Design and Evaluation of a Gas Chromatograph-Atmospheric Pressure Chemical Ionization Interface for an Exactive Orbitrap Mass Spectrometer

Abstract

Various separation and mass spectrometric (MS) techniques have furthered our ability to study complex mixtures, and the desire to measure every analyte in a system is of continual interest. For many complex mixtures, such as the total molecular content of a cell, it is becoming apparent that no one single separation technique or analysis is likely to achieve this goal. Therefore, having a variety of tools to measure the complexity of these mixtures is prudent. Orbitrap MSs are broadly used in systems biology studies due to their unique performance characteristics. However, GC-Orbitraps have only recently become available, and instruments that can use gas chromatography (GC) cannot use liquid chromatography (LC) and vice versa. This limits small molecule analyses, such as those that would be employed for metabolomics, lipidomics, or toxicological studies. Thus, a simple, temporary interface was designed for a GC and Thermo Scientific™ Ion Max housing unit. This interface enables either GC or LC separation to be used on the same MS, an Exactive™ Plus Orbitrap, and utilizes an atmospheric pressure chemical ionization (APCI) source. The GC-APCI interface was tested against a commercially available atmospheric pressure photoionization (APPI) interface for three types of analytes that span the breadth of typical GC analyses: fatty acid methyl esters (FAMEs), polyaromatic hydrocarbons (PAHs), and saturated hydrocarbons. The GC-APCI-Orbitrap had similar or improved performance to the APPI and other reported methods in that it had a lower limit of quantitation, better signal to noise, and lower tendency to fragment analytes.

Introduction

As the ability to measure fully all analytes from complex mixtures has progressed, it has become apparent that two main factors still hinder such efforts: The chemical diversity of these mixtures is high, making it difficult to probe the full complexity of the samples with a single analysis technique; and the identity of many analytes is still undetermined. The obstacles facing analysis of complex mixtures have been highlighted with the growth of interest in metabolomics and lipidomics. Often, biological, environmental, and other samples contain large mixtures of compounds that can be similar in composition, which limits the utility of techniques like nuclear magnetic resonance (NMR) to measure all chemical species. For global analyses of mixtures, separation and detection techniques such as liquid chromatography and gas chromatography–mass spectrometry (LC-MS and GC-MS, respectively) are employed since these methods are better suited for such endeavors. However, many studies still focus on analyses using only one type of separation, even though use of both techniques would give a higher coverage complex mixtures, such as the metabolome and lipidome.

Although not addressed herein, the identification of unknown compounds is still difficult via MS. Recently, a common workflow to aid in the identification of unknowns by LC-MS has been established [1], but GC-MS unknown identification has lagged behind. This discrepancy can be attributed to the frequent use of high-resolution mass spectrometry (HRMS) in LC-MS as well as the complexity of fragmentation generated by electron ionization (EI). Until recently, GC had not been routinely paired with HRMS as fast mass analyzers are required to fully utilize the high flow rates of GCs. In 2014, Peterson and coworkers published an article that detailed the coupling of a GC-MS (Thermo Scientific DSQ™ II) with an Orbitrap (Exactive™ Plus) as a possible tool for metabolomics analysis [2]. Thermo Fisher Scientific™ followed with the introduction of a Q Exactive™ GC hybrid quadrupole-Orbitrap mass spectrometer system [3]. These platforms pushed the boundaries of unknown identification, utilizing superior exact mass measurements; however, both still relied on matching fragmentation data to accomplish these tasks. More recently, companies such as Waters Corporation, Bruker, and Agilent have also developed GC-APCI sources for traditional HRMS analyzers, such as the QTOF, as well as low energy EI sources in the push for softer ionization of GC amenable compounds [4,5,6]. Atmospheric pressure photoionization sources (APPI) have alsobeen developed for GC-Orbitraps in order to provide softer, efficient ionization mechanisms for nonpolar systems [7]. These sources rely on analytes that have ionization energies (IE) less than the energies of the photons from the source (direct APPI) or the addition of a doping agent (dopant-assisted APPI), such as toluene, to achieve ionization. The addition of doping agents are most practical when target analytes are already known as the doping agents cause higher background ions and can possibly react with the matrix. Thus, for a complex biological or environmental mixture, direct APPI is best-suited for unknown identification.

In 1973, Horning and coworkers reported atmospheric pressure chemical ionization (APCI) [8], which has become an ionization technique of choice for LC-MS analyses of less polar molecules. Zuth and coworkers have also showed that an APCI-Orbitrap could be used for the detection of organic aerosols introduced through a unique apparatus, thus showing the utility of this instrument for the analysis of molecules introduced in the gas phase [9]. Herein, an interface to allow coupling of a Trace 1300 GC system to a Thermo Scientific™ Ion Max housing is reported. This allows an APCI source to be used for GC-HRMS analyses with the unique capabilities of instruments such as the Orbitrap™ and does not preclude the use of an LC on the same system. This simple, inexpensive GC-MS interface was tested using three well-known, diverse types of analytes [fatty acid methyl esters (FAMEs), polyaromatic hydrocarbons (PAHs), and saturated hydrocarbons] as they encompass the breadth of compounds that are routinely analyzed by GC-MS and are amenable to APCI. Moreover, the GC-APCI interface was benchmarked against a commercially available, aftermarket GC-APPI interface to evaluate and compare their analytical performance.

Experimental Section

Reagents and Materials

The 37 component FAME standard and the C8-C20 alkane standard solution were purchased from Sigma-Aldrich (St. Louis, MO). The 610 Calibration PAH Mix A standard was purchased from VWR International (Radnor, PA). HPLC-grade hexanes from Fischer Scientific (Hampton, NH) were used for preparations of all standards and syringe washes. Ultrahigh purity grade He was purchased from Airgas (Madison, WI). A stainless-steel sheet and silicon O-ring were purchased from McMaster-Carr (Elmhurst, IL). An 8-pin mini-din male to male cable was purchased from Thermo Fisher Scientific (Asheville, NC) and a 6-position, 3.81 mm Mini-COMBICON term block plug was included with the UPLC system.

Gas Chromatography–Mass Spectrometry

GC-MS experiments were conducted on a Trace 1300 gas chromatograph (Thermo Fisher Scientific, Waltham, MA) equipped with a Triplus RSH autosampler and aux temperature/cryo module. The GC interfaces were coupled to an Exactive™ Plus Orbitrap mass spectrometer (Thermo Fisher Scientific Waltham, MA). The commercially available GC-APPI interface, including the transfer line, was purchased from Gentech Scientific Inc. (Arcade, NY).

Fatty Acid Methyl Ester Analysis

FAMEs were separated using a (FAMEWAX, 30 m × 0.25 mm i.d. × 0.25 μm df) polyethylene glycol column (Restek, Bellefonte, PA) and helium as the carrier gas. The FAME standard mix was serially diluted five times in hexanes to obtain concentrations ranging from approximately 5 nM to 400 μM, depending on the individual analyte. Triplicate, 2 μL injections of the standards were directed into a programmable temperature vaporizer (PTV) injection port operated in CT split mode. Split flow was set to 5.1 mL/min (5:1 split ratio), and the temperature of the injection port was set to 240 °C. Analytes were separated on the column using a 1.7 mL/min He flow rate and a 26 min temperature program as follows: initial temperature 80 °C and hold for 5 min, 10.0 °C/min to 240 °C and hold at 240 °C for 5 additional minutes. The exact same standards were injected using both APCI and APPI sources. Analytes were transferred from the GC to the MS via the commercially available heated transfer line set to 240 °C.

While using the APCI ionization source, sweep gas flow rate was set to 3 arbitrary units, aux gas flow rate to 8, and sheath gas flow rate set to 10. The discharge current was set to 10.00 μA, ion transfer capillary temperature to 350 °C, S-lens RF level to 50.0, and vaporizer temperature set to 320 °C. The mass spectrometer was set to full scan from 50.0 to 700.0 m/z with a resolution of 140,000. Data was collected in positive mode, using 1 microscan, automated gain control (AGC) target of 3E6, and maximum inject time of 100 ms.

APPI parameters were set as follows: sheath gas flow rate to 8 arbitrary units, aux gas flow rate to 0, sweep gas flow rate to 0, ion transfer capillary temperature to 280 °C, S-lens RF level of 50.0, vaporizer temperature to 30 °C, and ionization in APPI mode. The remaining parameters—scan range, polarity, microscans, AGC, and maximum inject time—were kept the same as the APCI source.

Polyaromatic Hydrocarbon Analysis

PAHs were separated on a (Select PAH, 15 m × 0.15 i.d. × 0.10 μm df) capillary column (Agilent, Santa Clara, CA) using He as the carrier gas. The PAH standard mix was serially diluted seven times, resulting in approximately 0.1 nM to 800 μM concentrations depending on the individual analyte. Triplicate, 2 μL injections of the standards were directed into the PTV injection port operated in CT split mode. The injection port temperature was set to 315 °C with a split flow of 5.0 mL/min (5:1 split). Analytes were separated on the column using a 0.8-mL/min flow rate with the following 35.57-min temperature program: initial temperature of 100 °C and hold for 2.0 min, 7 °C/min to 230 °C and hold for 7 min, and 50 °C/min to 280 °C and hold for 7 min. The exact same standards were injected using both APCI and APPI sources. The settings or both were the same as those set for the FAME analyses. Analytes were transferred from the GC to the MS via the commercially available heated transfer line set to 280 °C.

Saturated Hydrocarbon Analysis

Alkanes were separated on a (TG1-MS, 30 m × 0.25 mm i.d., 0.25 μm df) capillary column (Thermo Fisher Scientific, Waltham, MA) using He as the carrier gas. The alkane standard mix was serially diluted six times in hexanes to obtain concentrations ranging from approximately 0,1 nM to 350 μM depending on the analyte. Triplicate, 2 μL injections of the standards were directed into the PTV injection port operated in CT split mode. The injection port was set to 330 °C with a split flow of 5.1 mL/min (5:1 split). Analytes were separated on the column using a 1.4 mL/min flow rate and the following 25 min temperature program: initial temperature of 55 °C and hold of 5.0 min and 15 °C/min to 280 °C and hold for 5 min. Each analysis required a new set of standards as the solutions evaporated noticeably. APCI and APPI settings were the same as those set for the FAME analysis. Analytes were transferred from the GC to MS via the commercially available heated transfer line set to 300 °C.

Data Analysis

Instrument control and data acquisition were supervised by Xcalibur software, Version 4.0.27.19. Data files generated by Xcalibur were of the .raw file extension, and these files were converted to .mzML format using msconvert [10] and loaded into MAVEN [11] (Metabolomic Analysis and Visualization Engine). In the case of [M + H]+ ions, peaks were picked based on m/z calculated from chemical formula in positive mode using a ± 5 ppm mass window. Peaks picked based on the most abundant ion for each analyte were based on retention times of the [M + H]+ ion and the exact m/z of the most abundant ion found in Xcalibur.

Results and Discussion

Interface Construction and Installation

The Ion Max housing is designed to allow both easy access and visibility, and the front of the unit contains a hinged, glass window that can be easily removed. Replacement of this window with a machined sheet of stainless-steel containing a hole for a transfer line provided a mechanism for the coupling of the GC to the MS (Figure 1). Earlier work by Lee and coworkers [12] also used this aperture to introduce the eluent from a GC into an Orbitrap fitted with an APPI source, although they did not add additional modifications to the housing other than to remove the window.

Figure 1
figure1

(a) Autocad rendering and (b) photograph of the fabricated stainless-steel insert that replaced the front window of the ion max housing. In (b), the orange, silicon O-ring can be seen recessed into the opening

The opening of the stainless steel piece was recessed, and a silicon O-ring was used to aid in the insertion and removal of the transfer line. A custom transfer line was built and used in the first iteration of the instrument (see Supplementary Data), although it was replaced with the transfer line purchased with the commercial APPI source as the latter was found to have superior temperature stability. This line was inserted to a depth of 8 cm through the stainless steel insert of the Ion Max housing. The other end of the transfer line was inserted into the precut duct of the GC, and the temperature of the transfer line was controlled by the GC via a Quickmate connection (Figure 2).

Figure 2
figure2

(a) Profile and (b) direct view of the completed APCI-GC-Orbitrap interface. The insertion depth of the interface can be seen in (a) and the stainless steel insert can be seen in (b)

Contact closures were used to synchronize data acquisition of the MS with the operation of the GC. These connections employed a male-to-male 8-pin mini-din cable, and one end was connected through the generic handshake port on the back of the GC. Each wire at the opposite end of the 8-pin mini-din cable was separated and stripped since the Orbitrap had an otherwise incompatible 6-pin input configuration and required direct plug in with its existing contact closures. Using the pin configuration of the generic handshake port provided by the manufacturer, as well as measuring the resistance of each wire, it was determined that the wires corresponding to pins 4 and 6 were required to make the contact closure. Pin 4 transmits the start run signal from the GC to the Orbitrap and Pin 6 transmits the end run signal from the GC to the Orbitrap. Through trial and error, the successful contact closure configuration consisted of connecting the wire from pin 4 to the positive digital input and the wire from pin 6 to the negative digital input using a Mini-COMBICON plug. GC devices software (Gentech Scientific Inc., Arcade, NY) was installed on the MS computer to allow for complete control of the autosampler, GC, and MS by ethernet connections. Additionally, it was necessary to change the software configuration of the GC, specifically “start of run out” to “low” and “end of run out” to “high.”

The first parameter to optimize was the placement of the transfer line in order to maximize the ionization for the APCI source. The configuration of the APCI-interface differed from that of either the APPI interface or of the normal source through introduction utilized by the Ion Max housing in that the eluent stream of the GC was pointed directly into the MS inlet rather than positioned at an angle. This was not deemed to be an issue since Thermo Scientific™ nano sources also utilize this geometry; therefore, it was postulated that the distance between the transfer line and the inlet was the more important parameter. It was noted that decreasing distance between the transfer line and the MS inlet had effects on both the number (observed as peak areas) and the composition (observed as m/z) of charged ions. When the transfer line was farther away from the corona discharge needle, there was a greater percentage of [M + H]+ ions, but a decrease in peak area and tailing of the analyte peaks were observed. Conversely, when the transfer line was closer to the corona discharge needle, the percentage of [M + H]+ ions was approximately equal to or less than the most abundant fragment ion for each analyte, and peak area increased by approximately two to threefold. Potential reasons for these changes in ion populations could be due to a variety of factors. A greater distance from the capillary column to the corona discharge would result in increased dispersion of analyte gas phase molecules and a higher probability of interacting with charged atmospheric ion species. When considering molecular interactions at a closer distance, there is a higher probability that an analyte will be directly ionized by the corona discharge and be detected as a charged radical cation as observed in these experiments (Figure 3). Due to the increase in signal and better peak shape, the final operating configuration for the APCI utilized the minimum distance to the corona needle necessary to avoid arcing, and all subsequent APCI analyses were performed done at that distance.

Figure 3
figure3

Side by side comparisons of TICs, EICs of m/z 229.2162, and MS from the transfer line being inserted to a depth of (a) 8 cm and (b) 4 cm. The difference in signal, peak shape, and ratio of [M + H]+ to M is dependent on the distance between the transfer line and corona discharge needle

FAME Analysis

In total, 36 of the 37 FAME standards were separated and quantitated using the chromatographic conditions discussed above (Fig. 4). The only analytes that were not separated were trans-9-elaidic acid methyl ester and cis-9-oleic acid methyl ester. These isomers were not suited for the stationary phase, and the summation of the peak area from both compounds is listed as methyl oleate/elaidate.

Figure 4
figure4

TICs of the 37 component FAME mixture ionized by (a) APCI and (b) APPI with selected S/N values

While the background generated by the APPI source was lower than that generated by the APCI source, the signal generated by the APCI source was more intense for each analyte and averaged one to two orders of magnitude larger. Comparing the performance of the APCI and APPI sources showed that the former performed as well as if not better than the latter in all cases. For the FAME analysis, the linear dynamic range for both ionization methods averaged four orders of magnitude. Correlation of determination (R2) values were greater than 0.9936 for analytes ionized by the APCI source and 0.9627 for analytes ionized by APPI in all cases. When using a S/N of 3:1, the limits of quantitation (LOQs) for each analyte were the same for both ionization methods. However, some analytes had detectable peaks at a lower concentration than the LOQ by APCI; the signal did not reach the required S/N. The LOQ for the APCI source was also good with an average of 111 ± 80 nM and a range of 52.5 to 393 nM, which matched the values obtained by APPI for the analysis of FAMES. The APCI source had increased sensitivity and better R2 values than the APPI source for most compounds, although the detection of the lower molecular weight compounds was more sensitive using APPI. At low concentrations, the variability for the detection of the FAMEs with APPI was of concern, and the apparent increase in sensitivity for this analysis is partially driven by fitting to the highly variable ion counts in this region (see Supplementary Data). Detection of FAMEs below the LOQ was not observed with APPI. The summation of these analytical data is presented in Figure 5.

Figure 5
figure5

Box and whisker plots depicting sensitivity, R2, and LOQ values based on average values for all 36 analytes in the FAME analysis

Detection of FAMEs by GC-MS is a routine, and a rigorous study by Dodds and coworkers reported limits of detection (LODs) for a selection of FAMEs based on flame ionization detection (FID) as well as quadrupole and trapping MS detection [13]. On average, Dodds and coworkers noted that the LOD for FAMEs average 439 ± 103, which compares favorably to the average LOQ observed in this study, and it is likely that the LODs for the reported GC-APCI interface would also be lower than this range since analyte was detected below the LOQ.

FAME Ionization

While each FAME produced a number of ions and fragments, several reproducible trends for these analytes were noted (a complete table of the most abundant ions species generated by each ionization technique can be found in Supplementary Data Table 6. In particular, the APCI source primarily generated m/zs indicative of [M + H]+ (M + 1.0072 AMU) and [M] (M − 0.0005 AMU) molecular ions as well as ions generated due to fragmentation of the FAMEs with a reproducible loss of − 32.0267 AMU ([M − 32.0267]+) (Supplementary Data Table 1). All but one of the 36 analytes had 40% or greater relative abundance of the [M + H]+ species using the APCI source. The APPI source also generated m/z indicative of [M + H]+ and [M], although the number of FAMEs whose base peak was the [M + H]+ was lower than for APCI, with only 22 of the 36 analytes having 40% or greater relative abundance of this species. The number of major fragments for the APPI source was also greater than those for the APCI source, and a fragment with a loss of − 45.0347 AMU ([M − 45.0347]+) was observed along with the [M − 32.0267]+ (Supplementary Data Table 1).

PAH Analysis

The analysis of a mixture of 16 PAHs again showed that APCI performed favorably when compared to APPI. The linear dynamic range for the APCI source averaged four and a half orders of magnitude, with a range of 18.1 to 78.4 nM while that for the APPI source averaged four orders of magnitude with a range of 18.1 nM to 56.1 μM. The dominate ionization mechanism for both APCI and APPI was loss of an electron yielding a radical cation; the sensitivity for this ion averaged 1.1153 with a range of 1.0802 to 1.2314 and 0.9819 with a range of 0.9114 to 1.0820 for APCI and APPI, respectively. The R2 values were better than 0.9979 in all cases. While the average LOQs for both APCI and APPI were the same for the [M] (35.0 nM), it was noted that APCI also produced a reasonable population of [M + H]+ that could be detected with only a one order of magnitude loss in performance with an average LOQ of 291 nM. Detection of the [M + H]+ via APPI was more difficult, with most LOQs being two orders of magnitude worse than those for the [M] (6.0 μM average). A summary of these data is presented in Figure 6. When both sensitivity and LOQ are taken into account, the APCI interface performs slightly better than the APPI interface.

Figure 6
figure6

Box and whisker plots depicting sensitivity, R2, and LOQ values based on average values for all 16 analytes in the PAH analysis. In the plot depicting LOQs, [M + H]+ values are assigned to the left axis and base peak values are assigned to the right axis

EPA Method 610 is widely utilized to detect PAHs and reports a GC-MS analysis. However, no analytical metrics for this technique are reported in the document. Alawi and coworkers have recently published a validated GC-MS method for detection of PAHs using a mass selective quadrupole detector in selective ion monitoring mode [14]. They reported a dynamic range of 0.01 to 5.0 μg/mL with LODs ranging from 0.0003 to 0.0089 μg/mL, and the performance of the APCI interface reported herein is nearly identical. These data confirm that Orbitrap MSs operated in a full scan mode can perform as well as or better than quadrupole instruments using targeted analyses.

PAH Ionization

Ionization of the PAHs by both the APCI and APPI sources was rather routine with both generating m/z indicative of radical cations ([M]) as the most abundant ion with little to no fragmentation (Supplementary Data Table 3A and 3B). However, APCI was also able to generate [M + H]+ ions, although with only an average relative abundance of 40 ± 5% (Supplementary Data Table 7). It is of note that the mix of [M] and [M + H]+ did not lead to a significant loss in detection sensitivity for the APCI as this source led to an overall higher average population of ions for the PAHs.

Saturated Hydrocarbon Analysis

Though previous work, such as that of Marotta and Paradisi [15], has shown that ionization of saturated hydrocarbons is possible by APCI, such applications have been overlooked when evaluating the usefulness of this ionization method for the detection of nonpolar compounds [16]. Therefore, a mixture of 13 alkanes was analyzed using both the constructed GC-APCI and commercial GC-APPI systems. It was observed that both APCI and direct-APPI sources ionized alkanes, although with differing base peaks. The average sensitivity for the alkanes was 1.1592 ± 0.2537 with a range of 0.3393 to 1.3288 for the APCI source and 1.2207 ± 0.0683 with a range of 1.0709 to 1.2712 for the APPI source. If the sole outlier by the APCI source was to be excluded, the averages and standard deviations for both sources would be nearly identical. The average LOQs for the APCI and APPI source were both 2.2 ± 0.6 μM with a range of 1.4 to 3.5 μM. These LOQs (Figure 7) were not as good as for the other compound classes studied; however, both the APCI and APPI sources provided detection at levels comparable to other reported methods [17].

Figure 7
figure7

Box and whisker plots depicting sensitivity, R2, and LOQ values based on average values for all 13 analytes in the saturated hydrocarbon analysis

Saturated Hydrocarbon Ionization

Saturated hydrocarbons present a unique analytical challenge since soft, atmospheric pressure chemical ionization processes typically rely on protonation for positive mode, which is not amenable to these analytes. Although GC-APPI sources generate photons to be absorbed by analytes for direct ionization and are amenable to hydrocarbon analyses, EI remains the most common ionization source used for alkane analyses. Due to the generation of a mixture of positively charged fragments, neutral radicals, and radical cations; EI sources produce several characteristic m/z values: a low intensity [M], a “ski-slope” pattern of small carbenium ions, and repeating peaks 14 m/z units apart [18]. In 1993, Bell and coworkers used an APCI source to ionize gas phase alkanes and reported [M − 1]+ and [M − 3]+ as characteristic ions, most likely generated via proton transfer pathways [19]. With these considerations, it was found that the GC-APCI ionized saturated hydrocarbons with a “ski-slope” pattern of small ions analogous to that of EI for each analyte (Supplementary Figure 5A). The ions closest in mass to the parent alkane were [M − 1.0073]+ and [M + 12.9707]+, and the latter species were on average 17.1 ± 5.5% of the base peak. The relative abundance of the [M − 1.0073]+ ions increased and the relative abundance of the [M + 12.9707]+ ions decreased as the analytes increased in molecular weight (Supplementary Data Table 8). The most abundant “ski-slope” ion for all compounds had a mass of 57.0707 m/z, and an average relative abundance of 96.6 ± 7.1% (Supplementary Data Table 8). The [M − 1.007276]+ ion could be detected for tetradecane and larger hydrocarbons, although with a very low relative abundance of no more than 3.3%. Overall, APCI produced characteristic alkane fragments similar to those observed with EI as well as unique species.

The APPI also produced ions with reproducible trends. The base peak for this ionization source was [M + 28.9655]+, although [M + 12.9707]+ species were also observed for each analyte with an average relative abundance of 23.9 ± 2.2% (Supplementary Data Table 8). The relative abundances of the “ski-slope” ions also increased with increasing molecular weight of the analyte (Supplementary Data Table 8). Finally, the most prominent ion (69.0706 m/z) only had a maximum relative abundance of 10% for lighter compounds and this peak decreased as the molecular weight of the analyte increased. Although APPI provides unique adducts for saturated hydrocarbons, identifying the analytes from complex samples using mass alone could be challenging since the ions would be hard to distinguish from those arising from oxidized hydrocarbons.

Role of Ionization in Identifying Unknowns

The analysis of complex mixtures still remains a challenge, and MS has become an important tool for measuring the entire molecular content of many systems. One of the biggest challenges faced in these experiments is the identification of analytes, which is hindered by a lack of knowledge concerning their structure as well as issues with detecting unmodified, intact molecules due to factors such as in-source fragmentation, oxidation, and adduct formation [16]. When considering unknown chemical species in a complex mixture, another important facet is also the confidence to accurately predict how the unknown will be ionized. For example, a single analyte can often be ionized as the result of protonation, deprotonation, the loss of an electron, etc., and a mix of ionization types can be observed under more energetic ionization conditions. Beyond this, fragmentation and rearrangement of the ions can occur, and this phenomenon is much more prevalent under harder ionization conditions [20]. Current databases and computation resources that incorporate tools, such as Fiehn’s seven golden rules [21], to help predict molecular formulae and/or structure require knowledge of predicted ionization mechanisms in order to determine the neutral mass of the detected ions and often require the user to input the expected neutral mass for each compound. Therefore, it is critical to not only understand the ionization chemistry when dealing with mixtures of unknown compounds but also to use ionization sources that produce reasonable populations of intact ions. With these criteria in mind, the APCI source is well suited for such experiments as it reliably produced [M + H]+ species for the majority of analytes with LOQs that are similar to or better than both the APPI source studied and other reported methods. Although saturated hydrocarbons posed a slightly more difficult analytical challenge due to the inability to reliably produce [M + H]+ ions, molecular ion peaks were still produced by APCI. When they were not the base peak, the molecular ions still had 16–50% of the intensity as the most abundant ion whereas, the populations of these ions for analytes analyzed by the APPI source were < 1% that of the base peak. Further investigation into the identity and mechanistic formation of these [M + 12.9707]+ and [M + 28.9655]+ species is ongoing.

Conclusion

These data show that GC-APCI-MS offers an alternative to traditional GC-EI-MS, and that the interface of this source to an Orbitrap utilizing the Ion Max housing is inexpensive and requires only temporary and simple modification of the instrument. Therefore, both LC and GC can be used on the same MS with minimal down time between analyses. Further, the analytical performance of the GC-APCI source compares favorably with both EI and APPI sources, while also providing relatively soft ionization of the analytes. Compounds of varying polarity (FAMEs, PAHs, and alkanes) that are routinely analyzed by GC-MS were studied, and as an underutilized technique in GC-MS analyses, APCI shows great potential for the detection of both known and unknown analytes.

References

  1. 1.

    Wang, L., Xing, X., Chen, L., Yang, L., Su, X., Rabitz, H., Lu, W., Rabinowitz, J.D.: Peak Annotation and Verification Engine for Untargeted LC–MS Metabolomics. Anal. Chem. 91, 1838–1846 (2019)

    CAS  Article  Google Scholar 

  2. 2.

    Peterson, A.C., Hauschild, J.-P., Quarmby, S.T., Krumwiede, D., Lange, O., Lemke, R.A.S., Grosse-Coosmann, F., Horning, S., Donohue, T.J., Westphall, M.S., Coon, J.J., Griep-Raming, J.: Development of a GC/Quadrupole-Orbitrap Mass Spectrometer, Part I: Design and Characterization. Anal. Chem. 86, 10036–10043 (2014)

    CAS  Article  Google Scholar 

  3. 3.

    Weidt, S., Pesko, B., Cojocariu, C., Silcock, P., Burchmore, R.J., Burgess, K.: Untargeted Metabolomics Using Orbitrap-Based GC-MS.

  4. 4.

    Portolés, T., Sancho, J., Hernández, F., Newton, A., Hancock, P.: Potential of Atmospheric Pressure Chemical Ionization Source in GC-QTOF MS for Pesticide Residue Analysis. J. Mass Spectrom. 45, 926–936 (2010)

    Article  Google Scholar 

  5. 5.

    Yan, Y., Rempel, D.L., Holy, T.E., Gross, M.L.: Mass Spectrometry Combinations for Structural Characterization of Sulfated-Steroid Metabolites. J. Am. Soc. Mass Spectrom. 25, 869–879 (2014)

    CAS  Article  Google Scholar 

  6. 6.

    Jaeger, C., Tellström, V., Zurek, G., König, S., Eimer, S., Kammerer, B.: Metabolomic Changes in Caenorhabditis elegans Lifespan Mutants as Evident from GC–EI–MS and GC–APCI–TOF–MS profiling. Metabolomics. 10, 859–876 (2014)

    CAS  Article  Google Scholar 

  7. 7.

    Kersten, H., Kroll, K., Haberer, K., Brockmann, K.J., Benter, T., Peterson, A., Makarov, A.: Design Study of an Atmospheric Pressure Photoionization Interface for GC-MS. J. Am. Soc. Mass Spectrom. 27, 607–614 (2016)

    CAS  Article  Google Scholar 

  8. 8.

    Horning, E., Horning, M., Carroll, D., Dzidic, I., Stillwell, R.: New Picogram Detection System Based on a Mass Spectrometer with an External Ionization Source at Atmospheric Pressure. Anal. Chem. 45, 936–943 (1973)

    CAS  Article  Google Scholar 

  9. 9.

    Zuth, C., Vogel, A.L., Ockenfeld, S., Huesmann, R., Hoffmann, T.: Ultrahigh-Resolution Mass Spectrometry in Real Time: Atmospheric Pressure Chemical Ionization Orbitrap Mass Spectrometry of Atmospheric Organic Aerosol. Anal. Chem. 90, 8816–8823 (2018)

    CAS  Article  Google Scholar 

  10. 10.

    Holman, J.D., Tabb, D.L., Mallick, P.: Employing ProteoWizard to Convert Raw Mass Spectrometry Data. Curr. Protoc. Bioinformatics 46, 13.24. 11-13.24. 19 (2014)

  11. 11.

    Clasquin, M.F., Melamud, E., Rabinowitz, J.D.: LC-MS Data Processing with MAVEN: a Metabolomic Analysis and Visualization Engine. Curr. Protoc. Bioinformatics. 37, 14.11. 11–14.11. 23 (2012)

    Google Scholar 

  12. 12.

    Lee, Y.J., Smith, E.A., Jun, J.H.: Gas Chromatography-High Resolution Tandem Mass Spectrometry Using a GC-APPI-LIT Orbitrap for Complex Volatile Compounds Analysis. Mass Spectro. Lett. 3, 29 (2012)

  13. 13.

    Dodds, E.D., McCoy, M.R., Rea, L.D., Kennish, J.M.: Gas Chromatographic Quantification of Fatty Acid Methyl Esters: Flame Ionization Detection vs. Electron Impact Mass Spectrometry. Lipids. 40, 419–428 (2005)

    CAS  Article  Google Scholar 

  14. 14.

    Alawi, M.A., Abdullah, R.A., Tarawneh, I.: Determination of Polycyclic Aromatic Hydrocarbons (PAHs) in Carbon Black-Containing Plastic Consumer Products from the Jordanian Market. Toxin Rev. 37, 269–277 (2018)

    CAS  Article  Google Scholar 

  15. 15.

    Marotta, E., Paradisi, C.: A Mass Spectrometry Study of Alkanes in Air Plasma at Atmospheric Pressure. J. Am. Soc. Mass Spectrom. 20, 697–707 (2009)

    CAS  Article  Google Scholar 

  16. 16.

    Li, D.-X., Gan, L., Bronja, A., Schmitz, O.J.: Gas Chromatography Coupled to Atmospheric Pressure Ionization Mass Spectrometry (GC-API-MS): Review. Anal. Chim. Acta. 891, 43–61 (2015)

    CAS  Article  Google Scholar 

  17. 17.

    Damas, E.Y.C., Medina, M.O.C., Clemente, A.C.N., Díaz, M.Á.D., Bravo, L.G., Ramada, R.M., Porto, R.M.d.O.: Validation of an Analytical Methodology for the Quantitative Analysis of Petroleum Hydrocarbons in Marine Sediment Samples. Química Nova. 32, 855–860 (2009)

  18. 18.

    Watson, J.T., Sparkman, O.D.: Introduction to Mass Spectrometry: Instrumentation, Applications, and Strategies for Data Interpretation (4th ed.). John Wiley & Sons, New Jersey (2007)

  19. 19.

    Bell, S.E., Ewing, R.G., Eiceman, G.A., Karpas, Z.: Atmospheric Pressure Chemical Ionization of Alkanes, Alkenes, and Cycloalkanes. J. Am. Soc. Mass Spectrom. 5, 177–185 (1994)

    CAS  Article  Google Scholar 

  20. 20.

    Portolés, T., Pitarch, E., López, F.J., Hernández, F., Niessen, W.M.A.: Use of Soft and Hard Ionization Techniques for Elucidation of Unknown Compounds by Gas Chromatography/Time-of-Flight Mass Spectrometry. Rapid Commun. Mass Spectrom. 25, 1589–1599 (2011)

    Article  Google Scholar 

  21. 21.

    Kind, T., Fiehn, O.: Seven Golden Rules for Heuristic Filtering of Molecular Formulas Obtained by Accurate Mass Spectrometry. BMC Bioinformatics. 8(105–105), (2007)

Download references

Acknowledgements

The authors thank Mr. Tim Free and Mr. Danny Hackworth for the assistance in machining parts and Mr. Matthew Nalepa for the help with the electrical components of the interface. We also acknowledge Dr. Hector F. Castro (Biological and Small Molecule Mass Spectrometry Core, UTK) and Dr. Brandon J. Kennedy for the helpful discussions and instrumental assistance as well as the reviewers for their insightful commentary into the first draft of the manuscript. Mr. Joshua B. Powers was supported by NSF award MCB-1615373. Instrumentation was provided by both NSF award DBI-1530975 and the University of Tennessee Institute of Agriculture.

Author information

Affiliations

Authors

Corresponding author

Correspondence to Shawn R. Campagna.

Electronic Supplementary Material

ESM 1

(PDF 1252 kb)

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Powers, J.B., Campagna, S.R. Design and Evaluation of a Gas Chromatograph-Atmospheric Pressure Chemical Ionization Interface for an Exactive Orbitrap Mass Spectrometer. J. Am. Soc. Mass Spectrom. 30, 2369–2379 (2019). https://doi.org/10.1007/s13361-019-02311-6

Download citation

Keywords

  • Gas chromatography
  • Atmospheric pressure chemical ionization (APCI)
  • Atmospheric pressure chemical ionization (APPI)
  • GC-MS
  • Fatty acid methyl esters (FAMEs)
  • Polyaromatic hydrocarbons (PAHs)
  • Saturated hydrocarbons
  • Interface
  • Orbitrap