Gerstel: Intelligent Automation for GC/MS and LC.MS

Multidimensional gas chromatography in combination with accurate mass, tandem mass spectrometry, and element-specific detection for identification of sulfur compounds in tobacco smoke

Introduction. Gas chromatography–mass spectrometry (GC/MS) has been an indispensable technique for identification of volatile compounds. However, one dimensional GC in combination with low resolution mass spectrometry is often insufficient for unequivocal identification of important trace components in complex samples like natural products due to co-elution of various compounds and non-specific electron ionization (EI) mass spectra. GC/MS with simultaneous selective detection can help to locate the region of interest within the complex chromatogram, but lack of sufficient resolution may still preclude reliable identification base on a pure mass spectrum, even after mass spectral deconvolution. An effective way to improve the chromatographic resolution and identification capability is through multidimensional (MD) GC with simultaneous mass spectrometric and element-specific detection.

Instrumentation. Analysis was performed on a 1D/2D GC–SCD/Q-TOF-MS. The Agilent 7890 gas chromatograph was equipped with a TDU thermal desorption unit (GERSTEL), a CIS programmable temperature vaporizing (PTV) inlet (GERSTEL), a MPS robotic autosampler (GERSTEL), a CTS cryo-trap system (GERSTEL), a dual low thermal mass (LTM)-GC system (Agilent), and a SCD (Agilent).

Identification of sulfur compounds. Thirty-five sulfur containing candidates were obtained in a list of 865 compounds with the NIST AMDIS search. Eighteen sulfur containing candidates were obtained in a list of 433 compounds with the Mass Hunter search using unit resolution. Twenty-five sulfur containing candidates were obtained in 837 compounds with the Mass Hunter search using high resolution.

The number one candidate from the NIST library search with all deconvolution conditions was always a non-sulfur compound and obviously an incorrect identification. Also, there were no sulfur compounds among the other candidate compounds from the library search lists for this sulfur peak.

It can be concluded that 1D GC–TOF-MS in combination with automated deconvolution capabilities, including a high resolution mode is often not enough for identification of trace sulfur compounds in tobacco smoke extract which contains thousands of compounds. Therefore, at least a 2D GC separation needs to be added.

Although 8 sulfur candidates were obtained from the NIST library search of the spectrum, the cross search with two different LRIs could narrow the candidates down to 3 compounds without taking elemental information into account. Using an additional filtering with the elemental information obtained from the SCD (sulfur must be present), only methional remained as possible candidate. Although methional was number 2 candidate in the NIST library search results and the ion of m/z 104.0296 in the EI mass spectrum corresponds to methional formula of C4H8OS with the mass error of −0.56 mDa (5.4 ppm), the match factors and the probability in the library search results showed low values and only 39.1% for the probability.

In order to identify additional sulfur compounds in the tobacco smoke extract, the other twenty-seven sulfur fractions selected from the 1D GC–SCD chromatogram were sequentially transferred to the second dimensional separation and then measured with 2D GC–SCD/EI–TOF-MS, 2D GC–SCD/PCI–TOF-MS and 2D GC–SCD/PCI–Q-TOF-MS (MS/MS). In some of the fractions, more than one sulfur compound could be detected in the SCD trace. Identification was performed with the NIST library search, 1D/2D LRI, molecular mass determination, formula calculation, and structure elucidation. Although the CAD mass spectra were not obtained for several sulfur compounds because of a lack of sensitivity in the MS/MS measurements, the combined approach with 1D/2D LRI, formula calculation, and the CAD mass spectrum of the protonated molecule, provided highly probable candidates for some sulfur compounds, which showed no or low NIST library match quality.

Finally the identity of fifteen sulfur compounds (3-1, 7-1, 7-2, 10-2, 10-3, 11-1, 11-2, 16-1, 16-3, 17-2, 18-1, 19-1, 23-2, 24-1, and 26-1) could be confirmed with authentic compounds, and another 15 sulfur compounds were tentatively identified with high probability. Of these 30 sulfur compounds, thirteen (2-2, 7-2, 10-1, 10-2, 11-2, 14-1, 16-2, 16-3, 17-1, 18-1, 21-1, 24-1, and 28-1) have not previously been reported in tobacco smoke.

Conclusion. The combination of LVI, 1D/2D GC, SCD, and Q-TOF-MS with EI and PCI, offers a very effective synergy for identifying trace sulfur compounds in a highly complex sample such as tobacco smoke. The method allows the combined approach using 1D/2D LRI, molecular mass determination, formula calculation, and structure elucidation as well as the NIST library search. Thirty sulfur compounds were tentatively identified with high probability in the flue-cured tobacco smoke extract by sequential heart-cuts of the 28 sulfur fractions using three MS detection modes (SCD/EI–TOF-MS, SCD/PCI–TOF-MS, and SCD/PCI–Q-TOF-MS), while maintaining greater system cleanliness with LVI using the TDU inlet and column back-flushing. Also, the best candidate molecular formulas could be obtained for 11 unknown sulfur compounds. Forty-one sulfur compounds could thus be determined at ng per mg TPM−1 levels.