Quasi-Simultaneous Acquisition of Hard Electron Ionization and Soft

Apr 5, 2011 - Single-Photon Ionization Mass Spectra during GC/MS Analysis by ... electron ionization (EI) and soft single-photon ionization (SPI) inte...
1 downloads 0 Views 5MB Size
Download PDF
ARTICLE pubs.acs.org/ac

Quasi-Simultaneous Acquisition of Hard Electron Ionization and Soft Single-Photon Ionization Mass Spectra during GC/MS Analysis by Rapid Switching between Both Ionization Methods: Analytical Concept, Setup, and Application on Diesel Fuel Markus S. Eschner,†,‡ Thomas M. Gr€oger,†,‡ Thomas Horvath,§ Marc Gonin,§ and Ralf Zimmermann*,†,‡,^ †

Joint Mass Spectrometry Centre, Cooperation Group “Analysis of Complex Molecular Systems”, Institute of Ecological Chemistry, Helmholtz Zentrum M€unchen, Ingolst€adter Landstrasse 1, 85764 Neuherberg, Germany ‡ Joint Mass Spectrometry Centre, Chair of Analytical Chemistry, Institute of Chemistry, University of Rostock, Dr.-Lorenz-Weg 1, 18059 Rostock, Germany § Tofwerk AG, Uttigenstrasse 22, 3600 Thun, Switzerland ^ BIfA—Bavarian Institute of Applied Environmental Research and Technology GmbH, Am Mittleren Moos 46, 86167 Augsburg, Germany ABSTRACT: This work describes the realization of rapid switching between hard electron ionization (EI) and soft single-photon ionization (SPI) integrated in a compact orthogonal acceleration time-of-flight mass spectrometer. Vacuum-ultraviolet (VUV) photons of 9.8 eV (126 nm) emitted from the innovative electron-beampumped rare-gas excimer light source (EBEL) filled with argon are focused into the ion chamber by an ellipsoidal mirror optic for accomplishing of SPI. This novel orthogonal acceleration time-of-flight mass spectrometer with switching capability was hyphenated to one-dimensional gas chromatography (GC) and comprehensive two-dimensional (2D) gas chromatography (GC  GC) for the first time. Within this demonstration study, a maximum switching frequency of 80 Hz was applied for investigation of a mineral-oil-type diesel sample. This approach allows the quasi-simultaneous acquisition of complementary information about the fragmentation pattern (EI) as well as the molecular mass (SPI) of compounds within a single analysis. Furthermore, by application of a polar GC column for separation, the SPI data can be displayed in a 2D contour plot, leading to a comprehensive 2D characterization (GC  MS), whereas the typical group-type assignment for diesel is also met.

T

he hyphenation of gas chromatography and mass spectrometry (GC/MS) represents a powerful tool for separating and identifying organic compounds. Conventionally, quadrupole mass spectrometers or magnetic sector instruments were utilized for GC/MS in the beginning.1,2 Nowadays, time-of-flight (TOF) mass spectrometers are gaining more and more attention for coupling with GC because of their high sensitivity, large mass range, and fast acquisition rates.3 With final acquisition frequencies of 100 Hz or even more, TOF mass spectrometers are especially suited as a detector for comprehensive two-dimensional (2D) gas chromatography (GC  GC).4,5 Moreover, orthogonal acceleration TOF (oaTOF) mass spectrometers allow the design of ultracompact instruments while maintaining a remarkable mass resolution. Furthermore, oaTOF mass spectrometry (oaTOFMS) instruments are especially suited for continuous ionization sources like electron ionization (EI).6 Typically, electrons emitted from a glowing filament are accelerated to a kinetic energy of 70 eV for ionization. Consequently, the typical ionization energies of 710 eV for organic molecules are exceeded and, thereby, appearance energies of intrinsic fragmentation pathways are reached. Hence, usually fragment ions are detected primarily, and the molecular ion (Mþ) is detected with minor intensity. r 2011 American Chemical Society

Therefore, EI is considered to be a hard ionization method. Fortunately, this ionization process is highly reproducible, yielding characteristic fragmentation patterns, which are used as fingerprints for the structural elucidation or identification of compounds via pattern recognition and statistical comparison with EI mass spectral libraries. However, this identification process can be hindered by a lack of the molecular-ion signal and/or coelution of multiple compounds, making proper deconvolution of the individual mass spectra very challenging. A rather promising approach is either to improve the separation power of the chromatographic system, e.g., by GC  GC, or to enhance the selectivity of the mass spectrometer by application of soft ionization methods, such as chemical ionization (CI),7,8 field ionization (FI)9,10 or photoionization (PI). The latter is comprised of resonance-enhanced multiphoton ionization (REMPI),11 ideally suited for aromatic profiling, and the more universal singlephoton ionization (SPI).1217 All of these soft ionization methods are similar in causing little to no fragmentation of the Received: February 10, 2011 Accepted: April 5, 2011 Published: April 05, 2011 3865

dx.doi.org/10.1021/ac200356t | Anal. Chem. 2011, 83, 3865–3872

Analytical Chemistry molecular ion. Therefore, molecules with different masses can be separated according to their molecular-ion masses within the same mass spectrum resembling a boiling point separation in classical GC using nonpolar columns. Furthermore, all of these soft ionization techniques have been applied as ionization sources in GC/MS instruments.1825 Recently, the application of PI methods in GC/MS analysis was reviewed,26 demonstrating the analytical usefulness of soft ionization MS in conjunction with GC. As was already mentioned, characterization of unknown compounds within a complex sample by GC/MS can be improved if, in addition to EI, also soft ionization methods are applied. This is particularly the case if the EI spectrum is minor in quality due to coelution, lack of the molecular ion, or little concentration of the analyte of interest. Therefore, it is very helpful to confirm a nonsatisfactory library match by determination of the molecular mass using one of the above-mentioned soft ionization techniques. Accordingly, many commercial GC/MS instruments nowadays are available with CI or FI sources in addition to standard EI. A further interesting feature of combining soft ionization methods with GC/MS was presented in 2005 by Wang et al.27 utilizing FI and in 2006 by Mitschke et al.28 utilizing laser-based PI. Because soft ionization MS resembles a volatility-type separation and the chromatographic separation is mainly due to polarity differences, a reasonably comprehensive 2D separation approach (GC  MS) could be realized. Additionally, the orthogonality of the two separation axes was increased by application of a retention time shift algorithm, resulting in a linear alignment of the normal alkanes. After transformation, a similar compound class grouping well-known for diesel could be clearly recognized and was comparable to GC  GC results. These ordered chromatograms simplify compound identification and allow the rapid screening of unknown samples. Qian et al.29 and Hejazi et al.30 employed GC  FIMS for a comprehensive 2D investigation of mineral oil and fatty acid methyl ester containing samples, respectively. Furthermore, comprehensive three-dimensional (3D) separation approaches (GC  GC  MS) were realized by the combination of GC  GC and laser-based SPI-MS31 as well as by utilization of the innovative vacuum ultraviolet (VUV) electron-beam-pumped rare-gas excimer light source (EBEL) for SPI-MS.32 Note that this comprehensive representation is only possible with real soft ionization methods like FI and PI, generating almost exclusively molecular ions. However, until now always two different GC/MS analyses have to be performed in order to compare the MS results yielded by soft and hard ionization. Unfortunately, even in two successive GC runs, small differences in the column performance may occur as a result of the variable column pressure or oven temperature, leading to slightly different retention times. This requires ingenious peak alignment algorithms, which can be avoided if rapid switching between the soft and hard ionization methods within the same analysis run are possible. Moreover, by application of switching, the analysis time can be reduced by 50% compared to consecutive runs. Automated switching between EI, laser-based SPI, and REMPI for online monitoring of the mainstream tobacco smoke was reported by M€uhlberger et al. in 2004 for the first time.33 However, in this approach only a maximum of one mass spectrum for each ionization method was obtained every second, being actually to slow for GC/MS hyphenation. In the present work, rapid switching with a frequency of 10 Hz between hard EI and soft EBEL-based SPI was realized within a single GC/MS analysis for the first time. Both confirmation of

ARTICLE

the MS library matches and a comprehensive 2D GC  MS illustration of a mineral-oil-type sample are presented in the following. Because an ultrafast oaTOFMS is utilized, this switching technique was also applied to GC  GC. In general, this approach represents an important step toward a deeper characterization of complex samples using quasi-simultaneously soft and hard ionization MS hyphenated to GC separation.

’ EXPERIMENTAL SECTION All experiments were carried out on a previously described GC/SPI-TOFMS instrument with GC  GC capability,32 consisting of a Trace GC  GC (Thermo Scientific, Milan, Italy) and an oaTOFMS (Tofwerk AG, Thun, Switzerland) with a mass resolution of 800. The diesel fuel investigated was purchased from Sigma-Aldrich (Steinheim, Germany). GC Methods. In the one-dimensional (1D) approach, GC separation was performed on a 30 m  250 μm  0.2 μm ionic liquid column SLB-IL59 (Supelco, Bellefonte, PA). A sample volume of 0.1 μL was injected in split mode (1:20) at 250 C. The oven temperature was programmed to increase from 40 to 300 C at a 5 C/min heating rate, resulting in a total run time of 52 min. The pressure program was as follows: 200 kPa ramped with 5 kPa/min to 460 kPa. For comprehensive 2D GC separation, the dual-jet liquid CO2 modulator was operated with a modulation period of 6 s. The same column as that in the 1D GC/MS approach was installed in the first dimension, connected to a 1 m  100 μm RTX-200 column (Restek, Bad Homburg, Germany) with a film thickness of 0.1 μm as the second dimension. The oven-temperature program was ramped from 40 to 300 C with a 4 C/min heating rate (total run time: 65 min). The pressure was programmed from 270 kPa with an increase of 3 kPa/min to a final pressure of 465 kPa. Again, a sample volume of 0.1 μL was injected in split mode (1:20) at 250 C. MS and Ionization Methods. Figure 1a exhibits an engineering drawing of the oaTOF mass spectrometer equipped with EI and EBEL-based SPI sources. For EI, electrons with a kinetic energy of 70 eV were emitted from a glowing tungsten filament. The functionality of the compact VUV light source is described in detail elsewhere.34,35 Thus, only a very brief description is given here. High energetic electrons (12 keV) excite and ionize Ar atoms separated from the vacuum region by a 300 nm thin SiNx membrane. The thereby formed excimer molecules decay under emission of VUV radiation with a central wavelength of 126 nm (9.8 eV). These VUV photons are collected by an ellipsoidal mirror and focused into the ionization chamber of the oaTOF mass spectrometer (Figure 1b). The VUV beam (purple) is arranged orthogonally to the electrons emitted from a tungsten filament (yellow), with both being placed perpendicularly to the effusive outlet (brown) of a deactivated transfer line capillary (0.5 m  100 μm), permanently heated to 270 C (see Figure 1c). Ions (red) are subsequently pulled/pushed out of the ionization region and guided to the orthogonal extractor by electronic lenses. The extraction frequency of the oaTOF mass spectrometer was set to 80 kHz to cover a mass range of m/z 35340. Data Acquisition and Realization of Switching. The MS stores the acquired data in hierarchical data format (abbreviated to “hdf5”). The hdf5 file format was developed for handling extremely large amounts of scientific data under optimized storage conditions by the HDF group,36 i.e., well suited for saving several thousands of TOFMS spectra. The hdf5 file consists of 3866

dx.doi.org/10.1021/ac200356t |Anal. Chem. 2011, 83, 3865–3872

Analytical Chemistry

ARTICLE

Figure 1. (a) Engineering drawing of the oaTOF mass spectrometer equipped with EI and EBEL-based SPI sources, enabling rapid switching between hard and soft ionization. (b) Engineering drawing of the innovative EBEL light source with an ellipsoidal mirror optic for focusing of the VUV beam. (c) Enlarged view of the ionization region and the arrangement of the ionization techniques. (d) Schedule for the quasi-simultaneous use of EI and EBELbased SPI for MS analysis. The emission current of the filament (red) is plotted on the left y axis, and the emission current of the EBEL (blue) is displayed on the right y axis.

two primary data objects: groups and data sets. Groups are, in principle, organizing folders in which the different data sets are located. Data sets are a multidimensional array of data elements. During MS analysis, the integrated and full MS spectra are stored in four-dimensional (4D) arrays as follows: time vector 1  time vector 2  segments  mass spectra. For 1D GC, time vector 1 and time vector 2 are multiplied to get the retention time, whereas for comprehensive 2D GC, time vector 1 equals the first dimension retention time and time vector 2 equals the modulation period. For rapid switching between the two ionization methods, two segments are used in order to separate the EIMS spectra from the SPI-MS spectra in the 4D array. Figure 1d depicts the schedule for quasi-simultaneous use of EI and EBELbased SPI for MS analysis. The emission current of the filament (red) is plotted on the left y axis, and the emission current of the EBEL (blue) is displayed on the right y axis. During the whole measurement, the excimer VUV light source was operated at 5 μA (yielding about 1013 VUV photons37). Because the total ion current (TIC) of an EI measurement is about 3 orders of magnitude higher than that with SPI, the minor contribution of SPI ions was neglected in this proof-of-concept study.

However, the SPI intensity contribution in each EI segment can be eliminated by subtraction of the signal intensity of the subsequent SPI segment. Furthermore, the EBEL technology also allows switching of the VUV light. For generation of an EIMS spectrum, an emission current of 200 μA is maintained. During the SPI segment, the voltage of the filament, typically 30 V, is drawn to the ion chamber voltage (typically þ40 V) by a fast electronic controller to avoid acceleration of electrons, thus blocking EI for this time. At the EI mode, the filament voltage is set to 30 V again to ensure acceleration of the electrons into the ionization chamber. Figure 1d also clarifies the sequence of acquired MS spectra. All odd-numbered MS spectra are generated by EI and saved in segment 1, while all even-numbered MS spectra are obtained by SPI and stored in segment 2. The measurement time for both ionization methods is indicated by the time index bar, whereas the reciprocal value of the segment time equals the switching frequency. The segment time for 1D GC was 100 ms (10 Hz spectral acquisition/switching frequency), resulting in an effective acquisition rate of 5 Hz for EI and SPI, respectively, while for GCGC, switching was performed with 80 Hz (12.5 ms segment time). 3867

dx.doi.org/10.1021/ac200356t |Anal. Chem. 2011, 83, 3865–3872

Analytical Chemistry

ARTICLE

Figure 2. Summed MS spectra of a diesel sample by switching between EI and SPI within a GC/MS analysis: (a) EI-MS spectrum obtained by summing over the total GC run (all EI segments) using 70 eV electrons; (b) SPI-MS spectrum (raw data) obtained by summing over all SPI segments using 9.8 eV photons generated by EBEL; (c) cross-section-corrected SPI-MS spectrum indicating the real molar concentration profile of the mineral oil distillate sample.

Data Postprocessing. The recorded hdf5 files were directly loaded into MatLab 7.11 R2010b (The Mathworks, Natick, MA) without any prior transformation. Subsequent data processing was performed on a personal computer (2.8 GHz CPU, quad core, Windows 64-bit, 8 GB RAM), employing a MatLab script written in-house. All presented figures were further treated in an Adobe Illustrator CS3 13.0.2 (Adobe System Inc., San Jose, CA) to generate publication-ready images.

’ RESULTS AND DISCUSSION Application of Rapid Switching between EI and EBELBased SPI for 1D GC/MS Analysis. SPI using VUV photons with

energies of around 10 eV enables universal and soft ionization of organic compounds, while most matrix gases, such as nitrogen or oxygen, are excluded from ionization. These inherent characteristics of SPI and the development of the compact, rugged, and brilliant excimer light source EBEL led to many online MS applications.34,3742 In the meantime, the first SPI-MS systems employing this innovative VUV light source are commercially available (Photonion GmbH, Schwerin, Germany). Now, this work highlights a further interesting and analytically useful feature when combining SPI and EI, i.e., rapid switching between these complementary ionization methods. By application of a homemade EBEL, a detection limit of 15 pg in the SPI mode (35 fg in the EI mode) for the diesel constituent phenanthrene, based on a signal-to-noise ratio of 3:1, was determined at a switching frequency of 10 Hz. However, this limit of detection, sufficient for the demonstration study presented, can be further lowered by using a commercial version of EBEL. For this concept study, a mineral-oil-type diesel sample was applied as a test matrix because of its ideal boiling range for GC separation and the content of a huge variety of compounds comprising saturates (alkanes and naphthenes) as well as mono-, di-, and triaromatics. Furthermore, it is a well-studied mixture because many different investigations have been accomplished on this type of fuel, being very well-referenced.19,27,32,43,44

First of all, to demonstrate the analytical suitability and the separation character of EBEL-based SPI-MS, Figure 2 displays three MS spectra of the diesel sample obtained by summing over all EI and SPI segments, whereas in panel c, the SPI raw data are cross-section-corrected. Thus, in principle, the same MS spectra should be obtained if the diesel sample is totally vaporized and analyzed all at once as a nonseparated mixture by this MS system. Figure 2a exhibits the 70 eV EI-MS spectrum dominated by hydrocarbon fragmentation peaks, revealing that this hard ionization method is not appropriate for online analysis of complex mixtures. Hence, no explicit deconvolution of the molecular composition can be obtained, except the sample primarily consisting of aliphatic hydrocarbons. In contrast, the EBELbased SPI-MS (9.8 ( 0.4 eV) spectrum in Figure 2b shows the molecular-ion profile of the diesel sample. Consequently, a homologous series of several compound classes, such as alkanes, naphthenes, alkylated benzenes, and indanes/tetralins, can be easily assigned because the principle composition of mineral-oiltype diesel fuels is known. Moreover, the VUV radiation band with a central wavelength of 126 nm (9.8 eV) and a full width at half-maximum (fwhm) of 9 nm of the EBEL also allows one to detect smaller alkanes or cycloalkanes, which normally cannot be ionized at a photon energy of 9.8 eV. Separation of compounds within the homologous series of the alkanes, for example, assigned with black stars in Figure 2b, is obtained by a mass difference of 14, corresponding to insertion of a CH2 group. Additionally, marginal alkane fragment signals with typical m/z values of 43, 57, or 71 are discernible in Figure 2b, indicating very little photoelectron-induced fragmentation. PI cross sections of different homologous series typically contained in diesel samples were recently determined by means of the EBEL-based SPI-GC/ MS setup using a nonpolar column for separation and benzene as the standard.45 For generation of the cross-section-corrected SPI-MS spectrum in Figure 2c, the assigned mass peaks of Figure 2b are multiplied by the inverse relative cross section of the individual compounds of the respective homologous series. Now, the molecular-ion intensity profile thus obtained reflects 3868

dx.doi.org/10.1021/ac200356t |Anal. Chem. 2011, 83, 3865–3872

Analytical Chemistry

ARTICLE

Figure 3. Left part: TIC of the EI (red) and cross-section-corrected SPI (blue) modes. Middle part: MS spectra of n-heneicosane for both ionization techniques. Right part: MS spectra of a branched alkane with a molecular mass of 254 for both ionization techniques.

Figure 4. (a) 2D representation of the cross-section-corrected GC/SPI-MS data by plotting the retention time (left panel) versus the molecular mass (upper panel). The two insets show enlarged views of selected aliphatics (blue frame) and alkylated monoaromatics (green frame). (b) 3D representation of transformed GC  MS data (the signal intensity is plotted on the z axis) and assignment of compound class separation. Here, the SPI raw data are displayed, indicating discrimination of the aliphatic compounds because of their reduced PI cross sections compared to aromatic molecules.

the real molar concentration distribution of the diesel constituents quite well. This signal intensity correction procedure for SPI is also applied to each spectrum during the GC separation by considering the retention time window of the particular isobars, e.g., alkanes and naphthalenes. On the left part of Figure 3, the TICs of the 1D GC analysis, recording EI-MS (red) and SPI-MS (blue) segments alternately with an acquisition frequency of 10 Hz, are compared, showing rather the same elution pattern. After application of the cross-section-corrected process, the normal alkanes, which exhibit very low PI cross sections at 9.8 eV, yielding a low detector response, are clearly assignable by SPI also. On the middle part, the MS spectra of n-heneicosane are displayed for both ionization techniques. The EI-MS spectrum exhibits the typical alkane fragment masses with m/z 43, 57, 71, 85, 99, etc., as well as the molecular mass of n-C21H44 at m/z 296 with considerably minor intensity. Because of the advantageous reproducibility of EI, leading to characteristic fragmentation patterns, many MS spectral libraries containing thousands of

molecules are available. By a comparison of the EI-MS spectrum with NIST MS Search v.2.0 software (NIST/EPA/NIH Mass Spectral Library, NIST Scientific and Technical Databases, Gaithersburg, MD), identification of n-heneicosane via pattern recognition is possible. Because this sample was investigated in the switching mode, also an SPI-MS spectrum was recorded at the same retention time, whereby the molecular mass of the n-alkane could be confirmed. The middle and right parts of Figure 3 demonstrate the complementarity of both ionization methods. While the EI-MS spectrum is dominated by fragment ions supplying information about the structure and functional groups of a compound, soft SPI-MS provides the opportunity to determine the molecular mass of a compound and, hence, to verify the MS match of an EI library search. The benefit of switching modes becomes even more obvious in the right part of Figure 3. The EI-MS spectrum of a branched alkane with a retention time of 18.47 min is dominated by the typical fragment masses of alkanes, while the molecular mass cannot clearly be determined. In contrast, the SPI spectrum exhibits a molecular 3869

dx.doi.org/10.1021/ac200356t |Anal. Chem. 2011, 83, 3865–3872

Analytical Chemistry

ARTICLE

Figure 5. (a) TIC of GC  GC analysis and assignment of compound class separation. (b) EI-MS (red) and SPI-MS (blue) spectra of the three different peaks encircled in black in panel a.

mass peak at m/z 254. The thermally induced fragmentation of this very fragile compound can be avoided by applying a supersonic jet inlet instead of the effusive capillary inlet system used.28 As already shown in Figure 2b,c, molecules with different masses can be separated according to their molecular-ion masses within the same MS spectrum using EBEL-based SPI, which resembles a boiling point separation, i.e., separation due to volatility. Therefore, if SPI-MS is combined with GC separation using a polar column, a comprehensive 2D separation approach (GC  MS) can be accomplished. This is illustrated in Figure 4a by plotting the retention time versus m/z. The left panel of Figure 4a depicts the cross-section-corrected TIC showing the typical pattern known from a classical GC/MS analysis (adapted from Figure 3). The upper panel exhibits the cross-sectioncorrected EBEL-based SPI-MS spectrum by summing over all SPI segments, demonstrating the separation power of the soft ionization MS technique (adapted from Figure 2c). The 2D illustration in the middle reflects the reasonably comprehensive 2D character of GC  MS. However, as discussed in depth in ref 32, a temperature dependency of the GC separation remains even when a very polar column is employed. Therefore, the two separation mechanisms, on the one hand, SPI-MS according to

volatility and, on the other hand, GC separation mainly due to polarity differences, are not sufficiently independent of each other, resulting in peak clustering within a homologous series along a diagonal. However, a reasonable peak assignment can already be met, as illustrated by the two insets. The blue-framed inset displays aliphatics ranging from C12 to C14 including n-alkanes, isoalkanes (dark blue), cyclic (green), and bicyclic alkanes (cyan). An enlarged view of the alkylated benzenes (orange) and indanes/tetralins (red) ranging from C10 to C12 is displayed in the green-framed inset. Furthermore, by application of the retention time transformation algorithm defined by Wang et al.,27 an entire orthogonality of both separation dimensions is obtained. The previous retention time axis is converted to a polarity scale with respect to the n-alkanes. Figure 4b exhibits a 3D representation of the transformed GC  MS data by a plot of the signal intensity in the z direction. By that, a compound class grouping similar to that for comprehensive 2D GC (GC  GC) is recognizable without using a sophisticated modulation device. This leads to a better differentiation of the single compound classes contained in the diesel sample. Note that in Figure 4b the raw data of SPI are displayed, indicating discrimination of the aliphatic compounds 3870

dx.doi.org/10.1021/ac200356t |Anal. Chem. 2011, 83, 3865–3872

Analytical Chemistry

ARTICLE

Figure 6. (a) 3D volume plot generated by plotting of the first and second dimension retention times against the molecular mass (third dimension) displaying the most intense peaks of the GC  GC  MS data. (b) SPI-MS (blue) and EI-MS (red) spectra of overlapping peaks (marked by blue ovals in panel a).

because of their reduced PI cross sections compared to aromatic molecules. Application of Rapid Switching between EI and EBELBased SPI for Comprehensive 2D GC  GC  MS Analysis. Because an ultrafast oaTOF mass spectrometer was used in this switching approach, the hyphenation to comprehensive 2D GC (GC  GC) could also be realized. The switching frequency was set to 80 Hz, and a modulation period of 6 s was employed. However, for faster separation in the second dimension, the data acquisition rate can be increased to a maximum of 200 Hz. Again, the ionic liquid column (SLB-IL59) is used for separation in the first dimension and was coupled to a medium-polarity column [RTX-200, (trifluoropropylmethyl)polysiloxane] for separation in the second dimension. As was already described in the Experimental Section, the data acquisition was accomplished in such a way that each data dimension (MS spectrum, ionization mode, and first and second separation dimensions) is saved in a separate dimension in the hdf5 file, facilitating import, analysis, and visualization of the acquired data. Figure 5a depicts the TIC 2D contour plot obtained by summing over the EI and SPI segments. Therefore, each chromatographic peak is comprised of two complementary MS spectra, which is exemplarily shown for three compounds. Figure 5b displays the EI- and SPI-MS spectra of each compound. The EI-MS spectra provide important information for the structure and functional groups of the molecules due to fragmentation. Additionally, a MS library search suggested the following compounds: n-undecane, 2-methyldecalin, and methyltetralin. The soft SPI-MS spectra verified the molecular masses of each compound. A further confirmation of this result is given by considering the position of each peak marked in the 2D separation plane because all molecules are located within their intrinsic chemical class, indicated by the white ovals (Figure 5a). According to the extension of a 1D GC/MS measurement to a comprehensive 2D GC  MS analysis, the GC  GC separation can be expanded to a 3D separation approach (GC  GC  MS) by counting the mass axis of the soft ionization as the third separation dimension (Figure 6). Therefore, sample constituents

are spread in the resulting 3D separation space, leading to a considerably improved dispersion of peaks. The upper right part of Figure 1 displays the SPI-MS spectrum at a retention time of 13.21 min in the first dimension and 1.14 s in the second dimension (indicated by the green vertical line). Three compounds overlapping in both GC dimensions are separated on the mass axis according to their molecular-ion peaks at m/z 148 and 162 (two alkylated benzenes) and m/z 160 (alkylated indane/ tetralin). When a high-resolution mass spectrometer is utilized in conjunction with SPI, the selectivity of the GC  GC  MS approach would be enhanced still further because isobaric compounds can be separated according to their molecular-ion peaks on the mass axis.46 The lower right part of Figure 6 exhibits the EI-MS spectrum at the same retention times comprising several overlapping fragment peaks from the nonseparated compounds. The mineral-oil-type diesel sample analyzed in this work does not represent the most challenging matrix for this switching approach combined with GC  GC analysis. However, the analytical suitability could clearly be demonstrated. Biological samples for metabolomic analysis, environmental aerosol particulate matter, and cigarette smoke represent even more complex mixtures of different compound classes and will be investigated comprehensively by employing rapid switching between EI and SPI in the near future.

’ CONCLUSION AND OUTLOOK 1D and comprehensive 2D GC were hyphenated to an oaTOF mass spectrometer with the capability of rapid switching between hard EI and soft SPI for the first time. The innovative EBEL provided VUV photons of 9.8 eV (126 nm) for SPI. A maximum switching frequency of 80 Hz was employed for investigation of a mineral-oil-type diesel sample. This novel approach allows quasisimultaneous acquisition of complementary information about the fragmentation pattern, as well as the molecular mass of compounds within a single GC analysis. Furthermore, if a comprehensive investigation of a sample using soft and hard ionization is 3871

dx.doi.org/10.1021/ac200356t |Anal. Chem. 2011, 83, 3865–3872

Analytical Chemistry desired, the measurement time can be reduced by half by this switching approach. Additionally, when a polar ionic liquid GC column is utilized for separation, the SPI data can be displayed in a 2D contour plot, leading to a comprehensive 2D characterization (GC  MS) of the investigated sample. The switching approach clearly demonstrated its analytical suitability also for GC  GC analysis and allows the extension to a 3D separation (GC  GC  MS). Prospective switching between EBEL-based SPI and EI in combination with GC  GC will be applied for the investigation of even more complex samples than mineral oil fractions.

’ AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected] or ralf.zimmermann@ helmholtz-muenchen.de. Tel.: þ49 (0) 381 498 6460. Fax: þ49 (0) 381 498 6461.

’ REFERENCES (1) McMaster, M. GC/MS: A Practical User’s Guide; Wiley & Sons, Inc.: Hoboken, NJ, 2008. (2) H€ubschmann, H.-J. Handbook of GC/MS: Fundamentals and Applications; Wiley-VCH: Weinheim, Germany, 2008. (3) Gross, J. H. Mass Spectrometry; Springer-Verlag: Berlin, Germany, 2004. (4) Mondello, L.; Tranchida, P. Q.; Dugo, P.; Dugo, G. Mass Spectrom. Rev. 2008, 27, 101–124. (5) Adahchour, M.; Beens, J.; Brinkman, U. A. T. J. Chromatogr. A 2008, 1186, 67–108. (6) Guilhaus, M.; Selby, D.; Mlynski, V. Mass Spectrom. Rev. 2000, 19, 65–107. (7) Munson, B. Int. J. Mass Spectrom. 2000, 200, 243–251. (8) Munson, M. S. B.; Field, F. H. J. Am. Chem. Soc. 1966, 88, 2621–2630. (9) Beckey, H. D. Angew. Chem., Int. Ed. 1969, 8, 623–639. (10) Lattimer, R. P.; Schulten, H. R. Anal. Chem. 1989, 61, 1201A–1215A. (11) Boesl, U.; Neusser, H. J.; Schlag, E. W. Z. Z. Naturforsch. 1978, 33a, 1546–1548. (12) Kuribayashi, S.; Yamakoshi, H.; Danno, M.; Sakai, S.; Tsuruga, S.; Futami, H.; Morii, S. Anal. Chem. 2005, 77, 1007–1012. (13) Tonokura, K.; Kanno, N.; Yamamoto, Y.; Yamada, H. Int. J. Mass Spectrom. 2010, 290, 9–13. (14) Tonokura, K.; Nakamura, T.; Koshi, M. Anal. Sci. 2003, 19, 1109–1113. (15) Yamamoto, Y.; Kanno, N.; Tonokura, K.; Yabushita, A.; Kawasaki, M. Int. J. Mass Spectrom. 2010, 296, 25–29. (16) Butcher, D. J. Microchem. J. 1999, 62, 354–362. (17) Ng, C. Y. Int. J. Mass Spectrom. 2000, 200, 357–386. (18) Abate, S.; Ahn, Y. G.; Kind, T.; Cataldi, T. R. I.; Fiehn, O. Rapid Commun. Mass Spectrom. 2010, 24, 1172–1180. (19) Briker, Y.; Ring, Z.; Iacchelli, A.; McLean, N.; Rahimi, P. M.; Fairbridge, C.; Malhotra, R.; Coggiola, M. A.; Young, S. E. Energy Fuels 2001, 15, 23–37. (20) Hejazi, L.; Ebrahimi, D.; Hibbert, D. B.; Guilhaus, M. Rapid Commun. Mass Spectrom. 2009, 23, 2181–2189. (21) Qian, K.; Dechert, G. J. Anal. Chem. 2002, 74, 3977–3983. (22) Imasaka, T.; Sakai, K.; Imasaka, T. Anal. Chem. 2008, 76, 5534–5538. (23) Sakoda, Y.; Uchimura, T.; Imasaka, T. Anal. Chem. 2010, 82, 1283–1287. (24) Zimmermann, R.; Rohwer, E. R.; Heger, H. J. Anal. Chem. 1999, 71, 4148–4153. (25) Tsuruga, S.; Suzuki, T.; Takatsudo, Y.; Seki, K.; Yamauchi, S.; Kuribayashi, S.; Morii, S. Environ. Sci. Technol. 2007, 41, 3684–3688.

ARTICLE

(26) Zimmermann, R.; Welthagen, W.; Gr€oger, T. J. Chromatogr. A 2008, 1184, 296–308. (27) Wang, F. C. Y.; Qian, K.; Green, L. A. Anal. Chem. 2005, 77, 2777–2785. (28) Mitschke, S.; Welthagen, W.; Zimmermann, R. Anal. Chem. 2006, 78, 6364–6375. (29) Qian, K.; Dechert, G. J.; Edwards, K. E. Int. J. Mass Spectrom. 2007, 265, 230–236. (30) Hejazi, L.; Ebrahimi, D.; Guilhaus, M.; Hibbert, D. B. Anal. Chem. 2009, 81, 1450–1458. (31) Welthagen, W.; Mitschke, S.; M€uhlberger, F.; Zimmermann, R. J. Chromatogr. A 2007, 1150, 54–61. (32) Eschner, M. S.; Welthagen, W.; Gr€ oger, T. M.; Gonin, M.; Fuhrer, K.; Zimmermann, R. Anal. Bioanal. Chem. 2010, 398, 1435– 1445. (33) M€uhlberger, F.; Hafner, K.; K€asdorf, S.; Ferge, T.; Zimmermann, R. Anal. Chem. 2004, 76, 6753–6764. (34) M€uhlberger, F.; Saraji-Bozorgzad, M.; Gonin, M.; Fuhrer, K.; Zimmermann, R. Anal. Chem. 2007, 79, 8118–8124. (35) Wieser, J.; Murnick, D. E.; Ulrich, A.; Huggins, H. A.; Liddle, A.; Brown, W. L. Rev. Sci. Instrum. 1997, 68, 1360–1364. (36) HDF group, http://www.hdfgroup.org/HDF5/. (37) M€uhlberger, F.; Streibel, T.; Wieser, J.; Ulrich, A.; Zimmermann, R. Anal. Chem. 2005, 77, 7408–7414. (38) M€uhlberger, F.; Wieser, J.; Morozov, A.; Ulrich, A.; Zimmermann, R. Anal. Chem. 2005, 77, 2218–2226. (39) Geissler, R.; Saraji-Bozorgzad, M. R.; Gr€oger, T.; Fendt, A.; Streibel, T.; Sklorz, M.; Krooss, B. M.; Fuhrer, K.; Gonin, M.; Kaisersberger, E.; Denner, T.; Zimmermann, R. Anal. Chem. 2009, 81, 6038–6048. (40) Saraji-Bozorgzad, M.; Geissler, R.; Streibel, T.; M€uhlberger, F.; Sklorz, M.; Kaisersberger, E.; Denner, T.; Zimmermann, R. Anal. Chem. 2008, 80, 3393–3403. (41) Schramm, E.; H€olzer, J.; P€utz, M.; Schulte-Ladbeck, R.; Schultze, R.; Sklorz, M.; Ulrich, A.; Wieser, J.; Zimmermann, R. Anal. Bioanal. Chem. 2009, 395, 1795–1807. (42) Schramm, E.; K€urten, A.; H€olzer, J.; Mitschke, S.; M€uhlberger, F.; Sklorz, M.; Wieser, J.; Ulrich, A.; P€utz, M.; Schulte-Ladbeck, R.; Schultze, R.; Curtius, J.; Borrmann, S.; Zimmermann, R. Anal. Chem. 2009, 81, 4456–4467. (43) Vendeuvre, C.; Ruiz-Guerrero, R.; Bertoncini, F.; Duval, L.; Thiebaut, D.; Hennion, M.-C. J. Chromatogr. A 2005, 1086, 21–28. (44) Adahchour, M.; Beens, J.; Vreuls, R. J. J.; Brinkman, U. A. T. Trends Anal. Chem. 2006, 25, 726–741. (45) Eschner, M. S.; Zimmermann, R. Appl. Spectrosc. 2011submitted for publication. (46) Reichenbach, S. E.; Tian, X.; Tao, Q.; Ledford, E. B., Jr.; Wu, Z.; Fiehn, O. Talanta 2011, 83, 1279–1288.

3872

dx.doi.org/10.1021/ac200356t |Anal. Chem. 2011, 83, 3865–3872