LC-MS with SMB

Electron ionization (EI) can significantly benefit LC-MS through the provision of automated library identification and extensive fragment information which is ideally suitable for LC-MS identification of unknown compounds. Thus, bringing back EI to LC-MS is highly valuable if a reliable and robust EI interface can be developed. Motivated by the above challenge we developed a new approach for LC-MS (Israel and USA patents, European and Japan patent applications), aimed at obtaining high quality library searchable electron ionization (EI) mass spectra for a broad range of samples in liquids.

Our novel EI-LC-MS approach is based on interfacing LC and MS with supersonic molecular beams (SMB) and electron ionization of the sample as vibrationally cold compounds in the SMB. The output of a LC was vaporized at about 1-2 Bar inside a glass tube which is connected to a supersonic nozzle via a short fused silica capillary transfer line. The fused silica transfer line serves as a flow impedance to enable effective high pressure sample vaporization combined with 0.1 Bar low pressure behind a 300 µm supersonic nozzle to suppress cluster formation during the supersonic expansion, yet to obtain efficient vibrational cooling. Sample vaporization is based on pneumatic assisted spray formation followed by fast, thermal vaporization of the sample compounds prior to their expansion from the supersonic nozzle. We named this new approach of EI-LC-MS as Capillary Separated Vaporization Chamber and Nozzle system (CSVCN), and we found that CSVCN provides a major improvement of the interface robustness and serviceability, which is similar to APCI. We note that since the sample velocity is over 100 times faster in the capillary transfer line then in the vaporization chamber, the vaporization chamber is the area in which sample thermal degradation should be minimized while it can be neglected in the transfer line, and as soon as the molecules expand from the supersonic nozzle, they are super-cooled and any further dissociation is avoided.

Our approach was experimentally tested and several thermally labile compounds that are not amenable for gas chromatography analysis were successfully analyzed, including underivatized steroids such as stanozolol and corticosterone, drugs such as reserpine sulfamerazine and erythromycin, a vitamin such as beta-carotene, explosives such as TATP, pesticides such as aldicarb and methomyl and polystyrene oligomers up to 1800 amu. High quality, library searchable, EI mass spectra were obtained for all these compounds and no peak tailing was observed with flow injection analysis.

Our novel approach of liquid sampling mass spectrometry with supersonic molecular beams possesses several unique advantages over the current state of the art:

1. Library searchable EI mass spectra are provided (unlike with ESI or APCI/APPI) for positive, legally defensible sample identification.

2. No matrix induced ion suppression or enhancements effects. The collision free conditions of EI of sample compounds in SMB ensure full elimination of any ion molecule reactions with their adverse suppression and/or enhancement effects.

3. The same MS system can serve for both GC-MS and EI-LC-MS with SMB with automated method based conversion from mode to mode of operation.

4. Enhanced molecular ion is provided together with the library searchable fragments for improved confidence level in the identification. Higher quality mass spectra are provided in comparison with Particle Beam-MS or any other liquid sampling MS methods. Cold EI (EI of vibrationally cold molecules in SMB) provides the ultimate structural and isomer mass spectral information and is the best tool for unknown sample identification.

5. Fast LC-MS analysis can be achieved by using AMDIS (automated mass spectral deconvolution identification software of NIST that is provided free with each NIST library) for enabling high throughput analysis combined with sample identification under co-elution conditions. The lack of ion suppression and/or enhancement effects further facilitate fast EI-LC-MS.

6. Non-polar compounds are amenable for LC-MS analysis in addition to the standard range of APCI compounds. Small, and even volatile compounds can be analyzed together with relatively large and thermally labile compounds.

7. Relatively uniform response is observed for a broad range of compounds, including non-polar compounds, due to the uniform ionization efficiency of EI, and transfer of the SMB interface. Thus, even without sample identification, one can obtain an estimate of the concentration of unknown samples and thus determine if their identification is needed.

8. Isotope abundance analysis can be performed for the provision of elemental and isotope labeling information with a unit mass resolution quadrupole mass analyzer.

9. No nitrogen gas generator is required (in contrast to APCI and ESI) since the vaporized solvent serves as the supersonic molecular beam carrier gas. Thus, the cost and laboratory space of the nitrogen gas generator is saved.

10. Higher EI sensitivity is obtained in comparison with Particle Beam-MS. Current minimum detected amount is 2 picogram of a compound at its m/z=774 molecular ion in SIM mode (Figure 5). This detection limit is far better than that of Particle Beam.

11. Low cost. Our vision is of a dedicated EI-LC-MS system that will use the MS system of a standard GC-MS with the addition of one differential pumping chamber and replacement of the EI ion source with our fly through cold EI ion source. The use of limited mass range mass analyzer, simpler ion optics, simpler vacuum system and elimination of nitrogen gas generator could enable the industry lowest cost LC-MS system. Furthermore, the same MS system could serve for both GC-MS and LC-MS with supersonic molecular beams.

Several LC-MS aspects were not yet investigated and require future research, including the use of non-volatile buffers, and real world LC-MS applications.

In Figure 1, the Cold EI mass spectrum of corticosterone in methanol solution is shown in the upper trace, and is compared with the standard NIST EI library mass spectrum shown in the lower trace. Note the similarity of the library mass spectrum to that obtained with the SMB apparatus. All the major high mass ions of m/z 227, 251, 269 and 315 are with practically identical relative intensity and thus good library search results are enabled with NIST library matching factor of 829, reversed matching factor of 854 and 86.5% confidence level (probability) in the corticosterone identification. In addition, the molecular ion at m/z = 346 is now clearly observed while it is practically missing in the library (very small in the shown mass spectrum and absent in the other three replica mass spectra). The relative abundance of the high-mass ion at m/z = 328 is also enhanced. On the other hand, low mass fragments are suppressed and the overall effect of vibrational cooling on the appearance of the Cold EI mass spectrum is of shifting intensity from low mass fragments to high mass fragments and in particular to the molecular ion. Overall, the obtained NIST matching factors are high but usually are not as high as obtained with standard EI. However, the probability factors and confidence level in the identification is actually higher with Cold EI as found in the analysis of many pesticides. We claim that the availability of the molecular ion is of critical importance for correct sample identification with high confidence level, particularly with thermally labile compounds. For example, in the NIST hit list other candidates appear after corticosterone, such as corticosteroneacetate (m/z = 388), which is listed as number two with 10.8% probability, since its standard EI mass spectrum is almost identical to that of corticosterone. The reason for this is that the standard EI MS of both compounds exhibit no molecular ion while having exactly the same fragments (and structure) of m/z = 315 and even a small m/z = 328. This observation is typical in a series of homologous compounds and well known for large aliphatic hydrocarbons that all show mostly m/z = 43, 57, 71 and 85 fragments. Consequently, having both a clear molecular ion peak at m/z = 346 and lack of a any ion peak at m/z = 388 unambiguously indicates that the sample compound is corticosterone and not corticosteroneacetate. Similarly, several other candidates in the NIST hit list are eliminated resulting in almost 100% probability that the sample is corticosterone or one of its isomers (1.2% probability).

In Figure 2 below we show the EI mass spectrum of selected (indicated) drugs, carbamate pesticides and non polar compounds obtained with our system. The molecular ion is clearly shown for all these compounds together with the library searchable high mass fragment ions. These EI mass spectra demonstrate the ability of our system to handle the analysis of thermally labile compounds, combined with the provision of high quality library searchable EI mass spectra to a broad range of polar and non polar compounds.

In Figure 3 below the flow injection cold EI mass spectral analysis of polystyrene oligomers is shown and compared with standard LC with diode array detection. Clearly, the ability to obtain molecular ion to all these oligomers enables the use of much faster flow injection analysis. We note that the demonstrated molecular ion with m/z = 1714 is to the best of our knowledge the highest molecular ion ever obtained in EI (with abundance over 1%). These polystyrene oligomers were also analyzed by LC-EI-MS but we found the flow injection as a faster and more effective way of analysis of these compounds.

In Figure 4 below LC-EI-MS analysis of the thermally labile triacetonetriperoxide (TATP) explosives is shown. The two TATP isomers (conformers) are properly separated and the cold EI mass spectrum that is included in the insert is characterized by a dominant molecular ion. We note that TATP cannot be analyzed by ESI and with APCI its analysis requires additives in order to show a clustered molecular ion.

In Figure 5 below we show our current sensitivity status with the indicated propeller shaped 774 amu compound. In addition, we found that our cold EI linear dynamic range extends well beyond four orders of magnitude.

A. Amirav and O. Granot., "LC-MS with Supersonic Molecular Beams" J. Am. Soc. Mass. Spectrom. 11, 587-591, (2000).

For further Supersonic LC-EI-MS information, please contact me through my E-mail: amirav@tau.ac.il