The structure of the adsorbed additive film formed on the Cu surface in a static state was evaluated by neutron reflectometry. Since hydrogen and deuterium are clearly distinguishable in neutron reflectometry, the thickness and film density of the adsorbed additive film can be evaluated by converting the hydrogen atoms in the additive to deuterium atoms (Hirayama et al., 2012; Hirayama and Yamashita, 2020). This was done using a lubricant with deuterated stearic acid (C₁₇D₃₅COOH) or deuterated stearyl alcohol (C₁₈D₃₇OH) with a 3 mM concentration, which corresponds to approximately 0.1 wt% in terms of the non-deuterated additive in PAO (30 mm²/s@40°C, Idemitsu Kosan). A mirror-polished Si block (50 × 50 mm, t10 mm) was prepared, and a Cu film was deposited on its surface using an ion beam sputtering system (KUR-IBS) to a 30 nm thickness (Hino et al., 2015).

Neutron reflectivity measurements were performed using the SOFIA time-of-flight neutron reflectometer in Beamline 16 at the Materials and Life Science Experimental Facility of the Japan Proton Accelerator Research Complex (Yamada et al., 2011; Mitamura et al., 2013). Figure 2 schematically illustrates the neutron reflectometry measurement. The theoretical value of the scattering length density (SLD), which is the coherent nuclear scattering length multiplied by the number density of the constituent atoms, is also shown in the figure. First, a Si block with a Cu film was placed in a holder, and the neutron reflectivity was measured while the Cu surface was immersed in PAO base oil. The base oil was then replaced with PAO containing deuterated stearic acid or deuterated stearyl alcohol. The reflectivity was measured again after 12 h at room temperature. The thickness of the Cu film and the thickness and film density of the adsorbed molecular film formed by each additive were evaluated by fitting to the reflectivity profiles obtained from each measurement using analysis software (Nelson, 2006).

AFM observations were first carried out to check that the proposed line-and-space pattern had formed correctly in the Cu film. The observations were carried out using an AFM (SPM-9700HT, Shimadzu) in dynamic mode in air. The maximum operating range of this AFM is 10 μm in the XY plane and 1 μm in the Z direction. The AFM exhibits a resolution of 0.2 nm in the XY plane and 0.01 nm in the Z direction. A point probe type Si cantilever (TAP300AL-G, BudgetSensors; spring constant = 40 N/m, tip radius <10 nm) was used.

A schematic illustration of the simultaneous measurement of friction and gap at the metal–lubricant interface using a specimen with a Cu line-and-space pattern is shown in Figure 3. A sphere-probe-type cantilever made of Si (SD-Sphere-CONT-L, NANOSENSORS), which typically has a 2 µm tip radius, was used for this contact mode AFM observation. Before the experiment, the cantilever was rinsed with specially developed HF for 1 min to inactivate its surface. A 3 µm wide Cu line was formed in the center of the 7.5 µm observation area, and observation was carried out in 50 µL PAO base oil. The applied loads were 25, 75, and 125 nN, with the maximum Hertzian contact pressures on the Cu film corresponding to 200, 280, and 330 MPa, respectively. Next, 50 µL of PAO containing 0.2 wt% of stearic acid or stearyl alcohol was supplied to the same substrate. The lubricant was gently stirred with a pipette, and the concentration of the additive in the lubricant on the substrate was adjusted to 0.1 wt%, after which the same measurements were carried out as for PAO base oil. For the simultaneous measurement of friction and gap at the metal–lubricant interface, 15 successive observations were performed at a load of 75 nN. From each obtained image, the average friction coefficients on the Si block and the Cu film and the step height from the Si reference plane to the Cu line were evaluated.

The AFM images were analyzed using Gwyddion open-source software for AFM data analysis (Nečas and Klapetek, 2012). The height distribution in each image was calculated, and the mean step heights from Si to Cu were found by Gaussian fitting to each height distribution peak. The friction coefficient was obtained by dividing the friction force mechanically calculated from the detected cantilever torsion by the total load of the pushing force calculated from the cantilever deflection and the adhesion force detected in the force curve measurement.

The thickness and density of the adsorbed molecular film on the Cu thin film were evaluated by neutron reflectometry. The reflectivity profiles obtained from the measurements are shown in Figure 4. Scattering vector Q _z (x-axis) is expressed as the following Equation 1. Q z = 4 π / λ   sin ⁡ θ , (1) where λ is the wavelength of the neutron beam and θ is the angle of incidence (Mitamura et al., 2013).

The SLD profiles obtained by fitting analysis for each reflectivity profile are shown in Figure 5. As the SLD is unique to each material, the film structure can be determined from the depth distribution of the SLD. The thickness of both Cu films formed on the Si blocks was around 30 nm, indicating that the films were successfully deposited at the desired thickness. The thickness of the adsorption film formed on the Cu surface by deuterated stearic acid was 1.9 nm and the film density was 41%. The molecular film formed by deuterated stearyl alcohol had a thickness of 2.2 nm and a film density of 14%. Both the stearic acid and stearyl alcohol has linear chains with 18 carbons, and their molecular length was approximately 2 nm. Therefore, both molecules were in a monolayer state.

The line-and-space patterns observed by optical microscopy and dynamic mode AFM in air with a point probe cantilever are shown in Figure 6. These images show that there was no photoresist or metal residue on the specimen surface and that the target line-and-space pattern was successfully fabricated.

Studies on Si/Si and Si/SiO₂ friction measured by AFM have been mostly carried out in air with controlled humidity and in water and have shown that Si-based materials are subject to abnormal wear due to interfacial reactions (Chen et al., 2013; Zhang et al., 2017; Zhang et al., 2018). Although this has not been demonstrated for non-polar solvents such as PAO, the moisture in lubricants may affect wear. In this study, the ambient atmosphere around the lubricant was not controlled, and the moisture content in the lubricant was not monitored. As a result, wear with an average depth of about 0.4 nm was rarely observed in the friction area, but data from such cases were not used for evaluation in subsequent experiments. Even if there was sub-nanometer wear that became invisible in the image processing phase, its depth was small compared with the molecular size of the additive and therefore had little effect on the step measurement.

Typical examples of AFM image and height distributions obtained for PAO base oil are shown in Figures 7A, B, respectively. The results of Gaussian fitting to the Si and Cu height distributions in the observed images are shown in Figures 7C, D, respectively. The distances between the peaks correspond to the measured thickness of the Cu film since the minimum oil film thickness calculated using the Hamrock–Dowson equation (Hamrock and Dowson, 1977) is usually extremely small (<10 p.m.). The same analysis was applied to all observed images to obtain an average value. The Cu films used in the experiments were around 17 nm thick for PAO base oil. Experiments were conducted using at least five specimens, and observations were made at five locations for each specimen.

Due to differences in the thickness of the deposited Cu film on each substrate, the average step height for PAO base oil was subtracted from that for PAO with an additive using the same substrate. Since no adsorbed molecular film of an additive was formed on the inactivated Si and since the AFM friction measurements were performed under boundary lubrication conditions with little or no oil film between the two sliding surfaces, the change in the step height from Si to Cu measured for PAO with or without an additive corresponds to the gap generated by the effect of the adsorbed additive film on Cu.

The average gap at the metal–lubricant interface obtained from the observations for each load is shown in Figure 8. Gap formation was clearly observed for PAO containing stearic acid, and the gap was greater at lower loads. At the lowest load the gap was approximately the same as the thickness of the adsorbed stearic acid film observed in neutron reflectometry, whereas at a load of 125 nN the gap was about half the thickness. In contrast, PAO containing stearyl alcohol showed a negative gap for the two lowest loads and only a slight gap increase for the highest load. A negative gap, even if the absolute value is small, indicates that stearyl alcohol may have adsorbed on the Si rather than the Cu or that there may have been a small measurement or statistical error.

Representative friction coefficients for Si and Cu extracted from each 7.5 × 7.5 µm image captured during 15 friction cycles measurement are shown in Figures 9A, B, respectively. The simultaneously measured step heights for PAO containing stearic acid or stearyl alcohol and those for PAO before supplying lubricant for the same specimen are shown in Figure 10. The results for PAO containing stearic acid are shown as three separate results (COOH 1–3) in Figures 9, 10 because the step height showed more variability than those measured for PAO containing stearyl alcohol and PAO base oil.

The step height measurement for PAO base oil showed that the step height was almost unchanged from the initial value, indicating that wear of the Cu does not occur at a load of 75 nN (Figure 10). It can also be seen that stearyl alcohol had no effect on the step height change on Cu. There was no clear difference in the friction coefficient on Cu for PAO base oil and for PAO containing stearyl alcohol (the differences ranged from 0.17 to 0.15). These results indicate stearyl alcohol has insufficient load-bearing capacity against friction for a 75 nN load.

The friction coefficient for Si in PAO containing stearic acid was slightly lower (around 0.19), whereas for PAO base oil and PAO containing stearyl alcohol it was generally around 0.21. According to Campen et al., there is significant friction reduction for lubricants containing stearic acid on mica despite the absence of film formation on its surface (Campen et al., 2015a; Campen et al., 2015b). In this study, the stearic acid slightly reduced the friction coefficient on Si, but the degree of reduction was smaller. This may be because HF treatment inactivates Si surfaces, which makes adsorption of additives more difficult.

For Cu, in PAO containing stearic acid, the friction coefficient was about 0.07, which is a drastic reduction in friction. When stearic acid was used, the step heights were larger than those for Cu film in PAO base oil, indicating that adsorbed stearic acid film on the Cu film increases the gap at the interface. Since the gap varied by 1–1.5 nm during the 15 friction cycles, which is less than the thickness of the molecular film obtained by neutron reflectivity, the tip of the AFM cantilever is considered to have scanned on or inside the film.

Neutron reflectometry showed that both stearic acid and stearyl alcohol formed adsorbed molecular film about 2 nm thick under static conditions, but the AFM measurements showed no significant step height change on the Cu film when stearyl alcohol was used, even with the lowest load used (Figure 8). Neutron reflectometry revealed that the film density with stearyl alcohol was 14%, which is clearly lower than that of stearic acid (41%). In general, adsorbed molecular film is stronger at higher film densities due to larger interactions between the alkyl chains of the adsorbed additives (Studt, 1981; Hironaka, 1988). Films with lower densities may have a larger contact area with the AFM cantilever probe, resulting in a stronger interaction and thus a higher shear force per unit area and higher friction. It has been suggested that sliding between loosely packed molecular films may cause molecular deformation and greater energy dissipation (Lee et al., 2000). In this study, there were no differences in the friction coefficient and step height for Cu with or without stearyl alcohol. Since the molecular film formed by stearyl alcohol has quite low density, it can be concluded that either the load-bearing capacity was low and could not bear the cantilever tip or that was removed from the Cu surface without any positive or negative effect on friction.

As the results of neutron reflectometry show, the density of the film formed by the stearic acid was relatively high (41%). Furthermore, fatty acids have a high adsorption capacity for Cu, forming a high density molecular film with chemical bonds (Greenhill, 1949). Many studies using steel as a sliding material have shown that fatty acids exhibit a lower friction coefficient than alcohols (Jahanmir, 1985; Studt, 1989; Simic and Kalin, 2013) and that this tendency appears to be basically the same for Cu. The carboxy groups of fatty acids include –OH and C=O groups. Stearic acid can be adsorbed on Cu via–OH and C=O groups in a mono- or bidentate mode, and bidentate adsorption increases after the friction-induced tribochemical reaction (Hu et al., 1992; Fischer et al., 1997). Therefore, the adsorption capacity of fatty acids is stronger than that of alcohols, which in principle can only be adsorbed in monodentate mode. For methyl and ethyl esters of fatty acids, the adsorption capacity is reported to be small compared with that of the original fatty acid (Moore, 1951; Studt, 1981; Liascukiene et al., 2012). It is important to have both C=O and –OH groups, and in this study, a stable chemically adsorbed film was formed on Cu, providing substantial friction reduction and a gap increase even under severe friction conditions beneath the AFM cantilever tip.

Figure 8 shows that the durability of the molecular film formed by stearic acid is load dependent. At a load of 25 nN, the change in step height for PAO with/without stearic acid was about 1.7 nm, which is comparable, as revealed by neutron reflectometry, to the film thickness in a static state. This indicates that the molecular film is strong enough to bear an average contact pressure of 200 MPa, in accordance with the Hertzian contact between the Si cantilever and the Cu film, with a contact diameter of 16 nm. At a load of 125 nN, the gap decreased to about 1 nm. That the gap became smaller than the thickness of the molecular film formed in a static state suggests that the film may have been compressed or tilted or that the alkyl chains may have been bent (Salmeron, 2001). However, the gap was still well maintained, indicating that the molecular film formed by stearic acid had a high load-bearing capacity.

In the friction measurement at a load of 75 nN using stearic acid, the friction coefficient remained stable regardless of the number of friction cycles (Figure 9B), and the gap did not decrease due to, for example, removal of the adsorbed molecular film by friction. However, the step height fluctuated around 18–18.5 nm during measurement. Since the step height for PAO base oil exhibited a constant value of around 17 nm (Figure 10), the gap fluctuated significantly with the formation of the adsorbed stearic acid film. Although this measurement technique cannot determine the molecular conformation of the adsorbed stearic acid film during friction, the results indicate that the state of the film may change from moment to moment. The stearic acid adsorbed molecular film on Cu changed to a softer metallic soap state due to friction (Hirayama et al., 2017), suggesting that the cantilever can scan on and within the film.

In the experiments with stearyl alcohol, the neutron reflectometry confirmed film formation under static state conditions, whereas AFM observation simulating single asperity contact did not show any gap increase due to film formation. This indicates that there is a clear difference between film formation ability under static state conditions and maintaining the film under friction conditions. Therefore, to investigate the tribological performance of additives under boundary lubrication conditions, it is highly effective to simultaneously measure the friction and gap at the metal–lubricant interface, as proposed here, in addition to using such methods as neutron reflectometry and quartz crystal microbalance to evaluate adsorption on the metal surface.

Adsorption of an additive can occur to a small extent as long as there is no repulsive force between the substrate and additive. That there was little friction reduction on Si when additives were supplied (Figure 9A) means that weakly and sparsely adsorbed additives are easily removed by applied friction with virtually no effect on the frictional properties. The adsorption of fatty acids, alcohols, and amines on active SiO₂ surfaces has been reported in experiments and molecular simulations (Fry et al., 2020; Shi et al., 2020). During the feasibility phase of this study, measurements of friction in lubricant containing stearic acid were performed using Si without HF treatment, and friction reduction was obvious.

In this study, the stearic acid was successfully prevented from adsorbing on Si by removing the Si native oxide film and hydrogen termination by treatment with a special HF, which does not damage Cu, immediately before starting the experiment (as described in Section 2). This indicates that the adsorption of the additive can be suppressed by using inactivated Si. A slight decrease in the friction coefficient was still observed for Si in PAO containing stearic acid (Figure 9A). However, the degree of reduction was much smaller than that for Cu, and no difference in surface topography was observed, suggesting that the amount of adsorption was small.

The feasibility of the proposed method was evaluated using Cu, which can maintain a line-and-space pattern without dissolving or peeling during HF treatment and is active to the adsorption of additives. In future experiments with other kind of metals, such as iron, it will be necessary to improve the selection of the base material, e.g., by using diamond-like carbon coated Si, which is inactive to additive adsorption (Simič et al., 2014). Future work includes a more detailed evaluation of the contact pressure dependence of the gap at the metal–lubricant interface formed by adsorbed molecular films created by various type of additives on metal films and visualization of the formation of tribofilms and metallic soaps under friction conditions. Simultaneous measurement of friction and gap under various tribological conditions provides a deep understanding of additive properties, facilitating the selection and development of additives that strongly adsorb to metals, resulting in low friction and wear. This method is useful for parameter studies in the field of tribology and contributes to improving the energy efficiency of machines.