Detailed chemical analysis of a fully formulated oil using dielectric barrier discharge ionisation–mass spectrometry

Rationale Fully formulated oils (FFOs) are integral to automotive lubrication; however, detailed compositional analysis is challenging due to high levels of chemical complexity. In particular, existing mass spectrometric approaches often target particular FFO components, leading to poor analytical coverage of the overall formulation, with increased overheads and analytical timescales. Methods Herein we report the application of a commercially available SICRIT SC‐20 dielectric barrier discharge ionisation (DBDI) source and Thermo Fisher Scientific LTQ Orbitrap XL to the analysis of an FFO. Nitrogen was used as a discharge gas for the DBDI source, and was modified using a range of commonplace solvents to tailor the experimental conditions for the analysis of various components. Results The reported method allowed analysis of a range of FFO components of interest, encompassing a wide range of chemistries, in under 1 min. By modifying the discharge gas used for ionisation, experiments could be optimised for the analysis of particular FFO components across positive and negative ion modes. In particular, use of water vapour as a discharge gas modifier with positive ion mode mass spectrometry permitted concomitant analysis of antioxidants and base oil hydrocarbons. Furthermore, case studies of selected linear alkanes and alkenes profile the differences in the range of ions formed across these saturated and unsaturated aliphatic compounds, giving insight into the fate of base oil hydrocarbons in FFO analyses. Conclusions A rapid method for analysis of FFO compositions has been developed and provides coverage of a range of components of interest. The results indicate that the method presented may be of utility in analysis of other FFOs or similarly challenging complex mixtures.

Owing to the overall complexity of the resultant product, obtaining detailed information on the chemical composition of FFOs using a single analytical technique is particularly challenging. Analysis of highly complex mixtures is a strength of mass spectrometry (MS), particularly when using mass analysers of high resolving power, or when coupled to a chromatographic technique such as gas or liquid chromatography. A range of ion sources are available to tailor an instrument to a particular application. Of those commonly available, electrospray ionisation (ESI) has well-documented use in the analysis of particular additives within FFOs. 5,6 Reported in the literature is the application of dielectric barrier discharge ionisation (DBDI)-MS to a range of analytes not easily accessible by ESI-MS, such as fluorinated alkanes and polycyclic aromatic hydrocarbons. 7,8 Dielectric barrier discharge sources are relatively simple and operate via the generation of a low-temperature plasma. Briefly, two electrodes separated by a dielectric barrier are held at high voltage, and the ionisation of the proximal discharge gas leads to the generation of charged species such as N 2 •+ , O 2 •+ and NO 2 À , amongst others. 7,9,10 Of these, N 2 •+ is principally responsible for much of the subsequent ionisation of analytes through reactions such as proton transfer, charge transfer, electron capture, hydride transfer and ion attachment. 11 DBDI sources are often set up to allow ambient sampling via the direction of the plasma at samples deposited on surfaces. In such a manner, the analysis of various engine oils available on the market has been reported, whereby the authors used a simple experimental set-up to allow chemometric fingerprinting of samples. 12 An alternative DBDI source design has been reported which instead uses an in-line geometry to minimise ion loss due to repulsion.
Not only does the development of a non-ambient DBDI source allow coupling to more automated sampling methods, such as gas chromatography (GC) and solid-phase microextraction (SPME), but also allows low limits of detection for particular analytes to be demonstrated. 13 n-Decane, 1-decene, n-dodecane and 1-dodecene were all purchased from Fisher Chemicals (Loughborough, UK) and diluted 1 in 10 using toluene for analysis. n-Decane and n-dodecane were of 99% purity, whilst 1-decene and 1-dodecene were of 96% purity (remainder isomers).

| Formulation
For each additive, between 110 and 123 mg of material was made up to 1 mL using toluene and vortexed to complete dissolution. From this, 1 mL of an FFO was prepared according to the formulation described in Table 1. For analysis, 125 μL of this solution was made up to 1 mL using toluene.

| Mass spectrometry
All MS analyses were conducted using a Thermo Fisher Scientific LTQ the instance of a pure N 2 discharge gas, or through a 1 L bubbler filled with approximately 300 mL of the corresponding solvent in the instance of modified discharge gases. A depiction of the set-up is included in Figure S1. The source voltage was set at 1.5 kV for positive ion mode analyses and 1.7 kV for negative ion mode analyses. The SPME heating unit was held at 275 C for all analyses, unless otherwise specified, and further modified by the attachment of a Varian CP3800 GC injector septum nut containing a 9 mm non-stick BTO septum (Restek Ltd, Buckinghamshire, UK) to ensure airtight sample injection. Analyses were conducted by injection of 1 μL of diluted FFO sample into the SPME heating unit from a gastight GC syringe.

| Data visualisation
Initial data analysis was conducted in the native Qual Browser When the source is supplied with dry N 2 , the exclusive formation of a fragment ion can perhaps be rationalised by acknowledging the range of reagent ions formed under these conditions, many of which can give rise to ionisation with concomitant dissociation. 10 The ions formed are summarised in Table 2 Table 2. The lack of ions observed for the viscosity modifier and antifoam additives may possibly be attributed to the high molecular weight of these materials or concentration of these analytes within the formulation.

| Negative ion mode DBDI-MS
When operating plasma-based sources using air as a discharge gas, several species have been identified that could contribute to the formation of reagent ions for negative ion mode DBDI, such as oxygen-based radical anions, in addition to NO 3 À and NO 2 À . 7,18 Eventual ionisation of analytes leads to a range of common ion types, such as radical and deprotonated anions, products of ion attachment and oxygenation as well as fragmentation in some instances. 7,19,20 Negative ion mode analyses were conducted that complement the conditions used for positive ion mode analyses reported above,  In all instances, ions were detected with mass differences of less than 2 mmu from theoretical masses. a Spectra for dispersant ions are shown in Figure S4 (supporting information). loss would enable greater stabilisation of a charge via resonance. An example spectrum of an FFO analysis using the conditions discussed above can be seen in Figure S5.
In addition, when water or an alcohol solvent vapour is presented to the source through the bubbler, a hydrocarbon distribution was also identified corresponding to the formation of highly oxygenated species from base oil, shown in Figure 2. This is in contrast with analyses in positive ion mode, where a hydrocarbon distribution is only observed when using water vapour; the ions generated in positive ion mode typically contain a lower oxygen content. The observation of O 2 À attachment ions has been reported for saturated alkanes analysed using DART in negative ion mode; however, using the set-up described for the present study, ions containing three or more oxygen atoms are predominately observed. 21 The range of ions formed in negative ion mode is summarised in Table 2, with spectra given in Figures S5-S7. Notably, no detergent ions were detected under any conditions in negative ion mode, despite these analytes being anionic in solution.

| Petroleomic visualisations
Comparison of ions generated in positive and negative ion modes for hydrocarbon base oil when using N 2 humidified with water vapour as the discharge gas shows two distinctly different distributions, shown in Figure 3. All following analyses focusing on analysing the hydrocarbon base oil used a SPME heater temperature of 320 C. This temperature was used to ensure maximum transfer of base oil hydrocarbons into the gas phase for ionisation within the source; however, it is accepted that at near-atmospheric pressure this temperature is unlikely to lead to vaporisation of the entire base oil hydrocarbon range.
In order to comprehensively profile differences between ions generated in both polarities, visualisation of the datasets was  Figure 4 with an asterisk. This is further discussed when analysing the dataset using oxygen-content bar plots below.
In a separate investigation, N 2 containing methanol or isopropanol vapours used as a discharge gas was also found to generate anionic species for hydrocarbon base oil and were compared to the range of ions generated when using N 2 humidified with water vapour. Visualisation of these datasets using van Krevelen plots shows many similar motifs to those obtained from analyses using water, given by the bottom right plot in Figure 4. Additional van Krevelen, KMD and bar plots are available in Figures S8 and S9 for this study. Group III base oils, such as that used in the present work, may contain less than or equal to 10% unsaturated hydrocarbon content, suggesting that the presence of some alkenes is likely. 25 Hydride abstraction has been posited as the initial step in the ionisation of F I G U R E 5 Oxygen content distribution for FFO hydrocarbon ions generated using N 2 humidified with water vapour as a discharge gas in both positive and negative ion modes. Discontinuous data at 25 carbons in both polarities arises from ionisation of residual phenolic antioxidant [Color figure can be viewed at wileyonlinelibrary.com] alkanes using DBDI, where the formed ions undergo additional reactions in many cases. 15 To investigate the hypothesis of stabilised hydride abstraction in alkene ionisation, two linear alkenes, namely 1-decene and 1-dodecene, in addition to two linear alkanes, namely ndecane and n-dodecane, were analysed. Visualisation of the ratios of ions generated allows the propensity of ion formation to be compared ( Figure 6). To export the mass list, elemental compositions were assigned within Qual Browser to ions over a mass window of 80-240 m/z and above a 2% relative intensity threshold. Inspection of ions below 2% relative intensity was conducted manually and in some instances confirmed the presence of some very minor products of alkane or alkene ionisation; however, these are poorly distinguished from the noise and so the threshold was maintained at 2% relative intensity for exporting mass lists. Data from five repeats were exported and used to plot 95% confidence intervals. Ions containing nitrogen were excluded from the data visualisation.
Comparing the ions formed for all C 10 and C 12 compounds in Figure 6, we see that the increased formation of zero-oxygen species from alkenes is statistically significant when compared to the same classes generated from alkanes. This is in contrast with one-and twooxygen species, which appear to yield similar ion intensities irrespective of being formed from either alkanes or alkenes. We also see that alkenes produce a markedly larger intensity of three-oxygen species compared to alkanes, suggesting that the presence of an alkene moiety promotes formation of zero-and three-oxygen species compared to alkane counterparts.

| CONCLUSIONS
The analysis of an FFO using active capillary DBDI-MS has been