In ﬂ uence of Biodiesel on Base Oil Oxidation as Measured by FTICR Mass Spectrometry

,


■ INTRODUCTION
Internal combustion engine oils undergo aging while in use, a major factor of which is thermo-oxidation which leads to the generation of oxidized species such as peroxides, alcohols, aldehydes, ketones, and carboxylic acids. The generation of carboxylic acids causes additional issues in the form of engine corrosion as a consequence of increased acidity. If oxidation is not controlled, the decomposition of the engine oil can result in oil thickening, 1−3 sludge, and deposit formation. 4 Excessive oil thickening may eventually lead to pumping failures and oil starvation.
To combat oil aging and to improve the physical and chemical properties of the lubricant oil, modern engine oils are usually a mixture of several different components. The main constituent is the base oil, with additives such as antioxidants and viscosity modifiers added to tune the properties of the lubricant oil to meet performance and service interval requirements.
Base oils can have varying properties depending on their composition. The American Petroleum Institute (API) defines five groups for base oils as shown in Table 1. Groups I, II, and III are produced from petroleum sources via various refining methods and mainly comprise hydrocarbons, such as alkanes, alkenes, alicyclics, and aromatics. Non-hydrocarbon species can also be present, of which organosulfur compounds are most prevalent compared to nitrogen-and oxygen-containing species. 5 API Group II and III oils are majority saturates (≥90%), with a higher percentage of normal-, iso-, and cyclo- paraffins (naphthenes) than solvent refined (Group I) oils, and as such performance and chemical properties of these oils are controlled by the relative abundance and chemistries of saturated species in the base oil. 6,7 Production capacity for Group II and III oils is growing at the expense of Group I as demand moves to higher-quality and longer-life oils. 8 The mechanism for the oxidation of hydrocarbons is typically described as a free radical chain process, comprising initiation, chain propagation, chain branching, and termination steps. 3,4,7,10 The radical chain is initiated by the reaction of dissolved oxygen with hydrocarbon compounds; the hydrocarbon compounds are likely to be either aromatics or unstable heterocyclics. 3 After the initiators are formed, propagation proceeds through the reaction between highly reactive aliphatic radicals and oxygen, forming hydroperoxides. In the chain branching stage hydroperoxides decompose to alkoxy and hydroxy radicals which further react, typically forming a variety of products such as ketones, aldehydes, alcohols, and carboxylic acids. The main product is typically ketones for paraffins or ketones and aldehydes in the case of benzene derivatives. 11 The ketone and aldehyde products themselves are readily oxidizable to acids, which in turn can react with alcohol species to form esters. In addition, intramolecular rupture reactions can occur, causing the release of low molecular weight volatile products such as CO 2 , CH 2 O, and CH 3 OH as well as H 2 O. 3 The final termination step occurs when two free radicals combine. The mechanism 12 of nonhydrocarbon base oils as well as their degradation products has also been investigated. 13,14 The oxidation of a fully formulated engine oil can be split into three stages. 15 In the initial inhibition stage minimal oxidation occurs due to antioxidant action from radical inhibitors or hydroperoxide decomposers; 3 this stage ends when the antioxidant species are depleted. In the second stage, oxidation is not influenced by antioxidants and oxidation of the hydrocarbon readily occurs. In the last stage, the oxidation rate slows due to increased oil viscosity resulting from the formation of high molecular weight condensed polymeric oxygenated compounds, 16 limiting the rate of oxygen diffusion into the oil. When applied to base oil alone, without any additives, only the latter two stages are observed; the base oil oxidizes rapidly with a slowing down of the oxidation rate as the viscosity of the oil increases.
One area of concern is fuel dilution of engine lubricant oils, particularly by biodiesel due to the growing trend of moving to higher proportions of biodiesel in diesel fuel blends. Increased fuel dilution occurs due to the higher boiling point of biodiesel, leading to increased engine oil degradation and engine deposit formation. 17 The presence of biodiesel is known to decrease the oxidative stability of the base oil 18 with increased oxidation of lubricant oil observed. 17,19 The oxidation of biodiesel occurs by an initial rapid oxidation of fatty acid methyl esters (FAMEs) to form peroxides which then lead to the formation of other secondary products such as aldehydes, ketones, and acids. 20,21 Further reactions lead to the eventual formation of polyoxygenated high molecular weight material; such material has been shown to be rich in ester content. 20 For the development of improved antioxidants, first a detailed characterization of the species present in oxidized base oils is needed, followed by an understanding of the oxidation processes occurring. Mass spectrometry has been previously used for the study of lubricant oils: 13,14,30,22−29 Fourier transform ion cyclotron resonance mass spectrometry (FTICR MS) offers the highest resolving power and mass accuracy of ultrahigh-resolution mass spectrometry (UHRMS) techniques and is well suited to the analysis of oxidized base oil. While techniques such as nuclear magnetic resonance (NMR) and Fourier transform infrared spectroscopy (FTIR) can give information about an average molecule occurring in a mixture, FTICR MS can give elemental compositional information to the thousands of individual species occurring within a sample. FTICR MS has previously been used for the analysis of biodiesel oxidation, 31 crude oil, 32−35 dissolved organic matter, 36,37 bio-oils, 38−40 and other complex mixtures in the field of petroleomics. 41 Electrospray ionization (ESI) has been previously used when targeting antioxidants in lubricant oil mixtures for degradation studies, as the bulk materials of the oil (hydrocarbon species) are not efficiently ionized by ESI, simplifying the analysis. 23,29 Additionally, negative-ion ESI is ideal for the analysis of the polar carboxylic acid species produced by the oxidation process. Atmospheric pressure photoionization (APPI) is well suited for the ionization of hydrocarbon molecules such as polycyclic aromatic hydrocarbons (PAHs); as such it can be used to characterize aromatic base-oil hydrocarbon species. Linear alkanes such as paraffins are not efficiently ionized by either APPI or ESI; however, several different ionization sources have been proposed for improved ionization of saturated species such as atmospheric pressure chemical ionization (APCI) 28,30 or the use of low-temperature plasma (LTP). 42 Alternative techniques to APPI and ESI such as easy ambient sonic-spray ionization (EASI) have been successfully used for the analysis of biodiesel oxidation. 31,43 In this study FTICR MS data are used in tandem with conventional analysis, providing further insights into the effect of differing levels of biodiesel upon the oxidation of a Group II base oil. While previous studies have used low resolution methods to investigate oxidation of biodiesel and additives, here we use FTICR MS for ultrahigh-resolution characterization of base oil samples to monitor oxidation as a function of time and biofuel content. This, in turn, provides insights into the oxidation processes and how differing levels of biofuels, which can come into contact with engine oils, can affect the rate of oxidation of the base oils. Group II base oils were selected for this study due to their commercial significance for heavy duty diesel (HDD) engine use. We have characterized a series of oxidized group II base oil samples, produced by a benchtop oxidation testing, with differing levels of biodiesel doping: 0, 15, and 100% (B0, B15, and B100, respectively). Petroleomics methods were used for the data analysis, which included additional analyses through modeling of the mean DBE and use of Upset plots.

Samples.
A Group II base oil mixture was produced from two commercial Group II base oils at a 15%:85% w/w ratio to give a final kinematic viscosity measured at 100°C (KV100) of 5.84 mm 2 /s. Properties for these two base oils are given in Table S1. A KV100 for the base oil in this range is required for the formulation of SAE 5W-30 and SAE 10W-40 viscosity grade engine oils. SAE 5W-30 and SAE 10W-40 have a high market share, making the study of this base oil mixture of high commercial importance. Samples with different levels of biodiesel were prepared by the addition of 5 vol % of a diesel mixture comprising mineral diesel (CEC DF-79-07) and biodiesel (80:20 mix of soybean methyl ester to rapeseed methyl ester). The diesel mixture added to each sample had the following compositions: Energy & Fuels pubs.acs.org/EF Article B0 (100% diesel), B15 (85% diesel, 15% biodiesel), and B100 (100% biodiesel). The prepared samples were then subjected to a modified benchtop Daimler oxidation test. The test procedure is described in Figure S1. In brief, a 500 mL three-necked flask equipped with a jacketed coil condenser, and an air inlet set at 10 L/h, was set up in an oil bath with the temperature set at 160°C, and 250 g of oil sample was added with 100 ppm (158.1 mg) of iron(III) acetylacetonate as a catalyst. The oxidation was monitored by samples taken at 0, 24, 48, 72, and 168 h time points.
Conventional Analysis. The kinematic viscosity at 100°C was measured according to the ASTM D7279 44 standard method by using an ISL Houillon viscometer VH1. Initial pH (i-pH) was measured at room temperature according to the ASTM D7946 45 standard method using a Metrohm 855 autosampler with a Metrohm LL-Solvotrode (LiCl salt in EtOH electrode). Oxidation was measured as the FTIR peak height at 1710 cm −1 according to DIN 51453 46 standard method using a PerkinElmer FTIR Spectrum 65 spectrometer.
FTICR MS Analysis. Samples were analyzed with a 12 T solariX FTICR mass spectrometer (Bruker Daltonik GmbH, Bremen, Germany) coupled to an APPI II ion source operating in positiveion (+) mode and a home-built nano-ESI (nESI) source operating in negative-ion (−) mode.
All samples for APPI analysis were prepared as 0.1 mg/mL concentration solutions in a 50:50% v/v mixture of propan-2-ol and toluene (HPLC grade 99.9%, Honeywell, Bracknell, UK). Samples were infused by using a flow rate of 600 μL/h, a drying gas temperature of 280°C, and a nebulizer temperature of 350°C. Ions were accumulated for 0.6 s in the collision cell before being transferred to the ICR cell for detection. Spectra were acquired by coadding 300 broadband mass spectra with a 4 MW data set size with a detection range of m/z 202.7−3000. The value of m/z 202.7 was chosen as it is known that an increased low mass cutoff results in increased resolution for FTMS instruments. 41 As such, it is beneficial to use a value that is optimal to both observe low m/z species present in base oil samples and achieve increased resolution.
All samples for nESI analysis were prepared in an 80:20% v/v mixture of propanol and toluene. The 0 h samples were prepared at 0.1 mg/mL concentration; all other samples were prepared by using a concentration of 0.05 mg/mL. Spectra were acquired by coadding 300 broadband mass spectra with a 4 MW data set size with a detection range of m/z 98.3−1500.
All mass spectra were internally calibrated in Data Analysis 5.0 (Bruker Daltonik GmbH, Bremen, Germany) with the quadratic mode by using a known abundant homologous series present in the sample. Mass spectra generated by (−)nESI were internally calibrated with O 2 [H] species, and APPI mass spectra were internally calibrated with series of HC [H] and O 3 [H] species; calibration mass lists can be found Tables S2 and S3. For both (−)nESI and (+)APPI data a mass accuracy of <1 ppm allowed assignment of a unique elemental assignment per composition. Three replicate data sets were acquired for B0 samples by (+)APPI (0, 24, 48, 72, and 168 h) to enable accurate determination of base oil oxidation trends; all other data sets were acquired once.
Data Analysis. The replicate data sets were processed by using an updated version of the in-house tool Themis, 47 termed Themis 2.0, to generate a single data set from each set of (three) replicates containing peaks only present in two or more replicates. Themis additionally removes noise peaks, enabling the assignment of species at much lower signal-to-noise values than possible in a standard mass spectrum, which may go undetected by using standard peak picking and left unassigned.
Composer 1.5.6 (Sierra Analytics Inc., Modesto, CA) was used for elemental compositional assignment with the following constraints: mass error <1 ppm, and the following limits for atoms; APPI: C 4 The root-mean square ppm error was <0.5 ppm for (−)nESI data and <0.3 ppm for (+)APPI data sets. Once elemental compositions were assigned, species can be further characterized by carbon number, double bond equivalents (DBE), and heteroatom content. DBE is calculated by the following formula: DBE = c − h/2 + n/2 + 1, where c, h, and n are the numbers of carbon, hydrogen, and nitrogen atoms, respectively. Heteroatom content can be summarized by a class label. For example, C 32  Once elemental assignment was completed, further visualization and data analysis were performed by using in-house tool KairosMS, 48 in combination with custom R 49 scripts; Upset plots were generated by using the Complex-Upset 50 package.
■ DISCUSSION Assessment of Base Oil Oxidation Observed in the B0 Samples. Ultrahigh-resolution mass spectrometry data were acquired by both (−)nESI and (+)APPI. The mass distribution for (−)nESI data is dominated by oxygenated species; spectra are shown in Figure S2, with a similar appearance in profile from 0 to 168 h. From an inspection of an enlarged section such as the 1 Da section as shown on the left of Figure 1, however, the spectra can be shown to comprise  Energy & Fuels pubs.acs.org/EF Article oxidation time, indicating the increasing relative concentration of these species with the ongoing oxidation process. The (+)APPI mass spectra, as shown in Figure S3, display the same characteristic "hump" profile as seen for petroleum distributions, with the distribution shifting toward higher mass with increasing oxidation time. Similarly, a zoomed 1 Da inset of a (+)APPI spectrum, as shown on the right of Figure 1, shows the emergence of peaks assigned to oxygenated species. At 0 h typically only HC and HC [H] species are observable; however, at 24 h onward multiple O x species are observed. Similarly to the (−)nESI spectra, typically a series of oxygenated species of the form C a H b O x with a spacing of 36.4 mDa are present; additionally, typically a pair of peaks corresponding to hydrocarbon species are observed: an oddelectron radical species and an even-electron protonated species with the mass split of 13 C 1 vs 12 C 1 H 1 (4.5 mDa). The typical even-electron species distribution per Da can then be assigned to a series of the form The number of monoisotopic species assigned per heteroatomic class can be summarized as a heteroatom class distribution as shown in Figure 2 for (+)APPI utilizing Themis-processed data. A corresponding class distribution for (−)nESI data can be found in the Supporting Information (see Figure S4). (−)nESI preferentially ionizes polar species such as carboxylic acids. In contrast, the (+)APPI mode enables access to less polar but more aromatic species, such as PAHs. APPI produces both even-electron and odd-electron radical species; however, evenelectron species typically dominate for heteroatom containing classes, whereas PAH species are typically observed as radical ions. 51 The use of (+)APPI has the advantage of ionizing hydrocarbon species that are present in the base oil and that are not ionized by (−)nESI, enabling the tracking of the reactions that may increase or decrease the number of species present in the HC [H] Figure S5. Similarly, an increase in the variety of oxidation products has been previously observed by FTIR measurements. 52 24 h was also observed, however, because of the formation of new hydrocarbon species. Formation of new aliphatic and aromatic species has been observed previously, 24 and it is possible such species are formed from the alkylation of hydrocarbon species by an alkyl radical 53 occurring during the initial stages of oxidation.
Some key metrics pertaining to (−)nESI and Themisprocessed (+)APPI data for B0 samples are given in Table 2; in general, it can be seen (+)APPI enables the assignment of ∼3 times the number of species compared to (−)nESI and ∼3 times the number of monoisotopic species. Species generated by (−)nESI typically have a lower carbon number than that of (+)APPI; additionally, the carbon number range covered by (+)APPI is greater. Energy & Fuels pubs.acs.org/EF Article molecular weight material from progressive polycondensation which eventually leads to sludge and varnish formation. 5 In addition, the successive alkylation of aromatic rings through the free-radical mechanism 53 leads to increases in the total number of carbon atoms in the molecule. We can observe the effect of processes that increase the carbon number of the molecule via mass spectrometry by investigating the changes in the carbon number ranges observed; however, it is not possible to determine whether species have a linear or branched structure with direct infusion mass spectrometry alone. Additional techniques such as gas chromatography (GC) 24,26,28 or two-dimensional mass spectrometry 54 would have to be employed to provide further insight into the structural arrangement of the carbon skeleton. Figure  3. After processing the replicates by using Themis, we used the data to construct a single plot to simplify analysis. Additional plots for each replicate can be found in the Supporting Information (Figures S6 and S7) 5 [H] and above display a clear trend of increasing carbon number with oxidation time, suggesting that these species could be the products of successive polycondensation reactions. Additive developments that then limit these successive polycondensation reactions may then limit the formation of high oxygenated material.

The maximum carbon number observed for a selection of protonated classes by (+)APPI; HC [H], and O 1 [H] to O 8 [H] with increasing oxygen incorporation is plotted in
The Figure 4 where a trend of increasing mean DBE with both oxygen content and oxidation time is observed. , which suggests each successively more oxygenated class contains an additional double bond such as the formation of a carbonyl. There also seems to be a trend of increasing mean DBE of a class with increasing oxidation time, such as O 5 [H], which shows a steady increase from 24 to 168 h (7.8 → 8.8 → 9.0 → 9.3) due to the formation of higher DBE material; this can be visualized by a plot of number of species for each DBE shown in Figure S8. Formation of increased DBE species of the same class suggests that more condensed species (with higher number of rings and double bonds) such as saturated ring structures 53   Energy & Fuels pubs.acs.org/EF Article the oxygen functionality present. To ensure the accurate determination of the mean DBE only species with an assigned 13 C isotopologue were included in the calculation, both replicate and Themis-processed data were included. The equation of each line can be found in Table 3, and a plot of the fitted line can be found in Figure S9 An increase in one DBE for an increase of two oxygen atoms would suggest the presence of an additional carboxylic acid group or ester, whereas an increase in one DBE per oxygen atoms suggests the formation of carbonyl species. The number of oxygen atoms required for an increase of one DBE decreases with oxidation time (from 1.58 to 0.95), as such with increasing oxidation time more oxygen content is likely to arise from the formation of carbonyl groups. The formation of additional condensed structures would also reduce the number of oxygen atoms required (as obtained by the linear model); nevertheless, increasingly oxygenated species have a greater chemical diversity due to the possible combinations of different oxygen-containing moieties. Therefore, highly oxygenated species would be expected to form more condensation products than less highly oxygenated species.

number of rings and double bond equivalents (DBE) can be used to infer the aromaticity of species or the degree of unsaturation. A boxplot of the double bond equivalents of the protonated classes, HC [H], and O 1 [H] to O 8 [H] is shown in
The  Figure 5. The species in blue are those that are unique to the 0 h samples; thus, they are no longer observed at 24 h, likely due to oxidative degradation. Species colored in green are those that are common between samples; as such, these species can be considered to remain in the sample after oxidation due to either high concentration or oxidative stability. Species colored in gray are species unique to the 24 h sample; therefore, they are species that have been generated from the oxidation of the 0 h sample. The high DBE species of HC (odd-electron) class are lost after oxidation; these are most likely structures with multiple aromatic cores. Multiring aromatics have been suggested to have an antioxidant effect due to formation of phenols; 7 however, as their concentration in Group II base oils is low, they are not expected to have a significant effect on the oxidation. 55 Effect of Biodiesel Addition. The oxidation of both base oil and biodiesel components produces increasingly complex mixtures due to the additional reactive pathways introduced by the biodiesel. As such, analysis of the biodiesel-doped base oil samples can prove more challenging than oxidized base oil alone.
Conventional Analysis (Kinematic Viscosity and FTIR). The plot of oxidation rating (an FTIR measurement of the carbonyl absorbance made according to the standard method DIN 51453) versus increasing oxidation time is given in Figure S10A. Kinematic viscosity at 100°C (KV 100) versus oxidation time plots are given in Figure S10B. Tabulated data are provided in Table S5. As expected, both the kinematic viscosity and oxidation rating increase with oxidation time, with minimal variation between B0, B15, and B100 samples, except for the recorded kinematic viscosity of the B15 sample at 168 h, which is much lower than that recorded for either B0 or B100. The is likely due to the increased polarity of the B15  Energy & Fuels pubs.acs.org/EF Article sample compared to B0, which improves the dispersion of oxidized material, thereby lowering the viscosity. In the case of B100 versus B15, an increased amount of polycondensation is expected in the B100 sample due to the increased biodiesel content, which leads to generation of higher viscosity material.
This is likely to have a more significant effect than the increased solvency from going from B15 to B100; hence, the measured viscosity increases for B15 relative B100. Similar solubility phenomena have been observed with measurements of deposit weight. 57 The pH values for the 168 h samples for  Energy & Fuels pubs.acs.org/EF Article each biodiesel level were also recorded at pH 1.6, 1.8, and 1.5 for B0, B15, and B100, respectively. This indicates that the most acidic sample at 168 h was B100 followed by B0, with the least acidic being B15. FTICR MS Analysis. The mass spectra profiles for B0, B15, and B100 by (+)APPI all share the same broad "petroleum" style distribution, and as such, it is not possible to easily identify biodiesel from the mass spectra profile alone. It is possible, however, to identify species that are present in the B15 and B100 samples that are not present in the B0 sample, as shown by Figure S13. To gauge the increased chemical diversity due to the biodiesel, an Upset plot 58 was constructed from the 0 h time points of B0, B15, and B100, as shown in Figure 6. For the following analyses to ensure parity between B0 and B15/B100 sample sets a single replicate (r2) was selected to represent B0 samples.
The Upset plot in Figure 6 displays the species in common and those which are unique for the 0 h samples for B0, B15, and B100. Each row in the matrix represents a sample (B0, B15, or B100), each column is an intersection, and the samples present in each intersection are signified by a filled black dot in their respective row; if multiple samples are present in an intersection, the black dots are connected by a solid black line. In similarity to a Venn diagram, the intersection size represents the number of species that are present in that intersection; additionally, the intersection size is then split into constituent heteroatom classes.
As shown in the Figure 6, 706 species, composed of hydrocarbon and oxygenated species, are common between all 0 h samples. These species are common between all 0 h samples as they originate from the base oil itself. An additional 140 species were found in common between the B15 and B100 samples, and most of these species are oxygenated species (O 1 −O 4 ); as such, these can likely be attributed to the fatty acid methyl ester (FAME) biodiesel present in the sample. A significant number of species are unique to the B100 sample (218), most being oxygenated (O 1 −O 4 ), and are likely species also present in the biodiesel mixture that were not detected in the B15 sample due to the decreased biodiesel concentration. Few (25) species were found to be common between B0 and B15 and are likely due to the petroleum diesel. In addition, 48 unique species were found in B15, mostly of hydrocarbon and oxygenated classes; these either represent species present in the diesel mixture or possibly products formed from mixing the biodiesel with the petroleum diesel. A small number of species (31) were also found in common between B0 and B100; these are most likely species that are present but were not detected in the B15 sample. The sample complexity increases from 784 species for the B0 sample to 919 for the B15 sample and 1095 for the B100 sample; it is clear that the addition of biodiesel increases the complexity of the sample.
The class distribution for assigned monoisotopic species for each level of biodiesel is given in Figure 7. The same trend of increasing number of O x species and a shift to higher values of x for O x with increasing oxidation time, as previously observed for B0 samples, are seen for both B15 and B100. B15 behaves very similarly to B0 with a similar number of species generated at each time point; as such, the additional biodiesel species at this level of doping appear to a have a limited effect on increasing the generation of oxidized products. B100, however, displays a significant increase relative to B0 and B15 in the number of oxygenated species present at 24 [H] are first observed at 48, 72, and 168 h, respectively, for both B0 and B15, whereas these three classes are first observed for B100 at 24 h, suggesting that B100 significantly increases the rate of oxidation and generation of polyoxygenated species. Previous studies 18 have found an increase in ester formation of biodiesel containing fuel blends with oxidation time due to the reaction of alcohols and carboxylic acids formed from FAMEs oxidation. Thus, some of the additional oxidized species observed for B15 and B100 samples are likely ester-containing species. At the 168 h time point all samples (B0, B15, and B100) have a very similar distribution of species across the heteroatom classes.
The ongoing oxidation of the base oil is expected to increase the chemical diversity in terms of the number of unique species generated. As such, the oxidation of the different biodiesel blends can be summarized and compared by the total number of monoisotopic species observed as shown in Figure 8A. At 0 h the B100 sample is the most complex in terms of assigned species, followed by B15 and then B0 due to additional species present in the biodiesel. At 24, 48, and 72 h, B100 is significantly more complex than either B15 or B0 due to the generation of additional oxidized material. B15 and B0 show a similar level of complexity, however; therefore, B15 does not cause a significant increase in complexity relative to B0. This observation is consistent with previous research which suggests low loading of biodiesel (B30 and below) 20 does not display as   Figure S11, indicates that the B100 sample contains highly oxygenated species, in particular O 8 [H], that are not present in either B15 or B0 samples. A DBE plot ( Figure S12) of these unique species suggests that B100 168 h unique species are highly oxygenated with a typically lower DBE than B0 or B15 unique species; these species are possibly polycarboxylic acids. However, further investigation is needed to confirm this. The oxygen content of the species can be summarized by the mean oxygenated class (mean x value of O x [H]) as plotted in Figure 8B. does not seem to cause the generation of significantly more oxygenated species than B0 alone. This is likely due to the low loading of biodiesel (15%) and high percentage of petroleum diesel (85%) in the B15 sample, making the B15 sample more similar to the B0 sample in terms of chemical composition; it would therefore be expected to have a similar behavior to the B0 sample. It is however possible that additional species were generated in the oxidized B15 sample but were subsequently lost due to their volatile nature.
At When using an isolation window in conjunction with an increased low-mass cutoff, as compared to broadband spectra, an increased resolving power results. To further probe the similarity of samples of differing biodiesel contents at 168 h, an isolation window with an m/z width of 40 and centered at m/z 700 was used for each of the B0, B15, and B100 samples. No significant differences were observed, however (see Figure  S14). It is possible B100 does display some additional complexity that is not observed due to ionization competition, which may be determined and resolved by either offline sample fractionation prior to direct infusion analysis or by using online chromatography coupled to FTICR-MS such as GC-59 or liquid chromatography (LC)-FTICR-MS. 60 Assessment of FTICR MS for Oxidized Base Oil Analysis. Previous studies have shown FTICR MS is well suited to base oil analysis, 22,28 with the use of (+)APPI or (+)APCI preferred as these techniques are both able to ionize the hydrocarbon base material. In this study, (+)APPI and (−)nESI methods were compared, and ionization of hydrocarbon components was found to be desirable in order track oxidation pathways. The oxidized material is both efficiently ionized by (+)APPI and (−)nESI and was found to contain species containing multiple oxygen atoms.

■ CONCLUSIONS
FTICR-MS has proved well suited to the analysis of the complex samples produced from oxidation of a Group II base oil via a benchtop oxidation method.
• Both (−)nESI and (+)APPI were efficient at ionizing oxygen-containing base oil oxidation products, with (+)APPI preferred due to the access to hydrocarbon species. • Base oil oxidation was found to produce polyoxygenated species containing 1−8 oxygen atoms, with the generation of increasingly oxygenated compounds favored with increasing oxidation time. This study provides a method for characterizing Group II base oil oxidation by FTICR MS; the detailed characterization affords further understanding of fundamental oxidation processes occurring in base oil samples. The findings may be potentially used for development of additional additives either to target material that is easily oxidizable and then therefore delay the onset of oxidation or to prevent the formation of highly oxygenated polymeric material which may be prone to deposit formation. This method can also be applied to other base oil groups such as Groups III/IV; the same methodology should also be applicable to other hydrocarbon-based oil groups. Alternative chemistries such as those found in Group V oils may require alternative optimal ionization methods. This method is also applicable to fully formulated engine oils; however, (+)APPI ionization may be preferred to avoid signal suppression occurring due to preferential ionization of additive components occurring in ESI. This method may be further expanded with the use of online chromatography methods such as GC-FTICR-MS to resolve isomeric species and verify the functionality and structure of oxidation products.