Effects of modular ion-funnel technology onto analysis of breath VOCs by means of real-time mass spectrometry

Proton transfer reaction time-of-flight mass spectrometry (PTR-ToF-MS) is a powerful tool for real-time monitoring of trace concentrations of volatile organic compounds (VOCs). The sensitivity of PTR-ToF-MS also depends on the ability to effectively focus and transmit ions from the relatively high-pressure drift tube (DT) to the low-pressure mass analyzer. In the present study, a modular ion-funnel (IF) is placed adjacent to the DT of a PTR-ToF-MS instrument to improve the ion-focusing. IF consists of a series of electrodes with gradually decreasing orifice diameters. Radio frequency (RF) voltage and direct current (DC) electric field are then applied to the electrodes to get the ions focused. We investigated the effect of the RF voltage and DC field on the sensitivity of a pattern of VOCs including hydrocarbons, alcohols, aldehydes, ketones, and aromatic compounds. In a proof-of-concept study, the instrument operating both as normal DT (DC-mode) and at optimal IF conditions (RF-mode) was applied for the breath analysis of 21 healthy human subjects. For the range of investigated VOCs, an improvement of one order of magnitude in sensitivity was observed in RF-mode compared with DC-mode. Limits of detection could be improved by a factor of 2–4 in RF-mode compared with DC-mode. Operating the instrument in RF-mode allowed the detection of more compounds in the exhaled air compared with DC-mode. Incorporation of the IF considerably improved the performance of PTR-ToF-MS allowing the real-time monitoring of a larger number of potential breath biomarkers. Graphical abstract Electronic supplementary material The online version of this article (10.1007/s00216-020-02846-8) contains supplementary material, which is available to authorized users.


Introduction
Proton transfer reaction mass spectrometry (PTR-MS) is an analytical technique that allows real-time monitoring of volatile organic compounds (VOCs) at low concentrations. It is widely used, e.g. in environmental sciences, food chemistry, homeland security, and breath analysis [1].
Since its introduction in the 1990s [2], PTR-MS has been improved in many ways. Inclusion of time-of-flight (ToF) mass analyzers has substantially overcome the limitations of the first generation of PTR quadrupole-MS (QMS) such as limited mass range and low mass resolution [3][4][5]. Modifications of the hollow cathode discharge ion source allowed to successfully use different chemical ionization agents such as H 3 O + , NO + , O 2 + , Kr + , Xe + , and NH 4 + , improving versatility and selectivity of the instrument [6,7].
Sensitivity of PTR-MS is not solely determined by mass analyzers and detectors but it also depends on the ability to effectively focus and transmit ions from the relatively highpressure drift tube (DT) to the low-pressure mass analyzer. As most of the ions crossing the DT do not exit through the small orifice at the MS interface, a large quantity of ion signal is lost. Ion-funnels (IF) represent a kind of ion guide that enhances sampling of ions through an orifice [8]. In an IF, a radio frequency (RF) voltage and a direct current (DC) electric field are applied to a series of electrodes with decreasing aperture sizes. The electrodes provide strong repulsive potentials at the edge of the electrode, radially focusing the ions. The first demonstration of an IF in PTR-MS was shown by Barber et al. [9]. In their instrument, the whole DT was set up as an IF with the first half used as a standard DT reactor running at a lower reduced electric field compared with the traditional DT. The RF electric field was only applied to the second section with decreasing orifice sizes. González-Méndez et al. [10] used the IF to manipulate the ion-molecule reactions and enhance the selectivity of PTR-MS. Brown et al. [11] reported that in this instrument, ion-focusing and proton transfer reaction both occurred in the IF region. This resulted in vastly different sensitivities for different compounds and in unusual fragmentation patterns. Recently, IONICON Analytik implemented a modular IF into proton transfer reaction time-offlight mass spectrometry (PTR-ToF-MS) instruments. The aim of the present study was to characterize and optimize the IF-PTR-ToF instrument for trace VOC analysis. In a proof-of-concept setup, the instrument was then applied for real-time breath analysis in human subjects.

Ion-funnel PTR-ToF-MS instrument
All investigations were carried out using an online PTR-ToF-MS instrument equipped with a modular IF (PTR-TOF 1000 ultra, IONICON Analytik GmbH, Innsbruck, Austria; firstgeneration model (2017)). Figure 1 shows a schematic view of the instrument. The general working principle of PTR-ToF-MS has been described in several studies [4,5]. Concisely, hydronium ions (H 3 O + ) are produced in a hollow cathode glow discharge ion source from electron ionization of water vapour and are drawn by an electric field into the ion-molecule reaction region (DT where ρ is the electrode orifice radius and δ = d/π where d is the electrode spacing [12]. A radio frequency (RF) voltage and a direct current (DC) electric field are then applied to the electrodes. The DC electric field drives the ions axially through the IF toward the exit aperture. An additional alternating current (AC) is superimposed on the electrodes, with the RF on neighbouring electrodes being phase-shifted by 180°. In this way, the RF field creates a strongly repulsive potential near the surface of each electrode. In combination with the progressively decreasing aperture size, this serves to focus the ions radially.  Within the standard PTR DT, E drift variation is about 15% due to the ratio of the inner diameter and axial distance between the drift rings [14]. The additional RF voltage considerably increases this variation and E/N cannot be properly calculated any more. However, it is possible to define certain voltage settings in RF-mode which enable branching ratios of distinct analytes to be obtained that are comparable with the ones obtained with a classic PTR-ToF-MS instrument operated at a certain actual E/N ("TRU-E/N" method). Nevertheless, in the present paper, the E/N definition is omitted.

Gas standard generation
The IF characterization was performed using a multicomponent gas VOC mix (IONICON Analytik GmbH, Innsbruck, Austria) including methanol, acetonitrile, acetaldehyde, ethanol, 2-propenal, acetone, isoprene, 2butenal, 2-butanone, benzene, toluene, o-xylene, chlorobenzene, α-pinene, and 1,2-dichlorobenzene at concentrations of~1ppmV. The VOC mix was subjected to a 100-fold dynamic dilution in pure nitrogen (purity 5.0, Linde, Vienna, Austria) by means of a liquid calibration unit (LCU, IONICON Analytik, Innsbruck, Austria) to generate a standard mixture with approximately 10 ppbV of each component. The working principle of the LCU involves the introduction of a liquid standard solution into a carrier gas stream, by forcing it through a nebuliser (X175, Burgener Research Inc., UK) and spraying the solution into an evaporation chamber at a defined temperature. This results in a rapid evaporation. The generated gaseous standard mixture can then be measured or collected directly at the output of the evaporation chamber. Two liquid ports (1-50 μl min −1 ), one carrier gas port (1-1000 ml min −1 ), and two additional gas ports (1-100 ml min −1 ), controlled by mass flow controllers (Bronkhorst High-Tech B.V., Ruurlo, Netherlands), enable the generation of complex standard mixtures. VOC standards can be prepared from either liquid solutions or gases, or even from both at the same time. In addition, defined amounts of humidity can be added by adding pure water via one of the two liquid ports [15].
In this study, the LCU flow was kept constant at 1000 ml min −1 for all experiments, the LCU temperature was 75°C and the humidity was adjusted by adding pure water (HPLC grade).

Experimental design
The standard mixture was introduced into the DT via a 1.5 m long polyether ether ketone (PEEK) transfer line (ID: 0.75 mm, Restek, Bellafonte, PA) that was directly connected to the outlet of the LCU. The transfer line temperature was 75°C and the sampling flow was 100 ml min −1 . The signal intensity was recorded for each m/z while the settings of the IF region were varied. Operating the instrument in RF-mode (RF on), the DC electric field applied to the IF was varied in the range of 4.5-27 V/cm while the RF voltage was varied in the range of 40-200 V peak-to-peak (V p-p ) at 4.5 MHz. These testing ranges were decided upon after preliminary measurements showed that these settings approximate the best operating conditions. The entire experimental design was repeated at two different E drift strength, 66 V/cm and 48 V/cm, and using both dry and humid samples (absolute humidity 47 g m −3 ) (ESM Table S2). These two sampling conditions will be referred in the text as "dry" and "humid" conditions, respectively.
When the instrument was operated only with the DC field applied to the IF region (DC-mode), the RF voltage was set to zero and the DC field was set at the same E drift value in order to have a homogeneous electric field along the DT/IF regions.
For the whole experimental design, the DT/IF pressure was 2.3 mbar, the DT temperature was 75°C, and the integration time was 1 s.
Three replicates were measured for each experimental setup, then the results were averaged and background signals were subtracted.

Human breath samples
All experiments were performed in accordance with the guidelines laid down in the Declaration of Helsinki and approved by the ethics committee at the University Medical Center Rostock. Informed consent was obtained from 21 healthy human subjects (aged between 20 and 45 years). Demographic parameters such as height, body weight, age, sex, and smoking habits were recorded for each volunteer (ESM Table S3). Volunteers were asked to breathe spontaneously and continuously over 3 min through a sterile mouth piece directly connected to the PTR transfer line in side stream mode by means of a T-piece. During the first minute of measurement, the PTR-ToF-MS instrument was operated in RF mode: E drift was 66 V/cm, RF voltage was 120 V p-p , and DC field was 13.5 V/cm. During the second minute of measurement, the operating mode of the instrument was switched from RF-mode to DC-mode. During the third minute of measurement, the instrument was operated in DC-mode: RF voltage was 0 V p-p and both E drift and DC field were 66 V/cm.
For breath measurements, the PTR transfer line temperature was 75°C, DT temperature was 75°C, and DT pressure was 2.3 mbar. For breath measurements, the integration time was 200 ms.

Data processing
The ion yields of all m/z were measured in counts per second (cps) and compounds were identified by means of their protonated m/z and isotopic patterns. The normalization of the measured ion intensities to the H 3 O + counts in combination with the water-cluster ion counts is standard practice in PTR-ToF-MS data treatment [16]. However, in the present paper, the normalization to reagent ions was omitted in order to reflect the actual sensitivity of the instrument, which would be masked by normalization.
Both breath and standard files were processed using the software PTR-MS viewer v. 3.2.8 (IONICON Analytik GmbH, Innsbruck, Austria).
For breath measurements, expiratory and inspiratory phases were recognized by means of a custom-made data processing algorithm called "breath tracker" (MATLAB version 7.12.0.635, R2011a). The function of the algorithm has been described previously [17]. Briefly, an endogenous compound that has a sufficiently abundant signal intensity in expiration is used as a tracker to differentiate between alveolar and inspired phases. Acetone, isoprene, or carbon dioxide is usually used for this purpose. Expiratory and inspiratory phases determined by means of the algorithm were then applied to all m/z of interest.    (Fig. 2 (a)) and at RF = 120 V under humid conditions (Fig. 2 (b) 2 and O 2 + intensities were up to six orders of magnitude smaller compared with that of the protonated water dimer. Figure 3 (a, b) shows the variation of the H 3 O + and H 3 O + · H 2 O intensities as function of DC field. Due to the large difference between the intensities of the two reagent ions, H 3 O + · H 2 O intensity is displayed on a second Y-axis. The H 3 O + intensity increased with increasing DC field with its maximum value at DC = 27 V/cm under both dry (Fig. 3 (a)) and humid ( Fig. 3 (b)) conditions. In contrast, H 3 O + ·H 2 O showed its maximum value at DC = 22.5 V/cm under both dry and humid conditions. ESM Fig. S1 (a, b) shows the variation of H 3 O + , protonated water clusters, and O 2 + and NO + measured intensities as function of RF voltage at E drift = 48 V/cm. ESM Fig. S2 (a, b) shows the variation of the H 3 O + and H 3 O + ·H 2 O intensities as function of DC voltage at E drift = 48 V/cm. H 3 O + intensity showed similar trends of those showed at E drift = 66 V/cm. In contrast, substantial differences were found for the protonated In DC-mode at E drift = 48 V/cm under humid conditions, H 3 O + ·H 2 O represent about 65% of the total water reagent ions. In contrast, under dry conditions, they represent about the 15% of the total water reagent ions. In DC-mode at E drift = 66 V/cm, protonated water clusters are present in low concentrations under both dry and humid conditions. Under humid conditions, H 3 O + ·H 2 O represent about 8% of the total water reagent ions; under dry conditions, they represent about 1% of the total water reagent ions.  Intensities of acetaldehyde, methanol, ethanol, 2propenal, and isoprene showed their maximum values at RF = 80 V p-p . In contrast, intensities of acetone, acetonitrile, 2-butenal, and butanone showed their maximum values at RF = 120 V p-p . Aromatic compounds, such as benzene, toluene, o-xylene, chlorobenzene, and α-pinene, showed their maximum intensities at RF = 160 V p-p . Intensity of dichlorobenzene showed its maximum at RF = 200 V p-p . Figure 6 shows the measured intensities of the acetaldehyde, acetone, benzene, and dichlorobenzene as function of DC field. Intensities of most of the investigated compounds showed a substantial increase when the DC field was increased from 4.5 to 13.5 V/cm; when the DC voltage was further increased up to 27 V/cm, they showed variations < 10%. This was with the exception of dichlorobenzene which showed steadily increasing intensity with increasing DC voltage, with its maximum at DC = 27 V/cm.

Fig. 3 Ion intensities in counts per second (cps) of H 3 O + and H 3 O + ·(H 2 O) present in the DT under dry (a) and humid (b)
conditions as a function of the DC field (V/cm). RF voltage was 120 V p-p and E drift was 66 V/cm Table 1 compares sensitivities and limits of detection (LODs) calculated for DC-mode and RF-mode for all the investigated VOCs. Sensitivity is expressed as ion count rate per second per part-per-billion volume mixing ratio of supplied analyte (cps/ppbV). LODs were calculated for 1 s of integration time using the "3σ method" with σ being the standard deviation of the background noise level [18]. The DC-mode data were collected at E drift = 66 V/cm, RF = 0 V p-p , and DC = 66 V/cm under humid conditions. The RF-mode data were collected at E drift = 66 V/cm, RF = 120 V p-p , and DC = 13.5 V/cm under humid conditions. At these conditions, switching from DCmode to RF-mode led to an improvement in sensitivity of about 1 order of magnitude for most of the investigated compounds with the exception of methanol, ethanol, and dichlorobenzene. In contrast, only an improvement by a factor of 2-4 was observed for the LODs in RF-mode compared with DC-mode.

Application in human breath samples
In a proof-of-concept study, the instrument operating both in DC-mode and RF-mode was applied for breath analysis of 21 human healthy subjects. Table 2 contains the list of VOCs that could be detected in exhaled breath. Compounds that could only be detected in RFmode are labelled using bold italic text. Concentrations of these compounds were below the LODs in DC-mode and above the LODs in RF-mode.
Concentrations and LODs and LOQs were calculated applying the kinetic theory [19,20].

Discussion
Incorporation of a modular IF adjacent to the DT led to a substantial improvement in sensitivity and LODs of the PTR-ToF-MS instrument. Improved sensitivities allowed the detection of a broader range of VOCs from human breath samples in real-time.
Intensities determined for water reagent ions (H 3 O + · (H 2 O) n , n = 0, 1, and 2) and for protonated VOCs showed a considerable dependence on RF voltage and DC field applied to the IF region. Highest intensities for H 3 O + were observed in the RF range 80-120 V p-p and at DC = 27 V/cm. At lower and higher RF voltages, the focusing effect of the funnel was lost and ion transfer was less efficient. High RF voltages increase the kinetic energy of molecules. As binding forces in the water clusters are weak when compared with normal chemical bonding, this will lead to collisional decomposition of water clusters long before fragmentation of chemical compounds occurs. Higher H 3 O + intensities in humid samples were most probably due to back diffusion of sample gas from the DT into the ion source generating additional H 3 O + [15,21,22]. O 2 + and NO + were present in low intensities and were observed in RF-mode only at high RF voltage (> 160 V p-p ) as a result of improved ion transmission [13,23].
VOCs showed maximum intensities at substance-specific DC field and RF voltage. In agreement with IF theory, cut-off values occurred at low (< 50 Vp-p) RF voltages for all VOCs and at high (> 160 Vp-p) RF voltages for low-mass compounds (m/z < 90). At low RF voltages, the focusing effect of IF is lost for low and high masses. In a substance-specific way, higher masses show maximum transmission at high (> 120 Vp-p) RF voltages due to the dependency of effective potential onto m/z. Decreasing transmission of low masses at high RF voltages is attributed to diffusional loss of molecules due to the relatively high kinetic energy of low-mass molecules under these conditions [12,24]. In addition, fragmentation may contribute to this effect, as we observed a 10% increase in acetaldehyde fragmentation with increasing RF voltage. Up to 50% fragmentation was reported by Barber et al. under similar conditions. In contrast to oxygen-containing aliphatic substances, aromatic compounds showed increasing ion yields of the protonated monomers even at high RF voltages where non-aromatic substances already exhibited decreasing intensities. Enhanced In contrast to previous setups, with the IF used in this study, sensitivity increases were rather uniform, i.e. approximately one order of magnitude for all investigated compounds. This is a strong indicator that the IF does not have major effects onto the ion chemistry within the DT itself. Therefore, the advantages of PTR-MS, e.g. quantification without calibration, are preserved. Although Brown and Barber et al. reported relative increases in sensitivity of 1-2 orders of magnitude for single compounds (acetaldehyde 45×, acetone 200×), absolute sensitivities for a broad range of compounds reported in our study were in general higher, e.g. 10 times higher for methanol and 2 times higher for acetaldehyde. Higher improvements in relative sensitivity as well as higher fragmentation rates Table 2 List of VOCs that could be detected from exhaled breath in real-time. Compounds that could only be detected in RF-mode compared with DC-mode are labelled using bold italic text. The DC-mode data were collected at E drift = 66 V/cm, RF = 0 V p-p , and DC = 66 V/cm. The RFmode data were collected at E drift = 66 V/cm, RF = 120 V p-p , and DC = 13.5 V/cm Characterization and optimization of DT conditions, RF voltage, and DC field and effects of humidity are of general importance for any IF-PTR-ToF instrument and can therefore be beneficial for the whole community [20]. In addition, the modular IF described in this study can be implemented into several PTR-ToF-MS instruments.
The impact of the IF onto quantification can be seen when LODs and LOQs are looked upon. As the applied IF will improve transmission of the ions, in parallel to the desired effects, increased ion yields will also induce growing background noise. Thus, the "raw" gain in ion counts will not directly translate into identical improvements of LODs and LOQs. LODs and LOQs substantially depend on noise inherent in a PTR-MS signal. This noise can be described by a Poisson distribution: the 1σ error in a measurement that is derived from counting a total of N ions is ffiffiffiffiffiffiffiffiffiffiffiffi ffi N •τ −1 p , with τ being the integration time [19,25,26]. Taking this into account, LODs and LOQs could effectively be improved by a factor of 2-4 when the instrument was switched from DCmode to RF-mode. Hence, just determining increases in ion yields may lead to overestimation of the instrument performances for quantitative analysis. For real-life applications, e.g. trace gas analysis in breath, LODs and LOQs have to be determined to take into account all effects of DC field and RF voltage applied within the IF.
Especially in diseased states, breath VOC concentrations may change quickly and abruptly. Hence, only real-time monitoring can provide complete and comprehensive information from breath VOC analysis [27][28][29]. PTR-ToF-MS with integration time of ≥ 200 ms enables breath-resolved continuous monitoring of breath volatiles. In this pilot study, the range of detectable volatile substances was significantly enlarged through application of IF technology.

Conclusion
The spectrum of detectable VOCs in real-time breath analysis was considerably enhanced through the application of IF technology. The IF can be tuned in order either to obtain the best operating conditions for a specific compound of interest or to realize operating conditions which represent the best compromise for the acquisition of a large number of compounds. In contrast to previous setups, the IF used in this study did not have major effects onto ion chemistry within the DT itself and therefore offers optimal conditions for VOC screening in biomedical applications.