Generating Multiple Independent Retention Index Data in Dual

Anal. Chem. 2006, 78, 8089-8097

Generating Multiple Independent Retention Index Data in Dual-secondary Column Comprehensive Two-dimensional Gas Chromatography Stefan Bieri and Philip J. Marriott*

School of Applied Sciences, RMIT University, G.P.O. Box 2476V, Melbourne, Victoria, 3001, Australia

A method producing simultaneously three retention indexes for compounds has been developed for comprehensive two-dimensional gas chromatography by using a dual secondary column approach (GC × 2GC). For this purpose, the primary flow of the first dimension column was equally diverted into two secondary microbore columns of identical geometry by means of a three-way flow splitter positioned after the longitudinally modulated cryogenic system. This configuration produced a pair of comprehensive two-dimensional chromatograms and generated retention data on three different stationary phases in a single run. First dimension retention indexes were determined on a polar SolGel-Wax column under linear programmed-temperature conditions according to the van den Dool approach using primary alcohol homologues as the reference scale. Calculation of pseudoisothermal retention indexes in both second dimensions was performed on low-polarity 5% phenyl equivalent polysilphenylene/siloxane (BPX5) and 14% cyanopropylphenyl/86% dimethylpolysiloxane (BP10) columns. To construct a retention correlation map in the second dimension separation space upon which Kova´ ts indexes can be derived, two methods exploiting “isovolatility” relationships of alkanes were developed. The first involved 15 sequential headspace samplings of selected n-alkanes by solid-phase microextraction (SPME), with each sampling followed by their injection into the GC at predetermined times during the chromatographic run. The second method extended the second dimension retention map and consisted of repetitive introduction of SPME-sampled alkane mixtures at various isothermal conditions incremented over the temperature program range. Calculated second dimension retention indexes were compared with experimental values obtained in conventional one-dimensional GC. A case study mixture including 24 suspected allergens (i.e., fragrance ingredients) was used to demonstrate the feasibility and potential of retention index information in comprehensive 2D-GC. The Retention Index (I) Notion. Very early in the history of gas chromatography, it was recognized that reproducibility of retention was a fundamental parameter for identification. Sup* Corresponding author. Phone: +61 3 99252632. Fax: +61 3 99253747. E-mail: [email protected]. 10.1021/ac060869l CCC: $33.50 Published on Web 10/31/2006

© 2006 American Chemical Society

porting this was the basic concept of the retention index system, which quantifies the elution position of a compound against two closely eluting and bracketing members of a homologous series, such as n-alkanes; alcohols; ethers; or in some specific cases, n-fatty acid methyl esters,1,2 specific polyaromatic hydrocarbons,3,4 (PAH) or polychlorinated biphenyl congeners.5,6 Kova´ts introduced this concept in 19587 by demonstrating that the logarithm of the adjusted retention time of components of a homologous series eluting under isothermal conditions increases linearly with their carbon number, thus forming a uniform reference scale. Retention indexes are essentially dependent only on the chromatographic phenomena (interaction between analyte and stationary phase at a given temperature), but they still require good control of conditions and columns (e.g., dissimilar performance of equivalent stationary phases from different manufacturers) to ensure the precision and accuracy of I data. Furthermore, to maximize retention index reliability, the polarity of the homologous series should be adapted to or compatible with the polarity of the column.8 This is to ensure good gas-liquid partitioning and avoid mixed retention mechanisms (e.g., interfacial adsorption of alkanes on polar stationary phases). It is noteworthy that the retention index system as defined by Kova´ts is restricted to isothermal elution conditions. van den Dool and Kratz9 overcame this restriction, extending it to linear programmed temperature retention indexes. When retention index information is available on different stationary phases, identification capabilities are close to those obtainable with mass spectrometric data.10 Nonetheless, the current situation is that the advent of mass spectrometry (MS) has led some to believe that the need for I values has diminished. Some areas of analysis (e.g., essential oils, fragrances, and flavor analysis) still depend critically upon retention index-based identification, usually in combination with MS. This is especially the case for the identification of closely related structures and isomers, (1) Miwa, T. K.; Mikolajczak, K. L.; Earle, F. R.; Wolff, I. A. Anal. Chem. 1960, 32, 1739-1742. (2) Woodford, F. P.; van Gent, C. M. J. Lipid Res. 1960, 1, 188-190. (3) Vassilaros, D. L.; Kong, R. C.; Later, D. W.; Lee, M. L. J. Chromatogr. 1982, 252, 1-20. (4) Lee, M. L.; Vassilaros, D. L.; White, C. M. Anal. Chem. 1979, 51, 768773. (5) Castello, G.; Testini, G. J. Chromatogr., A 1996, 741, 241-249. (6) Castello, G.; Testini, G. J. Chromatogr., A 1997, 787, 215-225. (7) Kovats, E. Helv. Chim. Acta 1958, 41, 1915-1932. (8) Castello, G. J. Chromatogr., A 1999, 842, 51-64. (9) van den Dool, H.; Kratz, P. D. J. Chromatogr. 1963, 11, 463-471. (10) Bicchi, C.; Binello, A.; D’Amato, A.; Rubiolo, P. J. Chromatogr. Sci. 1999, 37, 288-294.

Analytical Chemistry, Vol. 78, No. 23, December 1, 2006 8089

which often provide virtually superimposable electron impact mass spectra and, thus, cannot be unambiguously identified by MS. Therefore, incorporation of retention indexes in peak identification criteria strongly assists and complements MS matches or alternative confirmatory detection techniques, such as Fourier transform infrared spectroscopy or element-specific detectors.11,12 Retention Indexes in Comprehensive Two-Dimensional Gas Chromatography. Chromatographic developments search for greater separation power, pushing capillary GC to the limits for complex mixture characterization. Although this limit has essentially been reached for one-dimensional GC, multidimensional GC13 and, more particularly, comprehensive two-dimensional gas chromatography14-16 (GC × GC) have the potential to solve separation problems in ways previously not possible. The first GC × GC results were published by Phillips in 1991,17 and thus, they represent the practical achievement of the pioneering concept outlined by Giddings in 1983.18 The major benefit of this technique is its high peak capacity approximated as the product of the peak capacities of the two individual dimensions.19 To achieve this, the separations of the two columns are “decoupled” through use of a modulation system that continuously samples the first column effluent and rapidly delivers it to the second column; the period of modulation (PM) is usually less than the primary peak width and often is close to the peak width at halfheight, which leads to a modulation ratio (MR) of ∼4.20 As a result, the significantly greater separation performance achieved by GC × GC offers much greater scope for retention to be a more powerful identification property for molecules than in conventional single-column GC. Although retention index calculations are still relevant to singlecolumn GC, it is of interest to establish methods, approaches, and understanding of how indexes might be estimated and their reliability and practical utility might be investigated in GC × GC. However, there are, to date, only three papers that deal with concepts of retention indexes in both dimensions in GC × GC21-23 and show that conventional notions and bases for I calculation require considerable reevaluation. For GC × GC experiments, a reference map must be constructed in the second axis of the two-dimensional separation space upon which pseudoisothermal retention indexes can be derived (since the individual analyses on the second column are very fast, we assume that the temperature is effectively isothermal). For this purpose, isovolatility24 curves of members of a homologous series involving repetitive split injection during a chromatographic analysis proved to be an interesting approach, (11) Bicchi, C.; Frattini, C.; Raverdino, V. J. Chromatogr. 1987, 411, 237-249. (12) Bicchi, C.; Frattini, C.; Pellegrino, G.; Rubiolo, P.; Raverdino, V.; Tsoupras, G. J. Chromatogr. 1992, 609, 305-313. (13) Bertsch, W. J. High Resolut. Chromatogr. 1999, 22, 647-665. (14) Marriott, P.; Shellie, R. Trends Anal. Chem. 2002, 21, 573-583. (15) Bertsch, W. J. High Resolut. Chromatogr. 2000, 23, 167-181. (16) Dalluge, J.; Beens, J.; Brinkman, U. A. Th. J. Chromatogr., A 2003, 1000, 69-108. (17) Liu, Z.; Phillips, J. B. J. Sep. Sci. 1991, 29, 227-231. (18) Giddings, J. C. Anal. Chem. 1984, 56, 1258A-1270A. (19) Phillips, J. B.; Beens, J. J. Chromatogr., A 1999, 856, 331-347. (20) Khummueng, W.; Harynuk, J.; P. J. Marriott, Anal. Chem. 2006, 78, 45784587. (21) Arey, J. S.; Nelson, R. K.; Xu, L.; Reddy, C. M. Anal. Chem. 2005, 77, 71727182. (22) Western, R. J.; Marriott, P. J. J. Sep. Sci. 2002, 25, 832-838. (23) Western, R. J.; Marriott, P. J. J. Chromatogr., A 2003, 1019, 3-14. (24) Beens, J.; Tijssen, R.; Blomberg, J. J. High Resolut. Chromatogr. 1998, 21, 63-64.

8090

Analytical Chemistry, Vol. 78, No. 23, December 1, 2006

but showed some drawback due to the introduction of solvent at each injection.22 This leads to interfering solvent bands throughout the 2D GC × GC space. Alternatively, Arey et al.21 recently estimated second dimension retention indexes of diesel fuel solutes in GC × GC (nonpolar/polar column set) on the basis of calculated second dimension retention times of hypothetical n-alkanes (eluting at the same elution temperature, Te, as the solute whose retention is to be estimated) using experimental retention values of nearby n-alkanes present in the sample, as well as by exploiting experimental values of alkanes determined separately in conventional GC to develop a linear free energy relationship. These authors showed that two-dimensional retention data can be used to predict and estimate various partitioning properties for solutes analyzed by GC × GC with the developed mathematical approach. Because the second column in GC × GC is usually a polar column that often has unreliable index reproducibility and poor retention of alkanes, the present study aimed at demonstrating that reversal of the column set phase polarity offers considerable opportunities for some analyses and allows use of nonpolar homologues for retention indexes determination in the second dimension. Two SPME methods have been developed, rather than using consecutive solvent injections for the construction of the isovolatility curves, that permit extension of the retention map and also determination of separation numbers in the second dimension. Finally, to increase the retention-based identification capabilities, a GC × GC approach generating chromatographic retention on three different stationary phases in a single run was employed. A mixture containing 24 suspected allergens (SA) claimed as potentially leading to dermatological irritation by the European Commission’s Scientific Committee on Cosmetics and other Non-Food Products was used as a test solution.25 EXPERIMENTAL METHODS Instrumentation. Single dimension and GC × GC experiments were conducted using an Agilent 6890 gas chromatograph (Agilent Technologies, Burwood, Australia) equipped with a split/ splitless injector and two high-speed flame ionization detectors (FID). The detectors were operated at 20- and 100-Hz data acquisition rates during 1D-GC and GC × GC experiments, respectively. For GC × GC analysis, the GC was retrofitted with a longitudinally modulated cryogenic system (LMCS) from Chromatography Concepts (Doncaster, Australia). GC × GC data are converted and displayed as described elsewhere.26 Standard and Sample Preparation. A standard solution containing 24 SA and phenylacetaldehyde (see Table 1) was chosen as the test mix and prepared in hexane (0.1 mg/mL each compound). For determination of first dimension, linear temperature-programmed retention indexes, a solution of n-alcohols ranging from 1-hexanol to 1-eicosanol plus 1-docosanol (0.1 mg/ mL each) was prepared in dichloromethane. All compounds were obtained from Sigma-Aldrich in purity exceeding 97%, where available. Introduction of n-alkanes, necessary for the construction of the second dimension retention map during GC × GC analysis, was carried out by sampling the headspace (HS) of selected alkanes with a 30-µm film thickness (100% poly(dimethylsiloxane)), (25) Shellie, R.; Marriott, P.; Chaintreau, A. Flavour Fragrance J. 2004, 19, 9198. (26) Kinghorn, R. M.; Marriott, P. J.; Dawes, P. A. J. High Resolut. Chromatogr. 2000, 23, 245-252.

Analytical Chemistry, Vol. 78, No. 23, December 1, 2006

8091

1 2 3 4 5a 5b 6 7 8 9 10 11 12 13 14 15 16 17a 17b 18 19 20 21a 21b 22a 22b 23 24 25

limonene linalool phenylacetaldehyded methyl 2-octynoate citrale

120-51-4 118-58-1 103-41-3

31906-04-4

97-54-1 101-86-0 91-64-5 101-85-9

111-80-8 106-22-9 127-51-5 106-24-1 100-51-6 107-75-5 104-55-2 95-53-0 122-40-7 105-13-5 104-54-1 4602-84-0

5989-27-5 78-70-6 122-78-1 111-12-6 5392-40-5

CAS-RN 792 868 884 913 961 982 1001 1073 1076 1086 1169 1237 1364 1456 1464 1470 1526 1565 1532 1554 1604 1604 1724 1653 1664 1788 1932 -

792 862 881 908 957 979 1001 1066 1076 1084 1168 1230 1360 1450 1460 1466 1526 1565 1527 1548 1590 1602 1722 1648 1659 1779 1920 -

792 868 884 913 961 982 1001 1073 1076 1086 1169 1236 1363 1456 1463 1469 1526 1565 1531 1554 1604 1604 1723 1653 1664 1788 1932 -

I3T pseudo 1D-GC Wax-BP10 791 870 883 912 961 983 1002 1074 1076 1087 1172 1240 1363 1456 1467 1470 1527 1566 1533 1556 1604 1607 1726 1656 1666 1789 1931 -

I1D-WaxT GC× GC Wax-BPX5 791 870 884 915 964 986 1002 1075 1077 1088 1170 1237 1363 1454 1461 1469 1523 1563 1530 1553 1603 1603 1722 1651 1662 1786 1930 -

I1D-WaxT GC× GC Wax-BP10 1118 1085 1259 1290 1254 1268 1100 1324 1342 1406 1681 1377 1393 1720 1741 1518 1776 1578 1696 1731 1716 1722 1837 1956 -

method 1 I2D-BPX5 GC× GC BPX5 1042 1107 1077 1214 1258 1288 1313 1255 1491 1274 1107 1328 1344 1392 1681 1368 1385 1726 1751 1518 1789 1569 1698 1736 1721 1728 1838 1951 -

method 2 I2D-BPX5 GC× GC BPX5 1038 1096 1074 1212 1257 1283 1312 1234 1487 1260 1064 1304 1318 1385 1676 1326 1350 1709 1732 1490 1775 1518 1690 1718 1699 1707 1824 1935 -

Icalc 1D-GC BPX5

1386 1227 1471 1474 1535 1527 1543 1662 1722 1884 1887 1896 1981 -

1200 1200 -

method 1 I2D-BP10 GC× GC BP10

1064 1193 1188 1311 1368 1403 1415 1356 1599 1376 1237 1475 1480 1541 1810 1530 1545 1843 1868 1671 1908 1735 1838 1885 1885 1893 1974 2092 -

method 2 I2D-BP10 GC× GC BP10

Kova´ts indexes (second dimension)b

1055 1186 1179 1301 1362 1394 1402 1329 1590 1359 1194 1452 1454 1510 1792 1477 1496 1816 1841 1627 1893 1706 1815 1854 1859 1868 1949 2067 -

Icalc 1D-GC BP10

a van den Dool indexes from n-alcohol reference compounds. b Kova ´ ts indexes from n-alkane reference compounds. c (-) These compounds were not bracketed by two homologues in the reference series. d Not in the SCCNDFP list of 24 SA. e Compound purchased as a mixture of E and Z isomers.

isoeugenol hexylcinnamaldehyde coumarin amylcinnamic alcohol (impurity of 21a) lyral (impurity of 22a) benzyl benzoate benzyl salicylate benzyl cinnamate

methyl 2-nonynoate citronellol R-isomethylionone geraniol benzyl alcohol hydroxycitronellal cinnamaldehyde eugenol amyl cinnamaldehyde anisyl alcohol cinnamic alcohol farnesole

no.

compound

I2T pseudo 1D-GC Wax-BPX5

I1T 1D-GC Wax

van den Dool indexes (first dimension)a

Table 1. van den Dool and Kova ´ ts Retention Index Data for SA on First and Second Dimension Columns or Across the Combined Column Set (pseudo 1D)

Figure 1. Schematic of the dual-secondary column comprehensive two-dimensional gas chromatographic system (GC × 2GC). FS: fused silica.

solid-phase microextraction (SPME) fiber (Supelco, Bellefonte, PA). Prior to use, the fiber was conditioned according to the manufacturer’s instructions. Alkanes C9-C12 were individually sampled at room temperature (each ∼2 mg in separate 20-mL vials), C13 and C14 were sampled at 40 °C (∼5 mg in a 5-mL vial), C15-C17 at 80 °C (∼5 mg in a 5-mL vial), and C18-C20 at 120 °C (∼5 mg in a 5-mL vial). HS-SPME sampling times of individual alkanes varied between 1 and 25 s to sample similar amounts of alkanes for injection. Using this approach, any desired composite mixture of alkanes could be sampled. Following the sampling/ trapping step of a selected series of alkanes (sampling is commenced with the highest boiling alkane, proceeding to the more volatile alkane), they were desorbed in the split mode into the hot injector. The split ratio was set at 45:1 by initiating the gas saver flow rate at 60 mL/min, one min after the initial split injection of the liquid sample (this allowed a fast desorption and transfer time to be performed and led to sharp injection bands for each SPME alkane introduction). GC × GC experiments. GC × GC analyses were carried out using a dual secondary column system in which the first dimension column was coupled by means of a deactivated glass threeway splitter (BGB Analytik, Bo¨ckten, Switzerland) after the modulator to two second dimension columns of the same geometry but with different stationary phases (Figure 1). A piece of deactivated capillary (Agilent Technologies, 0.2 m × 0.25 mm i.d.) in which analytes were trapped by the LMCS joined the first dimension column to the flow splitter. This empty capillary was intended to avoid any extra contribution of stationary phase on retention indexes on both second dimension columns. The primary (1D) column was a polar poly(ethylene glycol) phase (SolGel-Wax, 30 m × 0.25 mm i.d., 0.25-µm film thickness; column 1); the first second dimension column (2D1) coupled to FID1 was a low-polarity, 5% phenyl polysilphenylene/siloxane phase (BPX5, 0.95 m × 0.1 mm i.d., 0.1-µm film thickness; column 2); and the other second dimension column (2D2) coupled to FID2 was a medium polarity, 14% cyanopropylphenyl/86% dimethylpolysiloxane phase (BP10, 0.95 m × 0.1 mm i.d., 0.1-µm film thickness; column 3). All columns were obtained from SGE International (Ringwood, Australia). The LMCS was operated at -40 °C using CO2 as coolant and with a modulation timing period of 5 s. The oven was linearly temperature-programmed from 65 to 245 °C at 3 °C/min. The carrier gas was hydrogen, which was supplied in 8092 Analytical Chemistry, Vol. 78, No. 23, December 1, 2006

Figure 2. Two-dimensional retention map obtained with multiple SPME injections (15 in total) of series of n-alkanes (C9-C20). Desorptions of alkanes were made at 2.5, 5.0, 8.0, 10.5, 15.0, 19.0, 23.0, 27.0, 30.5, 34.0, 38.0, 42.5, 45.0, 50.0, and 55.0 min; the alkane elutions for each mix in the GC × GC experiment are shown by the faint dotted lines. At t ) 0, an injection of a solution of n-alcohols ranging from C6 to C20, including 1-docosanol (C22), was carried out (heavy dashed line).

the constant flow mode with an initial head pressure of 11.8 psi. Split injections of the test mixture (1 µL, split ratio 10:1) were carried out at an injector temperature of 250 °C with a fast automatic liquid sampler (Agilent 7683 autosampler). Method 1 was used to produce the two-dimensional retention map using the same conditions as those employed for the separation of target compounds. For this purpose, after an initial injection of the mixture of n-alcohols, various suites of n-alkanes were sequentially introduced by SPME throughout the temperature-programmed chromatographic run. Thus, in a single run, 15 desorptions of varying alkane mixes were conducted at predetermined times (described in Figure 2 and discussed below), which formed the basis for generating the isovolatility curves in the second dimension separation space. Method 2 was used to extend the second dimension retention map by introducing various mixtures of n-alkanes by SPME, but under different isothermal conditions as follows: using flow conditions identical to those of method 1, an analysis consisting of 14 consecutive isothermal plateaus (80, 89, 98, 104, 116, 131, 143, 155, 167, 176, 188, 209, 221, and 236 °C) was used with three replicates of a selected alkane series introduced at each temperature setting. Although the temperatures chosen for this experiment are not critical, the alkane mixtures were essentially adapted for each isothermal condition so as to elute within the time range (5 s) of the second dimension separation (reported in Figure 4 and discussed below). After each isothermal plateau, the oven was quickly heated (at 30 °C/min) to the next temperature level. Average retention times (n ) 3) were used to generate the isovolatility map plotted as retention time versus elution temperature, Te. By calculating, for each selected temperature, the linear regression equation of the logarithm of adjusted retention time of each alkane versus carbon number z (R2 varied between 0.9980 and 0.9999 within the temperature range 89-236 °C for data on both secondary columns), we could calculate the retention time of subsequent higher homologues and so expand the second dimension separation space to 12 s. The holdup time in the second

dimension was calculated at 143 °C by using second dimension retention times (average of five measurements) of three consecutive alkanes (C10-C12) introduced by SPME during isothermal GC × GC conditions, according to the following equation:27 tM ) (tR1tR3 - t2R2)/((tR3 - tR2) - (tR2 - tR1)) ) 1.17 s. Conventional One-Dimensional GC. Linear temperatureprogrammed retention indexes (according to van den Dool and Kratz9) were determined in single dimension conditions (average of three consecutive measurements) on the SolGel-Wax stationary phase (column 1). Isothermal retention indexes were calculated (according to Kova´ts7) at various temperatures (from 50 to 210 °C at 10 °C intervals, as reported in Figure 6) on both a BPX5 (30 m × 0.25 mm i.d., 0.25-µm film thickness; column 4) and a BP10 phase (5 m × 0.1 mm i.d., 0.1-µm film thickness; column 5). Columns 4 and 5 were obtained from SGE International. RESULTS AND DISCUSSION To increase identification capabilities based on relative retention, a comprehensive two-dimensional gas chromatographic system that splits the primary column effluent toward two secondary columns (Figure 1) was used. This dual-secondary column system28 enabled determination of independent retention indexes on three stationary phases. A standard solution containing 25 fragrance ingredients (Table 1) was used as the test mixture. The first retention index series is based on linear temperatureprogrammed measurements on a polar first dimension column (SolGel-Wax) using the n-alcohols injected at t ) 0 as the reference scale. To obtain an adequate reference map upon which retention indexes in the second dimensions can be derived, we developed a headspace SPME method allowing solvent free introduction of selected n-alkanes at predetermined times during the GC × GC run. The resultant contour plot of 15 sequential SPME injections of n-alkane series together with the injection of the alcohol homologues (indicated by heavy dotted line) is shown in Figure 2. As the oven temperature increases, the mix of n-alkanes introduced by SPME is changed (to a higher alkane range, which suits the higher elution temperature), as indicated by the C-no. range of 14 of the 15 injections on the diagram. The first alkane mixture injection is of C9 + C10. The vertical lines at ∼2 min arise from the nontrapped solvent peak of the alcohol sample (injected at t ) 0) and justifies why an SPME method was developed for the alkane injections: it removes the interfering solvent peaks that would otherwise arise. At 10.5 min, an injection of C9-C12 was made, and these alkanes then eluted starting at ∼12 min. Each individual injected mix elutes within a 2.5-min time window from the first dimension column, indicated by the faint dotted lines linking the alkane suites of a given injection, which delineate almost parallel lines throughout the analysis due to the small affinity of the polar first dimension column toward n-paraffins. Linking the peak apex of identical alkanes via calculated exponential curves (Origin 7.5 plotting program was used for all data presentations and curve-fitting calculations) produced the so-called isovolatility lines,24 which were used as the reference map for retention index calculation in the second dimension separation space under pseudoisothermal conditions (Figure 3). The isovolatility line or curve is the line that describes the retention time of a given compound in the second dimension for varying (27) Ettre, L. S.; Hinshaw, J. V. Basic Relationships of Gas Chromatography, 2nd ed.; Advanstar Communications: Cleveland, 1993; pp 142-143. (28) Seeley, J. V.; Kramp, F. J.; Sharpe, K. S. J. Sep. Sci. 2001, 24, 444-450.

Figure 3. Exponential (second order) isovolatility curves for nalkanes ranging from C9 to C20 (data taken from Figure 2) superposed with the 2D contour plot of the suspected allergens. A2: BPX5 column in the second dimension. B2: BP10 stationary phase in the second dimension. A1: modulated chromatogram of n-alcohol set. B1: nonmodulated chromatogram of the 25 suspected allergens. Analytes annotated with an asterisk (/) suffered wraparound.

(increasing) oven temperatures. Pseudoisothermal conditions are those that hold for the short duration of each second dimension chromatogram, where oven temperature is essentially almost isothermal. Indeed, a temperature rate of 3 °C/min combined with a modulation period of 5 s leads to a temperature increase in the second dimension separation space of only 0.25 °C for each second dimension analysis. Therefore, second dimension chromatograms can essentially be considered as isothermal, which allows 2D retention indexes to be calculated according to Kova´ts’ equation.7 The isovolatility curves together with the contour plot of the 25 target compounds (see Table 1) are shown for the two 2D column in Figure 3A2 and B2. As a result, this approach generated three independent retention indexes for most compounds. The Analytical Chemistry, Vol. 78, No. 23, December 1, 2006

8093

first retention index dimension (ranging from 600 to 2200 retention units, Figure 3A1) is determined in linear programmed temperature conditions on a polar SolGel-Wax column according to van den Dool and Kratz9 using the suite of n-alcohols; Figure 3A1 is a modulated chromatogram of the n-alcohol sample. Second dimension indexes are based on the isovolatility curves of the n-alkanes (the map was from 900 to 2000 retention units) and were determined on the apolar BPX5 (Figure 3A2) and the mid-polar BP10 (Figure 3B2) columns. Figure 3B1 shows the nonmodulated chromatogram of the SA and reveals that the polar first dimension column is a good choice because almost all SA are separated. To calculate second dimension retention indexes of target compounds, 2D retention times of the bracketing alkanes that elute before and after each solute (2tR(z) and 2tR(z+1)) were determined from the individual isovolatility equations using values at the elution temperature of each compound of interest. First dimension linear retention indexes (vs alcohol reference) are calculated for three experiments: from 1D-GC (I1T), GC × GC conditions (I1D-WaxT) for both column sets, and over the total column set under nonmodulated conditions (I2T and I3T, termed pseudo-1DGC, Table 1). Second dimension indexes (method 1) determined by interpolation between isovolatility curves, on both secondary columns I2D-BPX5 and I2D-BP10, are also presented in Table 1. A limitation of the above approach (method 1) for generating the alkane-based correlation map was that an upper bound alkane reference line was not always available to bracket the elution position of each individual SA. Furthermore, compounds that wrapped around (elute later than the modulation period, and so are located and plotted during the subsequent modulation event; see asterisked solutes in Figure 3B2) could not be interpolated between two appropriate homologues. Therefore, a second strategy (method 2) was developed to produce alkane isovolatility curves in such a manner as to extend the retention map in the second dimension separation space and so overcome the limitation of the upper isovolatility curve range. For this purpose, a multiple isothermal plateau analysis was carried out using GC × GC conditions. At each isothermal plateau (14 in total), three consecutive SPME injections of a given n-alkane mixture were performed. The alkane map was then generated by plotting the second dimension retention time (average of three measurements) vs elution temperature, producing a true isothermal second dimension alkane separation space (Figure 4). Due to the linear Kova´ts relation of the logarithm of the adjusted retention time vs increasing carbon number (z) of a series of homologues injected at a given temperature, the second dimension retention times extrapolated for subsequent homologues beyond those in the injected series could be reasonably accurately predicted. For example, if at the temperature of 143 °C (dotted line, Figure 4) a mixture of C10-C15 was injected, the second dimension retention times of C16 and C17 could be calculated. Figure 4 shows calculated curves (dashed) fitted to experimental data for injection of n-alkanes using method 1, along with curves (solid) that included data for the extrapolated points. This method permitted extension of the second dimension retention map (Figure 4), shown as the upper clear part of the 2D space, beyond the 5-s shaded part of the space within which injected solutes eluted. This expansion was particularly useful in the case of the more polar second dimension column (BP10) in which several SAs wrapped around (compounds 5, 6, and 8). The latter could then be interpolated within two isovolatility curves 8094 Analytical Chemistry, Vol. 78, No. 23, December 1, 2006

Figure 4. Reconstructed two-dimensional retention correlation map plotting second-dimension retention time of n-alkanes vs elution temperature on the SolGel-Wax column. Solid line: exponential isovolatility curves of the individual alkanes. Solid points (b): average retention times obtained for three consecutive SPME desorptions under isothermal conditions. Open circles (O): extrapolated points; for example, the vertical dashed line at 143 °C uses retentions of C10-C15 to calculate extrapolated retention times of C16 and C17. Interrupted line: theoretical isovolatility curve of C21 constructed from extrapolated data. Dashed curves: isovolatility curves obtained with method 1 (see text). The clear section of the 2D space is the extrapolated region.

and their indexes estimated (shown by interrupted lines, Figure 4). It also allowed the isovolatility line of C21 to be plotted by extrapolation, which in turn permitted the index of SA 24 to be estimated. From Table 1, it can be concluded that both methods (refer to columns headed method 1 and method 2) led to fairly similar second dimension index values, with a maximum difference of