24th International Colloquium Tribology - January 2024 173 Tribochemical Properties of Glycerol as a Green Lubricant on Ferrous Substrates: Atomic-scale Study by Reactive Molecular Dynamics Simulation Vahid Fadaei Naeini 1,2* , J. Andreas Larsson 1 , Roland Larsson 2 1 Applied Physics, Division of Materials Science, Department of Engineering Sciences and Mathematics, Luleå University of Technology, Sweden 2 Machine Elements, Division of Machine Elements, Department of Engineering Sciences and Mathematics, Luleå University of Technology, Sweden * Corresponding author: vahid.fadaei.naeini@ltu.se 1. Introduction Annually, the leakage of oil-based lubricants into the environment presents significant ecological hazards [1] . When industrial lubricants spill, they can contaminate extensive areas, especially affecting water bodies, leading to widespread environmental damage [2] . The increasing demand for eco-friendly lubricants has made vegetable oils popular, despite their limitations in heat resistance and viscosity range. [3] . Glycerol, as a green and biocompatible base fluid, demonstrate better physical properties at cooler temperatures positioning it as a promising green substitute for vegetable oils. Recent research has delved into the complexities of tribofilm formation and its role in elastohydrodynamic lubrication. Zhang et al. examined the mechanochemical behaviors of zinc dialkyldithiophosphate (ZDDP) in full-film EHL, focusing on the critical influence of shear stress [4]. In this study, we present a computational framework based on reactive NEMD simulations, designed to enhance understanding of the tribochemical interactions of glycerol. 2. Computational Methods and System Setup 2.1 Stress-Augmented Thermally Activated Process Recent advances in nanoscale experiments, especially those involving asperities, have offered new perspectives for studying the single lubricant molecules [5]. Gosvami et al. employed atomic force microscopy (AFM) to investigate the growth rate of ZDDP tribofilms on aluminum-based interfaces, discovering that the formation rate is concurrently affected by temperature (T) and normal stress (σ). [6]. The tribofilm growth observed in these experiments follows a stress-augmented thermally activated (SATA) process described by a modified Arrhenius equation that is widely known as the Bell model: (1) In this equation, A represents the pre-exponential factor, E a the activation energy of the reaction, ∆V ∗ the activation volume, and k B the Boltzmann constant. This study aims to determine the mechanochemical factors that control the growth of glycerol lubricant films, using data from glycerol dissociation and the Bell model. 2.2 Simulation Method and System Setup Molecular dynamics (MD) simulations using the ReaxFF force field provide a reliable computational approach for studying physical and mechanochemical properties of lubricants, particularly excelling in simulating chemical reactions like bond breaking and formation. Its functional form, expanded by Aktulga et al. [7] , is expressed as: E ReaxFF = E bond + E angle + E tors. + E vdW + E Coulomb + E specific + E 0 (2) where E bond denotes the energy of bond formation, E angle and E tors. correspond to energies of three-body angles and four-body dihedrals. E Coulomb and E vdW represent electrostatic and van der Waals energies. E Specific covers system-specific properties like hydrogen bonding, while E over. prevents atom over-coordination. The dynamic atomic partial charges in the simulation are calculated using the charge equilibration method [7] . This study employs the ReaxFF parameters by Khajeh et al. for systems containing C/ H/ O/ Fe [8] . To evaluate the surface impact on tribochemical properties of glycerol, a layer containing 333 intact glycerol molecules was placed between α-Fe(110) and amorphous functionalized magnetite substrates (a-Fe 3 O 4 -OH), individually. The magnetite substrate was functionalized with hydroxyl groups. The simulation box measured 54 Å × 54 Å × 89 Å along the X, Y, and Z axes. After setting up the box, the system was equilibrated at T=300 K for 0.1 nanoseconds (ns), using the Nosé- Hoover thermostat [9] to be published. For the production run, the temperature was varied between 300-500 K and the normal pressure was increased from 1 to 4 GPa, as indicated in Figure 1. Figure 1. The initial configuration of the system To achieve the desired normal stress, a constant force was applied to the outermost atomic layers of each slab using steered MD simulation with a constant force. The sliding velocity was achieved by applying equal and opposite velocities (±5 m/ s) to the outermost layers of atoms along the x-axis. The heating, compression, and shear starin were all carried out simultaneously during the 1 ns production run using the LAM- MPS package [10], with a simulation timestep of 0.25 fs. 3. Results and Discussion Figure 2 displays the time evolutions of the number of intact glycerol molecules confined between a-Fe 3 O 4 -OH and 174 24th International Colloquium Tribology - January 2024 Tribochemical Properties of Glycerol as a Green Lubricant on Ferrous Substrates: Atomic-scale Study by Reactive Molecular Dynamics Simulation α-Fe(110) surfaces at T=300 K under various loading conditions. Numerical assessments show that glycerol dissociation between magnetite substrates adheres to a second-order reaction, while the decline in glycerol molecules between α-Fe slabs indicates a first-order reaction. Figure 2. The time evolution of the number of intact glycerol molecules for pure glycerol confined between: a. a-Fe 3 O 4 -OH surfaces and b. α-Fe(110) at T=300 K under various loading s. As depicted in Figure 3, we applied 3D surface fitting to the dissociation rate data of glycerol between a-Fe 3 O 4 -OH and α-Fe slabs to analyze the temperature and stress effects on reactivity. Figure 3. Panels a and b illustrate the dissociation rate alter-ations by temperature and shear stress for glycerol confined between a. a-Fe 3 O 4 -OH and b.α-Fe slabs, respectively. The approach depicted in Figure 3 paved the way to determine the key parameters of the Bell model like ΔV * , ln(A), and E a , as listed in Table 1. Table 1.The values ΔV*, ln(A), and Ea for glycerol confined between a-Fe3O4-OH and α-Fe(110) slabs Ea (kJ/ mol.) ln A (s-1) ΔV* (Å3) Gly._a-Fe3O4-OH 18.02±4.21 -3.75±0.40 8.71±0.49 Gly.-α-Fe(110) 8.27±1.67 3.78±0.75 12.21±2.33 4. Concluding Remarks In this study, molecular dynamics (MD) simulations with the ReaxFF reactive force field were used to study the tribochemical properties of glycerol between typical ferrous surfaces. The research focused on determining the decomposition rate of glycerol molecules and the mechanochemical factors affecting tribofilm growth. This computational method allows for validating results and assessing the performance of conventional lubricants under operational conditions. References [1] P. Nowak, K. Kucharska, and M. Kamiński, “Ecological and health effects of lubricant oils emitted into the environment,” Int. J. Environ. Res. Public Health, vol. 16, no. 16, p. 3002, 2019. [2] L. Spilsbury and R. Spilsbury, The oil industry. The Rosen Publishing Group, Inc, 2011. [3] M. A. Dandan, S. Samion, N. F. Azman, F. Mohd Zawawi, M. K. Abdul Hamid, and M. N. Musa, “Performance of polymeric viscosity improver as bio-lubricant additives,” Int. J. Struct. Integr., vol. 10, no. 5, pp. 634-643, 2019. [4] J. Zhang, J. P. Ewen, M. Ueda, J. S. S. Wong, and H. A. Spikes, “Mechanochemistry of zinc dialkyldithiophosphate on steel surfaces under elastohydrodynamic lubrication conditions,” ACS Appl. Mater. Interfaces, vol. 12, no. 5, pp. 6662-6676, 2020. [5] L. O. Laier et al., “Surface-dependent properties of α-Ag 2 WO 4: a joint experimental and theoretical investigation,” Theor. Chem. Acc., vol. 139, pp. 1-11, 2020. [6] N. N. Gosvami, J. A. Bares, F. Mangolini, A. R. Konicek, D. G. Yablon, and R. W. Carpick, “Mechanisms of antiwear tribofilm growth revealed in situ by single-asperity sliding contacts,” Science (80-. )., vol. 348, no. 6230, pp. 102-106, 2015. [7] H. M. Aktulga, J. C. Fogarty, S. A. Pandit, and A. Y. Grama, “Parallel reactive molecular dynamics: Numerical methods and algorithmic techniques,” Parallel Comput., vol. 38, no. 4-5, pp. 245-259, 2012. [8] A. Khajeh et al., “Statistical analysis of tri-cresyl phosphate conversion on an iron oxide surface using reactive molecular dynamics simulations,” J. Phys. Chem. C, vol. 123, no. 20, pp. 12886-12893, 2019. [9] S. Nosé, “A unified formulation of the constant temperature molecular dynamics methods,” J. Chem. Phys., vol. 81, no. 1, pp. 511-519, 1984, doi: 10.1063/ 1.447334. [10] S. Plimpton and L. S. National, “Fast Parallel Algorithms for Short-Range Molecular Dynamics,” J. Comput. Phys., vol. 117, no. June 1994, pp. 1-42, 1995.