Achieving the Lowest Limits of Identification – GC-MS with Cold EI versus Standard EI with High Efficiency Source

Aviv Amirav, Tel Aviv University and Aviv Analytical Ltd.

Executive Summary  

GC-MS sensitivities are specified with octafluoronaphthalene (OFN). However, for many GC-MS users the most important operational parameter is the sample limits of identification. We compared the Aviv Analytical GC-MS with Cold EI with the Agilent 5977B GC-MS with high efficiency ion source (HES) in sample identification. We found that Cold EI far outperforms the 5977B-HES in both detection and identification limits. Cold EI detected and identified thirteen impurity compounds in a given test mixture while the HES standard EI failed to detect most of these impurities and failed to identify any of them. In this article we demonstrate and discuss several Cold EI benefits of superior sensitivity, better identification capability, greater range of compounds amenable for analysis and faster speed of analysis. The graphical abstract figure above demonstrates the absence of impurity peaks in the TIC with HES versus at least thirteen detected peaks in Cold EI in which each peak exhibits a molecular ion and is amenable for trustworthy identification.   

Introduction - GC-MS Limits of Identification

GC-MS are specified today for their limit of detection (LOD) which is typically at the low and even sub one femtogram for octafluoronaphthalene (OFN) as measured via single ion monitoring (SIM) or full scan with reconstructed single ion monitoring (RSIM). On the other hand, the GC-MS limit of identification (LOI) is ignored although GC-MS is typically used for sample identification and its major strength over LC-MS is in having automated identification capability with names and structures via the use of the EI libraries such as NIST.

Achieving low limit of identification is far more demanding than low limit of detection since identification requires: a) Detection of the sample compound as a peak in the total ion mass chromatogram (TIC) which already is in the pg range; b) Having appropriate separation of the sample compound from other compounds; c) Having good ratio of peak to vacuum background and column bleed interferences; d) Having a trustworthy molecular ion; e) Having informative mass spectral fragment ions for good library based identification and/or structural information, particularly if the compound is not in the library.

In this article we compare the Aviv Analytical 5977-SMB GC-MS with Cold EI and the Agilent 5977 with its high efficiency ion source (HES) in sample identification of several compounds, both of high level and impurities. As will be described and demonstrated, Cold EI far outperforms standard EI in sample identification in all the important identification aspects.

              
Analytical conditions summary

GC-MS Systems: A) The Agilent 7890B + 5977B-HES system that was used is a new 5977B with HES ion source that is working about one month with its initially provided and assembled HP5MS column. The 5977B-HES passed all its installation OFN specifications and shortly after that we achieved with it impressive OFN LOD of 0.2 fg in SIM mode. B) 5977-SMB GC-MS with Cold EI system of Aviv Analytical, based on the combination of an Agilent 5977A MSD (and 7890B GC) from 2014 with the Aviv Analytical supersonic molecular beam interface and its fly-through ion source. The system used has a three years old fly-through ion source that was never serviced.

Sample: Aviv Analytical standard test mixture sample with 10 ppm each n-C16H34, methylstearate, cholesterol and n-C32H66 in hexane. This test mixture was custom made for us by Restek at 1000 ppm concentration and was home diluted to 10 ppm as above. Most of the identified impurities were incurred in this test mixture. 

Injection: 1 µL at 260ºC injector temperature and split 10 using the Agilent auto-sampler. The liners were 4.0 mm I.D. Agilent ultimate inert with gooseneck and glass wool.       

Columns: In the 5977B-HES it was 30 m long, 0.25 mm I.D. and 0.25 µ HP5MS film while in the Cold EI system it was 15 m, 0.32 mm I.D. and 0.1 µ DB1HT film. 

He column flow rate: 1.2 min in the 5977B GC-MS with HES and 8 ml/min with the 5977-SMB GC-MS with Cold EI system.   

GC Oven: For the 5977B-HES it started at 50ºC followed by 10ºC/min to 300ºC and wait 4 min for the total run time of 29 minutes. In the 5977-SMB GC-MS with Cold EI the temperature program start was 50ºC, its program rate was 40ºC/min and upper GC oven temperature was 300ºC for 1.75 min for the total of 8 min analysis time. 

Cold EI Source: 7 mA emission, 70 eV electron energy, 60 ml/min He makeup flow.

5977B-HES EI ion source: The Agilent HES was auto-tuned and operated at 70 eV with 100 µA emission current. The HES ion source temperatures was 300ºC (Cold EI has no relevant source temperature).

Transfer-lines temperature: 300ºC for the 5977B-HES systems and 250ºC with temperature program after 4 min at 10ºC/min to 270ºC for the 5977-SMB system.

Mass spectral range: 50-500 amu at about 3.2 Hz scanning frequency. 

Precautions: Exactly the same sample and vial were used in both systems, injections were automated in both cases with the same model of Agilent small autosampler, the washing solvents were fresh, each experiment was repeated several times, we injected blank samples, we also used both higher split ratio of 40 and splitless injections and we also evaluated the data with other GC-MS systems such as an older 5975-SMB GC-MS with Cold EI and a 5977 with standard EI with Inert ion source. 
 

Results

In the data presented below in Figures 1-4 we compare sample identification with Cold EI and the 5977-HES systems and clearly Cold EI is far superior to the HES in sample identification as it succeeded in the identification of all the 13 little "impurity" peaks while the 5977-HES failed in the detection and identification of all of them.

Figure 1. The analysis of a test mixture with 1 ng on-column (10 ppm split 10) each hexadecane (n-C16H34), methylstearate, cholesterol and n-C32H66 (in order of their elution times) by the Agilent 7890B GC and 5977B MS with HES ion source (upper left) and by the Aviv Analytical 5977-SMB GC-MS with Cold EI (bottom left). The generated mass spectra of n-C32H66 (last to elute compound) are presented at the right side and the upper right mass spectrum was obtained with the HES ion source at 300ºC while the bottom right mass spectrum was obtained with the Cold EI ion source.

 Figure 1 clearly demonstrates several major observations: 

1. Signal strength: The HES exhibits very high signal for 1 ng relatively volatile compounds such as n-C16H34 (the first to elute compound among the four). Furthermore, the total ion count (TIC) signal to noise for the n-C16H34 is ~6000 which is impressive. However, while the HES TIC sensitivity is very high for the volatile n-C16H34 its TIC signal to noise ratio for cholesterol (and n-C32H66) is only 160 while in Cold EI the cholesterol TIC signal to noise ratio is ~2000.   
2. Response uniformity: The response uniformity of the HES is poor even at 300ºC and for example its cholesterol peak height is only ~6% of that of the n-C16H34. Clearly the HES response is reduced with the sample compounds size and 300ºC is requires and even such high ion source temperature is not enough. On the other hand, Cold EI exhibits uniform compound independent response which is important for quantitation of unknowns.      
3. Mass Spectra: The mass spectrum obtained with the HES of n-C32H66 is of poor quality and has zero molecular ion abundance (undetected at any level) and it cannot be identified by the NIST library. Note that since it exhibits no molecular ion with the HES it can-not be identified not only at 1 ng but also at any higher level. In contrast, n-C32H66 exhibits dominant molecular ion in its Cold EI MS together with all the informative fragment ions and it is identified by the NIST library with 72% identification probability. In addition, the Cold EI mass spectrum of n-C32H66 enabled the use of the TAMI software that provided the n-C32H66 elemental formula, and improved the NIST library identification probability to 86% via using isotope abundance analysis with the limited 0.1 amu quadrupole MS mass accuracy. http://www.avivanalytical.com/Isotope-Abundance.aspx  

Figure 2. Zoom on mass chromatograms in the analysis of a test mixture with 1 ng on-column each hexadecane (n-C16H34), methylstearate, cholesterol and n-C32H66 (in order of elution). The upper mass chromatogram was obtained by the Agilent 5977B MS with HES ion source while the bottom mass chromatogram was obtained by the Aviv Analytical 5977-SMB GC-MS with Cold EI. The mass chromatograms were amplified 40 times and the time axis starts just before the elution of methylstearate and ends after the elution of cholesterol and n-C32H66. The names and arrows indicate the identified impurities in the Cold EI mass spectrum.

As demonstrated in Figure 2, clearly the Cold EI mass chromatogram is far richer in information than the HES mass chromatogram. Several observations are made from Figure 2 including:

1. Many more peaks in Cold EI.  The TIC mass chromatogram of Cold EI exhibits many more peaks that are far more abundant than those in the HES standard EI mass chromatogram. In fact, all the HES mass chromatogram peaks (aside the four main compounds) are relatively small, on top of major mass spectral background and none of them can be identified, aside a few siloxanes that emerge from pieces of septa at the liner and not from the sample. The tiny peaks that are indicated as hydrocarbons are indicated as such only via knowing about them from Cold EI and consequently searching peaks with RSIM on m/z=57. However, they remain as "not identified" since their TIC S/N is weak (about or below 1) and their mass spectra are dominated by vacuum background and column bleed and are without any molecular ions and or informative high mass fragments.      
2. Much better TIC sensitivity in Cold EI. Cold EI is much more sensitive in its TIC signal to noise ratio than the HES. For hydrocarbons Cold EI is about 10 times more sensitive while for the other groups of compounds (amides, cholestenone and Irganox 1076) they are detected in Cold EI with typical S/N of 40 for the amides while they are not detected at all in standard EI with the HES.  
3. Abundant molecular ions in Cold EI. As demonstrated in Figure 1 and Figure 4 below the Cold EI mass spectra are characterized by having enhanced molecular ions together with the standard EI fragment ions thus enable positive identification by the library which is supported by knowing the molecular ion identity. Thus, Cold EI provides far better identification than standard EI even with the HES.  

In Figure 3 below we zoom on the mass chromatograms of Figure 1 and 2 around the elution times of cholesterol and n-C32H66. As demonstrated, the HES mass chromatogram has no additional peaks of those sample compounds with low abundance and it is characterized by high degree of ion source related peak tailing that severely erode the separation, reduce the TIC signal and hampers the identification capability. In contrast, the Cold EI mass chromatogram reveals three sample compounds with low abundances of n-C30H62 (left of the cholesterol peak, cholestenone (also named cholesterone) at the right side of cholesterol and n-C31H64 before the big n-C32H64 peak. This is a clear demonstration of the superiority of Cold EI over standard EI (even with the HES) in sample identification. The hydrocarbons are easily identified via their molecular ions and the fragment ions indicate that they are linear chain hydrocarbons and not brunched isomers. In cholestenone the identification is not with full confidence but we know that the sample compound molecular ion is 384.3 which could be cholesterol with a loss of two hydrogen atoms thus tentatively assigned as the cholestenone steroid. 

In Figure 4 we show the Cold EI mass spectra of hexadecanamide, n-C29H60 and Irganox 1076 as obtained from the indicated small TIC peaks shown in Figure 2. The structures shown were taken from the NIST library. The estimated (assuming response uniformity with Cold EI) on-column amount of hexadecanamide is about 30 pg, that of n-C29H60 is ~6 pg and that of Irganox 1076 is ~24 pg. We note that these identifications were achieved at very low sample compounds levels yet due to the enhanced molecular ions and identification capabilities of Cold EI these identifications are with high confidence levels. Hexadecanamide was identified by the NIST library as #2 with 30% identification probability while its isomer Isobutyramide, N-dodecyl was rated as number 1 with 48% identification probability.

Figure 3. Zoom on mass chromatograms of Figure 1 and 2 above around the elution times of cholesterol and n-C32H66 (in order of elution). The upper mass chromatogram was obtained by the Agilent 5977B MS with HES ion source while the bottom mass chromatogram was obtained with the Aviv Analytical 5977-SMB GC-MS with Cold EI.

We are unable to properly differentiate between these two isomeric compounds and such isomer differentiation is the "final frontier" of sample identification. We note that at the Chemspider website we can find for the elemental formula C16H33NO of hexadecanamide a staggering number of 2907 isomers. n-C29H60 was identified by the NIST library as #4 in its list but the first 3 could be easily ruled out and the availability of its molecular ion at m/z=408.4 (high 0.4 amu mass defect) highly supports this identification and the range of fragment ions at the usual CnH2n-1 masses further confirms this identification. Irganox 1076 was not initially identified as we used upper mass range of 500 amu. However, since in Cold EI we expect to have abundant molecular ions and due to the late elution time we realized that we need to increase the upper mass range to 600 amu. With this mass spectral range we obtained dominant molecular ion and NIST identification probability of 97.1% and thus identification is assured.

Figure 4. Cold EI mass spectra of Hexadecanamide, n-C29H60 and Irgaphos 1076 with their structures taking from the NIST library. These mass spectra were obtained from the indicated small TIC peaks shown in Figure 2.

We can divide the "impurities" into five groups of compounds in terms of identification:

1. Amides. We identified two amides, hexdecanamide and octadecanamide. We note that these compounds were not found at all in standard EI with the HES even at the 30 pg on-column amounts, possibly since they are reactive with the ion source metallic surfaces. We also note that we failed to find them even with ten times higher on-column amounts using splitless injection and thus we conclude that they are not amenable for GC-MS with standard EI analysis without derivatization. In contrast, identification was very good with Cold EI with a dominant molecular ion and good NIST library identification but a few isomers remained a possibility. The origin of these amides impurities is unknown and probably they emerge from the solvent that we used.  
2. Hydrocarbons. We found a range of linear chain hydrocarbons from C25H52 up to C34H70 plus the branched squalane C30H62. These saturated hydrocarbons were at the few pg levels (1 up to 6 pg) and were all identified via the availability of their molecular ions and typical fragmentation pattern of CnH2n+1 ions. Squalane has a very typical fragmentation pattern and thus was identified by the NIST library with impressive 69% identification probability. The TIC S/N of these hydrocarbons with the HES standard EI ion source was one or below one and only with 10 times higher on-column amounts using splitless injection they could be identified as hydrocarbons. However, since they do not have molecular ions, compound identification failed with standard EI (with the HES) even at this x10 higher level.     
3. Irganox 1076. Irganox 1076 exhibited dominant molecular ion at m/z=530 and a characteristic high mass fragment ion at m/z=515 and was easily identified with the NIST library with great identification probability of 97.1% and thus identification is assured. In contrast, we failed to find Irganox 1076 in the TIC of the standard EI with HES mass chromatogram even in RSIM on its fragment ions. Thus, 5977-HES completely failed to identify and even failed to detect Irganox 1076 at the 26 pg on-column level which with Cold EI was detected with high TIC S/N of 12 and its RSIM on the molecular ion was detected with S/N of infinite due to zero baseline noise. We note that we failed to find Irganox 1076 with the 5977B-HES even in the splitless mass chromatogram with x10 times higher on-column amount. Since we found (with the x10 times higher amount) that n-C34H70 eluted at 28.4 min at the isothermal plateau of the analysis near its 29 min end, we assume that Irganox 1076 simply did not elute in its standard EI analysis. In contrast, in Cold EI the use of shorter column and 8 ml/min high column flow rate reduced the elution temperatures and thus extended the range of compounds that eluted and could be detected and identified to include Irganox 1076.    
4. Cholestenone. Cholestenone was detected and identified at the high time side of the cholesterol peak. It appeared as an impurity in the original Restek solution. We could easily obtain in RSIM on m/z=384.4 two separate peaks of the weak cholesterol M-2 fragment ion and the molecular ion of cholestenone. This compound MS also includes high mass fragment ions at m/z = 299 and 271. We can-not be sure that this compound is cholestenone but we are certain that its molecular ion is m/z=384 and that it is a steroid from the cholestenone and its isomers family. Cholestenone was fully missed by the 5977-HES that did not exhibit any such peak due to the big HES peak tailing even at 300ºC ion source temperature. In addition, RSIM on m/z=384.4 did not reveal any separate additional peak to the small M-2 cholesterol fragment ion and only at the splitless x10 times higher on-column amount mass chromatogram this cholestenone compound started to perhaps reveal itself with S/N ~1 in RSIM on its molecular and few fragment ions. Thus, this is another clear case of a complete miss by the HES compared with good detection and identification by Cold EI. 
5. OFN and easy to analyze compounds. The test mixture that was explored in this study also included octafluoronaphthalene (OFN) at 100 ppb thus 10 pg on-column amount with split 10 used, and it eluted before the elution times shown in figure 1. OFN is considered as a very easy compound to analyze, with relative very little mass spectral background and dominant molecular ion in standard EI. OFN was detected with the 5977B-HES with S/N (RMS) = 4800 using RSIM on m/z = 272 +- 0.3 and S/N (RMS) = 12600 while using m/z=272+-0.05. This is lower than the OFN specification but good considering almost doubled mass spectral range and different GC oven temperature program used for OFN specification. In Cold EI the S/N was infinite in both RMS and PTP using m/z = 272 +-0.3 since the noise was zero. NIST library identification of OFN with the 5977B-HES resulted in identification probability of 89.2% while in Cold EI it was much better with 97.9% identification probability. The reason for this superior Cold EI NIST identification probability is that the OFN TIC signal in standard EI was only 1.5 times higher than the vacuum background baseline level while in Cold EI it was ten times higher than the baseline.              

Conclusions and discussion               

In this paper we compared GC-MS with Cold EI and GC-MS with the most advanced standard EI ion source in the form of the Agilent 5977B with the HES ion source. As demonstrated, Cold EI far outperforms the HES in sample detection and identification. The test mixture used included four major compounds at the on-column level of 1 ng each and a range of unintended impurities that their detection and identification served for this performance comparison.

Sample identification is more challenging than detection yet it is usually ignored despite its importance as a performance capability. Sample identification requires: a) Detection of the sample compound as a peak in the total ion mass chromatogram; b) Having appropriate separation of the sample compound from other compounds; c) Having good ratio of the identified peaks to vacuum background and column bleed interferences; d) Having trustworthy molecular ions; e) Having informative mass spectral fragment ions for good library based identification and/or structural information if the compound is not in the library. We found that in all these five aspects Cold EI outperforms standard EI in the form of high efficiency ion source (HES) since: 

1. TIC detection. The response of standard EI is not uniform and in general declines with mass and polarity. In addition, the standard EI ion source response is highly non-linear and often vanishes at the low pg range as demonstrated and discussed in the Advanced GC-MS Blog Journal article titled "Linearity, Sensitivity and Response Uniformity Comparison of the Aviv Analytical 5975-SMB with Cold EI and the Agilent 5977A GC-MS with Standard EI".             http://blog.avivanalytical.com/2014/05/linearity-sensitivity-and response.html Accordingly, thirteen impurities were detected in Cold EI with good sensitivity while in standard EI only a few of the hydrocarbons were detected with S/N of about 1 and none was  identified.     
2. Separation. As demonstrated in Figure 3 standard EI with the HES is confronted with major peak tailing that eliminates the separation of cholestenone from cholesterol and similarly masks two other nearby eluting compounds. Cold EI was operated with 15 m 0.32 mm I.D. column and 8 ml/min column flow rate in order to speed up the analysis, lower the elution temperatures below the onset of column bleed and extend the range of compounds that can be analyzed (particularly with column flow program and higher GC oven temperatures that were not employed in this study). These operational conditions should result in reduced GC separation by a factor of 2.5-3 yet since Cold EI is used with a tailing-free fly-through ion source the GC separation with it is preserved and it is superior to that of the HES that is confronted with major ion source peak tailing.
3. Ratio of the peaks to vacuum background. With Cold EI vacuum background is eliminated (zero field ion source with directional sample compounds kinetic energy) and sample compounds elute before the onset of column bleed. Thus, the ratio of the identified peaks to vacuum background is far better in Cold EI than in the HES standard EI. For hydrocarbons this ratio is over 300 times better in Cold EI than in standard EI while the other four impurities are undetected in standard EI.  
4. Having trustworthy molecular ions. This is a prerequisite for good identification in which Cold EI excels while standard EI often fails. In addition, the HES is a harsh EI ion source that exhibits lower molecular ions abundances than other ion sources or the NIST library and a partial reason for this is that it requires higher operational temperatures such as 300ºC to reduce peak tailing which serves to reduce the molecular ions abundances. 
5. Having informative fragment ions and NIST library identification. Cold EI excels in having a combination of enhanced molecular ions and the standard fragment ions and thus it is the only soft ionization method that is fully compatible with NIST library identification. We found that the enhancement of the molecular ion actually improves the NIST library identification probabilities as described in details in "Tal Alon and Aviv Amirav "How Enhanced Molecular Ions in Cold EI Improve Compound Identification by the NIST Library" Rapid. Commun. Mass. Spectrom. 29, 2287-2292 (2015)." 

Accordingly, Cold EI far outperforms the 5977-HES in sample identification and in this paper we described how GC-MS with Cold EI serves to achieve the lowest limits of identification. 

In this paper we also demonstrated several additional benefits of Cold EI in which it is better than standard EI including the ten features below:

1. Significantly higher range of compounds amenable for analysis. This is the most important Cold EI benefit (hexadecanamide, octadecanamide, cholestenone and Irganox 1076 in this study).  
2. Enhanced molecular ions with Cold EI versus reduced molecular ion in the HES.
3. Improved NIST library identification probabilities. 
4. Much faster analysis. In this case the Cold EI analysis is four times faster than the HES (elution of the last to elute major peak at <6 min versus 27 min).    
5. Improved sensitivity particularly for difficult to analyze compounds, as demonstrates in this article.
6. Lowest vacuum background noise. 
7. Inherently inert ion source even for low pg range polar compounds. 
8. Highest (by far) ratio of TIC peaks to column bleed (better S/N and identification). A factor of 300 better Cold EI versus HES TIC to vacuum background and column bleed ratio was measured for hydrocarbons.  
9. No ion source peak tailing for better chromatography and separation.
10. Uniform response for better quantitation (as demonstrated in Figure 1).

Thus, contrary to some incorrect perception Cold EI is not just a method to enhance the molecular ions but it is a superior GC-MS technology in all aspects and its most important benefit is the extension of range of compounds amenable for analysis thereby closing the gap between GC-MS and LC-MS. Accordingly, if Cold EI will be adopted by the major vendors it will bring the future GC-MS revolution.