Anal. Chem. 2000, 72, 2788-2796
Application of the Thermally Tuned Tandem Column Concept to the Separation of Several Families of Environmental Toxicants Yun Mao and Peter W. Carr*
Department of Chemistry, University of Minnesota, Smith and Kolthoff Hall, 207 Pleasant Street, S.E., Minneapolis, Minnesota 55455
Separations of several families of environmental toxicants were optimized by means of the thermally tuned tandem column (T3C) concept. We use a tandem combination of an octadecylsilane (ODS) and a carbon-coated zirconia (C-ZrO2) column; and tune the selectivity by independently adjusting the isothermal temperatures of the two columns. This results in the change in the contribution that each column makes to the overall retention and selectivity. The separation was optimized by locating the optimum pair of column temperatures which give the best separation of the critical solute pair. For both triazine herbicides and carbamate pesticides samples, dramatically different selectivities and different critical pairs were observed for the two types of phases. Although neither individual phase gave adequate separation, the T3C approach provided baseline separations using only four preliminary trial separations. We also showed that, for the triazine samples, the T3C approach gave a better separation than did conventional mobile phase optimization with an ODS column. The combination of superior selectivity of T3C and high flow rate allows the baseline separation of complex mixtures in just a few minutes. In a previous paper we reported a novel approach to optimizing chromatographic selectivity in liquid chromatography: the thermally tuned tandem column (T3C) concept.1 Two columns with distinctively chemically different stationary phases are serially coupled and independently temperature controlled (see Figure 1). We have shown that the temperature affects selectivity in T3C through the continuous variation of the contribution that each column makes to the overall selectivity. T3C provides a new way to continuously and conveniently optimize resolution in reversedphase liquid chromatography (RPLC). Optimizing resolution is a common practice in RPLC method development. In most cases, resolution is improved by optimizing the sample retention (retention factor, k′), column efficiency (plate number, N) and band spacing (selectivity factor, R). Major emphasis is usually given to optimizing selectivity because it has the greatest impact on resolution.2 Useful method development strategies for optimizing selectivity are based on manipulating the (1) Mao, Y.; Carr, P. W. Anal. Chem. 2000, 72, 110-118. (2) Snyder, L. R.; Glajch, J. L.; Kirkland, J. J. Practical HPLC Method Development, 2nd ed.; Wiley-Interscience: New York, 1996.
2788 Analytical Chemistry, Vol. 72, No. 13, July 1, 2000
Figure 1. Block diagram of the T3C system.
experimental conditions, such as the mobile phase composition (B%), mobile phase type, stationary phase type, and sometimes the temperature. Although mobile phase composition is easy to change, it usually has a smaller effect on the effective chromatographic selectivity than does the type of mobile or stationary phase.3 However, several practical issues limit the use of varying the type of the mobile and stationary phases, including long equilibration times and, more importantly, the discontinuity and unpredictability of the changes in selectivity. One way to circumvent the discontinuity in the mobile phase type is to use ternary or quaternary mobile phases which blend different mobile phase additives in various proportions. However, due to the complex relationship between retention factor and mobile phase composition, large numbers of experimental runs are required for method development.2 In addition, the choice of the mobile phase type for RPLC is rather limited. The T3C concept allows us to tune the selectivity of the stationary phase type. It is quite analogous to the idea of using ternary mobile phases but with many more options in the type of stationary phase. Moreover, we have shown that, due to the simple relationship between retention factor and temperature, only four to five initial runs are needed for method development, which makes T3C very practical and convenient.1 Using a mixed stationary phase or tandem columns to vary selectivity is not at all a new concept in LC. Issaq and co-workers used a mixed reverse phase and β-cyclodextrin phase to separate antidepressants and anticonvulsants.4 Glajch and co-workers separated phenylthiohydantoin amino acid derivatives using coupled cyano and benzyl columns, and the selectivity was adjusted by varying the size of the two columns.5 McGuffin and co-workers used serially coupled columns and solvent modulation for the separation of isomeric polynuclear aromatic hydrocarbons.6 Recently, several groups optimized the selectivities for mixtures (3) Zhao, J. H.; Carr, P. W. Anal. Chem. 1999, 71, 2623-2632. (4) Issaq, H. J.; Mellini, D. W.; Beesley, T. E. J. Liquid Chromatogr. 1988, 11, 333-348. (5) Glajch, J. L.; Gluckman, J. C.; Charikofsky, J. G.; Minor, J. M.; Kirkland, J. J. J. Chromatogr. 1985, 318, 23-39. (6) Lukulay, P. H.; McGuffin, V. L. J. Chromatogr. A 1995, 691, 171-185. 10.1021/ac991435b CCC: $19.00
© 2000 American Chemical Society Published on Web 05/26/2000
containing chiral compounds by coupling a reversed and chiral stationary phases.7,8 They adjusted the selectivity by choosing different kinds of reversed phase columns. Although all these approaches provide significant control over varying the chromatographic selectivity, it is not convenient to prepare a mixed stationary phase or build to length expensive LC columns for specific separation problems. It is also not practical to vary selectivity by trying different combinations of LC columns. We believe T3C is novel and practical in that it uses temperature as a convenient variable to tune the selectivities. It is possible that the same T3C system containing two stationary phases with different selectivity can be used to solve different separation problems through the change in the two column temperatures. The objective of the present study is to evaluate the applicability of the T3C concept to some complex but tractable mixtures, extend the previously developed model to two-dimensional optimization, and further understand the advantages and limitations of T3C. EXPERIMENTAL SECTION Instruments. All chromatographic experiments were conducted with a Hewlett-Packard 1100 chromatography system, equipped with a quaternary pump, a vacuum degasser, an autosampler, a thermostated-column compartment, a variable wavelength UV detector, and a computer-based Chemstation (Hewlett-Packard S.A., Wilmington, DE). An additional experimental heating apparatus from Systec (Systec, Inc., Minneapolis, MN) was used as the second heating zone. This device consists of a mobile phase preheater assembly and insulating jacket. It allows the column to be heated to 200 °C. Analytical Columns. The carbon-coated zirconia (C-ZrO2) particles were provided by ZirChrom (ZirChrom Separation, Inc., Anoka, MN 55303). The C-ZrO2 particle has 1.6% carbon content, and the particle size is about 3 µm. A 5 cm × 4.6 mm i.d. C-ZrO2 column was packed in this lab using the upward stirred slurry method described in ref 9. Zorbax SB-C18 octadecylsilane (ODS) particles (5 µm) were purchased from Hewlett-Packard (Wilmington, DE). A 5 cm × 4.6 mm i.d. ODS column was packed using the method described in ref 10. Reagents. All reagents used here are reagent grade or better. ChromAR HPLC grade acetonitrile (ACN), methanol (MeOH), and tetrahydrofuran (THF) were purchased from Mallinckrodt Chemical Co. (Paris, KY). HPLC water was obtained from Barnsted Nanopure deionizing system (Dubuque, IA) with an “organic-free” cartridge followed by a 0.2 µm particle filter. The water was boiled to remove carbon dioxide. All solvents were filtered through a 0.45 µm filter (Lida Manufacturing Corp., Kenosha, WI) before use. Triazine herbicides were purchased from Supelco (Bellefonte, PA), and carbamate pesticides from AccuStandard (New Haven, CT). Other solutes used in this study were purchased from Aldrich (Aldrich Chemical Co., Milwaukee, WI). Chromatographic Conditions. All measurements were made at a flow rate of 1 mL/min unless otherwise stated. The injection (7) Lida, T.; Matsunaga, H.; Fukushima, T.; Santa, T.; Homma, H.; Imai, K. Anal. Chem. 1997, 69, 4463-4468. (8) Phinney, K. W.; Sander, L. C.; Wise, S. A. Anal. Chem. 1998, 70, 23312335. (9) Jackson, P. T.; Kim, T.-Y.; Carr, P. W. Anal. Chem. 1997, 69, 5011-17. (10) Tan, L. C.; Carr, P. W.; Abraham, M. H. J. Chromatogr. A 1996, 752, 1-18.
volume was 1 µL with a concentration of 1-2 mg/mL of the analyte dissolved in pure acetonitrile. Retention factors (k′) were calculated from the following equation:
k′ )
tr - t0 t0 - tex
(1)
Here tr is the retention time, t0 is the dead time, and tex is the time that a solute spends outside the column (in the injector, connection tubing, and detector). The dead time was determined by injecting uracil for the ODS and acetone for the C-ZrO2 column. The value of tex was determined to be 0.067 min by injecting uracil without any column under otherwise the same chromatographic conditions. The resolution is conveniently estimated on the basis of eq 2:
Rs )
xN tr2 - tr1 4 tr average
(2)
in which Rs is the resolution, tr average is the average of the retention times, and N is the theoretical plate number.11 N is assumed to be 5000 for a single 5 cm column and 10 000 for the two-column T3C set. We want to point out that, in this study, the mobile phase was prepared by weighing the proper amount of water and acetonitrile and premixed before use. We avoided the machinemixed mobile phase because we suspect that machine made mobile phase compositions might vary somewhat with a change in column backpressure. RESULTS AND DISCUSSIONS Environmental hazards such as herbicides and pesticides usually have high solubility in aqueous solution. The proper separation and analysis of these compounds has attracted great environmental attention. Conventional bonded phases often cannot provide enough resolution due to their limited selectivity toward polar solutes. Carbon-based sorbents, on the other hand, have been widely used in the petroleum industry and by chemical manufacturers to remove environmentally hazardous chemicals.12 We believe that a T3C system, which is comprised of an aliphatic and a carbon phase, should be useful for the separation of environmental toxicants. Initial Separations of Triazine Herbicides on ODS and C-ZrO2 Columns. Triazine herbicides are among of the most widely used herbicides in the United States. The persistent use of these herbicides has caused an increasing concern for their adverse effect on the environment. In this work, we optimize the separation of 10 triazine herbicides (see Figure 2 for their structures) using the T3C approach. Chromatograms of these triazines in a 30/70 acetonitrile/water mixture on an ODS column at 30 °C and on a C-ZrO2 column at 60 °C are shown in Figure 3. We choose these two stationary phases because ODS is the most widely used aliphatic RPLC material and C-ZrO2 exhibits unique selectivity toward polar compounds. As seen in Figure 3, neither phase could resolve the (11) Giddings, J. C. Unified Separation Science; Wiley-Interscience: New York, 1991. (12) Austin, G. T. In Chreve’s Chemical Process Industries, 5th ed.; McGraw-Hill: New York, 1984; Chapter 3.
Analytical Chemistry, Vol. 72, No. 13, July 1, 2000
2789
Figure 2. Structures of triazine herbicides.
Figure 3. Chromatograms showing the separation of triazines on (A) ODS at 30 °C and (B) C-ZrO2 at 60 °C. Experimental conditions: mobile phase, 30/70 acetonitrile/water; flow rate, 1 mL/min; detection, 254 nm. Solutes: 1, simazine; 2, cyanazine; 3, simetryn; 4, atrazine; 5, prometon; 6, ametryn; 7, propazine; 8, turbulazine; 9, prometryn; 10, terbutryn.
triazines under the indicated chromatographic conditions. The chromatogram with the ODS column (see Figure 3A) shows that the solute pairs simazine and cyanazine (1/2), simetryn and atrazine (3/4), and ametryn and propazine (6/7) are not baseline resolved. The triazines are much more retained on the carbon phase despite that lower carbon content of the C-ZrO2 phase (1.6%) as compared to the ODS phase (10%). This agrees with a linear solvation energy relationship (LSER) study of a C-ZrO2 phase.13 This work showed that C-ZrO2 phases are effectively much more hydrophobic than alkyl bonded phases. In addition, there is a significant electronic (π-π) interaction between the polar group on the solutes and the carbon surface. Both hydrophobic and π-π interactions probably contribute to the strong retention of triazines on C-ZrO2 column. We must use elevated temperature (60 °C) to elute the solutes from C-ZrO2 with k′ (13) Jackson, P. T.; Schure, M. R.; Weber, T. P.; Carr, P. W. Anal. Chem. 1997, 69, 416-25.
2790
Analytical Chemistry, Vol. 72, No. 13, July 1, 2000
values comparable to those on an ODS phase. The chromatogram on the C-ZrO2 column (see Figure 3B) indicates that three early eluting solutes, cyanazine, prometon, and propazine (2/5/7), elute closely, while significant peak tailing was observed for the latereluting peaks. As we will shown in detail later, the peak tailing was caused by the particular strong electronic interactions between the polar groups on the solutes and the carbon surface, and the peak shape was dramatically improved at higher temperature. Selectivity Comparison between ODS and C-ZrO2. Figure 3 shows that there are multiple changes in elution order when the ODS phase is replaced with the carbon phase. To quantitatively compare the selectivity difference between ODS and C-ZrO2 column, a κ-κ plot is shown in Figure 4A, in which the log k′ values of the triazines on C-ZrO2 at 60 °C were plotted versus the log k′ on ODS at 30 °C. κ-κ plots are commonly used to compare the energetics of retention on different bonded phases.14 A good linear correlation means the same or similar retention mechanism is operating on both columns, while a weak correlation indicates different retention mechanisms and different selectivities. Figure 4A shows that there is almost no correlation between the retention on ODS and that on C-ZrO2. The correlation coefficient is only 0.11, and the average residual is as high as 0.34. This confirms our previous observation that carbon-based stationary phases are in terms of selectivity very different from conventional bonded phases.1 A detailed comparison of the retention order for this set of triazines on ODS and C-ZrO2 columns is presented in Figure 4B, in which the retentions on the two phases are plotted against their log k′ values. Each crossover in this plot indicates a change in the elution order from one phase to another. Scrutiny of Figure 4B shows that, for this set of 10 triazines, 18 pairs of solutes changed their elution order when an ODS column was replaced by a C-ZrO2 column. More importantly, Figure 4B directly indicates the improvement for the critical pair (the worst separated pair) and the poorly separated pairs. We immediately observe that the difficult pairs on the ODS phase (pairs 1/2, 3/4, and 6/7) are well separated on the carbon phase. Similarly, solutes 2, 5, and 7 which elute closely on the carbon phase are well separated on the ODS column. (14) Melander, W.; Stoveken, J.; Horvath, C. J. Chromatogr. 1980, 199, 35-56.
Figure 5. Plot of log k′ for two triazine homologue series versus the number of substituted methyl groups on (A) ODS at 30 °C and (B) C-ZrO2 at 60 °C. The open circles represent the homologue triazines with an -SCH3 group (3, simetryn; 6, ametryn; 9, prometryn), and the filled circles represent homologue triazines with a -Cl group (1, simazine; 4, atrazine; 7, propazine).
Figure 4. (A) Plot of log k′ on C-ZrO2 at 60 °C versus log k′ on ODS at 30 °C for the triazines in 30/70 acetonitrile/water mixture. (B) Comparison of elution order (log k′) of the mixture on C-ZrO2 and ODS.
For serially connect columns, we have shown that the net selectivity (Rn) in T3C is related to the selectivities of the individual columns (R1 and R2) as follows:
Rn ) f1R1 + f2R2
(3)
Here f1 and f2 are the fractions of the total retention time on the first and second column, respectively.1 The log k′ data in the top and bottom row in Figure 4B represent the selectivity on ODS and C-ZrO2 columns, respectively; they define the extremes of the T3C selectivity. If we follow the lines in Figure 4B from ODS to C-ZrO2, a continuous change of T3C selectivity from one column to another is observed. For instance, Figure 4B shows that solute 1 and 2 closely elute on ODS. If we vary the temperature of the two columns and adjust the selectivity of T3C from ODS toward that of C-ZrO2, the selectivity factor first becomes worse, reaches a minimum (coelution) at a certain combination of the two columns, and then becomes better and reaches the best separation on the C-ZrO2 column. From eq 3 and Figure 4B, it is very easy to see that, for a particular pair of compounds, the selectivity factor provided by a tandem column set will always be less than that provided by the better of the two columns. This can be generalized to any combination of other chromatographic conditions, such as mobile phase composition. Thus if two columns have the same critical pair, T3C cannot improve the overall separation. Similarly, if a pair of compounds
is not baseline separated on either phase, it is not possible to achieve baseline separation by T3C. In order for T3C to work, we must have a sufficient difference in selectivities of the critical pair so that critical pairs on the two phases are different. To achieve baseline separation using T3C, every pair of solutes must be very well separated on at least one of the two phases. Figure 4B is important because it tells us immediately whether T3C will or will not work. Although carbon phases are known to be quite different from alkyl-bonded phases, they are still predominately reversed phase materials. It is rather surprising to see such massive changes in elution order for a set of triazines with so similar structures. To rationalize these large changes in selectivity from ODS to C-ZrO2, we compared the retention times of the two “homologue” series in the triazine sample which are series of solutes with the same functional group (Cl or SCH3) but with different numbers of methyl groups. Figure 5 is the plot of log k′ of the two “homologue” series versus the number of methyl group on the ODS and C-ZrO2, respectively. log k′ varies approximately linearly with the methyl group number for the two series on both phases. On the ODS column, the retention (log k′) increases linearly with the addition of a methyl group, which is characteristic of hydrophobic selectivity in reversed-phase mode. In contrast, on the carbon phase, the log k′s in both “homologue” series decrease linearly upon addition of a methyl groups. This means that retention decreases with an increasing in the solute’s hydrophobicity. This is not commonly observed for homologues on carbonbased phases. Mockel and co-workers studied the retention characteristics of porous graphite carbon (PGC) using a range of homologous alkane derivatives and found reasonably good correlations between log k′ and the number of methylene groups.15 Jackson and Carr compared the free energy of transfer for a methylene group on C-ZrO2 phase for some homologous alcohols. They also observed a linear increase in log k′ with increasing carbon number.9 We attribute the abnormal behavior observed here to the particularly strong electronic (π-π) interactions between the polar groups on the solutes and the carbon surface. Unlike conventional bonded phases with long alkyl chains which offer a partition-like retention mechanism,2,16,17 the carbon phases (15) Mockel, H. J.; Braedikow, A.; Melzer, H.; Aced, G. J. Liq. Chromatogr. 1991, 14, 2477.
Analytical Chemistry, Vol. 72, No. 13, July 1, 2000
2791
Table 1. Experimental Retention Factors of Triazine Herbicides on ODS and C-ZrO2 k′ (ODS)a
k′ (C-ZrO2)a
solutes
30 °C
80 °C
60 °C
130 °C
(1) simazine (2) cyanazine (3) simetryn (4) atrazine (5) prometon (6) ametryn (7) propazine (8) turbulazine (9) prometryn (10) terbutryn
4.67 5.12 9.08 9.84 13.9 19.4 20.6 27.2 41.8 51.6
3.44 3.14 6.22 6.77 9.67 12.3 13.2 15.2 24.2 26.9
12.5 5.91 42.7 8.45 5.93 27.0 5.80 10.4 17.9 31.7
2.75 1.49 6.59 2.04 1.39 4.65 1.52 2.50 3.33 5.39
a Measured in 30/70 acetonitrile/water mixture; flow rate, 1 mL/ min; detection, 254 nm.
Figure 7. Chromatograms showing the separation of a triazine mixture on T3C with the ODS column at 30 °C and the C-ZrO2 column at 125 °C with a flow rate of (A) 1 mL/min and (B) 3 mL/min. Experimental conditions: 30/70 acetonitrile/water mixture; detection, 254 nm. Solutes: 1, simazine; 2, cyanazine; 3, simetryn; 4, atrazine; 5, prometon; 6, ametryn; 7, propazine; 8, turbulazine; 9, prometryn; 10, terbutryn.
Figure 6. Window diagram showing the resolution of the critical pair versus the temperature of the ODS and C-ZrO2 columns.
are considered locally flat and provide an adsorption-like retention mechanism.9,13,18 The closer the solute’s polar group gets to the carbon surface, the stronger are the π-π interactions.18 This leads to the superior shape selectivity of the carbon phase toward structural isomers. Tanaka and co-workers found that the stereoselectivity of carbon phases for cis- and trans-isomers comes from difference in the planarity of the two isomers. The more planar isomer was more retained because planar molecules can most easily be accommodated to the “flat” surface of the carbon phase.19 Jackson also found extremely long retention and the absence of elution of polyaromatic hydrocarbons (PAHs) on C-ZrO2 phases due to the presence of the planar aromatic rings.20 The threedimensional structure of simazine shows that the molecule is rather flat. Addition of the methyl group to the triazine ring perturbs the planarity of the molecule. This inhibits the polar groups on the molecule from getting close to the stationary phase (16) Tan, L. C.; Carr, P. W. J. Chromatogr. A 1997, 775, 1-12. (17) Carr, P. W.; Li, J.; Dallas, A. J.; Eikens, D. I.; Tan, L. C. J. Chromatogr. A 1993, 656, 113-133. (18) Knox, J. H.; Ross, P. Adv. Chromatogr. 1997, 37, 73-119. (19) Tanaka, N.; Tanigawa, T.; Kimata, K.; Hosoya, K.; Araki, T. J. Chromatogr. 1991, 549, 29. (20) Jackson, P. Ph.D. Thesis, University of Minnesota, Minneapolis, MN, 1997.
2792
Analytical Chemistry, Vol. 72, No. 13, July 1, 2000
and leads to the decrease in retention. Figure 5 suggests that, for triazines, the interactions between the polar functional groups of the analyte and the carbon surface is the primary driving force for retention on C-ZrO2. It is the difference in retention mechanisms on ODS and C-ZrO2 that causes the dramatic change in elution order and selectivities. Selectivity Tuning in T3C. In T3C, selectivity can be adjusted by simultaneously varying both column temperatures. The optimization of T3C involves locating the temperatures on both the columns which give the best overall resolution. To implement selectivity tuning, it is necessary to determine the effect of temperature on the retention of both columns in the T3C set. First, we need to determine the effect of temperature on the retention for each solute on each column. We performed additional runs at higher temperatures, 80 °C on the ODS column and 130 °C on the carbon column. Table 1 summarizes all the initial k′ data that were used for T3C optimization. The relationship between the logarithm of the retention factor and the reciprocal temperature is linear in reversed-phase chromatography over a narrow temperature interval and can be described as follows:
log k′ ) A + B/T
(4)
Here A and B are solute-dependent constants. Fitting the retention data at low and high temperatures on each column to eq 4 allows us to estimate solute retention at any other temperature. Next, the overall T3C retention time for each solute at any specific temperatures (T1 and T2) was calculated as
tn,i ) t1,i + t2,i - tex
(5)
where tn,i is the net retention time on the T3C set, t1,i and t2,i are the retention times for solute i on the first and second columns, respectively, and tex is the time that a solute spends outside the
Figure 8. Chromatograms showing the separation of triazine mixture on an ODS column at (A) 30/70 THF/water, (B) 40/60 ACN/ water, and (C) 50/50 MeOH/water mixture. Experimental conditions: temperature, 30 °C; flow rate, 1 mL/min; detection, 254 nm. Solutes: 1, simazine; 2, cyanazine; 3, simetryn; 4, atrazine; 5, prometon; 6, ametryn; 7, propazine; 8, turbulazine; 9, prometryn; 10, terbutryn. Table 2. Experimental Retention Factors of Triazine Solutes on ODS Columns acetonitrile/watera methanol/watera
THF/watera
solutes
30/70
40/60
40/60
50/50
20/80 30/70
(1) simazine (2) cyanazine (3) simetryn (4) atrazine (5) prometon (6) ametryn (7) propazine (8) turbulazine (9) prometryn (10) terbutryn
4.67 5.12 9.08 9.84 13.9 19.4 20.6 27.2 41.8 51.6
2.23 2.23 3.82 4.15 5.05 7.10 7.63 9.13 13.2 15.0
8.06 6.27 18.8 17.7 38.8 41.0 38.6 50.8 90.6 117
3.44 2.39 6.99 6.72 12.7 13.3 13.1 15.7 26.8 31.9
10.6 13.9 17.3 24.4 17.6 40.2 55.4 67.4 91.8 104
a
4.46 5.12 6.13 8.64 5.76 11.7 16.2 17.0 22.1 21.5
Flow rate, 1 mL/min; detection, 254 nm.
column (in the injector, connection tubing, and detector). This equation has been described in detail previously.1 Finally, the optimization is based on calculating the resolution of the critical pair at each combination of the two column temperatures. The plot of the resolutions for the critical pair against the two temperatures on ODS and C-ZrO2 columns, respectively yields the two-dimensional window diagram shown in Figure 6. Given the fact that resolution larger than 1.5 means baseline separation, Figure 6 shows that many combinations of the temperatures in T3C will give quite adequate separations. The highest point in the window diagram corresponds to 30 °C on
Figure 9. Plot of minimum resolution versus the percentage organic modifier of (A) acetonitrile, (B) methanol, and (c) tetrahydrofuran in the mobile phase.
ODS and 125 °C on the carbon phase where the resolution is about 3. The chromatogram at the optimized conditions is shown in Figure 7A. All the solutes are well resolved in less than 30 min. We compared the observed retention times with the estimated values based on only the four initial runs, and the relative errors in retention times were all less than 1.2%. Comparison of the separation on T3C with those on the single phases (Figure 3) indicates that the overall separation was dramatically improved without a substantial increase in analysis time. If we had tried to improve the separation on the ODS column alone by using a longer column or by lowering the amount of organic modifier, a much longer analysis time would have resulted. Analysis time could be shortened by running at a flow rate of 3 mL/min at the optimized temperatures (see Figure 7B). The 10 triazines are baseline separated in less than 10 min. This is only possible because of the superior selectivity provided by the T3C, which allows us to sacrifice the plate count by using the high flow rate and yet maintain the baseline separation. The combination of the superior selectivity of the T3C set of column and high flow rate allowed the 10 triazines to be baseline separated in less than 10 min based on only four initial chromatographic runs. Comparison of T3C Separation with Mobile Phase Optimization. Varying the mobile phase type and composition is still the most popular separation optimization method. It is of interest to compare the separation optimized by T3C with optimizations by mobile phase adjustment. The mobile phase was optimized using ODS as the stationary phase because it is the most widely used RPLC material. We choose the three most common types of mobile phase modifiers, acetonitrile (ACN), methanol (MeOH), Analytical Chemistry, Vol. 72, No. 13, July 1, 2000
2793
Figure 10. Structures of urea and carbamate pesticides.
and tetrahydrofuran (THF). The initial chromatograms are shown in Figure 8. Although different elution orders were observed upon changing modifier, none of the conditions gave adequate separation. To further optimize the separation by varying B%, we did one additional run at a different mobile phase composition for each of the modifiers. Table 2 summarizes the retention factors used in the mobile phase optimization calculations. In RPLC, the log k′ is quasi-linearly related to the volume fraction (φ) of organic modifier by the linear solvent strength approximation:
log k′ ) log k0′ -Sφ
(6)
in which log k0′ and S are solute-dependent constants. Therefore, for each solute and each mobile phase type, two retentions at two different φ allow the estimation of retention at any other mobile phase composition and estimation of resolution. The resulting window diagrams are shown in Figure 9. For ACN, the highest achievable resolution is 1.0, which means that baseline resolution cannot be obtained, Even poorer separations are predicted for MeOH and THF, the best separation being 0.4 and 0.7, respectively. Although the window diagram only covers 20-40% of the whole mobile phase composition, we point out that the retention times will be extremely short or long outside this range. Therefore, complete separation cannot be achieved in a reasonable period of time by varying only the type and volumetric fraction of the mobile phase. This specific example demonstrates that T3C can be more powerful than mobile phase optimization. Although use of a ternary mobile phase might give a better separation than a binary mobile phase, the point is that T3C is a convenient optimization method because it can dramatically alter the selectivity and uses only four initial runs. Optimization of the Separation of a Mixture of Urea and Carbamate Pesticides. To investigate whether T3C can handle more diverse and highly polar analytes, we challenged it with another set of environmentally significant compounds. Urea and carbamate compounds belong to a heterogeneous group of 2794
Analytical Chemistry, Vol. 72, No. 13, July 1, 2000
Figure 11. Chromatograms showing the separation of carbamate mixture on (A) ODS at 30 °C and (B) C-ZrO2 at 90 °C. Experimental conditions: mobile phase, 40/60 acetonitrile/water; flow rate, 1 mL/ min; detection, 220 nm. Solutes: 1, oxamyl; 2, methomyl; 3, fenuron; 4, carbofuran; 5, fluometuron; 6, monuron; 7, methiocarb; 8, chlorpropham; 9, carbaryl; 10, diuron; 11, linuron; 12, swep; /, impurity.
synthetic pesticides that haven been used on a large scale during the past half century. Carbamate pesticides are used as fungicides, herbicides, and insecticides. Some insecticides such as carbofuran, methomyl, oxamyl, and carbaryl have high acute toxicity.21 Figure 10 shows the structure of the 12 carbamate pesticides used in this study. The separation of the mixture on a single column is shown in Figure 11. The chromatogram in Figure 11A shows the separation on an ODS column at 40/60 acetonitrile/water mixture and 30 °C. Coelutions of the solute pairs fluometuron and carbaryl (21) Engebretson, J. A.; Mourer, C. R.; Shibamoto, T. In Chromatographic Analysis of Environmental and Food Toxicants; Shibamoto, T., Ed.; Marcel Dekker Inc.: New York, 1998; Vol. 77, pp 259-289.
Figure 13. Window diagram showing the resolution of the critical pair versus the temperatures of the ODS and C-ZrO2 columns.
Figure 12. (A) Plot of log k′ on C-ZrO2 at 90 °C versus log k′ on ODS at 30 °C for the carbamate pesticides in 40/60 acetonitrile/water mixture. (B) Comparison of elution order (log k′) of the mixture on C-ZrO2 and ODS. Solutes: 1, oxamyl; 2, methomyl; 3, fenuron; 4, carbofuran; 5, fluometuron; 6, monuron; 7, methiocarb; 8, chlorpropham; 9, carbaryl; 10, diuron; 11, linuron; 12, swep. Table 3. Experimental Retention Factors of Urea and Carbamate Pesticides on ODS and C-ZrO2 k′ (ODS)a
k′ (C-ZrO2)a
solutes
30 °C
80 °C
90 °C
120 °C
(1) oxamyl (2) methomyl (3) fenuron (4) carbofuran (5) fluometuron (6) monuron (7) methiocarb (8) chlorpropham (9) carbaryl (10) diuron (11) linuron (12) swep
0.33 0.54 0.95 3.36 3.97 2.16 7.56 17.23 4.10 4.66 10.69 10.69
0.28 0.41 0.75 2.21 2.55 1.47 3.77 8.12 2.37 2.86 5.49 5.10
0.07 0.34 1.33 0.98 3.17 5.60 4.98 7.07 16.7 22.8 22.8 26.6
0.06 0.24 0.90 0.69 1.78 3.27 3.27 3.62 9.32 11.3 11.2 12.1
a Measured in 40/60 acetonitrile/water mixture; flow rate, 1 mL/ min; detection, 220 nm.
(5/9) and linuron and swep (11/12) are observed on the ODS column. Due to the polar group in the carbamate pesticides, the solutes are much more retained on the carbon phase. At low temperature on the C-ZrO2 column, extremely long retention time and tailed peaks were observed. The separation at 90 °C is shown in Figure 11B. The separations are generally good with decent peak shapes although diuron and linuron (10/11) coelute.
Figure 14. Chromatograms showing the separation of carbamate mixture on T3C with ODS column at 39 °C and C-ZrO2 column at 89 °C with the flow rate of (A) 1 mL/min and (B) 3 mL/min. Experimental conditions: 40/60 acetonitrile/water mixture; detection, 220 nm. Solutes: 1, oxamyl; 2, methomyl; 3, fenuron; 4, carbofuran; 5, fluometuron; 6, monuron; 7, methiocarb; 8, chlorpropham; 9, carbaryl; 10, diuron; 11, linuron; 12, swep; /, impurity.
A κ-κ plot, which compares the selectivity for the pesticides, is shown in Figure 12A. Substantial deviations from the regression line (r2 ) 0.716, sd ) 0.453) indicate large changes in selectivity from one phase to the other. In Figure 12B, we further compare the elution order on the two phases and observe significant changes in elution orders. Moreover, this plot shows that all the coeluted pairs on one phase are well separated at the other phase. This indicates that use of T3C will improve the separations from those observed on a single phase. To optimize the separation using the tandem column, we performed one additional run on both columns (80 °C for the ODS column and 120 °C for the C-ZrO2 column). The retention factors are summarized in Table 3. The average retention factors decrease 1.65- and 1.71-fold, respectively, for a 50 °C increase on the ODS column and 40 °C increase on the carbon column. Using a strategy Analytical Chemistry, Vol. 72, No. 13, July 1, 2000
2795
similar to that described above, we constructed the window diagram as shown in Figure 13. The highest point in the window corresponds to 39 °C on the ODS and 89 °C on the carbon column, and the predicted minimum resolution is 2.8. Figure 14A shows the chromatogram under the optimized conditions; all the components are well separated, and retention times agree well with the calculated ones (relative errors are all less than 1.7%). Furthermore, we found that the separation time could be significantly decreased to 8 min by using a flow rate of 3 mL/min. This example shows again that T3C is extremely powerful in separating complex mixtures. Although both examples showed here involved polar compounds, our ongoing research shows that, even for nonpolar compounds, a C-ZrO2 phase has significantly different selectivity compared to the conventional bonded phase, and selectivity can be improved when the critical pairs on the two phases are different. CONCLUSIONS The chief conclusions of this study are as follows: (1) The T3C combination of an aliphatic and a carbon phase is an extremely powerful technique in terms of its ability to tune selectivity. (2) As long as the critical pairs on the two phases are different, use of the thermally tuned tandem column approach will give a separation at the optimum temperatures superior to that on either
2796
Analytical Chemistry, Vol. 72, No. 13, July 1, 2000
column alone in the same eluent. (3) The two examples presented here show that T3C can dramatically improve the selectivity and give baseline separations even though the mixtures could not be separated on either individual phase. We also showed, in the case of the triazine herbicides, that T3C is more powerful than manipulating the mobile phase type and composition. (4) Because of the high selectivity often provided by T3C, analysis time can sometimes be dramatically shortened by using a high flow rate while still maintaining adequate separation. (5) For method development, only four initial experiments are required for locating the optimum temperatures. The estimated retention and selectivity agree well with the experimental ones. We believe that T3C provides a convenient and powerful new approach to method development in liquid chromatography. ACKNOWLEDGMENT The authors acknowledge the financial support from the National Institutes of Health.
Received for review December 15, 1999. Accepted March 29, 2000. AC991435B
