Chemical ionization of glyoxal and formaldehyde with H3O+ ions using SIFT-MS under variable system humidity

Glyoxal (C2H2O2) is a highly reactive molecule present at trace levels in specific gaseous environments. For analyses by chemical ionization mass spectrometry, it is important to understand the gas-phase chemistry initiated by reactions of H3O + ions with C2H2O2 molecules in the presence of water vapour. This chemistry was studied at variable humidity using a selected ion flow tube, SIFT. The initial step is a proton transfer reaction forming protonated glyoxal C2H3O2 . The second step, in the presence of water vapour, is the association forming C2H3O2 (H2O) and interestingly also protonated formaldehyde CH2OH . Hydrated protonated formaldehyde CH2OH (H2O) was also observed. Relative signals of these four ionic products were studied at the end of the flow tube where the reactions took place during 0.3 ms in helium carrier gas (1.5 mbar, 300 K) as the water vapour number density varied up to 10 cm . The data were interpreted using numerical kinetics modelling of the reaction sequences and the mechanisms and kinetics of the reaction steps were characterised. The results thus facilitate SIFT-MS analyses of glyoxal in humid air whilst drawing attention to ion overlaps with formaldehyde products.


Introduction
Glyoxal (C 2 H 2 O 2 ) is the simplest dialdehyde. It can polymerize from an anhydrous monomeric form into a series of oligomers and its vapour is highly reactive, contributing to aerosol formation. 1,2 Glyoxal plays a role in biological processes; it is known to be a cytotoxic metabolite of glucose, a product of lipid peroxidation and a marker of oxidation stress. 3 In addition, recent studies have shown that glyoxal is an intermediate product in the photocatalytic reduction of carbon dioxide to methane. 4,5 It is noteworthy that organic molecules present in the Earth's atmosphere can be transformed into glyoxal by oxidation or photo-oxidation. 6 Such processes were observed for toluene, 7-9 xylene, [8][9][10] trimethylbenzene, 9,10 isoprene 11 and a-pinene. 12 Daytime atmospheric glyoxal chemistry involves oxidation and may lead to the formation of secondary organic aerosols. 13 Chemical ionization mass spectrometry (CIMS) is a useful tool for the detection of glyoxal since the high reactivity and short lifetime of glyoxal impede the use of gas chromatography (GC) techniques. Nölscher et al. measured the total OH reactivity of tropical forest air and compared it to the reactivity expected from volatile organic compounds (VOCs) determined via protontransfer-reaction mass spectrometry (PTR-MS). 14 The measured OH reactivity exceeded the calculated values, indicating that either PTR-MS underestimated VOC concentrations or that not all relevant species were accounted for. This can also be caused by the different OH reactivity of isomers (which cannot be separated by PTR-MS) or the presence of molecules with high reactivity and low proton affinity (PA) compared to the PA of water (protonated water, H 3 O + , is the most common reagent ion in PTR-MS).
The PA of glyoxal was previously estimated by HF and DFT calculations to be 675 kJ mol À1 and 690 kJ mol À1 , respectively. 15 Since the PA is very similar to the PA of water (691 kJ mol À1 ), 16 proton transfer from hydronium (H 3 O + ) ions may not proceed at a collisional rate. 17 In addition, the reverse reaction could also take place: CIMS analysis of atmospheric glyoxal can be further hampered by isobaric molecules, e.g. acetone or propanal. These are typically present at much higher concentrations, causing signal overlaps. Understanding the ion chemistry may help to distinguish these isobars.
Michel et al. 18 previously studied reaction (1) using a selected ion flow tube (SIFT). The reported rate coefficient of 1.9 Â 10 À9 cm 3 s À1 is between the collisional values of the two isomers trans-glyoxal (1.3 Â 10 À9 cm 3 s À1 ) and cis-glyoxal (4.9 Â 10 À9 cm 3 s À1 ). In this study, C 2 H 3 O 2 + was found to decrease with increasing concentration of water vapour in the flow tube whilst protonated glyoxal hydrates did not increase in concentration and minority product ions CH 2  In a recent PTR-MS study, 20 CH 2 OH + was also observed in addition to C 2 H 3 O 2 + . The ratio of these ions changed, with less CH 2 OH + produced at increasing water vapour concentrations in the drift tube. The authors excluded the presence of neutral formaldehyde in the reagent mixture but did not discuss the mechanism leading to the formation of protonated formaldehyde. This unusual formation of CH 2 OH + is a particularly interesting ion chemistry phenomenon. In the SIFT, it increases with water vapour concentration while in PTR-MS it decreases. It is thus important to get insight into the sequences of ionmolecule reactions that can be responsible for this effect. Thus, we have carried out a detailed SIFT study with the aim to elucidate the influence of humidity on the glyoxal hydronium ion-chemistry. The experimental study was supplemented with the theoretical modelling of ion chemistry considering a sequence of individual ion-molecule reactions. We further studied the ion chemistry of formaldehyde under the same conditions to correctly account for the individual secondary reactions.

Materials
Glyoxal was prepared from a 40% glyoxal solution (Sigma-Aldrich) using a method similar to the isolation from glyoxal trimeric dehydrate. 18-21 10 ml of glyoxal solution was placed into a 250 ml flask and evacuated for 24 hours to remove the bulk of water. The resulting high viscose residuum was crystallised to bis(dioxolane) trimer. 1 Crystals were crushed and covered by a layer of P 2 O 5 (Sigma-Aldrich), separated by a 1 cm layer of 1 mm glass beads, to remove any remaining water, and slowly heated up to 150 1C until the yellow-green vapour characteristic for glyoxal appeared by thermal decomposition. 22,23 Glyoxal vapour was re-crystalized in a cold trap, and cooled by dry ice. The glyoxal crystals were stored in a dry ice cooled box to minimise exposure to atmospheric water vapour.
Formaldehyde was prepared by heating a closed vial containing paraformaldehyde powder (Sigma-Aldrich) up to 150 1C. 9 ml of the released formaldehyde vapour was injected into an B4 l Nalophan bag, filled with synthetic air (Messer).

Selected ion flow tube, SIFT
A Profile 3 SIFT-MS instrument (Instrument Science, Crewe, UK, see Fig. 1) [24][25][26] was used for the ion-chemistry study. Since SIFT-MS is described in detail in the literature, only a brief summary will be given here. A variety of reagent ions (H 3 O + , NO + or O 2 + ) is generated by the introduction of a water vapour/ air mixture into a microwave discharge. Reagent ions are selected by a quadrupole mass filter and injected into the 5 cm long flow tube through which a He carrier gas flow was established ( p = 1.5 mbar, T = 300 K). The neutral reactant gas mixture is injected at a known flow rate into the flow tube, where the ionmolecule reactions take place. Ions are subsequently separated using a quadrupole mass spectrometer and detected via a single channel electron multiplier (SCEM).
For glyoxal experiments, 2 mg of glyoxal crystals were placed into a 2 ml glass vial closed by a septum penetrated by a 5 cm long polyether ether ketone capillary (I.D. 0.25 mm), in a diffusion tube configuration. 27 A precision gas standard generator (PGSG; 491M, Kin-Tek) was used at 50 1C to release glyoxal vapour and dilute it with synthetic air to reach the concentration of 15 ppmv, which was then introduced into the flow tube. The humidity of the carrier gas was varied using a reservoir containing 5 ml of water connected to the supply line, thus avoiding the interaction of water with glyoxal prior to its introduction to the SIFT. The water reservoir temperature was slowly increased from À78.5 1C to 60 1C, releasing H 2 O number densities in 1.5 mbar of He between 4 Â 10 12 and 10 14 cm À3 (corresponding to 0.3% to 7.5% sample humidity at a typical SIFT-MS sample flow). To study the influence of He carrier gas on the ion distribution, we varied the flow tube pressure from 0.5 to 2 mbar. Measurements were then carried out at two different water concentrations: H = 0.03 (3 Â 10 12 cm À3 ) and H = 0.08 (1 Â 10 13 cm À3 ), see (4), estimated under standard experimental conditions.
For formaldehyde experiments, a 4 l Nalophan bag containing 5 ppmv formaldehyde vapour in synthetic air was used. Humidity was varied by adding a few millilitres of water at room temperature.
The water vapour concentration in the flow tube was determined from the association reactions of (hydrated) H 3 O + ions with H 2 O molecules, where n Z 1. The H 2 O number density in the flow tube [H 2 O] was calculated as where k 2eff is the rate coefficient for reaction (2) (n = 1), t is the reaction time, and the symbols in brackets correspond to respective ion count rates. Ion count rates were corrected for mass discrimination and for differential diffusion. 28 To compare the experimental results with the numerical model, it is useful to use the dimensionless logarithmic factor as a relative measure of water vapour concentration.

Numerical model KIMI
To model the ion-chemistry kinetics in a flow tube, we developed the Kinetics of Ion Molecular Interaction software (KIMI; using the Microsoft Visual Studio 2017, NET Framework 4.6.1). KIMI interactively solves sets of kinetic equations that describe the change in reagent and product ion concentrations along the axis of the flow tube and thus model ion concentrations at the end of the flow tube as a function of the reactant concentrations.
The input for the model is generated from a graphical interface that allows drawing reaction pathways and systems involving multiple ionic species. Each species has a defined initial concentration and each reaction path is described by a rate constant k. In addition, diffusion losses of ions are characterised by diffusion coefficients, D e , 29 and the characteristic radial diffusion length L. Concentrations of all species in the reaction system are calculated by solving a set of time dependent differential equations, exemplified by the proton transfer reaction using the Runge-Kutta method. 30,31 In the SIFT, the time coordinate directly corresponds to the axial position due to a constant ion velocity v i , estimated from the carrier gas velocity v g as 32 v g can be calculated from the carrier gas flow rate, F C , its pressure, p g , temperature, T g (in K) and the internal flow tube diameter d t as T g is assumed to be in equilibrium with the flow tube (usually around 300 K). The pressure is measured at the end of the flow tube. The pressure inside the flow tube is somewhat higher due to its finite conductivity according to the Hagen-Poiseuille equation Under conditions of the experiments (p = 1.5 mbar, T = 300 K, F = 880 sccm, L = 5 cm, flow tube radius R = 0.5 cm, and the dynamic viscosity of He Z = 2 Â 10 À5 Pa s at 300 K), the pressure difference is about 0.3 mbar. To estimate the average flow tube pressure, the measured pressure was thus increased by 0.15 mbar. The total flow rate was measured for several values of pressure in the flow tube. Under typical conditions, the gas bulk velocity according to (7) is equal to 105 m s À1 and thus the ion velocity calculated from (6) is 157.5 m s À1 . The reaction time for the reaction length L = 4 cm corrected by a positive ''end correction'' 32 e = 0.5 cm is thus 286 ms.

View Article Online
3 Results and discussion

Experimental observations in the SIFT
The mass spectra (see Fig. 2 The change in the relative intensities of product ions as a function of relative water vapour concentration H is shown in Fig. 3. For the lowest humidity (H o 0.1), protonated glyoxal was dominant (485%). With increasing humidity H 4 0.4, protonated formaldehyde concentration rose and became dominant. It is interesting that both hydrated formaldehyde (m/z 49) and hydrated glyoxal (m/z 77) concentration remained below 20%. Water vapour thus removes the glyoxal containing ions and leads to the production of formaldehyde containing ions. This is in stark contrast to analogous acetone ion chemistry that is well understood to terminate under the same conditions in water cluster ions at m/z 77 and m/z 95. 33 Fig . 3 indicates that protonated formaldehyde is formed when H 3 O + ions react with C 2 H 2 O 2 at 300 K in the presence of water molecules, without requiring additional collision energy. Glyoxal and formaldehyde have PA(C 2 H 2 O 2 ) = 675 to 690 kJ mol À1 15 and PA(H 2 CO) = 713 kJ mol À1 , 16 both close to the PA of water, PA(H 2 O) = 691 kJ mol À1 , 16 but much lower than the PA of the water dimer, PA((H 2 O) 2 ) = 808 kJ mol À1 . 34 (12) Reaction (11) and (12) Energetics of reactions (10)-(13) were calculated using the B3LYP/6-311G(p,d) level of theory in the Gaussian 09 software package as the change in the sum of the electronic and thermal free energies between optimized geometries of all reactants and products. According to this theory, reaction (10) is exergonic, DG = À64.81 kJ mol À1 , while reaction (13) 37 The rate coefficients for these reactions obtained in previous studies are given in Table 1.   Fig. 4 and are given in Table 2. These k values were then used for the modelling of formaldehyde and glyoxal ion chemistry.

Ion chemistry of glyoxal.
Ion kinetics of protonated glyoxal was modelled by considering two possible mechanisms leading to the formation of protonated formaldehyde: (10), or (12) and (13).   Table 2 give good fits. We were not able to determine the distribution of vibrational energy and potential barriers of the suggested reactions. Hence, we could not estimate whether the amount of internal energy stored within the association product of reaction (12) is   6 Numerical modelling of possible reaction channels and their influence on relative ion distribution using different reactions and rate constants. In all presented simulations, protonated glyoxal was formed according to (1), using a reaction rate of 1.35 Â 10 À9 cm 3 s À1 in both directions. (a): reaction (10) fixed at 1 Â 10 À10 cm 3 s À1 and reaction (12) variable (units cm 6 s À1 ) (b): reaction (12) fixed at 3.7 Â 10 À27 cm 6 s À1 and reaction (13) variable (units cm 3 s À1 ) (c): reaction sequence (14b) fixed at the association rate 1 Â 10 À10 cm 3 s À1 and the dissociation rate 1 Â 10 À9 cm 3 s À1 to (15) variable. (units cm 3 s À1 ) (d): reaction (13) at 1 Â 10 À9 cm 3 s À1 followed by (16) at k = 1 Â 10 À10 cm 3 s À1 (thick lines) and reaction (13) at 0.89 Â 10 À9 cm 3 s À1 followed by (17) at 1 Â 10 À9 cm 3 s À1 . sufficient for reaction (13) to proceed. Thus, the effect of dissociative association was explored numerically, testing whether C 2 H 3 O 2 + (H 2 O) can be formed in an excited state. Dissociation may be unimolecular: or induced by a collision with water molecules:  Fig. 3). The model explains these observations most appropriately by combining reactions (12) and (13) with rate constants of 10 À27 cm 6 s À1 and 10 À10 cm 3 s À1 , respectively. However, the data can also be modelled by reaction (14b), forming a reaction intermediate C 2 H 3 O 2 + (H 2 O)* with a rate constant of 10 À10 cm 3 s À1 followed by its reaction with water molecules at 10 À9 cm 3 s À1 and (15) at 10 À11 cm 3 s À1 . Fig. 6 shows model results for several combinations of rate constants. In addition, reactions (12) and (15) were studied under variable flow tube pressure (see Fig. 7) thus changing the He number density. For two different water concentrations, an increase in the carrier gas pressure leads to a decrease in the formation of protonated formaldehyde ions and a slight increase of C 2 H 3 O 2 + (H 2 O), as expected from (15). However, the water concentration was influenced by He pressure due to different He/H 2 O mixing ratios and also the diffusion of ions is accelerated at lower pressures. Therefore, these experiments cannot with certainty confirm three-body reactions, however they do not exclude them, either. The relative intensities of CH 2 OH + and CH 2 OH + (H 2 O) observed in the glyoxal experiment differ from those observed in the formaldehyde experiment (Section 3.2.2). The association of protonated formaldehyde and water (see k f3 in Table 1) alone cannot explain this observation. An additional reaction channel forming hydrated protonated formaldehyde is therefore required. The following two reaction channels describe the relative intensities observed in the glyoxal experiment: Both reactions contribute similarly to the model curve (see Fig. 5d) using rates of 10 À10 cm 3 s À1 and 10 À9 cm 3 s À1 for (16) and (17), respectively. Based on numerical simulations, we cannot favour one of them and even their simultaneous contribution is possible.
Finally, we have modelled the endothermic ligand switching reaction (11). The high number density of neutral water molecules will shift the equilibrium in the direction of the protonated water dimer. Modelling with different reaction rate constants showed that the reaction does not affect trends in the formation of targeted species. However, it does affect the equilibrium between protonated glyoxal and protonated formaldehyde (see Fig. 8). The ligand switching reaction with water has a significant effect on the total abundance of ions, thus this effect was demonstrated together with the effect of different reaction equilibrium in (1).
The aim of the study was to describe the main reaction channels in the ion chemistry of glyoxal with hydronium in SIFT-MS. The final model of glyoxal ion chemistry consists of thirteen individual kinetic equations, combining the hydronium and formaldehyde ion chemistry given in Table 1 with eqn (1) and (11)-(13) extended by (16) and (17). Even though we carefully considered each rate constant, the final set of rate constants will not be the only solution modelling the experimental observations. It is quite possible that the resulting set can be additionally tuned to better fit the experimental data and the model does not have to be accurate on the absolute scale. Nevertheless, the modelling shows the dramatic effect of neutral water concentration on the ion chemistry of glyoxal.

Conclusions
We investigated the H 3 O + /glyoxal ion chemistry for variable humidity using a SIFT-MS instrument. The study confirmed the secondary formation of protonated formaldehyde with increasing water concentration. For concentrations over 10 13 water molecules per cm 3 in the flow tube, the protonated formaldehyde became the dominant product ion. The abundance of total product ions was very sensitive to water concentration, quickly decreasing for higher humidity.
The observed experimental behaviour was described using numerical simulation of ion kinetics. An additional study of hydronium water clusters and formaldehyde ion chemistry was carried out to supplement the study of glyoxal ion formation. The numerical study of glyoxal ion chemistry showed that protonated formaldehyde is formed through the reaction of C 2 H 3 O 2 + (H 2 O) with water molecules. Alternative reaction channels involving water are possible, forming a protonated formaldehyde water cluster CH 2 OH + (H 2 O) either from protonated glyoxal (16) or from the protonated glyoxal water cluster (17).
Modifications in the glyoxal interaction with hydronium and hydronium clusters affect the total ion abundance, while changes in the reaction rate with hydronium clusters affect the reaction equilibrium as well. However, in all cases, we may observe a quick dissipation of ion signals for higher water concentration. The ion dissipation is maintained by ligand switching between protonated formaldehyde water cluster CH 2 OH + (H 2 O) and neutral water, forming H 3 O + ÁH 2 O and neutral formaldehyde. This makes the measurement of glyoxal in moist environments using CIMS particularly difficult. In addition, measurements in moist environments are much more sensitive to the influence of formaldehyde. The amount of protonated formaldehyde produced from glyoxal at H 2 O concentrations greater than 10 13 cm À3 corresponds to apparent 5% formaldehyde impurity in glyoxal.