Desorption of carboxylates and phosphonates from galvanized steel: Towards greener lubricants

This paper studies the removal of chemisorbed carboxylates and phosphonates from TiO2‐coated galvanized steel using NaOH(aq). XPS and FTIR data show that NaOH(aq) is effective at desorbing these species and so is an alternative to gas phase processes (eg, plasma cleaning). Tribological investigations show that NaOH(aq)‐treated surfaces show reduced friction and wear, relative to the “as‐received” galvanized steel. This is ascribed to carbonate (present as an impurity in NaOH) that adsorbs to the surface of the substrate during NaOH(aq) immersion. Carbonate removal through sonication in water generates surfaces that show friction similar to “as‐received” galvanized steel. This work is useful in areas (eg, automotive manufacturing), where the effective removal of lubricants following tribological contact is key to subsequent paint adhesion.

substrates. In addition, despite the large number of studies reporting SAM formation, there are relatively few reports of how SAMs can be removed from surfaces. The few reports there are describe thiol removal from Au using gas-phase processes (eg, plasma cleaning 8 and ozonolysis 17,18 ) or electrochemical methods. [19][20][21] By comparison, there are few reports describing carboxylate or phosphonate SAM removal from metal substrates although the removal of carboxylates from TiO 2 surfaces has been achieved through de-esterification, using bases like NaOH and Bu 4 NOH. 22 This study investigates the desorption of chemisorbed lauric (R 12 C) and dodecylphosphonic acid (R 12 P) from TiO 2 -coated automotivegrade galvanized steels using NaOH (aq) . Detailed surface characterization (X-ray photoelectron and infrared [IR] spectroscopy) allied to water contact angle (WCA) measurements to study surface wetting have been combined with scanning electron microscopy to study how NaOH (aq) treatment affects the galvanized steel surface. These data have been correlated with the tribological properties of the functionalized and NaOH-treated surfaces using linear friction testing (LFT) and confocal microscopy. In order to study the effectiveness of NaOH (aq) as a method of removing these chemisorbed species, we have evaluated this method against O 2 plasma cleaning. Exposure of surface-adsorbed species to O 2 plasma has been shown to be an effective method of removing compounds from surfaces. 8 Consequently, it was used as comparison to study the efficacy of the NaOH (aq) treatment. using an isopropanolic solution of Ti (O i Pr) 4 (100 mM) for 30 seconds before drying in air for 1 minute as described previously. 23,24 Samples were then immersed in 100 mM isopropanolic solutions of dodecanoic acid (R 12 C) or dodecane phosphonic acid (R 12 P) for 30 seconds and allowed to air dry. Physisorbed species were removed from the surfaces through rinsing the coated substrates with acetone for several minutes. Bands ascribed to physisorbed species (eg, C═O and O) were not observed after rinsing using IR spectroscopy. This indicated that the residual material was chemisorbed onto the surface. Desorption of chemisorbed R 12 C and R 12 P was achieved through immersing the acetone-washed surfaces in 100 mM NaOH (aq). Selected samples were then sonicated in deionized H 2 O for 1 minute followed by drying in air. Plasma cleaning was carried out using a radio frequency induced O 2 plasma in an Electronic Diener plasma cleaner. Before cleaning, the chamber was placed under vacuum (< 1 mbar) before the O 2 was injected (pressure < 10 mbar) and the samples exposed for 10 minutes on each side.
X-ray photoelectron spectroscopy (XPS) was studied using an Axis Supra XPS (Kratos Analytical) with a monochromated Al Kα source and large area slot mode detector (ca. 300 μm × 800 μm analysis area). Charge neutralization was used to limit differential charging, and the data calibrated with respect to the C 1s peak (284.8 eV).
Survey spectra (step size 1 eV, dwell time 0.1 second, pass energy 160 eV) were collected at three surface locations before highresolution spectra (step size 0.1 eV, dwell time 250 ms, pass energy 20 eV). Data were fitted using CASA software and Shirley backgrounds. Field emission gun scanning electron microscopy (FEG-SEM) was studied using a Hitachi S4800 at 1.0 kV. Confocal 3 | RESULTS AND DISCUSSION

| Surface characterization
The surface composition and morphology of the TiO 2 coated HDG-steel substrate have been in detail previously. 24 After deposition of R 12 C, the ATR-IR data show C-H stretching and bending bands of the alkyl chain, 25 along with asymmetric and symmetric carboxylate stretching bands for the linker group at 1540 and 1398 cm −1 , respectively 26-28 ( Figure 1A). The wavenumber gap between these carboxylate stretching bands suggests bridging mode coordination. 27 However, after immersion in NaOH (aq) , only very weak CH 2 symmetric and asymmetric stretching bands are observed after the R 12 C functionalized surface was immersed in the NaOH (aq) ( Figure 1B).
This suggests that almost all the R 12 C was desorbed but there may have been a trace amount of R 12 C which was not desorbed during the treatment. It was not possible to unequivocally confirm this because carboxylate stretching bands are coincident with the new and more intense carbonate asymmetric stretching band observed in the IR spectrum 29 ( Figure 1B). The trace CH 2 bands were not observed in the IR spectrum of the NaOH-treated surface after sonication in H 2 O, which could suggest that they were ascribed to a species that could be removed due to sonication. For the newly observed carbonate bands, these are ascribed to the ca. 2% impurity of Na 2 CO 3 that typically forms in NaOH during its manufacturing.
Hence, these data show that the substrate is altered so that carbonate species end up on the substrate surface during the base-catalyzed desorption of R 12 C in NaOH (aq) . The bands ascribed to carbonate stretching and bending modes (1422 and 879 cm −1 , respectively) are at similar positions to as in the IR spectrum of pure sodium bicarbonate. 26 Consequently, it is not possible to ascertain from the FTIR data ATR-IR spectra of the R 12 C functionalized surface A, after R 12 C deposition, B, after immersion in NaOH (aq) , and C, after immersion in NaOH (aq) and sonication in H 2 O. α = CH 3 asymmetric stretch, β = υ CH 2 asymmetric stretch, γ = υ CH 2 symmetric stretch, κ = carboxylate asymmetric stretch, η = CH 2 bend, ν = carboxylate symmetric stretch, ρ = υ CO 3 2− asymmetric stretch, and θ = υ CO 3 High-resolution XPS data of the Zn 3s and P 2p regions on the R 12 P functionalized surface A, after R 12 P deposition, B, after immersion in NaOH (aq) , and C, after O 2 plasma cleaning what carbonate species is on the surface; ie, whether it is a new carbonate phase. However, what is known is that the carbonate ion (CO 3 2− ) is a planar molecule with no alkyl chains like R 12 C or R 12 P.
Hence, if it does adsorb, it could either be perpendicular to the surface or in planar configuration. The SEM data ( Figure 4) do show rod-like features which do suggest a separate carbonate phase. Interestingly, it is known that carbonates can imbue lubricity when added as lubricant additives 30,31 which is in line with the later coefficient of friction testing for these surfaces.
In addition, in the 100 mM NaOH (aq) solution, the pH is 13 which greatly exceeds the pK a values for lauric acid (pK a = 5.3) or sodium bicarbonate (pK a1 = 6.4, pK a2 = 10.3) meaning that all ions are fully dissociated. In addition, the 100 mM concentration of hydroxyl ions from NaOH will greatly exceed the number of adsorbed molecules which helps drive the desorption process. At the same time, the ca.
A partitioning process between CO 3 2− (aq) and CO 3 2− (sorbed) will take place. Our previous work on dyeing metal oxides 22 shows that 2 mM is more than sufficient to drive partitioning towards adsorbed species. Thus, whilst chemical desorption of carbonate was not possible in NaOH (aq) , sonication of the NaOH (aq)-treated samples in H 2 O resulted in no carbonate stretching or bending bands in the ATR-IR spectra ( Figure 1C). Instead, the data are very similar to that of the "as-received" HDG substrate (ESI Figure 2). This shows that adsorbed carbonate could be removed from the surface through sonication. Whilst it is possible that sonication could selectively remove adsorbed carbonate, it is more likely that weakly held particles of surface zinc oxide are physically removed and the carbonate inevitably is removed along with this. This creates a pristine zinc metal surface which does not dissolve because the pH of the water is neutral and instead rapidly re-oxidizes to ZnO.
The IR spectrum for the R 12 P-treated sample (ESI Figure 3A) shows C-H stretching and bending modes from the alkyl chain of R 12 P, and asymmetric and symmetric P-O stretching bands (1156 and 1083 cm −1 , respectively) from the linker group. 3,32 However, the absence of a O stretching band at ca. 1220 cm −1 suggests the phosphonate chemisorbs by tridentate coordination as observed in previous studies. 23,24 After immersion of the substrate in NaOH (aq) , no R 12 P bands are observed, and only carbonate stretching and bending modes are present (ESI Figure 3b) as for the analogous R 12 C sample. This indicates that the phosphonate had desorbed from the surface and has been replaced by adsorbed carbonate.
Due to the ubiquitous presence of adventitious C on surfaces, it was not possible to unambiguously determine whether NaOH (aq) had desorbed R 12 C from the surface using XPS data. The positions of the O-C═O peaks in the XP spectra of an untreated R 12 C and an R 12 C sample after immersion in the NaOH (aq) were both observed at 288.9 eV. Consequently, it is not possible to differentiate whether the carbonaceous material remaining on the surface was R 12 C, carbonate, or some other form of adventitious carbon from the XPS data.
However, the atomic % values for C, Zn, and Ti observed for R 12 C samples after NaOH (aq) treatment were similar to the unfunctionalized substrate ( Table 1) which suggests that NaOH (aq) does desorb R 12 C from the surface, in line with the IR data. The Zn:C ratio of the NaOH (aq) -treated sample also changes to 0.9:1.0 from 0.7:1.0 for the R 12 C-treated sample which is a closer ratio to the untreated sample (1.2:1.0) but not identical which is ascribed to carbon from the surface carbonate which the IR data show is present after NaOH (aq) treatment (Table 1). The desorption of R 12 P was also further studied by XPS.
After R 12 P deposition, the XPS data (Figure 2A) show 2s and 2p phosphorus peaks at 191.3 and 133.7 eV, respectively. Notably, these values are in agreement with related studies that report phosphonate FIGURE 3 Water contact angle images for HDG substrate A, after initial cleaning, B, after R 12painitng C deposition, C, after R 12 P deposition, and D, after NaOH (aq) treatment binding on alloy and metal oxide surfaces. 3,34 After NaOH (aq) treatment, very weak phosphorus peaks are observed in the highresolution XP spectrum. However, the atomic percentage of P was observed to be 0.0 ± 0.0% using CASA XPS software which indicates that the amount of P remaining after the NaOH (aq) treatment is below the limit of detection of the software. This suggests that, at least the majority of, the chemisorbed phosphonates are removed from the surface during this treatment, in line with the IR data. Analysis of the C 1s XPS spectra was also performed. However, the line shapes of the C 1s peak envelopes of surfaces before and after the NaOH (aq) treatment were very similar. Consequently, this made it very difficult to elucidate the nature of the carbonaceous material present on the surfaces.
Water contact angle (WCA) data of the HDG substrate ( Figure 3 and Table 2) were 63.2 ± 6.2°, which increases to values between 105°and 115°following chemisorption of R 12 C or R 12 P as reported in previous studies. 3,25 After NaOH (aq) treatment, the surfaces became substantially more hydrophilic (WCA < 50°) which is ascribed to the presence of surface adsorbed carbonate. Carbonate is a charge ion with polarized carbon-oxygen bonds which will hydrogen bond to water reducing the surface energy and increasing hydrophilicity.
Finally, after sonication in H 2 O, the WCA data for either R 12 C or R 12 P samples were very similar to the untreated HDG substrate suggesting that both chemisorbed R 12 C and R 12 P, and carbonate had been removed the surfaces. Surface roughness data have also been measured using confocal microscopy for these treatments ( Table 2).
The data do not show any substantial changes between the treatments, and so this is not thought to be a major influence on the hydrophobicity or surface lubricity of these samples.
SEM of the HDG steel substrate before and after immersion in NaOH (aq) was studied to investigate to effects of the NaOH (aq) solution on the zinc galvanic layer of the substrate. The data show that the surface of the NaOH (aq) -treated substrate ( Figure 4B) is substantially rougher than the as-received HDG steel ( Figure 4A). This suggests that, in addition to desorbing R 12 C and R 12 P, when the surface is treated with NaOH (aq) , the amphoteric surface ZnO dissolves. This exposes the underlying zinc metal which is also dissolved by the NaOH (aq) solution. AAS was used to further study this. The zinc concentration in the initial NaOH (aq) solution was 275 μg Zn L −1 which is ascribed to trace impurities during its manufacture. This zinc concentration increases to 1404 μg Zn L −1 for R 12 C and 1704 μg Zn L −1 for R 12 P (ESI Figure 4). Taking into account the increase in zinc concentration and the 100 mL of NaOH (aq) used, this suggests that 112.9 μg of Zn was removed from the R 12 C sample and 142.9 μg of Zn from the R 12 P sample. This confirms that NaOH (aq) removes Zn from the HDG steel through etching, and this affects the surface morphology of the material, in line with previously reported data. 35 Assuming a 99.7% zinc galvanic coating of 7 to 10 μm, the total zinc present on 1 × 2 cm samples will be ca. 10 to 14 mg. As such, these data suggest that ca. 10% of the galvanic zinc layer is removed during this etching process. Rod-like features (labelled α but present all over the surface in Figure 4B) were also observed in some areas of the NaOH (aq) -treated galvanized steel.
XRD data of these samples showed no new diffraction peaks relative to the untreated HDG substrate (ESI Figure 5), suggesting that this material was either poorly crystalline or amorphous. Alternatively, it could be re-deposited Zn and/or ZnO which would not show up as extra peaks because these phases were already present in the HDGsteel substrate. Figure 4C then shows SEM data for R 12 C-coated HDG which has been acetone washed and then NaOH treated. The data do not show the same rod-like features which are observed on the NaOH-treated HDG. Instead, the NaOH-treated R 12 C surface appears more like TiO 2 -coated HDG ( Figure 4A) albeit slightly more textured which would be expected after a strong alkali treatment of a zinc-rich surface.
By comparison, the WCAs of the R 12 C and R 12 P functionalized surfaces after O 2 plasma cleaning were 58.4 ± 4.0°and 27.2 ± 16.9°.
These low WCA values show that the surface is changed by the O 2 plasma and suggest that more organic material is removed presumably by oxidation to produce surface largely composed of metal oxide. 8,36  1 Atomic percentages (calculated using sensitivity factors) and Zn:C ratios calculated the combined areas of the Zn 2p1/2 and Zn 2p 3/2 peaks versus the C 1s peak from XPS on selected surfaces (data are average of three analyses per surface ± standard deviation). Trace contaminants (Si, N, Na, Ca) on TiO 2 -coated HDG and on the R 12 C surfaces were ascribed to contamination from laboratory gloves 33  In line with this, the IR spectra of the plasma-cleaned surfaces show very weak C-H stretching bands, indicating that the majority of the chemisorbed organic material had been removed by the plasma (ESI Figure 6). However, XPS data of the NaOH-treated R 12 P sample still show the presence of P 2s and P 2p photoelectron peaks, suggesting that the phosphonate linker group remains on the surface ( Figure 2C). The ratio of the peak areas of the P 2p to the adjacent Zn 3s peak (P:Zn) of this sample was observed to be 4.8:1. An untreated R 12 P sample was observed to display a P:Zn of approximately 3.1:1, whereas an R 12 P sample immersed in the NaOH (aq) displayed a P:Zn of 0

| Tribological testing
The tribological properties of the surfaces were studied using LFT.
As discussed in our previous studies, LFT is an aggressive tribological test that is designed to simulate the sliding conditions galvanized steels experience during sheet metal forming operations. 22,23 The data show the average coefficient of friction (μ) of the TiO 2 -coated HDG steel substrate is 0.26 ± 0.06. Treatment of the substrate with either R 12 C or R 12 P reduced μ to 0.11 ± 0.01 ( Figure 5A,B). This large reduction is ascribed to alkyl chains of R 12 C or R 12 P acting like molecular springs or brushes during tribological contact. 1,11 Interestingly, the μ values of the R 12 C and R 12 P functionalized surfaces did not increase significantly after immersion in NaOH (aq) . This suggests that the carbonate that adsorbs to the surface during caustic treatment provides a lubrication effect and acts as an effective barrier against surface asperity contact. To study this further, LFT measurements were also performed on NaOH (aq) -treated surfaces that had been sonicated in H 2 O (which our ATR-IR data show removes the adsorbed carbonate). The LFT data show that the average μ value of the surfaces increases substantially after sonication to between 0.17 and 0.25 ( Figure 5A,B). These data do suggest that carbonate is responsible for the lower μ values and so does imbue surface lubricity in its own right. In line with this, confocal microscopy shows substantially deeper scratches on the sonicated surfaces relative to samples that were solely subjected to NaOH (aq) (Figure 5C-E). This severe galling behaviour 14 Table 3 shows surface roughness data.
Before LFT, the arithmetic mean surface roughness (S a ) varies from  during R 12 C and R 12 P desorption. Importantly, less than 10% of the galvanic layer is dissolved in this process so there will be little effect on the corrosion protection. But this also means that the dissolved Zn 2+ (aq) should be easily recoverable either by electrolytic recovery or precipitation. At the same time, the surface morphology will change through NaOH etching, but NaOH is already widely used in automotive manufacturing, eg, to remove surface-adsorbing additives in current metal forming lubricants (albeit the NaOH is typically at a lower concentration). Additionally, confocal microscopy shows that the macroscopic surface roughness of the NaOH-treated surfaces is similar to the as-received HDG. Because macroscopic surface roughness has a greater impact on paint appearance than roughness of smaller length scales, we do not anticipate any problems with paint adhesion or surface finish for SAM-processed substrates.
In a further interesting observation, whilst the NaOH (aq) treatment desorbs alkyl carboxylates and phosphonates, these are replaced by adsorbed carbonate which itself imbues similar lubricity to the substrates. This is important because while sodium carbonate is a similar price (£99 for 5 kg) to R 12 C (£70 for 5 kg), it is much cheaper than R 12 P (£54 for 1 g)-price data from Aldrich. This would also be a good end use for fossil fuel-related CO 2 removed from the atmosphere into water and, in the context of the potential impact of green lubricants, the processing methodologies described in this paper should be compatible with existing processes on assembly lines without requiring any additional steps. In addition, much less SAM material needs to be deposited compared with oil-based lubrication processes and, as stated earlier, the SAM can be recovered and re-used.
However, the mode of action of SAMs is very different to traditional lubricants because it arises from a thin, solid film on the surface rather than a liquid. Whilst a solid lubricant brings benefits in terms of reduced liquid waste, it cannot remove debris or frictionrelated heat from the contact area in the same way as a liquid can.
However, this can be overcome by using water with the SAMs during forming rather than running them dry. As the WCA data show, the SAM surfaces are hydrophobic, and further LFT data show they work just as effectively when combined with water as when they are used dry (eg, the μ of the R 12 C samples run dry were 0.11 ± 0.005, whereas with water the μ of R 12 C was 0.12 ± 0.01). By comparison, the lubricity for the carbonated surfaces is believed to arise from carbonate acting either as an electrostatic (due to high polarity) and/or as a physical barrier coating. 30,31 The rod-like features on the NaOH (aq)-treated surface ( Figure 4B) suggest that these could contribute to lowering the coefficient of friction. This is further confirmed by AFM data (ESI Figure 15) which shows no particles on the as-received DX56 surface, then many particles of 300 to 400 nm size on the NaOH (aq) -treated surface and then that these particles are removed by sonication in water.

| CONCLUSIONS
This paper has shown that, not only can chemisorbed alkyl carboxylates or phosphonates act as an inherent, solid lubricants on galvanized steel, but these compounds can be easily desorbed by briefly dipping the substrates in aqueous NaOH using processes which are compatible with current steel manufacturing processes. Whilst detailed characterization confirms the desorption of the vast majority of the alkyl carboxylates or phosphonates, the coefficient of friction of the substrates does not increase as expected. Instead, not only is a significant adsorption of carbonate observed, but this new layer also appears to imbue surface lubricity. The fact that such a simple and low cost adsorbent as carbonate can influence surface properties is both important and surprising. Also, our data show that NaOH (aq) only etches 10% of the galvanic zinc layer which does not affect surface roughness sufficiently to affect painting or surface finish.