Applied Catalysis A: General 411–412 (2012) 60–69 Contents lists available at SciVerse ScienceDirect Applied Catalysis A: General journal homepage: www.elsevier.com/locate/apcata Ultrasound assisted preparation of stable water-based nanocrystalline TiO2 suspensions for photocatalytic applications of inkjet-printed films I. Fasaki a, K. Siamos a, M. Arin b, P. Lommens b, I. Van Driessche b, S.C. Hopkins c, B.A. Glowacki c, I. Arabatzis a,∗ a NanoPhos S.A., PO Box 519, Science and Technology Park of Lavrio, Lavrio 19500, Attica, Greece SCRiPTS, Department of Inorganic and Physical Chemistry, Ghent University, Krijgslaan 281 (S3), 9000 Ghent, Belgium c Department of Materials Science and Metallurgy, University of Cambridge, Pembroke Street, Cambridge CB2 3QZ, United Kingdom b a r t i c l e i n f o Article history: Received 29 June 2011 Received in revised form 14 October 2011 Accepted 15 October 2011 Available online 20 October 2011 Keywords: Stability of water based suspensions TiO2 nanoparticles Ultrasonication Self-cleaning surfaces Inkjet printing a b s t r a c t The use of titania photocatalytic materials in industrial applications is strongly dependent on the stability, nanoparticle size distribution, ease of deposition and cost of the relevant titania precursor solutions or suspensions. The present contribution presents the preparation of inkjet-printed titania films, derived from stable water-based suspensions. The suspensions were synthesized by applying a “top-down” synthetic strategy, namely the ultrasonication of commercially available titania powder (Evonik Aeroxide P25). Crucial parameters, such as suspension stability, energy input requirements, particle size distribution, surface characteristics, compatibility with industrially proven inkjet systems and photocatalytic performance were investigated. The developed synthetic procedure proves environmentally friendly, low cost and most suitable for large scale production of titania thin films, by inkjet printing commercially available ceramic tiles. © 2011 Elsevier B.V. All rights reserved. 1. Introduction Titanium dioxide heterogeneous photocatalysis has proved to be among the most effective advanced oxidation processes (AOPs). The ability of solar or artificial light to promote the production of strongly oxidizing species (OH• and O2 •− radicals) on the semiconductor surface enables applications with a low operating temperature, low energy consumption and low cost [1]. This property makes the material – especially on the nano-scale – suitable for a wide range of applications including natural contaminated systems [2,3], air purification [4] and antimicrobial protection [5]. TiO2 on the nano-scale has previously been used in solar cells [6], gas sensors [7], dielectric applications [8] and extensively as a photocatalytic material for self-cleaning surfaces such as glass [9,10] and fabric [11]. The photocatalytic performance of the TiO2 coating strongly depends on the preparation method and the application method. The application of TiO2 suspensions on surfaces for the formation of thin films has previously been demonstrated by the doctor blade method [12], electrophoretic deposition [13], plasma spraying [14] and inkjet printing [15–17]. Even though photocatalytic applications appear very promising, the deposition of large surface area titania films for industrial ∗ Corresponding author. Tel.: +30 22920 69312; fax: +30 22920 69303. E-mail address: iarabatz@NanoPhos.com (I. Arabatzis). 0926-860X/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.apcata.2011.10.020 or commercial use has been limited, as titania suspension precursors have not been adequately coupled with an industrially proven deposition technique, such as spraying, rotary drum deposition or inkjet printing. Chemical vapour deposition (CVD) provides good quality and transparent films but at high cost; while doctor blade deposition is an affordable and simple deposition technique, but it produces thick and opaque films: simply unacceptable properties in most everyday applications. In parallel, crucial factors that affect the industrial adoption rate of photocatalytic applications have largely been neglected: the cost, volatile organic content, flammability, stability and shelf life of titania precursor suspensions. The sol–gel synthetic technique is promising in fine-tuning specific materials properties; however, the (i) high cost of raw metal alkoxides or salts, (ii) large shrinkage during processing, (iii) health hazard of alcohol or non-polar organic solvents and (iv) extended processing time do not lower the industrial adoption barriers. This fine, “bottom-up” synthesis strategy (from alkoxides or salts to nanoparticles) has been thoroughly studied for the preparation of titania nanoparticles, even though it is susceptible to temperature variations or raw materials purity. This contribution presents the synthesis of stable, completely water-based titania suspensions, specially designed to be applied by inkjet printing. Instead of applying the “bottom-up” sol–gel process, the “top-down” ultrasonication of commercially available titania powder has been selected. The selection of a physical, I. Fasaki et al. / Applied Catalysis A: General 411–412 (2012) 60–69 deaggregation method is energy efficient and fast enough for the scaling up of suspension production. Additionally, the whole process remains unaffected by the high cost of metal alkoxides or their inherent impurity. In contrast to the sol–gel process of titanium alkoxides/salts (titanium (IV) butoxide, isopropoxide or chloride), there are no harmful by-products (alcohols or hydrochloric acid) and the final suspension remains extremely low in volatile organic compound (VOC) content. The development of the proposed synthetic method was based on qualitative and quantitative criteria. An orthogonal array for the design of the experiments was adopted to reduce experimental effort and maximize the quantitative assessment of quality characteristics. In particular, optimising the following parameters was crucial for the development of the presented suspensions: • • • • Stability of titania suspensions Energy requirements during the ultrasonication process Requirements for stabilization or dispersion reagents Rheological and surface tension characteristics for successful inkjet printing • Transparency of the deposited titania films • Enhanced photocatalytic performance The stability of water-based titania suspensions has been extensively investigated during the last decade [18–27]. Stabilization of powders in liquid media is still a major problem for various technological processes, since numerous properties of the final product depend strongly on colloidal stability of particles and their distribution in a certain liquid volume. A particle size distribution in the nanoscale is a very crucial parameter for the stabilization of the particles as sedimentation velocity is directly proportional to the particle radius squared. On the other hand, decreasing the particle size (at constant particle number density) increases the relative surface area and therefore the tendency to aggregation is enhanced [28]. Attractive London–van der Waals dispersion forces are the origin of the driving force for colloids to aggregate. The source of the repulsive forces needed to stabilize the suspension against these attractive forces is usually one of two kinds: (i) coulombic repulsion due to electric charges on the particles, i.e. electrostatic interactions between the ionic double layers surrounding the particles and (ii) steric repulsion introduced by large molecules or polymeric chains adsorbed on the particles. The combination of the two mechanisms of stabilization is classified under the labels ‘electrostatic’ or ‘steric’, and their combination ‘electrosteric’ [29]. Various electrolytes, such as the salts NaCl [20,24], Na4 P2 O7 [22], or acids and bases [30], have previously been used for electrostatic stabilization of titania colloidal systems. Steric stabilization can be achieved by the addition of surfactants or dispersants, polymers that adsorb at the solid–liquid interface, e.g. Gemini surfactants [11,23] or polyethylene glycol (PEG) dispersant [19,26]. Finally, electrosteric stabilization can be achieved by the addition of polyelectrolytes [21,25,31]. As already mentioned, the particle size influences the stability of the colloidal suspension. Additionally, the particle size affects the application to self-cleaning surfaces. The smaller the size of the particle, the larger the active surface area and thus the greater the photocatalytic activity. When the titania powder is dispersed in water the nanoparticles create large agglomerates. Energy in the form of ultrasound is necessary for the breakage of these agglomerates and their dispersal, which is controlled predominantly by the specific energy input, E/V (energy per dispersion 61 volume) and frequency [18,32]. The specific energy (E/V) can be calculated according to Eq. (1): E P·t = V V (1) where P (W) is the power of the ultrasonication, t (s) is the ultrasonication time, E (J) is the energy and V (L) is the suspension volume. Inkjet printing is a printing common deposition technique for text and images, typically on paper or other substrates. In the last few years it has been used as a free-form fabrication method for building three dimensional parts and is being explored as a way of printing sol–gel materials, polymers, ceramics and nanoparticles [33]. This simple deposition method offers a low cost, fast and convenient route for the controllable growth of coatings with low material waste, and hence it is considered a promising deposition technique for industrial applications. Inkjet printing can be used for the printing of suspended nanoparticles. The most important parameters for successful printing of these materials are the particle size and stability of the suspension, to avoid blocking the nozzle, and the viscosity and surface tension to permit reliable printing. Since sedimentation must be avoided, stability is crucial and particle size should be less than a micrometer. The dispersion must exhibit qualitative characteristics to avoid any aggregation that would increase the viscosity, thus severely limiting the maximum, small sized particle volume fraction. In this work, the electrostatic and steric stability of titania nanoparticles in water suspension and their application on glass and glazed tiles for photocatalytic thin films is discussed. Ultrasonication was applied to the suspensions for the dispersal of titania nanoparticles. The influence of dispersants, titania concentration and ultrasonication time was studied. TiO2 films were deposited by the doctor blade method and by inkjet printing on glass. The degradation of methyl orange under UV illumination was measured for the determination of photocatalytic activity of the coatings. 2. Experimental details 2.1. Materials and preparation The investigated suspensions were obtained by adding titania nanopowder (Evonik Aeroxide P25), dispersants (Air Products) and a salt (Na4 P2 O7 , Panreac) in deionized water. The commercially available dispersants Zetasperse 1200, Zetasperse 3100 and Surfynol CT-231 were purchased from Air Products and were used without any further purification. Their use offers outstanding wetting and efficient particle stabilization providing the advantage of low dispersion viscosity, high pigment load, gloss and color development improvement. The final mixture was set to a volume of 1 L and the temperature during the whole mixing was held constant at 25 ◦ C. The modified suspensions were ultrasonicated at 1000 W for several minutes by a sonotrode operating at 20 kHz [34]. The ultrasonication system consists of a UIP1000hd ultrasonic processor (Hielscher Ultrasonics GmbH, adjustable power 500–1000 W) with a robust stainless steel reactor vessel. The system is connected to a continuous stirred-tank reactor (CSTR). Coupling of the ultrasonicator with the CST reactor was designed and performed at NanoPhos SA, so that the system allows continuous recirculation, for scaling up to the production of large volumes (100 L or more) of nanoparticles suspensions. In order to benchmark the performance of the TiO2 thin films obtained by the suspensions with a commercial photocatalytic glass, Saint Gobain Bioclean Glass was purchased and its photocatalytic activity was measured without any treatment of the surface. 62 I. Fasaki et al. / Applied Catalysis A: General 411–412 (2012) 60–69 2.2. Characterization of the suspensions The particle size distributions of the suspensions were measured using a MasterSizer 2000 (Malvern Instruments) with Laser Light Scattering and the zeta potential of the suspensions was measured by a ZetaSizer 3000 with Phase Analysis Light Scattering (Malvern Instruments) at a pH range 2.5–10 which was adjusted by standard solutions. The viscosity of the suspensions was measured by Viscometer VM2, Sheen Instruments equipped with a low viscosity adaptor. The surface tension of the samples was measured by an optical tensiometer Attension, Theta Lite, KSV Instruments in pendant drop shape analysis. Transmission electron microscopy (TEM) (JEOL JEM-2200FS) was applied onto the suspensions in order to measure the particle size and to investigate the crystallinity. Specimens for TEM studies were prepared by depositing a drop of diluted aqueous suspension (1 mL of sol in 9 mL of ethanol) onto a 300 mesh holey carbon copper grid. The suspensions were dried at 120 ◦ C (until weight difference among three subsequent measurements was not observed) and the resulting powder was measured without any further treatment. X-ray diffraction (XRD) patterns of titania Evonik Aeroxide P25 powder and powder obtained from an ultrasonicated sample using a Siemens D 5000 in the –2 mode were recorded. Cu K␣ radiation ( = 1.5418 Å) was chosen for excitation. Nitrogen adsorption experiments were performed and isotherms were recorded at 196 ◦ C with a Belsorp-mini II gas analyzer. The surface area was determined using the Brunauer–Emmett–Teller (BET) method by adsorption/desorption of liquid N2 monolayer on the surface. 2.3. Thin films by doctor blading TiO2 films were deposited on glass microscope slides by the doctor blade method and were heated at 500 ◦ C for 1 h. This application method was applied first, in order to optimise the experimental parameters. The surface and morphology of the films was studied by atomic force microscopy (AFM) (Molecular Imaging, PicoPlus) in non-contact mode and scanning electron microscopy (SEM) (FEI Philips Quanta 200). The thickness of the films was measured by a spectroscopic ellipsometer (SE, J.A. Woollam Co. Inc., M-2000FI) where the Cauchy dispersion model was used to fit the optical constants. 2.4. Photocatalysis assessment The photocatalytic activity of the obtained films was evaluated by following the degradation of methyl orange (in aqueous solution, prepared from Sigma–Aldrich powder) under UV illumination. The experiments were carried out in round-bottomed photocatalytic cells with a near UV-transparent window (cut off below 340 nm). A laboratory constructed irradiation box equipped with four Sylvania GTE 15W F15W/T8 blacklight blue fluorescent light tubes was used. The photon source has a maximum emission at 360 nm and emits 71.7 ␮W cm−1 at a distance of 25 cm. The azo-dye solutions were used without prior oxygen gas bubbling. The concentration was correlated to the absorption of the methyl orange solution at 464 nm, using a single beam Shimadzu UV 1240 spectrophotometer. The titania coated microscopy slides were accurately shaped to achieve a surface area of 1.0 cm2 (cut after the deposition) and inserted in the photocatalytic cell. Photocatalysis experiments took place under stirring. Table 1 Plan of the experiments. The values 1–4 specified for each parameter (E/V, dispersant and TiO2 concentration) correspond to the definitions in Table 2. Suspension E/V Dispersant TiO2 1 2 3 4 5 6 7 8 9 10 11 12 1 1 1 2 2 2 3 3 3 4 4 4 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 2 3 1 3 1 2 tests with a drop-on-demand system using a solenoid microvalve modified in the University of Cambridge from a commercial Domino MacroJet 2 printer. This electromagnetic printing device was fitted with a 90 ␮m diameter jewel orifice, and was selected for its compatibility with a wide range of ink viscosities and particle sizes. The MacroJet technology is also industrially proven for high throughput low resolution printing. The jetting parameters (pressure and valve opening time) were optimised, and reliable jetting confirmed, using a drop visualisation system comprising a high-sensitivity camera with 1292 × 964 px resolution at 30 frames s−1 (Allied Vision Technologies, Stingray F-125B) and a telecentric zoom lens (ML-Z07545, Moritex). Collimated, strobed LED illumination was used, synchronized with droplet ejection with a selectable delay time, such that each frame corresponded to the drop geometry a chosen time after ejection. Image analysis was used to characterize droplet break-up and quantify the ejected ink volume and velocity. The electromagnetic nozzle, mounted on an X–Y positioning system with drive electronics and printing software developed in house, was used to print test coatings on both glass slides and commercially available ceramic tiles. The deposition was performed at room temperature without a controlled atmosphere. 3. Results and discussion 3.1. Titania suspensions The experimental parameters under investigation (Table 1) were selected to be (i) the titania concentration, (ii) the choice of commercial dispersant and (iii) the ultrasonication specific energy (by adjusting the duration of ultrasound application). Three different titania concentrations, three commercial dispersants with a concentration fixed at 10% (w/w) of the titania concentration and four different ultrasonication specific energies (energies per volume) were investigated (Table 2). A preliminary set of experiments indicates that 10% (w/w) is the ideal ratio between the dispersant and titania. Increasing the ratio in favour of the dispersant is limited by its solubility and compatibility in the system. At a ratio below 10% (w/w), the steric hindrance of agglomeration is limited and subsequently the system gets destabilized. The same applies for the electrostatic stabilization for Table 2 The values of the experimental parameters (specific ultrasonication energy, titania concentration and dispersant) tested. 2.5. Inkjet printing Following the optimisation of suspension rheological properties, the particles size distribution and the photocatalytic performance, a scalable suspension was selected for inkjet printing 1 2 3 4 E/V (J/L) Dispersant TiO2 (%, w/w) 1.8 × 105 3.0 × 105 4.2 × 105 5.4 × 105 Zetasperse 1200 Zetasperse 3100 Surfynol CT-231 – 0.1 0.5 1.0 – I. Fasaki et al. / Applied Catalysis A: General 411–412 (2012) 60–69 Fig. 1. Schematic illustration of the definition of agglomerates and aggregates (a) and schematic of the possible breakage mechanism of the powder (b). sodium pyrophosphate at a concentration level of 5 × 10−4 M. The combination of a dispersant and a salt provides an effective stabilization medium, allowing the minimization of variables that would affect the preparation of ultrasound assisted titania suspensions for photocatalytic applications. A design of experiments approach based on an orthogonal array was applied in order to minimize the number of trials required to determine the preferred processing conditions (Table 1). This range of titania concentration was chosen to combine both deposited films transparency and enhanced photocatalytic activity. The values of specific ultrasonication energy were carefully chosen in order to be easily scalable (accelerated production rate) and, at the same time, obtain the desired properties of the titania particles. It has been demonstrated [35] that, when ultrasonication is applied to a given liquid formulation with identical processing parameters, the result depends only on the specific energy and not the scale of processing. Therefore, when 5.4 × 105 J/L is applied to 1 L of TiO2 suspension for 9 min, the equivalent time for a production of 100 L is 15 h, already long enough for an industrial process of this volume. Titania nanopowders are usually described as non-eroding particles or aggregates. The term ‘agglomerate’ relates to powders with weak inter-particle bonds that can be re-dispersed in a solvent, and the term ‘aggregate’ or ‘hard agglomerate’ describes powders with primary particles held together by strong attractive forces (Fig. 1a). Agglomerates break by an erosion mechanism, in which bonds between primary nanoparticles are breached and nanoparticles or small agglomerates are separated one by one. Mechanical attrition of aggregates, on the other hand, occurs by fracture mechanisms. Fracture of an aggregate initiates at a surface flaw or imperfection. Large aggregates are broken down into smaller aggregates until there are no more defects to initiate the breakage, after which further energy input results in no significant attrition [36,37]. In Fig. 2, the particle size distribution of titania powder in water without ultrasonication, salt and dispersant is presented. It is clear that the titania nanoparticles create large agglomerates and 63 aggregates when dispersed in water, indicating the necessity for ultrasonication of the suspensions. After ultrasonication and the addition of the salt and dispersant the particle size distribution of the titania particles changed dramatically into a bimodal distribution with peaks near 100 nm and 1 ␮m (Fig. 2b). The position and relative magnitude of the peaks are most strongly affected by the specific energy applied, but the majority of the particles are in the sub-micron range (D90 < 0.7 ␮m) for all the samples. An increase of the ultrasonication specific energy results in the increased breakage of agglomerates, decreasing both the mean particle size and the width of the distribution, the tail of the distribution above 2 ␮m becoming negligible for specific energies of 3.0 × 105 J/L and above. At the highest tested specific energy, 5.4 × 105 J/L, aggregates with size close to 1 ␮m break to smaller ones but never disappear, which indicate that they are primary particles or more energy is required for the breakage of the aggregates. A schematic of the effect of the ultrasonication on the particles is illustrated in Fig. 1b. As proved by extending the ultrasonication duration to several hours, the same tendency persists (increase of particle size distribution around 100 nm and decrease of the 1 ␮m peak), even though the shoulder around 1 ␮m never disappears. Moreover, the photocatalytic activity enhancement, after prolonged ultrasonication, is insignificant, so a combination between performance of the coating and equivalent process production time can be achieved. For example, when doubling the energy per volume from E/V = 12 × 105 J/L (the latest corresponding to 33 h of ultrasonication for 100 L) the gain in the reaction rate constant is only 5%. The size distribution was only slightly affected by the type of dispersant and the titania concentration. As expected, these commercially available non-ionic dispersants, mainly consisting of diols, have proved suitable for the stabilization of titania particles in aqueous dispersions. One part of the chain is anchored on the particle and the other is free in the liquid medium hindering the contact with other TiO2 particles. Although it is known that the higher the concentration of particles the more difficult it is to obtain a stable suspension, in the present contribution, the titanium dioxide concentration (0.1–1%, w/w) level did not affect the stability of the suspension. A common way to evaluate the stability of colloidal dispersions is by determining the zeta potential magnitude. The zeta () potential is a function of the surface charge of the particles. In dispersions where the value of the  potential is close to zero (isoelectric point), particles tend to agglomerate. At highly negative or positive values of  potential (more than 30 mV or less than −30 mV) particles in dispersions tend to repel each other and no agglomeration occurs. The actual values needed to prevent re-agglomeration are not well established, and depend on the solvent (in our case, water), ionic strength, effective pH, and the functional groups on the surface of nanoparticles. The  potentials of the TiO2 powder alone, TiO2 with salt, TiO2 with dispersant, and TiO2 with salt and dispersant were measured versus pH in order to determine the influence of each on the Fig. 2. Particle size distribution of titania powder in water and suspensions ultrasonicated with different specific energies (suspensions 2, 5, 7, and 12). 64 I. Fasaki et al. / Applied Catalysis A: General 411–412 (2012) 60–69 Table 3 Viscosity and surface tension of the suspensions. titania powder titania with salt titania with dispersant 1 titania with salt and dispersant 1 15 10 5 ζ potential (mV) 0 2 -5 3 4 5 6 pH 7 8 9 10 11 -10 -15 -20 -25 -30 Suspension Viscosity (cP) Surface tension (mN/m) 1 2 3 4 5 6 7 8 9 10 11 12 1.43 1.35 1.42 1.36 1.40 1.41 1.40 1.37 1.38 1.43 1.36 1.37 69.25 61.13 55.11 65.20 57.78 49.67 50.09 42.53 58.30 36.96 50.28 43.65 -35 -40 Fig. 3. Zeta potential of titania suspensions in water versus pH. surface charge of the particles (Fig. 3). The diagram displayed in Fig. 3 is for suspensions with Zetasperse 1200 (dispersant 1). The same behaviour was observed for the other dispersants as their chemical structure is similar. The measured isoelectric point of the TiO2 powder occurs at a pH value of 4.9 and it was observed that the  potential has more negative values in an alkaline environment. The measured values of the isoelectric point for Evonik Aeroxide P25 powder from previous reports are in the 5.5–6.0 pH range [11,22,38]. The addition of the dispersant leads to marginal change of particle surface charge, leaving the isoelectric point practically unchanged. As expected, the addition of the salt in the suspension has a much stronger effect than any other additive on the titania particles surface charge. In the presence of the salt, the zeta potential is only slightly affected by changes in pH and no isoelectric point was observed in the measured range. The surface charge is controlled by the salt, which can dissociate into multiply charged ions. If a particle is ionic or has highly polar bonds, multiply charged ions may be adsorbed by the particle in an aqueous environment, leading to an increase in particle surface charge and zeta potential. The ionic strength I in the suspension was calculated, using Eq. (2), as 5 × 10−3 M: n I= 1 2 ci zi 2 (2) i=1 where ci is the molar concentration of ion i (mol/L), zi is the charge number of that ion, and the sum is calculated over all ions in the solution. The salt concentration was carefully chosen as according to the literature the ionic strength range 10−2 –10−4 M usually assures the most stable suspension conditions [22,39–41]. Actually, sodium pyrophosphate dissociates in bulky and multicharged ions that heavily impact the ionic strength. This was one of the main reasons for the selection of the specific salt type. Ionic strength is an important parameter that controls the colloidal stability. Minimal or zero ionic strength values result in nanoparticles coming in close contact and form a large agglomerate. By increasing the ionic strength, a large repulsive barrier prevents the particles from falling into their original condition (agglomerates) and thus they are kept dispersed in the suspension. This is a much expected effect, as the increase of ionic strength causes a decrease in the effective thickness of the diffuse layer. For even higher electrolyte concentrations, the double layer becomes thinner, and colloidal particles approach each other, increasing the probability of coagulation. Elevated values of ionic strength adversely affect the colloidal stability by inducing attractive forces among the charged nanoparticles. In Table 3 the measured values of surface tension and viscosity of the suspensions are presented. Thorough analysis of the results reveals the correlation of the titania concentration and the particle size distribution. In particular, the TiO2 concentration increase leads to a surface tension decrease, due to both the hydrophilic character of titania particles and the chemical nature of dispersants, which assists on the wetting behaviour. This fact is evident by comparing samples of the same energy per volume (for example suspensions 1, 2, 3 or 10, 11, 12) but different TiO2 concentration. Increasing the ultrasonication energy per volume, small nanoparticles decrease the surface tension of the suspension. It is interesting to observe that suspension 1 (lowest energy per volume and titania concentration) has the highest surface tension, close to the value of pure water which is 72.8 mN/m. In addition, the viscosity of all suspensions is very close to water (1.00 cP), with a poor correlation among them. It is important to mention that, for the titania suspensions either containing only sodium pyrophosphate and no dispersant, or only dispersant and no salt, there was sedimentation of solids within a month. The combination of both stabilizers, which was applied for all the twelve presented suspensions, no agglomeration was observed after at least 9 months, and the particles of titania were sufficiently dispersed in the aqueous medium. All the dispersions could be fully recovered to their initial form after stirring or shaking. Even though the selected dispersants offer exceptional wetting and chemical system compatibility, the selection of Zetasperse 1200 was based on qualitative experimental criteria: in specific, the performance of Surfynol CT-231 and Zetasperse 3100 proved inadequate after 12 months of storage. On the other hand Zetasperse 1200 helps in maintaining the homogeneity of the original suspensions for at least 18 months, justifying its selection as a superior dispersant for extended shelf life. The crystalline structure of Evonik Aeroxide P25 powder was examined by X-ray diffraction (XRD). The anatase (A) diffraction pattern (Fig. 4) dominates but rutile (R) is also present, with an approximate composition of 75% anatase and 25% rutile weight fractions calculated by the following formula: xR =  1 + 0.8 · IA IR −1 × 100% (3) where IA and IR are the intensities of anatase and rutile respectively. Comparing the two patterns two important findings are clear: the crystalline structure of the TiO2 does not change by the treatment with ultrasounds (exactly the same peaks are observed) and the crystallite size has been reduced. In particular, the grain size was determined from the FWHM of the A (1 0 1) peak, according to the Scherrer formula assuming no crystal distortion in the lattice: D= k· FWHM · cos  (4) I. Fasaki et al. / Applied Catalysis A: General 411–412 (2012) 60–69 65 Table 4 RMS roughness (area 2 ␮m × 2 ␮m) and thickness of the TiO2 thin films. Fig. 4. XRD patterns of TiO2 powder before and after ultrasonication (US). where k is a constant equal to 0.9,  is the X-ray wavelength 1.54 Å, FWHM is the full width at half maximum and  is the diffraction angle. The calculated crystallite size value of TiO2 Evonik Aeroxide P25 powder without any treatment is 25 nm and the calculated value for the dry powder of suspension 10 is 18 nm. TEM images obtained by suspension 3 and 10, with energy per volume 1.8 and 5.4 × 105 J/L respectively, are shown in Fig. 5. The TiO2 nanoparticles exhibit an irregular shape rather than a sphere, but this is the typical appearance of Evonik Aeroxide P25 without any treatment [42]. The effect of ultrasounds is in accordance to the results obtained by XRD, as smaller nanoparticles are visible in the image of suspension with the higher energy per volume, although there is not a narrow particle size distribution in both cases. It is important to distinguish the particle size measured by Laser Light Scattering, X-ray diffraction and TEM analysis. With the first technique what is measured is the hydrodynamic diameter, namely the aggregates or agglomerates of nanoparticles in water, in our case ranging from 35 nm to 2 ␮m (Fig. 2). By applying the Scherrer formula from XRD patterns the crystallite size can be calculated, but the accuracy is limited by assumptions. By using TEM the crystalline structure and nanoparticles of TiO2 with a size of 15–30 nm can be clearly seen (Fig. 5). The active surface of the TiO2 powders was calculated by BET method at 47 m2 /g for untreated powder, remaining practically unchanged by the ultrasonication process in every case. On the contrary, an increase of the mean pore diameter with the increase of ultrasonication energy per volume was observed which means that the interparticle distance is affected. The value for untreated TiO2 powder is at 23 nm while it reaches a value of 51 nm for the powders obtained with the highest E/V. The same tendency was observed also for total pore per volume, increasing from 0.28 cm3 /g to 0.48 cm3 /g respectively. 3.2. TiO2 thin films by doctor blading The deposition of TiO2 suspensions by doctor blade resulted in thin films with scratch resistance and adherence as only strong mechanical force can remove titania from the glass substrate. The SEM image (Fig. 6) illustrates the rough, sponge-like structure of doctor-blade TiO2 films, consisting of particles which form a complex network of high mountains and deep valleys. No obvious difference on the length scales imaged by SEM was observed between the samples. AFM images of coatings prepared from suspensions with 1% (w/w) TiO2 (suspension 10) and 0.5% (w/w) TiO2 (suspension 12) are illustrated in Fig. 7. Observing the 2D image, Fig. 6c, TiO2 Suspensions 11 12 10 TiO2 concentration RMS roughness (nm) Thickness (nm) 0.1% 4 106.0 ± 0.6 0.5% 8 107.6 ± 0.3 1.0% 11 119.5 ± 0.6 particles having a size of 100–200 nm on the surface can be seen. The RMS roughness values of films with each titania concentration are listed in Table 4, along with thicknesses obtained by ellipsometry. The increase of the titania concentration leads to an increase of the film roughness but minor change in coating thickness. The ultrasonication specific energy and the type of dispersant did not affect the surface analysis results. The photocatalytic decomposition of methyl orange was studied as a function of time by measuring the absorbance change of a solution of the dye in contact with TiO2 under UV-illumination. The selection of this specific azo-dye was not solely based on the fact that it can be quantified spectrophotometrically. Environmentally, azo-dyes account for more than 50% of the effluents of the textile industry and one of the most prominent among them is methyl orange [43]. Both methyl orange and the by-products of its hydrolytic decomposition in the environment are mutagens [44]. Additionally, it is most important that methyl orange does not exhibit absorption bands near the irradiation wavelength of the lamps in the irradiation box (360 nm). Consequently, the sensitized photocatalysis decomposition mechanism [45–47], according to which the azo-dye absorbs photons and transits to an excited state before oxidizing and transferring electrons to the semiconductor, can be excluded. The photocatalytic process takes place exclusively by the interaction of photons with the titania semiconductor; a phenomenon that induces charge separation of electrons and positive holes (e− /h+ ). The photo-induced charge-separation is responsible for the creation of oxidative radicals, which attack the azo-dye structure, gradually decomposing it. It is reasonably concluded that the photocatalytic mechanism follows first order kinetics, according to the Langmuir–Hinshelwood model [48]. The kinetics is described according to Eq. (5): dC = kobs  dt (5) where  is the coverage of the surface. Therefore, the photocatalytic process follows a pseudo-firstorder kinetic mechanism, especially at the millimolar concentration range of methyl orange (as in the present case). Integrating and taking logarithms of Eq. (5) leads to Eq. (6): ln C C0 = −k · t (6) C being the concentration of methyl orange after photocatalysis time t, C0 is the initial methyl orange concentration and k is the kinetic constant of the photocatalytic reaction. The concentration of the dye can be calculated from the absorbance, A, by combining the Lambert–Beer Law (A = e·b·C) with Eq. (6) (see Eq. (7)): ln A A0 = −k · t (7) e is the molar absorptivity, b is the path length and c is the concentration of the compound in the solution. The photodecomposition kinetics of methyl orange by thin films obtained from deposition of suspension 10 both by doctor blade and inkjet printing is shown in Fig. 8 in comparison to a commercial photocatalytic glass Saint Gobain Bioclean. In all cases, is clear that under the experimental conditions applied, the photocatalytic curves follow first-order reaction kinetics and the initial reaction 66 I. Fasaki et al. / Applied Catalysis A: General 411–412 (2012) 60–69 Fig. 5. TEM images of suspension 3 (a) and 10 (b). Fig. 6. SEM image of doctor-bladed TiO2 thin film (secondary electron image, 20 kV acceleration voltage) at two magnifications prepared by suspension 10 (left: scale bar 2 ␮m, right: scale bar 1 ␮m). conditions provide a reasonable approximation of the photodegradation rates. The calculated rate constant for each sample is listed in Table 5 and plotted in Fig. 9 where the effect of each parameter, titania concentration, ultrasonication energy per volume and type of dispersant, is evident. The ranking of the samples by photodegradation rate is consistent with expectations based on the titania concentration. This parameter influences proportionally the film photocatalytic activity, comparing the obtained values for 0.1, 0.5 and 1% TiO2 , not only because of the mass load on the glass surface but also the surface roughness as the AFM examination revealed. The ultrasonication specific energy is the parameter that most crucially affects the particle size distribution and consequently the performance for self-cleaning surfaces. Even though BET analysis does not reveal significant changes in terms of total surface area, the interparticle porosity and total pore volume increases significantly. The ultrasound treatment of TiO2 particles leads to a distance increase among them and a porosity opening. It is important to notice that ultrasound treatment did not result in significant change of nanoparticles size: both XRD and TEM results suggest that nanoparticles of titania decrease in size, nevertheless this is attributed mostly on the aggregates cleavage, rather than Table 5 Reaction rate constant of the TiO2 films. Suspension TiO2 concentration (%, w/w) Dispersant E/V (J/L) Rate constant (10−4 min−1 ) 1 2 3 4 5 6 7 8 9 10 11 12 0.1 0.5 1.0 0.1 0.5 1.0 0.5 1.0 0.1 1.0 0.1 0.5 Zetasperse 1200 Zetasperse 3100 Surfynol CT-231 Zetasperse 1200 Zetasperse 3100 Surfynol CT-231 Zetasperse 1200 Zetasperse 3100 Surfynol CT-231 Zetasperse 1200 Zetasperse 3100 Surfynol CT-231 1.8 × 105 1.8 × 105 1.8 × 105 3.0 × 105 3.0 × 105 3.0 × 105 4.2 × 105 4.2 × 105 4.2 × 105 5.4 × 105 5.4 × 105 5.4 × 105 −1.5 −2.6 −3.1 −1.8 −2.7 −3.3 −2.8 −3.3 −2.0 −3.5 −2.2 −2.8 I. Fasaki et al. / Applied Catalysis A: General 411–412 (2012) 60–69 67 Fig. 7. 3D topographic AFM images of the surface of films prepared by doctor-blading on glass of (a) 1.0% (w/w) TiO2 (sample 10) and (b), 0.5% (w/w) TiO2 (sample 12) and (c) a higher-resolution 2D AFM image of 1.0% (w/w) TiO2 (sample 10). nanoparticles splitting. The porosity extension is responsible for an enhancement of the photocatalytic activity: more free volume is available for the pollutants to diffuse within the solid photocatalyst and get adsorbed. The increase of the photocatalytic rate under the Langmuir–Hinshelwood kinetic model for photocatalysis underlines the porosity change after ultrasonication (on the contrary, Eley-Ridal model does not require physical absorption and therefore structural changes would not lead to the increase of photocatalytic rate). The third parameter, the type of dispersant used for the titania nanoparticles dispersion and stabilization, is a less significant parameter in controlling the stability of the suspension and the 0,0 3,5 -0,1 3,0 -1 k 10 (min ) -0,3 2,5 -4 ln(A/Ao) -0,2 Suspension 10 doctor blade k=-3.5·10-4min-1 Suspension 10 inkjet prinƟng k=-3.2·10-4min-1 SGG Biocleank=-6.4·10-5min-1 -0,4 2,0 0.1% 0.5% 1.0% -0,5 1,5 -0,6 0 200 400 600 800 1000 1200 1400 1600 t (min) 1,5 2,0 2,5 3,0 3,5 4,0 4,5 5,0 5,5 5 E/V (10 J/L) Fig. 8. Photodecomposition kinetics of methyl orange for a film of suspension 10 (1% TiO2 , E/V = 5.4 × 105 J/L, Zetasperse 1200 dispersant) obtained with doctor blade and inkjet printing compared to commercial photocatalytic glass Saint Gobain Bioclean. The films were irradiated by a UV 71.7 ␮W cm−1 lamp and the assessment was realized by the photodegradation of methyl orange solution. Fig. 9. Rate constant versus energy per volume for the various TiO2 concentrations. Note that the dispersants were not the same for each suspension (see Table 5). The films were irradiated by a UV 71.7 ␮W cm−1 lamp and the assessment was realized by the photodegradation of methyl orange solution. 68 I. Fasaki et al. / Applied Catalysis A: General 411–412 (2012) 60–69 Fig. 10. Images of the evolving drop shape during jetting from the electromagnetic nozzle (300 mbar pressure, 200 ␮s opening time). The images have been cropped to remove the nozzle, and the background subtracted, for clarity. The delays after ejection have been selected to show key changes in the drop shape and are not at equal intervals: (a) 200 ␮s, (b) 800 ␮s, (c) 1250 ␮s, (d) 1300 ␮s, (e) 1500 ␮s, (f) 1700 ␮s, (g) 1800 ␮s, (h) 2000 ␮s, (i) 2100 ␮s, (j) 2150 ␮s, and (k) 2200 ␮s. photocatalytic activity of the coating. This is partly due to the dominant role of the salt and its optimised concentration, and partly due to the dispersant agent. For evaluating the critical parameter of ultrasonication, a film with the same experimental parameters of suspension 10, but without being subjected to ultrasounds was prepared and the degradation rate of methyl orange was experimentally calculated. The calculated rate constant was one order of magnitude less than the presented values of the twelve suspensions. Therefore, it is clearly evident, that the ultrasound treatment affects the physical properties of titania nanoparticles, enhancing their photocatalytic activity. 3.3. TiO2 thin films by inkjet printing After succeeding in producing stable suspensions and confirming the photocatalytic activity of TiO2 thin films produced by the doctor blade method, the next step was to confirm that these suspensions were suitable for inkjet printing and to demonstrate that coatings could be printed on glass and ceramic coatings. The viscosity of the suspensions was in the range 1–2 cP, in the range reported as suitable for both electromagnetic and piezoelectric printing technologies from a wide range of manufacturers. A representative suspension (suspension 10) was selected for jetting optimisation and the effect of pressure (200–500 mbar) and micro-valve opening time (200–500 ␮s) was investigated by drop visualisation. Stable jetting was achieved for pressures in the range 300–500 mbar, with 200 mbar being insufficient to accelerate the ink away from the nozzle. Image analysis confirmed that the ink velocity increased with increasing pressure and the ejected volume increased with both increasing pressure and opening time. To optimise the behaviour for printing thin coatings or high resolution patterns, a small drop volume is preferred, and it is important to avoid the formation of smaller satellite droplets. During the early stages of drop ejection, a continuous ink stream is formed below the nozzle, and at some critical length (or time after ejection) this ligament typically breaks up into small droplets, particularly for low viscosity inks ejected at high velocity. In order to obtain a single drop on the substrate, the maximum length of the ink stream must not exceed the distance of the nozzle from the substrate, and any droplets which do form must coalesce within this distance. With the ink and nozzle configuration tested here, this was achieved in a distance just over 3 mm for a pressure of 300 mbar and an opening time of 200 ␮s. Fig. 10 shows the drop shape at selected intervals after drop ejection for these printing settings, demonstrating the growing ink stream (a–c) detaching from the nozzle (d), breaking up into droplets (e and f) and recoalescing (g–k). With the settings used in Fig. 10, the drop volume was calculated as 10.3 nL (close to the smallest possible with this nozzle) and the velocity as 2.1 m s−1 using quantitative image analysis. Having demonstrated that inkjet printing was possible and selected suitable jetting parameters, test coatings of the suspension were successfully printed on both a glass microscope slide and a glazed ceramic tile. After the same heat treatment applied to doctor blade coatings, uniform films were produced (Fig. 11). The photocatalytic activity of the coatings on both substrates was measured, and the resulting rate constants were found to be very similar to those of the film produced by the doctor blade method. It has therefore been demonstrated that these suspensions can be deposited by inkjet printing to form photocatalytic coatings of similar effectiveness to those deposited by conventional methods, but the coating thickness and roughness will require further investigation to optimise transparency and photocatalytic performance. As demonstrated here, inkjet printing provides the flexibility to adjust the drop size and precisely control drop placement, giving considerably more control over these parameters than doctor blading without requiring reformulation of the suspension. Fig. 11. Photograph of the printed TiO2 thin film of suspension 10 on (a) a microscope glass slide and (b) a glazed ceramic tile. I. Fasaki et al. / Applied Catalysis A: General 411–412 (2012) 60–69 4. Conclusions Photocatalytic thin TiO2 films were obtained by the deposition of stable TiO2 suspensions in which water was the primary solvent and the other constituents were non-volatile substances. Ultrasonication succeeded in disrupting the nanoparticle agglomerates, resulting in a bimodal particle size distribution largely in the sub-micron range (D90 < 0.7 ␮m), with both the modal particle sizes and the volume fraction associated with the peak at larger particle size decreasing with increasing ultrasonication specific energy. The fact that the combination of the stabilizers (salt and dispersant) provide an electrosteric stabilization in combination to the obtained particle size distribution was proved both experimentally ( potential measurements) and practically as the suspensions remained stable for 18 months. The type of dispersant and the titania concentration, slightly affected the stability, because the chemical technology of these commercial dispersants is compatible with pigments in aqueous dispersions, and the titania weight concentration (0.1–1%) is modest. The stabilization additives proved that they do not affect the photocatalytic performance of the films as their concentration in the suspension is kept low. The crucial parameters that most strongly influence the photocatalytic activity of the films are the titania concentration and the ultrasonication specific energy. Large amounts of TiO2 mass load lead to rougher thin film surface, while the increase of the interparticle distance results in more porous coating and both contribute in the enhancement of the photocatalytic activity. It was shown that stable suspensions with a viscosity in the range suitable for drop-on-demand inkjet printing (electromagnetic and piezoelectric) can be produced by this method, and stable jetting without satellite drops has been demonstrated by drop visualisation using an electromagnetic printing device. Test coatings with adequate photocatalytic performance have been produced on both glass and glazed ceramic tiles. The proposed synthesis method is patent pending [49] and has already been scaled up for producing up to 1000 L per day of stable titania suspensions that can be used as inks for inkjet printing for industrial deposition on glass or ceramic tiles. Acknowledgements This research was carried out under Efficient EnvironmentallyFriendly Electro-Ceramics Coating Technology and Synthesis (EFECTS), a project funded by the European Union, FP7-NMP-2007SMALL-1 grant no 205854. References [1] V.A. Sakkas, I.M. Arabatzis, I.K. Konstantinou, A.D. Dimou, T.A. Albanis, P. Falaras, Appl. Catal. B: Environ. 49 (2004) 195–205. [2] A.T. Shawaqfeh, F.A. Al Momani, Sol. Energy 84 (2010) 1157–1165. [3] N. Miranda-Garcia, M.I. Maldonado, J.M. Coronado, S. Malato, Catal. Today 151 (2010) 107–113. 69 [4] J. Taranto, D. Frochot, P. Pichat, Sep. Purif. Technol. 67 (2009) 187–193. [5] C.J. Chung, C.C. Chiang, C.H. Chen, C.H. Hsiao, H.I. Lin, P.Y. Hsieh, J.L. He, Appl. Catal. B: Environ. 85 (2008) 103–108. [6] G. Katsaros, T. Stergiopoulos, I.M. Arabatzis, K.G. Papadokostaki, P. Falaras, J. Photochem. Photobiol. A 149 (2002) 191–198. [7] I. Fasaki, I. Hotovy, A. Rehakova, J. Hotovy, V. Rehacek, M. Kompitsas, F. RoubaniKalantzopoulou, Thin Solid Films 518 (2009) 1103–1108. [8] W. Yang, C.A. Wolden, Thin Solid Films 515 (2009) 1708–1713. [9] K. Murugan, T.N. Rao, A.S. Gandhi, B.S. Murty, Catal. Commun. 11 (2010) 518–521. [10] A. Kafizas, S. Kellici, J.A. Darr, I.P. Parkin, J. Photochem. Photobiol. A 204 (2009) 183–190. [11] N. Veronovski, P. Andreozzi, C. La Mesa, M. Sfiligoj-Smole, Surf Coat. Technol. 204 (2010) 1445–1451. [12] M.E. Simonsen, H. Jensen, Z. Li, E.G. Sogaard, J. Photochem. Photobiol. A 200 (2008) 192–200. [13] L. Grinis, S. Dor, A. Ofir, A. Zaban, J. Photochem. Photobiol. A 198 (2008) 52–59. [14] N. Berger-Keller, G. Bertrand, C. Filiatre, C. Meunier, C. Coddet, Surf Coat. Technol. 168 (2003) 281–290. [15] S. Nishimoto, A. Kubo, K. Nohara, X. Zhang, N. Taneichi, T. Okui, Z. Liu, K. Nakata, H. Sakai, T. Murakami, M. Abe, T. Komine, A. Fujishima, Appl. Surf. Sci. 255 (2009) 6221–6225. [16] P. Blazdell, J. Mater. Process. Technol. 137 (2003) 49–54. [17] M. Arin, P. Lommens, N. Avci, S.C. Hopkins, K. De Buysser, I.M. Arabatzis, I. Fasaki, D. Poelman, I. Van Driessche, J. Eur. Ceram. Soc. 31 (2011) 1067–1074. [18] N. Mandzy, E. Grulke, T. Druffel, Powder Technol. 160 (2005) 121–126. [19] J.L. Deiss, P. Anizan, S. El Hadigui, C. Wecker, Colloid Surf. A 106 (1996) 59–62. [20] I.N. Martyanov, E.N. Savinov, K.J. Klabunde, J. Colloid Interface Sci. 267 (2003) 111–116. [21] S. Fazio, J. Guzman, M.T. Colomer, A. Salomoni, R. Moreno, J. Eur. Ceram. Soc. 28 (2008) 2171–2176. [22] J. Jiang, G. Oberdorster, P. Biswas, J. Nanopart. Res. 11 (2009) 77–89. [23] N. Veronovski, P. Andreozzi, C. La Mesa, M. Sfiligoj-Smole, V. Ribitsch, Colloid Polym. Sci. 288 (2010) 387–394. [24] A. Fernandez-Nieves, F.J. de las Nieves, Colloid Surf. A 148 (1999) 231–243. [25] N.H. Tkachenko, Z.M. Yaremko, C. Bellmann, M.M. Soltys, Colloid Surf. A 279 (2006) 149–158. [26] P. Alphonse, R. Bleta, R. Soules, J. Colloid Interface Sci. 337 (2009) 81–87. [27] N. Van de Velde, M. Arin, P. Lommens, D. Poelman, I. Van Driessche, Thin Solid Films 519 (2011) 3475–3479. [28] Z.M. Yaremko, N.H. Tkachenko, C. Bellmann, A. Pich, J. Colloid Interface Sci. 296 (2006) 565–571. [29] C.S. Hirtzel, R. Rajagopalan, Chem Eng. Commun. 33 (1985) 301–324. [30] J. Widegren, L. Bergstrom, J. Eur. Ceram. Soc. 20 (2000) 659–665. [31] Q. Ran, S. Wu, J. Shen, Polym. Plast. Technol. 46 (2007) 1117–1120. [32] B. Neppolian, Q. Wang, H. Jung, H. Choi, Ultrason. Sonochem. 15 (2008) 649–658. [33] P. Calvert, Chem. Mater. 13 (2001) 3299–3305. [34] http://www.hielscher.com/ultrasonics/i1000 p.htm. [35] T. Hielscher, Ultrasonic Production of Nano-size Dispersions and Emulsions, ENS Paris, France, 2005, February. [36] K. Kusters, S. Pratsinis, S. Thoma, Powder Technol. 80 (1994) 253–263. [37] B. Park, D. Smith, S. Thoma, Powder Technol. 76 (1993) 125–133. [38] J.C. Yu, Z.T. Jiang, H.Y. Liu, J. Yu, Colloid J. Interface Sci. 262 (2003) 97–100. [39] P. Velikovska, P. Mikulasek, Sep. Purif. Technol. 58 (2007) 295–298. [40] I. Pochard, J. Boisvert, J. Persello, A. Foissy, Polym. Int. 52 (2003) 619–624. [41] J. Widegren, L. Bergstrom, J. Am. Ceram. Soc. 85 (2002) 523–528. [42] J.F. Porter, Y.-G. Li, C.K. Chan, J. Mater. Sci. 34 (1999) 1523–1531. [43] H. Zollinger, Color Chemistry: Synthesis, Properties and Applications of Organic Dyes and Pigments, VCH Publishers, New York, 1987, p. 12. [44] K.-T. Chung, C.E. Cerniglia, Mutat. Res. 277 (1992) 201–220. [45] F. Kiriakidou, D.I. Kondarides, X.E. Verykios, Catal. Today 54 (1999) 119–130. [46] F. Zhang, J. Zhao, T. Shen, H. Hidaka, E. Pelizzetti, N. Serpone, Appl. Catal. B: Environ. 15 (1998) 147–156. [47] F. Chen, Y. Xie, J. Zhao, G. Lu, Chemosphere 44 (2001) 1159–1168. [48] H. Al-Ekabi, N. Serpone, J. Phys. Chem. 92 (1988) 5726–5731. [49] Patent application number 2011-00389, Applicant: NanoPhos S.A., Inventors: I. Arabatzis, I. Fasaki, Stable titanium dioxide suspension for photocatalytic, self-cleaning, self-sterilizing and super-hydrophilic applications.