Combined use of neutron reflectometry and frequency-modulation atomic force microscopy for deeper understanding of tribology

The neutron beam is not familiar to most of researchers, but several features of neutron beams make them well suited for tribological studies.

• Neutrons can penetrate most common materials because they physically interact not with electrons but with atomic nuclei. Their use makes it easy to directly analyze the interface between a metal surface and lubricant.
• Neutrons can detect even light atoms such as hydrogen or carbon ones, in contrast to X-rays, making neutron beams suitable even for studies of lubricants composed of hydrocarbons.
• Labelled molecules by deuterium substitution are clearly distinguished from non-deuterated molecules by neutrons. Therefore, the thickness of an additive adsorption layer can be determined by using, for example, a normal base oil and a deuterated additive.
• Neutron energy is comparatively weak compared with that of X-rays or other beams, enabling non-destructive analysis of the target.

Reflectometry is a classical way to obtain angstrom-level information on the vertical structure, such as the distributions of density and thickness, through the reflectivity profile from the target surface (or interface).^44–46) The vertical structure at the interface can be estimated by taking the optimum fitting for the experimentally-obtained profile by the theoretical reflectivity profile on the Parratt's theory with a physical model⁴⁷⁾ because the contribution of "thickness" and "density" of the target layer for reflectivity profile is dependent in principle.

The typical procedure for interfacial analysis by NR is shown in Fig. 2. First, a sample substrate was set in the sample holder to obtain a neutron reflectivity profile from the interface between the substrate surface and air. Next, a base oil was flowed into the sample holder (=liquid cell) and a reflectivity profile from between the surface and the base oil was similarly obtained. Finally, an additive-mixed base oil was flowed and replaced with the normal base oil in the sample holder, and the reflectivity profile was again obtained. The neutron optics and sample holder were not touched during these steps, so the differences in the reflectivity profiles between steps were completely recognized as being provided by the existence of the base oil and/or additive at the interface. To clearly distinguish the additive layer from base oil by neutron, a deuterated base oil or a deuterated additive must be used as mentioned above.

Zoom In Zoom Out Reset image size

Typical neutron reflectivity profiles from the sample surface in air, in a base oil, and in a base oil with deuterated additive we experimentally-obtained are shown in Fig. 3(a).⁴⁸⁾ The reflectivity profiles in the figure were obtained by the angular-dispersive neutron reflectometer "SUIREN", which is an angle dispersive reflectometer, operating at JRR-3M in the Japan Atomic Energy Agency (JAEA).⁴⁹⁾ The sample surface was a Cu layer (∼75 nm thick) deposited on a silicon block. The base oil and additive were poly-alpha-olefin (PAO) and deuterated acetic acid in this case, and the concentration of acetic acid in the PAO was 0.1 mass%. The reflectivity on the vertical axis corresponds to the ratio of the reflected beam intensity to the incident beam intensity. The scattering vector q on the horizontal axis was calculated from the incident beam angle θ and the wavelength of the detected neutrons λ

$$\begin{eqnarray}&&q=4\pi \,\sin \,\theta /\lambda .\end{eqnarray} \tag{ 1 }$$

The figure shows that though there was no difference in the reflectivity profiles between the first step (in air) and second step (in PAO), the reflectivity profile obtained in the third step (in PAO + d-acetic acid) was slightly different; the fringe interval was clearly narrower than in the other profiles. The intervals of the interferential fringes in the profiles obtained in the first and second steps precisely corresponded to the thickness of the metal coating; the thickness of the Cu layer was estimated to be exactly 74.0 nm by taking the optimum fitting on a physical model. Fringe interval narrowing was observed only in the profile obtained in the third step, meaning that a homogeneous layer had formed on the metal surface. The layer could only be an additive layer formed by the deuterated acetic acid. Using the optimum fitting revealed that the adsorbed layer on the Cu surface was quite thin, only 2.0 nm, as illustrated in Fig. 3(b)⁴⁸⁾ with the estimated profile of scattering length density (SLD).

Zoom In Zoom Out Reset image size

Besides being useful for in situ analysis, NR has another important advantage: the environmental conditions are easily changeable and controllable. Typical reflectivity profiles from the target interface obtained in another experiment are shown in Fig. 4(a).⁵⁰⁾ The bottom, middle and top profiles were obtained for the sample surface in PAO, in PAO with deuterated palmitic acid (d-PA), and in PAO with d-PA under a hydrostatic condition (3.0 MPa), respectively. In this experiment, the sample surface was a Cu layer deposited on a silicon block as well, and the concentration of d-PA in PAO was 0.1 mass%. The reflectivity profiles were obtained by a time-of-flight (TOF) neutron reflectometer in the CN-3 beamline at the Institute for Integrated Radiation and Nuclear Science, Kyoto University. The hydrostatic condition was created by injecting Ar gas through the top hole in the liquid cell. The optimum fitting lines based on Parratt's theory are also shown in the figure.

Zoom In Zoom Out Reset image size

A comparison of the reflective profiles for PAO (bottom) and PAO with d-PA (middle) clearly shows that they are slightly different; the fringe interval in the profile for PAO with d-PA is clearly narrower than in the profile for PAO. The scattering length density (SLD) profiles and images of the layer structure at atmospheric and high pressures (3.0 MPa) are shown in Fig. 4(b).⁵⁰⁾ Fitting based on Parratt's theory showed that the adsorbed additive layer on the Cu surface at atmospheric pressure was about 1.4 nm thick and that the density of the additive in the layer was almost the same as that of pure deuterated palmitic acid. This means that a monolayer of palmitic acid molecules formed on the Cu surface with high grafting density, which is almost the same result as with pure palmitic acid. Moreover, the fringe interval in the top profile in Fig. 4(a) was much narrower than that in the profile at atmospheric pressure (middle profile). Fitting showed that the adsorbed additive layer at high pressure was about 5.9 nm thick and that the density of the additive in the layer was almost the same as that of pure deuterated palmitic acid, as it was at atmospheric pressure. These results interestingly demonstrate that a high-pressure condition may results in a thicker additive layer due to the change in equilibrium state of additive molecules in lubricant. It is a well-known fact that the structure (thickness and density) of the adsorbed additive layer dominates the friction coefficient. NR is a unique and powerful technique for quantitative analysis for the layer structure in lubricant even under various conditions.

Atomic force microscopy (AFM) has two basic operation modes: contact and dynamic. Dynamic mode is further divided into "amplitude-modulation (AM) mode" and "frequency-modulation (FM) mode". The contact mode AFM is widely used in many researches in various fields. Particularly in the field of tribology, lateral force microscopy (LFM) that is modified from the normal AFM to measure a horizontal force has been recognized as an invaluable technique for evaluating the friction coefficient with taking the 3D surface profile simultaneously in a microscopic view. Such microscale studies have been positioned as being fundamental ones for clarifying the mechanisms of friction and wear.^51–54) Recent advances with AFM for tribology studies is more remarkable, and classical problems are being moved toward clarification day by day.⁵⁵⁾

Regarding the dynamic mode AFM, on the other hand, the use of AM-AFM is more widespread than that of FM-AFM, especially in the soft matter physics field. AM-AFM is quite useful for measuring stiffness and damping properties of soft materials. FM-AFM is still minor because of the difficulty in controlling its operation, but FM-AFM is more sensitive in estimating the structure of soft materials covering liquid due to its higher signal-to-noise ratio.^56–58) In the FM-AFM apparatus, a cantilever is oscillated at the frequency of self-excited vibration by using a rectangular piezoelectric element. When the head of the cantilever comes close to the surface, the interaction force between the substrate and top of the cantilever shifts the self-excited frequency of the cantilever oscillation. The most attractive feature of FM-AFM is its higher sensitivity and its force resolution is less than 10 pN in principle.^59,60) The surface topography of materials is obtained by moving the cantilever along the z-axis to each (x,y) point, as shown in Fig. 5(a),⁶¹⁾ thereby keeping the shift, Δf, constant. Moreover, with FM-AFM it is possible to draw a cross-sectional density map of liquid molecules from the Δf map by x–z plane scanning, as shown in Fig. 5(b), owing to its high force sensitivity, which cannot be done with any other method.

Zoom In Zoom Out Reset image size

To obtain a cross-sectional image of the metal/lubricant interface by FM-AFM, the target lubricant is first dropped onto a substrate. It is desirable for the substrate to have an ultraflat surface. The lubricant is kept between the cantilever and substrate by its surface tension when the cantilever comes close to the substrate surface. A cross-sectional image of the interface between the substrate surface and pure base oil is captured first, and a similar image is captured again after the lubricant is changed to base oil with additive. The choice of cantilever is quite important for obtaining clear images. One should be selected from the viewpoints of tip sharpness and resonant frequency.

A typical example of an image captured by FM-AFM is shown in Fig. 6.⁶¹⁾ In the experiment, the substrate was an ultraflat Cu layer made by sputtering on a silicon wafer, and the base oil and additive are hexadecane and palmitic acid (PA), respectively. The concentration of PA in hexadecane was 0.01 mass%. A commercially available FM-AFM apparatus (SPM-8000FM, Shimadzu) and a cantilever (PPP-NCHAuD, Nanosensors) suitable for the interfacial observation were used in this case. A cross-sectional image of the interface between the substrate and lubricant with additive taken immediately after addition of the additive is shown in Fig. 6(a). On the basis of the FM-AFM principle, the bright area corresponds to a "high-density" area, which should show the adsorbed additive layer. We can see a clear layer structure formed by the palmitic acid on the Cu surface. After 2 h of continuous scanning, we can see that the layer grew to a thickness of 20 nm [Fig. 6(b)]. In contrast, after 2 h without continuous scanning, the layer did not grow. This indicates that the external stimulus of continuous scanning caused the additive layer to grow. FM-AFM revealed that the adsorbed additive layer easily grew into a multilayer under tribological conditions. When a silicon wafer was used as a substrate material instead of a Cu layer, a bright area was not seen. This demonstrates that an adsorbed additive layer forms only on an oxidized metal surface, as expected.

Zoom In Zoom Out Reset image size