Effect of hydrocarbon adsorption on the wettability of rare earth oxide ceramics

Condensation is observed frequently in our environment and routinely used in industry as an effective means of transferring heat. Water condensation on typical industrial condenser metal surfaces and their respective high-surface-energy oxides, e.g., CuO, Al₂O₃, and Fe₂O₃, results in the formation of a film of condensate that spreads over the condenser surface, termed filmwise condensation.¹ This filmwise mode of condensation imposes a thermal resistance across the film, which limits heat transfer. Conversely, water condensation on a low-surface-energy material, e.g., polytetrafluoroethylene (PTFE), parylene, and poly-(1H,1H,2H,2H-perfluorodecyl acrylate) (p-PFDA), results in the formation of discrete condensate droplets that, when under gravity-driven convection, shed as their size approaches the capillary length (≈2 mm for water), termed dropwise condensation.² The shedding of droplets refreshes the surface for renucleation and offers an improvement in heat transfer performance of 5 − 7x compared to filmwise condensation.³ 

State-of-the-art techniques to promote dropwise condensation rely on the application of low-surface-energy hydrophobic coatings to the condenser surface.^3,4 Coatings as thin as a monolayer (≈1 nm) of long-chain fluorocarbon molecules or fatty acids can be applied to induce hydrophobicity, but these are often not robust over extended periods of time and therefore unsuitable in industrial applications.⁵ Thicker polymer coatings, e.g., ≈20 μm of PTFE, have shown the potential to maintain robust hydrophobicity, but have a characteristically large thermal resistance that can negate the advantage gained by achieving dropwise condensation.³ More recently, plasma enhanced chemical vapor deposition (PECVD) and initiated chemical vapor deposition (iCVD) have been used to grow ultra-thin (<40 nm) conformal coatings of polymers on surfaces with success in achieving dropwise condensation.^5,6 However, the longevity of these ultra-thin coatings remains a question due to the lack of extended or accelerated testing to assess mechanical durability and long-term stability.

An alternative to the direct application of low-surface-energy coatings relies on surface contamination due to energetically favorable hydrocarbon adsorption, particularly on high thermal conductivity noble metals (i.e., gold and silver).⁷ These metals are wetting when clean, but reduce their surface energy by adsorbing hydrocarbons from air, enabling dropwise condensation when used as condenser surfaces. The robustness of this approach is well-documented, with one paper demonstrating continuous dropwise condensation on gold for over 5 years in a closed system.⁸ Unfortunately, the high price of noble metals prohibits this approach in practice.

Researchers have recently demonstrated rare earth oxides (REOs) as potential candidates for condenser surface coatings due to the their apparent intrinsic hydrophobicity⁹ and costs approaching 1% of gold.¹⁰ However, reported contact angles on REOs are inconsistent. Advancing contact angles ranging from 17° to 134° have been observed, with a study reporting 94°–134° on rough electroplated ceria coatings.¹¹ Meanwhile, another study reports 120° on a rough ceria membrane, but 17° on ceria after oxidation by heating cerium foil in air and 31° on a rough ceria membrane which has been sonicated in ethanol to destroy the nanostructure.¹² 

Furthermore, the underlying mechanism of REO hydrophobicity does not seem to be well-understood. The initially reported intrinsic hydrophobicity of REOs asks for a comparison with the debate in scientific literature regarding the intrinsic wettability of gold in the 1960s.⁹ Erb and Fowkes asserted that gold was intrinsically hydrophobic in 1964,^8,13 which Zisman contradicted the following year with experiments demonstrating that the contact angle on a gold surface with hydrocarbons desorbed and oxide removed (by heating in a hydrogen gas stream with <1 ppm hydrocarbons) was ≈0°.¹⁴ Though Erb initially disputed the claim,¹⁵ subsequent studies determined that gold is intrinsically hydrophilic but rapidly adsorbs hydrocarbons from the ambient environment, resulting in an increased contact angle.¹⁶ The idea of achieving hydrophobicity via hydrocarbon adsorption has been extended for a wide class of materials, including ceramic metal oxides^17,18 as well as pristine monolayer graphene.¹⁹ However, in the case of REOs such as ceria, while previous work has shown that methane adsorbs to the surface²⁰ and hydrocarbon adsorption increases on roughened surfaces with more available surface area,²¹ adsorption of hydrocarbons besides methane and the subsequent effect on contact angle have not been investigated. In this work, we show through experiments and modeling that REO hydrophobicity occurs due to a similar hydrocarbon adsorption mechanism observed previously on noble metals. To investigate adsorption dynamics under ambient conditions, we studied two REOs with different oxidation states, holmia (Ho₂O₃) and ceria (CeO₂), along with control samples of silica on a silicon wafer substrate and gold; both chosen because literature values on the effect of hydrocarbon adsorption were readily available.

REO samples were fabricated by pressing and sintering powders (Sigma-Aldrich: holmia, 99.9% pure, 100 nm; ceria, 99.9% pure, 5 μm) in accordance with the procedure described in the study which first reported REO hydrophobicity.⁹ First, the powders were dry-pressed into ≈2 mm thick chips at 270 MPa and then at 350 MPa in a 13-mm-diameter steel pellet die (REFLEX evacuable pellet die). The chips were then sintered for 4 h at 1600 °C and 1560 °C for holmia and ceria, respectively, in a box furnace (Blue-M, Thermo Scientific). Immediately after sintering, the advancing and receding contact angles were measured as ≈0° and were also ≈0° on a freshly fractured surface. Field emission scanning electron microscopy (FESEM) images of the grains formed during sintering are shown in Figure 1 along with atomic force microscopy (AFM) scans of the surface. The surface roughness, defined as the ratio of actual surface area to projected surface area, was determined from AFM to be less than 1.05 for both samples, which indicates that surface roughness did not significantly impact wettability.

To ensure a pristine surface, the samples were cleaned with argon plasma (Harrick PDC-001) until no contaminants were present as evidenced by X-ray photoelectron spectroscopy (XPS) (Thermo Scientific K-Alpha), and the samples were further bombarded by argon ions inside of the XPS chamber before the first (pristine) measurement of surface composition. Argon plasma was used because it is inert and removes contamination by physical bombardment as compared to oxygen-containing plasma, which reacts chemically with the surface.²² Furthermore, argon plasma has been shown to remove adsorbed hydrocarbons and does not significantly increase surface roughness.²³ XPS analysis after argon ion bombardment showed that the pristine surfaces exhibited the expected stoichiometric ratios for their respective oxidation states, with gold in its elemental state (Table I).

After cleaning, the samples were exposed to laboratory air (MIT Rohsenow-Kendall Heat Transfer Laboratory, ambient temperature = 25 ± 2 °C, relative humidity = 35% ± 10%) and the advancing and receding contact angles and XPS spectra were measured at multiple time points. Contact angles were obtained by microgoniometric measurement (Kyowa MCA-3, see supplementary Figure S2 (Ref. 24)). Droplet vibrations induced by the piezoelectric dispenser head²⁵ did not affect the measurements.²⁴ The advancing and receding contact angles are presented as opposed to the equilibrium contact angle to thoroughly describe the surface wettability²⁶ and to characterize the force needed to hold the droplet stationary on an inclined condensing surface against the force of gravity, which directly affects condensation heat transfer^3,27 (see supplementary material²⁴). To determine the amount of adsorbed hydrocarbon, the surface carbon percent was measured from the relative peak magnitudes of the surface components observed from XPS spectra taken at each time point. Representative XPS spectra for holmia and ceria are presented in Figure 2. Comparison between the XPS spectra for pristine holmia and ceria and at 96 h after cleaning reveals that a sharp carbon peak develops, often referred to as the “adventitious carbon” peak, which is indicative of adsorption of hydrocarbons onto the surface.²⁸ Note that hydrogen cannot be explicitly detected by XPS because it only has valence electrons, which are indistinguishable from other elements upon excitation and for which the binding energy is influenced by environment; therefore, determination of the average hydrocarbon chain length was not possible.²⁹ 

The average advancing contact angle measurements for each sample as a function of time after argon plasma cleaning started at ≈0° and increased asymptotically over time for every sample (Fig. 3), with the REOs displaying hydrophobic (θ_a > 90°) behavior after 4 days. Note that both the advancing and receding angles for all of the surfaces except gold were less than 10° immediately after cleaning, and, as a result, the contact angle hysteresis was also initially less than 10° (the advancing/receding contact angles on gold were 46°/10° at the time of the first measurement). At 2448 h (102 days) after cleaning, the average advancing angle reached 103° for holmia and 95° for ceria, which are within 10% of the previously reported values.⁹ The advancing angles on gold and silicon reached 66° and 44°, respectively, which are in good agreement with the literature values for hydrocarbon contamination of these surfaces after cleaning to their pristine state and exposing to laboratory air.^8,15,30 Representative images of the advancing contact angle increase over time on the REOs are shown in Figures 3(b) and 3(c).

The increase in advancing contact angle over time suggests that the surface energy decreases over time, which can be attributed to the lower surface energy of the adsorbed hydrocarbons.³¹ This trend has been previously shown for a variety of non-noble metal oxide materials including zirconium dioxide (ZrO₂) and titanium dioxide (TiO₂), among others, and occurs due to physisorption of hydrocarbons to OH⁻ groups and other energetically favorable sites present on the surface,^17,32 where van-der-Waals and hydrogen bonding are typical³³ but covalent bonding is also possible.³⁴ The results of the XPS analysis conducted here show that the amount of carbon present on the surface is indeed increasing over time, indicating that hydrocarbons adsorb to the cleaned surface after exposure to air. As shown in Figure 4(a), the surface atomic percent of carbon increased from ≈0% immediately after cleaning to an asymptotic value of between 12% and 34% depending on sample type.

To explore the relationship between advancing contact angle and hydrocarbon adsorption, the measured advancing and receding angles are shown as a function of the surface atomic percent carbon for holmia and ceria in Figures 4(b) and 4(c). The advancing and receding contact angles increased with surface atomic percent carbon, where the advancing and receding contact angles are positively correlated with surface atomic percent carbon with a Pearson product-moment of at least 0.93 for all of the samples studied here. This correlation is in agreement with previous work for metals and metal oxides (see supplementary Figure S3 (Ref. 24)). The mechanism for this relationship can be explained by considering the adsorbed hydrocarbons to be hydrophobic defects on an initially hydrophilic surface.³⁵ If the hydrocarbons are approximated as circular hydrophobic defects, then the advancing angle is predicted by

where θ_A is the advancing contact angle as a function of the surface coverage of hydrocarbons, f_max, which is determined from the surface atomic percent carbon and the relative sizes of the adsorbed hydrocarbons and the surface atoms, θ_1,A is the advancing contact angle of the hydrophilic surface with no adsorbed hydrocarbons (≈0°), and θ_2,A is the advancing contact angle on the surface once it has become saturated with hydrocarbons (approximated as the advancing contact angle at 2448 h) (see expanded explanation in supplementary material²⁴). The curve obtained from this model is shown to match well with the experimental data, indicating that hydrocarbon adsorption results in the observed increase contact angle. Modeling the receding angle in this case yields less useful information due to adhesion hysteresis of the adsorbed hydrocarbons and the variability in receding behavior as a function of time that the droplet remains on the surface.³⁶ 

This work demonstrates that the hydrophobicity of REOs is due to hydrocarbon adsorption, as shown by the relationship between the increasing contact angle and surface carbon percent over time upon exposing a pristine surface to atmosphere. Similar to noble metals and more typical metal oxides, pristine REOs have high surface energy, making them intrinsically hydrophilic. This study on the evolution in wetting behavior suggests that REOs can serve as coatings to induce dropwise condensation for improved heat transfer performance through the spontaneous adsorption of hydrocarbon material and subsequent effect on wetting behavior.

The potential of REOs as functional surface coatings for condensers due to their hydrophobicity after hydrocarbon adsorption is promising, but also raises concerns. The large contact angle hysteresis of the REO surfaces studied here (∼60°–70°), and shown previously,⁹ will act to increase the size of departing droplets, which negatively impacts heat transfer.³ Another challenge is the thermal expansion coefficient mismatch between REOs (4–10 μm/m·K) and many industrial condenser metals (10–25 μm/m·K), which could result in fracturing of thin and brittle REO coatings due to temperature fluctuations.³⁷ However, the relatively low cost and moderate thermal conductivities of REOs (2.4–13.3 W/m·K, see supplementary material²⁴) offer a potentially unique advantage over traditional promoter coatings. Layers of hydrophobic polymers (PTFE) have been shown to give excellent dropwise condensation behavior but have only been found to be sufficiently durable when the thickness (δ) of the low-conductivity polymer (k_p ∼ 0.2 W/m·K) layer is so large (δ_p ≈ 20 μm) as to offset the advantage of dropwise condensation.³⁸ The larger coating thickness is typically required in order to increase adhesion to the metal substrate and enhance resistance to oxidation and moisture. Gold coatings have been shown to give excellent dropwise condensation but have only been found to be sufficiently durable (≈5.7 years of operating time) when the thickness of the gold is so large (≈50 μm) as to make the approach economically unfeasible.³⁹ On the other hand, REOs strike a balance between the two previous approaches in terms of cost and thermal conductivity. The moderate thermal conductivity of REOs allows for a ∼25x thicker coating than conventional polymer layers while maintaining a comparable thermal resistance with the added benefit of potentially greater adhesion and durability.⁹ Furthermore, the reduced cost of REOs compared to gold coatings makes their application to industrial materials more economically feasible.⁴⁰ In the future, more rigorous calculations of the expected condensation heat transfer are needed based on existing high fidelity models in the literature.^3,41

This study provides insight on the wetting mechanism of the REO material group that suggests potential for implementation in other fields which make use of hydrophobic materials, including self-cleaning surfaces,⁴² anti-icing surfaces,⁴³ water desalination,⁴⁴ and enhanced heat transfer surfaces.^4,45 Furthermore, our work highlights the importance of controlling hydrocarbon adsorption for material wetting characterization.

We thank Dr. Jay J. Senkevich for fruitful discussions and Professor Kripa K. Varanasi for constructive feedback regarding hydrocarbon adsorption. We gratefully acknowledge funding support from the Office of Naval Research (ONR) with Dr. Mark Spector as program manager and the MIT S3TEC Center, an Energy Frontier Research Center funded by the Department of Energy, Office of Science, Basic Energy Sciences, under Award No. DE-FG02-09ER46577. D. J. Preston acknowledges funding received by the National Science Foundation Graduate Research Fellowship under Grant No. 1122374. Any opinion, findings, conclusions, or recommendations expressed in this material are those of the authors(s) and do not necessarily reflect the views of the National Science Foundation. R. Enright acknowledges funding received from the Irish Research Council for Science, Engineering, and Technology, cofunded by Marie Curie Actions under FP7. Bell Labs Ireland thanks the Industrial Development Agency (IDA) Ireland for their financial support. We also acknowledge the support from the National Science Foundation through the Major Research Instrumentation Grant for Rapid Response Research (MRI-RAPID) for the microgoniometer. This work made use of the MRSEC Shared Experimental Facilities at MIT, supported by the National Science Foundation under Award No. DMR-08-19762. This work was performed in part at the Center for Nanoscale Systems (CNS), a member of the National Nanotechnology Infrastructure Network (NNIN), which was supported by the National Science Foundation under NSF Award No. ECS-0335765. CNS is part of Harvard University.