Enhanced Adsorption of Epoxy‐Functional Nanoparticles onto Stainless Steel Significantly Reduces Friction in Tribological Studies

Abstract Epoxy‐functional sterically‐stabilized diblock copolymer nanoparticles (ca. 27 nm) are prepared via RAFT dispersion polymerization in mineral oil. Nanoparticle adsorption onto stainless steel is examined using a quartz crystal microbalance. Incorporating epoxy groups within the steric stabilizer chains results in a two‐fold increase in the adsorbed amount, Γ, at 20 °C (7.6 mg m−2) compared to epoxy‐core functional nanoparticles (3.7 mg m−2) or non‐functional nanoparticles (3.8 mg m−2). A larger difference in Γ is observed at 40 °C; this suggests chemical adsorption of the nanoparticles rather than merely physical adsorption. A remarkable near five‐fold increase in Γ is observed for ca. 50 nm epoxy‐functional nanoparticles compared to non‐functional nanoparticles (31.3 vs. 6.4 mg m−2, respectively). Tribological studies confirm that chemical adsorption of the latter epoxy‐functional nanoparticles leads to a significant reduction in friction between 60 °C and 120 °C.


Introduction
It is well-documented that the addition of oil-soluble polymers to automotive engine oils confers various benefits.For example, polyolefins or poly(n-alkyl methacrylates) confer superior lubrication performance, [1][2][3][4] polystyrenebased diblock copolymers can act as effective diesel soot dispersants [5] and star diblock copolymers can be used to control the viscosity-temperature profile of engine oils. [6]n 2010, Zheng and co-workers reported that corecrosslinked acrylic diblock copolymer nanoparticles reduced friction within the boundary lubrication regime as judged by mini-traction machine (MTM) studies. [7]Such nanoparticles were postulated to provide a physical barrier between the planar stainless steel surface and the stainless steel ball bearing that was placed in direct contact with it.However, these acrylic nanoparticles were prepared by a multi-step process that involved protecting group chemistry and selfassembly in dilute solution via solvent transfer from THF to mineral oil. [7]Thus they are not readily amenable to industrial scale-up.[13][14][15][16][17][18] Importantly, such sterically-stabilized poly(stearyl methacrylatepoly(benzyl methacrylate)-poly(ethylene glycol dimethacrylate) (PSMA 31 -PBzMA 200 -PEGDMA 20 ) nanoparticles also significantly reduced friction within the boundary lubrication regime in tribological experiments. [8]erein we demonstrate that introducing epoxy groups into similar methacrylic nanoparticles significantly enhances their adsorption from n-dodecane onto planar stainless steel.This is achieved simply by statistical copolymerization of glycidyl methacrylate (GlyMA) with lauryl methacrylate (LMA) when preparing the steric stabilizer precursor block, prior to its chain extension using either BzMA or methyl methacrylate (MMA).Quartz crystal microbalance with dissipation (QCM-D) and scanning electron microscopy (SEM) are used to compare the adsorption of such epoxyfunctional nanoparticles onto a model planar stainless steel substrate with that of (i) nanoparticles containing epoxy groups located within the cores and (ii) nanoparticles bearing no epoxy groups.Finally, tribological experiments are conducted to assess whether nanoparticle adsorption leads to lower friction within the boundary lubrication regime.

Synthesis and Characterization of Spherical Nanoparticles
RAFT dispersion polymerization in mineral oil was used to produce ca.27 nm diameter epoxy-functional PLMA 63 -PGlyMA 89 and P(LMA 50 -stat-GlyMA 9 )-PMMA 67 nanoparticles, as well as the corresponding non-functional PLMA 63 -PMMA 67 nanoparticles (see Scheme 1).Similarly, both epoxy-functional P(LMA 50 -stat-GlyMA 9 )-PBzMA 245 and non-functional PLMA 63 -PBzMA 245 spheres of approximately 50 nm diameter were also prepared in mineral oil (see Scheme 1).In each case, a high final monomer conversion (� 95 %) was achieved within 6 h.THF GPC analysis indicated comparable M n values for similar copolymer compositions, as shown in Figure S1.A narrow molecular weight distribution (M w /M n � 1.18) and a high blocking efficiency was obtained for each synthesis (see Figure S1).

QCM-D Studies of Nanoparticle Adsorption onto a Planar Stainless Steel Substrate
26][27][28][29][30] There are also various QCM-D studies of the adsorption of small-molecule surfactants, such as glyceryl monooleate (GMO), [31] fatty amines (e.g., octadecylamine, N-tallowalkyl-1,3-propanediamine) [31,32] or fatty acids (e.g., stearic-, oleic-or linoleic acids) [33,34] onto stainless steel, iron oxide or silica substrates from n-alkanes.38] Recently, Gmür et al. used QCM-D to study the adsorption of a series of poly(lauryl methacrylate)-based diblock copolymers onto a planar iron oxide substrate from hexadecane.Such copolymers were prepared by RAFT solution polymerization using pentafluorophenyl methacrylate as reactive repeat units to enable introduction of nitrodopamine functional groups.The resulting anchor block led to a relatively high adsorbed mass in the form of a brush-like surface layer and excellent friction reduction in tribological experiments. [3]Similar observations were reported by the same team for poly(lauryl methacrylate)-based diblock copolymers bearing a carboxylic acid-based anchoring block. [4]Herein we report the first QCM-D study of the adsorption of epoxy-functional diblock copolymer nanoparticles onto planar stainless steel from a nonpolar solvent (n-dodecane).
First, adsorption of the ca.27 nm epoxy-functional PLMA 63 -PGlyMA 89 nanoparticles, epoxy-functional P(LMA 50stat-GlyMA 9 )-PMMA 67 nanoparticles and non-functional PLMA 63 -PMMA 67 nanoparticles were compared via QCM-D experiments performed at 20 °C.Initially, pure n-dodecane was introduced into the QCM-D cell to obtain a baseline (see Figure 2), followed by switching to a 1.0 % w/w nanoparticle dispersion.Nanoparticle adsorption occurred at this stage (see Figure 2a, first red arrow) and subsequently returning to a flow of pure n-dodecane enabled removal of weakly adsorbed nanoparticles from the surface (see Figure 2a, second red arrow).Typical dissipation data obtained during adsorption of the P(LMA 50 -stat-GlyMA 9 )-PMMA 67 nanoparticles is shown in Figure S2.
According to Figure 2a, employing the P(LMA 50 -stat-GlyMA 9 )-PMMA 67 nanoparticles resulted in a significantly greater change in frequency (Δf 3 = À 129 Hz) relative to that recorded for either the PLMA 63 -PGlyMA 89 nanoparticles (Δf 3 = À 63 Hz) or the non-functional PLMA 63 -PMMA 67 nanoparticles (Δf 3 = À 65 Hz).In each case, the Δf 3 data provided in Figure 2 were average values obtained from two consistent measurements (see Table S2).The adsorbed mass per unit area can be calculated from Δf 3 (see Figure 2, black double-headed arrows) using the Sauerbrey equation: [39] Dm where C is a mass sensitivity constant, which is equal to 0.177 (mg (m 2 Hz) À 1 ) for a 5 MHz crystal, [40] and n is the overtone number.To determine the final adsorbed mass of nanoparticles in mg m À 2 , we chose to use the third harmonic (n = 3) of the resonance frequency.This is because the fundamental frequency (n = 1) is rather sensitive to experimental artifacts. [41,42]  is emphasized that the Sauerbrey model assumes a rigid adsorbed layer and hence neglects viscoelastic effects.If the layer is appreciably viscoelastic, employing the Sauerbrey equation leads to an underestimate of the mass of the adsorbed layer. [29,43] onitoring the dissipation (D) in real time provides useful information regarding the viscoelasticity of the adsorbed nanoparticle layer. [44,45] n principle, the adsorbed layer can be assumed to be rigid provided that the change in dissipation (ΔD) is relatively small compared to Δf. [46,47] Furthermore, Reviakine et al. reported that the Sauerbrey equation can be safely used if ΔD n /(À Δf n /n) ! 4 × 10 À 7 Hz À 1 for a 5 MHz sensor. [43,47] s shown in Table S2, all ΔD 3 /(À Δf 3 /3) data were well below this threshold value.Thus the Sauerbrey equation should be valid for the nanoparticle adsorption experiments performed herein.Accordingly, adsorbed amounts of 7.6, 3.7 and 3.8 mg m À 2 were calculated for the P(LMA 50 -stat-Gly-MA 9 )-PMMA 67 , PLMA 63 -PGlyMA 89 and PLMA 63 -PMMA 67 nanoparticles, respectively.Thus, introducing epoxy groups into the steric stabilizer block resulted in a two-fold increase in the adsorbed mass per unit area compared to the nonfunctional PLMA 63 -PMMA 67 nanoparticles.However, introducing almost ten times more epoxy groups into the nanoparticle cores produced no discernible increase in the adsorbed amount relative to these reference nanoparticles.Hence the precise spatial location of the epoxy groups within the nanoparticles is of critical importance for achieving their enhanced adsorption onto stainless steel.Moreover, our preliminary experiments (data not shown) suggest that placing a single epoxy group at the end of each steric stabilizer chain using an epoxy-functional RAFT agent [48] is not sufficient to promote greater nanoparticle adsorption.Thus it seems that multiple epoxy groups per copolymer chain are required to achieve this objective.In future work, it would be interesting to establish the optimum number of epoxy groups per copolymer chain that are required for maximum nanoparticle adsorption.
To estimate the corresponding fractional surface coverage (Θ) in each case, the QCM-D adsorbed amount (Γ) was compared to the theoretical adsorbed mass (Γ t ) estimated for a fully-coated substrate of unit surface area.Γ t was calculated using Equation 2: where A is equal to 1 m 2 , D is the overall nanoparticle diameter (ca.27 nm in this case) and ρ is the density of either the PGlyMA core (ρ ~1.07 g cm À 3 at 20 °C) or the PMMA core (ρ = 1.18 g cm À 3 at 20 °C).Accordingly, a Θ of 0.24 was determined for P(LMA 50 -stat-GlyMA 9 )-PMMA 67 nanoparticles, 0.13 for PLMA 63 -PGlyMA 89 nanoparticles and 0.12 for PLMA 63 -PMMA 67 nanoparticles.There are two possible explanations for the significantly higher surface coverage obtained for the epoxy-functional P(LMA 50 -stat-GlyMA 9 )-PMMA 67 nanoparticles.In principle, the surface hydroxyl (i.e., Fe-OH) groups on the stainless steel substrate can simply form hydrogen bonds with the polar epoxy groups.Alternatively, such surface hydroxyl groups can ring-open the epoxy groups to form covalent bonds.It is important to note that physical adsorption via hydrogen bonding should be less favorable at higher temperature.In contrast, such conditions should promote chemical adsorption.
To distinguish between these two adsorption mechanisms, the QCM-D experiments were repeated at 40 °C, which is the upper limit temperature for our instrument setup (see Figure 2b).This higher temperature led to a modest but discernible increase in adsorbed amount for the P-(LMA 50 -stat-GlyMA 9 )-PMMA 67 nanoparticles (Γ = 8.9 mg m À 2 ; Θ = 0.28), whereas a significant reduction was observed for both the PLMA 63 -PGlyMA 89 nanoparticles (Γ = 2.7 mg m À 2 ; Θ = 0.09) and the PLMA 63 -PMMA 67 nanoparticles (Γ = 2.5 mg m À 2 ; Θ = 0.08).Thus these experiments suggest that a chemical reaction between the epoxy and hydroxyl groups occurs when employing the P(LMA 50 -stat-GlyMA 9 )-PMMA 67 nanoparticles, whereas the other two types of nanoparticles merely undergo physical adsorption (most likely via hydrogen bonding interactions between the methacrylic ester groups on the PLMA chains and the Fe-OH groups present at the surface of the stainless steel).This is consistent with our recent observation that higher temperature enhances the ring-opening of epoxy groups by water. [49]Moreover, both  [50,51] Furthermore, Xu and co-workers recently used sum-frequency generation (SFG) vibrational spectroscopy to study the chemical reaction between epoxy rings and surface Al-OH groups when curing an epoxy adhesive on sapphire. [52]Moreover, the same team reported enhanced adhesion for an epoxy-amine adhesive when cured in contact with a stainless steel substrate.Again, direct spectroscopic evidence was provided for a chemical reaction between the epoxy rings and the Fe-OH groups. [53]ext, the adsorption of ca.50 nm diameter epoxy-functional P(LMA 50 -stat-GlyMA 9 )-PBzMA 245 nanoparticles onto stainless steel was compared to that obtained for non-functional PLMA 63 -PBzMA 245 nanoparticles of similar size.To produce larger spheres in mineral oil, PBzMA was chosen as a core-forming block instead of PMMA.This is because targeting higher PMMA DPs leads to the formation of colloidally unstable aggregates. [54]As shown in Figure 3, increasing the particle size only resulted in a modest increase in adsorbed amount (Γ = 6.4 mg m À 2 ) when using the PLMA 63 -PBzMA 245 nanoparticles.In contrast, an approximate five-fold increase in adsorbed amount (Γ = 31.3mg m À 2 ) was obtained for the P(LMA 50 -stat-GlyMA 9 )-PBzMA 245 nanoparticles (see Table S3).Using Equation 2, assuming an overall nanoparticle diameter of 50 nm and taking the density of PBzMA to be 1.179 g cm À 3 , fractional surface coverages of 0.11 and 0.53 were calculated for the non-functional and epoxy-functional nanoparticles, respectively.This remarkable difference is consistent with the observations made for the ca.27 nm nanoparticles.Moreover, it indicates that nanoparticle adsorption is independent of the chemical nature of the core-forming block (i.e., PMMA vs. PBzMA).Furthermore, the ca.50 nm diameter nanoparticles are sufficiently large to be imaged via SEM after their

Research Articles
adsorption onto stainless steel (see Figure 3).Such SEM images can be used to estimate surface coverage (Θ) using ImageJ software.

Tribology Experiments Using 50 nm Nanoparticles
Previously, Zheng et al. [7] and Derry et al. [8] reported excellent lubrication performance for acrylic and methacrylic core-crosslinked diblock copolymer nanoparticles, respectively.In these prior studies, Stribeck curves suggested a dramatic reduction in the friction coefficient at low entrainment speeds (i.e.within the boundary lubrication regime) compared to glycerol monooleate (GMO), which is a well-known friction-reducing agent.Zheng et al. [7] postulated that this was the result of elastic deformation of the nanoparticles once they diffused within the asperity contact area. [7,8] erein, we sought to determine whether there is a direct relation between lubrication performance and nanoparticle adsorption.Accordingly, we compared the lubrication performance of the ca.50 nm epoxy-functional P-(LMA 50 -stat-GlyMA 9 )-PBzMA 245 nanoparticles with that of the non-functional PLMA 63 -PBzMA 245 nanoparticles.
Initially, Stribeck curves were recorded for 2.5 % w/w nanoparticle dispersions at 40, 60 and 80 °C while reducing the entrainment speed (v e ) from 3000 to 30 mm s À 1 using a slide-to-roll ratio (SRR) of 50 % under a constant applied load of 37 N (see Figure S3).At 40 °C, very similar data were obtained for the nanoparticle dispersions (see Figure S3c).Moreover, these experiments indicate that a v e of approximately 200 mm s À 1 lies within the so-called 'mixed lubrication' regime at this temperature.This intermediate regime lies between the hydrodynamic (full fluid film) and boundary (metal-on-metal contact) lubrication regimes. [55]In principle, lowering the entrainment speed reduces the film thickness created by the oil between the metal surfaces, which should lead a higher number of asperity contacts and hence a higher friction coefficient (see Figure S3c).
Since the QCM-D data suggests enhanced adsorption for the epoxy-functional nanoparticles at higher temperature (see Figure 2), we examined more demanding MTM test conditions.Thus, the entrainment speed was fixed at 200 mm s À 1 and a slide-to-roll ratio (SRR) of 50 % was employed under an applied load of 37 N while heating from 40 to 120 °C.In principle, the viscosity of the nanoparticle dispersion should be lower at higher temperature, which means that the onset of the "mixed lubrication" regime should occur at a higher entrainment speed (see Figure S3ab).Hence additional asperity contacts and a higher friction coefficient were expected at elevated temperature.
For the 40-60 °C regime, the two sets of data are essentially identical.A gradual increase in friction coefficient is observed for both the epoxy-functional P(LMA 50stat-GlyMA 9 )-PBzMA 245 nanoparticles and the non-functional PLMA 63 -PBzMA 245 nanoparticles (see Figure 4a).Above 60 °C (which corresponds to the boundary lubrication regime at a v e of 200 mm s À 1 ; see Stribeck curves shown in Figure S3a-b), a significant reduction in the friction coefficient from 0.09 to 0.04 was observed for the epoxyfunctional nanoparticles, whereas there is almost no change in this parameter for the non-functional nanoparticles.This suggests that higher temperatures lead to more epoxyfunctional nanoparticles becoming located within the asperity contact area, despite the relatively harsh experimental conditions.Indeed, inspecting the surface of the MTM disks (within the test area) after these tribology experiments by SEM revealed a significant difference in the nanoparticle surface coverage, see Figure 4b.
Digital image analysis using ImageJ software indicated a relatively high fractional surface coverage of 0.45 for the epoxy-functional P(LMA 50 -stat-GlyMA 9 )-PBzMA 245 nanoparticles but only 0.09 for the corresponding non-functional nanoparticles.In the former case, it is perhaps worth emphasizing that the chemically-adsorbed nanoparticles remain intact despite the relatively harsh conditions to which they were subjected during the tribological experiments.These MTM studies and SEM observations are consistent with our QCM-D findings and provide further (albeit indirect) evidence for the likely chemical adsorption of the nanoparticles onto stainless steel, see Figure 4c.
The highest adsorbed amount was obtained when epoxy groups were incorporated within the steric stabilizer chains, rather than being located within the nanoparticle cores.Conducting these QCM-D experiments at 40 °C led to a significantly higher adsorbed amount for the epoxy-functional P(LMA 50 -stat-GlyMA 9 )-PMMA 67 nanoparticles.For purely physical adsorption, a lower adsorbed amount should be obtained under such conditions.Thus, such temperaturedependent studies suggest that nanoparticle adsorption involves a chemical reaction between the epoxy groups and the Fe-OH groups located at the stainless steel surface.Indeed, there is some recent literature precedent for such surface chemistry. [52,53] ncreasing the nanoparticle diameter from ca. 27 nm to ca.50 nm resulted in an approximate five-fold increase in the adsorbed amount for P(LMA 50 -stat-GlyMA 9 )-PBzMA 245 nanoparticles compared to that for the non-functional PLMA 63 -PBzMA 245 nanoparticles, with the corresponding QCM-D fractional surface coverages estimated to be 0.53 and 0.11, respectively.Furthermore, replacing the core-forming PMMA block with PBzMA confirmed that the enhanced adsorption of such epoxy-functional spheres is independent of the nature of the core-forming block.
The ca. 50 nm diameter nanoparticles were then examined for tribology experiments using an MTM set-up.When heating from 60 °C to 120 °C at a constant entrainment speed and slide-to-roll ratio, the epoxy-functional P(LMA 50 -stat-GlyMA 9 )-PBzMA 245 nanoparticles reduced the friction coefficient dramatically compared to the corresponding nonfunctional PLMA 63 -PBzMA 245 nanoparticles.In this case, postmortem SEM studies indicated a significantly higher surface coverage for the adsorbed epoxy-functional nanoparticles.Such studies provide strong evidence for chemical adsorption of such nanoparticles via an epoxy-hydroxyl ringopening reaction at elevated temperature.
Finally, given their relative ease of synthesis directly in mineral oil using potentially scalable chemistry, such epoxyfunctional nanoparticles appear to offer important advantages compared to other polymer-based lubricating systems reported in the literature such as polymer brushes. [56,57] ydrogen bond formation between the polar epoxy rings and the surface hydroxyl groups (i.e., Fe-OH) or (ii) epoxy ring-opening reaction with the same surface hydroxyl groups to produce covalent bonds.The experimental evidence reported herein supports the latter interaction, for which there is also some literature precedent. [52,53]

Figure 2 .
Figure 2. QCM-D analysis showing the change in frequency (Δf 3 ) observed during the adsorption of PLMA 63 -PMMA 67 nanoparticles (blue data), PLMA 63 -PGlyMA 89 nanoparticles (green data) or P(LMA 50 -stat-GlyMA 9 )-PMMA 67 nanoparticles (red data) in turn from 1.0 % w/w copolymer dispersions in n-dodecane onto a stainless steel substrate at a flow rate of 0.5 mL min À 1 : (a) at 20 °C and (b) at 40 °C.In part (a), the first red arrow indicates when the nanoparticle dispersion was first introduced into the QCM cell while the second arrow corresponds to the rinsing step with pure n-dodecane.Each curve is shown for a single measurement but good reproducibility was observed for duplicate experiments.Black doubleheaded arrows indicate the final change in frequency (Δf 3 ) in each case, from which the adsorbed mass can be calculated.

Figure 3 .
Figure 3. QCM-D data obtained for the adsorption of (a) PLMA 63 -PBzMA 245 nanoparticles and (b) P(LMA 50 -stat-GlyMA 9 )-PBzMA 245 nanoparticles in turn from 1.0 % w/w copolymer dispersions in n-dodecane onto a stainless steel substrate at a flow rate of 0.50 mL min À 1 at 20 °C.Each curve is shown for a single measurement but good reproducibility was observed for duplicate experiments.The corresponding SEM images recorded for each nanoparticle-coated stainless steel substrate after these QCM-D experiments are also shown.The black doubleheaded arrows indicate the final change in frequency (Δf 3 ).

Figure 4 .
Figure 4. (a) Friction coefficient vs. temperature data obtained for a 2.5 % w/w dispersion of PLMA 63 -PBzMA 245 nanoparticles (blue data) and epoxy-functional P(LMA 50 -stat-GlyMA 9 )-PBzMA 245 nanoparticles (red data).Data were recorded at an entrainment speed of 200 mm s À 1 with a slide-to-roll ratio (SRR) of 50 % under an applied load of 37 N. (b) SEM images recorded for the MTM disks following these tribology experiments when using the PLMA 63 -PBzMA 245 nanoparticles (blue frame) and the P(LMA 50 -stat-GlyMA 9 )-PBzMA 245 nanoparticles (red frame), respectively.The epoxy groups in the latter nanoparticles clearly promote much stronger adsorption, which correlates with the significant reduction in friction coefficient observed above 60 °C.(c) Two possible mechanisms for the enhanced adsorption of epoxy-functional P(LMA 50 -stat-GlyMA 9 )-PBzMA 245 nanoparticles onto stainless steel: (i)hydrogen bond formation between the polar epoxy rings and the surface hydroxyl groups (i.e., Fe-OH) or (ii) epoxy ring-opening reaction with the same surface hydroxyl groups to produce covalent bonds.The experimental evidence reported herein supports the latter interaction, for which there is also some literature precedent.[52,53] Hatton et al. and Docherty et al. reported that hydroxyl groups can ring-open epoxy groups