Evaluation of a Rapid-Scanning Quadrupole Mass Spectrometer in an

Sep 27, 2010 - Find my institution. Blank image Log In ...... Peter Q. Tranchida, Luigi Mondello, Samuel D. H. Poynter, Robert A. Shellie. Comprehensi...
2 downloads 0 Views 2MB Size
Anal. Chem. 2010, 82, 8583–8590

Evaluation of a Rapid-Scanning Quadrupole Mass Spectrometer in an Apolar × Ionic-Liquid Comprehensive Two-Dimensional Gas Chromatography System Giorgia Purcaro,†,‡ Peter Quinto Tranchida,† Carla Ragonese,† Lanfranco Conte,‡ Paola Dugo,†,§ Giovanni Dugo,† and Luigi Mondello*,†,§ Dipartimento Farmaco-chimico, Facolta` di Farmacia, Universita` di Messina, viale Annunziata, 98168 Messina, Dipartimento di Scienze degli Alimenti, Universita` degli Studi di Udine, via Sondrio 2, 33100 Udine, and Universita` Campus-Bio-medico, via Alvaro del Portillo 21, 00128 Roma, Italy Comprehensive two-dimensional gas chromatography (GC×GC) is a powerful technique which can enable a great increase in GC peak capacities. However, since secondary-column separations are very rapid, detectors with a fast acquisition rate are mandatory. Such a requirement has certainly limited the use of the quadrupole mass spectrometer in the GC×GC field. The present research is focused on the evaluation of a novel rapid-scanning quadrupole mass spectrometry (qMS) detector, characterized by a 20 000 amu/s scan speed and a 50 Hz scan frequency, using a 290 amu mass range (40-330 m/z). The performance of the MS system was assessed by analyzing mixtures of 24 allergens, as well as a perfume sample, through GC×GC/qMS. The MS parameters evaluated at different acquisition rates (50, 33, and 25 Hz), as well as in the (simultaneous) scan/selected ion monitoring (SIM) mode, were the number of data points per peak, mass spectrum quality, peak skewing, and sensitivity. Two GC×GC/qMS methods, using the 50 Hz acquisition rate and the scan/SIM mode, were validated. Both methods provided similar results in terms of repeatability, accuracy, and linearity, while a great increase in sensitivity was observed (ca. a factor of 10) under scan/SIM conditions. The validated method proved to be suitable for the analysis of perfume allergens, according to the requirements of Directive 2003/15/EC. Comprehensive two-dimensional GC (GC×GC), a revolutionary GC invention,1 is becoming well-established and has been described in recent reviews.2,3 Skipping a superfluous description of the GC×GC process, attention can be devoted to seconddimension analyses, essentially very fast GC analyses (4-8 s). * To whom correspondence should be addressed. Phone: +39-090-6766536. Fax: +39-090-358220. E-mail: [email protected]. † Universita` di Messina. ‡ Universita` degli Studi di Udine. § Universita` Campus-Biomedico. (1) Liu, Z.; Phillips, J. B. J. Chromatogr. Sci. 1991, 29, 227. (2) Adahchour, M.; Beens, J.; Brinkman, U. A. Th. J. Chromatogr., A 2008, 1186, 67. (3) Cortes, H. J.; Winniford, B.; Luong, J.; Pursch, M. J. Sep. Sci. 2009, 32, 883. 10.1021/ac101678r  2010 American Chemical Society Published on Web 09/27/2010

Fast detectors are mandatory to accurately reconstruct modulated peaks (typically 100-600 ms at the baseline), separated under very fast GC conditions. In terms of accurate peak reconstruction, a variety of opinions exist.4-7 For example, Hinshaw4 stated that 10 data points across the width at half-height are required, while Dyson5 concluded that at least 15-20 data points/peak are needed for reliable quantification. Adahchour et al.,6 evaluating the performance of a rapidscanning quadrupole mass spectrometer in GC×GC analysis, affirmed that at least 7 data points/peak (above the baseline) were necessary for accurate peak reconstruction. However, it is generally agreed within the chromatography community that 10 data points/peak are sufficient for reliable peak reconstruction.7 The “fast” flame ionization detector (FID) dominated the GC×GC scene in the 1990s. However, as the bidimensional technique became more popular, the use of the mass spectrometer became a necessity. Throughout the 12-year GC×GC/MS history, quadrupole and low-resolution time-of-flight (TOF) systems have dominated the scene, almost entirely. The first GC×GC/MS combination was reported by Frysinger and Gaines in 1999.8 A quadrupole mass spectrometry (qMS) system was employed which generated 2.43 scans/s, far too slow for GC×GC requirements. In that first investigation the authors stated, with hindsight, that the best solution would be the use of a TOF instrument. A year later, van Deursen et al.9 were the first to employ GC×GC/ TOF MS (kerosene analysis): a 50 Hz acquisition rate was used, a compromise to maintain good sensitivity, to generate a sufficient number of data points per peak, and to limit the data file size. During 2001 and 2002, further GC×GC/TOF MS works appeared:10-14 in particular, Dalluge et al.10 described a cigarette (4) Hinshaw, J. V. LCGC North Am. 2003, 21, 268. (5) Dyson, N. J. Chromatogr., A 1999, 842, 321. (6) Adahchour, M.; Brandt, M.; Baier, H.-U.; Vreuls, R. J. J.; Batenburg, A. M.; Brinkman, U. A. Th. J. Chromatogr., A 2005, 1067, 245. (7) Poole, C. F. The Essence of Chromatography; Elsevier: Amsterdam, 2003; pp 66-67. (8) Frysinger, G. S.; Gaines, R. B. J. High Resolut. Chromatogr. 1999, 22, 251. (9) van Deursen, M.; Beens, J.; Reijenga, J.; Lipman, P.; Cramers, C.; Blomberg, J. J. High Resolut. Chromatogr. 2000, 23, 507. (10) Dallu ¨ ge, J.; van Stee, L. L. P.; Xu, X.; Williams, J.; Beens, J.; Vreuls, R. J. J.; Brinkman, U. A. Th. J. Chromatogr., A 2002, 974, 169. (11) Dallu ¨ ge, J.; van Rijn, M.; Beens, J.; Vreuls, R. J. J.; Brinkman, U. A. Th. J. Chromatogr., A 2002, 965, 207.

Analytical Chemistry, Vol. 82, No. 20, October 15, 2010

8583

smoke experiment, emphasizing the enormous amount of data generated (100 spectra/s), the problems encountered during partially automated data processing, and the necessity of dedicated software. Further GC×GC/qMS experiments were described three years after the first paper:15,16 Frysinger et al.15 used the same qMS instrument described previously,8 encountering similar problems. Shellie and Marriott16 focused their attention on the use of a 0.25 mm i.d. secondary chiral column: enantiomer separations were achieved by exploiting the vacuum outlet conditions. Rather wide peaks were generated, suiting the slow qMS acquisition rate (8.33 Hz). The year 2003 was marked by the introduction of a “complete” comprehensive GC/TOF MS instrument, with a quad-jet modulator and fully integrated software for instrument control and entirely automated data processing.17 The first appearances of the use of fast-scanning qMS systems in the GC×GC field occurred also in 2003.18-20 In particular, Shellie et al.18 applied GC×GC/qMS to the analysis of ginseng volatiles. Mass spectra were acquired at 20 Hz, in a reduced range of 41-228.5 m/z, to enable the detection of molecular ions up into the oxygenated sesquiterpene region. The authors reported that the 3-4 data points/peak attained were sufficient only for identification purposes. The first example of quantification in a comprehensive GC application, by using a qMS system, was reported by Debonneville and Chaintreau in 2004:21 fragrance allergens were analyzed by selected ion monitoring (SIM); a detection frequency of 30.7 Hz was reported and affirmed to be sufficient for peak quantitation. It is obvious that the use of the SIM mode is only applicable to known target analytes. In 2005, Koryta´r et al.22 were the first to use a rapid-scanning qMS instrument, with chemical ionization [electron-capture negative ion (ECNI) mode], in a GC×GC experiment. The performance of the MS instrument was studied, finding it capable of producing 23 spectra/s, using a 300 amu mass range. Shortly after, the same research group used a newly introduced TOF MS system, also operated in the ECNI mode.23 Methane was used as the reagent gas, the mass range was 50-700 m/z, and the acquisition rate was 40 Hz. Also in 2005, Adahchour et al. reported a study on the principles, practicability, and potential of rapid-scanning qMS instrumentation.6 The detector was characterized by a maximum scan speed of 10 000 amu/s and could reach a frequency of 50 spectra/s at an excessively restricted mass range (95 amu). The ultimate goal of 7 data points/peak, affirmed by the authors to be sufficient for quantification, was achieved through the application (12) Dallu ¨ ge, J.; Vreuls, R. J. J.; Beens, J.; Brinkman, U. A. Th. J. Sep. Sci. 2002, 25, 201. (13) Adahchour, M.; Beens, J.; Vreuls, R. J. J.; Batenburg, A. M.; Rosing, E. A. E.; Brinkman, U. A. Th. Chromatographia 2002, 55, 361. (14) Shellie, R.; Marriott, P.; Morrison, P. Anal. Chem. 2001, 73, 1336. (15) Frysinger, G. S.; Gaines, R. B.; Reddy, C. M. Environ. Forensics 2002, 3, 27. (16) Shellie, R.; Marriott, P. J. Anal. Chem. 2002, 74, 5426. (17) Dimandja, J-M. D. Am. Lab. 2003, 35, 42. (18) Shellie, R. A.; Marriott, P. J.; Huie, C. W. J. Sep. Sci. 2003, 26, 1185. (19) Shellie, R. A.; Marriott, P. J. Analyst 2003, 128, 879. (20) Kallio, M.; Hyo ¨tyla¨inen, T.; Lehtonen, M.; Jussila, M.; Hartonen, K.; Shimmo, M.; Riekkola, M.-L. J. Chromatogr., A 2003, 1019, 251. (21) Debonneville, C.; Chaintreau, A. J. Chromatogr., A 2004, 1027, 109. (22) Koryta´r, P.; Parera, J.; Leonards, P. E. G.; de Boer, J.; Brinkman, U. A. Th. J. Chromatogr., A 2005, 1067, 255. (23) Koryta´r, P.; Parera, J.; Leonards, P. E. G.; Santos, F. J.; de Boer, J.; Brinkman, U. A. Th. J. Chromatogr., A 2005, 1086, 71.

8584

Analytical Chemistry, Vol. 82, No. 20, October 15, 2010

of a restricted mass range (m/z 50-245), which enabled the generation of 33 spectra/s. Under such MS operational conditions, the smallest peak width allowed for correct quantification was about 200 ms. During the 2006-2009 period, although the appearance of rapid-scanning qMS instrumentation still continued (ca. 13% of the published work), GC×GC/TOF MS became increasingly established (ca. 83% of the published work). TOF MS approaches can be divided essentially into two groups: one is related to instruments characterized by a high data acquisition frequency (i.e., 100-500 Hz) and low resolution, generally in the 1000-1500 range. The other class is characterized by systems with a high mass accuracy (i.e., 5-10 ppm) and a much lower data acquisition frequency.24 Worthy of note is that a few high-resolution (HR) TOF applications have also been described in the GC×GC literature.25,26 Ochiai et al. (2007) employed an HR TOF mass spectrometer to identify atmospheric nanoparticle constituents. The HR TOF MS instrument was operated using electron-based ionization and provided full-scan spectra in the 45-500 m/z mass range at a 25 Hz data acquisition frequency.25 A GC×GC/HR TOF MS experiment, focused on the analysis of polychlorinated dibenzo-p-dioxins and dibenzofurans, was reported by Shunji et al. (2008). A total of 25 full-scan spectra/s were generated using a 35-550 m/z mass range; the mass resolution was 5000 (fwhm).26 It must be noted that GC×GC/MS analysts normally look for unitmass resolution to generate spectra suitable for MS library matching. Quadrupole MS and low-resolution TOF analyzers are fine for such a scope. A series of “one-time” GC×GC experiments, using specific MS systems, have also been described:27,28 in particular, an ion-trap MS instrument has been coupled to a portable 2D GC system, capable of either heart-cutting or GC×GC analyses (2003). The MS acquisition speed was very low, namely, 1.5 s per scan;27 a single GC×GC/triple-quadrupole MS experiment using supersonic molecular beam electron ionization has also been reported (2008). Full-scan experiments were carried out using a 50-400 amu mass range and a 6.25 Hz data acquisition frequency (scan speed 2100 amu/s).28 Summarizing the GC×GC/MS history, it can be affirmed that (a) low-resolution TOF systems have gained a firm position, (b) qMS instruments, although characterized by a series of limitations, have been used in several investigations, and (c) other MS detectors have had a much minor impact. Considering point b, the data acquisition rate of qMS instruments is limited by two parameters, namely, the scan speed and the interscan delay. The scan process is composed of the scan time plus the time delay between consecutive scans (Figure 1). The scan frequency is easily attained by dividing the scan speed by the duration of a single scan process. Current-day rapid-scanning qMS systems are characterized by scan speeds in the 6000-12500 range; obviously, (24) Dallu ¨ ge, J.; Roose, P.; Brinkman, U. A. Th. J. Chromatogr., A 2002, 970, 213. (25) Ochiai, N.; Ieda, T.; Sasamoto, K.; Fushimi, A.; Hasegawa, S.; Tanabe, K.; Kobayashi, S. J. Chromatogr., A 2007, 1150, 13. (26) Shunji, H.; Yoshikatsu, T.; Akihiro, F.; Hiroyasu, I.; Kiyoshi, T.; Yasuyuki, S.; Masa-aki, U.; Akihiko, K.; Kazuo, T.; Hideyuki, O.; Katsunori, A. J. Chromatogr., A 2008, 1178, 187. (27) Wahl, J. H.; Riechers, D. M.; Vucelick, M. E.; Wright, B. W. J. Sep. Sci. 2003, 26, 1083. (28) Poliak, M.; Fialkov, A. B.; Amirav, A. J. Chromatogr., A 2008, 1210, 108.

Figure 1. Scheme of the qMS scan process, composed of the scan time plus the time delay between consecutive scans (volt. ) voltage).

the duration of the interscan delay (a fixed parameter, measured in milliseconds) and the extension of the mass range both have a great influence on the acquisition frequency. For example, in GC×GC/qMS experiments Cordero et al. reported an interscan delay of 30 ms and a “restricted” 40-240 m/z mass range (11 111 amu/s), to attain 18.52 spectra/s,29 while Mondello et al. applied a 40-400 m/z mass range (10 000 amu/s), reporting the generation of 20 spectra/s (the interscan delay was not reported).30 The present investigation is focused on the evaluation of a very fast scanning qMS instrument, herein used as a GC×GC detector. The novel mass spectrometer can operate at a 20 000 amu/s scan speed and generate 50 spectra/s using a 290 amu mass range (40-330 m/z). GC×GC/qMS applications were directed to the analysis of perfume allergens. The latter substances are regulated by European Directive 2003/15/EC,31 with maximum residue limits for “leave-on” and “rinse-off” products fixed at 0.001% and 0.01%, respectively. Several GC×GC papers21,29,30,32 have dealt with perfume allergens; however, apart from the FID, which gave reliable quantitative results,32 and the aforementioned SIM application,21 all qMS detectors suffered the same limitation, namely, a low spectral acquisition frequency. In this study, two GC×GC/qMS methods have been developed and validated using a 50 Hz acquisition frequency and the simultaneous scan/SIM mode. A nonpolar column was employed in the first dimension, while an ionic-liquid capillary was installed in the second. Ionic liquids (ILs) represent a class of organic nonmolecular solvents, usually consisting of an organic cation containing N or P (i.e., alkylimidazolium, phosphonium) counterbalanced by an anion of organic or inorganic nature.33 ILs have been widely employed in several chemistry fields and have received increasing interest as GC stationary phases.34,35 Indeed, IL columns have already been applied in a number of applications, such as polyaromatic hydrocarbons,36,37 chlorinated pesticides,36 essential oils,38 fatty acid methyl esters,39-41 and flavors and fragrances,39,42 as well as in classical multidimensional GC,43 GC×GC,44 and GC×GC×GC systems.45 (29) Cordero, C.; Bicchi, C.; Joulain, D.; Rubiolo, P. J. Chromatogr., A 2007, 1150, 37. (30) Mondello, L.; Casilli, A.; Tranchida, P. Q.; Dugo, G.; Dugo, P. J. Chromatogr., A 2005, 1067, 235. (31) Directive 2003/15/EC of the European Parliament and of the Council of 27 February 2003. (32) Shellie, R.; Marriott, P.; Chaintreau, A. Flavour Fragrance J. 2004, 19, 91. (33) Welton, T. Chem. Rev. 1999, 99, 2071. (34) Armstrong, D. W.; He, L.; Liu, Y.-S. Anal. Chem. 1999, 71, 3873. (35) Dhanesar, S. C.; Coddens, M. E.; Poole, C. F. J. Chromatogr. 1985, 349, 249. (36) Anderson, J. L.; Armstrong, D. W. Anal. Chem. 2005, 77, 6453. (37) Anderson, J. L.; Armstrong, D. W. Anal. Chem. 2003, 75, 4851. (38) Qi, M.; Armstrong, D. W. Anal. Bioanal. Chem. 2007, 388, 889.

EXPERIMENTAL SECTION Sample and Standard Solution Preparation. A commercial perfume was purchased in Messina (Italy). A stock solution of 24 allergens, plus 2 internal standards (1,4-dibromobenzene and 4,4′dibromobiphenyl), all supplied by Sigma-Aldrich (Milan, Italy), was prepared in methanol. The stock solution was then diluted to six working solutions for the construction of the calibration curves. GC×GC/qMS Analyses. All applications were carried out on a GC×GC/QP2010-Ultra qMS instrument (Shimadzu, Kyoto, Japan). The GC instrument was equipped with an AOC-20i autoinjector and a split-splitless injector (250 °C). (GC temperature program, 50-260 °C at 5 °C/min; initial He pressure (constant linear velocity), 140 kPa; injection volume, 1 µL; split ratio, 10:1. The primary column, an SLB-5 ms, 30 m × 0.25 mm i.d. × 0.25 µm df [silphenylene polymer, virtually equivalent in polarity to poly(5% biphenyl/95% methylsiloxane)], was connected to a 1.4 m × 0.25 mm i.d. uncoated capillary segment (used to create a double loop) by using an SGE SilTite miniunion (SGE, Ringwood, Victoria, Australia). The uncoated capillary was then connected to a segment of a custom-made ionic liquid SLB-IL59 (1.0 m × 0.1 mm i.d. × 0.08 µm df) column by using another mini-union. All the columns were provided by Supelco (Milan, Italy). Modulation was carried out every 4.5 s by using a loop-type modulator (under license from Zoex Corp., Houston, TX). The duration of the hot pulse (325 °C) was 375 ms. MS Parameters. The sample was analyzed in the full scan mode with a scan speed of 20 000 amu/s and a mass range of 40-330 m/z and in the contemporaneous scan/SIM mode. The interface and ion source temperatures were 250 and 200 °C, respectively. The MS ionization mode was electron ionization. qMS Performance Evaluation and Method Validation. The standard mixture, at the concentration level of about 10 mg/L for each compound, was tested twice at 33 and 25 Hz acquisition rates and three times at 50 Hz and in the contemporaneous scan/ SIM mode to evaluate the peak reconstruction, mass spectral quality, peak skewing, signal-to-noise ratio (S/N), and analytical repeatability. The scan/SIM mode is achieved by acquisition in the scan mode for 20 ms using the same 290 m/z mass range and by acquiring three specific ions for the following 20 ms; obviously, the ions varied according to the nature of the specific analyte. Method validation was carried out both at a 50 Hz acquisition rate and in the scan/SIM mode. Calibration curves were built at six concentration levels, performing two replicates at each level. The repeatability and accuracy were evaluated by performing the analysis of a mixture of known concentration three (39) Payagala, T.; Zhang, Y.; Wanigasekara, E.; Huang, K.; Breitbach, Z. S.; Sharma, P. S.; Sidisky, L. M.; Armstrong, D. W. Anal. Chem. 2009, 81, 160. (40) Ragonese, C.; Tranchida, P. Q.; Dugo, P.; Dugo, G.; Sidisky, L. M.; Robillard, M. V.; Mondello, L. Anal. Chem. 2009, 81, 5561. (41) Ragonese, C.; Tranchida, P. Q.; Sciarrone, D.; Mondello, L. J. Chromatogr., A 2009, 1216, 8992. (42) Huang, K.; Han, X.; Zhang, X.; Armstrong, D. W. Anal. Bioanal. Chem. 2007, 389, 2265. (43) Sciarrone, D.; Tranchida, P. Q.; Ragonese, C.; Schipilliti, L.; Mondello, L. J. Sep. Sci. 2010, 33, 594. (44) Seeley, J. V.; Seeley, S. K.; Libby, E. K.; Breitbach, Z. S.; Armstrong, D. W. Anal. Bioanal. Chem. 2008, 390, 323. (45) Siegler, W. C.; Crank, J. A.; Armstrong, D. W.; Synovec, R. E. J. Chromatogr., A 2010, 1217, 3144.

Analytical Chemistry, Vol. 82, No. 20, October 15, 2010

8585

consecutive times. Data were collected by the GCMS Solution software (Shimadzu); bidimensional visualization was carried out by using the ChromSquare v.1.2 software (Chromaleont, Messina, Italy). The MS library used for spectral matching was FFNSC 1.4 (Shimadzu). RESULTS AND DISCUSSION GC×GC Separation. The most popular set of columns used for fragrance analyses is a conventional nonpolar column in the first dimension and a fast polar column in the second (commonly, polyethylene glycol). In the present research, a conventional 30 m × 0.25 mm i.d. nonpolar first column was combined with an ILcoated second-dimension column. The IL phase employed is characterized by a polarity comparable to that of polyethylene glycol and a higher thermal stability. Evaluation of the polarity of the SLB-IL59 column has been achieved by calculating the McReynolds constants. As well-known, McReynolds constants (∆I) represent the difference between the Kovats indices of chosen probes, obtained on a given stationary phase, and those obtained on a nonpolar reference column, the squalane stationary phase; the sum of the first five probe ∆I values measures the overall polarity (P) of the stationary phase. The IL column showed a P value of 2624, still in the same range of polarity though a little higher than that obtained for polyethylene glycol (2324). The main advantage of this ionic liquid stationary phase is the higher thermal stability (300 °C vs 280 °C) and the negligible bleeding if compared to polyethylene glycol, the most widely used polar phase at present. Bleed evaluation has been achieved by measuring noise intensity at the highest operating temperature on both stationary phases: a value of 3 × 105 pA was obtained for the IL column at 300 °C, while the noise intensity at 280 °C was 1.85 × 107 using the polyethylene glycol capillary. The peak capacity was also evaluated by calculating the separation number (SN) for a series of homologous hydrocarbons (from C7 to C30) and saturated fatty acid methyl esters (FAMEs; from C4 to C24) on both columns; the SN values were slightly lower for the IL column (443 vs 514 for alkanes and 461 vs 553 for FAMEs), mainly due to the fact that elution is faster on the IL column. In fact, triacontane elutes on the IL column in about 52 min (206 °C), while it elutes in about 61 min (233 °C) on polyethylene glycol. With regard to FAMEs, C24:0 elutes in about 58 min (223 °C) on the IL phase and in about 60 min (229 °C) on polyethylene glycol. Such results demonstrate the possibility to analyze longer chain hydrocarbons and FAMEs (C43 and C32:0, respectively) on the IL phase (also accounting for its higher operating temperature) compared to polyethylene glycol, where the C40 alkane and C27:0 FAME elute potentially at 280 °C. The aforedescribed results, characterizing the IL and polyethylene glycol phases, derive from previous (unpublished) research. Although the benefits of flow-splitting between the two GC×GC dimensions have been demonstrated in recent research,46-49 a conventional GC×GC approach was used here; this is simply because such GC×GC methods are by far the most common and, hence, appear the most suitable to assess MS performance. Chromatographic conditions were optimized by tuning the operational parameters in the analysis of a perfume sample (Figure 2). The head pressure generated gas linear velocities of ca. 24 and 200 cm/s in the first and second dimensions, respectively. 8586

Analytical Chemistry, Vol. 82, No. 20, October 15, 2010

The slight wrap-around observed under such conditions was corrected using the GC×GC software “wrap-around correction” option, which is simply a visual function. The optimized chromatographic conditions were used to evaluate the qMS performance and to construct calibration curves for 24 perfume allergens. qMS Performance. As reported in the introduction, the performance of the qMS instrument as a detector in the GC×GC field has been widely studied. The investigations carried out demonstrated that the qMS systems available could achieve, or were close to reaching, the requirements for analyte quantification only by using some obliged limitations (in particular, SIM acquisition or the application of a reduced mass range). In the present investigation, no restrictions were applied to the MS parameters: at a scan speed of 20 000 amu/s and an interscan delay of maximum 5 ms and using a 40-330 m/z range, a spectral production frequency of 50 Hz was attained. The extent of peak reconstruction was assessed by counting the number of data points per peak above the baseline and over the half-height of the peak, comparing three different acquisition frequencies (50, 33, and 25 Hz), and maintaining the same mass range. For such a purpose, four compounds were chosen, namely, linalool, eugenol, lilial, and benzyl salicylate; the results are shown in Table 1 and are related to the main modulated peak of each compound. The 10 data point/peak requirement can be satisfied by using an acquisition frequency of 25 Hz, only if the peak width is above 400 ms, while in such a case more than 20 data points/ peak were obtained through the application of a 50 Hz frequency. Such a high acquisition frequency also enables higher analytical flexibility: if high MW compounds are analyzed, then a wider mass range can be applied while a satisfactory acquisition frequency is maintained. For example, if a 40-440 m/z range is used, then up to 33 spectra/s can be generated, which in all cases in the present study would enable reliable peak reconstruction. The quality of the mass spectra, relative to the four selected compounds, was assessed at the three acquisition frequencies tested: the similarity match (MS %) with MS library-contained compounds was evaluated at each data point. The intensity ratio between the target ion (T) and the qualifier ions (Q1 and Q2) was also calculated to evaluate the consistency of the mass spectral profiles. Peak skewing, in particular, was evaluated by considering the ion intensity ratio between the highest (HM) and the lowest (LM) mass ions present, with a significant abundance in the mass spectrum (an ion above 50 m/z was chosen to reduce the noise contribution to the relative abundance). As an example, three graphs (one for each acquisition frequency) containing the four aforementioned spectral parameters, relative to those of linalool (MS %, T/Q1, T/Q2, Q2/LM; LM ) 55 m/z) and derived from sequential data points across the upper peak half, are reported in Figure 3. To show the consistency of the spectral parameters, MS (%) and T/Q2 values at each data point acquired are illustrated in the 50 Hz graph. Among the four compounds, linalool is shown (46) Tranchida, P. Q.; Casilli, A.; Dugo, P.; Dugo, G.; Mondello, L. Anal. Chem. 2007, 79, 2266. (47) Tranchida, P. Q.; Purcaro, G.; Conte, L.; Dugo, P.; Dugo, G.; Mondello, L. J. Chromatogr., A 2009, 1216, 7301. (48) Tranchida, P. Q.; Purcaro, G.; Conte, L.; Dugo, P.; Dugo, G.; Mondello, L. Anal. Chem. 2009, 81, 8529. (49) Tranchida, P. Q.; Purcaro, G.; Fanali, C.; Dugo, P.; Dugo, G.; Mondello, L. J. Chromatogr., A 2010, 1217, 4160.

Figure 2. GC×GC/qMS chromatogram of a real sample of perfume. Twelve allergens are identified, along with the two internal standards (IS-1 and IS-2). Table 1. Target Ion (T) and Qualifier Ions (Q1 and Q2) Evaluated for Each Compound and Peak Width at the Baseline and at Half-Height and Number of Data Points Acquired for the Main Modulated Peak of Linalool, Eugenol, Lilial, and Benzyl Salicylate 50 Hz

compd

T

Q1

Q2

linalool 93 71 121 eugenol 164 103 149 lilial 189 147 204 benzyl salicylate 91 228 65

33 Hz

25 Hz

peak peak peak width no. of half-peak no. of width no. of half-peak no. of width no. of half-peak no. of LM (ms) points (ms) points (ms) points (ms) points (ms) points (ms) points 55 55 57

360 480 420 480

18 24 21 24

180 240 190 240

because it is characterized by a poor peak shape at the lower acquisition frequencies. Indeed, at 25 and 33 Hz the linalool peaks appear to be “beheaded”, a random phenomenon which could reduce the S/N and the retention time repeatability. With regard to the spectral quality observed for the four compounds considered, MS % values were nearly always constant at each data point and T/Q ratios were comparable under all the tested conditions while the coefficients of variation relative to HM/LM ion ratios were between 5% and 10% when using 33 and 50 Hz and in the 10-15% range when using 25 Hz. From a practical viewpoint, it is well known that a decrease of sensitivity occurs when the acquisition rate is increased.6 Therefore, sensitivity was evaluated, in terms of S/N values, by injecting the allergen standard mixture at the concentration level corresponding to the legal limit for leave-on products (0.001%, corre-

9 12 10 12

360 480 420 480

12 16 14 16

180 240 180 240

6 8 6 8

360 480 360 480

9 12 9 12

180 240 180 240

4 6 5 6

sponding to 10 mg/L). The analysis was carried out at different acquisition rates, 50, 33, and 25 Hz, as well as using the contemporaneous scan/SIM mode. In the latter condition, the detector performs a scan acquisition for 20 ms and an SIM acquisition in the following 20 ms if the number of selected ions is limited to 3-5 masses (ion dwell time and interchannel delay range both between 1 and 5 ms in relation to the ion masses selected). Table 2 reports the S/N values calculated automatically by the instrument software, considering the extracted target ion of each compound. The calculation of the signal-to-noise ratios in the scan/SIM mode was obtained in two ways: the first by calculating the S/N value of the target ion, extracted from the scan data, and the second by directly using the target ion acquired in the SIM mode. In general, there was no significant sensitivity variability at the different acquisition rates, while a significant Analytical Chemistry, Vol. 82, No. 20, October 15, 2010

8587

Table 2. Comparison of the Signal-to-Noise Ratios (S/N) at Different Scan Acquisition Rates and in the Scan/SIM Mode and the Target Ion (T) and Qualifiers (Q1 and Q2) Used for Each Compound S/N compd limonene benzyl alcohol linalool methyl 2-octynoate citronellol citral (neral) geraniol citral (geranial) cinnamaldehyde anisyl alcohol hydroxycitronellal cinnamyl alcohol eugenol coumarin isoeugenol isomethylionone lilial amylcinnamaldehyde lyral amylcinnamyl alcohol farnesol hexylcinnamaldehyde benzyl benzoate benzyl salicylate benzyl cinnamate

Figure 3. Spectral quality evaluation of linalool at 25, 33, and 50 Hz in terms of mass spectrum similarity (MS %) at each data point acquired and ion ratios across the width at half-height: T, 93 m/z; Q1, 71 m/z; Q2, 121 m/z; LM, 55 m/z. In the 50 Hz graph, MS and T/Q2 values at each data point acquired are shown.

increase of sensitivity was observed using the scan/SIM mode, in some cases about 50-fold, with an average increase of a factor of 10. This last result is very interesting since a great increase in sensitivity is obtained, without loss of full-scan information. Method Validation. The two optimized GC×GC/qMS methods, namely, the 50 Hz and the scan/SIM methods, were validated. For such a purpose, two internal standards were used: 1,4dibromobenzene (IS-1) and 4,4′-dibromobiphenyl (IS-2), as requested by the International Fragrance Association (IFRA).50 Regression curves for the 24 allergens were built using the leastsquares method. Tables 3 (scan data) and 4 (SIM data) report information related to the calibration range, regression curve parameters (slope, intercept, and regression coefficient), repeatability [coefficient of variation (10 mg/L level), CV (%)], accuracy [A (%)], and limits of detection and quantification (LOD and LOQ, estimated as 3 and 10 times the S/N value, respectively) under both MS conditions. The significance of the intercept was established by running a t test (significance level 5%, highlighted (50) Analytical procedure for the GC/MS quantitation of potential fragrance allergens in fragrance compounds; version 1. Analytical Procedures; International Fragrance Association: Brussels, Belgium, 2003; pp 1-10.

8588

Analytical Chemistry, Vol. 82, No. 20, October 15, 2010

target ion Q1 Q2 50 Hz 33 Hz 25 Hz scan 68 108 93 95 69 69 69 69 131 138 59 92 164 146 164 135 189 202 192 133 69 216 105 91 131

93 79 71 123 95 84 123 94 132 137 71 134 103 118 149 206 147 201 136 115 93 215 212 228 192

67 107 121 79 81 94 93 109 103 109 43 115 149 89 131 150 204 129 149 204 81 129 194 65 193

25084 4362 4582 9899 1151 1276 1230 1470 1293 1025 2800 875 3164 767 935 2690 4482 2312 389 520 354 2590 14906 8204 1090

17863 5346 3847 8038 1068 1032 1666 1693 1508 1273 2906 768 4342 829 860 10657 4364 2799 372 368 414 2347 11290 10011 1016

22621 3154 4737 8644 1607 1228 2763 1496 1992 1156 3280 1098 7657 931 877 13384 4302 2679 395 413 780 3356 15818 11221 1497

17115 3014 3747 1126 1649 925 1874 1667 1338 966 1602 607 6973 708 2136 6235 4075 1040 162 234 560 3395 6429 11727 1964

SIM 200427 18866 51419 53717 7292 4929 6040 10997 15379 6801 52252 6834 33174 5800 7939 57061 47058 32191 5138 4007 4802 10532 23383 43044 3898

in the tables with footnote a): only three compounds were characterized by significant values, namely, limonene, linalool, and citronellol, when using the 50 Hz acquisition rate and only two under scan/SIM conditions (limonene and linalool). Regression coefficients were all above 0.982 for both GC×GC/qMS methods. The repeatability of the area values (expressed as the CV (%) of three replicates) was similar for both methods, though slightly better when the scan/SIM mode was employed; however, CV values always lower than 12% were obtained for the 50 Hz method. Accuracy was better than ±10% for all the allergens tested. As expected, the scan/SIM mode was characterized by a significant sensitivity increase: the S/N values increased by about 10-fold, with respect to the scan mode, where quantification was achieved through an extracted ion (e.g., the LOD and LOQ of eugenol decreased from 19.3 to 1.8 µg/L and from 64.5 to 6.1 µg/L, respectively). However, even in the scan mode, LODs and LOQs were far below the lower legal limit: values always lower than 80 and 265 µg/L were obtained, respectively. Perfume Analysis. A commercial perfume sample was analyzed by employing the method validated using the 50 Hz acquisition rate. In this specific case the increase in sensitivity obtained by employing the scan/SIM mode was not required, since the allergen legal limits are far higher than the measured LOQs. The GC×GC/qMS profile obtained, shown in Figure 2, and the allergens identified and quantified are reported in Table 5. Twelve allergens, all declared on the label, were identified at concentrations higher than the legal limit for leave-on products. The retention time repeatability was evaluated to consider possible matrix effects; differences lower than 0.05 min and than 0.09 s were observed in the first and second dimensions, respectively.

Table 3. Calibration Curves, Calculated at Six Concentration Levels, Regression Coefficients (R2), Repeatability (Expressed as the Coefficient of Variation, CV), Accuracy (A), and Limits of Detection (LOD) and Quantification (LOQ), Calculated by Using an Acquisition Frequency of 50 Hz compd

concn range (µg/L)

slope

intercept

R2

CV (%) (n ) 3)

A (%) (n ) 3)

LOD (µg/L)

LOQ (µg/L)

limonene benzyl alcohol linalool methyl 2-octynoate citronellol citral (neral) geraniol citral (geranial) cinnamaldehyde anisyl alcohol hydroxycitronellal cinnamyl alcohol eugenol coumarin isoeugenol isomethylionone lilial amylcinnamaldehyde lyral amylcinnamyl alcohol farnesol hexylcinnamaldehyde benzyl benzoate benzyl salicylate benzyl cinnamate

54 × 103 to 4.3 224 × 103 to 25.7 204 × 103 to 22.3 244 × 103 to 12.3 220 × 103 to 95.6 154 × 103 to 60.7 188 × 103 to 76.4 154 × 103 to 52.7 204 × 103 to 78.9 228 × 103 to 111.2 276 × 103 to 49.3 200 × 103 to 114.3 408 × 103 to 64.5 236 × 103 to 153.8 352 × 103 to 188.2 236 × 103 to 43.9 224 × 103 to 25.0 236 × 103 to 51.0 204 × 103 to 262.4 236 × 103 to 226.9 204 × 103 to 258.3 248 × 103 to 42.5 360 × 103 to 12.1 484 × 103 to 29.5 220 × 103 to 101.0

0.0189 ± 0.0003 0.0075 ± 0.0002 0.0056 ± 0.0001 0.0040 ± 0.0001 0.0056 ± 0.0001 0.0054 ± 0.0001 0.0113 ± 0.0003 0.0123 ± 0.0004 0.0083 ± 0.0002 0.0039 ± 0.0001 0.0092 ± 0.0001 0.0028 ± 0.0001 0.0070 ± 0.0001 0.0061 ± 0.0002 0.0064 ± 0.0002 0.0077 ± 0.0002 0.0093 ± 0.0001 0.0030 ± 0.0001 0.0009 ± 0.0001 0.0018 ± 0.0001 0.0054 ± 0.0002 0.0026 ± 0.0001 0.0189 ± 0.0002 0.0256 ± 0.0007 0.0066 ± 0.0002

0.1609 ± 0.0103a -0.0151 ± 0.0153 0.0134 ± 0.0058a -0.0140 ± 0.0089 -0.0198 ± 0.0072a -0.0117 ± 0.0099 -0.0523 ± 0.0250 -0.0290 ± 0.0297 -0.0166 ± 0.0244 -0.0220 ± 0.0160 -0.0120 ± 0.0160 -0.0208 ± 0.0108 -0.0003 ± 0.0169 -0.0130 ± 0.0255 -0.0331 ± 0.0382 -0.0029 ± 0.0185 -0.0037 ± 0.0156 -0.0031 ± 0.0120 -0.0041 ± 0.0029 -0.0183 ± 0.0081 -0.0683 ± 0.0299 -0.0041 ± 0.0071 -0.0219 ± 0.0233 -0.0826 ± 0.1235 -0.0386 ± 0.0219

0.998 0.996 0.999 0.997 0.999 0.995 0.995 0.991 0.992 0.988 0.999 0.986 0.999 0.988 0.989 0.996 0.998 0.989 0.991 0.986 0.985 0.995 0.999 0.995 0.992

1.7 4.1 2.3 0.9 6.9 8.8 5.3 10.5 8.9 2.5 10.8 9.7 2.2 5.1 4.4 9.6 5.5 8.4 6.0 7.2 11.3 3.9 10.6 8.0 6.8

0.3 -2.0 -3.9 -5.8 -3.4 3.1 -0.5 6.3 -0.4 8.1 -0.4 7.3 -3.8 1.5 3.8 -9.8 -4.3 2.1 4.4 9.4 7.3 4.1 -6.3 5.8 5.3

1.3 7.7 6.7 3.7 28.7 18.2 22.9 15.8 23.7 33.4 14.8 34.3 19.3 46.2 56.5 13.2 7.5 15.3 78.7 68.1 77.5 12.7 3.6 8.8 30.3

4.3 25.7 22.3 12.3 95.6 60.7 76.4 52.7 78.9 111.2 49.3 114.3 64.5 153.8 188.2 43.9 25.0 51.0 262.4 226.9 258.3 42.5 12.1 29.5 101.0

a

Significative intercept (p < 0.05).

Table 4. Calibration Curves, Calculated at Six Concentration Levels, Regression Coefficients (R2), Repeatability (Expressed as the Coefficient of Variation, CV), Accuracy (A), and Limits of Detection (LOD) and Quantification (LOQ), Calculated by Using the Scan/SIM Mode (SIM Data) compd

concn range (µg/L)

slope

intercept

R2

CV (%) (n ) 3)

A (%) (n ) 3)

LOD (µg/L)

LOQ (µg/L)

limonene benzyl alcohol linalool methyl 2-octynoate citronellol citral (neral) geraniol citral (geranial) cinnamaldehyde anisyl alcohol hydroxycitronellal cinnamyl alcohol eugenol coumarin isoeugenol isomethylionone lilial amylcinnamaldehyde lyral amylcinnamyl alcohol farnesol hexylcinnamaldehyde benzyl benzoate benzyl salicylate benzyl cinnamate

108 × 103 to 0.5 224 × 103 to 5.9 204 × 103 to 2.0 244 × 103 to 2.3 220 × 103 to 15.1 154 × 103 to 15.7 188 × 103 to 15.6 154 × 103 to 7.0 204 × 103 to 6.6 228 × 103 to 16.8 276 × 103 to 2.6 200 × 103 to 14.6 204 × 103 to 6.1 236 × 103 to 20.3 352 × 103 to 22.2 236 × 103 to 2.1 224 × 103 to 2.4 118 × 103 to 3.7 102 × 103 to 19.9 236 × 103 to 29.4 248 × 103 to 25.8 110 × 103 to 10.4 180 × 103 to 7.7 484 × 103 to 5.6 110 × 103 to 28.2

0.0269 ± 0.0010 0.0072 ± 0.0003 0.0071 ± 0.0001 0.0049 ± 0.0002 0.0064 ± 0.0002 0.0065 ± 0.0002 0.0090 ± 0.0004 0.0122 ± 0.0003 0.0100 ± 0.0003 0.0035 ± 0.0002 0.0106 ± 0.0003 0.0021 ± 0.0001 0.0057 ± 0.0003 0.0038 ± 0.0002 0.0043 ± 0.0001 0.0079 ± 0.0002 0.0061 ± 0.0002 0.0019 ± 0.0000 0.0006 ± 0.0000 0.0006 ± 0.0000 0.0032 ± 0.0001 0.0021 ± 0.0001 0.0191 ± 0.0006 0.0201 ± 0.0013 0.0173 ± 0.0007

0.3606 ± 0.0628a -0.0155 ± 0.0316 0.0352 ± 0.0101a 0.0023 ± 0.0210 -0.0293 ± 0.0275 0.0040 ± 0.0129 -0.0573 ± 0.0421 0.0232 ± 0.0250 0.0288 ± 0.0322 -0.0345 ± 0.0187 -0.0181 ± 0.0451 -0.0153 ± 0.0124 0.0357 ± 0.0293 0.0154 ± 0.0175 -0.0183 ± 0.0196 0.0537 ± 0.0275 0.0399 ± 0.0246 0.0013 ± 0.0005 -0.0008 ± 0.0009 -0.0060 ± 0.0046 -0.0332 ± 0.0142 0.0030 ± 0.0036 0.0342 ± 0.0545 0.1314 ± 0.1467 -0.0150 ± 0.0341

0.991 0.990 0.999 0.992 0.993 0.996 0.993 0.996 0.993 0.987 0.994 0.983 0.988 0.986 0.994 0.993 0.992 0.999 0.997 0.988 0.995 0.995 0.994 0.987 0.993

0.9 0.6 2.1 1.9 3.0 2.5 0.4 9.6 4.9 2.4 4.7 1.2 2.2 3.4 2.0 1.2 0.1 2.5 2.3 5.6 3.9 2.2 4.0 0.4 1.4

6.7 -2.3 7.0 -0.7 0.3 0.7 -1.1 0.5 0.2 -1.4 0.4 -1.9 6.9 -3.8 -7.7 2.1 2.0 1.7 -4.2 6.3 -3.3 9.4 1.2 8.6 -3.7

0.2 1.8 0.6 0.7 4.5 4.7 4.7 2.1 2.0 5.0 0.8 4.4 1.8 6.1 6.7 0.6 0.7 1.1 6.0 8.8 7.7 3.1 2.3 1.7 8.5

0.5 5.9 2.0 2.3 15.1 15.7 15.6 7.0 6.6 16.8 2.6 14.6 6.1 20.3 22.2 2.1 2.4 3.7 19.9 29.4 25.8 10.4 7.7 5.6 28.2

a

Significative intercept (p < 0.05).

CONCLUSIONS To the best of the authors’ knowledge, the present paper is the first report of true full-scan quantification in GC×GC/qMS using a standard comprehensive 2D GC setup. Under the reported

analytical conditions, more than 15 data points/peak were attained, meeting the requirements related to reliable peak reconstruction. Future research will be focused on the employment of the qMS instrument in other, less common GC×GC systems, such as “splitAnalytical Chemistry, Vol. 82, No. 20, October 15, 2010

8589

Table 5. Allergens Identified and Quantified in a Commercial Perfume (n ) 3)a compd limonene linalool citronellol citral (neral) geraniol citral (geranial) hydroxycitronellal cinnamyl alcohol eugenol coumarin isomethylionone hexylcinnamaldehyde

concn CV (%) 1D tR (mg/L) (n ) 3) (min) 4102.7 2721.1 73.3 455.1 56.1 519.3 12.5 54.8 301.9 187.9 506.9 462.1

8.1 5.3 7.5 3.0 3.9 3.6 7.7 11.4 11.3 1.7 1.8 2.3

12.797 14.755 18.437 18.895 19.163 19.671 20.202 20.889 22.052 24.622 25.213 31.492

SD

2D tR (sec)

SD

0.000 0.001 0.002 0.001 0.044 0.043 0.001 0.002 0.001 0.001 0.044 0.001

2.826 3.288 3.683 4.232 3.777 4.266 1.607 2.304 0.127 1.319 3.770 3.984

0.012 0.047 0.081 0.065 0.035 0.042 0.053 0.071 0.058 0.046 0.046 0.042

a Retention times in the first and second dimensions (1D tR and 2D tR) are reported, along with the standard deviation (SD).

flow” cryogenic systems and flow-modulation instrumentation. Furthermore, the qMS detector will be used in the GC×GC analysis of high molecular weight analytes, such as pesticides. It

8590

Analytical Chemistry, Vol. 82, No. 20, October 15, 2010

is the authors’ opinion that a 33 Hz scan frequency, which will be employed under wider mass range conditions, is sufficient for quantification in a properly optimized GC×GC experiment. Over the past 10 years, in the GC×GC/MS field, it has been affirmed, time and time again, that quadrupole mass spectrometry is fine for identification purposes, but time-of-flight MS is necessary for accurate peak reconstruction. The validity of TOF MS instrumention, for GC×GC analysis, is out of discussion; however, the possibility to use a quadrupole MS instrument for quantification is now an additional option. ACKNOWLEDGMENT We gratefully acknowledge Shimadzu Corp. and Supelco Corp. for continuous support.

Received for review June 25, 2010. Accepted September 5, 2010. AC101678R