A novel coupling technique based on thermal desorption gas chromatography with mass spectrometry and ion mobility spectrometry for breath analysis

Seven ketones, such as 2-butanone, 2-pentanone, 2-hexanone, 2-heptanone, 2-octanone, 2-nonanone, and 2-decanone purchased at Sigma–Aldrich (Taufkirchen, Germany) and all having a purity of at least 95%, were used as external standard compounds. Methanol was taken as the solvent and was purchased from Carl Roth (Karlsruhe, Germany) in sufficient purity (99.95%). The in-house generator supplied the nitrogen gas with a purity of at least 99.999%. Helium 5.0 (purity at least 99.999%) as carrier gas was purchased from Messer Industriegase (Siegen, Germany). Gases were filtered with a hydrocarbon trap (Supelpure HC from Supelco, Bellefonte, PA, USA) to increase their purity. For the liquid stock solution, the seven ketones listed above were dissolved in methanol, at a concentration of 0.01 g l⁻¹ for each of the ketones. Pure chemicals and the stock solution were stored at 4 °C. The stock solution was prepared in methanol and renewed every four weeks or whenever contamination was detected.

The developed TD-GC-MS-IMS consists of five main components: a TD unit (TD-30 R, Shimadzu Corporation, Kyoto, Japan), a 4-Port-Splitter (SilFlow 4-Port-Splitter, Shimadzu Corporation, Kyoto, Japan), a GC with a single quadrupole MS (GCMS QP2020 NX, Shimadzu Corporation, Kyoto, Japan), and a drift tube IMS (G.A.S. mbH, Dortmund, Germany). A schematic overview of the setup is shown in figure 1. The following operating parameters of the TD-GC-MS-IMS method are summarized in the supplementary material (table S.1).

Zoom In Zoom Out Reset image size

Analytes were adsorbed on TD tubes, with the procedure described in section 2.3. The samples prepared for analysis, as described in section 2.1, were analyzed in two steps according to TD conditions as follows. During the desorption process, the TD tubes were heated to 250 °C (50 °C min⁻¹) with a helium flow rate of 60 ml min⁻¹ to desorb the VOCs from the sorbent and to focus them onto the Tenax TA trap at −20 °C. Then, the focus trap was rapidly heated to 250 °C and temperature was maintained for 2 min. The analytes entered the GC system with carrier gas for analysis and a split ratio of 1:10. A Rxi-624Sil MS column (30 m × 0.25 mm × 1.4 μm, Restek GmbH Bad Homburg, Germany) was used in the GC operated with a constant pressure of 155 kPa and an initial column gas flow of 1.94 ml min⁻¹ of helium. The oven program was as follows: 40 °C (5 min) → 5 °C min⁻¹ → 150 °C → 10 °C min⁻¹ → 230 °C (5 min) (total run time: 40 min).

Sample detection was performed by a single quadrupole MS and a drift tube IMS in parallel. To achieve parallel detection of MS and IMS, a 4-Port-Splitter was placed behind the GC column. The splitter is placed in the GC oven and, therefore, has a temperature according to the GC oven program. The connection allocation of the ports is displayed in figure 2.

Zoom In Zoom Out Reset image size

The split ratio between the two detectors is determined by the pressure applied to the advanced pressure controller (APC, AFC-2030, Shimadzu Corporation, Kyoto, Japan) and by the length and inner diameter of the restriction capillaries. By adding helium as make-up gas, a defined pressure is built up by the APC. The restriction capillary to the IMS is 1.50 m with an inner diameter of 0.15 mm, and the capillary to the MS has a length of 0.60 m with an inner diameter of 0.10 mm. The determination of capillary dimensions was performed using the detector splitting creator software (Advanced Flow Technology software, Version 1.02, Shimadzu Corporation, Kyoto, Japan). This creates a split ratio of 1:1 at an APC pressure of 40 kPa. It should be mentioned that no pressure below 38 kPa should be selected in order to avoid a backflush and thus a contamination of the APC as 38 kPa represents the pressure inside of the splitter. At 40 kPa a low flowrate of 0.20 ml min⁻¹ avoids any kind of dead volumne at the APC port. Heated transfer lines for MS and IMS were operated at 230 °C. IMS parameters were as follows: drift tube dimension was 15.2 mm × 53 mm, ionization by ³H-source, at 100 °C, a nitrogen drift gas flow rate of 150 ml min⁻¹, and field strength of 500 V cm⁻¹. Spectra were recorded in positive polarity mode with a duration of 21 ms per scan, an injection pulse width of 150 μs, and a blocking and injection voltage of 140 and 2500, respectively, which are both unitless software settings. To reduce data size, six single spectra were averaged, respectively. The MS was equipped with an EI source operated at 230 °C with 70 eV ionization energy scanning a mass range of m z⁻¹ 20–250.

Liquid and breath samples were loaded onto deactivated glass sorbent tubes (Restek GmbH, Bad Homburg, Germany; 6.35 od, lenght 89 mm) filled with Tenax^® TA (60/80 mesh). Prior to use, the tubes were conditioned with the TD CLEAN-CUBE (SiM GmbH, Oberhausen, Germany) under a nitrogen flow of 100 ml min⁻¹ with a temperature program as follows: 40 °C (2 min) → 40 °C min⁻¹ → 270 °C (60 min). For sample collection, a sophisticated device under controlled flow and temperature conditions was developed (flow- and temperature-controlled adsorption unit, FxT-AU). With the FxT-AU, loading both, liquid and gaseous samples, onto TD tubes is possible. The sample application of the liquid standards with the FxT-AU shown in figure 3 works as follows: The TD tube is inserted into a home-made aluminum heating block with the aid of a T-piece, which is heated to 40 °C using temperature sensors (PT100, B + B Thermo-Technik GmbH, Donaueschingen, Germany).

Zoom In Zoom Out Reset image size

1 μl of stock solution was sampled with a 10 μl syringe (Shimadzu Corporation, Kyoto, Japan) and transferred directly to the adsorber material via the septum flushed with nitrogen (100 ml min⁻¹). A mass flow control unit (smart controller GSC-A, Vögtlin Instruments GmbH, Muttenz, Switzerland) maintains a constant and properly adjusted flow through the system. With the nitrogen inlet and outlet, the solvent is blown off for 5 min. Tubes were closed with 1/4 in aluminum caps equipped with a Viton O-ring coated with PTFE (Shimadzu, Corporation, Kyoto, Japan) until analysis.

The breath sample was taken from one volunteer (self-experiment of one of the authors). For this purpose, an ALTEF gas sampling bag (Restek GmbH, Bad Homburg, Germany), was conditioned twice with nitrogen prior to breath sampling. The volunteer then delivered 1 l of exhaled air into the empty bag. This was done by blowing into the bag several times, skipping the first 5 s of breath (mixed expired). The filled gas bag was then connected to the TD tube, which was located in the FxT-AU at 40 °C. A vacuum pump (Microsart mini.vac, Sartorius Lab Instruments GmbH & Co. KG, Göttingen, Germany) was used to transfer all of the exhaled air onto the TD-Tube with a controlled flow rate of 60 ml min⁻¹.

GC-IMS data was analyzed using the software VOCal Version 0.1.4 (G.A.S. mbH, Dortmund, Germany). Peak positions were determined in the 'Live Monitor' by picking the maximum of the selected peaks.

Total ion chromatograms (TIC), acquired by the mass spectrometer (MS), were analyzed using GCMSsolutions 4.52 from Shimadzu. The NIST/EPA/NIH Mass Spectral Library 14 from the NIST of the U.S. Department of Commerce was used for peak identification.

For the determination of reproducibilities, the ketone mixture at a concentration of 0.01 g l⁻¹ was measured ten times, and the coefficient of variation (CV) was calculated. Therefore, the peak height (intensity) after baseline subtraction of each ketone was determined for the MS and the IMS analysis. For calculating the CV for each ketone in percent, the absolute standard deviation of the peak heights was divided by the mean value of the peak heights.

The peak height for each ketone in the MS and IMS TIC was determined for the APC pressure experiment at the splitter. Two measurements were made per pressure level (40, 50, 60, 70, 80, 90, and 100 kPa) used, and the mean value of the signal intensities of each ketone was determined. In addition, the absolute standard deviation of the signal intensities for each ketone was calculated from the duplicates.

The retention indices were calculated using the method proposed by van den Dool and Dec Kratz [35]. A detailed description of the calculation can be found in Augustini et al [29]. This is based on the retention times of the homologous series of n-alkanes (in our case, butane to tetradecane).

As a reference, a mixture of seven ketones (2-butanone, 2-pentanone, 2-hexanone, 2-heptanone, 2-octanone, 2-nonanone, and 2-decanone) in MS and IMS have been examined. The homologous series of ketones was chosen for calculating retention time indices in IMS [29] and are also used here as analytes for the system evaluation.

The measurement results from the GC-IMS are three-dimensional. The first dimension is given by the retention time of the GC, the second dimension is the drift time of the analyte ions detected by the IMS, and the third dimension is obtained from the intensity value of IMS data. Thus, data are typically presented in a topographic plot (here called GC-IMS chromatogram). However, like in figure 4, the topographic plot can be displayed in two dimensions. In this case the intensity is shown with a colored gradient from blue (low intensity) to red (high intensity).

Zoom In Zoom Out Reset image size

Figure 4 shows the TD-GC-IMS chromatogram of a ketone mixture with a concentration of 0.01 g l⁻¹ per substance in methanol. Each number (1–7) represents one compound: (1) 2-butanone, (2) 2-pentanone, (3) 2-hexanone, (4) 2-heptanone, (5) 2-octanone, (6) 2-nonanone, and (7) 2-decanone. Monomers and dimers can be differentiated by the suffixes 'm' and 'd', respectively. The 'continuous' signal of the IMS at a drift time of t_d = 4.7 ms is the positive reaction ion peak (RIP) formed by the ionization of traces of water in the carrier and drift gas. In positive ion mode, the RIP consists mainly of stable protonated water clusters H⁺(H₂O)_n [6]. Detailed information of the retention and drift times are given later in table 2.

Temperature is a critical parameter affecting the adsorption capacity of an adsorbent [36]. Therefore, a sophisticated device for sample collection under controlled flow and temperature conditions was developed (FxT-AU, section 2.3, figure 3). With the FxT-AU, loading both, liquid and gaseous samples, onto TD tubes is possible. To test the reproducibility of sampling, the ketone mixture mentioned before (with a concentration of 0.01 g l⁻¹ for each ketone) was adsorbed onto ten TD tubes using the FxT-AU unit and measured with the TD-GC-IMS-MS. Then, the signal intensities were determined. The calculated CV for the intensities of the seven ketones for MS and IMS measurements can be found in table 1. The coefficients of variation scope from 3.4% to 7.6% for the MS and from 2.2% to 5.3% for the IMS. Although the values for the MS are slightly higher than for the IMS, the sampling and subsequent measurement can be regarded as well reproducible.

If the splitter is used as described in section 2.2 (MS:IMS splitting → 1:1 at 40 kPa), the eluent gas flow (helium) from the restriction capillaries is distributed equally among the detectors. To achieve maximum sensitivity on the mass spectrometric side, the eluent gas flow at the splitter can be shifted entirely towards the MS. If the pressure at the APC is increased (figure 2), the back pressure at the IMS rises, which means that the eluent gas stream with the analytes is directed from the column to the MS to a greater extent. At a defined pressure, the sample is completely directed to the MS. Thus, the IMS is discriminated. The pressure setting of the APC required for this must be determined experimentally.

Before this experiment, it was already known that a particular helium concentration in the ionization region of the IMS could generate plasma due to the high voltage applied. As a result, damage is caused to the ionization region [37]. According to the manufacturer's information, a flow of approx. 10 ml min⁻¹ of helium is tolerable. Therefore, the first step was to determine the flow rate of helium in the IMS as a function of APC pressure. The results are shown in figure 5. At the minimum required pressure for operating the APC of 40 kPa, 0.2 ml per minute of helium flows into the IMS in addition to the carrier gas. For example, if the pressure is increased to 80 kPa, this flow rate goes up to 1.7 ml min⁻¹. The system is operational at APC pressures from 40 kPa to 200 kPa since the total tolerable limit of up to 10 ml min⁻¹ is not exceeded.

Zoom In Zoom Out Reset image size

After establishing that the APC pressure can be maintained to a value of 200 kPa without risking damage to the IMS by igniting a plasma, the effect of APC pressure on signal intensities in the TIC in the MS and IMS was investigated. For the TIC of the GC-MS measurements, the retention times are plotted against the sum of the intensities of the entire range of masses in the spectra of the same scan. Similarly, in IMS, the retention time is plotted against the sum of the data points in the ion mobility spectrum in the range of 4.9 ms–11.0 ms without considering reactant ions. Therefore, the ketone mixture was measured at different APC pressures (40, 50, 60, 70, 80, 90, and 100 kPa). The results can be seen in figure 6, upper part for MS and lower part for IMS.

Zoom In Zoom Out Reset image size

With increasing APC pressure, the IMS signal decreases, and the signal intensities in the MS increase. By changing the APC pressure at the splitter from 40 kPa to 70 kPa, the MS signal intensities for the ketones increased by 44%–67%. Above a pressure of 70 kPa, the IMS is completely discriminated, and for none of the analytes, neither a monomer nor a dimer signal is obtained. If the APC pressure exceeds 80 kPa, a loss of signal intensity can be noted at the MS due to dilution effects, as a significant amount of the make-up gas helium is injected. Thus, between 70 kPa and 80 kPa, the highest MS signal intensity is achieved. It is also noteworthy that higher-chain ketones such as 2-nonanone and 2-decanone can be discriminated in the IMS at a lower APC pressure than short-chain ketones such as 2-butanone and 2-pentanone. The dilution effect has a proportionally stronger effect on the ketones with low volatility than those with higher volatility. It can be summarized that at a pressure of 80 kPa, the maximum sensitivity for very volatiles in the MS can be reached since the focus is mainly on substances found in breath. After showing the possibility of discrimination, the matching in retention time of the MS and IMS measurements is displayed. The TIC of MS and IMS of the ketone mixture at a concentration of 0.01 g l⁻¹ and an applied APC pressure of 40 kPa (MS:IMS, 1:1) are displayed in figure 7, upper part, respectively lower part. Detailed information about the retention and drift times and the retention indices can be found in table 2.

Zoom In Zoom Out Reset image size

When comparing the chromatograms of the ketone analysis from MS and IMS, the retention times of the seven ketones almost coincide, but there are some characteristics. However, the absolute difference in retention times between MS and IMS increases with increasing chain length of the ketones. This is due to a weakness in the design of the sample gas inlet tube of the IMS detector. The sample is transferred to the IMS via a transfer line from the gas chromatographic column. Still, before the entrance of the detector, there is a short unheated piece of tubing with an effect comparable to a so-called cold trap. This also explains the peak broadening of peaks 6 and 7 in the TIC of the IMS in figure 7. Since this does not affect the objective of the topic presented here, it will be considered later.

Higher-chain ketones are retained more strongly, so the absolute difference in the retention time deviation increases. In the range of very volatiles, which is essential for breath gas analysis a very high agreement of the retention times is achieved so that a reliable identification via the correlation is possible. As expected, the retention indices calculated according to van den Dool and Dec. Kratz method also match between MS and IMS. The results obtained here can be related to the coupling system HS GC-MS-IMS by Brendel et al [30] using a dean switch. The comparison of the retention times from MS and IMS for common terpenes and terpene alcohols shows deviations between 0.05 min and 0.26 min in a retention time window of 13 min. The solution presented here, which is more affordable compared to Brendel et al, thus shows the same good performance, but even a retention time range of up to 30 min can be covered very well without any significant differences. For higher retention time shifts, retention indices, according to Augustini et al [29], can be used. This method uses the 2-alkanones as retention time markers to normalize the differing retention times of the GC-IMS and GC-MS so that they can be compared, and the chromatograms can be converted to the same timescale. However, as this method uses 2-alkanones as retention time markers, only the time range covered by these markers (2-butanone to 2-decanone) is available for the correlation. Therefore, substances were tentatively identified using the correlated mass spectra. When available, reference substances were used to confirm the identification, comparing the retention and drift times from the GC-IMS.

In summary, this section showed that using a simple 4-Port-Splitter, an MS and IMS detector can be operated in parallel after a gas chromatographic separation. This has the advantage that the MS databases can be used to identify unknown IMS signals due to the retention time correlation. Furthermore, to decrease the limit of detection of the MS, the IMS can be discriminated against if necessary.

To show a first application of the developed TD-GC-MS-IMS in the field of breath analysis, the author has used his own exhaled air. 1 l of breath collected in a Tedlar Bag was transferred to a TD tube at 60 ml min⁻¹. Figure 8, Part A shows the TD-GC-IMS chromatogram of the collected breath. Especially in the early retention time range (3.0–13.5 min) many signals can be detected. Part B shows a zoomed section of the retention time range from 2.9 to 6.3 min. In this, the first five substances are identified by means of the correlation with the MS measurement and the use of the NIST database as explained in the previous section, thus they are allocated to be ethanol, isoprene, acetone, 2-propanol, and 1-propanol. Additionally, all substances identified via the MS database in the IMS measurements were confirmed by reference substances. Retention and drift times are assigned in table 3.

Zoom In Zoom Out Reset image size

The mentioned compounds are well known for their presence in human breath [38]. Acetone, for example, is used as a marker substance in human breath which can be found in elevated concentrations in patients with type 1 diabetes. The monitoring of acetone can be seen as an opportunity to diagnose this disease non-invasively [39]. Isoprene is one of the most studied volatile compound in respiratory research, not only because it is the primary endogenous hydrocarbon found in exhaled human breath but also because of its potential use as a noninvasive marker for studying various metabolic processes in the body. Its origin was uncertain until now, but Sukul et al [40] have discovered the actual origin of human exhaled isoprene by multi-omic analysis of genes and metabolites. Exhaled 2-propanol in human breath reflects health conditions: There is a relationship with 2-propanol blood concentration and diabetes, so a link of its concentration in the breath can also be suspected [41].