Water-based lubrication provides cheap and environmentally friendly lubrication and, although hydrophilic surfaces are preferred in water-based lubrication, often lubricating surfaces do not retain water molecules during shear. We show here that hydrophilic (42° water contact angle) quartz surfaces facilitate water-based lubrication to the same extent as more hydrophobic Si crystal surfaces (61°), while lubrication by hydrophilic Ge crystal surfaces (44°) is best. Thus surface hydrophilicity is not sufficient for water-based lubrication. Surface-thermodynamic analyses demonstrated that all surfaces, regardless of their water-based lubrication, were predominantly electron donating, implying water binding with their hydrogen groups. X-ray photoelectron spectroscopy showed that Ge crystal surfaces providing optimal lubrication consisted of a mixture of –O and =O functionalities, while Si crystal and quartz surfaces solely possessed –O functionalities. Comparison of infrared absorption bands of the crystals in water indicated fewer bound-water layers on hydrophilic Ge than on hydrophobic Si crystal surfaces, while absorption bands for free water on the Ge crystal surface indicated a much more pronounced presence of structured, free-water clusters near the Ge crystal than near Si crystal surfaces. Accordingly, we conclude that the presence of structured, free-water clusters is essential for water-based lubrication. The prevalence of structured water clusters can be regulated by adjusting the ratio between surface electron-donating and electron-accepting groups and between –O and =O functionalities.
Water-based lubrication strategies offer a relatively cheap and environmentally friendly way of lubrication [1–3] and have been extensively considered [4–7] for use in technological and biomedical applications [5,7,8]. Water molecules, however, are much more difficult to retain on lubricating surfaces than hydrocarbon-based lubricating molecules , which has hitherto impeded water-based lubrication from general application.
In hydrocarbon-based lubricated machineries, contact pressures can range up to 1000 MPa  and accordingly, in order to facilitate water-based lubrication, lubricating surfaces have been modified in various ways to more strongly attract and retain water molecules during lubrication. Polymer brushes [7,11] or ion-modified surfaces  are known to retain high numbers of water molecules that are difficult to remove during shear and can withstand contact pressures up to 7.5 MPa . In water-based lubrication, water molecules act as nano-ball-bearings to lower friction. Similar hydration shells can also form around hydrophilic functional groups in surfactants  or liposomes  to facilitate lubrication. The lubricating properties of water have been mostly studied on hydrophilic surfaces, as these are generally considered to more strongly attract water molecules and facilitate lubrication than hydrophobic surfaces. Yet, surface hydrophilicity solely expresses the degree of spreading of a water droplet on a surface, but it does not indicate the degree to which water molecules are free or bound to a surface and, when bound, whether they are bound with their electron-donating oxygen groups or with their two electron-accepting hydrogen groups. Differences in the degree of binding of water to lubricating surfaces and the structure of bound water may have an influence on lubrication, but these have not yet been studied in any depth during lubrication because of the lack of appropriate instrumentation.
The Tribochemist is a combination of a tribometer and a Fourier transform infrared spectrometer (FTIR). The instrument is based on a standard pin-on-plate tribometer , in which the plate has been replaced by an attenuated total reflection (ATR) crystal made of either Ge (germanium) or Si (silicon). This set-up enables real-time and in situ ATR-FTIR spectra and friction data to be obtained simultaneously in order to monitor changes in the structure and composition of lubricating films under tribological conditions [14,15].
The aim of this paper is to monitor the structuring of water on and near smooth Ge- and Si-ATR crystals under lubricating conditions using the Tribochemist and relate the amount of bound and free water and their structuring with the surface properties of the lubricating surfaces.
2. Material and methods
The Ge- and Si-ATR crystals (72 × 10 × 6 mm, angle of incidence 45°) and quartz slides (76 × 25 × 1.5 mm) were commercially obtained from Pike Technology, Fitchburg, Wisconsin, USA, and Lowers Hapert Glastechniek, Hapert, The Netherlands, respectively. All surfaces were cleaned in ultrapure water and isopropanol with a cotton stick and air-dried at room temperature, overnight, before the experiments.
Deionized water used in this study was purified by a Sartorius arium 611 DI water purification system (Sartorius AG, Goettingen, Germany), yielding ultra-pure water with pH 6.9, resistivity at 17.8–18.1 MΩ·cm and the total organic content was less than 4 ppb.
2.2. Surface characterization
The arithmetic averages with standard deviations of the surface roughness Ra of the different surfaces were measured in triplicate using a white light profilometer (SCANTRON Proscan 2000; Monarch Centre, Taunton, UK). The range, axial resolution and lateral resolution of the profilometer settings were 0.3 mm, 0.1 µm and 4 µm, respectively. An area of 2 × 0.1 mm (step length X-axis 0.008 mm and Y-axis 0.001 mm) at the centre of each crystal surface was scanned at a 300 Hz scan rate.
Contact angles with different liquids were measured on cleaned Ge and Si crystal and quartz surfaces with water, formamide, α-bromonaphthalene and methyleneiodine. The contact angles were recorded by a fixed camera about 5 s after placing a 0.5 µl liquid droplet on a surface. The droplet contours were detected by grey-value thresholding and contact angles were calculated from the digitized contours using home-made software. The contact angles on each surface were converted to a Lifshitz–van der Waals (γLW) and acid–base (γAB) surface free energy component, while the acid–base components were split up into an electron-donating (γ−) and an electron-accepting (γ+) parameter  according to:
where γLW is the Lifshitz–van der Waals surface free energy component and γ− and γ+ are the surface free energy parameters, respectively, of the four liquids applied or the material surface (see subscripts). The total surface free energy is denoted as γ, while θ represents the contact angles. Note that we have chosen to use two a-polar liquids (α-bromonaphthalene and methyleneiodine) in order to more reliably determine the Lifshitz–van der Waals component from the average value provided by both a-polar liquids.
The thickness of the oxide layer on Ge and Si crystal surfaces was determined using an ELX-02C ellipsometer (DRE-Dr. Riss Ellipsometerbau GmbH, Ratzeburg, Germany), equipped with a HeNe laser source (λ = 632.8 nm). Measurements were taken at an angle of incidence of 70°. The thickness of the oxide layer was calculated using the refractive indices of the crystals and their oxide layers. For bare Ge and Si, the refractive indices used were 4.06 and 3.44, respectively, whereas for the oxide layer on the Ge and Si crystals the respective refractive indices were 1.58 and 1.46. The measurements were averaged from 10-point measurements on each sample.
In order to determine the type of oxygen functionality present in the oxide layer on the Ge and Si crystal and quartz surfaces, samples were placed in an X-ray photoelectron spectroscope (XPS; S-probe; Surface Science Instruments, Mountain View, CA, USA), operated at a vacuum of 10−9 Pa. The X-ray (10 kV, 22 mA) beam was produced using an aluminium anode and had a spot size of 250 × 1000 µm. The wide-scan spectrum in the binding energy range of 1–1300 eV was measured at a resolution of 150 eV, after which narrow scans were made of the O1s binding energy peak in the range of 520–540 eV at a resolution of 50 eV. Next, the O1s peak was decomposed into two components with a full width at half maximum set at 1.75 eV due to oxygen involved in =O (O531.5 eV) and –O (O532.5 eV) functionalities (17); the resulting peak fractions were multiplied by the at% O measured in the wide scan to obtain the at% O531.5 eV and at% O532.5 eV.
2.3. Tribochemistry of water molecules
The Tribochemist is a novel in situ technique that provides information on the dynamics of adsorbed layers during friction . The Tribochemist is an integrated device consisting of an ATR-FTIR spectrometer (Cary 600 series FTIR Spectrometer; Agilent Technologies, Santa Clara, CA, USA) and a tribometer (sliding wear tester TR-17; Ducom Instruments Pvt. Ltd, Bangalore, India). The FTIR spectrometer is used for acquiring IR spectra of the surface layer and the tribometer measures the coefficient of friction (figure 1).
In the tribometer, a linear motion drive using a stepper motor (VEXTA Oriental Motor, model PK56 W; Oriental Motor Pvt. Ltd, Bangalore, India) enables reciprocating sliding of the polydimethyl siloxane (PDMS) pin (semi-hemispherical geometry, radius of 3 mm) that is loaded on the ATR crystal. A bi-directional load cell (Anyload model 108AA; Anyload Transducer Co. Ltd, Burnaby, BC, Canada) with a maximum load capacity of 5 N was used to measure the friction force Ff. The resolution of the load cell was 0.03% of the maximum load, i.e. 1.5 mN. For the current experiments, stroke length was set to 45 mm, sliding speed to 1 mm s−1, load to 450 mN and the duration was 10 min, as adjusted using the Winducom 2010 (Ducom Instruments Pvt. Ltd) software developed using the LabVIEW platform (National Instruments Corporation, Austin, TX, USA). The friction force data acquisition rate was fixed at 2 kHz. The coefficient of friction µ was subsequently calculated by using in which Fn is the normal load.
FTIR spectra were collected within the wavenumber range of 400–4500 cm−1 at a resolution of 4 cm−1, with one spectrum being averaged from 12 interferograms. Backgrounds were taken of dry Ge and Si crystal (Pike Technologies) before the addition of water. One millilitre of water was added on the crystal surface and an IR spectrum was taken prior to applying tribological conditions (control). Then IR spectra were taken during and immediately after the friction measurements. The IR laser was kept on during the entire measurement. An IR spectrum was also collected on a water sample on the same ATR crystal after 10 min IR irradiation in the absence of shear, serving as a control spectrum. Decomposition and fitting of the absorption bands in IR spectra were done using the Origin Pro v. 9.0 program (Origin Lab Corporation, Northampton, MA, USA).
3.1. Physico-chemical surface characteristics and coefficients of friction
Coefficients of friction were compared on Ge and Si crystals as well as on a quartz surface, although the latter could not be used in ATR-FTIR. First, the arithmetic average roughness of the surfaces was measured using profilometry and found to be similar on Ge and Si crystal and quartz surfaces (0.13 ± 0.02 µm, 0.12 ± 0.01 µm and 0.13 ± 0.01 µm for Ge and Si crystal and quartz surfaces, respectively). This ensures that possible differences in coefficients of friction are not due to differences in surface roughness.
Figure 2 shows the coefficients of friction for repetitive strokes on all three surfaces and their corresponding average root mean squared (RMS) values. The coefficient of friction on a Ge crystal surface is approximately three times lower (significant at p < 0.01; Student's t-test) than that on a Si crystal or quartz surface, which have similar coefficients of friction.
Next, we set out to relate the surface hydrophobicity/hydrophilicity by water contact angles of the three surfaces (table 1) with their coefficients of friction. The Ge and quartz surfaces were hydrophilic with similarly low water contact angles of 44 ± 2° and 42 ± 3°, respectively, while the Si crystal surface was most hydrophobic with a water contact angle of 61 ± 3°. By comparison with figure 2, water films on more hydrophilic surfaces (i.e. Ge crystal and quartz surfaces) need not necessarily provide better lubrication than hydrophobic surfaces, i.e. the Si crystal surface. However, according to surface thermodynamics, hydrophobicity/hydrophilicity is not uniquely determined by the water contact angle. Therefore, contact angles were also measured with formamide, possessing a different ratio of electron-donating and electron-accepting groups than water and two completely a-polar liquids [17,18]. The measured contact angles were subsequently converted into surface free energy components and parameters using the Lifshitz–van der Waals/acid–base approach [17,18] (see also table 1). Interestingly, like the coefficients of friction, the ratio between electron-donating and electron-accepting groups on an Si crystal surface (17.5) was virtually identical to the one on a quartz surface (21.9), while this ratio was significantly higher on a Ge crystal surface (40.5), possessing a threefold lower coefficient of friction than the other two surfaces. As all surfaces have a much higher electron-donating parameter than their electron-accepting one, water molecules must be predominantly bound with their hydrogen groups facing the surface. However, the degree to which water molecules are bound with their hydrogen groups to the Ge crystal will be much higher than for Si crystal and quartz surfaces. Therefore, it can be concluded that water is differently structured on a Ge crystal surface than on Si crystal and quartz surfaces.
These differences in structure of bound water were further related to the chemical composition of the different surfaces. The quartz surface possesses bulk-oxygen groups by nature, while both Ge and Si crystals possess spontaneously formed oxide skins that were found, using ellipsometry, to be almost three times thinner on a Ge crystal surface (0.30 nm) than on the Si crystal (0.86 nm).
The composition of the outermost oxide layer differed considerably among the three surfaces (see also table 1) and decomposition of the O1s electron binding energy peak in XPS (see figure 3 for an example of differently shaped O1s electron binding energy peaks) demonstrated that the oxide layer on a Ge crystal is composed of a mixture of =O and −O functionalities, while on an Si crystal and quartz surface solely −O was found (see the electronic supplementary material, figure S1, for spectra of the Ge and Si electron binding energy peaks). Comparison of the ratio γ−/ γ+ and the occurrence of =O (table 1) shows that electron-donating groups on the Ge crystal surface are due to =O functionalities , while Si crystal and quartz surfaces are composed solely of –O functionalities. Accordingly Ge crystal surfaces are composed of a mixture of GeO and GeO2 , while Si crystal surfaces possess only SiO2 .
3.2. Bound-water films and free water
FTIR-ATR absorption spectra for Ge and Si crystal surfaces in water showed absorption bands between 2500 and 4000 cm−1 indicative of stretching of the O–H bond in water molecules (note that no spectra could be taken on quartz surfaces). For Ge and Si crystal surfaces, the band could be decomposed into two characteristic absorption bands, identified through the second derivatives of their IR spectra (see the electronic supplementary material, figure S2), representative of bound and free water, as shown in figure 4a,b, respectively, revealing the wavenumber for bound water at 3309 cm−1 (Ge) and 3306 cm−1 (Si) and the wavenumber for free water at 3472 cm−1 (Ge) and 3488 cm−1 (Si). The bound-water band is at a lower wavenumber than the free-water band, because bonding with two hydrogen groups of a single water molecule to the crystal surfaces limits the vibrational freedom of the water O–H groups, leading to a lower wavenumber of the band. Accordingly, the data suggest (figure 4c) that water is slightly less strongly bound to the Ge crystal surface than to the Si crystal surface. In addition, as can be concluded from a comparison of the absorption band areas, fewer water molecules bind to the Ge than to the Si crystal surface. Since the ratio γ−/ γ+ is far above unity for Ge and Si crystal surfaces, this implies that on both crystal surfaces water will bind with its electron-accepting hydrogen groups. Owing to the possession of a mixture of =O and –O functionalities on more hydrophilic Ge crystal surfaces, water will bind less strongly and with structuring extending over thinner layers than on more hydrophobic Si crystal surfaces with only –O functionalities. Structuring of water is indeed suggested  to extend over 10- to 20-fold higher distances—up to 13 nm on more hydrophobic surfaces than on more hydrophilic ones.
Both for the Ge crystal as well as for the Si crystal, the evanescent IR wave penetrates about 500 nm into the adjacent fluid , which exceeds the maximum distance reported for the structuring of water on surfaces, regardless of their hydrophobicity . Hence, influences of the crystal surface properties will be reflected in the IR absorption bands representative of free water. Opposite to what was found for bound water, the absorption band for free water on the Ge crystal surface is located at a much lower wavenumber than on the Si crystal surface (figure 4d). This is likely to be due to the development of free, but internally bound, structured water clusters in the close vicinity of the Ge crystal surface, while surface-induced structuring of bound water extends much further into the bulk fluid on Si crystal surfaces. However, the absorption band areas for free water are approximately similar on both crystal surfaces, which is basically as expected since the amount of free water probed by FTIR is in the micrometre range.
3.3. Behaviour of bound-water films and free water under lubricating conditions
In order to examine whether the degree of structuring of bound and free water was impacted by lubricating conditions, FTIR-ATR absorption spectra were collected in the Tribochemist prior to, during and immediate after lubrication (figure 5).
Lubricating conditions have no impact at all on IR absorption characteristics of free water on either surface (figure 4c,d), but tend to loosen water binding on Ge and Si crystal surfaces (IR absorption bands shift to slightly higher wavenumbers, see figure 5a), without any impact on the amount of bound water probed (figure 5b).
In this paper, we reveal that hydrophilicity is not a sufficient condition of surfaces to allow water-based lubrication, since hydrophobic Si crystal surfaces with a high (61°) water contact angle had a similarly high coefficient of friction than a much more hydrophilic quartz surface (water contact angle 42°). Alternatively, a Ge crystal surface with a similarly low water contact angle of 44° to as a quartz surface provided much better lubrication. Further surface thermodynamic analyses demonstrated that only a hydrophilic surface (Ge crystal) with a high ratio of electron-donating over electron-accepting groups yielded a bound-water structure that provides low friction, as compared with a more hydrophobic Si crystal surface with a lower ratio of electron-donating over electron-accepting groups. Moreover, Ge crystal surfaces were composed of a mixture of –O and =O functionalities, while Si crystal and quartz surfaces solely possessed –O functionalities, causative of lower electron-donating surface free energy parameters of the surface to be lubricated . Comparison of FTIR-ATR absorption bands of the crystals in water indicated that water bound less strongly and in lower amounts to hydrophilic Ge than to hydrophobic Si crystal surfaces, on which water structures were formed over 10- to 20-fold longer distances than on Ge crystal surfaces. Oppositely, absorption bands for free water on the Ge crystal surface were located at a lower wavenumber and indicated a more pronounced presence of structured water clusters near the Ge crystal surfaces than near Si crystal surfaces.
In figure 6, we provide a translation of the above in a schematic manner for two hypothetical surfaces consisting solely of =O functionalities, such as Ge-crystal surfaces, or –O ones, such as Si crystal surfaces. The models depicted in figure 6 place a major emphasis on the development of structured, internally bound free-water clusters, as evidenced by the IR absorption wavenumbers of free water (figure 4d). Note that theoretically, free water has a higher absorption band wavenumber than bound water, but in the water clusters envisaged in figure 6 water molecules are only bound to each other with one hydrogen group, which is different from surface-bound water being bound with two hydrogen groups in one water molecule. This much more strongly impedes O–H vibration than in the internally bound water clusters shown in figure 6 . Clearly, more extensive, structured and internally bound free-water clusters in close vicinity to the surface to be lubricated act as nano-ball-bearings facilitating lubrication.
Our friction coefficients were measured with a PDMS pin. PDMS is a highly hydrophobic material. In tribopairs comprising hydrophilic–hydrophilic, hydrophobic–hydrophilic and hydrophobic–hydrophobic pairs , it was found that friction was least when a hydrophobic surface slid over a hydrophilic one. PDMS has zero electron-donating and electron-accepting surface free energy parameters [25,26] and hence our conclusions regarding the role of structured water, as related to the surface to be lubricated, are not influenced by the pin as it does not facilitate structuring of water molecules. Importantly, IR absorption characteristics of free water are not affected by lubricating conditions, while IR characteristics of bound water are only marginally affected.
Water-based lubrication offers a relatively cheap and environmentally friendly way of lubrication, but water molecules are much more difficult to retain on lubricating surfaces than hydrocarbon-based lubricating molecules, impeding general application of water-based lubrication. Although hydrophilic surfaces are generally considered to better retain water molecules than hydrophobic surfaces, the results of this study show that the presence of structured, free-water clusters acting as nano-ball-bearings are essential for water-based lubrication. The prevalence of structured water clusters can be regulated by adjusting the ratio between surface electron-donating and electron-accepting groups and between –O and =O functionalities. We believe that the existence of water clusters acting as nano-ball-bearings would unite all our experimental data, although other possible explanations might exist. Furthermore, although demonstrated for quartz, Ge and Si crystal surfaces, we believe that the conclusions of this study can be extrapolated to other materials, based on a characterization of their electron-donating and electron-accepting groups.
J.H. carried out the laboratory work and data analysis of all experiments except the XPS experiments. J.d.V. carried out the XPS experiments. All authors except J.d.V. contributed to the design of the study, interpretation of the results and preparation of the manuscript. All authors gave final approval for publication.
The authors declare no potential conflicts of interest with respect to authorship and/or publication of this article. H.J.B. is a director of consulting company SASA BV. D.H.V. is a manager of Ducom Instrument Europe B.V.
This study was entirely funded by UMCG, Groningen, The Netherlands.
The opinions and assertions contained herein are those of the authors and are not construed as necessarily representing the views of the funding organization or their respective employer(s).
We thank Prof. Jacob Klein from the Department of Materials and Interfaces in the Weizmann Institute of Science (Israel) for his advice in preparing this manuscript.
Electronic supplementary material is available online at https://dx.doi.org/10.6084/m9.figshare.c.3473631.
- Received July 12, 2016.
- Accepted September 8, 2016.
- © 2016 The Author(s)
Published by the Royal Society. All rights reserved.