Morphology dependent adsorption of methylene blue on trititanate nanoplates and nanotubes prepared by the hydrothermal treatment of TiO₂

Dye compounds are common and difficult to remove pollutants from several industries, including textiles and printing. The removal of dye compounds from wastewater has therefore received considerable attention in recent years and requires environmentally friendly materials and potentially easy separation or regeneration techniques (Ramakrishna & Viraraghavan 1997; Crini 2006). Photocatalytic and biological degradations of the dyestuffs can result in the production of toxic products due to the complex nature of the aromatic dye molecules, therefore adsorption is a promising technique for the removal of dyes from water (Zhao & Liu 2008). A wide variety of adsorbents have been employed for several different applications, including activated carbon and mesoporous molecular sieves, however, the operating and regeneration costs are high, leading to a search for low cost, high capacity absorbents (Widjaja et al. 2004; Namane et al. 2005; Parida et al. 2006; Wang et al. 2006). Nanostructured materials are of great interest due to their exceptional electronic and mechanical properties, as well as exhibiting high surface areas: a useful property for adsorption. Research pertaining to inorganic nanotubes has been extensive (Tenne et al. 1992; Feldman et al. 1995; Nakamuara & Matsui 1995; Hoyer 1996; Pu et al. 2001; Goldberger et al. 20013), with TiO₂ receiving a great deal of attention.

Titania nanotubes were first prepared by the hydrothermal treatment of TiO₂ with 10 M NaOH (Kasuga et al. 1998). The nanotubular material was produced by treating crystalline TiO₂ with 10 molar NaOH. It was concluded that washing the alkali treated specimen with water followed by further reaction with HCl are vital steps in the formation of nanotubes (Kasuga et al. 1999). Work has later shown (Chen et al. 2002) that the nanotubes could be prepared via a single hydrothermal alkali treatment of crystalline TiO₂.

We have shown that the reaction temperature and phase composition of the starting material are important factors in determining the morphology of the product (Dawson et al. 2010). The anatase component of the starting material is easily converted to trititanate nanotubes at 140 °C, where, as the rutile, remains unreacted. After the temperature is increased to 170 °C, the morphology is exclusively trititanate nanoplates, with the tubular structure destroyed. The high adsorption capacity of trititanate nanotubes for basic dyes has been reported (Lee et al. 2007; Hsieh et al. 2008), with the dye is adsorbed on the nanotube via an ion exchange mechanism. Modification of the nanotube surface by dopamine imparts photocatalytic properties on the composite material (Dawson et al. 2012; Liu et al. 2016), however little is known about the adsorption process and interaction.

In this study, we have compared the adsorption properties of trititanate tubes and plates prepared by the hydrothermal treatment of different phase TiO₂. It was found that nanotubes with high surface area and pore volume exhibited exothermic monolayer adsorption of methylene blue (MB) governed by external mass transport. Nanoplates with low surface area and pore volume yielded a higher adsorption capacity through endothermic multilayer adsorption governed by pore diffusion. We believe these nanostructured materials have interesting potential applications in wastewater treatment and absorption.

The two different nanostructures were prepared following our previously reported method (Dawson et al. 2010) at 140 °C for 72 h for anatase (TiNT) and 170 °C for seven days for rutile (TiNP).

The products morphologies were investigated using field emission scanning electron microscopy (FESEM) (FEI Quanta 400 FEG) and transmission electron microscopy (TEM) (FEI Tecnai F20 operating at 200 kV). Powder X-ray diffraction (XRD) was performed with CuKα radiation on a X'Pert Pro MPD operation at 40 kV and 30 mA. Zeta potentials were measured using a Colloidal Dynamics Zeta Probe. Specific surface areas, pore size and pore volumes were determined by nitrogen adsorption at −196 °C on a Micromeritics ASAP 2020 apparatus. The surface area was determined by the Brunauer–Emmett–Teller (BET) method.

The equilibrium adsorption isotherms were analysed using the Langmuir and BET models.

For adsorption from the liquid phase, the partial pressure term can be replaced by the saturation concentration (Parker 1995). In this study, we have used the initial concentration of MB, C_i, in accordance with (El-Halwany 2010). Thus, plotting M.C/(C_s-C)x versus C/C_s will give rise to a straight line from whose slope and intercept the BET monolayer can be calculated. The value of the correlation coefficient, r², determines which model is the best fit for the data.

The specific surface area and pore volumes of the two samples were measured using the BET and Barrett–Joyner–Halenda (BJH) methods, respectively. The surface area of TiNT was found to be 250 m²/g. The TiNP sample exhibited a surface area of 40 m²/g, much lower than the tube sample. In the TiNT sample, the pore size peak at 3.4 nm is attributed to the internal tube cavity, which is in agreement with the TEM observations (Figure 1). The pore distribution is wide, with the larger pores between 10 and 100 nm being attributed to voids formed between nanotube aggregates. The distribution for the TiNP sample is also large, with small pore peaks at 2.1 and 3.7 nm. From the TEM shown in Figure 1, we can see that there are no pores visible in the plate structures themselves, so it can be concluded that the mesopores come from the aggregation and stacking of plates, with the larger pores created in the voids between the particles. The pore volume for TiNT was high, 0.9 cm³/g, but TiNP had a low pore volume of 0.19 cm³/g.

The stability of the samples in aqueous solution was measured by zeta potential, and are shown in Table 1. TiNT had a potential of −12 mV at a pH of 6.7, and TiNP exhibited a value of −23 mV at pH 6.0. TiNP exhibit a higher –ve zeta potential, and are therefore more stable in suspension.

The adsorption characteristics and kinetics of the two morphologies were compared and are detailed below.

As can be seen, the Langmuir model is a good fit for the TiNT data, with all r² values >0.99. While the Langmuir model reflects monolayer adsorption, a good fit to the BET isotherm indicates multilayer adsorption. The TiNP data are a good fit for the BET model, with a calculated monolayer capacity of 15 mg/g, which is in keeping with the theoretically calculated value from the SSA data. The calculated adsorption capacity maximums for the nanotubes using the Langmuir model were 122.65, 106.84 and 101.42 mg/g at 25, 40 and 60 °C, respectively.

The experimental adsorption capacities measured in 100 mL of 10 mg/L MB solution at 25 °C were found to be 120 mg/g for TiNT and 130 mg/g for TiNP samples. This is in good agreement with the Langmuir calculated values for the TiNT sample. The Langmuir model gives lower capacities than observed, however, this model is mostly suitable for monolayer adsorption.

The value of the zeta potential is a measure of the charge on the electric double layer, but can be taken as the surface charge. Therefore, TiNP have a higher –ve surface charge, and attract the positively charged MB ions more strongly. This stronger interaction, along with the flat plate structure compared to the curved tube structure results in the formation of multilayers of MB, and only monolayers in the TiNT adsorption. The lower surface charge may be a result of the scroll structure ‘hiding’ some of the surface within the tube, where it is balanced by H⁺ and Na⁺ ions and does not contribute to the surface negativity in the zeta potential measurement. There may also be a contribution from the lowering of strain through layer coupling in the scroll structure. However, this will require further investigation.

As can be seen from Figure 2(a), the adsorbed amount for TiNT decreases with increasing temperature, that is to say, the reaction is an exothermic process. In the case of TiNP, the adsorbed amount increases with increasing temperature, and therefore the adsorption reaction is endothermic in nature. Physical adsorption is generally an exothermic process, however endothermic adsorptions of MB have been reported on activated carbons and clays (Lin & Teng 2002; Hong et al. 2009). Because of this interesting observation, the adsorption was investigated further.

The heats of adsorption, ΔH, were calculated from the van 't Hoff equation (Equation (5)) and are −39.83 kJmol⁻¹ for TiNT and +25.98 kJmol⁻¹ for TiNP. These values indicate a physical adsorption, which is generally of the range <40 kJmol⁻¹ (Lin & Teng 2002). The Gibbs free energy was calculated from Equation (10), and was –ve for both morphologies, indicating a spontaneous reaction in both cases. The values of ΔG at different temperatures are listed in Table 3. The change in entropy for both systems was also calculated as −86 and 130 JK⁻¹ mol⁻¹ for TiNT and TiNP, respectively. The different signs of the entropy for both reactions is interesting. The numerical values are both relatively small, and around the order of values seen for phase changes. Both reactions are spontaneous, as shown by the negative values of the Gibbs energy. A decrease in entropy for an adsorption process would be expected, as is seen for the TiNT, where the negative entropy change is accounted for by the increase in order of the system through adsorption, as the condensed state is more ordered than in bulk solution. The exothermic nature of the adsorption overcomes the increase in order to give a negative value of Gibbs energy, and a spontaneous reaction. The positive value of ΔS for TiNP overcomes the exothermic value of ΔH, resulting in a negative Gibbs energy and a spontaneous reaction. This entropy change reflects the affinity of the plates for the dye and suggests some structural changes in the system. The adsorption of MB will result in a negative change in entropy, however each MB will replace several water molecules, resulting in an overall positive entropy change for the system (Hong et al. 2009).

Assuming that MB molecules can occupy the entire surface area, we can calculate the maximum amount of MB adsorbed in a monolayer (Hong et al. 2009). Rather than lying flat on the surface, MB has been shown to adsorb onto clay surfaces at a tilted angle (Krishna et al. 2004). The dimensions of an MB molecule are given as 1.65 nm by 0.77 nm wide and 0.4 nm thick. Therefore, the area that one molecule of MB takes up on a surface can therefore vary from 1.27 nm² for flat lying cations to 0.66 nm² for molecules perpendicular to the surface. These values correspond to a capacity of 102–240 mg/g for TiNT and 16.4–38.6 mg/g for TiNP. From these figures, we can conclude that MB is adsorbed on the nanotube surface in a monolayer at a low tilting angle, whereas the adsorption mode for the nanoplates appears to be multilayer adsorption, with not enough surface layer available to provide the high capacity observed in a monolayer configuration.

The adsorption capacities of the TiNT and TiNP were compared with similar complex nanomaterials in the literature, shown in Table 4. Both materials showed a favourable adsorption capacity.

After the capacity of adsorption was reached, the particles aggregated and deposited on the base of the beaker. This is an interesting property for applications of water treatment where separation of the adsorbant is an issue. This occurs because the adsorption is an electrostatic interaction, and repulsion between the particles helps in stabilising the suspension. When the charge is neutralized by MB adsorption, the particles aggregate and deposit on the base of the vessel. The adsorption capacities of macrosized bulk TiO₂ and H₂Ti₃O₇ was also measured and found to be negligible. During the measurements of adsorption, the sample was suspended in the MB solution by sonication. After adsorption, the particles aggregated together to form m sized clumps and deposited on the bottom of the vessel. As a comparison, TiNT dispersed in water aggregate and deposit over hours, but TiNT after MB adsorption aggregate and deposit within 5 min.

Two distinct trititanate nanomorphologies were prepared by the hydrothermal treatment of TiO₂ with strong base. The two morphologies, plates and tubes, had very different structural and adsorption properties. Nanotubes with high surface area and pore volume exhibited exothermic monolayer adsorption of MB. Nanoplates with low surface area and pore volume nanoplates yielded a higher adsorption capacity through endothermic multilayer adsorption governed by pore diffusion. The stronger electrostatic interaction between the TiNP surface and the MB molecules results in multilayer adsorption, and the different modes of adsorption of the two morphologies result in the exothermic and endothermic character for TiNT and TiNP, respectively.

The morphology dependent adsorption characteristics of two materials prepared by a facile and low cost production method have interesting potential applications in wastewater treatment. It is also interesting and informative to observe such different adsorption properties within the same material phase of different nanomorphologies.

We would like to thank the Fellowship for Young International Scientists from Chinese Academy of Sciences, National Science Foundation of China (50950110349 and 10704051), and Xi'an Jiaotong Liverpool University Research Development fund.