The detail of staining technique and observation equipment were reported in reference Tokoroyama et al. (2024), the brief and important procedure of staining the silica particles is summarized below. The fluorescent staining of silica particles was accomplished using a combination of a silane coupling agent (KBM-4803 from Shin-Etsu Silicone, Japan) and Rhodamine B (from Fujifilm Wako Pure Chemical Corporation, Japan). These silica particles came in three sizes, with average diameters of 0.8, 1.0, and 3.0 μm (supplied by Potters Ballotine Co., Ltd., Japan).

Friction tests were carried out using an in situ observation system called “iFLAT” (Intelligent Fluorescent Light Assisted Tribometer, which equipment can be allowed to use 1.0 N as limitation of normal load, 5.0 mm/s as maximum of sliding speed, and 2,000 fps as frame rate). A schematic of this system is depicted in Figure 1A, with an enlarged view of the observation area shown in Figure 1B. The iFLAT system is equipped with a 532 nm laser source (GLK32200, LASOS, Germany). This laser travels through a light guide from the source, enters an objective lens of an optical microscope, and is directed to the friction area via a sapphire and a quartz disk (1.0 mm thickness, 50 mm diameter, with a surface roughness of approximately 2.0 nmRa, respectively). These transparent materials are used for the visualization of the in situ observation of the surface, and their surface roughness is smaller than that of actual sliding surfaces (such as honing or coating surfaces). However, by changing the surface roughness of the counterpart material’s ball, it will be necessary in the future to clarify the relationship between roughness and particle entry. A fixed SUJ2 bearing ball (approximately 8.0 mm in diameter, with a surface roughness of approximately 7.0 nmRa, it is same property of AISI53200 stainless steel) within a ball holder is placed on a friction cantilever. A dead weight is used to apply a normal load of 0.2 N. The bearing ball reflects the incident laser, and the fluorescently stained particles emit fluorescent light. Both sets of light pass through the sapphire disk before reaching a half-mirror. Subsequently, both lights are directed to a WIG filter, which is a long-pass filter with a cutoff wavelength at 580 nm, effectively eliminating reflected laser light. Finally, the fluorescent light is captured by a CMOS camera, operating at a frame rate of 500 frames per second (fps). The in situ observations were conducted multiple times, and we selected 60 frames (0.12 s) that recorded the highest particle counts within the observation area during the observations. The summary of those conditions are listed in Table 1.

The friction test involved the interaction between an SUJ2 ball and a sapphire or a quartz disk lubricated with PAO8 lubricant, which possesses a viscosity of 46.0 mPs s at 40°C. The i-FLAT system was utilized for this test. In the previous report (Tokoroyama et al., 2024), an outside oil film (Hi) was formed by applying a load to the lubricating oil placed between two disks and sliding the two disks to achieve a film of a certain thickness. However, there were issues with the controllability and reproducibility of the oil film thickness. Therefore, a uniform oil film was formed using a spin coater (ACT-220AII). When using a spin coater, the theoretical oil film thickness is calculated by the following Equation 1 (Alfred et al., 1958): h = h 0 1 + 4 ω 2 h 0 2 t 3 ν (1) here, h is the oil film thickness, h ₀ is the initial film thickness, ω is the rotational speed, t is the rotational time, and ν is the kinematic viscosity (Emslie et al., 1958). An oil film was formed using PAO8 at different rotational speeds and times, and the oil film thickness was measured using a reflection spectroscopic film thickness meter. The results are shown in Figure 2. The calibration curve was created using a power approximation so that it would be a function of time t in the theoretical formula, and the power values of t ranged from −0.427 to −0.567, which were close to the theoretical value of −0.5. The experimental Equations 2–4 obtained from the experiment is shown below. h 3000 rpm = 21177 t − 0.427 (2) h 5000 rpm = 13935 t − 0.469 (3) h 7000 rpm = 14202 t − 0.567 (4)

The sapphire disk with the lubricant was secured onto the disk holder, with the ball specimen attached to it. A dead weight of 0.2 N was added to the ball specimen. Fluorescently stained particles, comprising a colloidal liquid of ethanol with a volume of 0.1 μL (assuming that the stained particles are Uniformly dispersed in the solvent, the number of particles contained in 0.1 μL is 20,000 when the diameter is 0.8 μm), were applied onto the oil film, positioned at the same radius as the friction test rotation, just ahead of the contact point. Prior to commencing the in situ friction test, several minutes were allowed to pass to account for ethanol vaporization. The sliding speed was set at 5.0 mm/s (with a rotation radius of approximately 5.0 mm), and the camera operated at a speed of 500 frames per second (fps). We captured a continuous sequence of 60 frames (0.12 s) to tally the number of particles when the highest concentration of particles was observed near the contact point. The sliding distance covered in these 60 frames was approximately 0.6 mm.

A friction test was conducted for 5 s until the fluorescent-stained particles were trapped between the contact surfaces, and the area of deposited particles was measured. Since the particles present on the area emit light due to Rhodamine B, the observed images were converted into 256-level grayscale images, and binarization was performed to calculate the area. A threshold value of 60 was used in this process. The projected area created by particles with a diameter of 0.8 μm on the observation area was approximately 0.5 mm², and even with adjustments of ±2 levels around 60, the particle area fluctuation was less than 1 mm², so this value was adopted.

To identify the position where the lubricant generates vortices that cause particles to detach from the disk surface, the flow generated within the thin lubricant film was visualized using Navier-Stokes method simulations. Quartz was used as the disk material, and PAO8 was used as the lubricant. The contact angle between the lubricant and the quartz was set to 11° (the value measured using a contact angle meter). The movement speed of the disk was 5 mm/s, the same as in the experiment, and in the simulation, the lubricant and other test specimens were set to room temperature (23°C).

In the Navier-Stokes simulation (simulations were performed using Ansys Fluent 2020 R1 (ANSYS, Inc., US).), a contact region as shown in Figure 3 was prepared. The outside oil film thickness was set to values (Hi = 0.4 as half of particles, 1.2 as three times thicker than 0.4 μm and it is within the value of 1 ≤ r ≤ 2, and 2.0 μm as five times thicker than 0.4 μm and it is the value of r > 2). In the figure, i) represents the Hertz contact circle generated between the contacting ball and disk, ii) is a cross-section divided at the center of the Hertz contact circle by the incoming lubricating oil, assuming that the flow occurring in the oil film is symmetrically present. The cross-section of the incoming lubricating oil is represented by iii), and iv) is assumed to symmetrically exist as a pressure outlet. The interface between air and oil was defined as VOF (volume of fluid) = 0.5, assuming that each medium exists in equal amounts.

To clarify the effect of outside oil film thickness on the number of particles entering between contact surfaces, friction tests were conducted using glass particles with a diameter of approximately 0.8 μm and employing sapphire disks and quartz disks as counter materials. Figure 4 shows the results of experiments conducted under dry conditions after the solvent dropped before the observation area had dried. The areas covered by particles deposited on the surface of the sapphire disk (Figure 4A) and the quartz disk (Figure 4B) were almost the same, and the numerical results of the areas (Figure 4C) also yielded nearly identical values. The results for when the outside oil film thickness was set to 0.8 μm are shown in Figure 5. It is evident that the number of particles deposited is higher on the sapphire disk and very low on the quartz disk. Next, the results for when the outside oil film thickness was set to 1.4 μm are shown in Figure 6. There was a tendency for the number of particles deposited on the quartz disk to increase compared to the 0.8 μm case. Finally, the results for when the outside oil film thickness was set to 5.0 μm are shown in Figure 7. The number of particles deposited decreased on both the sapphire and quartz surfaces. The relationship between the measured number of particles and the outside oil film thickness is shown in Figure 8. For both the sapphire disk and the quartz disk, the number of deposited particles was the same under conditions without lubricating oil, and the number of particles decreased as the outside oil film thickness increased. However, the number of particles deposited on the quartz disk was consistently lower than on the sapphire disk. From these results, it is shown that the number of deposited particles varies depending on the disk material.

It is believed that the flow field generated near the contact point affects the particle’s movement speed and retention near the contact point. Since the retention time is considered to be influenced by the rotation speed of the disk specimen, experiments were conducted by reducing the sliding speed from 5 mm/s used in the previous tests to 0.5 mm/s. Additionally, to clarify the effect of normal load acting between the two contact surfaces on the number of particle deposits, experiments were conducted by increasing the normal load to 1.0 N. The results are shown in Figures 9, 10, respectively.

By reducing the rotation speed of the disk from 5.0 mm/s to 0.5 mm/s, the number of particles deposited between the contact surfaces increased significantly compared to Figure 6. Approximately twice as many particles were deposited on the sapphire disk, and about 1.5 times as many particles were deposited on the quartz disk. From these results, it can be concluded that the flow generated near the contact surface affects particle deposition.

Next, Figure 10 shows the observation results when the normal load was increased from 0.2 N to 1.0 N under a sliding speed of 5.0 mm/s and an outside oil film thickness of 5.0 μm. Compared to Figure 7, where a similar experiment was conducted at 0.2 N, there are almost small number of particles present between the contact surfaces, similar to the results shown in Figure 7. From these results, it can be concluded that under boundary lubrication conditions, the normal load applied between the contact surfaces does not affect the number of particles entering.

As previously mentioned, it is evident that the number of particles deposited near the contact point varies depending on the material of the disk specimens used and the small outside oil film thickness, using particles with a diameter of 0.8 μm. The different amounts of particle deposition may be due to the inherent physical properties of the disk materials (e.g., van der Waals forces or wettability).

However, it is not clear to what extent these physical properties affect the flow near the contact point. Therefore, particles with a diameter of 0.02 μm were used to observe the flow on the surfaces of sapphire and quartz disks. The results are shown in Figure 11. The yellow circles in the figure indicate positions where the geometric gap between the SUJ2 ball and the disk surface is 0.8 μm. Calculating the particle movement speed on each disk surface from the subtraction between frames, the speed was approximately 5.0 mm/s, which is similar to the disk speed in the sliding direction, indicating that the flow speed is almost the same regardless of the disk material.

The flow near the contact point is unaffected by the disk material, but the amount of particle deposition near the contact point shows the influence of the disk material. Therefore, it is inferred that van der Waals forces, as inherent material properties, play a role (Regarding wettability mentioned earlier, the contact angles of quartz and sapphire with PAO8 are both approximately 11°, showing no difference, thus having no effect.) Nikas et al. has demonstrated through numerical analysis that frictional forces significantly contribute to the trapping of particles between two sliding surfaces, and that for very small particles below 5 μm, the influence of van der Waals forces on particle behavior becomes comparable to fluid forces (Nikas, 2007; Nikas, 2020). Therefore, assuming forces other than the fluid pushing against the deposited particles, Figure 12 shows the geometrical model of the forces acting on a particle when it is trapped. T is the frictional force, N is the normal force, F is the fluid force, G is the van der Waals force acting between the disk and the particle, and θ is the angle at which the disk and the particle contact at a geometric gap of 0.8 μm. To investigate the change in the number of particles deposited by the disk, the van der Waals force represented by G is assumed to only affect when the distance between the particle and the disk is very close, and it is assumed to act only between the disk and the particle. Under these conditions, it is considered that the particle will be trapped when the sum of the frictional forces T ₁ and T ₂ is greater than the reverse fluid force F. When a particle is trapped between the sphere and the disk, it is considered that the particle will move into the contact area when the total frictional force of the particles deposited near the contact surface exceeds the horizontally decomposed force of the load.

Here, we use the following Equations 5–7 described in the report by Nikas et al., F x = 3 π η 0 u 1 + u 2 2 h c − π 4 d − h c (5) G = − H d 12 s 2 1 1 + d 2.97 σ + 1 1 + 1.485 σ s 2 (6) H ≅ h d − h f h p − h f (7)

Fx is the fluid force in the sliding direction generated near the contact area, η ₀ is the viscosity of the lubricating oil, u ₁ and u ₂ are the velocities of the sliding surfaces, h _c is the minimum gap at the contact area, d is the particle diameter, G is the van der Waals force acting between the particle and the disk, H is the Hamaker constant, and s is the minimum distance between the particle and the disk, and σ is the surface roughness. The h _d , h _f , and h _p are Hamaker constant of each element as disk, fluid, and particle. To calculate the forces acting on the particles, we substituted the values shown in Table 2 for the Hamaker constant, lubricating oil viscosity, and other parameters of the test specimens used in the experiment. As a result, the forces Fx and G moving the particles within the contact surface were Fx = 0.64 nN, G _sapphire = 0.18 nN, and G _quarts = 0.077 nN, respectively. These results indicate that sapphire disk has higher affection for particles and the force is also same order of fluid force.

The flow direction of lubricating oil containing particles at different outside oil film thicknesses was calculated. The results of averaging the velocity vectors in the vertical direction (defined as the y-axis) within the contact surface, accompanying the entry of lubricating oil into the contact surface at an outside oil film thickness Hi = 0.4 μm, are shown in Figure 13A. An air-liquid interface is formed between the two surfaces in contact at a position approximately 95 μm (x = 95 μm) from the centre of the Hertz contact. Assuming that the likelihood of particle entry increases when particles move towards the disc side due to downward flow in the y-axis direction, the flow is strongest downwards around x = 93 μm. However, as will be discussed later, since the geometric distance between the contacting surfaces equals the particle diameter of 0.8 μm at x = 160 μm, in the case of Hi = 0.4 μm, particles with a diameter of 0.8 μm first contact the surface of the SUJ2 ball. Despite the presence of lubricating oil, particles on the disc are assumed to contact the SUJ2 ball in nearly dry conditions, resulting in a similar amount of particle deposition around Hi = 0.4 μm regardless of the change in disc material, as shown in Figure 8.

Next, the results for an outside oil film thickness Hi = 1.2 μm are shown in Figure 13B. The geometric distance between the contacting surfaces equals the particle diameter of 0.8 μm at x = 160 μm. Particles near the disc surface ride the upward flow in the y-axis direction at x = 178 μm, but are influenced by the downward flow in the y-axis direction in the range of x = 159–174 μm. This downward flow just before particles are geometrically trapped is assumed to push the particles against the disc surface. This result is thought to affect the amount of particle deposition and is consistent with the observation that the number of SiO₂ particles deposited peaks around an outside oil film thickness of 1.4 μm, as shown in Figure 8.

Finally, for an outside oil film thickness Hi = 2.0 μm as shown in Figure 13C, an upward flow in the y-axis direction is generated at x = 205 μm where the air-liquid interface is formed between the contacting surfaces, suggesting that particles are drawn away from the disc surface. Additionally, an upward flow in the y-axis direction occurs again around x = 160 μm where the geometric gap equals the particle diameter, making it difficult for particles to remain.

These results align with the authors’ previous findings (Tokoroyama et al., 2024) as shown in Figure 14, which identified the generation of vortices as the cause of a peak in the number of particles entering within the range of 1 ≤ r ≤ 2 times the particle diameter, when the outside oil film thickness reaches an extremum.

The results obtained in this study demonstrate how the van der Waals forces and the components that cause particles to separate from the incident flow act when using sapphire and glass disks. However, the effects of the incident flow, the shape, the hardness of the materials, and the hardness of the particles formed on the surfaces of the two contacting faces have not yet been considered. In the future, it will be necessary to widely apply observation techniques to anisotropic particles, ductile metal particles, and biomaterials, making it possible to design structures that suppress particle penetration. Furthermore, the wear particle problems such as human joint which is suffered by wear particles which cause undesirable reaction due to physically and chemically reaction in the body (Kovochich et al., 2018), cell-accelerated corrosion due to metal alloy wear debris (Bijukumar et al., 2020), immune toxicity caused by wear particles (Gill et al., 2012) are the formidable problems to be solved in near future. So, development of observing method of those particles or collecting them (Tokoroyama et al., 2018) should be achieved by several perspectives.