Tribology of Copper Metal Matrix Composites Reinforced with Fluorinated Graphene Oxide Nanosheets: Implications for Solid Lubricants in Mechanical Switches

The potential for the use of copper coatings on steel switching mechanisms is abundant owing to the high conductivities and corrosion resistance that they impart on the engineered assemblies. However, applications of these coatings on such moving parts are limited due to their poor tribological properties; tendencies to generate high friction and susceptibility to degradative wear. In this study, we have fabricated a fluorinated graphene oxide–copper metal matrix composite (FGO-CMMC) on an AISI 52100 bearing steel substrate by a simple electrodeposition process in water. The FGO-CMMC coatings exhibited excellent lubrication performance under pin-on-disk (PoD) tribological sliding at 1N load, which reduced CoF by 63 and 69%, compared to the GO-CMMC and pure copper coatings that were also prepared. Furthermore, FGO-CMMC achieved low friction and low wear at higher sliding loads. The lubrication enhancement of the FGO-CMMCs is attributed to the tribochemical reaction of FGO with the AISI 52100 steel counterface initiated by the sliding load. The formation of an asymmetric tribofilm structure on the sliding track is critical; the performance of the FGO/Cu tribofilm formed in the track is boosted by the continued fluorination of the counterface surface during PoD sliding, passivating the tribosystem from adhesion-driven breakdown. The FGO-CMMC and GO-CMMC coatings also provide increased corrosion protection reaching 94.2 and 91.6% compared to the bare steel substrate, allowing for the preservation of the long-term low-friction performance of the coating. Other influences include the improved interlaminar shear strength of the FGO-containing composite. The excellent lubrication performance of the copper matrix composite coatings facilitated by FGO incorporation makes it a promising solid lubricant candidate for use in mechanical engineering applications.


GO and FGO powder characterisation
Characterization of the GO and FGO powders was carried out using a variety of techniques.
Transmission electron microscopy (TEM) was performed using a Tecnai T20 scope. Samples were prepared by drop casting dispersions of 0.1 mg/ml (F)GO in 1:1 water, IPA onto a lacey carbon support. SEM was carried out using a Zeiss Ultra 55 FEG scope. Samples were prepared by drop casting dispersion of 1 mg/ml (F)GO in 1:1 water, IPA onto a silicon wafer support. Contact angle measurements were taken on a Biolin Scientific, Theta Lite optical tensiometer. A 2 mL water droplet was pipetted onto the pelletized powder (13mm discs, consisting of 0.5g of powder, pressed at 10 tons for 5 minutes using a hydraulic press). Laser diffraction particle sizing was carried out using a Malvern Mastersizer, with the HydroEV attachment. Samples were dispersed in 1:1 water, IPA solution until appropriate obscuration values were reached and measured post two rounds of 120s sonication. Bulk FTIR spectra were obtained from 32 co-averages collected in transmission mode using an ATR-FTIR spectrometer (Nicolet iS50 spectrometer, Thermo Scientific) with a diamond crystal window, operating at 1 cm -1 resolution across a 650-4000 cm -1 range. Raman Spectroscopy was carried out on a Renishaw InVia with a 532 nm laser and a 50x scope. The StreamHR mapping function was used to map 480 spectra over a ~2000 µm 2 area. X-ray Photoelectron Spectroscopy (XPS) was performed on an Axis Ultra Hybrid spectrometer (Kratos Analytical) using monochromated Al Kα radiation (1486.6 eV, 10 mA emission at 150 W, spot size 300 x 700 μm) with a base vacuum pressure of ~5 × 10 -9 mbar. Charge neutralization was achieved using a filament. Analysis was carried out in CasaXPS. 1 Binding energy scale calibration was performed using C-C in the C 1s photoelectron peak at 284.7 eV. The high-resolution (HR) C 1s peaks were fitted using a finite Lorentzian (LF) peak shape, with a damping parameter used to quash the peak tail. 2  HR-F 1s . The transitions with the largest peak areas were selected for quantification. In the case of Cu 2p , K 2p and P 2p , the doublet pairs were used and scaled using the total relative S-3 sensitivity factor (RSF). 1 The K 2p doublet overlaps with the C 1s peak, as such a synthetic peak model was fitted to apportion the area. 1 Since K 2p spin-orbit splitting gives rise to a doublet, the component areas of 1/2 and 3/2 are constrained to a ratio of 0.5.

S-5
The detailed characterization of FGO and GO powders is summarized in Figure S1. A TEM image of a typical GO flake is shown in (a), its transparent nature indicating few-layer material, the inset (top left) selected area electron diffraction pattern (SAED), taken at the centre of the flake shows a hexagonal pattern of spots (annotated with Miller-Bravais hkil notation) extending across the captured image. The linear intensity profile of the SAED pattern in the inset (top right), shows that the intensity ratio between 1100 and 2110 is 1.7, this is consistent with few-layer behaviour 3 and follows literature values for GO. 4 No diffraction spots from the oxygen moieties are seen, as expected since these do not produce the requisite ordered lattice arrays. 4 The distribution of lateral size can be visualized in the SEM image of GO flakes drop cast onto a silicon wafer in (b). The consistent contrast indicates flakes of similar thickness, while a range of flake sizes can be observed from < 1 µm, to > 10 µm. The particle size distribution in (f) shows an asymmetric distribution for GO and FGO, as the blended freeze-dried powder had been sieved to < 50 µm, to minimize the presence of agglomerates. The D 50 is 28.5 and 30.5 µm for GO and FGO respectively. The slight increase could be indicated by larger aggregates in the FGO, possibly due to worse dispersion in the water and IPA solution due to the presence of hydrophobic fluorine groups. The characteristic (001) peak can be observed for both GO (at 10.9°) and FGO (at 11.0°) in the XRD spectra in (e), from which the interlayer spacing is calculated to be 8.1 and 8.0 Å respectively, suggesting that there is minimal effect on sheet-stacking post-fluorination.
The effect of fluorination can be most clearly observed in contact angle measurements, as shown in (c, d) for GO and FGO respectively. For the former, the hydrophilic surface leads the water to spread across the field of view, and so the contact angle is too small to accurately measure. Whereas for the latter, significant hydrophobic behaviour is observed due to the presence of fluorine groups increasing the contact angle to 135.2°.
The average FGO and GO spectra for the mapped area are shown in (g). Each spectrum shows a characteristic graphitic G mode at ~1566 cm -1 and defect-induced D mode at ~1342 cm -1 and weak 2D mode at ~2930 cm -1 . The I D /I G ratio increases from 0.93 (±0.02) for GO to 0.97 (±0.05) for FGO. However, since the FWHM of the G peaks is above 30 (61.2 and 56.8 cm -1 for GO and FGO respectively), the material is in the second stage of defect evolution, wherein an increase in I D /I G indicates a reduction in disorder. 5 Indicating that the plasma S-6 fluorination acts to clean the graphitic surface of amorphous debris and/or passivate defects.
The (F)GO powders were analysed under ATR-FTIR, as shown in (h). Both materials share a similar spectrum, hydroxyl (-OH) groups display a broad peak of ~3400 cm -1 , while other carbon lattice and oxygen functionality-derived peaks appear between 2000 -1000 cm -1 .
These consist of carboxylic stretching from C=O at 1720 cm -1 and C-OH at 1380 cm -16 , aromatic C=C stretching at 1610 cm -1 7 , and epoxy (C-O-C) deformations and stretching at 1200 and 1050 cm -1 respectively 8,9 . Fluorination appears to slightly dampen both carboxylic and epoxy peaks, implying functionalisation occurs via the substitution of these oxygen moieties with fluorine.  The ATR-FTIR spectra of the (F)GO Filtrate , have been plotted alongside the relevant powder in Figure S2 to better discern spectral changes. In both cases, the hydroxyl and aromatic C=C peaks are unaffected, however, the intensity of the other oxygen functionality peaks is dampened, in particular the carboxyl C=O (1720 cm -1 ) peak. The lack of bands in the alkyl C-H region (2900-2750 cm -1 ) eliminates organic functionalisation, suggesting this is due to carboxyl-edge functionalisation by an inorganic component.

Figure S3: (a) XPS survey scan and (b-d) high-resolution C 1s , O 1s and F 1s scans for (F)GO Filtrate samples.
The XPS analysis for the (F)GO Filtrate samples is summarised in Figure S3.  The HR-C 1s peak for the (F)GO Filtrate samples overlaps with K 2p peaks so both have been fitted simultaneously in (b). A new peak appears at ~288.2 eV assigned which can be assigned to C 1s signals for potassium carbonate/carboxylate, further evidencing functionalisation with the potassium metal in dispersion. 13 The shake-up, CF 2 and O-C=O, S-9 CHF peaks have a higher relative area in the FGO scan, which could be attributed to any fluorine that remains post-deposition. The HR-O 1s peak for the (F)GO Filtrate samples, shown in (c), broadens towards lower binding energies, to encompass a potassium carbonate/carboxylate peak (~ 531 eV). The O-F x peak weakens in the FGO post-deposition, in line with the drop in atomic fluorine concentration. In the HR-F 1s peak for FGO Filtrate , shown in (d), a new peak appears at 685.5 eV, indicating the presence of metallic fluoride bonds. 14 This suggests that the displaced fluorine groups may go on to react with potassium ions in solution and filter out as salts on the FGO Filtrate .   A summary of all the results presented in Figure S7a and b, are given in Table S4 below.