Due to restrictions relating to accelerating and guiding, the intensity of the incident neutron beam is usually rather low, and so in order to avoid long measurement times, samples with a large surface area (typically around 50 × 50 mm²) need to be used. In addition, a low surface roughness (ideally R_a < 2 nm) and a low coating thickness (preferably around 50 nm and up to maximum of 500 nm) are required to resolve thin adsorbed layers with sub-nanometer precision, whereas, the adhesive interlayers during coating deposition are avoided to facilitate the analysis of the reflectivity profiles.

The coatings used in our studies were hydrogenated (a-C:H) and silicon-doped (a-C:H:Si) DLC coatings. The coatings were deposited on ultra-flat silicon blocks with a surface roughness of R_q = 0.3 nm (Yamanaka Semiconductor Co., Japan), using a 13.56-MHz radio-frequency plasma-assisted CVD (RF PACVD) process. The thickness of the DLC coatings was in the range 50–60 nm for the adsorption studies of oiliness additives and in the range of 300–350 nm for the adsorption studies of the ZDDP additive.

A contrast between the specimen of interest and the surrounding media is a prerequisite for the successful detection of thin, adsorbed layers. Good contrast in the scattering length density (SLD, the definition is given later in subsection The Neutron Reflectometer and the Measuring Procedure) was achieved by employing deuterated hexadecanoic acid or hexadecanol (98 at. % D, Sigma–Aldrich Co., LLC.) with non-deuterated PAO6 oil (Neste Oil, Espo, Finland). In contrast, more complex and application-based molecules of ZDDP are not easy to deuterate, so a non-deuterated ZDDP additive (The Lubrizol Corporation, Ohio, USA) was used in deuterated decane/octadecane (98.4 at. % D, CDN Isotopes, Quebec, Canada), in order to obtain the relevant contrast. A solution of 0.7 wt. % oiliness additives and a solution of 1.0 wt. % ZDDP were used in our experiments.

The NR experiments were performed at the continuous spallation source (SINQ) at the Paul Scherrer Institute (PSI, Switzerland), using the Apparatus for Multi-Option Reflectometry (AMOR) operated in a neutron wavelength range of 3.5 Å < λ < 12 Å in the time-of-flight (TOF) detection mode (Gupta et al., 2004). A schematic of the AMOR reflectometer is shown in Figure 1. The measurements were performed using a horizontal sample-plane geometry. At the instrument position, the continuous neutron beam is pulsed using two choppers. A set of three slits was used to obtain the desired beam footprint on the sample's surface. The slits after the beam reflection were used to ensure that only specular reflection at an angle 2θ with respect to the incident beam reaches the detector. The sample's reflectivity was calculated as the ratio of the reflected beam to the incident beam intensities. In the case of specular reflection, the reflectivity is measured as a function of the momentum transfer vector perpendicular to the surface q_z (reflectivity profile), which is defined as qz=(4πλ)·sinθ. In order to obtain the reflectivity profile in the q_z range 0.01 Å-1<qz<0.08 Å-1, measurements were performed at two angles of incidence, i.e., θi= 0.5° and 1.3°. The variation of the neutron reflectivity with the momentum-transfer vector depends on the contrast in the scattering length density ρ, across the interface. The scattering length density is defined as (1)ρ=∑ibini, where b_i is the scattering length of the nuclear species i and n_i is its number density.

In our experiments the reflectivity was measured under static conditions and after the in-situ tribological testing (in dynamic conditions) using the tribotester rig shown in Figure 2. The test rig has two openings, which allows the neutron beam to enter the silicon block on one side and reflect from the backside of the coating. A PTFE counter body with dimensions of ∅ = 10 mm × 40 mm was connected via a holder to a loading stage to reach a normal load of 160 N, which corresponds to 10 MPa of the maximum initial Hertzian contact pressure. The holder with the loading stage is mounted on a slider driven by an actuator that simultaneously ensures a reciprocating sliding movement, thus creating the tribological contact. The tribotester rig also has a heating unit to enable high-temperature testing.

In order to obtain the thickness and the density of the adsorbed layer with the highest accuracy, the reflectivity measurements for all the samples were performed according to the following procedure (Hirayama et al., 2012; Kalin et al., 2014; Simič et al., 2014). In the first step, the reflectivity for the air/DLC interface was measured to determine the thickness and the vertical distribution of the scattering length density (SLD) of the coating itself (Figure 3A). In the second step, the sample surface was covered with base liquid/oil. This helped us to check whether any interactions between the pure liquid and the coating or liquid diffusion into the coating take place (Figure 3B). In the final step, the pure base liquid was replaced by the additive solution in the base liquid (Figure 3C). In the case of adsorption, a change in the reflectivity profile with respect to those measured in the first two steps is expected. In static measurements, the last step was repeated to monitor changes in the additive adsorption with the exposure time, whereas, in dynamic experiments, the reflectivity was measured in sequence after each application of rubbing for 30 min (383 cycles). The thickness and the density of the adsorbed layers were analyzed by fitting the model reflectivity profile to the measured profile according to Parratt's theory (Parratt, 1954).

The results in the form of graphs are presented as reflectivity profiles, i.e., the reflected beam intensity, normalized to 1, as a function of the momentum transfer vector q_z. The reflectivity profiles, obtained in all the subsequent steps are offset along the y axis to provide better clarity. The optimum fitting curves and the SLD profiles extracted from these fitting results are also presented. The χ² values for the fits were found to be in the range from 2.0 × 10⁻² to 7.8 × 10⁻³, meaning that this technique can provide a reliable thickness determination on a small scale. The uncertainty in the thickness calculation is, in principle, given by the chosen Δq/q of the experiment and is usually found to be in the range of ± 2–3 Å. The fitting parameters are the thickness, the density, and the roughness of the investigated structure, i.e., a monolayer, a bilayer, or a more complex multi-layer. To avoid any ambiguities in the construction of the theoretical model, additional measurements of the layer's thickness, roughness, surface coverage, and layer structure can be made with other complementary surface-sensitive techniques, such as AFM, X-ray reflectivity, XPS, AES, TOF-SIMS, etc.

It should also be noted that the measured NR provides an averaged value over a large surface area (the scanning footprint in our case was ~40 × 40 mm), and therefore, the spatial resolution in the lateral dimension is per se non-existent. For this reason NR is insensitive to the local thickness of the film and/or the distribution of defects.

Figure 4A shows the measured reflectivity profiles for the a-C:H coating exposed first to the base oil and then to the solution of hexadecanoic acid at 25°C. A shift of the reflectivity profile toward lower q_z values was observed upon exchanging the base oil with the additive solution, which is an indication of the interactions between the hexadecanoic acid and the a-C:H coating. The fitting of the reflectivity profiles revealed that an ~0.9-nm-thick layer was formed on the a-C:H coating during the 3-h exposure to the solution of hexadecanoic acid. This can be better seen from the SLD profile and schematic of the layer structure in Figure 4B that was extracted from the fitted reflectivity curve. The SLD value of the adsorbed layer was calculated to be ~3.2 × 10⁻⁶ Å⁻², which corresponds to a layer density of about 50% of the bulk value (the SLD of hexadecanoic acid = 6.4 × 10⁻⁶ Å⁻²). In addition, the fitting revealed that the DLC coating is relatively homogeneous over the whole coating thickness, which was found to be 54.2 nm.

The reflectivity profiles for the a-C:H DLC coating, exposed to air, base oil, and to the solution of hexadecanol in the final step, are presented in Figure 5A. In comparison to the hexadecanoic acid, hexadecanol exhibits very poor adsorption to the a-C:H DLC coating. This is evident from the absence of the phase shift between the reflectivity profiles for the a-C:H coating exposed to the base oil and to the additive solution. The SLD profile in Figure 5B also shows that here the coating was homogeneous over the whole coating thickness, which was calculated to be around 53.4 nm.

The same measurement procedure was also conducted for the a-C:H:Si DLC coating for both oiliness additives at 25°C. The obtained results contrast to what was observed in the case of the a-C:H DLC coating. Figure 6 shows first the measured reflectivity and the calculated SLD profiles for the adsorption of the hexadecanoic acid. The absence of a phase shift indicates that no continuous layer was formed during the coating's exposure to this additive.

On the other hand, the base oil's substitution with the hexadecanol solution causes a shift of the reflectivity profile toward lower q_z values, as can be seen in Figure 7A. This shift corresponds to the formation of an ~1.3-nm-thick layer (Figure 7B). The SLD value of the adsorbed layer, extracted from the fitted reflectivity curve, was around 2.2 × 10⁻⁶ Å⁻², which corresponds to a layer density of around 37% of the density of the bulk hexadecanol (SLD = 5.9 × 10⁻⁶ Å⁻²).

In contrast to the a-C:H DLC, the silicon-doped coatings were found to be less homogenous over the coating thickness, as can be seen from the SLD profiles (Figures 5B, 6B). These coatings consist of roughly two layers, a bottom thinner layer with a thickness of around 5 nm and the SLD of around 4.1 × 10⁻⁶ Å⁻², and a topmost thicker layer with a thickness of around 48 nm and an SLD of around 3.4 × 10⁻⁶ Å⁻².

Figure 8A shows the reflectivity profiles for a-C:H DLC coating exposed first to air, then pure deuterated decane and finally to the ZDDP solution in deuterated decane at 25°C for different exposure times. The corresponding SLD profiles, extracted from fitted curves, are presented in Figure 8B. We can see that the reflectivity profiles remained unchanged upon decane substitution with the ZDDP solution, regardless of the exposure time. This indicates poor adsorption of the ZDDP molecules on the a-C:H DLC coating at 25°C.

The reflectivity and the corresponding SLD profiles obtained at elevated temperature, i.e., 120°C, are presented in Figure 9. Under these conditions, differences in the reflectivity profile upon exchanging pure decane with the ZDDP solution occurred during the first 2 h of the exposure to ZDDP. The phase shift toward lower q_z values that appeared only at q_z values above 0.03 indicates that interactions between the ZDDP and the a-C:H DLC coating take place. The fitting revealed that this phase shift corresponds to the formation of an ~0.8-nm-thick layer. With longer exposure times, i.e., 4–6 h, the phase shift toward lower q_z values became more pronounced than that observed during first 2 h. It also appeared at q_z values below 0.03. The fitting of the reflectivity curves indicates that during this time a 1.2-nm-thick layer was formed. The presence of the interaction layer upon coating exposure to the ZDDP and its growth with further exposure (4–6 h) can be seen in the SLD profiles in Figure 9B.

The SLD profiles in Figures 8B, 9B also show that the a-C:H DLC coatings used in this study were not perfectly homogenous over the coating thickness. This can be seen from the presence of layers of various densities in the film. In addition, it was observed that the SLD values of the individual layers changed when decane was substituted with the ZDDP solution. These changes could be attributed to the diffusion of the ZDDP solution into the coating through the coating imperfections (e.g., defects, pinholes).

Figure 10A shows the NR profiles for the a-C:H coating measured at 100°C during exposure to air, then pure octadecane, after that to ZDDP solution for 4 h and finally after each application of rubbing in the presence of ZDDP. The corresponding SLD profiles are presented in Figure 10B.

The coating's exposure to ZDDP for the first 4 h resulted in the shift of the neutron reflectivity toward lower q_z values with respect to that obtained in pure octadecane. As obtained from fitting the data, this difference in the NR corresponds to the thickness change due to the adsorption of the ~1.2-nm-thick layer. Similar to that observed at 120°C in section Adsorption of ZDDP on an a-C:H DLC Coating at Various Temperatures, the SLD values of the individual layers changed upon exchanging the pure base liquid (i.e., octadecane) for the ZDDP solution.

The reflectivity profile obtained after the application of rubbing for 30 min (after 383 cycles) shifted further toward lower q_z values, when compared to the reflectivity measured in octadecane and ZDDP during the first 4 h of exposure. The fitting showed that the thickness increased from ~1.2 to 1.7 nm. Upon further exposure to rubbing for another 30 min (in total 60 min of rubbing and 766 cycles) no major changes in the reflectivity were observed (data not shown).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.