Particulate Characterization and Lubricating Oil Tribology Investigations of a Gasoline Compression Ignition Engine Using Low Octane Fuels

Economists have predicted total annual global energy consumption will increase by 1.3% in 2023 [1]. At the same time, renewable energy and natural gas consumption would rise by ∼2.3% and 1.4% per annum, respectively [2]. The share of petroleum-based fuels will continue to increase in the global energy supply [3]. Conventional fuel-powered internal combustion (IC) engines play a vital role in industrial, transportation, and agricultural sectors due to their reliability and performance for various applications. Reciprocating IC and jet engines propel >99% of transport vehicles worldwide [4]. Compression ignition (CI) engines exhibit higher efficiency, whereas spark ignition (SI) engines emit lower harmful emissions. Further research and development of CI engines can significantly improve their efficiency and reduce emissions [5]. Several advanced combustion techniques have been proposed and investigated to control emissions [6]. In this context, low-octane gasoline in CI engines as fuel has the potential to deliver significantly improved engine performance. This can be done by using a novel engine combustion technique called gasoline compression ignition (GCI), which can simultaneously control the emissions of nitrogen oxide (NOx) and particulate matter (PM) as well [7]. GCI engines incur lower operating costs than comparable CI engines, and fewer hardware modifications are required in the production-grade CI engines to operate them in GCI mode [8]. Optimizing the charge stratification, fuel quality, and injection strategy improved the GCI combustion. For stable combustion, preheated air with optimum exhaust gas recirculation (EGR) and boost pressure are required [9]. Boost pressure shows good correlations with the test fuel octane number. A high boost with less EGR is required at high loads to shorten the ignition delay [10]. EGR addition may decrease or increase the PM emissions and particle number concentrations (PNC) by inhibiting/supporting particle growth. However, EGR substitution could affect nucleation mode particle (NMP) and accumulation mode particle (AMP) formations [11].

Mixing-controlled combustion plays a significant role in PM formation because it alters charge homogeneity and combustion temperature [12]. In GCI mode, the combustion phasing is closely associated with the fuel injection strategy. GCI engines show improved combustion stability at mid-loads, but high engine speeds and loads show combustion instability issues [13]. Typically, GCI engines emit lower PM emissions than diesel engines [14]. However, studies on particle size, PNC distribution, and chemical characteristics are scarce and required. These studies could help design an efficient diesel particulate filter (DPF) for further emission control from GCI engines, enabling them to meet the most stringent emission norms. Diffusion combustion of highly aromatic fuel and premixed combustion of low aromatic fuel showed low PNC [15]. Unsaturated hydrocarbon fuels resulted in higher emissions of AMPs from the GCI engines [12].

Advanced fuel injection timing, high EGR, and fuel composition alter the lubricating oil characteristics. Pilot injections in GCI mode lead to engine cylinder wall-wetting and fuel dilution issues [16]. Fuel–air mixture entering the squish region undergoes incomplete combustion and engine blow-by [17]. Gasoline has lower lubricity than diesel, producing higher engine components' wear and tear [18]. This is due to easy oil squeezing between the component's mating surfaces [19]. The thermal decomposition of lubricating oil significantly increases semi-volatile combustion byproducts [20]. The used lubricating oils contain toluene, xylenes, $C3$ to $C4$-benzenes, $C2$ to $C4$-naphthalene, pyrene, and fluoranthene in low concentrations. Liquid chromatography-mass spectrometry (LC–MS), Fourier transform infrared (FTIR), nuclear magnetic resonance spectroscopy (NMR), and inductively coupled plasma-mass spectroscopy (ICP-MS) are used for lubricating oil analysis. Proton (¹H) and carbon (¹³C) NMR spectroscopy identify aromatic and aliphatic carbon and hydrogen in the lubricating oil [21]. The analysis of lubricating oil compounds using gas chromatography reports ∼76 and 68% acyclic and monocyclic alkanes [22]. Lubricating oil contamination with PM and water increases the $−OH$ radical concentration in the lubricating oil [23]. An increase in engine runtime raises trace concentration of alcohols, aromatics and organic acids in the lubricating oil [24]. Sejkorova et al. [25] identified the $−OH$⁠, nitro and sulfate groups in used lubricating oils. Used lubricating oils showed different spectra in the sulfonate detergent peaks (1172 and 1156 cm⁻¹) and zinc dialkyl dithiophosphates (ZDDP) additive peaks (970 and 656 cm⁻¹) than fresh lubricating oils. Changes in lubricating oil characteristics affects fuel consumption, lubricity, and soot formation in the engine.

In an engine, the fuel and soot mix with the lubricating oil film on the cylinder liner, eventually reaching the oil sump [26]. Lubricating oils contain ash and trace metals because of contamination by the combustion byproducts and wear debris. Tekie et al. [27] examined trace metals in lubricating oils using dry ash and ICP-MS and reported reasonably good results. The soot contamination resulted in oil film breakdown in the engine due to increased lubricating oil viscosity [28], which increased the engine wear due to oil pumping issues. Rocca et al. [27] investigated the soot-in-oil and reported that soot aggregates between 50 and 130 nm had a primary particle diameter $(dp)$ of 10–35 nm and an inner core of ∼8–15 nm. Fractal dimension $(Df)$ of soot aggregate informs about soot behavior [29]. More porous and complex aggregates show a higher $Df$ than compact aggregates. High $Df$ exhibits high oxidative reactivity due to increased surface area and more complex structure.

This study investigates the comparative tribological characteristics of lubricating oil and particulates formed in GCI vis-à-vis conventional diesel combustion (CDC) engines. The experiments were performed in a medium-duty CI engine running at 5 and 10-bar brake mean effective pressure (BMEP) at 1500–2500 rpm. Engine exhaust particle sizer (EEPS) was used to measure the PNC. Changes in the lubricating oil properties in GCI and CDC engines were compared. LC–MS, FTIR, NMR, and ICP-MS techniques were used to analyze the changes in the chemical characteristics of lubricating oils drawn from GCI and CDC engines.

Figure 1 shows the schematic of the experimental setup. The test engine is a four-stroke, four-cylinder, turbo-charged, water-cooled compression ignition engine (Table 1). Stock electronic control unit (ECU) with predefined calibration maps operated the engine in CDC mode. An open ECU was used to optimize various parameters (fuel injection pressure, injection timing, mass, etc.) for the GCI mode operation. Detailed installation procedure of open ECU in a production-grade engine can be found in Ref. [30].

A 30:70% (v/v) diesel and gasoline blend (G70) was used as GCI mode test fuel, while baseline diesel was used as CDC mode test fuel. The engine had a common rail fuel injection system and solenoid injectors. A triple injection strategy was used in GCI mode to achieve partially premixed combustion (PPC). Fuel injection strategy, intake air temperature (IAT), and EGR rate were varied in each operating condition for maintaining the combustion phasing at the desired crank angle position (Table 2).

The GCI engine was operated for the three cases by adjusting the timings of the two pilot injections $(θpilot−1,θpilot−2)$ and one main $(θmain)$ injection. M1, M2, and M3 were the three operating cases for 1500 rpm at 5 bar BMEP in GCI mode, as shown in Table 2, and CDC was the corresponding baseline case for comparison. Injection timing, fuel quantity, and EGR rate were unknown for the CDC cases since the original equipment manufacturer (OEM)-configured ECU was used. For PNC analysis, EEPS data were acquired for 60 s at 1 Hz frequency, and an average of the 60 data sets was used. Specifications and working principles of the EEPS can be found in Ref. [31]. In GCI and CDC mode experiments, Universal Gold oil 20W40 was used as a lubricating oil. The engine was operated in GCI and CDC modes for 20 h each, per the test matrix shown in Table 2, while keeping the lubricating oil temperature <120 °C. Table 2 shows the engine run time in each of the GCI and CDC modes, totaling 20 h before the lubricating oil samples were drawn for physical and chemical characterization. Fresh lubricating oil was also taken for the baseline physical and chemical characterization.

GCI and CDC engine lubricating oil samples were analyzed according to ASTM procedures to determine their physicochemical properties. A kinematic viscometer (Stanhope-SETA), copper corrosion bath (Setavis), and a flash point apparatus (SETA-flash) were used to measure the lubricating oil viscosity (ASTM D2270) [32], copper corrosion (ASTM D130) [33] and flash point (ASTM D92) [34], respectively. Furnace with a silica crucible and Ramsbottom test were used to determine the ash content (ASTM D482) [35] and carbon residue (ASTM D189) [36] in the lubricating oil samples. A high-speed oil centrifuge (REMI) was used to determine the pentane and toluene insolubles (ASTM D893) [37] in lubricating oil samples. The test procedures for physical property characterizations of lubricating oil samples are well established [19,38]. The cracking test is a standard laboratory test to indicate the water content in the lubricating oil. In the cracking test, a drop of oil was placed on a hot plate held at ∼150 °C. The dry ash method with ICP-MS was used to determine the trace metals in the lubricating oils [27]. A 2.0 g oil sample was taken in a silica crucible and placed in a furnace at 550 °C for 1.5 h. The sample was then cooled in the furnace, and the ash was weighed. Ash was dissolved in 2 ml of conc. $HNO3$ and placed on a hot plate at 120 °C till complete evaporation of acid. 10 ml solution (1% $HNO3$ mixed milli-Q) was added to this dissolved ash and filtered through 0.2-micron PTFE syringe filters. This sample was further diluted with 25 milli-Q (18.2 MΩ). 14 ml of the sample was taken in 15 ml sterile centrifuge tubes and centrifuged for 5 min. Then the tubes were transferred to ICP-MS (Thermo Scientific, X Series) for trace metals analysis.

LC–MS with electrospray ionization (ESI-MS) identifies the components, structure, and chemical properties of molecules in the lubricating oil samples [39]. A mass spectrometer (Agilent portfolio-LC single quadrupole) was used to obtain the mass spectra. 5 µl of a lubricating oil sample was injected into a mobile phase with methanol solvent. The inlet capillary and chamber currents were 0.07 and 0.3 µA. The nitrogen drying gas flowrate was maintained at 11 l/min. The gas temperature environment was maintained at 320 °C. MestReNova-14 software was employed for the chromatogram analysis.

FTIR technique was used to investigate the structural changes in lubricating oil's functional groups. FTIR spectrometer (Perkin Elmer) with the KBr pellet method was used for the FTIR analysis of the lubricating oil samples in the wave number range of 4000–400 cm⁻¹. A total of four scans were acquired at a resolution of 4 cm⁻¹ and converted to absorbance spectra. Oil samples (1 wt%) were mixed with KBr powder and placed on a transparent disc to measure the transmittance IR.

Major hydrocarbon components can be quantified using NMR spectroscopy. NMR analysis was conducted at 24 °C on a spectrometer (JEOL 500 MHz) using an inverse detection probe in a z-direction. For proton NMR, 0.2 ml oil was diluted with 0.4 ml of deuterated chloroform (CDCl₃) [24]. The lubricating oil sample was diluted with CDCL3 until saturation for carbon NMR. The signals were divided into different regions for classifying the proton-containing functional groups based on chemical shifts (ppm) in ¹H NMR. They are aliphatic protons (0.8–1.8 ppm), alpha to unsaturated hydrocarbons {H–C–C = O} (1.8–3.2), alpha to heteroatoms {H–C–O} (3.2–4.4), Anomeric {O–CH–O} (5–6), Aromatics {H–Ar} (6.5–8.5), and Aldehydes (9.5–10.1) [23]. Similarly, the signals were divided into groups, such as Aliphatic (10–60), Oxygenates (50–80), group of Phenols, PAHs and Olefins (120–136), Aromatics (141–160), COO/N–C = O groups (160–188) and R(C = O)R’ groups (188–230) in ¹³C NMR [40]. The area under resonance peaks in NMR spectra refers to ¹H or ¹³C atoms of corresponding chemical types. Lubricating oils contain many primary, secondary, and tertiary aliphatic and aromatic hydrocarbons [41]. Various chemical shifts and their affiliated chemical groups (Table 3) are found in Ref. [42].

The soot reaches the oil sump via blow-by gas or the lubricating oil layer on the cylinder liner, increasing its viscosity [28]. Carbon residue, FTIR, insoluble tests, etc., are used to quantify the presence of soot in the lubricating oils. The morphological structure of soot can be analyzed by TEM using suitable solvents [43]. For this, 1 ml of lubricating oil was diluted with 60 ml n-heptane in a centrifuge vessel. The vessels were placed in the centrifuge (REMI) and centrifuged for 2 h at 1500 rpm. The n-heptane/oil supernatant was removed without disturbing the soot sediment (∼3 ml). The above procedures were repeated six times to remove the oiliness from the soot. The sample was transferred to a 2 ml centrifuge tube, followed by ultrasonication for 5 min. Rocca et al. [43] reported that the aggregates' macro-size increased with centrifugation.

In contrast, ultrasonic bathing breaks the large aggregates into small clusters/chains. Two drops of the sample were dropped on TEM grids using a micropipette. A transmission electron microscope (FEI-Technai G2 Twin 120 KV TEM) was used for soot imaging. Figure 2 shows the methodology for identifying primary particles of soot.

Spray quenching and wall-wetting effects increase HC and CO emissions in GCI engines. Hence, non-reactive split-injection simulations were performed to observe the fuel quenching using Converge computational fluid dynamics (CFD). A summary of models used for engine modeling is given in Table 4. Iso-octane and n-heptane were taken as surrogates for gasoline and diesel, respectively. A reduced primary reference fuel (PRF) mechanism (73 species and 296 reactions) was used for GCI combustion [47], whereas a reduced mechanism (42 species and 168 reactions) was used for CDC [48]. Based on OEM data, a single injection was used in the CDC. A trapezoidal injection rate shape was used for CDC and GCI mode cases [49].

Figure 3(a) shows the engine's computational domain near TDC. Adaptive mesh refinement was used to refine the grid based on temperature and velocity gradients [46]. The images were taken at two horizontal plane positions in the combustion chamber (Fig. 3(b)). Figure 3(c) shows that the developed model is validated using the experimental in-cylinder pressure data.

PM emissions from GCI and CDC engines are discussed initially. Changes in the physicochemical properties of lubricating oil samples are investigated after that. Then, the trace elements and soot particles in lubricating oils are discussed in detail. Finally, the modeling study examined fuel distribution in the engine cylinder, and the results are discussed in the following sections.

The particles are generally classified into three categories: nano-particles (NP; mobility diameter $(dm)<10nm$⁠), nucleation mode particles (NMP; 10 < d_m < 50 nm), and accumulation mode particles (AMP; d_m > 50 nm) [8,50]. Figure 4 shows low NPs at 5 bar BMEP for the baseline case. A bi-model distribution of NPs and AMPs was observed for GCI cases. Higher PNC was observed for GCI than baseline due to PPC. Increasing engine speed reduces the volumetric efficiency and available time for soot oxidation, increasing the small particles and clusters [51].

Retarded injection timings (M1 to M3 sweep) showed an insignificant effect on the particulate emissions. However, at 2000 rpm, retarded injection timing (M6) exhibited higher PNC than advanced injection timing (M4). Yang et al. [52] reported that PNC increased with retarded injection timing at no (0%) EGR. PNC increased and then decreased while retarding the injection timing at 20% EGR. PNC emissions depend on in-cylinder dynamics. Increasing engine load increased PNC for baseline and GCI modes. At high loads, more NMPs were generated and transformed into AMPs [53]. Increased fuel–air ratio and temperature resulted in diffusion combustion, producing soot [51]. With increasing EGR, both NMPs and AMPs increased at medium loads and speeds, as reported in Ref. [54]. H4 showed higher PNC than H6 due to incomplete combustion. Advanced injection led to spray wall impingement, causing liquid fuel film formation and subsequent pool fire leading to soot formation [52].

GCI mode showed 97–99.7% and 3–98% higher NP concentrations than baseline CDC mode at 5 and 10 bar BMEP, respectively (Fig. 5). With increasing engine speed, PNC increased, and the peak concentration shifted to AMPs region [55]. GCI mode at high engine load exhibited higher NP due to more fuel burning, oxidation of larger soot particles, and condensation of heavy hydrocarbons in the dilution line [56]. Reduction in EGR rate improved the combustion of fuel. Therefore, lower NPs were observed at 2500 rpm than 2000 rpm at 10 bar BMEP (Table 2). At all engine operating conditions, fuel injection timing significantly impacted the formation of NPs. At 5 bar BMEP, NMPs were significantly higher (10–67 times) than NPs for baseline and GCI modes. Increasing the engine speed increased the NMPs by 0.6–10% and 0–55% for baseline and GCI modes, respectively. Higher engine speed produced heavier hydrocarbons, producing higher NMPs [57]. GCI mode engine operation at 10 bar BMEP showed 4.5–63% higher NMPs than baseline CDC mode, except for H4 and H5, which showed 2–20% lower NMPs. Gasoline PPC combustion produced more NMPs at higher loads and speeds [58]. Typically, with increasing engine speed, NMPs decreased, and AMPs increased due to coagulation of NMPs [59]. In this study, NMPs and AMPs increased with increasing engine speed. Fuel injection timings did not significantly impact NMPs at high engine speeds.

AMPs were higher for baseline CDC and GCI modes than NMPs (Figs. 5(e) and 5(f)). At 5 bar BMEP, AMPs were 0.4–4.8 times higher for GCI mode than baseline CDC mode. In contrast, Chen et al. [60] reported higher AMPs for baseline CDC mode than diesel-gasoline blends. 10 bar BMEP showed 0–4.84 and 1.2–7.2-times higher AMPs than 5 bar BMEP for baseline CDC and GCI mode cases. Increasing the engine load increased the AMPs due to higher surface growth and coagulation of soot [57]. 10 bar BMEP and 2000 rpm showed 3–10.3% more AMPs than 1500 rpm. Increasing the engine speed initiated frequent particle collisions, increasing the formation of AMPs [58]. With increasing EGR, AMPs increased due to reduced oxygen availability in fuel-rich combustion zones [61]. Due to the absence of EGR, 2500 rpm engine speed exhibited 18–34% lower AMPs than 2000 rpm. Not-so-strong fuel evaporation, wall-wetting, and pool fires on the piston surface induce large AMPs, increasing the PM mass emission [62].

Figures 5(a) and 5(b) show lower total particle number (TPN) for baseline CDC than GCI mode. With increasing engine speed, primary particles have a shorter time available to form and grow, leading to a smaller count mean diameter (CMD) of the soot formed [56]. TPN remains unaffected by fuel injection timings at lower engine loads and speeds. Too advanced and retarded fuel injection timings and small pilot-main injection intervals can affect the TPN [63]. 10 bar BMEP showed 1.3–1.9 (baseline CDC) and 1–5.6 (GCI mode) times higher TPN than 5 bar BMEP. Fuel-rich high-temperature zones result in pyrolysis of fuel and lubricating oil, increasing the particulate emissions [64]. TPN and CMD of soot particles increased with engine speed at high loads due to the dominance of diffusion combustion [56]. Since GCI engines emitted higher AMPs, the TPM was higher than the baseline CDC engine [58]. A slight and 2.7–8.4 times increment in TPM was observed for baseline CDC and GCI modes upon increasing the engine speed. CMD was higher for GCI mode than baseline CDC mode except for 5 bar BMEP and 1500 rpm. At low loads, CMD showed lesser dependence on engine speed but more dependency on engine load [58]. At 5 bar BMEP, GCI mode showed 30–35% larger CMD for 2500 rpm than 1500 rpm. Typically, larger particles can be trapped easily by the DPF. CMD was slightly and 21–92% higher for baseline CDC and GCI modes at 10 bar than 5 bar BMEPs. Increasing soot formation and aggregation at high engine loads yielded larger particles [64]. The CMD range was between 55 and 149 nm, and the highest values were observed at high engine speeds. Finally, TPN and TPM were higher for GCI mode, requiring a more effective after-treatment device to meet the stringent and upcoming emission norms.

Wear debris, moisture, fuel dilution, and chemical changes due to heating alter the lubricating oil's properties. Dense contaminants, gums, asphaltic and phenolic compounds potentially alter the lubricating oil's viscosity. GCI engine lubricating oil showed higher density than CDC engine lubricating oil (Table 5) due to pyrolyzed fuel components and PM mass mixing. Lower viscosity of the lubricating oil affected the lubricating oil film formation, oil film thickness, and load-bearing capacity, thus increasing the engine component wear.

GCI engine lubricating oil showed lower viscosity than CDC engine lubricating oil. Oxidation, fuel dilution, moisture addition, insoluble contamination, depletion of additives, and addition of lighter fractions reduce lubricating oil's viscosity. Lubricating oil's viscosity may increase or decrease with usage, depending on the dominance of any of these factors [65]. A lower flash point temperature was observed for GCI engine lubricating oil than for CDC engine lubricating oil due to higher fuel dilution [65]. This phenomenon was also discovered in the CFD simulation results in Sec. 3.8. Higher ash content and carbon residues were observed for CDC engine lubricating oil than for GCI engine lubricating oil. The ash was 0.51, 17.54, and 8.77 wt% for unused, CDC, and GCI engine lubricating oils, respectively. Atmospheric dust, wear debris from component wear, decomposed fuel, and lubricating oil additives lead to the formation of carbon residues in the lubricating oil. Carbon residues increased by 45.45% and 25% for CDC and GCI engine lubricating oils, respectively, than unused lubricating oil. The copper corrosion test showed a 3a grade (Dark Tarnish) level for CDC and GCI engine lubricating oils. The copper corrosion test indicated the corrosiveness of lubricating oil to the copper-containing components [65]. Pentane- and toluene-insoluble tests can determine suspended contaminants in the lubricating oils [38]. Insignificant insoluble matter was observed in the lubricating oils from both engines. GCI lubricating oil contains more fuel traces and lower wear debris than CDC lubricating oil.

$C17–C35$ carbon atoms form significant lubricating base oil components (80–90%), and the rest include organo-metallic additives (10–20%) [22]. Alkanes and olefins are saturated and unsaturated hydrocarbons commonly used as base oils. Various additives are dissolved in the lubricating oil to give it desirable properties. During usage, lubricating oils may get contaminated and undergo physicochemical changes. LC-MS analysis is used to identify the changes or deterioration over time in the lubricating oil. The peak at m/z 422 ([M + H]⁺) in the methanol atmosphere refers to di-nonyl diphenylamine, an antioxidant [66]. Used lubricating oils from GCI and CDC mode engines showed 75.1 and 77.5% lower m/z 422 count than unused lubricating oil (Fig. 6). Peaks at m/z 296 were mono-nonyl diphenylamine, and m/z 369 and 397 were trimethylolpropane ester additives [66]. With usage, antioxidant decay results in the formation of several undesirable byproducts. GCI engine lubricating oil showed 0.4 and 1.85 times higher m/z 369 and 397 counts, respectively, than unused lubricating oil. They were 14 and 97% higher for CDC mode engine lubricating oil than unused lubricating oil. At the same time, fragmentation of an ion of high m/z can result in the production of multiple small ions with lower m/z values. The base lubricating oils m/z can vary from 50 to more than 1000. Hydrocarbon molecules with an m/z 100–450 are majorly present in the lubricating oil. The concentration of components in the m/z 100–400 range was 17.2% higher and 89.4% lower for GCI and CDC engine lubricating oils, respectively, than unused lubricating oil. The peaks at m/z 196, 224, 280, 322, and 350 correspond to $C11,C13,C17,C20,andC22$ n-alkanes, respectively [67]. Hence, fuel dilution to lubricating oil can increase the m/z 100–400 components. Alcohols (-OH), organic acids (-COOH), and amides (-CONH₂) are minorly present in lubricating oils. Alcohols act as solvents to base oils and additives, whereas organic acids are corrosion inhibitors and pH adjusters [68]. Amines are used as friction modifiers or viscosity improvers. Lubricating oil contains various viscosity improvers (m/z 500–3000), molybdenum compounds (m/z 350–500), corrosion inhibitors (m/z 119–300), and diarylamine antioxidants (m/z 170–207) [69,70]. These components were detected in higher concentrations in the CDC engine lubricating oil than in the GCI engine lubricating oil. The fragmentation of -dioctyl diphenylamine and molybdenum complex can form the peak at m/z 322, representing the benzylic tertiary carbon [71]. Olefins, aromatic components, fatty acid esters, and thiophosphates are present in the base oils, and their peaks occurred between m/z 450–750 [72]. They were 139% higher and 32.8% lower for GCI and CDC engine lubricating oils, respectively than the unused lubricating oil. Dibenzothiophenes, phthalic acid esters, hydrogenated polydecene and alkylated aromatics compounds have ∼m/z 800, and they are present in the base oils. Large molecules such as polyalphaolefins and alkylated naphthalenes have ∼ m/z 1000–2000, detected in the base oils [73].

Relative concentrations of peaks at m/z 200–400 were higher for GCI mode than baseline CDC mode. The used lubricating oil contains higher polynuclear aromatic hydrocarbons than unused lubricating oils, increasing peaks in m/z 300–500 [74]. Lubricating oils from gasoline engines showed 22 times higher PAHs than diesel engines [75]. Figure 6(d) shows a chromatogram of G70, showing m/z range between 100–1400. Diesel has higher concentrations of compounds with lower m/z than gasoline [76]. Barnett et al. [77] detected gasoline fuel additives at 320–450 °C with m/z 600–810. Peaks at m/z 146, 217, 274, 325, and 517 were observed in G70 and GCI engine lubricating oil. Peaks at m/z 196, 736, 792, and 858 might belong to baseline diesel. Dilution of lubricating oil by diesel led to the detection of these components in oil. The gasoline components can react with lubricating oil whereas diesel components remain suspended [78]. Concentrations of base oil and additives were lower in GCI engine lubricating oil than in CDC engine lubricating oil, requiring more frequent lubricating oil replacements.

In Fig. 7, 3500–3350 and 1700–1600 cm⁻¹ bands exhibited the O-H stretching and bending vibrations, respectively [79]. Due to moisture and ethanol, GCI engine lubricating oil showed high transmittance in these bands. The lubricating oils showed bands at 3000–2850 cm⁻¹, an intense peak at 1460 cm⁻¹, and a less strong peak at 1377 cm⁻¹ due to hydrocarbon compounds (C–H bands) [78]. Spectral bands at 1760–1710 cm⁻¹ inferences about the in-service oil deterioration kinetics due to radical oxidation [80], and found to be higher in GCI engine lubricating oil. Bands within 1650–1600 cm⁻¹ indicate nitration products in the lubricating oils. Nitration and carbonyl groups (N = O, C = O, etc.) were lower in CDC engine lubricating oil than in GCI engine lubricating oil. GCI engine lubricating oil showed higher transmittance at 1300–1150 cm⁻¹. Bands at 1300–1250 cm⁻¹ are associated with sulfonate salts in the lubricating oils [80]. The peak at 1156 cm⁻¹ related to polymethacrylate, shows a pour-point depressant additive of the lubricating oil [78]. Peaks at 1172 and 1156 cm⁻¹ also indicate the presence of sulfonate groups $(SO3−)$ [25]. Bands at 1040–1020 cm⁻¹ confirm the existence of sulfonate groups. Unused lubricating oil contains higher sulfonate group elements than used lubricating oils. During engine combustion, sulfur compounds are combusted, resulting in lower sulfonate components in the lubricating oil.

Bands at 1050–920 cm⁻¹ related to the P–O–C in ZDDP are effective antioxidants. The absorption bands at around 990–980 cm⁻¹ are generally associated with the P–O–C starting vibrations of ZDDPs [25]. Lubricating oil undergoes thermal and oxidative stresses, destroying ZDDPs. CDC engine lubricating oil exhibited lower ZDDPs than GCI engine lubricating oil, increasing the component wear and leading to engine degradation. Bands at 990–900 cm⁻¹ characterize the C–H stretching vibration of unsaturated hydrocarbons [80]. GCI engine lubricating oil showed higher transmittance than unused lubricating oil in this band. The peaks at 720 and 654 cm⁻¹ signify the P = S stretching of ZDDPs [81]. Peaks corresponding to ZDDPs were higher for GCI than CDC engine lubricating oils. Bands at around 760–740 cm⁻¹ signify the presence of C = C, -CH = CH-, S–O, and halo components in the lubricating oils. GCI engine lubricating oil contained more hydrocarbon components than CDC engine lubricating oil. Diesel and G70 showed almost similar FTIR spectra (Fig. 8).

Asymmetric and symmetric stretching of $CH,CH2,CH3$ were detected at 2924 and 2854 cm⁻¹, respectively [82]. They were higher for G70 than baseline diesel. Similarly, peaks at 1450 and 1375 cm⁻¹ represent $CH3$ and $CH,CH2$ bending. Diesel and G70 showed similar carbonyl group bond (C = O) stretching at 1742 cm⁻¹. Axial asymmetric deformation peak (O–C–C) was observed at 1195 cm⁻¹. Diesel contained more C–C bonds than G70. GCI engine lubricating oil showed higher bending vibrations of the aromatic compounds (C–H bond) at 875–825 cm⁻¹ [80]. Gasoline contains aromatics as octane number boosters. The absorption peaks at 810, 780, and 750 ± 20 cm⁻¹ correspond to the C–H bend of alkanes [82]. Diesel and gasoline contain high long-chain and short-chain alkanes, respectively. Short-chain alkanes have higher bending frequencies than long-chain alkanes due to the lower stiffness of C–H bonds [83]. Gasoline contains higher alkenes than baseline diesel, while diesel may contain alkenes as impurities. Peaks at 970, 890, 840–790, 730–665 cm⁻¹ correspond to the C = C bend of alkenes. Hence, the C = C bend was higher for G70 than for diesel. Halo components (850–550, 690–515, and 600–500 cm⁻¹) are impurities in the gasoline and diesel, reducing their stability. However, some fuel additives contain halogenated compounds. Lubricating oils show peaks at 985 and 654 cm⁻¹ corresponding to additives that were not observed in the test fuels. Overall, the GCI engine lubricating oil contains more C–H groups (∼1460, 1380, and 1400 cm⁻¹) and soot. These components could degrade the lubricating oil quality.

Figure 9 shows the proton spectra of unused and used lubricating oils. At 0–2 ppm spectral regions, dominant signals were observed. Lubricating oils contain a diverse range of alkyl chains (saturated and unsaturated) that exhibit variations in size and chemical composition [42].

The signals in the regions of 1.7–3.2 ppm (aliphatic), 3.2–4.4 ppm (oxygenates), 4.4–6.0 ppm (olefins, amines, and esters), and 6.4–8.8 (aromatics) were unseen in the lubricating oils. Signals in aliphatic and oxygenated species regions increase with lubricating oil aging [24]. ¹³C NMR determines the quaternary carbon atoms and functional groups, and ¹H NMR could not detect them [21]. ¹³C NMR signals were observed at 10–40 ppm (alkanes) (Fig. 10). However, it is challenging to obtain clear information on alkanes and cycloalkanes from ¹³C NMR [21]. The signals in the regions of 50–90 ppm (alcohols, esters, and alkyne), 100–150 ppm (aromatic), and 190–210 ppm (ketone) were undetected in the lubricating oil samples.

The intensity of an NMR signal corresponds to the number of nuclei contributing to the signal. The area under an NMR signal corresponds to the number of nuclei in a specific chemical environment [42]. Primary alkyl hydrogen-containing functional groups (5–22 ppm) were higher than secondary alkyl (20–30 ppm) counterparts for unused lubricating oil. Alkyl chains in lubricating oils usually differ in size and chemical structure from saturated to unsaturated carbon chains. Despite chemical changes in lubricating oil during usage, there will always be many $−CH3$ and $=CH2$ groups [24]. GCI and CDC engine lubricating oils showed higher –OH molecules (50–90 ppm) than unused lubricating oil [78]. Metals form metal oxides when exposed to oxygen for a prolonged period [84]. Figure 11 shows proton and carbon NMR for G70. More CH₂ and CH₃ protons (0.8–1.4 ppm) were observed than CH protons (1.4–2.1 ppm) in G70 (Table 3).

Significant aromatics (2.1–4.0 ppm) and aromatic ring protons (6.2–9.2 ppm) were observed in G70. Carbon NMR confirms the presence of C = C, CH, CH₂, and CH₃ in G70. Ether and alcohol carbon (50–90 ppm) were detected in G70 due to 12% ethanol in commercial gasoline used for blending. Finally, no significant changes in hydrogen and carbon functional groups were observed between used and unused lubricating oils from the two engines.

ICP-MS multi-element technique analyzes the trace elements in the lubricating oils [85]. Trace metals are low in new lubricating oils and increase with usage. Around 26 elements were investigated, and 16 were quantified (Fig. 12).

Ca, Zn, and P traces were majorly observed in the lubricating oils. These trace elements were 27–34% and 2–4% higher for CDC and GCI engine lubricating oils, respectively, than unused lubricating oil. Mg and Ca are present in detergent additives, alloys, wear particles, dust, and water-based contaminants [86]. P is present in the anti-wear additive (ZDDP) and phosphate ester. Brass, ZDDP additives, and filter canisters contain Zn traces. Lubricating oil decays due to thermal effects and can reduce P, Zn, and Ca trace concentrations. Increasing fuel dilution can further reduce their trace concentrations. Fe, Na, and Al trace elements were 228, 55.5, and 179% higher in CDC engine lubricating oil than unused lubricating oil, whereas they were 92, 504, and 74% higher in GCI engine lubricating oil. High Na traces in the GCI engine lubricating oil infer that it comes from G70 (fuel) via fuel dilution. Cu traces were 19.6 and 7.69 times higher for CDC and GCI engine lubricating oils, respectively, than the unused lubricating oil. Fe and Cu traces were higher for CDC engine lubricating oil than for GCI engine lubricating oil. Pb, Sr, and Cr traces increased by 48, 2.3, and 17 times in CDC engine lubricating oil than unused lubricating oil, whereas they increased by 26, 0.25, and 3.36 times for the GCI engine lubricating oil.

Wear metals (Fe, Cu, Cr, and Pb) trace concentrations increased with engine runtime [86,87]. K, B, Cr, and P traces are generally associated with antifreeze additives ingress in the lubricating oils. Minor amounts of Rb, Mn, Ba, and Bi trace were detected in the lubricating oils. Mn is a metallic element, whereas Ba traces are present in fuel detergent additives and grease. Pb, Cr, and Bi are hazardous elements, observed to be ∼63.6 and 30.3 ng/g of lubricating oil in the CDC and GCI engine lubricating oils, respectively. CDC engine lubricating oil contains higher trace elements than GCI engine lubricating oil. Fuel dilution of the lubricating oil could reduce the trace metals in the lubricating oil. Higher concentrations of trace metals in the lubricating oils can increase engine wear and PM emissions.

The soot aggregates were between 140–650 nm in size and had a modest branched structure (Fig. 13). The average aggregate width and length were 130 and 500 nm, respectively, while the maximum width and length were 200 and 565 nm, respectively (for GCI soot in the lubricating oil).

Apart from chain aggregates, compact aggregates of ∼650 nm were observed. Aggregates length-to-width (L/W) ratio varied between 1.1 to 1.51 and 1.0 to 2.16 for CDC and GCI engines, respectively. Skeleton width $(Wsk)$ and skeleton length $(Lsk)$ give more information about the soot shape, especially for chain-like aggregates [88]. The $Lsk/Wsk$ ratio was between 2 and 10 for the GCI engine, which was higher than the CDC engine. Agglomerates are defined as a cluster when L/W < 2.5; hence, most soot falls into this group. Whenever the skeletal ratio $(Lsk/Wsk)$ is used, more aggregates can be classified as chains.

The aggregate perimeter versus area is shown in Fig. 14(a). More small-size aggregates were observed in CDC engine lubricating oil than in GCI engine lubricating oil. The soot aggregate perimeter for CDC and GCI engine lubricating oils were ∼2000nm and 2500 nm, respectively. The high combustion temperature of CDC emits more small-size particles, whereas low-temperature combustion regions in GCI produce more chain-like aggregates [46]. The number of primary particles per aggregate (N) is plotted against the corresponding normalized radius of gyration $(2Rg/dp)$ [89], in Fig. 14(b). Slope and intercept of logarithmic $2Rg/dp$ and N provide $Df$ and $Kf$ (fractal prefactor) [46]. The soot $Df$ was observed to be between 1.1 to 1.9 (accepted universal values) and lower $Df$ represents chain-like aggregates [89]. Soot in the GCI and CDC engine lubricating oils showed 1.21 and 1.15 $Df$⁠, respectively. A high $Kf$ indicated a higher number of primary particles and a larger overall size of soot aggregates [29].

The $dp$ distribution of soot from CDC and GCI engine lubricating oils is shown in Fig. 14. For the CDC engine soot in the lubricating oil, the minimum and maximum $dp$ were 19.6 and 46.7 nm, respectively. These were 18.6 and 49 nm for GCI engine soot in the lubricating oil. Rocca et al. [43] reported that primary particle sizes extracted from the lubricating oil range from 20.2 to 35 nm. With increasing in-cylinder temperature and oxygen, mean $dp$ decreases [88]. The soot in the CDC engine lubricating oil showed smaller primary particles than the primary particles in the soot from GCI engine lubricating oil. Molecular (aromatic) structures influenced the soot formation in direct-injection PPC engines [90]. GCI engine lubricating oil contained more soot, which increased its viscosity, requiring more frequent oil changes.

In-cylinder pressure and temperatures were slightly lower for GCI than the CDC mode engine (Fig. 15(a)). Higher latent heat of vaporization of gasoline decreases the intake charge temperature. The chamber temperature was ∼330, 620, and 750 K at 100, 30, and 14 °bTDC injection timings. Iso-octane's critical temperature is ∼550 K, leading to a longer spray for the pilot injections. Figure 15(b) shows that the gasoline spray hits the cylinder walls and is converted into vapors. Le Coz et al. [91] reported gasoline spray penetrated up to 7 cm at 700 K and 11 bar chamber conditions. Ray et al. [49] reported diesel liquid and vapor penetration lengths lie within 4 cm. Fuel trapped in the squish region was negligible in the CDC engine due to the absence of pilot injection. Moreover, n-heptane exhibited lower liquid spray mass due to higher temperature and fuel-rich conditions.

Observations were made for in-cylinder fuel mass fractions at different crank angles, starting after the start of injection up to the TDC. Fuel from pilot-1 strikes the cylinder walls, causing the wall-wetting (at 89 °bTDC). Fuel from Pilot-2 injection hits the piston lip and splits into two parts: squish and bowl regions. A portion of the fuel–air mixture trapped in the squish region moves to the bowl region during the compression stroke. A significant fraction of fuel in the squish and crevice regions can undergo incomplete combustion [26]. Fuel from the main injection enters the bowl region and forms a richer fuel–air mixture. The GCI mode engine's spray droplet radius is 10–20 microns. Yan et al. [92] reported Sauter mean diameter (SMD) of iso-octane was between 5 and 12.5 microns. Adding n-heptane to iso-octane increased the SMD of spray droplets and spray droplet velocity. In CDC mode, the fuel was targeted into the bowl region at ∼800 K gas environment, which underwent dominant diffusion combustion. It was observed that n-heptane did not evaporate fully at the TDC position. More fuel was trapped in the piston bowl region, and subsequent combustion led to higher CO emissions [26].

This study investigated the effects of GCI and CDC mode engines on particulate emissions and lubricating oil deterioration. Advanced fuel injection timings led to wall-wetting and interaction with the cylinder liner and the lubricating oil. GCI engine burns more lubricating oil, resulting in higher particulate emissions. GCI mode engine exhibited 97–99.7% and 4.5–58% higher NPs and NMPs than the baseline CDC mode engine. AMPs increased with the engine load and speed. A small fraction of fuel mixed with the lubricating oil causes crankcase dilution, reducing the lubricating oil's viscosity. Particulates also interact with lubricating oil, increasing its viscosity. GCI engine lubricating oil showed 8.78%, 7.46%, and 14% lower viscosity, ash, and carbon residue than CDC engine lubricating oil. Unused lubricating oil showed higher base oil and additive components, whereas used lubricating oils contained more diesel molecules than gasoline. Due to their chemical characteristics, gasoline molecules evaporated or reacted with lubricated oils. CDC engine lubricating oil showed higher base oil components than GCI engine lubricating oil. FTIR analysis revealed that hydroxide or moisture was lower in CDC engine lubricating oil than in GCI engine lubricating oil. NMR showed that primary alkyl hydrogens were 57.9% and 68.34% higher than secondary alkyl hydrogens for GCI and CDC engine lubricating oils. GCI and CDC engines showed dominant clusters and chain-like aggregates in the TEM images of soot in the lubricating oils. Larger primary particles and fractal dimensions were observed in the soot from the GCI engine lubricating oil than from the CDC engine lubricating oil. Sixteen trace elements were detected in the lubricating oil samples. GCI engine lubricating oil showed lower (∼ 20 mg/kg of oil) trace metals than the CDC engine lubricating oil. GCI engine showed a longer fuel spray penetration length due to lower in-cylinder temperature, resulting in the possibility of wall-wetting by the fuel sprays, which will adversely affect the lubricating oil, as seen in the oil characterization studies. Therefore, GCI combustion engines would either require special lubricating oil formulations to address these issues or more frequent oil changes.

This work is supported by the J C Bose Fellowship by the Science & Engineering Research Board, Government of India (Grant No. EMR/2019/000920) and SBI endowed Chair Professorship from the State Bank of India to Professor Avinash Kumar Agarwal. Financial support from the SERB, under the National Postdoctoral Fellowship scheme to Dr M. Krishnamoorthi (PDF/2021/001209) is acknowledged. The authors are grateful to the Department of Chemistry, Center for Environmental Science and Engineering, and the Advanced Imaging Centre of IIT Kanpur for allowing us to use their sophisticated analytical facilities. The authors gratefully acknowledge the assistance of Harsimran Singh, Abhijit Saha, Vaibhav Singh, Ankur Kalwar, Sam Joe, and Vasudev Chaudhari in this work. The assistance of the Engine Research Laboratory staff in conducting the exhaustive series of experiments is gratefully acknowledged.

There are no conflicts of interest. This article does not include research in which human participants were involved. Informed consent was obtained for all individuals. Documentation is provided upon request. This article does not include any research in which animal participants were involved.

The datasets generated and supporting the findings of this article are obtainable from the corresponding author upon reasonable request.

Figures 16 and 17.