Photoinduced Superhydrophilicity: A Kinetic Study of Time Dependent

Article pubs.acs.org/Langmuir

Photoinduced Superhydrophilicity: A Kinetic Study of Time Dependent Photoinduced Contact Angle Changes on TiO2 Surfaces Philip S. Foran,† Colin Boxall,*,† and Kieth R. Denison†,‡ †

Engineering Department, Lancaster University, Lancaster LA1 4YR, United Kingdom Oxley Developments Co. Ltd., Priory Park, Ulverston, Cumbria LA12 9QG, United Kingdom

‡

S Supporting Information *

ABSTRACT: Transparent TiO2 thin films were prepared on quartz substrates via a reverse micelle, sol−gel, spin-coating technique. The time dependence of the TiO2 film photoinduced superhydrophilicity (PISH) was measured by goniometric observation of the contact angle, θ, of sessile water drops at the film surfaces. In these measurements, the TiO2 substrate was illuminated by 315 nm light and drops were sequentially applied at a range of illumination times. Using a model for the wetting of heterogeneous surfaces derived by Israelachvili and Gee, these measurements were used to calculate the time dependence of f 2, the fractional surface coverage of the TiO2 surface by adventitious contaminating organics (Israelachvili, J. N.; Gee, M. L. Contact angles on chemically heterogeneous surfaces. Langmuir 1989, 5, 288). Extending this model to include a Langmuir−Hinshelwood based kinetic analysis of f 2 as a function of time allowed for calculation of an expected value for θ immediately prior to illumination, that is, at illumination time t = 0. Such expected values of θ at t = 0 were calculated using two possible values of θ1, the contact angle on a pristine unilluminated homogeneous TiO2 surface: (i) θ1 = 4° as suggested by, inter alia, Zubkov et al. (Zubkov, T.; Stahl, D.; Thompson, T. L.; Panayotov, D.; Diwald, O.; Yates, J. T. Ultraviolet Light-Induced Hydrophilicity Effect on TiO2(110)(1 × 1). Dominant Role of the Photooxidation of Adsorbed Hydrocarbons Causing Wetting by Water Droplets. J. Phys. Chem. B 2005, 109, 15454); and (ii) where θ1 = 25°, as suggested by Fujishima et al., representative of a more hydrophobic homogeneous TiO2 surface that reconstructs upon exposure to ultraband gap illumination into a hydrophilic surface where θ1 → 0° (Fujishima, A.; Zhang, X.; Tryk, D. A. TiO2 photocatalysis and related surface phenomena Surf. Sci. Rep. 2008, 63, 515). Analysis of data from our experiments and from selected literature sources demonstrates better agreement between these calculated and experimental values of θ at t = 0 when θ1 is taken to be 4°, implying that an uncontaminated TiO2 surface is inherently hydrophilic. The results of this study are discussed in the context of the current debate over the origin of the photoinduced superhydrophilic effect.

1. INTRODUCTION Since the discovery of the photoinduced superhydrophilic effect observed on UV-illuminated semiconducting TiO2 films in 1995, two distinct schools of thought have evolved concerning its origin.3 The first holds that the superwet nature occurs due to a surface reconstruction of the TiO2 film resulting in an increase in hydroxyl groups.4,5 The second is of the view that photocatalytic destruction of trace surface organic contaminants leads to the presentation of an inherently wet underlying metal oxide surface.2 In 2003, Shultz et al.6 used visible-infrared sum frequency generation spectroscopy to analyze the surface composition of TiO2 films spin coated from aqueous suspensions of anatase nanoparticles and demonstrated for the first time the existence of a hydrocarbon film on the surface of as-prepared TiO2 coatings under ambient conditions in air. Their results also showed that these trace organics are removed during UV illumination. Additionally, they also reported a simultaneous increase in the measured surface hydroxyl count on the same samples during the same experiments. This second result is not © 2012 American Chemical Society

inconsistent with the possibility that both surface reconstruction and trace organic destruction may contribute to the photoinduced superhydrophilic effect. Sessile drop contact angle measurements by goniometry have been widely used to study the wettability of surfaces and thus have been used to interrogate the mechanisms and kinetics of photoinduced superhydrophilicity on TiO2.7,8 Within the context of work by Zubkov et al.2 (discussed in detail below) and Yan et al.,9 Fujishima et al. have suggested that there are two distinct stages that may be observed during time dependent contact angle studies of surface wettability on illuminated TiO2.3 These stages involve the photocatalytic destruction of organic contaminants that are present on the semiconductor surface, which lowers the measured contact angle from the vicinity of 60 o to 20−35°, and a separate effect that lowers the latter value down to 0−5°.3 Quantitative analysis of results from Received: July 3, 2012 Revised: November 19, 2012 Published: November 28, 2012 17647

dx.doi.org/10.1021/la3026649 | Langmuir 2012, 28, 17647−17655

Langmuir

Article

anneal the TiO2. After cooling, the TiO2 coated substrates were sonicated in ethanol followed by distilled water for 5 min each, dried in a stream of filtered N2, and then subsequently irradiated for 8 h under ultraband gap illumination. All samples were stored in a desiccator in darkness at room temperature prior to characterization/use. TiO2 phase identification was achieved through Raman spectroscopy using a Renishaw Ramascope 1000 (Renishaw, Gloucestershire, U.K.) with backscattering geometry and a 17 mW He−Ne 632.8 nm laser. Consistent with earlier studies in our laboratory,7 the annealed layers were found to be predominantly anatase in structure (Supporting Information: Figure S1). UV−vis spectra (HP 8452A diode array spectrophotometer, Hewlett-Packard, U.K.) indicate a band gap of ∼3.54 eV (Supporting Information: Figure S2), blueshifted from the bulk value of 3.2 eV3 as the films are composed of particles small enough to exhibit the quantum size effect.7,12 Surface topography was assessed using atomic force microscopy (AFM; QScope 250, Quesant, CA), Figure 1. Again consistent with our earlier

surface goniometry with respect to the former of these two effects requires the availability of a mathematical model of the system that relates the measured contact angle during an irradiation experiment to the concentration of organic remaining on the semiconductor surface. Such a model already exists. Israelachvili and Gee1 have derived from first principles an equation that relates the equilibrium contact angle observed on a flat, nonporous chemically heterogeneous surface, comprising, for example, a mixture of hydrophobic and hydrophilic components, to the fractional coverage of the surface by each component and the contact angles observed on the pure homogeneous surfaces of each component. Järn et al.10 have recently used Israelachvili and Gee’s equation, and an earlier treatment by Cassie,11 to relate the contact angle observed on a fluoroalkylsilane (FAS)functionalized TiO2@SiO2 nanopatterned surface under ultraband-gap illumination to the FAS fractional surface coverage as a function of time. In so doing, they consider this surface to be composed of a hydrophilic TiO2@SiO2 component and a hydrophobic FAS component. Here we report on the extension of this approach to include a Langmuir−Hinshelwood-based model of the dynamics of the photocatalytic destruction of surface adsorbed organics. This then allows for kinetic analysis of time dependent contact angle measurements on illuminated semiconductor surfaces. We also report on the extension of this model to a limited class of mesoporous TiO2 surfaces and describe the conditions under which this extension may be applied. This paper is then structured as follows: for the convenience of the reader, we begin by summarizing Israelachvili and Gee’s model, its advantages with respect to Cassie’s treatment, and the collation of data needed to implement that model, specifically, the contact angles recorded at homogeneous surfaces of the constituent phases of the heterogeneous system. We then deploy the model on data drawn from both our own experiments and an array of literature sources, allowing for the construction of surface hydrocarbon concentration versus time curves which are then subjected to the aforementioned Langmuir−Hinshelwood-based kinetic analysis. In so doing, we are for the first time providing a framework within which contact angle versus time data from photoinduced superhydrophilicity experiments may be processed to yield both meaningful quantitative kinetic information and insights into the surface properties of the underlying semiconductor material.

Figure 1. Three dimensional intermittent contact mode AFM images of the surface of mesoporous TiO2, spin-coated on quartz, and annealed at 500 °C as described in the text. studies,7 the annealed layers were found to be composed of interconnected, monodispersed, spherical primary particles. The resultant mesoporous structure was found to have a measured average roughness (Ra) of 0.74 nm and measured height of 26 nm, similar to the height obtained by Yu et al.12 2.3. Preparation and Characterization of Stearic Acid Multilayer Films on TiO2. Substrates. Stearic acid was dissolved in chloroform to a concentration of 30 g L−1. Using a Pasteur pipet, 30 μL of this solution was deposited onto a TiO2 sensitized quartz substrate and subsequently spun at 500 rpm for 10 s. The resultant sample was then transferred to a ceramic boat and placed in an oven at 100 °C for 40 s to remove any residual chloroform. Stearic acid has a melting point of 69.6 °C. Thus, after drying in the oven and cooling to room temperature, it forms a waxy multilayer film on the substrate surface. Once cooled, the sample was ready for contact angle measurements. Stearic acid film formation was confirmed using Fourier-transform infrared analysis conducted using a Bruker Alpha-P ATR Diamond FTIR spectrometer (Coventry, U.K.). Consistent with other studies,13 expected FTIR peaks for stearic acid were measured at 2958, 2923, and 2852 cm−1; see Supporting Information: Figure S3. Quartz crystal microbalance studies of m-TiO2 coated substrates preand post-stearic-acid film deposition show that the mass change associated with application of the film is consistent with it being composed of multilayer13 stearic acid; that is, the TiO2 layer is completely covered with a homogeneous layer of stearic acid. See Supporting Information: section S3. 2.4. Contact Angle Measurements. All contact angle measurements were performed using a custom-built camera-based goniometer described previously.7 In summary, the substrate stage of the goniometer was placed within a sealed enclosure to allow for control of the relative humidity (RH) during measurement. An air pump (Hi-

2. EXPERIMENTAL SECTION 2.1. Materials and Reagents. All reagents were of analytical grade or better and used as received. All solutions were prepared using doubly distilled water, further purified by a deionization system (Epure model 04642, Barnstead/Thermodyne, Dubuque, IA) to a resistivity of 1.8 × 105 Ωm. 2.2. Mesoporous TiO2 Layer Preparation and Characterization. Quartz substrates were coated with a mesoporous TiO2 (mTiO2) layer via sol−gel spin coating. A sol−gel, as described by Yu et al.,12 was prepared by vigorously mixing Triton X-100 (26.0 g) and cyclohexane (150 mL) to form a reverse micellar solution. After 30 min, water (1.08 g) is added and the solution becomes turbid. This clears upon addition of titanium(IV) isopropoxide (99.999%, 23 g). The solution was then stirred for 1 h at 20 °C, so forming a colloid of TiO2 nanoparticles. Acetylacetone (10 mL) is then added to stabilize the solution. The resultant sol−gel is applied to substrates by spin coating for 10 s at 2900 rpm using an inverted model 636 rotating disk electrode system (Princeton Applied Research, Oak Ridge, TN). Coated substrates are then fired in a furnace at 500 °C for 1 h to 17648


Langmuir

Article

Tech 1500, Weiko UK, Berkshire, U.K.) was used to transfer moist air from a Drechsel bottle fitted with a no. 1 filter into the enclosure. The pump, with an adjustable valve to regulate air flow, was fitted in-line between the enclosure and the bottle, with the bottle being immersed in a temperature controlled water bath (Clifton NE4-22D, Bennett Scientific, Devon, U.K.). Control of RH was provided by adjustment of the in-line valve and the bath temperature, the initial value of which, for the target RH, was determined from the van’t Hoff isochore. A thermohygrometer (Rotronic A1, Rotronic, West Sussex, U.K.) was used to measure chamber RH and temperature. Pump speed and bath temperature were fine adjusted until the desired RH and temperature were reached. In all experiments, a sessile drop of distilled, doubly deionized water with a volume of 1 μL was vertically deposited onto the as prepared TiO2 thin films using a 5.0 μL microsyringe. The edge on profile of the drop was then filmed, either as an instantaneous image or as a function of time. Accordingly, two types of contact angle measurement were made on test substrates: static/discrete measurements and continuous measurements. Discrete measurements followed a protocol wherein the semiconductor surface is first illuminated for a period of 60 s and illumination then removed by closing of the shutter, followed by application of the drop and the recording of an image for subsequent measurement of θ, this cycle being repeated over the period of illumination or until θ ≈ 0. The clock recording illumination time is paused while the shutter is closed, and the image of the drop recorded. While the surface is being illuminated, the microsyringe and camera optics are then repositioned to another spot on the illuminated surface in preparation for the next reading. Thus, for any one reading, the shutter is closed for no more than 30 s. Continuous measurements followed the protocol wherein a single drop is first applied to the TiO2 substrate surface and then illuminated. θ of that single drop is then continuously measured as a function of time. In the case of continuous measurements, the RH and chamber temperature were preset to 99% and 20 °C, respectively, in order to minimize drop evaporation. In both experiment types, image capture was performed using a Sony XC-ST50CCD camera and Studio DC10plus v8.0 image editing software (Pinnacle Systems, Middlesex, U.K.) at 25 frames s−1. To aid visualization, drops were back lit with a 100 W quartz iodine lamp (Rank Bros, Cambridge, U.K.) fitted with a borosilicate glass UV cutoff filter (L G Optical Ltd., St. Leonards-on-Sea, Sussex, U.K.) to prevent its light inducing any photoinduced effects. Images were then analyzed and contact angle θ determined using Screen Caliper and Screen Protractor software (http://www.iconico.com). Contact angles reported are an average of contact angles measured on a minimum of three separate locations on the test surface. In both illuminated discrete and continuous experiments, monochromated light was directed perpendicularly “top down” onto the test slide via a quartz light guide coupled to a 75 W xenon arc lamp (Photon Technologies International, Ford Road, Sussex, U.K.) fitted with a monochromator. The illumination wavelength was 315 nm, incident intensity 0.04 mW cm−2 measured using a CA-2 laboratory thermopile (Equinox Instruments, Lincoln, Lincolnshire, U.K.) with an accuracy of ±0.01 mW cm−2.

on Young’s equation, Cassie’s equation is phenomenological and not based on any modeling of atomic level microheterogeneity of surface composition, a shortcoming that Troughton et al. have suggested limits its quantitative utility.14 Israelachvili and Gee have addressed the concerns of Troughton et al. through applying the Young−Dupre equation to each surface component within a treatment that averages the polarizabilities, dipole moments, and surface charges of those components within the heterogeneous surface under study. This gives rise to eq 2 for flat, nonpoorous heterogeneous surfaces when the chemical heterogeneities are of atomic or molecular dimensions: (1 + cos θ )2 = f1 (1 + cos θ1)2 + f2 (1 + cos θ2)2

Järn et al.10 have recently found that, for FAS-functionalized TiO2@SiO2 nanopatterned surfaces, both the Cassie and Israelachvili models satisfactorily describe the static/discrete contact angle measurements over a broad range of f 2 values where, in this case, f 2 describes the fractional surface coverage by the FAS. Thus, if θ1 is the contact angle measured on a pure homogeneous uncontaminated TiO2 surface and θ2 is the contact angle measured on a homogeneous surface comprised of the organic, and both θ1 and θ2 are known, then, as f1 = 1 − f 2, eqs 1 and 2 can be readily applied to the quantitative analysis of contact angle data measured on TiO2 surfaces contaminated with trace organic. The next section describes the determination of these component θ values. In using these values of θ in the interpretation of photoinduced contact angle changes on TiO2, we will restrict our data treatment to application of eq 2 only. This is both because of Troughton’s reservations about Cassie’s treatment described above and because adsorbed organic-contaminated TiO2 surfaces are best described using the molecular-scale chemical heterogeneity-based model of Israelachvili and Gee.1 3.2. Determination of a Value of θ2. The contact angle for θ2, the angle observed on a pure organic surface, is most easily obtained from discrete, static measurements on a multilayer film of stearic acid, chosen as it is widely regarded as a model pollutant when studying self-cleaning surfaces.15−17 θ2 was determined experimentally by taking the average of the measured contact angles of 1 μL water droplets on a film of stearic acid, prepared as described in section 2.3. From measurements conducted on both sides of three separate drops, Figure 2, an average value of θ2 = 90° ± 1° is obtained. This value is in good agreement with a value of 85° reported by Sakatani et al.18 for lauric acid coated TiO2 surfaces. 3.3. Determination of values of θ1. A value for θ1 for the uncontaminated TiO2 surface is provided by the data of Zubkov et al.2 who, in a set of continuous contact angle measurements, studied the time dependence of the contact angle of an illuminated rutile (110) surface, Figure 3. Experiments were conducted in a UHV chamber filled with oxygen at 1 atm pressure as a function of the concentration of a deliberately added model gas phase hydrocarbon contaminant, hexane. Before drop placement, the rutile surface had been cleaned by Ar+ sputtering and then annealed at 900 K in dry O2. Their results showed that in the absence of deliberately added organic contaminant, that is, at 0 ppm hexane, the contact angle measured at the onset of illumination was 25° which, after ca. 10 s of continuous illumination, had reduced to approximately 4°.

3. RESULTS AND DISCUSSION 3.1. The Israelachvili and Gee Model. One of the earliest descriptions of the effect that heterogeneity of surface composition can have on measured contact angles of sessile drops was that of Cassie11 who deduced that if two phases (phase 1 and 2) with different surface energies and so contact angles exist on the same surface, the apparent contact angle is described by eq 1: cos θ = f1 cos θ1 + f2 cos θ2

(2)

(1)

where θ is the measured contact angle of a liquid on a heterogeneous surface composed of two fractions f1 and f 2 where θ1 and θ2 are the contact angles on homogeneous surfaces of phase 1 and phase 2, respectively. Although based 17649


Langmuir

Article

supporting the view that the rapid contact angle change recorded by Zubkov et al. at 0 ppm hexane could be due to surface reconstruction. Thus, as suggested by Fujishima et al.,3 it is possible that, in the absence of illumination, a clean TiO2 surface is hydrophilic with a contact angle of ∼25−30°, this hydrophilicity increasing upon illumination with a consequent reduction in contact angle to 4°. The former value is in approximate agreement with a contact angle value of 32° measured by Hennessey et al.4 on HF etch-preapared rutile (110) surfaces in a dry air atmosphere. The latter value is supported by the final contact angles measured by Zubkov et al. on their same rutile (110) surface, after its deliberate exposure to the hexane contaminant, also being approximately 4° after photocatalytic removal of said contaminant, Figure 3. Thus, opinion as to the value of θ1 for the uncontaminated TiO2 surface is broadly divided into two schools of thought: (i) the surface is intrinsically hydrophilic and θ1 ∼ 4° or less, and any increase above this value is due entirely to the adsorption of trace hydrophobic organic contaminant; (ii) the surface is slightly hydrophobic with a θ1 of 25−30°, with this value decreasing to ∼4° upon illumination due to surface reconstruction. Both of these values will be used and tested in the analysis presented in sections 3.5 and 3.6 below. 3.4. Results of Continuous Measurement Experiments. The form of the time dependence of the continuously measured contact angle for hexane-contaminated surfaces reported by Zubkov et al. and shown in Figure 3 is similar to that which we have observed from sessile water drops exposed to continuous illumination on nominally clean (i.e., in the absence of any deliberately added organic contaminant, be it stearic acid, hexane, or any other substance) mesoporous anatase surfaces (Figure 4); that is the drops exhibit a

Figure 2. Discrete measurements of contact angles of a single 1 μL sessile drop of water on a film of stearic acid covering a mesoporous layer of TiO2: (A) left, 90.8° and (B) right, 89.0°.

Figure 3. Continuous measurements of water contact angle evolution at a rutile (110) surface during UV irradiation in an oxygen atmosphere with different hexane concentrations, according to Zubkov et al.2

Figure 4. Continuous measurements of the contact angle evolution of a 1 μL sessile water drop placed on a nominally clean mesoporous TiO2 surface, in a 99% RH atmosphere at 20 °C, under 0.04 mW cm−2 illumination at 315 nm, immediately after drop deposition and exposure to light.

Zubkov et al. attributed this reduction in contact angle to the photocatalytic removal of trace contaminant organics. However, Fujishima et al.3 have suggested that this reduction in contact angle may be due to a surface reconstruction leading to an increase in surface hydrophilicity as per Shultz, and not due to trace organic destruction. Supportive of Fujishima’s view is the fact that surface reconstructions in such a short time period, 10 s, are not unheard of − an example being the place exchange process that takes place at the surface of metal electrodes during electrochemical oxidation to metal oxides.19 Additionally, it is well reported that photocatalytic destruction of organics is a relatively slow process (see, e.g., ref 13 and more generally refs 3, 5, and 20 and references therein), again

photoinduced stick−slip behavior. The thermodynamic driving force for this behavior is the departure of the system from capillary equilibrium as, with increasing illumination time, the equilibrium contact angle of the drop decreases. Movement of the drop triple line in response to this change in surface energy is opposed by chemical and/or physical inhomogeneities on the surface, so-called “stick”. Only once the stored Gibbs free energy associated with the departure from capillary equilibrium exceeds the potential energy barrier associated with the surface 17650


Langmuir

Article

inhomogeneities does movement of the triple line occur, socalled “slip”. In the case of our experiments on nominally clean mesoporous TiO2 surfaces (Figure 4), our earlier studies of the thermodynamics of photoinduced slip7 have shown that the potential energy barrier, U, opposing movement is by eq 3: U=−

γLV ⎛ rY 2 rX 2 (r 2 − rX 2) ⎜ − − Y rX ⎝ 1 + cos θY 1 + cos θX 2

⎞ cos θY ⎟ ⎠

(3)

where θX, rX and θY, rY are the contact angles and measured contact radii of the drop immediately before and after slip, respectively. γLV is the surface tension of the liquid−vapor interface. For the data of Figure 4, U is found to be 4.1 × 10−6 J m−1, in good agreement with a value of 6.6 × 10−6 J m−1 obtained in our earlier studies.7 From the latter, the potential energy barrier opposing movement was shown to be predominantly derived from the physical roughness of these films, a finding supported by the AFM image of Figure 1. In the case of the data of shown in Figure 3 and obtained on a flat single crystal rutile surface, Zubkov et al. propose that the potential energy barrier opposing movement is primarily derived from the organic contaminant, that is, chemical sources.2 This means that the effect of illumination in the experiments of Figure 3 is twofold, simultaneously decreasing both the equilibrium contact angle through eq 2 and the energy barrier opposing movement. The Israelachvili and Gee- based approach adopted here provides a means to extract quantitative kinetic information concerning the former of these effects, complementing our earlier study of the thermodynamics of the latter (vide supra).7 3.5. Relationship between Contact Angle θ and Hydrocarbon Coverage. Using the two putative values of θ1, at unilluminated clean TiO2 surfaces 4° and 25°, the experimentally measured value of θ2 = 90°, and solving eq 2 for all integer values of θ between θ1 and θ2 for both values of θ1 allows us to construct plots of predicted measured contact angle θ versus the calculated fractional surface coverage by the hydrocarbon, f 2, Figure 5. Using this data, we can calculate estimated values of f 2 from experimentally measured contact angles on TiO2 sample surfaces. Measured contact angles on TiO2 in the dark are most commonly reported as falling in the range 55−60°,3 with 60° being the value obtained by Hennessey et al. after exposing HF etched rutile (110) surfaces with initial contact angles of 32° to ambient air.4 There are two points to note from Figure 5. The first point is that at the upper end of this range, 60°, f 2 ≈ 0.6 regardless of whether θ1 = 4° or 25°. This indicates that the effect of any surface reconstruction/increase in surface titanol concentration on measured contact angle during the early part of a period of surface illumination by ultraband gap light will be masked by the effect of the submonolayer of organic contaminant. We shall return to this below. The second point to note is that if θ1 = 4°, then little more than 10% organic surface coverage will yield a measured θ of 25°, illustrating just how susceptible contact angle measurements are to surface contamination on these materials. Again, we shall return to this below. 3.6. Application and Extension of the Israelachvili and Gee Model. Use of Israelachvili and Gee’s equation in the

Figure 5. Predicted experimental contact angle θ on a TiO2 surface as a function of f 2, the fractional surface coverage of the surface by a contaminant hydrocarbon, calculated using eq 2 and θ2 = 90°, θ1 = 4° and 25°.

analysis of measured θ versus illumination time for TiO2 surfaces presupposes that the measured θ values are equilibrium contact angle values rather than the nonequilibrium values that obtain preslip in the data of Figures 3 and 4. Fortunately, such values are readily obtainable through use of a static or discrete measurement protocol (see section 2.4) rather than the continuous measurement method used in Figures 3 and 4. Thus, Figure 6A shows the time dependence of the contact angles, measured using the discrete method, for a series of sessile water drops deposited on the same nominally clean mesoporous anatase surface used in the continuous experiment of Figure 4, that is, in the absence of any deliberately added organic contaminant, be it stearic acid, hexane, or any other substance. The anatase surface is illuminated with ultra-bandgap light throughout the experiment of Figure 6A, with the measured value of θ continuously decreasing with increasing illumination time, rather than in a stepwise slip event as per Figure 4. Data from an analogous discrete experiment on a CVD-prepared anatase surface, using 0.5 μL drops in a 60% RH atmosphere at 25 °C conducted by Miyauchi et al.21 are presented for comparison, Figure 6B. In a previous study,7 we have previously shown that, on mesoporous TiO2 surfaces identical to those used in the study of Figure 6A, evaporation of sessile drops in the dark exhibit a so-called vigorous phase in the final stages of that evaporation, with this vigorous phase being characterized by a decrease in drop height and radius occurring at near-constant contact angle. Both Bourges-Monnier and Shanahan,22 and Zhang and Yang23 have pointed out that observation of a vigorous phase during drop retreat is indicative of an effectively smooth surface. Thus, again in our previous study, we concluded that observation of a vigorous stage during drop evaporation experiments on our mesoporous TiO2 surfaces in the dark meant that the surface could be treated as being essentially smooth within the context of the Young−Dupre equation. The Israelachvili and Gee model for composite surfaces is based on the Young−Dupre equation, so justifying the use of eq 2 in the analysis of the data of Figure 6A. Data of Figure 3 suggest that any surface reconstruction that occurs is complete within 10 s illumination at an intensity similar to that employed in both Figures 4 and 6. This in turn suggests that any change in θ that occurs after 10 s in the 17651


Langmuir

Article

Figure 6. Discrete measurements of the contact angle of (A) 1 μL sessile water drop placed on a nominally clean mesoporous anatase surface, in a 99% RH atmosphere at 20 °C, under 0.04 mW cm−2 illumination at 315 nm, immediately after drop deposition and exposure to light. (B) 0.5 μL water droplet on an anatase surface, in a 60% RH atmosphere at 25 °C, data extracted from Miyauchi et al.21

Figure 7. Time dependence of surface hydrocarbon coverage using θ2 = 90°, θ1 = 4° of (A) nominally clean mesoporous TiO2 film under illumination, based on data from Figure 6A. (B) TiO2 film under illumination based on data from Figure 6B.

Figure 8. Time dependence log surface hydrocarbon coverage using θ2 = 90°, θ1 = 4° of (A) m-TiO2 film under illumination, based on data from Figure 7A. (B) TiO2 film under illumination based on data from Figure 7B.

and is thus first order with respect to the surface concentration of the organic, then it can be written that

discrete measurement data of Figure 6 is almost entirely due to the photocatalytic removal of adventitious surface organics. Using eq 2/Figure 5, θ2 = 90°, and θ1 = 4°, it is possible to plot fractional hydrocarbon coverage f 2 versus time for Figure 6A and B given in Figure 7A and B, respectively. If it is assumed that the photocatalytic removal of the surface organic contaminants accords with the Langmuir−Hinshelwood-based kinetic model developed by Turchi and Ollis,24

−df2 dt

= kf2

or

ln(f2 )t = ln(f2 )t = 0 − kt

(4)

where k is the associated pseudo-first-order rate constant, ( f 2)t=0 is the fractional surface coverage by the organic at illumination time t = 0 and (f 2)t is the surface coverage after 17652


Langmuir

Article

Table 1. Experimentally Measured Values of θ at t = 0 s from Four Separate Literature Sources with Their Respective Calculated Values Using the Israelachvili and Gee Based Analysis of eq 2 and 4 with Both θ1 = 4° and 25° data source

sample preparation method

sample atmosphere/ conditions

Miyauchi et al.21 Sakai et al.25

MOCVD

ambient

sol−gel

Watanabe et al.26 Yu et al.12

sol−gel

unspecified, assumed ambient ambient

sol−gel

ambient

sample surface roughness

measured θ at t=0

calculated θ at t = 0 (using θ1 = 4°)

calculated θ at t = 0 (using θ1 = 25°)

3.1 nm

57°

57°

61°

49°

53°

57°

35°

38°

45°

66°

64°

67°

0.56 nm

illumination time t. Plotting ln( f 2)t versus t at t > 10 s for the data of Figure 7A and B should then yield straight lines with slopes of −k and extrapolated intercepts of ln( f 2)t=0 (Figure 8A and B, respectively). Values of k of 0.023 and 0.002 s−1 (1.36 and 0.09 min−1) are found for Figure 8A and B, respectively, reflecting, inter alia, the different photocatalytic activities arising from the various synthetic routes used for these samples (e.g., sol−gel, MOCVD; see Table 1) and/or the different experimental conditions the samples were exposed to while under illumination (pO2, humidity, light intensity, etc.). However, only the values of ln( f 2)t=0 obtained from the extrapolated intercepts of the plots of Figure 8 are of relevance to this study; thus, further analysis of k values will not be attempted here. For Figure 8A, ln( f 2)t=0 is found to be −0.4714 or ( f 2)t=0 = 0.624. Substituting this hydrocarbon coverage into eq 2 and using parameters θ1 = 4° and θ2 = 90°, we calculate an initial contact angle of θ = 62° at t = 0 s, slightly higher than the experimentally measured value of θ = 56° at t = 0 s, Figure 6A. When using parameters θ1 = 25° and θ2 = 90° we calculate an initial contact angle of θ = 65° at t = 0 s, even higher than the measured angle. Following the same protocol with Figure 7B yields Figure 8B with ln( f 2)t=0 = −0.6312 or (f 2)t=0 = 0.5320. Using θ1 = 4°, θ2 = 90°, this equates to a calculated initial contact angle of θ = 57° at t = 0 s, the same value measured experimentally, Figure 6B. When using parameters θ1 = 25°, θ2 = 90° we calculate an initial contact angle of θ = 61° at t = 0 s, higher than the experimentally measured angle. Obviously, at t = 0 s, the TiO2 surface cannot have undergone reconstruction due to absence of any illumination. Thus, in accordance with Fujishima’s interpretation of Zubkov’s data, the pure TiO2 surface at this time would be expected to have a contact angle of θ1 = 25° (vide supra). However, analysis of the data shows a closer fit between the experimentally measured contact angle and that calculated from the extrapolation of Figure 8 when using θ1 = 4° compared to θ1 = 25°; that is, the data suggests that at t = 0, a single component TiO2 surface will have a contact angle of θ1 = 4° and so is intrinsically “wet”. Measured contact angles in excess of this value are then due to the adsorption by adventitious surface contaminant as originally suggested by Zubkov et al. in reference to their own data. To determine the robustness and range of application of this conclusion, the analysis procedure described above was applied to an array of data from literature sources. The discrete, static measurement method is by far the most commonly used approach in photoinduced superhydrophilicity experiments in the literature; however, not all data sets are appropriate for such analysis. The criteria used to select data sets for analysis were then as follows: (i) the experimental data set must be gathered

using the discrete/static measurement methodology and not the continuous method; (ii) there must be a sufficient number of data points, arbitrarily set at five, between the initial measured and final contact angle measurements during any one period of illumination; (iii) the material under study must be TiO2; (iv) the surface behaves as being effectively flat and nonporous, so conforming with the requirements for use of the Young−Dupré equation and, in turn, the model of Israelachvili and Gee. We found four data sets that match these criteria: Miyauchi et al.21 (vide supra), Sakai et al.,25 Watanabe et al.,26 and Yu et al.12 The analysis of these data sets is summarized in Table 1. Recently, data sets have been reported on mesoporous surfaces, and especially TiO2 surfaces with a controlled, regular mesoporosity such as those prepared using block copolymer templating systems.18,27 However, it is expected that such surfaces will not generally obey the Young−Dupré equation that underpins the model derived by Israelachvili and Gee, rather, such surfaces would be better described using the Wenzel equation,28 requiring the rederivation of the Israelachvili and Gee model in this context. This is beyond the scope of this current communication and will form the basis of further work. Thus, restricting our analysis to effectively flat nonporous surfaces, Table 1 summarizes the results of the Israelachvili and Gee based kinetic analysis of the four above-mentioned data sets. Fortunately, these covered a range of TiO2 sample and experiment conditions (see Table 1) so allowing us to test the generic validity of this analysis. All samples were nonporous anatase with the contaminant being adventitious atmospheric organics. The details of the analyses of the individual data set are given in the Supporting Information (SI4−7). From Table 1, it can be seen that the calculated values for θ at t = 0 s obtained from Miyauchi, Sakai, and Watanabe data sets using eq 2 and 4 all fit more closely with the experimentally measured values when using θ1 = 4° in eq 2 than when using θ1 = 25°. This fits with the observations from our own data. Miyauchi’s starting θ is relatively close to the limiting starting contact angle of 60° (vide supra, Figure 5); as such, the calculated values only differ slightly between the two different θ 1 models. In contrast in Watanabe’s data set, the experimentally measured starting contact angle is much lower, measured at 35°. Here, the calculated values of θ exhibit greater separation with θ at t = 0 being found to be 38° when θ1 = 4° and 45° when θ1 = 25°. Yu’s data set again demonstrates a potential limitation of the analysis when starting at large contact angles close to the value most commonly observed in the literature, 60°. As can be seen from Figure 5, above this value of θ, the model cannot easily discriminate between the two θ1 values for this system and hence gives relatively similar calculated values for θ. 17653


Langmuir

Article

minimal, with the main cause of contact angle reduction upon illumination being the photocatalytic destruction of organic contaminants at the TiO2 surface, exposing an underlying inherently wet TiO2 film. These conclusions are then consistent with those drawn by Shultz et al. from their SFG study and it is especially interesting to note that, in that study, Shultz reports that illumination of a TiO2 surface, prepared by nanoparticulate TiO2 deposition, results in the doubling of surface concentration of free OH groups. As the signal associated with such free OH groups is easily attenuated by dissociative adsorption of an adsorbing entity, hydrogen bonding with an adsorbed entity, or displacement of the IR dipole from the surface normal by an adsorbing entity, it is not unreasonable to conclude that the observed increase in free OH population reflects the removal of such adsorbing entities/contaminants. If such is the case then the doubling of free OH population upon illumination reported by Shultz is reflective of a 50% fractional surface coverage by contaminant organic preillumination. From Figure 5, this value of f 2 corresponds to a measured contact angle of 55°, a value that is within the 55−60° range commonly reported for as prepared TiO2 surfaces in the literature (vide supra).

One further possible limitation of this analysis is sensitivity to the value of θ2 used in eq 2. We have used a value of 90° as provided by the measurements on a stearic acid coated TiO2 surface of Figure 2. However, contact angles of near to 100° have been measured for hydrocarbon based surfaces and it might be expected that polar functional groups would reduce the value of θ2. Thus, we have repeated the analysis of Figures 6−8 and S4−S7 (Supporting Information) using values of θ2 of 80° and 100° (not shown). We found that while the values of f 2 at t = 0 do change (typically by ±10%, also not shown), values of θ at t = 0 calculated using these values of f 2 and θ1 = 4° and 25° exhibit the same results as the analysis conducted using θ2 = 90°, that is, that values of θ at t = 0 calculated using θ1= 4° and θ2 = 80° or 100° are closer to the actual measured θ at t = 0 than values calculated using θ1 = 25° and θ2 = 80° or 100°. Our analysis therefore appears to be insensitive to the value of θ2 in the range 80°-100°. In summary, analysis of data from our own experiments and from a range of analogous experiments described in the literature indicate that, at t = 0 s and before any irradiation, effectively flat and nonporous TiO2 surfaces are inherently hydrophilic with a contact angle θ1 close to 4°. As discussed in section 3.3, other works either assert or report a value for θ1 in the range of 25°−32°.3,4 However, inspection of Figure 5 reveals that, for a TiO2 surface with an underlying θ1 value of 4°, a fractional surface coverage by a contaminating organic of only 12% is enough to yield a measured contact angle, θ, of 25°. A fractional surface coverage of only 19% yields a value of θ = 32°. This dramatically illustrates just how susceptible to surface contamination contact angle measurements on TiO2 surfaces are and supports Zubkov et al.’s interpretation of their own data presented in Figure 3, that is, that the photoinduced change in measured contact angle from 25° to 4° observed on a rutile (110) surface under 1 atm O2 in the absence of deliberately added hexane is wholly due to the photocatalytic removal of adventitious trace organics.2 Such a view is also supported by Henderson who recently restated the fact, widely known by UHV scientists, that even brief exposure of surfaces to atmospheres of 10−6 Torr or higher results in surface contamination by a variety of strongly bound species that include organics.29 The rate of this adsorption by contaminant organics is not specified by Henderson; however, it has been quantified by Mills and Crow.8 They recorded an increase in θ of 4°/h for photocatalytically cleaned TiO2 surfaces left in the dark on the laboratory bench, that is, under quiescent conditions similar to those employed during the static/discrete experiments of Figure 7. As described above each, static/ discrete contact angle measurement of Figure 7 takes less than 30 s to accomplish leading us to expect minimal contact angle increase and thus readsorption of contaminant organic during each short measurement period. In contrast, the above-cited work of Hennessey et al. and Zubkov et al. involves samples being subjected to a postcleaning annealing treatment with associated cooling periods in dry O2 air. Such cooling periods would typically be of several hours, during which it is entirely conceivable that enough adventitious contaminant organic can be adsorbed at the nominally clean titania surface to raise the initial contact angles measured by Hennessey et al. and Zubkov et al. to the vicinity of 25−30°. Indeed, the latter specifically attributes the elevated value for their initial contact angle to just such a phenomenon. Taken in total, these observations are then strongly suggestive of photoinduced surface reconstruction being

4. CONCLUSIONS The time dependence of photoinduced superhydrophilicity (PISH) on mesoporous TiO2 films, previously shown to behave as being effectively flat and nonporous, was measured by goniometric observation of the contact angle, θ, of sessile water drops at the film surfaces. In these measurements, the TiO2 substrate was illuminated by 315 nm light and drops were applied at a range of illumination times. From these measurements, the time dependence of f 2, the fractional surface coverage of the TiO 2 surface by adventitious contaminating organics, was calculated using a model for the wetting of flat, nonporous heterogeneous surfaces derived by Israelachvili and Gee.1 Extending this model to include a Langmuir−Hinshelwood based kinetic analysis of f 2 as a function of time allowed for calculation of k, the pseudo-first-order rate constant for the photocatalytic destruction of adventitious hydrocarbon contaminant adsorbed at the TiO2 surface. In so doing, we are for the first time providing a framework within which contact angle versus time data from photoinduced superhydrophilicity experiments may be processed to yield quantitative kinetic information relating to the photocatalytic behavior of the underlying semiconductor material. This approach also provides insight into the intrinsic surface properties of the unilluminated TiO2 material as it also allows for the determination of a value of f2, and thus f1, at t = 0. These quantities were used in conjunction with θ2 = 90° (measured from a TiO2 surface coated with stearic acid as a model contaminant) and eq 2 to calculate two predicted values for the experimentally observed contact angle θ at t = 0 using θ1 values of 4° and 25°. These two calculated values of θ at t = 0 were then compared with the experimentally observed value of θ in order to determine which value of θ1 gives closest agreement and so is representative of θ1 at t = 0. The two values of θ1 of 4° and 25° were chosen as they are representative of the two common schools of thought regarding this quantity in the literature. Use of this approach in the analysis of both our θ versus time data and analogous data sets for flat, nonporous TiO2 samples derived from literature sources suggests that at t = 0, a single 17654


Langmuir

Article

component TiO2 surface will have a contact angle of θ1 = 4° and so is intrinsically “wet”/hydrophilic. Measured contact angles in excess of this value are then due to the adsorption by surface contaminant as originally suggested by, inter alia, Zubkov et al. particularly in reference to their own data.2 These observations are strongly suggestive of photoinduced surface reconstruction being minimal, with the main cause of contact angle reduction upon illumination being the photocatalytic destruction of organic contaminants at the TiO2 surface, exposing an underlying inherently hydrophilic TiO2 film. We have found that this approach can be exported to mesoporous surfaces that behave as if they are flat, nonporous surfaces, that is, those that exhibit a vigorous phase during sessile drop evaporation experiments in the dark. However, we also note that the general extension of this approach to mesoporous surfaces and especially those with a controlled mesoporosity would require the rederivation of the Israelachvili and Gee model using the Wenzel equation rather than the Young−Dupré equation. This will form the basis of further work.

■

Generation: Trace Hydrocarbons and Hydroxyl Groups. Langmuir 2003, 19, 7330. (7) Denison, K. R.; Boxall, C. Photoinduced ″stick-slip″ on superhydrophilic semiconductor surfaces. Langmuir 2007, 23, 4358. (8) Mills, A.; Crow, M. A study of factors that change the wettability of titania films. Int. J. Photoenergy 2008, 470670. (9) Yan, X.; Abe, R.; Ohno, T.; Toyofuku, M.; Ohtani, B. Action spectrum analyses of photoinduced superhydrophilicity of titania thin films on glass plates. Thin Solid Films 2008, 516, 5872. (10) Järn, M.; Xu, Q.; Lindén, M. Wetting Studies of Hydrophilic− Hydrophobic TiO2@SiO2 Nanopatterns Prepared by Photocatalytic Decomposition. Langmuir 2010, 26, 11330. (11) Cassie, A. B. D. Contact angles. Discuss. Faraday Soc. 1948, 3, 11. (12) Yu, J. C.; Yu, J. G.; Ho, W. K.; Zhao, J. C. Light-induced superhydrophilicity and photocatalytic activity of mesoporous TiO2 thin films. J. Photochem. Photobiology, A 2002, 148, 331. (13) Mills, A.; Wang, J. S. Simultaneous monitoring of the destruction of stearic acid and generation of carbon dioxide by selfcleaning semiconductor photocatalytic films. J. Photochem. Photobiol., A 2006, 182, 181. (14) Troughton, E. B.; Bain, C. D.; Whitesides, G. M.; Nuzzo, R. G.; Allara, D. L.; Porter, M. D. Monolayer films prepared by the spontaneous self-assembly of symmetrical and unsymmetrical dialkyl sulfides from solution onto gold substrates: structure, properties, and reactivity of constituent functional groups. Langmuir 1988, 4, 365. (15) Mills, A.; Elliott, N.; Parkin, I. P.; O’Neill, S. A.; Clark, R. J. Novel TiO2 CVD films for semiconductor photocatalysis. J. Photochem. Photobiol., A 2002, 151, 171. (16) Sawunyama, P.; Fujishima, A.; Hashimoto, K. Photocatalysis on TiO2 Surfaces Investigated by Atomic Force Microscopy: Photodegradation of Partial and Full Monolayers of Stearic Acid on TiO2(110). Langmuir 1999, 15, 3551. (17) Simonsen, M. E.; Jensen, H.; Li, Z. S.; Sogaard, E. G. Surface properties and photocatalytic activity of nanocrystalline titania films. J. Photochem. Photobiol., A 2008, 200, 192. (18) Sakatani, Y.; Grosso, D.; Nicole, L.; Boissiere, C.; Soler-Illia, G.; Sanchez, C. Optimised photocatalytic activity of grid-like mesoporous TiO2 films: effect of crystallinity, pore size distribution, and pore accessibility. J. Mater. Chem. 2006, 16, 77. (19) Shackleford, S. G. D.; Boxall, C.; Port, S. N.; Taylor, R. J. An in situ electrochemical quartz crystal microbalance study of polycrystalline gold electrodes in nitric acid solution. J. Electroanal. Chem. 2002, 538−539, 109. (20) Mills, A.; LeHunte, S. An overview of semiconductor photocatalysis. J. Photochem. Photobiol., A 1997, 108, 1. (21) Miyauchi, M.; Kieda, N.; Hishita, S.; Mitsuhashi, T.; Nakajima, A.; Watanabe, T.; Hashimoto, K. Reversible wettability control of TiO2 surface by light irradiation. Surf. Sci. 2002, 511, 401. (22) Bourges-Monnier, C.; Shanahan, M. E. R. Influence of Evaporation on Contact Angle. Langmuir 1995, 11, 2820. (23) Zhang, N.; Yang, W. J. Trans. ASME, Ser. C 1982, 104, 56. (24) Turchi, C. S.; Ollis, D. F. Photocatalytic degradation of organic water contaminants: Mechanisms involving hydroxyl radical attack. J. Catal. 1990, 122, 178. (25) Sakai, N.; Wang, R.; Fujishima, A.; Watanabe, T.; Hashimoto, K. Effect of ultrasonic treatment on highly hydrophilic TiO2 surfaces. Langmuir 1998, 14, 5918. (26) Watanabe, T.; Nakajima, A.; Wang, R.; Minabe, M.; Koizumi, S.; Fujishima, A.; Hashimoto, K. Photocatalytic activity and photoinduced hydrophilicity of titanium dioxide coated glass. Thin Solid Films 1999, 351, 260. (27) Wang, J.; Li, H.; Li, H.; Zuo, C.; Wang, H. Thermal Stability and Optimal Photoinduced Hydrophilicity of Mesoporous TiO2 Thin Films. J. Phys. Chem. C 2012, 116, 9517. (28) Wenzel, R. N. Resistance of Solid Surfaces to Wetting by Water. Ind. Eng. Chem. 1936, 28, 988. (29) Henderson, M. A. A surface science perspective on photocatalysis. Surf. Sci. Rep. 2011, 66, 185.

ASSOCIATED CONTENT

S Supporting Information *

S1, S2, and S3: Raman and UV/vis spectra of as prepared TiO2 films and FTIR of stearic acid films, respectively. S4−S7: Detailed Israelachvili and Gee based analysis of the four cited literature data sets. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*Telephone: +44 1524 593109. Fax: +44 1524 381707. E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors wish to thank the Engineering and Physical Science Research Council UK for provision of a CASE award to support P.S.F., and Oxley Developments Ltd. and The Lloyds Register Educational Trust for additional financial support. The Lloyds Register Educational Trust is an independent charity working to achieve advances in transportation, science, engineering and technology education, training and research worldwide for the benefit of all.

■

REFERENCES

(1) Israelachvili, J. N.; Gee, M. L. Contact angles on chemically heterogeneous surfaces. Langmuir 1989, 5, 288. (2) Zubkov, T.; Stahl, D.; Thompson, T. L.; Panayotov, D.; Diwald, O.; Yates, J. T. Ultraviolet Light-Induced Hydrophilicity Effect on TiO2(110)(1 × 1). Dominant Role of the Photooxidation of Adsorbed Hydrocarbons Causing Wetting by Water Droplets. J. Phys. Chem. B 2005, 109, 15454. (3) Fujishima, A.; Zhang, X.; Tryk, D. A. TiO2 photocatalysis and related surface phenomena. Surf. Sci. Rep. 2008, 63, 515. (4) Hennessy, D. C.; Pierce, M.; Chang, K. C.; Takakusagi, S.; You, H.; Uosaki, K. Hydrophilicity transition of the clean rutile TiO2 (110) surface. Electrochim. Acta 2008, 53, 6173. (5) Fujishima, A.; Rao, T. N.; Tryk, D. A. Titanium dioxide photocatalysis. J. Photochem. Photobiol., C 2000, 1, 1. (6) Wang, C.-y.; Groenzin, H.; Shultz, M. J. Molecular Species on Nanoparticulate Anatase TiO2 Film Detected by Sum Frequency 17655


Photoinduced Superhydrophilicity: A Kinetic Study of Time Dependent

Recommend Documents