Universal Method for Online Enrichment of Target ... - ACS Publications

Cryotrap (CT) methods in GC are well-known approaches to collect volatile compounds. Purge-and-trap prior to injection,(14) multidimensional GC (MDGC)...
1 downloads 0 Views 2MB Size
Download PDF
ARTICLE pubs.acs.org/ac

Universal Method for Online Enrichment of Target Compounds in Capillary Gas Chromatography Using In-Oven Cryotrapping Sung-Tong Chin, Bussayarat Maikhunthod, and Philip J. Marriott*,† Centre for Green Chemistry, School of Chemistry, Monash University, Wellington Road, Clayton 3800, Australia, and College of Pharmacy (WCU), †Chung-Ang University, Seoul 156-756, Republic of Korea

bS Supporting Information ABSTRACT: A method is described that permits automated online enrichment of injected compounds in multidimensional gas chromatography by using a microfluidic heart-cut (H-C) device to direct target compounds into a cryogenically cooled internal trap (cryotrap, CT). By performing multiple injections of a sample, selected compounds or regions of a primary column separation can be collected in the CT. Remobilizing the trapped species allows elution and further resolution on the second column. Using a well-balanced H-C device, compounds can be fully excluded from the collection step or quantitatively transferred to the CT. Peak areas of the remobilized compound correlate well with the number of sample injections. Trapping on various column phases shows the method is suited to quantitative trapping of alkanes of mass greater than about dodecane and fatty acid methyl esters greater than the C8 homologue. Caffeine and menthol standards of concentration 100 μg mL 1 gave peak area correlation coefficients for 1 10 and 1 50 replicate split injections of 1 μL volume of 0.999 and 0.996, respectively. Peak height correlations were less favorable as a result of peak broadening on the second column, presumably due to overloading at greater collected mass. The method was applied to 0.2% solutions of peppermint oil (menthol; a major component; 44%) and 1.0% lavender oil (R-terpineol and neryl acetate; minor components of 1.05 and 0.42% abundance). The minor components gave good area and height correlations, and good recovery around 90% was observed for menthol compounds recovered from 15 accumulations. Response amplification was further demonstrated for menthol from mint oil headspace sampling using solid phase microextraction. This approach should be a valuable adjunct for improved detection specificity, for detectors of low sensitivity, and when prior sample concentration provides insufficient response of selected target analytes.

C

ompound detection in gas chromatography (GC) requires sufficient compound mass to satisfy the needs for adequate signal measurement. Different detectors possess specific demands with respect to compound detectability, depending upon the sensing mechanism. A minimum requirement for sample mass depends upon the magnitude of the signal (S) in comparison to the noise (N) level associated with the detector, and other attendant parameters that affect the measurement of signal.1 Sensitive detection may be needed for discovering specific biological activity or functionality.2,3 In all cases, trace components may often be below the detection limits (DL) of the detector used. Detection in GC may be integrated with the GC instrument (online) or remote from the instrument (off-line).4,5 Frequently off-line detection requires greater sample mass for adequate detection of the compound, since postcollection sample handling may accompany dilution6 or other sample losses prior to spectroscopic analysis,7 and background contamination may arise.8 Offline methods may include those unsuitable for hyphenation with GC, such as nuclear magnetic resonance,9 X-ray methods,10 accelerator mass spectrometry,8 or synchrotron spectroscopy. Not all spectroscopic methods may be available as a hyphenated r 2011 American Chemical Society

instrument in a laboratory, such as FTIR, so isolation of the analyte makes the use of stand-alone instruments possible. Certain GC detectors suffer from poor DL and require considerable quantities of sample to be injected. In the analysis of volatile stereoisomers, successful online GC NMR detection at 400 MHz was reported to require a sample amount of 1.1 μmol of cis/trans-1,2-dimethylcyclohexane (about 12 μg).11 GC coupled with multiplex coherent Raman spectroscopy has the advantage of greater spectral resolution compared with GC FTIR;12 however, a 5.5 μg sample had a S/N ratio of 5. Such methods, with DL for benzene in the tens of micrograms range, have significantly poorer DL than that of common GC detectors. Prior research using GC surface-enhanced Raman scattering (SERS)13 apparently achieved a lower DL of ∼50 ng for benzene, but a later report12 described limitations of the GC SERS approach. Thus, there can be a disconnect between maximizing the chromatographic resolution and the satisfactory detection of solutes. Analytes in trace Received: February 5, 2011 Accepted: July 18, 2011 Published: July 18, 2011 6485

dx.doi.org/10.1021/ac200973z | Anal. Chem. 2011, 83, 6485–6492

Analytical Chemistry amounts normally require greater sample mass or alternative presampling enrichment to improve detection and characterization. Cryotrap (CT) methods in GC are well-known approaches to collect volatile compounds. Purge-and-trap prior to injection,14 multidimensional GC (MDGC) with heart-cutting (H-C) and focusing of analyte at the head of a 2D column,2 comprehensive 2D GC with cryomodulation,15 and collecting effluent at the outlet of a column16 represent approaches to which a CT may be applied in GC. Signal amplification of silylated sugars in GC-FID (flame ionization detection)17 involved three repeat injections (totaling 13 μL) into a packed precolumn, cumulative CT, then separation in a glass capillary column as 2D. Bicchi et al.18 performed a 4-fold accumulation into a CT after H-C of minor components from Anthemis nobihs and Tagetes nana essential oil to obtain mass spectral data for sulfur compounds. Roeraade et al.19 injected 800 aliquots of a compound into prep-GC to collect sufficient isolated component mass in an external trap to conduct 13C NMR experiments. Eyres et al.6 employed a prep-scale capillary MDGC experiment using an internal CT storage loop and external CT, with multiple injections to isolate individual essential oil and crude oil components for NMR analysis. The same group10 reported crystal structures for a number of prep-cap-GC-collected catalyst products. Nojima16 examined a range of uncoated and phase-coated capillary traps, evaluating their performance to collection of a range of analyte classes. This study was designed to enhance the in situ compound sample mass in capillary MDGC analysis, allowing multiple repeat injections to be conducted with high-resolution isolation of a specific component from a sample by use of automated Deans switch operation. Component CT at the head of a 2D column allows the mass of the component to be readily increased, with further separation from any cotransferred solute permitting high purity individual compounds to be detected. Compound enrichment of up to 50-fold, with more than 10 μg on-column mass, is reported for menthol compounds here, although further enrichment is possible.

’ EXPERIMENTAL SECTION Chemicals and Materials. Peppermint and lavender essential oils, and some of their individual component standards (L-menthol, R-terpineol (TP), and neryl acetate (NA)) were provided by Australian Botanical Products (Hallam, Australia). The SPME assembly included a 65 μm polydimethylsiloxane/divinyl benzene absorbent (PDMS/DVB) fiber and a manual fiber holder (Supelco, Sigma-Aldrich, Bellefonte, PA, USA). Alkanes (C10 C20), caffeine (purity >97%), and a 37 component fatty acid methyl ester mixture (FAME; C4 C24) were purchased from SigmaAldrich. GC grade acetone was obtained from Merck Chemical (Kilsyth, Australia). Prior to GC analysis, the essential oils and all standard compounds were dissolved and diluted with acetone solvent to an appropriate level. Heart-Cut MDGC Configuration. Figure 1 illustrates the system configuration utilized in this work. An Agilent 6890A GC (Agilent Technologies, Nunawading, Australia) was equipped with a split/ splitless inlet, dual flame ionization detectors (FID), retrofitted with an Everest model longitudinally modulated cryogenic system (LMCS; Chromatography Concepts, Doncaster, Australia). MDGC separation was performed with two different column sets: (1) a SLB 5 ms capillary column (15 m length, 0.32 mm internal diameter (i.d.), 1.00 μm film thickness) and a SLB-IL100 ionic liquid (IL) column

ARTICLE

Figure 1. Instrument schematic for MDGC arrangement. DS, Deans switch; DFS, deactivated fused-silica transfer line; 1D, first dimension column; 2D, second dimension column; CT, cryotrap. Collect and release positions shown for movement of the CT; inlet injector, FID1; 2, flame ionization detector.

(30 m  0.25 mm  0.20 μm), both from Supelco as 1D and 2D phase columns, respectively, and (2) a DB-FFAP column (30 m  0.25 mm  0.25 μm) and a DB-1301 column (10 m  0.18 mm  0.40 μm), both from Agilent Technologies as 1D and 2D phase columns, respectively. The CT of the LMCS was positioned at the beginning of the 2D column, which focused and concentrated the effluent H-C from the target regions of 1D to 2 D. A microfluidic Deans switch device (DS; Agilent, part no. G2855B) for H-C effluent switching was used to interface the end of the 1D column to the start of the 2D column and a length of 0.10 mm i.d. deactivated fused-silica tubing (DFS; 0.8 m for column set A, 1.0 m for column set B) as the transfer line. The DFS and 2D column were terminated at FID1 and FID2, respectively. Hydrogen carrier gas was used throughout. GC Operation for Online Compound Enrichment Analysis. Sample injection (1 μL) was performed using an Agilent 7683 series autosampler into the GC inlet at 280 °C. Both FIDs (Figure 1) were operated at 250 °C with acquisition rate of 20 Hz. The standard mixture or essential oil sample was separated on 1D, with transfer of compounds via DFS to FID1, allowing estimation of the switching time for the H-C event to be entered into the events software. The functional parameter is the injected amount of solute, rather than the injection method. Note here the CT functions by fast movement relative to the column to instantaneously remobilize collected compound into the downstream 2D column. This is in contrast to other CT devices that turn off or interrupt the cryogen supply while heating the column segment.20 22 The Supporting Information provides further details on H-C and GC operation, with Table S1 reporting programmed events for ChemStation software control. Trapping Efficiency over the Molecular Range of Alkanes and FAME. The alkane and FAME mixtures (both at 100 μg mL 1) were used with column set 1. Standard mixtures were injected in a split mode of 20:1 as a convenient way to deliver less analyte to the column to provide smaller compound amounts. For alkane trapping, the GC oven program was 80 °C for 1 min, ramp to 230 at 25 °C/min, hold for 10 min. For FAME trapping, the program was 80 °C for 2 min, ramp to 180 at 25 °C/min, then to 6486

dx.doi.org/10.1021/ac200973z |Anal. Chem. 2011, 83, 6485–6492

Analytical Chemistry 230 at 15 °C/min. During both analyses, selected discrete individual compounds from 1D were H-C to 2D then either (i) eluted directly from the 2D column without CT, (ii) trapped and released to 2D after 1 min, or (iii) trapped after multiple injections then released to 2D in a subsequent run. Quantitative Comparison for Caffeine and Menthol Enrichment vs Number of Injections. Caffeine and menthol standards at 100 μg mL 1 were used with column set 1. To isolate caffeine, the oven program was 150 °C, ramp at 25 °C/min to 230 °C, hold for 7 min; for menthol, 80 °C, ramp to 180 at 12 °C/min. The H-C compound was accumulated in the CT zone after multiple injection cycles (1 50 times) prior to its release in the final quantification run. In-Oven Enrichment of Odor-Impact Compounds from Essential Oils. Peppermint oil (0.2%) was chromatographed using column set 2 with an oven program of 60 °C for 1 min, ramp to 180 at 12 °C/min. The menthol peak from each peppermint oil injection was transferred and enriched in the CT and assessed by its total 2D peak response after conclusion of the repeated injection sequence. Both TP and NA compounds in lavender oil (1.0%) were targeted as low-abundance components for enrichment using column set 1 using an oven program of 80 °C for 1 min, ramp to 200 at 12 °C/min. After a sufficient number of injections, the oven was cooled to 100 °C, the compounds were released by moving the CT, and the oven was ramped at 12 °C/min to 175 °C. In-Oven Enrichment of Menthol from Peppermint Oil Sampling with Headspace (H/S) SPME. Five microliters of oil was placed in a 4 mL H/S vial held at 25 °C throughout the study. The conditioned SPME fiber was exposed to the H/S for 5 min, and the fiber was retracted and introduced into the GC inlet for 5 min. SPME desorption was at 250 °C, in splitless mode for 2 min. Column set 2 was used, and the oven program was 60 °C for 2 min, ramp to 180 at 12 °C/min. Column backflushing was performed at 60 psi for 2 min after each injection/elution of menthol. After the desired number of SPME injections, the oven was cooled to 100 °C, the compound was released by moving the CT, and the oven was ramped to 175 at 12 °C/min.

’ RESULTS AND DISCUSSION Capability for Alkane and FAME Trapping. Solute trapping and remobilization in GC has been successfully achieved using a LMCS system that is conceptually simple and well served by a wide range of applications.2,9,15,23 A temperature-adjustable cylindrical trap moving longitudinally relative to the capillary column that passes through its cooled region permits unique modulation effects. Collection of a volatile compound in an open-tubular capillary is feasible, provided adequate trapping temperature (cooling) has been set. Figure 2 shows results for solute transfer and trapping experiments of an alkane and FAME suite. System balance for heart-cut operation without leakage to the other channel is confirmed (Supporting Information Figure S1). Decane (Figure 2A), was diverted (FID1) to the CT and the 2 D column, but breakthrough of decane occurs, as evidenced by the decane bleed peaks emerging from the cold trapping region in the FID2 trace. The total area of the small baseline fluctuations is 231 pA 3 s, close to that of the C10 peak in the absence of trapping (208 pA 3 s). For tridecane (Figure 2B), complete transfer (FID1 trace) and essentially full trapping of C13 (FID2) arises, with excellent peak shape shown in the inset. Here, the C13 peak is retained in the CT region about 1.5 min prior to moving the CT,

ARTICLE

with effective narrowing of the C13 peak due to rapid delivery from the CT to the 2D column. Beyond C13, all alkanes are well trapped. Figure 2C shows C6 FAME with some evidence (FID2 trace) of incomplete trapping. The C10 FAME is apparently well trapped (Figure 2D, FID2 trace). Figure 2E reports the FID1 result for the FAME reference sample but with 3 components (C10, C13, and C17:0 FAME) H-C to 2D and CT. This single injection produced the result in Figure 2F after trapping then eluting the three FAME on the 2D column to FID2. With three injections and eluting the FAME after the third injection, the response increased about 3-fold (Figure 2G, FID2). Relative responses and retentions are well reproduced for Figure 2F and G, with neither preferential loss of the lighter C10 component nor total collected amount, even though the oven was cycled over its temperature range a number of times. The quantitative trapping efficacy of the combined effect of phase type, thickness, and temperature for other classes of compounds can be established. The work of Nojima16 is a useful guide; with an uncoated (deactivated) capillary trap device at room temperature (RT) external to the oven, a C14 ester, C16 hydrocarbon, and C17+ alcohol could be trapped. Lighter compounds could not be quantitatively trapped. A 0.5 μm df methyl siloxane phase capillary trap improved trapping such that a C9 ester, C12 hydrocarbon, and C9 alcohol could be collected. Thicker films progressively shift trapping capabilities to lower molar mass compounds. Blomberg and Roeraade24 proposed that the retention volume of a trapped component on a phasecoated capillary is proportional to the retention factor (k), which can be increased by lowering the phase ratio or cooling the trap. Multiple Injection Trapping and Enrichment; Caffeine and Menthol. The caffeine peak from a single split injection (5 ng on column) was passed directly through the 1D-DFS arrangement (area and height 20 pA 3 s and 3.4 pA respectively), then the DS was switched just prior to caffeine elution from the end of 1D to pass through the dual column 1D 2D ensemble, giving responses 20 pA 3 s and 2.4 pA, respectively. Caffeine was injected multiple times (1, 5, and 10) to test the linearity of the trapping procedure and confirm no breakthrough. Figure S2A and B (Supporting Information) reports the good balance of the system. Caffeine is a solid at RT; it should be well trapped at the CT temperature. Table 1 reports respective data for both area and height of the trapped peaks, and the results of elution of different collected injected masses of caffeine (5 50 ng) is shown in Figure S2C. No internal standard (IS) was used for this study. A constant level IS could be incorporated into all mixtures and heart-cut to the CT or monitored on FID1 to gauge injection reproducibility. The area correlation data, giving a coefficient of 1.000, demonstrates the excellent recovery of the multiple collected injections. Peak shape was good on the IL column, with an asymmetry value around 0.83 (estimated at 10% peak height). In contrast to the alkanes, caffeine is strongly retained in the 2D column (retention factor of 4.35), so the broader peak means that the height increase is not as large as for alkanes or FAME. The linear correlation coefficient for height data was still good, at 0.998. Menthol (5 ng injected mass) revealed a rather more tailing peak on the IL column (asymmetry value of 0.41; no CT), and system balance is confirmed (Figure S3A and B of the Supporting Information). Cryotrapping improves peak shape, since the effect of dispersion on 1D is negated, and the series of traces in Figure 3 show results for 1, 10, 20, and 50 repeat injections of menthol standard (for which 5, 50, 100, and 250 ng 6487

dx.doi.org/10.1021/ac200973z |Anal. Chem. 2011, 83, 6485–6492

Analytical Chemistry

ARTICLE

Figure 2. Chromatograms of C10 C20 alkanes and C4 C24 FAME mixture to test CT trapping capability. (C4 FAME is coeluted with solvent front, while late-eluting FAME are back-flushed.) (A) Decane switched to the 2D column, with cryotrapping; lower plot FID1, upper plot FID2 response. (B) Tridecane switched to the 2D column, with cryotrapping. (C) Methyl hexanoate switched to the 2D column, with cryotrapping. (D) Methyl decanoate switched to the 2D column, with cryotrapping. Insets show expanded plots to display baseline response on FID2. (E) 1D-DFS-FID1 response of the FAME mixture with C10, C13, and C17:0 methyl esters switched to the 2D column. (F) FID2 response for elution of the cryotrap for one injection of FAME sample. (G) FID2 response for elution of the cryotrap for three injections of FAME sample.

amounts were collected). Table 1 reports a linear correlation of 0.996 for area data and 0.996 for height data fitted to Y = mXn. Results in Table 1 suggest that peak height response vs number of injections correlates better with a power relationship, presumably because of peak broadening, overloading with increased mass due to the relatively high phase ratio of the 2D column, or both. The area comparison for 1 with 50 injections is close to the expected value of 50 (42 vs 1818 pA 3 s). A recovery of over 75% (and often greater than 92% for other solutes) was estimated on the basis of comparison of the area of the 10 injection result

compared with the single injection multiplied by 10, suggesting good efficiency for this enrichment approach. Repeat Trapping for Enrichment of Flavor Compounds from Essential Oil Samples. Menthol is the most abundant compound in the essential oil of peppermint (∼44% estimated by total GC response, or 0.8 μg in 0.2% oil according to external standard calibration). The above approach can be applied to menthol in peppermint. Figure 4A is the 1D-DFS-FID1 response of the whole peppermint sample (menthol asterisked, confirmed by GC-MS and tR). In Figure 4B, FID1 response shows that 6488

dx.doi.org/10.1021/ac200973z |Anal. Chem. 2011, 83, 6485–6492

Analytical Chemistry

ARTICLE

Table 1. Quantitative Evaluation of Trapped Compound Amount (area and height) vs Number of Injections standard compound caffeine number of injections, X

area, Y1

1

20

3

61

height, Y2 3.4

peppermint oil

menthol area, Y1 42

lavender oil

menthol

height, Y2

area, Y1

height, Y2

6459

1739

45

11

R-terpineol

neryl acetate area, Y1

height, Y2

area, Y1

height, Y2

24

20

58

48

67

50

158

120

5

98

17

165

130

114

73

230

158

10

198

33

324

248

223

130

531

330

15

87752

20

631

393

50

1818

3989c

regression a recovery

0.999 99%

0.998b

0.996 77%

0.996b 89%

6837

0.999 93%

0.999b

0.991 92%

0.991b

a

Recovery estimated as percent obtained from peak area value of 10 injections/10 times peak area value of 1 injection, or 15 injections for peppermint oil. b Regression obtained from the relationship Y2 = mXn. c Data point omitted due to change in experimental conditions for elution of the trapped peak, which varied the height; area data not affected .

Figure 3. FID2 response for elution of the cryotrap, for 1, 10, 20, and 50 injections of menthol as indicated.

menthol has been effectively H-C from the sample. Figure 4C and D show 1 and 15 repeat injections of the 0.2% peppermint sample (∼0.8 and 12.0 μg of menthol injected mass, respectively). Retention time is ∼11.33 min on the 1D column and 2.69 min using the temperature program for the 2D column elution. The large mass of menthol (∼12 μg) collected in the 15 trapping process leads to an overloaded peak for menthol on the 2D column and the expected increase in tR of the peak maximum. This would be similar to injecting a 3% solution of peppermint without splitting onto the column, which apparently would produce overloading. The collected mass range here would be sufficient for GC SERS detection of 50 ng DL (of benzene),25 GC FTIR verification at 15 ng for nitrogen mustard derivatives,26 and potentially adaptable for current online GC NMR analysis11 that required 10s of micrograms of solute. In all cases, a higher phase load on the 2D column would reduce the tendency for peak overloading. The 1 and 15 experiments produced excellent quantitative recovery of the

trapped menthol, with a 15-fold increase in areas (6459 vs 87752 pA 3 s, respectively). Figure 5A records menthol (indicated by asterisk) recovered by using SPME with splitless injection from 5 μL of mint oil equilibrated in a 4 mL closed vial. The DS effectively cuts menthol (Figure 5B) to 2D, with the absence of menthol in 1 D. Figure 5C and D reports chromatograms of 1 and 5 trapping of the selected menthol zone. The peak response for menthol after 5 trapping was 110617 pA 3 s; that is, an ∼5 increment in peak area compared with the single trapping result (19 141 pA 3 s). One goal of this study was to improve general detection of lowlevel compounds. Lavender oil was studied to increase the in situ response of minor components TP and NA (∼1.05% and 0.42% in the oil, respectively), which were targeted for analysis using a novel switchable GC  GC/MDGC system described recently.27 Figure 6A is the 1D-DFS-FID1 result (to estimate DS switching times for TP and NA); TP and NA have a relative 6489

dx.doi.org/10.1021/ac200973z |Anal. Chem. 2011, 83, 6485–6492

Analytical Chemistry

ARTICLE

Figure 4. Chromatograms of 0.2% peppermint. (A) 1D-DFS-FID1 response of whole peppermint sample; menthol indicated by asterisk. (B) 1D-DFSFID1 response of peppermint sample with menthol heart-cut to 2D. (C) FID2 response for elution of the cryotrap for a single injection of peppermint (1 collection of menthol). (D) FID2 response for elution of the cryotrap for 15 injections of peppermint (15 collections of menthol).

Figure 5. Chromatograms of peppermint volatiles H/S sampled by SPME. (A) 1D-DFS-FID1 response of peppermint; menthol is indicated by asterisk. (B) 1D-DFS-FID1 response of peppermint with menthol heart-cut to 2D. (C) FID2 response for elution of the cryotrap for a single collection of menthol. (D) FID2 response for elution of the cryotrap for 5 collections of menthol.

response of ∼2.5:1. Figure 6B shows both TP and NA quantitatively transferred to the 2D column and CT. The sequence from

Figure 6C1 C4 illustrates 1, 3, 5, and 10 repetitive injections of lavender with elution of TP and NA. In each case, 6490

dx.doi.org/10.1021/ac200973z |Anal. Chem. 2011, 83, 6485–6492

Analytical Chemistry

ARTICLE

Figure 6. Chromatograms of 1% lavender with collected R-terpineol and neryl acetate. (A) 1D-DFS-FID1 response of whole lavender sample; TP and NA indicated by first and second asterisk, respectively. (B) 1D-DFS-FID1 response of lavender sample with TP and NA heart-cut to 2D. (C) FID2 response for elution of the cryotrap for C1, 1 (i.e., 1 collection of TP + NA); C2, 3; C3, 5; C4, 10 injections of lavender.

the ratio of TP/NA is ∼2.4, so there is no bias between collection of the two peaks. Comparison of 1 and 10 injection results for TP and NA shows a greater than 9-fold area increase for the 10 injection. Heights are only 6 7-fold greater. Multiple injections but with varied DS switching during the sequence allows collection or omission of different peaks. For instance, if it was desired to enrich a particular (e.g., abundant) compound a few times (so that it does not suffer overloading) but to collect a minor component over an extended number of injections, this simply requires developing a different event table and operating the DS under this timed switching arrangement. For most compounds reported here, linear regressions (Table 1) are shown for area correlations with the number of

injections, but a Y = mXn relationship fits height data better (Figure S4 of the Supporting Information), suggesting nonlinearity due to peak broadening at higher injected mass (increasing number of trapping events). These data support the quantitative H-C and trapping performance of the approach for compound enrichment from complex samples. Trapping is demonstrated to be successful for diverse functional groups and over a range of volatility. By increasing the mass load of the components, without increasing matrix interferences that might arise from preconcentration methods or large volume injection techniques, the MDGC system maintains good resolution that is suitable for spectroscopic characterization. As an example application, recent work28 demonstrated an off-line preparative packed 6491

dx.doi.org/10.1021/ac200973z |Anal. Chem. 2011, 83, 6485–6492

Analytical Chemistry column GC (5  80 μL) approach with eight collected gradient temperature fractions to increase the abundance of trace compounds in spirits for improved GC/MS response of trace compounds. Final extracts were made up in 25 μL CH2Cl2 (16-fold concentration factor) for capillary GC/MS analysis. Such a procedure could potentially be conducted automatically using the approach described here.

’ CONCLUSION Automated operation of the MDGC system provides online and in situ enrichment of compounds isolated by use of a microfluidic DS, directed to a 2D column and cryotrapped. Multiple collections of target peaks allow compound mass to be increased in the capillary column prior to launching the enriched compound into the downstream 2D column. Correlation of the number of injections with peak response demonstrated linear area data for the compounds over a range of up to 50 repeat injections. This method will be suitable for increasing the detectability of compounds that have a low S/N response or for detectors that require a larger injected mass of compounds to provide adequate identification. This should be especially the case in which preconcentration of the sample is either impractical or is limited by matrix effects. Compounds to be enriched should be sufficiently nonvolatile at the temperature of the CT to permit collection over the duration of repeated injections. An IL column phase was used for most studies, and a DB-1301 phase was used for the SPME experiment. Other stationary phases and film thicknesses will vary the trapping efficacy accordingly, potentially extending the trapping range to higher volatility compounds. The method may find application to olfactometric detection in which a region of a sample possesses an odor-impact activity, but no peak can be found on a GC trace; enrichment of this region may reveal otherwise unidentified components. Although the method allows a regular film thickness 1D column to be used, which should permit high-resolution separation and isolation once a high-abundance solute is H-C to the 2D column from multiple injections, it is likely to overload the phase (as shown in some examples here) unless a thick film column is employed. ’ ASSOCIATED CONTENT

bS

Supporting Information. Table S1: Additional operational procedures for MDGC and online enrichment of target compounds. Figure S1: Alkane and FAME balance data. Figure S2: Caffeine DS balance and multiple injection data. Figure S3: Menthol DS balance data. Figure S4: Calibration graphs of multiple injections of standard and samples for both area and height data. This material is available free of charge via the Internet at http://pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*WCU Distinguished Visiting Professor, Chung-Ang University. Phone: +61 3 99059630. Fax: +61 3 99058501. E-mail: [email protected].

ARTICLE

Government for a Ph.D. scholarship. The work reported here is part of our association with the Australian Centre for Research on Separation Science (ACROSS).

’ REFERENCES (1) Scott, R. P. W. Chromatographic Detectors: Design, Function, and Operation; Marcel Dekker, Inc: New York, 1996; Vol. 70. (2) Eyres, G. T.; Urban, S.; Morrison, P. D.; Dufour, J.-P.; Marriott, P. J. Anal. Chem. 2008, 80, 6293–6299. (3) Nojima, S.; Schal, C.; Webster, F. X.; Santangelo, R. G.; Roelofs, W. L. Science 2005, 307, 1104–1106. (4) Brinkman, U. A. T., Hyphenation: Hype and Fascination; Elsevier, 1999. (5) Wilson, I. D.; Brinkman, U. A. T. J. Chromatogr., A 2003, 1000, 325–356. (6) Eyres, G. T.; Urban, S.; Morrison, P. D.; Marriott, P. J. J. Chromatogr., A 2008, 1215, 168–176. (7) Ochiai, N.; Sasamoto, K. J. Chromatogr., A 2011, 1218, 3180–3185. (8) Eglinton, T. I.; Aluwihare, L. I.; Bauer, J. E.; Druffel, E. R. M.; McNichol, A. P. Anal. Chem. 1996, 68, 904–912. (9) Ruehle, C.; Eyres, G. T.; Urban, S.; Dufour, J.-P.; Morrison, P. D.; Marriott, P. J. J. Chromatogr., A 2009, 1216, 5740–5747. (10) Ruehle, C. P. G.; Niere, J.; Morrison, P. D.; Jones, R. C.; Caradoc-Davies, T.; Canty, A. J.; Gardiner, M. G.; Tolhurst, V.-A.; Marriott, P. J. Anal. Chem. 2010, 82, 4501–4509. (11) Kuehnle, M.; Kreidler, D.; Holtin, K.; Czesla, H.; Schuler, P.; Schaal, W.; Schurig, V.; Albert, K. Anal. Chem. 2008, 80, 5481–5486. (12) Chen, P. C.; Joyner, C. C.; Patrick, S. T.; Royster, R. M.; Ingham, L. L. Anal. Chem. 2003, 75, 3066–3072. (13) Cawon, K. T.; Kennedy, B. J. Anal. Chem. 1995, 67, 3353–3356. (14) Haefliger, O. P.; Jeckelmann, N.; Ouali, L.; Leon, G. Anal. Chem. 2009, 82, 729–737. (15) Marriott, P. J.; Kinghorn, R. M. Anal. Chem. 1997, 69, 2582–2588. (16) Nojima, S.; Apperson, C. S.; Schal, C. J. Chem. Ecol. 2008, 34, 418–428. (17) Schomburg, G.; Husmann, H.; Weeke, F. J. Chromatogr., A 1975, 112, 205–217. (18) Bicchi, C.; D’Amato, A.; Frattini, C.; Nano, G. M.; Pisciotta, A. J. High Resolut. Chromatogr. 1989, 12, 705–708. (19) Roeraade, J.; Blomberg, S.; Pietersma, H. D. J. J. Chromatogr. 1986, 356, 271–284. (20) Adahchour, M.; Beens, J.; Vreuls, R. J. J.; Brinkman, U. A. T. TRAC, Trends Anal. Chem. 2006, 25, 540–553. (21) Milman, G.; Barnes, A. J.; Lowe, R. H.; Huestis, M. A. J. Chromatogr., A 2010, 1217, 1513–1521. (22) MacNamara, K.; Leardi, R.; Hoffmann, A. LCGC North Am. 2004, 82–91. (23) Marriott, P. J.; Eyres, G. T.; Dufour, J.-P. J. Agric. Food Chem. 2009, 57, 9962–9971. (24) Blomberg, S.; Roeraade, J. J. Chromatogr. 1987, 394, 443–453. (25) Chen, P. C.; Cottingham, K. A.; Royster, R. M.; Joyner, C. C. Appl. Spectrosc. 2005, 59, 1310–1314. (26) Garg, P.; Purohit, A.; Tak, V. K.; Dubey, D. K. J. Chromatogr., A 2009, 1216, 7906–7914. (27) Maikhunthod, B.; Morrison, P. D.; Small, D. M.; Marriott, P. J. J. Chromatogr., A 2010, 1217, 1522–1529. (28) Ledauphin, J.; Saint-Clair, J.-F.; Lablanquie, O.; Guichard, H.; Founier, N.; Guichard, E.; Barillier, D. J. Agric. Food Chem. 2004, 52, 5124–5134.

’ ACKNOWLEDGMENT S.-T.C. acknowledges the provision of a Dean’s postgraduate scholarship from the School of Chemistry and Faculty of Science, Monash University. B.M. gratefully thanks the Royal Thai 6492

dx.doi.org/10.1021/ac200973z |Anal. Chem. 2011, 83, 6485–6492