Buffer Gas Modifiers Effect Resolution in Ion Mobility Spectrometry through Selective Ion-Molecule Clustering Reactions

Abstract

RATIONALE

When polar molecules (modifiers) are introduced into the buffer gas of an ion mobility spectrometer, most ion mobilities decrease due to the formation of ion-modifier clusters.

METHODS

We used ethyl lactate, nitrobenzene, 2-butanol, and tetrahydrofuran-2-carbonitrile as buffer gas modifiers and electrospray ionization ion mobility spectrometry (IMS) coupled to quadrupole mass spectrometry. Ethyl lactate, nitrobenzene, and tetrahydrofuran-2-carbonitrile had not been tested as buffer gas modifiers and 2-butanol had not been used with basic amino acids.

RESULTS

The ion mobilities of several diamines (arginine, histidine, lysine, and atenolol) were not affected or only slightly reduced when these modifiers were introduced into the buffer gas (3.4% average reduction in an analyte's mobility for the three modifiers). Intramolecular bridges caused limited change in the ion mobilities of diamines when modifiers were added to the buffer gas; these bridges hindered the attachment of modifier molecules to the positive charge of ions and delocalized the charge, which deterred clustering. There was also a tendency towards large changes in ion mobility when the mass of the analyte decreased; ethanolamine, the smallest compound tested, had the largest reduction in ion mobility with the introduction of modifiers into the buffer gas (61%). These differences in mobilities, together with the lack of shift in bridge-forming ions, were used to separate ions that overlapped in IMS, such as isoleucine and lysine, and arginine and phenylalanine, and made possible the prediction of separation or not of overlapping ions.

CONCLUSIONS

The introduction of modifiers into the buffer gas in IMS can selectively alter the mobilities of analytes to aid in compound identification and/or enable the separation of overlapping analyte peaks.

INTRODUCTION

Drift tube ion mobility spectrometry (IMS) is a time-of-flight analytical technique that uses the distinctive mobilities of ions in a gas phase for the separation and identification of analytes. Ion mobility spectrometry has been applied to the separation and detection of a wide variety of chemicals such as heroin and cocaine,^[1] explosives,^[2] halogenated compounds,^[3] explosives in hair,^[4] veterinary drugs in meat,^[5] methamphetamines in human serum,^[6] pharmaceuticals for equipment cleaning verification,^[7] carbohydrate isomers,^[8] complex peptide mixtures,^[9] large non covalent protein-ligand and protein-protein complexes,^[10] metabolites,^[11] and human plasma proteins.^[12] Several comprehensive reviews of IMS applications are available^[13,14] that demonstrate the breath of applications of this rapidly developing analytical separation technique. As with all separation methods, however, some compounds (ions in the case of ion mobility spectrometry) elute at a similar time and are not resolved from one another. The objective of this work is to investigate methods of modifying the buffer gas to effect the resolution of difficult-to-resolve ions.

In IMS, separation is based on the different velocities that ions acquire under the influence of an electric field. Because the velocities of ions are an inverse function of their size-to-charge ratios, ions of similar size have similar velocities and are difficult to separate by IMS. In electrospray ionization ion mobility spectrometry, singly charged, stable gas-phase ions are created in an electrospray ionization source. To generate these ions, ion mobility spectrometers used with electrospray ionization have a desolvation region where the charged droplets that are generated from the electrospray process are stripped of solvent molecules by a countercurrent of preheated buffer gas. Ions are pulsed into the drift region, where they are accelerated by an electric field, but are continuously decelerated by collisions with the buffer gas. This combination of collisions and accelerations thermalizes the ions and averages their velocities to distinct values that can be used to calculate a characteristic parameter called the mobility constant, K:^[15]

$$K=\frac{\nu }{E}=\frac{L^2}{V.t_d}$$ (1)

where v is the velocity of the ion in cm s⁻¹, E the electric field in the drift region in V cm⁻¹, L the length of the drift region in cm, V the total voltage drop in volts across the drift region, and t_d the time that the ion spends traveling the distance L in s. The velocity, ν, is proportional to E for electric fields of less than approximately 500 V cm⁻¹ (at ambient pressure).^[16] In IMS, ion mobilities are normalized to standard pressure and temperature as reduced mobilities (K₀ , cm²V⁻¹s⁻¹), which are constants useful for identification purposes. This standardization allows comparison of results in different laboratories by correcting for variations in environmental and instrumental conditions:

$$K_0=K\frac{P}{760}\frac{273}{T}$$ (2)

where P is the pressure in the drift tube in Torr and T the buffer gas temperature in Kelvin.^[16] However, as noted as early as the beginning of the 20^th century, these mobilities can be affected by the presence of neutral contaminants such as moisture in the buffer gas of an ion mobility drift tube.^[17,18] Recently, modification of the buffer gas by the introduction of small quantities of organic vapors has been used to vary the arrival times of analyte ions. The addition of ketones to the buffer gas allowed the separation of ammonia from hydrazines^[19,20] and the addition of 2-butanol enabled the separation of enantiomers.^[21]

In this paper the term “modifier” refers to traces of vapors added to the buffer gas to modify the ion separation characteristics while the term “dopant” refers to traces of vapors added to the ionization region of the IMS system to produce selectivity. Dopants mainly have been introduced to preferentially ionize compounds with higher proton affinities (as in the case of drugs) or higher electronegativities (as in the case of explosives). ^[20] Dopants that have been added to ion mobility spectrometers include ammonia,^[22] chloride ions,^[23] dichloromethane, methyl iodide, acetic acid, dimethyl sulfide, acetonitrile,^[24] acetone,^[25-27] dimethylsulfoxide,^[26-28] water,^[28] 5-nonanone,^[27] 4-heptanone,^[19] and ketones.^[20] A comprehensive review on the introduction of dopants in IMS was published by Puton et al. in 2008.^[29]

In this work, we evaluated three new modifiers, ethyl lactate, nitrobenzene, and tetrahydrofuran-2-carbonitrile, and compared their effects on ion mobilities with that of 2-butanol, which was used previously. In addition, the mechanism of mobility changes as a function of buffer gas modification was investigated.

EXPERIMENTAL SECTION

Instrument

An electrospray-ionization atmospheric-pressure ion mobility spectrometer coupled to a quadrupole mass spectrometer through a 40-μm pinhole was used in this investigation (Fig. 1). The mass spectrometer was an ABB Extrel (Pittsburgh, PA, USA) 150-QC quadrupole (m/z 0-400), and was equipped with a Keithley amplifier (Model 427, Keithley Instruments, Cleveland, OH, USA) that amplified data from the electron multiplier detector and sent it to the acquisition systems. Merlin software (version 3.0, ABB Extrel) controlled the mass spectrometer and collected the mass spectral data.

Figure 1.

Sketch of the electrospray ionization-atmospheric pressure ion mobility-mass spectrometer.

The ion mobility spectrometer was built at Washington State University (Pullman WA, USA), and has been described in detail elsewhere.^[30] Briefly, the drift tube consisted of a desolvation and a drift region separated by a Bradbury-Nielsen ion gate. Both regions comprised 2.2”-OD, 2.0”-ID stainless steel rings, insulated from each other by alumina rings of the same size (99.6% Al₂O₃, Advalue Tech., Tucson, AZ, USA). A counterbore into each metallic ring supplied a support to hold the neighboring ceramic insulator. The insulating spacers and steel rings were horizontally stacked in an interlocking design. All rings were kept together in a 2.5”-OD, 2.3”-ID alumina tube placed in an aluminum heating case. Steel rings were connected in series by high temperature resistors. The resistors were 1-MΩ (drift region) or 0.5 MΩ (desolvation region) (Caddock Electronics Inc., Riverside, CA, USA, ±1%). When an electrical potential was applied to the first ring, an electric field of 432 V cm⁻¹ developed in the drift tube.^[31] A target screen was placed at the first ring of the desolvation region. This screen helped to electrospray the samples, and was made out of 2-mm stainless steel mesh with a 0.5-cm round hole in the center. A countercurrent of preheated N₂ buffer gas was introduced at the low voltage end of the drift tube through a stainless-steel tube. The buffer gas was heated by passing it through a 2-m long stainless-steel tube coiled inside an aluminum heating block (Fig. 1).

The gate had approximately 80 parallel 75-μm Alloy 46 wires (California Fine Wire Co., Grove Beach, CA, USA) separated 0.6 mm from each other. Ions were prevented from passing to the drift region by applying a closure potential that was 40 V higher for a set of wires and 40 V lower for the other set than the voltage of the gate when it was open. For these experiments the gate voltages were 10840 V and 10760 V. The open-gate voltage was that of a steel ring in the position of the gate in the tube, 10800 V. This closure voltage was disconnected for 0.1 ms so that a narrow pulse of ions entered the drift region. The mobility spectrometer was run at ambient pressure (685-710 Torr in Pullman, WA, USA). LabView software (National Instruments, Austin, TX, USA), modified in lab, collected the IMS data and controlled the ion gate. Igor Pro 5.0.3 (WaveMetrics, Portland, OR, USA) was used to process spectral data text files. The electronics for IMS data acquisition were built at Washington State University.^[32]

Typical operating parameters used with this instrument were: ESI flow rate, 3 μL min⁻¹; reaction region length, 7.5 cm; drift tube length, 25.0 cm; ESI voltage, 15.6 kV; voltage at first ring, 12.12 kV; voltage at the gate, 10.80 ± 0.01 kV; gate closure potential, ±40 V; gate pulse width, 0.1 ms; scan time, 35 ms; pressure, 680-710 Torr; buffer gas, nitrogen; buffer gas temperature, 150 ± 2 °C; buffer gas flow rate, 1 L min⁻¹; modifier flow rates, 1 to 50 μL hr⁻¹ (Table 1).

Table 1.

Ion mobility spectrometer operating conditions summary

Parameter	Settings
Reaction region length	7.5 cm
Drift region length	25.0 cm
ESI voltage	15.6 kV
Voltage at first ring	12.12 kV
ESI flow	3 μL min−1
Voltage at the gate	10.80 ± 0.01 kV
Gate closure potential	±40 V
Gate pulse width	0.1 ms
Scan time	35 ms
Buffer gas	Nitrogen
Buffer gas temperature	150 ± 2 °C
Buffer gas flow	930 mL min−1
Modifier flow rate	1 to 50 μL hr−1

Modes of operation

Ion mobility spectra were obtained in two modes. In IMS mode, the DC voltages to the mass spectrometer are removed and the quadrupole was operated in the RF-only condition, allowing all the ions to pass through the mass spectrometer to the ion detector. Total ion mobility spectra were obtained in this mode. In SIMIMS mode (single ion monitoring), the DC and RF voltages in the mass spectrometer are set to allow only ions of a specific mass or a selection of specific masses to reach the detector. Mass-selected ion mobility spectra were obtained in this mode. To obtain a complete mass spectrum of a sample, the gate of the ion mobility spectrometer was held open and all the ions migrated through the IMS region to the mass spectrometer without pulsing.

Materials and Reagents

2,4-Dimethylpyridine (2,4-lutidine), 2,6-di-tert-butyl pyridine (DTBP), 2-butanol, arginine, atenolol, desipramine, ethanolamine, ethyl lactate, histidine, lysine, nitrobenzene, serine, tetrahydrofuran-2-carbonitrile, the drug valinol, and tetrabutylammonium (TBA), tetraethylammonium (TEA), tetramethylammonium (TMA), and tetrapropylammonium (TPA) chlorides (ACS reagent grade, ≥98% purity) were purchased from Sigma Aldrich Chemical Co. (Milwaukee, WI, USA). Ethyl lactate, tetrahydrofuran-2-carbonitrile, 2-butanol, and nitrobenzene were selected as modifiers because they have different functionalities and produce different steric effects. Nitrobenzene has a low proton affinity and does not charge through proton transfer reactions with analytes. Therefore, nitrobenzene does not compete for charge and can serve as a comparison with other modifiers that might neutralize the analytes as they drift through the buffer gas. The analytes were chosen to have a range of sizes, functionalities, and steric effects.

Sample preparation and introduction

100 μM standard solutions of the analytes were prepared in ESI solution (47.5 % methanol: 47.5 % water: 5 % acetic acid). Liquid samples or blank solution (ESI solution) were continuously infused by electrospray ionization using 250-μL syringes (Hamilton, Reno, NV, USA) at a flow rate of 3 μL min⁻¹ into a 30-cm-long, 100-μM ID capillary (Polymicro Technologies, Phoenix, AZ, USA). This capillary was connected, through a stainless steel union (Valco, Houston, TX, USA) to a 50-μm ID silica capillary. The end of this capillary was centered at the target screen placed at the entrance of the mobility spectrometer. A voltage of 15.6 kV (or 3.5 kV bias with respect to the target screen) was applied to the stainless steel union to produce positive electrosprayed ions. To prevent cross contamination between analytes, different syringes and capillaries were used for every compound whenever possible.

Modifier introduction

The liquid modifiers were injected into the buffer gas line (pumped by a KD Scientific pump, model 210, Holliston, MA, USA) before the buffer gas heater using gas tight syringes to avoid leaking. Modifiers were introduced through a 10-cm-long, 50-μm ID silica capillary into the buffer gas line using a cross-junction (Fig. 1). The purpose of introducing the modifier before the buffer gas heater was to provide a longer path in order to obtain a homogeneous mixture of the modifier with the buffer gas. Heating tape (OMEGA Engineering, Stamford, CT, USA) was wrapped around the buffer gas tube and cross to help vaporize the modifier.

Calibration

To account for errors produced by inaccurate measurement of drift tube length, temperature, voltages, and pressure, the reduced mobilities used in this study were determined from the known reduced mobility values of standards^[33] using the relation

$$\frac{K_{o\left(\text{unknown}\right)}}{K_{o\left(s\text{tandard}\right)}}=\frac{t_{d\left(\text{standard}\right)}}{t_{d\left(\text{unknown}\right)}}$$ (3)

where K₀ is the reduced mobility in cm²V⁻¹s⁻¹ and t_d the drift time in ms. 2,4-Lutidine and DTBP were used as chemical standards to calibrate the reduced mobilities of the analytes used in this study.

Purity of buffer gas

Because these experiments investigated the modification of the buffer gas it was essential to ensure that the buffer gas was pure prior to modification. The purity of the buffer gas was evaluated using two IMS standards, DTBP and 2,4-lutidine (24L). DTBP is insensitive to buffer gas contamination while 24L is known to be sensitive to buffer gas contamination. The reduced mobility value was calculated for 24L using DTBP as the known standard in equation 3 above. If the reduced mobility calculated for 24L matched its literature value, the instrument was considered to be free of contamination. If the reduced mobility calculated for 24L did not match its literature value, the instrument was cleaned and the test for buffer gas purity was repeated until the buffer gas was free of contamination.^[34]

Identification of compounds

All analytes were detected in the mass spectrometer as MH⁺ or cluster ions. Analytes were identified in the mass spectra by their molecular weights determined from these protonated molecules or clusters. Analyte peaks in the mobility spectrum were identified using SIMIMS and by comparing their reduced mobilities with literature values.

RESULTS AND DISCUSSIONS

1. Ethyl lactate as a buffer gas modifier

1a. Effect of ethyl lactate modifier on the ion mobility of ethanolamine

Figure 2 illustrates the reduction in the ion mobility of ethanolamine when ethyl lactate was introduced as the modifier into the buffer gas of the ion mobility spectrometer. The drift time of ethanolamine ions increased more than 10 ms as the ethyl lactate concentration increased from 0.0 to 2.8 mmol m⁻³ (104 ppmv), which corresponded to a 41% reduction in mobility. Figure 2 shows clusters of ethanolamine (E) ions with ethyl lactate (L), sodium ions, and reactant ions in the buffer gas in the mobility and mass spectra; sodium ions are ubiquitous, arising as trace contaminants from the electrospray solvent. LNa⁺ ions were formed because ethyl lactate was much more concentrated in the buffer gas than analytes or reactant ions (more than 100 times); analyte-Na⁺ adducts were not visible.

Figure 2.

Reduction in analyte's mobility with the introduction of ethyl lactate modifier in the buffer gas.

Figures 2a and 2b are the mobility and mass spectra, respectively, when no modifier was added to the nitrogen buffer gas. Three mobility peaks are seen in Fig. 2a; the first and last peaks in the mobility spectrum (Fig. 2a) were those of the solvent background ion while the middle peak was protonated ethanolamine (m/z 62). As expected, 1) the intensity of EL₂H⁺ increased with respect to ELH⁺ as the concentration of ethyl lactate in the buffer gas increased, and 2) the EL₃H⁺ and L₃Na⁺ ions only appeared at high concentrations

All the reactant ions disappeared from the spectra when the ethyl lactate modifier was added to the buffer gas; ethyl lactate was protonated and also clustered to ethanolamine and sodium to form the species L_nH⁺, EL_nH⁺, L_nH₃O⁺, EL_nH₃O⁺ and L_nNa⁺. The L_nNa⁺ species was seen in all the spectra where modifier was added to the buffer gas.

When ethyl lactate was added as the modifier, clusters of the ethanolamine were seen in the mass spectrum and the analyte peak shifted to longer drift times. Clusters with ethanolamine occurred at 24.5 ms, EL_nH⁺ and 26.2 ms EL_nH₃O⁺ in the mobility spectrum (Fig. 2c). These clusters also appear in Figs 2e, 2g, and 2i at longer drift times when the concentration of ethyl lactate increased in the buffer gas. Clusters of ethanolamine ions with one to three ethyl lactate molecules occurred at m/z 180 and 298 (Figs 2d, 2f, 2h, and 2j), and m/z 416 (Figs 2h and 2j) in the mass spectra. The mass spectra also show that the ratio EL₂H⁺ to ELH⁺ increased with the concentration of ethyl lactate in the buffer gas and that EL₃H⁺ only appeared at high concentrations of modifier, indicating that there was increasing clustering as a function of increased ethyl lactate concentration.

A maximum of three molecules of ethyl lactate can cluster to ethanolamine ions; this number corresponds to the number of hydrogen atoms attached to the positively charged nitrogen of protonated ethanolamine, C₂H₄ONH₃⁺ , available for binding modifier molecules. These results agree with the findings of Bollan et al.^[20] who reported that the number of ketone molecules binding to hydrazines and ammonia analytes when the buffer gas was modified depended on the number of hydrogen atoms on the protonated nitrogen of the analytes (four for ammonia, three for hydrazine, two for monomethyl hydrazine, and one for 1,1-dimethylhydrazine).

1b. Effect of ethyl lactate modifier on the ion mobility of serine

Figures 3a and 3b show the background ions when only ethyl lactate was added to the buffer gas. Under these conditions the background ions were independent of the electrospray solvent and were controlled by the excess of modifier added to the buffer gas. The presence of ethyl lactate depleted all reactant ions by clustering with H⁺ and H₃O⁺. In addition, the sodium ion cluster of ethyl lactate was observed. Three background ion species were identified: L_nH⁺, L_nH₃O⁺, and L_nNa⁺ with drift times of 25.0 ms, 25.5 ms, and 29.1 ms, respectively.

Figure 3.

Ethyl lactate clustering with ESI solvent and serine ions. (a) and (c) IMS and (b) and (d) mass spectra of ESI solvent and serine (S) when ethyl lactate (L) modifier was introduced into the buffer gas at a concentration of 1.7 mmol m⁻³ in (a) and (b) or 0.33 mmol m⁻³ in (c) and (d). (a) and (b) ESI solvent produced extensive clustering due to the relatively small size and low proton affinity of reactant ions. (c) and (d) Serine formed abundant clusters as a consequence of its lack of steric hindrance and small size. SH⁺ and SL_nH⁺ had the same drift times and appeared as a single peak in the mobility spectrum. Reactant ions were completely stripped of their charge by the modifiers and disappeared from the spectra.

Figures 2c, 2e, 2g, and 2i show a tail between the two peaks at ~25 ms, which means that the ion at 26.2 ms, EL_nH₃O⁺, was converting into that at 24.5 ms, EL_nH⁺, during the drift time. The tail can be used to calculate rate constants of ion-molecule or decomposition reactions.^[35] However, EL_n.H⁺ and EL_n.H₃O⁺ were expected to overlap. In IMS, there should be a fast forward and backward reaction between water molecules and cluster ions in the drift region:

$${\mathrm{ELH}}^{+}+{\mathrm{H}}_2\mathrm{O}\leftrightarrow {\mathrm{ELH}}_3{\mathrm{O}}^{+}$$

Nevertheless, the reactant ions, all protonated analytes, and Na⁺ ions were “sequestered” by ethyl lactate due to its high concentration and their clusters were available mostly for equilibration with the free modifier, which separated EL_nH⁺ and EL_nH₃O⁺ clusters. Figures 2 and 5 show how ethyl lactate depleted all the reactant ions by clustering with H⁺ and H₃O⁺ and probably with water, sequestering it, making water unavailable for the equilibrium depicted above. Ethyl lactate also separated the pair of ions, L₂H⁺ and L₂H₃O⁺, from the pair, L₃H⁺ and L₂H₃O⁺ (Fig. 4), which were also expected to overlap. The identity of these clusters in the IMS spectra was confirmed in SIMIMS mode. That the strongest reactant ion peak clusters were not LH⁺, LH₃O⁺, or LNa⁺ but L₂H⁺, L₂H₃⁺, or L₂Na⁺ confirmed this sequestration. No clusters with two or more water molecules were observed in the mass spectra here or in earlier work when modifiers were introduced into the buffer gas,^[34,36] which also confirmed the sequestration of water. On the contrary, clusters of analyte with up to eleven molecules of water were seen when water was introduced as a modifier.^[34]

Figure 5.

Clustering in DTBP, 2,4-lutidine, and tetraalkylammonium ions. (a) and (c) IMS and (b) and (d) mass spectra of 2,4-lutidine, DTBP, and tetraalkylammonium ions when 1.1 mmol m⁻³ (42 ppmv) of ethyl lactate (L) modifier were introduced into the buffer gas. (a) and (b) 2,4-lutidine produced a large MS cluster with the modifier and a small protonated molecule; In contrast, DTBP ions did not show clustering with ethyl lactate in the mass spectrum; these features agree with the fact that the ion mobility of 2,4-lutidine was more affected by the modifier than that of DTBP ions and that from previous work with 2-butanol.^[37] (c) and (d) Tetraalkylammonium ions did not cluster due to steric hindrance; TBA ions and L_nNa⁺ overlapped in the mobility spectrum.

Figure 4.

Changes in mobilities of selected ions with modifiers in the buffer gas. (a) and (c) The mobilities of tetraalkylammonium and arginine ions did not change, and for DTBP and atenolol ions changed minimally with modifiers in the buffer gas due to steric hindrance, which impeded the attachment of modifier molecules to the positive charge of these ions. Analytes ions with lower steric hindrance such as 2,4-lutidine, desipramine, ethanolamine, serine, and valinol showed larger changes in mobility. (b) Changes in mobilities of ethyl lactate cluster ions with increasing concentrations of ethyl lactate. The mobilities of both cluster ions coincided within families: protonated, hydrated, and sodiated clusters had the same mobility.

Clustering with ethyl lactate was evident also in the mobility and mass spectra of the amino acid serine (Figs 3c and 3d). Clusters of ethyl lactate with the reactant ions occurred at 25.0 ms, L_nH⁺, 25.5 ms, L_nH₃O⁺, and 29.1 ms, L_nNa⁺ , in the mobility spectrum (Fig. 3a); in the mass spectrum (Fig. 3b), the clusters occurred at m/z 119 LH⁺, 136 LH₃O⁺, 237 L₂H⁺, 255 L₂H₃O⁺, 259 L₂Na⁺, 355 L₃H₃O⁺, 373 L₃H₃O⁺, and 377 L₃Na⁺ . Most of these ions in the IMS and MS spectra also appear in Fig. 2 but clustered to ethanolamine. Ethyl lactate was present mainly as LH⁺ and LH₃O⁺ based on the peak intensities shown in Fig. 3a. The mechanism of formation of these ions was clustering with the reactant ions followed by stripping of the charge. Reactant ions were completely stripped off their charge by the modifiers and disappeared from the spectra. LNa⁺ was formed but not analyte-Na⁺ clusters, because ethyl lactate was present at a much higher concentration in the buffer gas than the analytes. Serine-ethyl lactate clusters, SL_nH⁺, occurred at 28.1 ms in the mobility spectrum (Fig. 3c) and at m/z 224 SLH⁺ and 342 SL₂H⁺ in the mass spectrum (Fig. 3d). SL_nH⁺ ions were formed in the reaction region and not at the atmospheric/vacuum interface of the IMS system with the mass spectrometer because the drift times of serine ions increased with the concentration of ethyl lactate in the buffer gas, indicating the formation of larger serine ionethyl lactate clusters, SL_nH⁺, as a function of increasing modifier concentration.

SH⁺ and SL_nH⁺ had the same drift times (Fig. 3c), appearing as a single peak in the mobility spectrum (Fig. 3c) and indicating that these ion species were in equilibrium and the charged species migrated through the drift tube. Based on these results, the following equilibria (Scheme) are proposed to occur in the drift tube:

Scheme.

Clustgering equilibribia

These reactions must be rapid during the passage of ions through the drift tube. Thus, the drift time of serine ions was a weighted average of the drift times of its fast protonated molecule and the slow cluster ions. Similar arguments can be used to demonstrate that peaks in the mass spectra of other analytes corresponded to cluster peaks.

In summary, when ethyl lactate was used as a modifier for nitrogen buffer gas, the ion mobility of serine decreased due to the formation of serine ion-ethyl lactate clusters. The maximum cluster size was limited by the number of hydrogen atoms attached to the amine nitrogen. The ion – ethyl lactate exchange rate was sufficiently rapid that the ion migrated as an average ion cluster through the buffer gas.

1.c Effect of ethyl lactate on the mobility of other ions

Changes in mobilities with the introduction of ethyl lactate in the buffer gas are shown in Figs 4a and 4b. These changes are summarized as %ΔK₀ values, where %ΔK₀ was defined as the percentage difference between K₀ in N₂-only buffer gas and K₀ when a modifier was introduced into the buffer gas at a given concentration. The test ions showed the following %ΔK₀ values: 2,4-lutidine (24%), atenolol (7.0%), desipramine (12%), DTBP (1.5%), ethanolamine (41%), serine (30%), TBA (0.3%), TEA (0.5%), TMA (1.0%), TPA (0.1%), and valinol (28%) when the ethyl lactate concentration increased in the buffer gas from 0.0 to 1.7 mmol m⁻³, and L_nH⁺ (6.4%), L_nH₃O⁺ (10%), and L_nNa⁺ (4.3%) as ethyl lactate increased in the buffer gas from 0.33 (12 ppmv) to 2.8 mmol m⁻³ (1.0×10² ppmv). All ion mobilities decreased (average %ΔK₀ value of 13%, excluding the cluster ions) when ethyl lactate was introduced into the buffer gas at 1.7 mmol m⁻³; tetraalkylammonium ions and DTBP showed only reduced %ΔK₀ values (≤1.5%) at this ethyl lactate concentration.

2.Effect of other modifiers on ion mobility

In general, the formation of ion clusters decreased ion mobilities by increasing the collision cross sections of the ions. Table 2 summarizes the changes in mobilities of ions of 2,4-lutidine, arginine, DTBP, ethanolamine, serine, valinol, the drugs atenolol and desipramine, and the TBA, TEA, TMA, and TPA ions when modifiers were introduced into the buffer gas. In these experiments, 2-butanol, ethyl lactate, nitrobenzene, and tetrahydrofuran-2-carbonitrile were used as modifiers. Ethyl lactate, nitrobenzene, and tetrahydrofuran-2-carbonitrile had not previously been tested as buffer gas modifiers, and 2-butanol had not been used as a modifier with basic amino acids.

Table 2.

Percentage decrease in K₀ values, %ΔK₀, for selected ions when modifiers were introduced into the buffer gas. The concentrations of modifier increased in the buffer gas from 0 to 1.7, 0.95, and 6.8 mmol m⁻³ for ethyl lactate, nitrobenzene, and 2-butanol, respectively; for ethyl lactate clusters, ethyl lactate increased in the buffer gas from 0.33 (12 ppmv) to 1.7 mmol m⁻³. Differences of less than 0.32 units in %ΔK₀ may arise from the maximum accepted standard deviation of the drift time measurements (0.05 ms).

	Modifiers
Ions	Ethyl lactate	Nitrobenzene	2-butanol
2,4-Lutidine.H+	24	13	5.6b
LnH+	6.4
LnH3O+	10
LnNa+	4.3
Arginine.H+	9.8	2.7	0.3
Atenolol.H+	7.0	2.0	0.7a
Desipramine.H+	12	3.0	1.1
DTBP.H+	1.5	0.1	2.2b
Ethanolamine.H+	41	37	19
Histidine.H+	4.1	2.3	1.1
Lysine.H+	4.3	2.1	1.1
Serine.H+	30	25	14b
TBA	0.3	0.1	0.3b
TEA	0.5	0.3	1.0b
TMA	1.0	1.0	1.0b
TPA	0.1	0.5	0.0b
Valinol.H+	28	21	9.8b

^aAverage %ΔK₀ at 100, 150, 200, and 250°C.

^bReference 36. Differences of less than 0.32 units in %ΔK₀ may arise from the standard deviation of the drift time measurements (0.05 ms).

2.a. 2-butanol as a buffer gas modifier

When the 2-butanol concentration increased in the buffer gas from 0.0 to 6.8 mmol m⁻³ (2.5×10² ppmv), the %ΔK₀ values for diamines ions were: atenolol (0.7%), arginine (0.3%), and histidine and lysine (1.1%) (Table 2). In a previous work,^[36] the mobilities of a series of analytes decreased (average %ΔK₀ value of 10%) when 2-butanol was introduced into the buffer gas at the same experimental conditions as used here. In this work, only diamines were tested with 2-butanol, and they showed a low responsiveness to the introduction of this modifier into the buffer gas (average %ΔK₀ value of 0.8%).

2.b. Nitrobenzene as a buffer gas modifier

The changes in mobilities with the introduction of nitrobenzene in the buffer gas are shown in Fig. 4c. Ions of the test compounds produced the following %ΔK₀ values when the nitrobenzene concentration increased in the buffer gas from 0.0 to 1.0 mmol m⁻³ (36 ppmv): 2,4-lutidine (13%), arginine (2.7%), atenolol (2.0%), DTBP (0.1%), ethanolamine (37%), TBA (0.1%), TEA (0.3%), TMA (1.0%), TPA (0.5%), and valinol (21%) (Table 2). In general, decreases in ion mobility were observed when 1.0 mmol m⁻³ nitrobenzene were introduced into the buffer gas (average %ΔK₀ value of 7.8%). Only small reductions in ion mobility were observed for diamines, DTBP, and tetraalkylammonium ions (%ΔK₀ values ≤ 2.7%).

2.c. Tetrahydrofuran-2-carbonitrile as a buffer gas modifier

When tetrahydrofuran-2-carbonitrile (tHFCN) was used as a modifier, ions were not observed in the mass spectrum for α-amino acids and valinol at concentrations lower than 3.3 mmol m⁻³ (125 ppmv) of this modifier in the buffer gas. However, tetraalkylammonium and DTBP ions were observed at this concentration. The absence of some analyte peaks from the mass spectrum when tHFCN is used as a modifier can be attributed to charge stripping. As the protonated analyte molecules drift through the buffer gas, they behave as reactant ions and protonate the neutral modifier. The proton affinity of tHFCN stripped off the charge of the analyte ions except for those from DTBP, where steric hindrance deterred the approach of the modifier to the positive charge (as explained in section 4), and those inherently ionic species such as tetraalkylammonium ions (data not shown).

3. Effect s of steric hindrance, modifier saturation, and analyte size on changes in ion mobility due to clustering with modifiers

In Fig. 4, several common characteristics of changes in mobility are observed when modifiers were introduced into the buffer gas. The ion mobilities of the two pyridines changed to different extents when the modifiers were introduced into the buffer gas. The mobilities of DTBP ions decreased less than those of 2,4-lutidine ions (average %ΔK₀ value of 0.8% and 19%, respectively) with ethyl lactate and nitrobenzene modifiers. Figures 5a and 5b show the mobility and mass spectra, respectively, of a DTBP/2,4-lutidine mixture when the buffer gas was spiked with 1.1 mmol m⁻³ of ethyl lactate. No cluster peaks of DTBP ions were visible, but a 1:1 2,4-lutidine ion-ethyl lactate cluster occurred at 22 ms in the mobility spectrum and m/z 226 in the mass spectrum. This lack of clustering of DTBP ions was due to the strong steric hindrance exerted on the charge by the large tert-butyl substituents, in positions 2 and 6 on the ring. These substituents shielded the protonated nitrogen atom of DTBP from modifier molecules. The two small methyl groups of 2,4-lutidine, located at positions 2 and 4 on the ring, shielded the nitrogen atom less effectively. This propensity of DTBP ion to resist cluster formation explains its small mobility change with modifiers in the buffer gas.

Figure 4 also shows that the mobilities of tetraalkylammonium ions decreased by ≤1% when the buffer gas was spiked with ethyl lactate or nitrobenzene. When 1.1 mmol m⁻³ of ethyl lactate was introduced into the buffer gas (Fig. 5c), tetraalkylammonium ions showed single and sharp peaks (i.e. they did not produce clusters) occurring at 16.7 (TMA), 20.3 (TEA), 24.8 (TPA), and 29.4 ms (TBA) in the mobility spectrum and at m/z 75.2 (TMA), 131.2 (TEA), 186.4 (TPA), and 243.5 (TBA) in the mass spectrum (Fig. 5d). This lack of clustering was a consequence of the steric hindrance exerted by the four alkyl substituents in tetraalkylammonium ions that deterred the attachment of ethyl lactate molecules to the nitrogen positive charge and delocalized this charge, thus weakening the ion-modifier interactions.^[15,37] This lack of clustering of tetraalkylammonium ions explains their small mobility change with modifiers in the buffer gas.

Figure 4 shows that the largest reductions in mobility were obtained with the first introduction of modifier, and only small reductions were achieved with further increments of modifier concentration, which caused a plateau in the K₀ values at high concentrations of modifier. In Figure 2a, the largest increase in the drift time of ethanolamine ions (60%) was obtained when the ethyl lactate concentration increased from 0.0 to 0.33 mmol m⁻³ (12 ppmv), and only a small reduction (8.4%) was obtained for a larger increment of concentration (from 0.33 mmol m⁻³ to 1.7 mmol m⁻³). Small reductions in mobility with the introduction of ethyl lactate into the buffer gas may indicate modifier saturation of the hydrogen atoms available for ligand binding on the positive charge in ethanolamine ions, due to overloading of the buffer gas with ethyl lactate. This ligand saturation deterred the attachment of additional molecules of ethyl lactate, and no further decrease in the mobility of ethanolamine ions was obtained when the concentrations of ethyl lactate increased.

Table 2 shows that there was a tendency towards large %ΔK₀ values when the mass of the analyte decreased. For analytes with relatively small steric hindrance on the positive charge, such as desipramine, serine, valinol, 2,4-lutidine, and ethanolamine ions, the mass to %ΔK₀ correlation coefficient was −0.91 when the ethyl lactate concentration increased from 0.0 to 1.7 mmol m⁻³ in the buffer gas. This inverse relationship between mass and %ΔK₀ originated in the large increase of size with the attachment of modifier molecules to small analytes. Increased clustering was also expected as the size decreased, as a consequence of localization of the positive charge.^[15,37] Steric hindrance, of course, would obscure this mass %ΔK₀ trend by deterring clustering in spite of the size of the analyte ions as found for TMA ions, which showed %ΔK₀ values of only 1%, which were small compared with the values of 41 and 37% for ethanolamine ionsof similar mass.

In summary, ion mobilities decreased, (average %ΔK₀ value of 9.1% for all the modifiers), due to the formation of analyte-modifier clusters. The mobilities of the DTBP and tetraalkylammonium ions showed only small changes (average %ΔK₀ value of 0.54% with ethyl lactate and nitrobenzene modifiers) as a consequence of steric hindrance and charge delocalization, which hindered clustering. In addition, ligand saturation of binding sites limited the decrease in mobility as a function of modifier concentration in the buffer gas. Finally, changes in ion mobility due to clustering were large for low mass ions.

4. Effect of intramolecular bridges on ion mobility

From Fig. 4 it can be seen that the mobility of the atenolol ion was only slightly affected by the addition of a modifier to the buffer gas. The atenolol ion was essentially unaffected by the addition of 2-butanol to the buffer gas. Table 2 shows that the mobility was only slightly reduced on the addition of nitrobenzene. While the tetraalkylammonium ions, discussed above, were essentially unaffected by the introduction of a modifier, the reduction in the mobility of the atenolol ion was much less than that of many of the other analytes investigated. This is most clearly demonstrated in Fig. 6 using a mixture of desipramine and atenolol.

Figure 6.

Clustering of desipramine and atenolol ions and intramolecular bridges. The drift time separation between desipramine and atenolol ions increased when the ethyl lactate concentration in the buffer gas was increased from (a) 0, to (b) 0.33, and (c) 1.7 mmol m⁻³. (d) and (e) MS spectrum of desipramine (Des) and atenolol (Ate) at 150 °C when 0.66 mmol m⁻³ of ethyl lactate (L) were introduced into the buffer gas. (f) Structure and (g) 3D model of the formation of the bridge in atenolol: the arrow signals the hydrogen bridge. The steric hindrance and charge delocalization caused by the bridge deterred clustering to ethyl lactate molecules. (h) 3D model of desipramine; an arrow shows the nitrogen between the aromatic rings which sterically hindered the formation of the bridge with the nitrogen in the alkyl chain. The models were generated using the program Chem3D Pro 11.0 (PerkinElmer Inc. Cambridge, MA, USA).

These drugs have the same nominal molecular weight (266 g mol⁻¹), and, therefore, their mobilities are expected to be affected to the same extent by clustering. Desipramine is an antidepressant and atenolol is a β-blocker drug used to prevent angina and to reduce the risk of heart attacks by reducing heart rate and high blood pressure. The IMS spectra of the mixture of these drugs show a small separation of the peaks in N₂-only buffer gas (Δt_d = 0.8 ms, α = 1.03) in Figure 6a; when 0.33 mmol m⁻³ of ethyl lactate was introduced into the buffer gas, the separation increased (Δt_d = 1.6 ms, α = 1.05, Fig. 6b) and when the ethyl lactate concentration increased to 1.7 mmol m⁻³, the separation increased even more (Δt_d = 3.2 ms, α = 1.11, Fig. 6c).

Figures 6d and 6e provide an explanation of why the mobility of atenolol ions was less affected than that of desipramine ions after the introduction of ethyl lactate into the buffer gas. In these Figures, the mass spectra of solutions of desipramine and atenolol with 0.66 mmol m⁻³ of ethyl lactate modifier in the buffer gas show that:

• The intensity of protonated desipramine (m/z 267) was lower than that of its cluster with ethyl lactate (m/z 385); for atenolol, however, the peak intensity of the protonated molecule (m/z 267) was stronger than that of the cluster with ethyl lactate (m/z 385)

• A cluster of desipramine with two ethyl lactate molecules occurred at m/z 503, DL₂.H⁺ but not the cluster of atenolol with two ethyl lactate molecules, AL₂.H⁺ at m/z 503.

These two facts indicate a lower clustering of ethyl lactate with atenolol ions than with desipramine ions.

The reduced interaction of atenolol ions with the modifiers may be related to the formation of an intramolecular proton bridge between its two amine functionalities as illustrated in Figs 6f and 6g. The access of ethyl lactate to the positive charge of atenolol would be restricted by the ring on one side and the propyl group on the opposite side when this bridge was formed. Desipramine ions clustered more than atenolol ions because the formation of the intramolecular bridge in desipramine was sterically hindered due to the position of one of the amine moieties between large aromatic rings (Fig. 6h). This steric hindrance to the formation of the bridge would cause desipramine ions to adopt an open structure, prone to more collisions and clustering, which explains the larger ion mobility change observed for this ion than for the atenolol ion with ethyl lactate in the buffer gas. Steric hindrance to the formation of the proton bridge, similar to that of desipramine ions, was observed in the ions of the diamine tryptophan, which also has an amine moiety as part of a ring. The tryptophan ion mobility also was affected by modifiers in the buffer gas.^[36]

Evidence for the ability of atenolol to form an intramolecular proton bridge is that the mobility of the atenolol ion in the pure nitrogen drift gas was higher than that of the desipramine ion although their masses are the same. The formation of an intramolecular proton bridge would cause the ion structure to be more compact. The possibility of the formation of intramolecular proton bridges in diamines was reported by Karpas who found that the mobilities of α,ω-diamines were higher than those of normal monoamines as a consequence of a cyclization reaction between the two amine moieties.^[30]

Additional evidence of the formation of intramolecular proton bridges in diamines was obtained by studying basic amino acids. The mobilities of arginine, and histidine and lysine ions only decreased by by 0.3 and 1.1%, respectively, when the 2-butanol concentration increased from 0.0 to 6.8 mmol m⁻³ (2.5×10² ppmv) in the buffer gas (Table 2). These changes in the %ΔK₀ values of ions of basic amino acids were small compared with the average value of 10.2% for ions of non-basic amino acids such as methionine, serine, tryptophan, and tyrosine in similar experimental conditions.^[36] In addition, the mobility of arginine ions only decreased by 2.7% with 1.0 mmol m⁻³ of nitrobenzene in the buffer gas, which was small compared with the %ΔK₀ values of ethanolamine (37%) and valinol ions (21%) under the same conditions. Figure 7 shows the mass spectra of arginine, histidine, and lysine when 6.8 mmol m⁻³ (2.5×10² ppmv) of 2-butanol (B) were introduced into the buffer gas. These basic amino acids produced large MH⁺ ion peaks (HisH⁺ m/z 156, LysH⁺ 147, and ArgH⁺ 175) and only small analyte-2-butanol cluster peaks (HisBH⁺ at m/z 230, LysBH⁺ at m/z 221, which was overlapped with the modifier trimer at m/z 223, and ArgBH⁺ at m/z 249). The limited mobility shift of basic amino acids to the introduction of modifiers into the buffer gas may be related to this low clustering capacity. Figure 7d compares the high clustering of phenylalanine (Phe) with the low clustering of basic amino acids with 2-butanol (B) modifier in the mass spectrum. The PheBH⁺ and PheB₂H⁺ clusters occurred at m/z 240 and 314. This lack of ion-modifier clustering of basic amino acids may be due to the formation of an intramolecular proton bridge between the two amine functionalities, deterring attachment of modifier molecules to the positive charge of basic amino acids by steric hindrance and charge delocalization.

Figure 7.

Low clustering in basic amino acids. (a) Mass spectra of arginine, (b) histidine, and (c) lysine with 6.8 mmol m⁻³ (2.5×10² ppmv) of 2-butanol (B) in the buffer gas. Basic amino acids showed large MH⁺peaks (ArgH⁺ m/z 175, HisH⁺ m/z 156, and LysH⁺ m/z 147) and only small analyte-modifier peaks (HisBH⁺ m/z 230, LysBH⁺ m/z 221, which was overlapped with the modifier trimer at m/z 223, and ArgBH⁺ m/z 249) due to steric hindrance produced by an intramolecular bridge. (d) Mass spectrum of a 100-μM solution of phenylalanine (Phe) at the same 2-butanol concentration as in (a) to (c); cluster peaks of 2-butanol were evident at m/z 149 (dimer), 223 (trimer), and 297 (tetramer), and the clusters of phenylalanine with 2-butanol occurred at m/z 240 and 314.

5. Resolution of IMS overlapping ions by selective clustering

Application of drift gas modifiers to effect resolution in IMS was demonstrated using differences in intramolecular proton bridging of diamines and monoamines (Fig. 8). Figure 8 demonstrates the use of a buffer gas modifier, 2-butanol, to effect the base line resolution of partially overlapping ion peaks. Figure 8a shows the IMS spectra of a mixture of arginine (22.0 ms) and phenylalanine (22.5 ms), which partially overlap in N₂-only buffer gas. Figure 8b demonstrates that the two ions can be base line resolved by introducing 2-butanol as a modifier into the nitrogen buffer gas at a concentration of 6.8 mmol m⁻³ (250 ppm v/v). In this case the addition of the modifier did not affect the mobility of the arginine ion but did reduce the mobility of the phenylalanine ion such that it eluted 3.4 ms after the arginine ion, increasing the resolution of these two ions from about 1.0 to about 8.5. This dramatic difference in mobility as a function of modifier addition was attributed to the ability of the arginine ion to form intramolecular proton bridges between the two amine groups while the amine protons of the phenylalanine ion could interact with the modifier to produce phenylanlanine.H⁺(2butanol)_n ion clusters. Mass spectral evidence for the formation of these clusters is given in Fig. 7.

Figure 8.

IMS separations by introducing 2-butanol into the buffer gas. (a) IMS spectra (average of 1600) of a 100-μM mixture of arginine (Arg) and phenylalanine (Phe) in N₂-only buffer gas and (b) when 6.8 mmol m⁻³ of 2-butanol (B) were injected into the buffer gas. (c) IMS spectra, (average of 1600) of a 50-μM mixture of alanine (Ala), serine (Ser), threonine (Thr), isoleucine (Ile), and lysine (Lys) in N₂-only buffer gas. The inset demonstrates the separation of a mixture of lysine and isoleucine ions with 6.8 mmol m⁻³ of 2-butanol (B) in the buffer gas. 2-butanol made possible the baseline resolution of these two pairs of ions by forming IleB and PheB clusters (shown in Fig. 7d for Phe), which reduced the mobilities of Ile and Phe. The steric hindrance in arginine and lysine, caused by an intramolecular bridge, deterred extensive formation of clusters, and these amino acid mobilities were not affected by the modifier.

A second example of separation is shown in Fig. 8c where the mobility of the isoleucine ion is reduced as a function of modifier addition to the nitrogen buffer gas. The full IMS spectrum in Fig. 8c shows the IMS separation of a mixture of the amino acids alanine, serine, threonine, isoleucine, and lysine. In this spectrum the isoleucine ion elutes before the lysine ion but is not completely separated from it. The insert in Fig. 8c shows the baseline resolution of lysine and isoleucine ions after the addition of 6.8 mmol m⁻³ of 2-butanol to the nitrogen buffer gas. With the addition of 2-butanol the drift time of the isoleucine ion increased by 1.2 ms but the drift time of lysine ion increased only by 0.2 ms, producing the separation of these two amino acids with a resolution of about 2.0. As with the arginine ion, in the example above, lysine has the ability to form intramolecular proton bridges which apparently reduces its ability to cluster with 2-butanol while the isoleucine ion forms 2-butanol clusters that increase its size and reduce its mobility. Note: The separation of these amino acids could not be obtained at 250 °C (data not shown), perhaps because the decreased isoleucine-2-butanol interactions at high temperature did not allow clustering of isoleucine.

CONCLUSIONS

Formation of ion-modifier clusters was found when liquid modifiers such as ethyl lactate, nitrobenzene, tetrahydrofuran-2-carbonitrile, and 2-butanol were vaporized into the buffer gas of an ion mobility spectrometer. Analytes clustered to different extents with a modifier, depending on their size and the formation of intramolecular bridges. Intramolecular bridges limited clustering in diamines by hindering the attachment of modifier molecules to the positive charge of those analytes, for which diamines only experienced small changes in ion mobilities with the addition of modifiers into the buffer gas. Steric hindrance caused by bulky substituents and large size also limited changes in ion mobility with modifiers in the buffer gas. This difference in clustering slowed down the ions with abundant clustering and only slightly affected the mobility of ions with limited clustering. These different clustering behaviors were applied to separation of mixtures of compounds that overlapped in IMS by selectively changing their ion mobilities. Modifiers also produced different changes in mobilities: ethyl lactate had a larger effect than nitrobenzene or 2-butanol on ion mobilities due to larger formation of clusters. Finally, tetrahydrofuran-2-carbonitrile stripped off the charge of the analytes, except for those with steric hindrance that deterred clustering, for which it was considered to be inappropriate as a modifier for shifting drift times of ions. This selective charge stripping suggests a possible use of this modifier as a means of simplifying sthe pectra of complex mixtures of analytes.