Zijian Liua,
Dustin Leiningera,
Amir Koolivanda,
Alex I. Smirnova,
Olga Shenderovab,
Donald W. Brennera and
Jacqueline Krim*a
aNorth Carolina State University, Raleigh, NC 27695-7218, USA. E-mail: jkrim@ncsu.edu; Tel: +1 919 513 2684
bInternational Technology Center, 8100 Brownleigh Dr, Raleigh, NC 27617, USA
First published on 11th September 2015
Surface uptake and nanotribological properties of positively and negatively charged 5–15 nm diameter nanodiamonds dispersed in water have been studied in real time by means of an in situ Quartz Crystal Microbalance (QCM) technique. The frequency and dissipative properties (mechanical resistance) of a QCM with gold surface electrodes immersed in water were monitored upon addition of the nanodiamonds. Measurements were also performed with the QCM electrode in macroscopic contact with stainless steel ball bearings in the presence or absence of the positively or negatively charged nanodiamonds dispersed in water surrounding the contact. The nanodiamonds were found to have a profound effect on the tribological performance at both nanometer and macroscopic scales. The tribological effects were highly sensitive to the sign of the nanodiamond electrical charge: negatively (positively) charged particles exhibited weaker (stronger) adhesion. Positively charged particles consistently increased friction at the solid–liquid interface, while negatively charged particles of comparable size were observed to decrease interfacial friction. For the macroscopic contacts with the gold electrodes, negatively charged nanodiamonds appeared to be displaced from the contact, while the positively charged ones were not. Overall, the negatively charged nanodiamonds were more stable in an aqueous dispersion over extended time periods while the positively charged nanodiamonds coagulated into larger particles and formed precipitates more quickly.
Currently, different types of nanoparticulates are being actively pursued as alternatives to the traditional oil-based lubricant additives. Specifically, nanoparticulates composed of metals (e.g., Zn, Al, Cu Ti, Fe) and their oxides,10–16 fluorinated compounds, graphitic nanoparticulates, and nanodiamonds17–19 have all been shown to yield significant reductions in both friction (typically 10–20%) and wear when used as lubrication additives. Studies of nanoparticulate additives have also been extended to nonpetroleum-based fluids, including water (e.g. fullerenols,20 ZnO and Al2O3 nano-particulates11,13,21) and ionic liquids (e.g. functionalized multi-walled carbon nanotubes22). All of the abovementioned studies reported both a reduction in friction and an improved wear resistance.
One emerging class of effective anti-friction and wear-preventing nanoparticulates is comprised of nanodiamonds obtained mechanically (i.e., by crushing diamond crystals23) or through detonation of an oxygen-deficient mixture of explosives.24 Crushed nanodiamonds are typically 100–500 nm in diameter and have potential as solid lubricants in vacuum23 whereas the detonation nanodiamonds (DNDs) are much smaller, typically 4–7 nm in diameter,24 with a narrow size distribution. DNDs have shown to be effective as anti-friction and anti-wear additives to oil-based lubricants.17,25–28 Recently, Osawa described eight- to tenfold reduction in the friction coefficients upon addition of 4 nm NDs to several polar liquids including water, ethylene glycol, and DMSO.29 Given that the conventional additives to oil are insoluble or ineffective in aqueous and ionic liquid solvents, development of NDs as additives may ultimately allow for replacing the centuries-old oil-based lubricating technologies altogether after the essential technical criteria are met.29
While the empirical search for the best combination of nanoparticulates, lubricating fluids, and applicable surfaces continues, the fundamental understanding of the atomic scale mechanisms responsible for the macroscopic tribological performance is currently lacking in the literature. While experiments in vacuum with larger (>100 nm in diameter) NDs suggest that these hard spherical particle could play the role of rolling spacers between the contacting surfaces,23 atomic scale surface phenomena30 are expected to play a larger role for much smaller (ca. 5 nm diameter) DNDs, especially when used as additives to liquid lubricants. Suggested mechanisms include changes in the lubricant viscosity and thermal transport properties, formation of protective surface films, and surface smoothing through polishing and/or filling of the spaces between the contacting asperities. In addition to the “boundary lubrication” regime, the benefits of such nanolubricants have recently been demonstrated in elastohydrodynamic lubrication, where a reduced surface roughness in the rolling contact was achieved through polishing by the nanoparticulates.31 Uptake of the nanoparticulates on surfaces could also alter the slip conditions at the fluid–solid interface, thus, changing the system's friction and wear attributes.32–35 Here, we examine the impact of the sign of the ND's surface charge on the tribological properties in aqueous solutions, as virtually all the dispersed nanoparticulates require surface charge to prevent coagulation and the eventual precipitation from a solution. At the macroscopic scale, both positive and negatively charged NDs are observed to reduce the friction coefficients, with surface polishing suggested as a possible mechanism28,36 despite the fact that surface smoothening in general does not necessarily result in lower friction coefficients.32 Further investigations are therefore necessary to establish the underlying physical mechanisms.
To be effective as lubricant additives, nanoparticulates must also exhibit long-term colloidal stability. Stability of such dispersions is known to be determined by the surface charge density that is typically characterized by the electrokinetic potential, or “zeta-potential”. It is the electrostatic repulsive force that prevents the dispersed nanoparticulates from coagulation and consequent precipitation. Typically, a zeta potential of 25–30 mV (positive or negative) is sufficient to electrically stabilize a colloid.37 DNDs can be produced with both positive and negative zeta potentials, thus, allowing for investigating the effects of the sign of the surface charge on tribological performance. Moreover, DND surfaces can be functionalized with a number of chemical groups to vary the zeta potential without compromising the diamond core properties.18,29,38–43 Functionalization so as to produce DNDs with like-charged surfaces does not, however, guarantee that the DNDs will exhibit a mutual repulsion. There are numerous literature reports of an attraction between like-charged colloids in solutions and/or other environments where the charge rearrangements are possible when the particles approach each other.44–47
Here we report on a Quartz Crystal Microbalance (QCM) study of surface uptake and the nanotribological properties of DNDs ca. 5–15 nm in diameter dispersed in water. QCM is an established tool for probing the tribological performance of material–liquid–nanoparticulate systems.32,33,35,48,49 It consists of a quartz single crystal that is electrically driven at its resonance frequency f0 in a transverse shear motion (Fig. 1). Changes in the resonant frequency, δf, and the inverse quality factor, δ(Q−1), of the crystal are reflective of the material uptake and the mechanical dissipation properties (e.g., flexibility, sliding friction, density) of materials deposited onto its surface electrodes and/or drag forces and slip lengths of the fluids by which it may be surrounded.
QCM measurements were performed with a crystal whose gold surface electrodes were immersed in aqueous suspensions of positively and negatively charged DNDs. As illustrated schematically in Fig. 1, it is anticipated that mobile conduction electrons within the gold electrode may react differently to the uptake of the negatively and the positively charged DNDs. Specifically, an electrostatic attraction force between the positively charged DNDs and the negatively charged electrons is expected to result in relatively strong adhesion compared to a neutral nanodiamond. The negatively charged DNDs are expected to be relatively poorly bound to the surface and possibly reduce the friction through a reduction of drag forces with the surrounding liquid. Measurements were also performed with stainless steel ball bearings loaded onto the upper gold QCM electrode in the presence of a water layer containing either positively or negatively charged DNDs (Fig. 2). Our experimental data demonstrate profound changes in the tribological performance at both nano- and macroscopic scales in the presence of DNDs, and also highlight significance of the electrostatic phenomena as the tribological parameters were found to be strongly affected by the sign of the DNDs' zeta potential.
![]() | ||
Fig. 2 Photograph of the QCM holder held horizontally with sixteen ball bearings loaded onto its surface electrode. |
Ball bearings, 316 stainless steel and 5/32′′ in diameter, model 316-5/2, were purchased from Bearing Ball Store (Orlando, FL).
DNDs were provided by the International Technology Center (Raleigh, NC). The samples were prepared by a detonation of an oxygen-deficient explosive mixture of trinitrotoluene with hexogen (40:
60 wt%) in a closed steel chamber using ice as a cooling media.18 The product, detonation soot, was a mixture of up to 75% diamond particulates with other carbon allotropes, as well as metallic impurities. DNDs were purified by oxidation of the soot in an ozone-enriched air at 150–200 °C over 72 h.38 The residual content of incombustible impurities in DNDs was estimated as ≈0.8 wt%. After the ozone purification, the resulting DND powder was light-grey in color. DNDs (10 wt%) were suspended in deionized (DI) water by sonication. Consequently, the suspensions were processed in a Retsch (Haan, Germany) planetary mill for 4 h using 100 μm zirconia beads. After the milling the product was treated in HF to remove contaminations from zirconia, washed with DI water, and re-suspended in DI water at 1 w/v% by sonication. Centrifugation at 25
000 × g for 2 h was used to extract DND particles with 5 ± 2 nm in diameter as measured by dynamic light scattering (DLS) and zeta potential ζ ≈ −45 mV at neutral pH (sample (−S1), Table 1). An additional centrifugal fractionation was employed to separate DNS fractions with 10 ± 2 and 15 ± 2 nm diameters (samples (−S2) and (−S3), respectively, Table 1).
Sample | DND preparation | Diameter (nm) | Zeta (ζ) potential, mV |
---|---|---|---|
(−S1) | As prepared | 5 ± 2 | −45 ± 2 |
Stored ca. 2 years | 10.7 ± 3.3 | −25.1 ± 1.6 | |
Sonicated after ca. 2 years | 5.3 ± 2.5 | −41.8 ± 2.8 | |
(−S2) | As prepared | 10 ± 2 | −35 ± 2 |
Stored ca. 2 years | 25.1 ± 7.4 | −26.0 ± 0.8 | |
Sonicated after ca. 2 years | 17.2 ± 5.1 | −21.5 ± 2.5 | |
(−S3) | As prepared | 15 ± 2 | −40 ± 2 |
Stored ca. 2 years | 17.1 ± 5.3 | −22.1 ± 1.1 | |
Sonicated after ca. 2 years | 16.9 ± 4.9 | −23.4 ± 1.6 | |
(+S1) | As prepared | 5 ± 2.0 | +35 ± 2 |
Freshly sonicated | 8.4 ± 2.5 | +33.4 ± 2.1 |
DNDs with a positive zeta potential (ζ ≈ +35 mV) were produced by a reduction reaction.38 A suspension of the 5 nm DND with ζ ≈ −45 mV was initially air-dried. The dry powder (5 g) was added to a round-bottom flask and further dried using a Schlenk line and at least three nitrogen purge-and-refill cycles. Consequently, 20 ml of degassed anhydrous tetrahydrofuran (THF) and 50 ml of 2.0 M solution of LiAlH4 in THF was added by a cannula. The sample was stirred under a nitrogen atmosphere at room temperature overnight. The reaction was quenched by a drop-wise addition of 1 M HCl, which solubilized the lithium and aluminum, and then pH of the solution was adjusted back to neutral. The product was collected by centrifugation and rinsed several times with water followed by a re-suspension in DI water by sonication and fractionation by centrifugation at 25000 × g for 2 h to extract 5 ± 2 nm primary particles. While the 5 nm fully de-agglomerated particles were used for the reduction reaction, during the functionalization some of the particles formed larger agglomerates requiring an additional fractionation to separate the 5 ± 2 nm primary particles (sample (+S1), Table 1). All DNDs were stored as 5 w/v% colloids in DI.
Immediately before the QCM measurements, DND suspensions were sonicated for 20 min using an Aquasonic Model 75D (VWR International, Radnor, PA) bath sonicator at 23 °C. For Dynamic Light Scattering (DLS) measurements the suspensions were diluted twentyfold to reduce light scattering and drawn into a polystyrol/polystyrene cuvette (10 × 10 × 45 mm3). The size distributions and ζ-potential were measured at 25 °C using Zetasizer (Nano-ZS series, Malvern Co, U.K).
QCM measurements were carried out at room temperature in a glass beaker containing 150 ml of DI water. Temperature was further stabilized by placing the beaker into a Styrofoam™ container. After fully immersing the QCM holder into water and an initial stabilization of the QCM frequency, 1 weight% DND aqueous suspension was added portionwise (5 portions of 1.05 ml each at approximately 30 min intervals) via a pipette. Each portion contained 1.05 mg of DNDs. The color of the aqueous suspension became uniform in ca. 30 s after the injection indicating an effective mixing.
For the experiments with stainless steel ball bearings, the QCM crystals with gold electrodes were freshly cleaned by the procedure described in the Section 2.2, placed in the Teflon holders in air, and then covered with 1.5 ml of either pure DI water, or an aqueous suspension containing positively or negatively charged DNDs. Sixteen ball bearings were placed atop the electrode in a pattern displayed in Fig. 2 and the QCM response was recorded. The order in which water and ball bearings were added atop the electrode did not impact the QCM final response.
![]() | (1) |
If the adsorbed film slips on the QCM surface in a response to the oscillatory motion, the magnitude of the frequency shift δffilm will be lower than that given by the eqn (1) for the rigidly attached film. There will also be a decrease in the QCM's quality factor, Q, since the friction associated with the film's sliding results in additional energy dissipation. For a film characterized by a frictional force per unit area F/A = ηv, where v is the sliding speed, a characteristic “slip time”, τ = ρ2/η parameterizes the strength of the friction coefficient η and can be inferred from the relation δ(Q−1) = 4πτ(δf) if the shifts in both frequency and the dissipation associated with the presence of the film are monitored.53
For a QCM immersed in a fluid with bulk density ρ3 and viscosity η3, one would also observe δf and δ(Q−1) shifts from additional viscous drag forces and an increased inertia of the oscillator that under no-slip boundary conditions are given by:54
![]() | (2) |
Therefore an immersion of one side of f0 = 5 MHz QCM in water at room temperature (ρ3 = 1 g cm−3, η3 = 0.01 poise) results in δf = −714 Hz drop in the resonant frequency and an increase of δ(Q−1) = 2.85 × 10−4 in the dissipation. For a QCM with quality factor Q = 50000 in air this corresponds to a drop to Q = 3280 after an immersion in water.
Although the viscous drag forces on the QCM electrode are mechanical in nature, a decrease in Q is manifested as an increase in the series resonant resistance Rm of the QCM resonator that can be measured electrically. For a QCM electrode exposed to a fluid from one side under non-slip conditions:54
![]() | (3) |
A rigidly adhering monolayer of nanoparticles on a planar surface is expected to lower f0 by an amount given by eqn (1) with no additional shift in Q if the no-slip boundary condition is satisfied. If, however, the QCM surface is not perfectly planar, or the surface roughness and/or slip conditions at the boundary change upon the nanoparticles' uptake (surface polishing or filling by nanoparticles resulting in a smoother topology, heterogeneous adhesion of the nanoparticles yielding a rougher surface, nanoparticles sliding at the interface, etc.), then the observed δf and δ(Q−1) shifts would reflect details of the underlying physical mechanisms.
The smallest negatively charged DNDs (−S1) yielded remarkable resilience to the long term storage: while average diameter of 5 ± 2 nm as-prepared particles increased to d = 10.7 ± 3.3 nm after ca. 2 years in storage, the initial diameter was fully recovered after sonication (5.3 ± 2.5 nm, Table 1). After the sonication, the zeta-potential ζ = −41.8 ± 2.8 mV also recovered to ζ = −45 mV specified for as-prepared particles. An interesting feature of the DND suspensions is that the particle size distributions after ca. 2 in storage either before or after the secondary sonication did not trend with the original as-prepared DNDs with the exception of sample (−S1). We therefore denote the samples according to their original sizes (Table 1).
For DND suspensions stored for ca. 2 years the secondary sonication decreases the average particle diameters for all but the (−S3) sample. However, the zeta potential did not change significantly except for the smallest 5 nm DNDs (−S1) which exhibit the highest magnitude of the zeta potential among all the samples studied. The general trend here is the smaller the DND, the higher the absolute value of the zeta potential. Fig. 4 shows the absolute value of the average zeta potential vs. average DND diameters for both negatively and positively charged particles. The zeta potential increases with decreasing particle diameter. This might be attributed to a higher charge density at the surface in smaller NDs that is essential for long-term stability of nanoparticle dispersions.37
The QCM response was strikingly different for DNDs of opposite electrical charges. Fig. 5 shows characteristic frequency shift and mechanical resistance data with time t = 0 set to the moment of the first addition of DNDs and subsequent additions approximately every 30 minutes thereafter. Notably, observed changes in the frequency and the mechanical resistance cannot be attributed to an increase in viscosity of the liquid surrounding the QCM (cf. eqn (2) and (3)), as doubling and quadrupling the DND content failed to result in an additional significant response (Fig. 5). The data show a clear trend in both resistance and frequency shift, with more uptake and more resistance for the large sized particles. In no case however does the uptake become comparable to that of the response of the positively charged sample denoted by +S1, whose original size as prepared was 5 nm, but was more likely to be closer to 10 nm in size as the measurements were recorded given the poor size stability of the positively charged samples.
Fig. 5 (lower) demonstrates that after ca. 3 h the frequency response of QCM stabilizes at δf ≈ −60 Hz for the positively charged sample, corresponding to approximately one monolayer of the positively charged d ≈ 12.3 nm DND particles attached to the electrode while the frequency shift associated with the addition of the negatively charged DNDs was negligible and even, in some cases, positive. The positive sign of δf is indicative of particles being present near the surface (since there was an impact on Rm), but exhibiting a high degree of slip and, thus, no significant increase in the inertia of the oscillator due to the negligible mass loading.
The mechanical resistance Rm of the QCM–liquid interface meanwhile increased by approximately 3 Ω for the positively charged DNDs and decreased by approximately the same 3 Ω when negatively charged 5 nm DNDs were injected. The experiments clearly revealed a dramatic influence of the DND surface charge arising from various chemical groups on the nanoparticles' surface38 on the system's nanotribological response. A possible explanation of the observed changes in f and Rm is that the there is a strong electrostatic attraction between positively charged DNDs and the gold electrode surface while the negatively charged particles are physically adsorbed to the surface but also undergo significant slip in response to the lateral oscillations of the QCM. The effect clearly weakens for negatively charged nanoparticles whose zeta potentials are significantly smaller (Table 1). This scenario is depicted schematically in Fig. 1, where mobile conduction electrons within the gold electrode are attracted or repelled from the surface, depending on the charge of the adsorbed nanodiamonds. This scenario would also explain the higher frictional drag forces observed for the surrounding liquid, as the surface topology could be more corrugated with bound, and more rigidly associated DNDs at the solid–liquid contact than the bare gold electrodes.
In some experiments the QCM response reached the values as high as δf = ±200 Hz. Typical values for the QCM response upon immersion of one of the gold electrodes in liquid followed by loading of sixteen close-packed ball bearings are summarized in the Table 2. This is consistent with the mobile negatively charged nanodiamonds being displaced from the tribological contact, since the response for water with negatively charged ND is quite similar to pure water alone. In particular, the frequency increases when the ball bearings are placed on the electrode, indicating that contact stiffness effects are dominant over mass loading effects.55,56 The reverse is true however for the case of positive nanodiamonds, where mass loading effects as reflected by negative frequency shifts are dominant. This is consistent with the positive nanodiamonds remaining within the tribological contact (Table 2).
f (Hz) | δRm (ohm) | |
---|---|---|
Pure water | +102 ± 22 | 8 ± 20 |
Water with −ND | +76 ± 28 | 13 ± 22 |
Water with +ND | −9 ± 2 | 6 ± 20 |
For the macroscopic contacts with the gold electrodes, the negatively charged nanodiamonds appeared to be entirely displaced from the contact, while the positively charged nanodiamonds were not.
Overall, the study shows the potential of QCM to study tribological properties of nanoparticle dispersions and nanodiamonds in particular.
This journal is © The Royal Society of Chemistry 2015 |