developed: log k = (log k)0 + p(PmN − PsN ). This model allows the calculation of the retention factor (k) of any non-ionized compound using one parameter which describes the polarity of the solute (p), another one for the polarity of the mobile phase (PmN ) and two more parameters for the characterisation of the stationary phase ((log k)0 and PsN ). The selected set of compounds allows the determination of (log k)0 and PsN of stationary phases and it has been used to characterise two commercial columns (Symmetry C18 from Waters and Chromolith Performance RP-18 monolithic from Merck). Column parameters, together with those of the mobile phase permit successful transfer of retention data between chromatographic systems. Prediction of retention of a variety of non-ionized analytes has been also successfully achieved using the column descriptors and p values of solutes from a previously established p data base. © 2005 Elsevier B.V. All rights reserved. Keywords: Column polarity; Retention prediction; Polarity model; Column characterisation
1. Introduction In RPLC, an accurate prediction of the analyte retention factor before the experimental analysis allows a significant simplification of optimization procedures for the experimental conditions. Several models have been proposed to explain the chromatographic behaviour of the solutes in RPLC. Some of these approaches are based only on solute-mobile phase interactions, where descriptors such as the organic modifier volume fraction in the mobile phase (ϕ) [1] or mobile phase polarity expressed through Dimroth and Reichardt parameter (ETN ) [2,3] were related with the logarithm of the retention factor. Although acceptable predictions could be achieved using these models, accurate predictions could be done only in a specific range of mobile phase compositions. Further, it is quite obvious that chromatographic retention is too complex to be explained considering solely solute-mobile phase interactions. More complete models try to explain the chromatographic behaviour using mobile phase-solute-stationary phase interac-
tions. One approach was introduced by Abraham and coworkers [4–7] and extended by Carr and coworkers [8–10]. They developed a solvation equation, based in a linear free energy relationship, which accounts for the interactions of the solute with the mobile and stationary phases, due to hydrophobicity, polarity, polarizability, hydrogen-bond acidity and hydrogen-bond basicity. Snyder and coworkers [11–14] proposed an empirical linear-free-energy equation of form similar to Abraham’s equation to characterise the column selectivity in terms of five retention-related parameters that accounts for column hydrophobicity, steric selectivity, hydrogen-bond acidity, hydrogen-bond basicity and cation exchange behaviour. Our model (Eq. (1)) was directly derived from the Abraham–Carr equations [15,16] and predicts the retention factor of any non-ionized solute from polarity parameters of the N ) and stationary ((log k) and P N ) phases and the mobile (Pm 0 s polarity of the chromatographed solute (p) [16–21], according to N log k = (log k)0 + p(Pm − PsN )
(1)
or ∗
Corresponding author. Tel.: +34 934021284; fax: +34 934021233. E-mail address: [email protected] (E. Bosch).
0021-9673/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.chroma.2005.12.019
N log k = q + pPm
(2)
P. Izquierdo et al. / J. Chromatogr. A 1107 (2006) 96–103
2. Experimental
where q=
97
(log k)0 − pPsN
(3)
(log k)0 and PsN are related to the stationary phase, although they also depend on the organic modifier used in the mobile phase. PsN includes hydrophobicity, dipolarity and hydrogen-bond ability of the stationary phase and it is a generic retentivity descriptor of the column. (log k)0 is the intercept of the correlation and it accounts for the retention in the hypothetical case where mobile and stationary phases show the same polarity. It mainly depends N is a polarity parameter easily calculaon the phase ratio. Pm ble from the volume fraction of organic modifier in the mobile phase (ϕ) by means of Eqs. (4) and (5) for acetonitrile–water and methanol–water mobile phases, respectively [16,18,20,21]. N Pm = 1.00 −
2.13ϕ 1 + 1.42ϕ
(4)
N = 1.00 − Pm
1.33ϕ 1 + 0.47ϕ
(5)
The polarity of the solute, p, mainly depends on the compound, but the organic modifier used in the mobile phase and the stationary phase itself produce small changes of its values as well. A wide database, which contains p values for more than 250 compounds, and referred to a Waters Spherisorb ODS2 (100 mm × 5 mm) column and acetonitrile–water mixtures as mobile phases, was built [18,19]. The database p value for a solute is linearly related to p of the same solute in any different chromatographic system (column and/or mobile phase) according to psystem,2 = a psystem,1 + b
(6)
where psystem,1 is the p value in the database of any solute and psystem,2 is the p value of the same solute in any other chromatographic system. Further, a and b are constants that depend only on the related chromatographic systems. Then, a proper column characterisation allows accurate prediction of retention and retention data transfer between columns and mobile phases, as has been widely demonstrated [16,18]. The aim of this paper is to propose a fast, simple and reliable procedure to characterise reversed-phase chromatographic columns, that is, to obtain (log k)0 and PsN parameters of a column for acetonitrile/water and methanol/water mobile phases. Thus, on the basis of the p database, a representative characterisation set of 12 compounds has been selected. The column parameters for the own chromatographic system used to build the database have been recalculated using the selected set. Results agree with those derived from the retention of more than 150 compounds and 600 experimental retention measurements. Therefore, the usefulness of the selected set has been validated in this way, and the selected compounds have been used for the characterisation of two commercial C18 columns with different stationary phases, one particulate and the other one monolithic. Good results for both retention prediction and retention data transfer between chromatographic systems, have been achieved using the derived parameters for both columns and acetonitrile and methanol as mobile phases organic modifiers.
2.1. Apparatus pH measurements were taken with a combined Ross electrode Orion 8102 (glass electrode with a 3.0 mol L−1 KCl solutions as a salt bridge) in a Crison micropH 2002 potentiometer with a precision of ±0.1 mV (±0.002 pH units). A Shimadzhu (Kyoto, Japan) HPLC system equipped with two LC-10ADvp dual pumps, a SIL-10ADvp autoinjector fixed to 10 l, a SPD-10AVvp UV detector, a CTO-10ASvp column oven at 25 ± 0.1 ◦ C and a SCL-10Avp system controller was employed. The detection wavelengths were 200 nm for KBr, 254 nm for benzene derivatives and 282 nm for phenols. 2.2. Columns Retention data were measured on a Waters Symmetry C18, 5 m, 150 mm × 4.6 mm column and a Merck Chromolith Performance RP-18 monolithic 100 mm × 4.6 mm column. Characteristics of the studied columns are given in Table 1. 2.3. Chemicals Methanol and acetonitrile were HPLC grade from Merck and water was purified by the Milli-Q plus system from Millipore. Other chemicals were reagent grade or better and were from Aldrich, Carlo-Erba, Flucka, Merck, Normapur or Sigma. Solutions of 100 g mL−1 of each substance were prepared in methanol. 2.4. Procedure Each compound was chromatographed on both columns at 25 ◦ C, using a variety of methanol–water and acetonitrile–water mobile phases (10, 20, 30, 40, 50, 60, 70, and 80% in volume of organic modifier). Most mobile phases were prepared with aqueous 0.1 M acetic acid (pH 2.7) and the organic modiTable 1 Chromatographic features of the chromatographic columns studied Features
Chromolith Performance
Symmetry C18
Silica type Structure type Particle size (m) Total porosity (%) ˚ Median pore diameter (A) Macropore size (m) ˚ Mesopore size (A) Surface area (m2 g−1 ) Pore volume (ml g−1 ) Total carbon Endcapped Column length (mm) Column diameter (mm) pH stability range Temperature stability (◦ C) Pressure stability (bar) Flow rate (ml min−1 )
High purity Monolithic – >80 – 2 130 300 1 18 Yes 100 4.6 2–7.5 max. 45 max. 200 2
High purity Particulate 5 65 86 – – 346 – 19.57 Yes 150 4.6 2–8 – – 1
98
P. Izquierdo et al. / J. Chromatogr. A 1107 (2006) 96–103
fier, although mobile phases based in aqueous phosphate buffer (pH 7.3) were employed for aniline derivatives. A solution of 100 g mL−1 of KBr was used as void volume marker. The flow rate was 2 mL min−1 on the Chromolith Performance column and 1 mL min−1 on the Symmetry C18 column. A 10 L aliquot of each sample was injected. All the measurements were taken in triplicate. 2.5. Data treatment Starting values were obtained from retention data of a group of compounds which was chromatographed on each column using acetonitrile–water and methanol–water with different contents of organic modifier as mobile phases. N , which is According to Eq. (2), log k is linearly related to Pm easily calculable from Eqs. (4) or (5). Then, Eq. (2) allowed the computation of q and p parameters for each solute and organic modifier. Paired values, q and p, of the totality of compounds were fitted to Eq. (3), which provided a first approach to the column parameters, (log k)0 and PsN . To refine the (log k)0 and PsN values and also the descriptor p of each compound, an iterative process was carried out by comparison of predicted and experimental log k values. Initial (log k)0 and PsN were those calculated by Eq. (3) and initial p for each compound were those computed by Eq. (2). In a particular step, predicted log k values are those calculated by Eq. (1) from the
Table 2 Characterisation set and their p values from the database [8] Substances
p (ACN–water)
p (MeOH–Water)
1,2-Dihydroxybenzene Benzamide 3-Methylphenol Propiophenone Methyl benzoate 4-Nitrotoluene Butyrophenone Naphthalene Propylbenzene Heptanophenone Butylbenzene Chrysene
1.79 1.69 2.82 3.70 3.59 3.95 4.19 4.86 5.56 5.78 6.13 7.33
1.98 2.10 2.98 3.88 3.89 4.08 4.36 5.12 5.71 6.06 6.32 8.46
refined parameters of the preceding step. Compounds with a p residual value higher than three times the standard deviation were withdrawn in the following step. Progressively, the (log k)0 and PsN of the column and p for each solute were improved together and the new values were substituted in Eq. (1) to continue with the iterative process. Step by step, in subsequent iterations the sum of residual squares value (δ2 ) decreased until it reached a minimum and those polarity parameters in the minimum were accepted as the optimal. To implement the iterative process, the Solver tool of Microsoft Excel was used.
Fig. 1. Prediction of retention factors of 152 compounds on Spherisorb ODS-2 column. Column parameters determined through the 12 compounds set and (a) acetonitrile–water and (b) methanol–water mobile phases. Column parameters determined through the 152 compounds set and (c) acetonitrile–water and (d) methanol–water mobile phases.
P. Izquierdo et al. / J. Chromatogr. A 1107 (2006) 96–103
3. Results and discussion 3.1. Proposal of a characterisation set of compounds As it was previously demonstrated, when the chromatographic system characterisation parameters were obtained using a large number of compounds, the polarity model, Eq. (1), provided very good retention predictions [16–18]. Thus, this procedure guarantees accurate values for both column ((log k)0 and PsN ) and compound (p) parameters. On the basis of literature data [22–31] and our own measurements [6,15,16], we reported a wide p database, for more than 250 compounds. This database is related to a specific column, a Waters Spherisorb ODS-2, and acetonitrile and methanol as the organic modifiers of the mobile phase. These chromatographic systems were used by Smith and Burr [22–27] to determine retention factors of compounds with different chemical structure and the reported data were the source for a preliminary p database of 152 compounds. Some compounds of this data set were also chromatographed by other authors [6,15,16,28–31] on a variety of stationary phases, that allowed us to get their p values related to each chromatographic system and, also, to correlate them with p values of the original database. On the basis of these correlations the p database was extended to more than 250 compounds [18,19]. The Spherisorb ODS-2 column was characterised according to Eq. (1), using the 152 compounds chromatographed by Smith and Burr. The column parameters (log k)0 and PsN , after
99
a refining process, were (−1.04, −0.03) for acetonitrile–water and (−1.26, −0.08) for methanol–water mixtures, respectively, and they allowed to predict successfully the retention of many other compounds on this column [18]. However, in practice, it would be desirable to get column descriptors of similar quality by means of a faster characterisation procedure, which involves a smaller number of experiments. This means the proposal and validation of a small set of representative compounds, able to carry out the characterisation process and to get properly the parameters of any column with any mobile phase. Then, a complete statistical study was made with groups of 10 compounds randomly selected among the 152 compounds belonging to the initial database and satisfactory results were obtained with most of them. Thus, derived values for (log k)0 and PsN were (−1.01, −0.03) for acetonitrile–water and (−1.17, −0.06) for methanol–water mixtures, respectively [18]. In this work, a set of 12 compounds has been selected from the whole database, composed by more than 250 compounds. This group includes solutes of diverse polarity, homogenously distributed through all the available p range, and contains compounds that show different functional groups. It includes a significant fraction of highly hydrophobic compounds (p > 5) to ensure correct predictions for this kind of solutes, and this is a major difference with respect to the previous studied characterisation sets mentioned above. Further, these compounds are commonly used in laboratory practice. The selected compounds and their p database values for acetonitrile–water and methanol–water mobile phases are presented in Table 2. In order to check the
Table 3 Initial solute parameters obtained by Eq. (2) Compound
Acetonitrile–water
Methanol–water
q
p
r2
n
q
p
r2
n
Symmetry column 1,2-Dihydroxybenzene Benzamide 3-Methylphenol Propiophenone Methyl benzoate 4-Nitrotoluene Butyrophenone Naphthalene Propylbenzene Heptanophenone Butylbenzene Chrysene
−1.01 −1.29 −1.03 −0.82 −0.83 −0.86 −0.86 −0.82 −0.73 −0.70 −0.57 −0.51
1.88 2.23 3.13 3.53 3.46 3.96 4.23 4.82 5.34 5.67 5.51 6.23
0.917 0.982 0.999 0.999 0.999 0.999 0.999 0.999 0.999 1 1 1
6 6 6 5 5 5 4 3 3 2 2 2
−1.05 −1.15 −0.87 −0.90 −0.87 −0.79 −0.89 −0.66 −0.61 −0.84 −0.60 −0.85
2.05 2.21 2.91 3.62 3.60 3.65 4.26 4.72 5.40 6.39 6.15 9.35
0.989 0.981 0.995 0.999 0.999 0.999 0.999 0.999 1 1 1 1
8 8 8 6 6 6 5 4 3 3 2 2
Chromolith column 1,2-Dihydroxybenzene Benzamide 3-Methylphenol Propiophenone Methyl benzoate 4-Nitrotoluene Butyrophenone Naphthalene Propylbenzene Heptanophenone Butylbenzene Chrysene
−1.40 −1.52 −1.41 −1.24 −1.25 −1.16 −1.28 −1.24 −1.17 −1.24 −1.12 −1.14
1.86 1.94 3.00 3.45 3.39 3.57 4.14 4.69 5.24 5.90 5.71 6.58
0.986 0.939 0.999 0.999 0.999 0.996 0.999 0.999 0.999 0.999 0.999 0.999
6 6 6 6 6 6 5 5 4 3 3 3
−1.56 −1.67 −1.38 −1.37 −1.30 −1.24 −1.36 −1.11 −1.06 −1.35 −1.08 −0.96
2.08 2.36 3.00 3.68 3.58 3.59 4.29 4.61 5.24 6.51 6.08 7.19
0.972 0.972 0.998 0.999 0.999 0.999 0.999 0.999 0.999 0.999 1 0.999
8 8 8 8 8 7 7 6 5 4 4 3
100
P. Izquierdo et al. / J. Chromatogr. A 1107 (2006) 96–103
Fig. 2. Correlations between the solute polarity parameters, p, from the database and the data obtained in each working chromatographic system on Symmetry C18 column in (a) acetonitrile–water and (b) methanol–water mobile phases and on Chromolith RP-18 column in (c) acetonitrile–water and (d) methanol–water.
capabilities of this set to obtain accurate (log k)0 and PsN parameters, the Spherisorb ODS-2 column has been characterised again using the selected compounds and the data treatment described in Section 2.5. The results obtained have been: (−1.02, −0.03) for acetonitrile–water and (−1.13, −0.05) for methanol–water mixtures, respectively, which are consistent with those already calculated and given in the previous paragraph. Therefore, the suitability of the proposed 12 compound set to characterise this column with a high accuracy has been demonstrated.
Fig. 1 shows the agreement between calculated and experimental retention values of the 152 compounds eluted on the Spherisorb ODS-2 column when the column parameters ((log k)0 and PsN ) were obtained from the selected 12 compounds and from the 152 compounds of Smith and Burr. Solute parameters, p, have been taken from the database. As previously observed, the standard deviation values are slightly greater for methanol–water than for acetonitrile–water mobile phases.
Fig. 3. Correlation between solute polarity parameters, from acetonitrile–water and methanol–water mobile phases when (a) Symmetry C18 and (b) Chromolith RP-18 were used.
P. Izquierdo et al. / J. Chromatogr. A 1107 (2006) 96–103 Table 4 Initial characterisation parameters Parameters
(log k)0 PsN r2 SD n
Symmetry C18
Chromolith RP-18
ACN–water
MeOH–water
ACN–water
MeOH–water
−1.39 −0.13 0.791 0.10 12
−1.03 −0.04 0.276 0.15 12
−1.52 −0.06 0.629 0.08 12
−1.73 −0.10 0.616 0.14 12
3.2. Characterisation of two C18 columns using the proposed set of compounds Two different C18 RPLC columns, Waters Symmetry C18 with a particulate packing, and Merck Chromolith Performance RP-18 with a monolithic packing, have been characterised by means of the established set of 12 compounds. The different steps of the characterisation process are shown in this section. The data treatment is described in Section 2.5. Initially, the retention of each compound, log k, is plotted N , calculated through against the polarity of the mobile phase, Pm Eqs. (4) or (5). As shown in Table 3, good linear correlations are obtained for all the compounds of the group, with a regression coefficient (r2 ) usually higher than 0.99. Also, q and p parameters for each compound corresponding to the intercept and the slope of Eq. (2) are shown. As expected, the values of p are slightly dependent on both, the mobile and the stationary phase and they are homogenously distributed through the entire p range (1.86–6.58 for acetonitrile–water, and 2.05–9.35 for methanol–water mobile phases). Normally, p values are larger for methanol than for acetonitrile. Thus, a specific change in the composition of the mobile phase would produce greater variations in methanol–water chromatographic retention than in acetonitrile–water mixtures. According to Eq. (3), the linear regression between q and p provided initial (log k)0 and PsN values. The results obtained Table 5 Parameters of the column and polarity parameter of each substance after the refining procedure Parameters
(log k)0 PsN p(1,2-Dihydroxybenzene) p(Benzamide) p(3-Methylphenol) p(Propiophenone) p(Methyl benzoate) p(4-Nitrotoluene) p(Butyrophenone) p(Naphthalene) p(Propylbenzene) p(Heptanophenone) p(Butylbenzene) p(Chrysene)
Symmetry C18
Chromolith RP-18
ACN– water
MeOH– water
ACN– water
MeOH– water
−1.40 −0.14 2.13 1.88 3.04 3.66 3.60 3.93 4.14 4.65 5.21 5.49 5.64 6.28
−1.31 −0.12 2.07 2.06 3.03 3.59 3.63 3.80 4.10 4.92 5.54 5.73 6.10 7.70
−1.49 −0.06 1.84 1.65 2.85 3.54 3.46 3.83 4.09 4.66 5.28 5.69 5.81 6.52
−1.76 −0.12 2.05 2.07 3.05 3.63 3.63 3.76 4.14 4.80 5.41 5.77 6.01 7.11
101
for each chromatographic system are shown in Table 4. It is noticeable that the regression coefficients (r2 ) are very different from 1 because the slope of these regression lines is close to 0. However, the standard deviations obtained are quite good. Values given in Table 4 are used as initial values in the iterative refining procedure. The initial p values for each chromatographic system are those shown in Table 3. The columns and solutes parameters after the refining process are given in Table 5. These values have been chosen as the optimal characterisation values and they provide, by means of Eq. (1), specific equations for each studied chromatographic system. The comparison of PsN parameters for the two solvated microparticulate columns shows that Spherisorb is more polar than Symmetry and Chromolith column shows an intermediate polarity. 3.3. Transfer of p values between columns The feasibility of transferring retention data from a given column to another column filled with a different stationary Table 6 List of 40 solutes used to test the accuracy of the predictions Butyrophenone Heptanophenone Propiophenone Valerophenone Phenol 2-Bromophenol 2,4-Dichlorophenol 2,4,6-Trichlorophenol Pentachlorophenol 3-Methylphenol 3-Nitrophenol Aniline N-Ethylaniline 2-Nitroanilne Benzaldehyde Benzamide 1,2-Dihydroxybenzene Benzene Biphenyl Chlorobenzene Propylbenzene Butylbenzene Hexachlorobenzene Methoxybenzene 1,4-Dimethylbenzene n-Hexylbenzene Methyl benzoate Methyl 4-hydroxybenzoate n-Propyl 4-hydroxybenzoate 4-Nitrotoluene Toluene Benzoic acid 3-Methylbenzoic acid 1-Naphthoic acid Thymol Chrysene Naphthalene Anthracene Pyrene 2-Naphthol
102
P. Izquierdo et al. / J. Chromatogr. A 1107 (2006) 96–103
phase was already shown in previous work [8]. Thus, p values of a series of compounds from two different columns are linearly related by means of Eq. (6). This equation can be easily employed to transfer data from the Spherisorb ODS-2 column (system1) to any of the columns characterised here (system2). For this purpose, the p values of the 12 solutes (Table 5) have been plotted against the corresponding p values from the database (Table 2). Good correlations, which show statistical parameters slightly better for acetonitrile mobile phases, have been obtained in all instances. The correlations for each system are shown in Fig. 2. 3.4. Transfer of p parameter between solvent systems Similarly to Section 3.3, the feasibility of retention data transfer between mobile phases with different organic modifier has been tested. The characterisation set is employed to establish, through Eq. (6), the linear relationships between p values obtained in the same column but different organic modifier in the mobile phase (Table 5). The plots are shown in Fig. 3. Fair statistical parameters are obtained on both columns. 3.5. Prediction of retention of several solutes In this section, the accuracy of the model is tested by comparison between predicted and experimental retention factors.
For this purpose, solute polarity parameters, p, for 40 compounds listed in Table 6 have been taken from the database. These p values have been transferred to the appropriate chromatographic system employing the equations shown in Fig. 2. The appropriate column parameters given in Table 5 have been N substituted in Eq. (1) for each chromatographic system. The Pm values for each specific mobile phase have been calculated by means of Eqs. (4) or (5). Finally, predicted data have been plotted against experimental data. The predictions were done in 10–60% acetonitrile–water and 10–80% methanol–water mobile phases. The results are shown in Fig. 4. Generally, predictions are better in mobile phases that contain acetonitrile than those with methanol, but the model shows good results in methanol too. Further, predictions of similar quality could be obtained on both columns although slightly better precision is achieved on the Symmetry C18 than on the Chromolith RP-18 column. This is clearly noticeable when mobile phases with a small amount of organic solvent (10%) are used on the Chromolith column. In these instances, the results obtained were not as accurate as those for other mobile phase compositions. Therefore, an improvement of the statistical parameters associated with the Chromolith column was attempted when the predictions for the 10% organic modifier containing mobile phase have been withdrawn. A r2 = 0.975, SD = 0.098 for a population of n = 180 points (Fig. 4c) in acetonitrile–water and r2 = 0.963, SD = 0.106 when n = 233 points (Fig. 4d) in methanol–water
Fig. 4. Predictions in the different studied chromatographic systems: (a) Symmetry C18 in acetonitrile–water mobile phases, (b) Symmetry C18 in methanol–water mobile phases, (c) Chromolith Performance RP-18 in acetonitrile–water mobile phases, (d) Chromolith Performance RP-18 in methanol–water mobile phases. Legend: () 80%, () 70%, () 60%, (×) 50%, (*) 40%, (䊉) 30%, (+) 20%, (−) 10% of organic modifier in the mobile phase.
P. Izquierdo et al. / J. Chromatogr. A 1107 (2006) 96–103
have been obtained. On the other hand, the results obtained on the Symmetry column when low concentrations of organic modifier are employed are equivalent to those obtained for the rest of mobile phase compositions. 4. Conclusions On the basis of the model to predict retention from the polarity of stationary and mobile phases and the polarity of the own analyte, a representative characterisation set of 12 compounds that allows the successful characterisation of RPLC columns is proposed. The solutes included in the characterisation set are homogenously distributed through the available polarity range and have different chemical nature. Column parameters for the Spherisorb ODS-2 column determined from the proposed set agree with those estimated from a much bigger number of compounds. This result validates the selected set of compounds to evaluate the polarity of RPLC columns. Two C18 columns with different silica support structure, particulate and monolithic (Waters Symmetry C18 and Merck Chromolith Performance RP-18) have been characterised using the proposed set and acetonitrile and methanol as organic modifiers of the mobile phase. Moreover, the linear equations that allow the data transfer between the studied chromatographic systems (columns and solvents) have been determined, too. The results obtained are used to predict the retention factors of 40 different substances on both working columns and a variety of acetonitrile–water and methanol–water mobile phases and they are consistent with the experimental results. Predictions using mobile phases with acetonitrile are more accurate than those with methanol. Otherwise, an increase in dispersion was observed when low proportions of organic modifier are present in the mobile phase for the Chromolith column. Acknowledgements We thank financial support from the Ministerio de Ciencia y Tecnolog´ıa of the Spanish Government and the Fondo Europeo de Desarrollo Regional of the European Union (Project CTQ2004-00633/BQU) and from the Catalan Government (Grant 2001SGR 00055).
103
References ´ [1] C.F. Poole, The Essence of Chromatography, Elsevier, Amsterdam, 2003. [2] C. Reichardt, Solvents and Solvent Effects in Organic Chemistry, third ed., Wiley, Weinheim, 2003. [3] B.P. Johnson, M.G. Khaledi, J.G. Dorsey, Anal. Chem. 58 (1986) 2354. [4] P.C. Sadek, P.W. Carr, R.M. Doherty, M.J. Kamlet, R.W. Taft, M.H. Abraham, Anal. Chem. 57 (1985) 2971. [5] M.H. Abraham, Chem. Soc. Rev. 23 (1993) 660. [6] M.H. Abraham, M. Ros´es, J. Phys. Org. Chem. 7 (1994) 672. [7] M.H. Abraham, M. Ros´es, C.F. Poole, S.K. Poole, J. Phys. Org. Chem. 10 (1997) 358. [8] W.J. Cheong, P.W. Carr, Anal. Chem. 60 (1988) 820. [9] W.J. Cheong, P.W. Carr, Anal. Chem. 61 (1989) 1524. [10] M. Reta, P.W. Carr, P.C. Sadek, S.C. Rutan, Anal. Chem. 71 (1999) 3484. [11] N.S. Wilson, M.D. Nelson, J.W. Dolan, L.R. Snyder, R.G. Wolcott, P.W. Carr, J. Chromatogr. A 961 (2002) 171. [12] N.S. Wilson, M.D. Nelson, J.W. Dolan, L.R. Snyder, P.W. Carr, J. Chromatogr. A 961 (2002) 195. [13] N.S. Wilson, J.W. Dolan, L.R. Snyder, P.W. Carr, L.C. Sander, J. Chromatogr. A 961 (2002) 217. [14] J.J. Gilroy, J.W. Dolan, L.R. Snyder, J. Chromatogr. A 1000 (2003) 757. [15] M. Ros´es, E. Bosch, Anal. Chim. Acta 274 (1993) 147. [16] E. Bosch, P. Bou, M. Ros´es, Anal. Chim. Acta 299 (1994) 219. [17] M. Ros´es, P. Bou, E. Bosch, K. Siigur, Org. React. 29 (1995) 51. ´ [18] J.R. Torres-Lapasi´o, M.C. Garc´ıa-Alvarez-Coque, M. Ros´es, E. Bosch, J. Chromatogr. A 955 (2002) 19. ´ [19] R. Bosque, J. Sales, E. Bosch, M. Ros´es, M.C. Garc´ıa-Alvarez-Coque, J.M. Torres-Lapasi´o, J. Chem. Inf. Comput. Sci. 43 (2003) 1240. ´ [20] J.R. Torres-Lapasi´o, M.C. Garc´ıa-Alvarez-Coque, M. Ros´es, E. Bosch, A.M. Zissimos, M.H. Abraham, Anal. Chim. Acta 515 (2004) 209. ´ [21] J.R. Torres-Lapasi´o, M.C. Garc´ıa-Alvarez-Coque, E. Bosch, M. Ros´es, J. Chromatogr. A 1089 (2005) 170. [22] R.M. Smith, C.M. Burr, J. Chromatogr. 475 (1989) 57. [23] R.M. Smith, C.M. Burr, J. Chromatogr. 475 (1989) 75. [24] R.M. Smith, C.M. Burr, J. Chromatogr. 481 (1989) 71. [25] R.M. Smith, C.M. Burr, J. Chromatogr. 481 (1989) 85. [26] R.M. Smith, C.M. Burr, J. Chromatogr. 485 (1989) 325. [27] R.M. Smith, C.M. Burr, J. Chromatogr. 550 (1991) 335. [28] T. Hanai, J. Hubert, J. Chromatogr. 302 (1984) 89. [29] T. Hanai, J. Hubert, J. Chromatogr. 290 (1984) 197. [30] T. Hanai, J. Hubert, J. High Resolut. Chromatogr. Chromatogr. Commun. 6 (1983) 20. [31] A. Kaibara, M. Hirose, T. Nak, Chromatographia 29 (1990) 551.