In Silico Modeling of Hundred Thousand Experiments for Effective

Jan 17, 2016 - In Silico Modeling of Hundred Thousand Experiments for Effective Selection of Ionic Liquid Phase Combinations in Comprehensive Two-Dime...
1 downloads 9 Views 1MB Size
Download PDF
Subscriber access provided by NEW YORK UNIV

Article

In silico modelling of hundred thousand experiments for effective selection of ionic liquid phase combinations in comprehensive two-dimensional gas chromatography Yada Nolvachai, Chadin Kulsing, and Philip John Marriott Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.5b03688 • Publication Date (Web): 17 Jan 2016 Downloaded from http://pubs.acs.org on January 24, 2016

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Analytical Chemistry is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 8

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

In silico modelling of hundred thousand experiments for effective selection of ionic liquid phase combinations in comprehensive twodimensional gas chromatography Yada Nolvachai, Chadin Kulsing, and Philip J. Marriott* Australian Centre for Research on Separation Science, School of Chemistry, Monash University, Wellington Road, Clayton, VIC 3800, Australia. ABSTRACT: Selection of best column sets is one of the most tedious processes in comprehensive two dimensional gas chromatography (GC×GC) where a multitude of choices of column sets could be employed for an individual sample analysis. We demonstrate analyte/stationary phase dependent selection approaches based on the modified linear solvation energy relationship (LSER), which is a reliable concept for study of interaction mechanisms and retention prediction with a large database pool of columns and compounds. Good correlations between our predicted results, with experimental results reported in the literature, were obtained. The developed approaches were applied to simulation of 157,920 individual experiments in GC×GC, focusing on application of 30 non-ionic liquid and 111 ionic liquid (IL) stationary phases for separation of some example sets of model compounds present in practical samples. Best column sets for each sample separation could then be extracted according to maximizing orthogonality, which estimates the quality of separation.

The ability to identify and quantify compounds in complex matrices is paramount for many industrial applications including pharmaceutical and illicit drugs, petroleum, food, beverages and environmental. Analysis of analytes at low levels presents especial difficulties, particularly in multi-component samples. GC×GC is one of the most powerful techniques to obtain molecular-based information of complex samples.1-3 This technique employs two columns providing different selectivity towards sample components connected sequentially with a device offering a modulation process. Compared to 1DGC, GC×GC is superior in terms of enhanced resolution and analyte peak capacity, and improved detection limit.1,4 Increasing use of GC×GC is noted with >100 papers per year, and >1,000 total publications published thus far (SciFinder database). Conventional poly(siloxane) and poly(ethyleneglycol) stationary phases employed in GC provide separation mechanisms mainly based on vapor pressure and polarity difference of analytes. In GC×GC, this might not maximize peak capacity or resolve target analyte peaks. Even with the selectivity of mass spectrometry (MS), some samples cannot be adequately analyzed if separation is insufficient.4 This inherent limitation can be overcome by introducing stationary phases with alternative selectivity. Often, specific phases are employed for particular applications, such as high cyano-content phases being preferred for fatty acid methyl ester (FAME) analysis due to better retention towards unsaturated groups;5 high phenyl-content phases are usually preferred for analysis of mineral oils which have high aromatic solute content. Beside application in liquid phase extraction procedures,6,7 more than one hundred IL have been synthesized for use in GC8-11 improving tuneability of peak distribution in separation due to their desirable properties such as customizable molecular structures

(providing tuneable selectivity in separation) and high thermal stability12-14 as well as good thermal and surface sensitivity where a single IL phase can result in good separation in GC×GC.15 Phosphonium- and imidazolium-based IL columns have been commercialized,9,14,16 and offer improved selectivity in separation of many analytes, such as FAME, polychlorinated biphenyls (PCB), pesticides and sulfur or nitrogen containing compounds in different sample matrices, as compared with other phases.10,17,18 The increasing trend of IL column synthesis over the years expands the number of GC phases available in the ‘world library’. This has an analogy in the expansion of solvent types to be employed in general extraction methods, with a subsequent need to establish specific useful applications.19 Thus, understanding their use and achieving maximum performance for unique selectivity of these phases is as important as making the new phase, especially in GC×GC where the selection of column sets critically affects orthogonality (O, overall separation performance). However, random column selection arising from lack of insight into separation mechanisms may prove inefficient, and testing novel column phases that may not improve separation results over conventional nonpolar/polar column sets is time-wasteful. There is clearly a need for separation mechanism-directed GC phase selection to inform the most effective use of 2D space in the GC×GC experiment. Generally, it is necessary to perform a number of trial and error experiments during the optimization process related to the column selection and separation condition for the best separation result. However, there is no guarantee that a truly best column set has been selected. At least 150 phases have been proposed for capillary column GC analysis. Even with a single analytical run for a sample – without temperature (T)

ACS Paragon Plus Environment

Analytical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 8

program or flow rate optimization, the possible number of experiments that incorporates all column couples used in GC×GC could be >2,000,000 experiments a year. The amount of time, energy and consumables cost, wear-and-tear, and potentially obsolete column set combinations involved in deciding the best column set is highly impractical. The process to achieve the “best” result is thus critical and if based purely on an experimental trial-and-error basis, appears to be impossible. Among various approaches, reliable simulation of an experimental result in order to reduce the reliance on rule-of-thumb or trial-and-error experimentation is an effective way to achieve a more effective outcome. This study thus develops a new approach to inform column selection in GC×GC for a given set of solutes and stationary phases based on linear solvation energy relationship (LSER) which is a well-established concept with a large database for retention prediction,13 understanding of separation mechanisms or characterization of a wide range of stationary phases in solid phase extraction,20 liquid chromatography,21,22 micellar electrokinetic chromatography,23 and GC.10,24 The examples for illustration of the established approach are then given for separation of some selected sets of analytes observed in practical samples.

phases and analytes, respectively. ki,j values were converted to normalized retention times (tR,normalized) which is expressed as

EXPERIMENTAL SECTION Simulations in this study were compared to normalized experimental data, obtained from previously reported experimental GC×GC results.12,25-27 The comparison focused on separations of 254 hydrocarbons in kerosene, 209 PCB congeners, 59 compounds in gasoline, and 63 compounds with different functionalities, all employing different column sets mostly comprising at least one IL columns (Supporting Information, Table S1–S2). A Microsoft Visual Basic script was developed herein to manually select (mouse click) the position (pixel) of each peak in the .jpg format reference 2D chromatograms, with correction of wraparound. The coordinates of all peak positions in each reference 2D result were automatically recorded and converted into the 1D and 2D normalized retentions (1tR,normalized; 2tR,normalized) for all detected analytes.

where k,min and k,max are k values of the earliest and latest, respectively, eluted analytes in the separation. The calculated tR,normalized values were then used to calculate orthogonality (O). O is a scale describing analyte peak distribution in 2D chromatograms. Recent progress on the description of peak distribution and evaluation of orthogonality in GC×GC has been reported.40,41 O can be obtained by computing the ratio of area containing analytes in a normalized 2D chromatogram to the statistically maximum area covered by the analyte peaks.41,42 Correlation 1tR,normalized and 2tR,normalized can then be incorporated into the calculation of O for more accurate description of peak distribution.41 Alternatively in order to reduce computational time, a simple and reliable algorithm was herein selected for the calculation of orthogonality which is obtained by using asterisk equations as recently introduced.43 The orthogonality in this approach is defined on an Ao scale (ranging from 0 to 1) capturing peak distribution around four main axes in 2D space in the directions of x-axis, y-axis and the other two diagonal lines.43 Perfectly orthogonal separation (expected to provide the best overall separation) corresponds to Ao = 1. The Microsoft Visual Basic script was developed for the prediction of Ao values (Ao,pred) for a given set of analytes and stationary phases according to their different LSER descriptors.

Simulation of GC× ×GC results according to the LSER at 100 °C. One hundred and forty-one stationary phases are focused on here. Without requiring experimental investigation, the approach to obtain the best column couples within the 19,740 possible 2D column sets (141!/[2!(141– 2)!]×2) for each sample separation was established from the LSER database of all the columns at 100 °C, and based on calculated O values (see Supporting Information, Table S1 for the compiled LSER database of all columns).10,12,28-38 Briefly, retention factors (k) of all analytes on a particular column are calculated as log ki,j(T) = ei(T)Ej + si(T)Sj + ai(T)Aj + bi(T)Bj + li(T)Lj + ci(T) (1)

where c is an intercept constant. e, s, a, b and l (stationary phase descriptors) represent the stationary phase contributions to the interactions being dispersity, dipolarity, H-bond with acid functionalities, H-bond with basic functionalities and dispersion/cavity formation for gas to liquid phase, respectively. The corresponding analyte descriptors are E, S, A, B and L which can be calculated (see Supporting Information, Table S2 for all analyte descriptors used in this study).39 The subscript i and j are indices for the simulated set of stationary

, =

 ,

(2)

,,

where tR is analyte retention time, tR,min and tR,max are retention times of the earliest and latest, respectively, eluted analytes in the separation. In order to obtain the relationship be   tween tR,normalized and analyte retention (k =   ,where t0 is void time), eq 2 can be transformed into , =



 −  − ,  +  , −  − ,  + 

 −   − ,  −  !  , =    , −  ! − ,  −  ! , =

" #" ! $", #" %  " "

", #" ! $", #" %  " "

=

&&'()

&'*+ &'()

(3)

RESULTS AND DISCUSSION The developed approach was evaluated by comparison with experimental data reported elsewhere.12,25-27 Ao values were then calculated for each experiment (Ao,exp). Ao,pred values calculated from the predicted retention in isothermal separation at 100 °C (k,pred,100°C,L) by using eq 1 were much less than the experimental values in temperature programmed separation as observed with the slope of the linear relationship being less than 1, Figure 1A. Isothermal separation at low temperature (100 °C) often leads to higher boiling point analytes eluting later in both 1D and 2D separation. This behavior forces the analyte peak distribution to be along the diagonal line in 2D plots, reducing orthogonality.

ACS Paragon Plus Environment

Page 3 of 8

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Figure 1. Correlation between Ao,exp and Ao,pred calculated from (A) k,pred,100°C,L ( ), (B) k,pred,100°C,V ( ) and (C) Ipred ( ) values with some snapshot of experimental ( ) vs predicted ( ) GC×GC results shown on the right for the separation of 209 PCB congeners (i), 59 hydrocarbons in gasoline (ii) and 63 model analytes (iii) using SPB-Octyl (30 m × 0.25 mm I.D. × 0.25 µm df) × SLB-IL59 (1.8 m × 0.10 mm I.D. × 0.24 µm df), Petrocol (100 m × 0.25 mm I.D. × 0.5 µm df) × SLB-IL59 (1.8 m × 0.10 mm I.D. × 0.24 µm df), trihexyl(tetradecyl)phosphonium bis(trifluoromethylsulfonyl)imide ([P66614][NTf2]) (30 m × 0.25 mm I.D. × 0.15 µm df) × HP-5 (5 m × 0.25 mm I.D. × 0.25 µm df) respectively.25,27,44 The average errors (differences between predicted and experimental values) for the orthogonality values in A-C were 0.35±0.17, 0.23±0.09 and 0.07±0.04, respectively. The average errors for 1tR,normalized and 2tR,normalized for the chromatograms shown on the right are provided in Table 1.

On the other hand, since all analyte retentions decrease at higher temperature distorting the analyte distribution from the diagonal line, peak distribution is experimentally observed to be better in an optimized temperature programmed separation. The prediction of retention at 100 °C (k,pred,100°C,V) from the descriptor V (cavity formation/dispersion for liquid to liquid phase), instead of L, can also be performed in order to reduce the vapor pressure influence on the k calculation. 1D and 2D k of all analytes were calculated for the separation employing all the 2D column sets according to the LSER concept as -./ 01,2 3 = 41 352 + 61 372 + 81 392 + :1 3;2 + 1 3 (4) where v is stationary phase descriptors which represent the contributions of the stationary phase to the interactions related to effects of dispersion/cavity formation for liquid to liquid phase, respectively. Ao,pred values calculated from the k,pred,100°C,L by using eq 1 were much less than the Ao,pred in T programmed separation as observed with the correlation slope being 30 analytes,28 which can be analysed in a single run, e.g. with GC hyphenated with mass spectrometry, are used as the representation of any analytes to be separated. #min,exp,LSER is calculated as #€1},Bq@,DOM = ‚#rƒT„€} u # L @BA rƒT„€} …Y†st + |P}PT~N1N u

!stust

(11)

tR,normalized

(iii) (ii) 1C

tR,normalized

2

Page 6 of 8

z u 2 u #Bq@ @BA NB u (10)

where #exp,direct GC×GC, #column and #exp per set are the total number of experiments in direct GC×GC approach required to obtain the best column set with the best experimental condition, the number of columns available in the lab and the number of experiments per a column set used in direct GC×GC. ranalysis and y are analysis rate (sample per year) and the number of years in the future, respectively. The factorial term in bracket represents the number of possible column sets to be employed in GC×GC and the factor 2 (outside the bracket) represents the inverse configuration of each column set. There is a need for fundamental experiments in order to allow a successful time dependent LSER simulation. The total number of experiments required for the simulation to obtain the best column set with the best experimental condition (including the confirmation of the best GC×GC experiment for each sample) based on LSER approaches lies between #min,exp,LSER and #max,exp,LSER values, where #min,exp,LSER and

where #T per column is the number of temperatures investigated in constant temperature 1DGC. The left terms subscribed as 1DGC in eq 11 represented the total numbers of 1DGC experiments required for the simulation; whilst, the right term |P}PT~N1N u ! represented the number of GC×GC exstust periments required to perform a real separation in publications.

Figure 4. The required number of 1DGC experiments for the T dependent LSER simulation for newly developed phases. (A), and minimum reduced number of GC×GC experiments with the different number of available columns within the next years (B). (eq 10–12 for the calculation)

For more accurate LSER simulation approach, a specific simulation approach when each specific set of standards being the representation for each set of analytes to be separated is applied, e.g. the use of FAME standards for the calculation of phase descriptors which was then used for simulation of FAME separation.15 This approach requires no more than #max,exp,LSER with the addition of the term |P}PT~N1N u !, taking into account the case when the number of specific sets of standards used is equal to the number of analysis in the next y years, which can be calculated as #€Pq,Bq@,DOM = ‚ #rƒT„€} u # L @BA rƒT„€} ! u |P}PT~N1N u !…Y†st + |P}PT~N1N u ! (12) stust

The minimum reduced number of GC×GC experiments obtained by using T dependent LSER approach (#exp,direct GC×GC #max,exp,LSER) was then calculated and plotted as functions of the number of available columns and the period of application of this approach. The number required for long term optimization in trial-and-error GC×GC is much more than the number of experiments required in the T dependent LSER approach (Figure 4A) resulting in a reduction in the ‘million experiments’ in the future, as shown in Figure 4B. In fact, providing that the database is available and the newly synthetic columns are test-

ACS Paragon Plus Environment

Page 7 of 8

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

ed isothermally in 1DGC at different T, no further experiment is required for prediction using this approach. CONCLUSIONS A new approach integrating a large LSER database for column selection allowing prediction of GC×GC separations of all manner of samples was demonstrated. The developed script allows the input of any number of columns and compounds to be separated in order to match either the best column sets worldwide, or those that might be available in a research group. Such capability is especially useful when hyphenation with MS is applied with extracted mass analysis, to selectively detect particular compound classes, which results in the best column sets depending on the extracted mass values for a particular sample. Good correlation with experimental results reveals crucial evidence of the usefulness of the approach with the potential to reduce millions of GC×GC experiments to a manageable experimental task in the future. The exact match of a predicted and an experimental absolute 2D peak position is not the most crucial requirement here – although would be desirable – but most important is to predict a robust column selection that is the best (providing that the maximum limit of operating T is appropriate) in order then to install such a column set in the knowledge that this should be the one that offers best separation performance. Then a straightforward optimization and refinement of conditions will produce a best separation for the target analysis.

ASSOCIATED CONTENT Supporting Information The Supporting Information is available free of charge on the ACS Publications website. List of LSER descriptors for stationary phases and analytes; detailed calculation and computational method (PDF) Complete list of column combinations for all analyte sets (XLSX)

AUTHOR INFORMATION Corresponding Author * Email: [email protected]

Notes The authors declare no competing financial interests.

ACKNOWLEDGMENT The authors acknowledge Prof. Paul Haddad, University of Tasmania (ACROSS) for highlighting the importance of best column selection. YN acknowledges Monash Institute of Graduate Research, School of Chemistry and Faculty of Science for scholarship support. We acknowledge funding from ARC Discovery and Linkage program grants DP130100217 and LP130100048. PM acknowledges ARC funding for a Discovery Outstanding Researcher Award.

REFERENCES (1) Marriott, P. J.; Chin, S. T.; Maikhunthod, B.; Schmarr, H. G.; Bieri, S. TrAC, Trends Anal. Chem. 2012, 34, 1–20. (2) Chin, S.-T.; Marriott, P. J. Chem. Commun. 2014, 50, 8819–8833. (3) Mitrevski, B.; Marriott, P. J. Anal. Chem. 2012, 84, 4837– 4843.

(4) Nolvachai, Y.; Kulsing, C.; Marriott, P. J. Crit. Rev. Environ. Sci. Technol. 2015, 45, 2135–2173. (5) Harynuk, J.; Wynne, P. M.; Marriott, P. J. Chromatographia 2006, 63, S61–S66. (6) Mokhtar, S. U.; Chin, S. T.; Vijayaraghavan, R.; MacFarlane, D. R.; Drummer, O. H.; Marriott, P. J. Green Chem. 2015, 17, 573–581. (7) Aguilera-Herrador, E.; Lucena, R.; Cárdenas, S.; Valcárcel, M. Anal. Chem. 2008, 80, 793–800. (8) Poole, C. F.; Poole, S. K. J. Sep. Sci. 2011, 34, 888–900. (9) Armstrong, D. W.; He, L.; Liu, Y.-S. Anal. Chem. 1999, 71, 3873–3876. (10) Poole, C. F.; Lenca, N. J. Chromatogr. A 2014, 1357, 87– 109. (11) Ho, T. D.; Zhang, C.; Hantao, L. W.; Anderson, J. L. Anal. Chem. 2014, 86, 262–285. (12) Hantao, L. W.; Najafi, A.; Zhang, C.; Augusto, F.; Anderson, J. L. Anal. Chem. 2014, 86, 3717–3721. (13) Kulsing, C.; Nolvachai, Y.; Zeng, A. X.; Chin, S.-T.; Mitrevski, B.; Marriott, P. J. ChemPlusChem 2014, 79, 790–797. (14) Anderson, J. L.; Ding, R.; Ellern, A.; Armstrong, D. W. J. Am. Chem. Soc. 2005, 127, 593–604. (15) Nolvachai, Y.; Kulsing, C.; Marriott, P. J. Anal. Chem. 2015, 87, 538–544. (16) Breitbach, Z. S.; Armstrong, D. W. Anal. Bioanal. Chem. 2008, 390, 1605–1617. (17) Zeng, A. X.; Chin, S. T.; Nolvachai, Y.; Kulsing, C.; Sidisky, L. M.; Marriott, P. J. Anal. Chim. Acta 2013, 803, 166– 173. (18) Kulsing, C.; Nolvachai, Y.; Hügel, H. M.; Marriott, P. J. LCGC Europe 2015, 28, 434–440. (19) Lapkin, A. A.; Peters, M.; Greiner, L.; Chemat, S.; Leonhard, K.; Liauw, M. A.; Leitner, W. Green Chem. 2010, 12, 241–251. (20) Nolvachai, Y.; Kulsing, C.; Boysen, R. I.; Matyska, M. T.; Pesek, J. J.; Marriott, P. J.; Hearn, M. T. W. Food Chem. 2014, 174, 434–439. (21) Zhao, J.; Carr, P. W. Anal. Chem. 1998, 70, 3619–3628. (22) Kulsing, C.; Yang, Y.; Matyska, M. T.; Pesek, J. J.; Boysen, R. I.; Hearn, M. T. W. Anal. Chim. Acta 2015, 859, 79– 86. (23) Muijselaar, P. G.; Claessens, H. A.; Cramers, C. A. Anal. Chem. 1997, 69, 1184–1191. (24) Anderson, J. L.; Ding, J.; Welton, T.; Armstrong, D. W. J. Am. Chem. Soc. 2002, 124, 14247–14254. (25) Zapadlo, M.; Krupčík, J.; Kovalczuk, T.; Májek, P.; Špánik, I.; Armstrong, D. W.; Sandra, P. J. Chromatogr. A 2011, 1218, 746–751. (26) Krupčík, J.; Gorovenko, R.; Špánik, I.; Bočková, I.; Sandra, P.; Armstrong, D. W. J. Chromatogr. A 2013, 1301, 225– 236. (27) Seeley, J. V.; Seeley, S. K.; Libby, E. K.; Breitbach, Z. S.; Armstrong, D. W. Anal. Bioanal. Chem. 2008, 390, 323–332. (28) Rodríguez-Sánchez, S.; Galindo-Iranzo, P.; Soria, A. C.; Sanz, M. L.; Quintanilla-López, J. E.; Lebrón-Aguilar, R. J. Chromatogr. A 2014, 1326, 96–102. (29) Li, Q.; Poole, C. F.; Kiridena, W.; Koziol, W. W. Analyst 2000, 125, 2180–2188. (30) Kiridena, W.; Koziol, W. W.; Poole, C. F. J. Chromatogr. A 2001, 932, 171–177. (31) Li, Q.; Poole, C. F. J. Sep. Sci. 2001, 24, 129–135. (32) Kiridena, W.; Koziol, W. W.; Poole, C. F.; Nawas, M. I. Chromatographia 2001, 54, 749–756. (33) Poole, C. F.; Li, Q.; Kiridena, W.; Koziol, W. W. J. Chromatogr. A 2001, 912, 107–117. (34) Poole, C. F.; Poole, S. K. J. Chromatogr. A 2008, 1184, 254–280.

ACS Paragon Plus Environment

Analytical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(35) Weber, W.; Andersson, J. T. Anal. Bioanal. Chem. 2014, 406, 5347–5358. (36) Poole, C. F.; Li, Q.; Kiridena, W.; Koziol, W. W. J. Chromatogr. A 2000, 898, 211–226. (37) Zhao, Q.; Eichhorn, J.; Pitner, W. R.; Anderson, J. L. Anal. Bioanal. Chem. 2009, 395, 225–234. (38) Twu, P.; Zhao, Q.; Pitner, W. R.; Acree Jr, W. E.; Baker, G. A.; Anderson, J. L. J. Chromatogr. A 2011, 1218, 5311–5318. (39) Lang, A., Abraham Descriptor Prediction from SMILES Model 003. http://showme.physics.drexel.edu/onsc/models/AbrahamDescripto rsModel003.php (accessed June 11, 2014). (40) Poole, S. K.; Poole, C. F. J. Sep. Sci. 2008, 31, 1118– 1123. (41) Zeng, Z. D.; Hugel, H. M.; Marriott, P. J. Anal. Chem. 2013, 85, 6356–6363. (42) Davis, J. M. J. Sep. Sci. 2005, 28, 347–359.

Page 8 of 8

(43) Camenzuli, M.; Schoenmakers, P. J. Anal. Chim. Acta 2014, 838, 93–101. (44) Krupčík, J.; Májek, P.; Gorovenko, R.; Sandra, P.; Armstrong, D. W. J. Chromatogr. A 2011, 1218, 3186–3189. (45) Seeley, J. V.; Libby, E. M.; Edwards, K. A. H.; Seeley, S. K. J. Chromatogr. A 2009, 1216, 1650–1657. (46) Seeley, J. V.; Seeley, S. K. J. Chromatogr. A 2007, 1172, 72–83. (47) Jiang, M.; Kulsing, C.; Nolvachai, Y.; Marriott, P. J. Anal. Chem. 2015, 87, 5753–5761. (48) Nosheen, A.; Mitrevski, B.; Bano, A.; Marriott, P. J. J. Chromatogr. A 2013, 1312, 118–123. (49) Gu, Q.; David, F.; Lynen, F.; Vanormelingen, P.; Vyverman, W.; Rumpel, K.; Xu, G.; Sandra, P. J. Chromatogr. A 2011, 1218, 3056–3063.

Table of Contents (TOC)

ACS Paragon Plus Environment