Theoretical and experimental studies on the adsorption of intercalant dyes on anatase and rutile

Colloids and Surfaces A: Physicochem. Eng. Aspects 386 (2011) 71–78

Contents lists available at ScienceDirect

Colloids and Surfaces A: Physicochemical and Engineering Aspects journal homepage: www.elsevier.com/locate/colsurfa

Theoretical and experimental studies on the adsorption of intercalant dyes on anatase and rutile C.E. Zubieta a,b , E. German b,c , I. López Corral b,c , A. Juan b,c , P.C. Schulz a,∗ , P.V. Messina a,b , M. Dennehy a a b c

Departamento de Química e Instituto de Química del Sur (INQUISUR, CONICET) – Universidad Nacional del Sur-Avda. Alem 1253, 8000 Bahía Blanca, Argentina CONICET, Argentina Departamento e Instituto de Física-Universidad Nacional del Sur-Av. Alem 1253, Bahía Blanca, Argentina

a r t i c l e

i n f o

Article history: Received 25 March 2011 Received in revised form 26 June 2011 Accepted 30 June 2011 Available online 7 July 2011 Keywords: Adsorption Anatase Rutile Computational simulation Intercalant dyes Acriflavine Acridine orange

a b s t r a c t The adsorption and photodegradation of acridine orange (AO) and acriflavine (AF) on two titania crystalline phases, anatase and rutile were experimentally studied and compared with results of a computational simulation. The adsorption capacity of rutile was higher than that of anatase, while the reverse is observed for the photodegradation of both dyes. The adsorption of AO on both adsorbents was higher than that of AF, which was related to the higher basic character of the dimethylammonium groups of AO in comparison with the amine groups of AF. Taking into account that the computational simulation was made in vacuum and the experimental results in an aqueous medium, the results of both approaches are general in agreement with each other. The heterogeneity of the adsorbent surface of rutile, the possible cause of the higher adsorbent capacity of rutile and higher catalytic capacity of anatase when compared are explained by the computational model. A factor that the computational simulation did not take into account was the macroporosity of rutile that could increase the apparent adsorption capacity. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Textile dyes are extremely soluble in water and highly persistent in the environment, being scarcely biodegradable. In particular, acridine dyes as acriflavine (AF) and acridine orange (AO) are heterocyclic compounds containing nitrogen atoms (see Scheme 1). Acridine dyes are generally yellow or orange. They are widely employed in printing, dyeing, leather, and other industries [1]; and are also employed in biological dyeing. Toxicological research indicates that aminoacridines have mutagenic effects because they join the DNA molecules and intercalate inside the double chain causing local interruptions of the nitrogen bases unions [2]. These interruptions cause errors in the following replications. Acridine dyes are difficult to remove from water by conventional methods such as coagulation, chemical oxidation, precipitation, flocculation and biodegradation. It must be mentioned that dyestuffs are highly structured organic compounds which are difficult to break down biologically [3–8]. The release of such polluted waters to the ambient has a dramatic effect on the environment, with perturbation of aquatic life and visual contamination [9]. Advanced oxidation is one of the most promising technologies for the removal of dye-contaminated wastewater due to its high

efficiency. It is based mainly on the oxidative reactivity of hydroxyl radicals generated by various methods which could completely degrade the dyes to harmless compounds by photocatalysis under normal temperature and air pressure. Then it is predicted that these techniques will soon become one of the most effective means of depollute dye wastewaters [10]. The TiO2 -mediated photocatalysis process has been successfully used to degrade pollutants in recent years [11–18]. Some advantages of using TiO2 as a photocatalyst are its nontoxicity, photochemical stability and low cost [19,20]. Then, a better comprehension of the factors affecting adsorption and degradation of dyes with titania is desirable. In this work we studied both experimentally and theoretically the adsorption of acridine dyes on two different crystalline forms of titania (TiO2 ): anatase and rutile. The photocatalytic activity of TiO2 depends on its crystalline structure, its specific area and the structure of the pores. Many studies have pointed out that the photocatalytic activity of anatase is superior to that of rutile for water and air purification, disinfection and elimination of dangerous materials [20,21]. However, it is not clear why the above mentioned factors are responsible for the advantage of anatase against rutile. 2. Experimental

∗ Corresponding author. Tel.: +54 291 4548305; fax: +54 291 4595159. E-mail address: [email protected] (P.C. Schulz). 0927-7757/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.colsurfa.2011.06.030

Materials: acriflavine (AF) and acridine orange (AO) were from Sigma–Aldrich and employed as received. Anatase (Aldrich, 99%)

72

C.E. Zubieta et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 386 (2011) 71–78

H2 N

N

NH2

45 ◦ C in dark. Samples were analyzed at different times. At the end of each adsorption period the supernatant was centrifuged for 5 min at 3500 min−1 . The supernatant AO and AF concentration before and after adsorption was determinated using a Spectronic 20 UV–Vis spectrophotometer at 449 and 450 nm, respectively. A KH2 PO4 –K2 HPO4 buffer was employed to maintain the pH = 8.00. 2.2. Adsorption data analysis

Acriflavine (AF)

The equilibrium adsorption capacity qe (mmol/g) of AO and AF was calculated with the following equation: qe =

N

N

N

Acridine orange (AO) Scheme 1. Structure of the acridine dyes.

was used as received. Sodium dioctyl sulfosuccinate (Aerosol OT; AOT) 99% was from Sigma. All solutions were made with tri-distilled water. To obtain mesoporous rutile, a reverse microemulsion solution was prepared by mixing 1.1276 g of AOT and 1.5 g of water, and then the sample was left for 3 h to allow the surfactant hydration. In this case the water/surfactant ratio (R) was R = 30. Then 80 mL of n-hexane (Carlo-Erba, p.a) was added and the system was sonicated in an ultrasonic bath to produce the microemulsion. Next 1.4 mL of TiCl4 (Carlo Erba, 99%, ı = 1.722 g/cm3 ) was added and left by three days to react following the reaction: 2H2 O (l) + TiCl4

(l) → TiO2

(s) + 4HCl (g).

The excess of HCl and n-hexane was eliminated by evaporation under vacuum, and the resulting material was left for 3 h in hydrothermal treatment at 70 ◦ C in autoclave. A white powder formed by titania nanoparticles surrounded by AOT was obtained. Subsequently the material was calcined during 7 h at 540 ◦ C under air flux. The material was characterized by X-ray diffraction (XRD), nitrogen adsorption, transmission electron microphotography (TEM) and scanning electron microphotography (SEM). For aqueous dye solutions double distilled water was used. Solution pH was kept constant and equal to 8.0 by employing a phosphate buffer. Powder ray diffraction (XRD) data were collected via a Philips ˚ and PW 1710 diffractometer with Cu K␣ radiation ( = 1.5418 A) graphite monochromator operated at 45 kV; 30 mA and 25 ◦ C. The nitrogen adsorption isotherms at 77.6 K were measured with a Micrometrics Model Accelerated Surface Area and Porosimetry System (ASAP) 2020 instrument. Each sample was degassed at 373 K for 720 min at pressure of 10−4 Pa. Scanning electron microscopy was performed using a JEOL 35 CF, Tokyo, Japan. 2.1. Dye adsorption experiments Dye adsorption tests were performed by immersion of 50 mg of each adsorbent in 5 mL of dye solution in sealed flasks. The range of dye concentration was 0.020–0.061 mM for acriflavine and 0.041–0.098 mM for acridine orange. These concentrations were selected to ensure low error in the measurement by the UV–Vis technique. Then the flasks were shaken at 25, 35, and

(Co − Ce )V m

(1)

where Co is the initial concentration (mmol/dm3 ), Ce the residual concentration at equilibrium (mmol/dm3 ), V the solution volume (dm3 ), and m is the adsorbent mass (g). Two isotherm equations are used to analyze the data. One of them is the Langmuir equation, based on the hypothesis that the maximum adsorption corresponds to a saturated monolayer of adsorbate molecules on the adsorbent surface, with constant energy of interaction: qe =

qmon KL Ce 1 + KL Ce

(2)

where KL is the Langmuir constant related to the energy of adsorption (KL = Ae−H/RT , where A is a constant, R the gas constant, T the absolute temperature and H the energy of adsorption, which must be constant in the Langmuir model); and qmon (mmol/g of adsorbent) is the maximum amount of adsorption corresponding to a complete coverage on the surface by a monolayer of dye. It must be emphasized that the theoretical interpretation of Langmuir equation parameters must be done cautiously, since the fitting of experimental data to this isotherm is not a sensitive test of the model [22]. In particular, in many cases the Langmuir equation gives adequate results in systems in which surface heterogeneity is known to be present. The other model here employed is the Freundlich isotherm, which can be used for non-ideal adsorption that involves heterogeneous adsorbent surfaces [23], and is expressed by the equation: 1/n

qe = KF Ce

(3)

where KF is related with the adsorption capacity and 1/n is related to the adsorption intensity. In general, as KF increases the adsorption capacity of an adsorbent for a given adsorbate augments. KF can be related to the surface energy of adsorption: KF = RTAe−H/RT , where A is a constant. The magnitude of the exponent 1/n gives an indication of the favorability of adsorption. The isosteric adsorption enthalpies (Hads ) were computed from the adsorption dependence on temperature. 2.3. Photocatalytic activity of titania The photocatalytic activity of anatase and rutile was investigated using an aqueous solution of AF 0.061 mM and other of AO 0.096 mM. Ten millilitres of diverse dilutions of these solutions were put in vials containing 0.2 g of adsorbent. The vials were shaken in a thermostatized bath at 25 ◦ C, exposed to UV light with a lamp DESAGA UV 131000 ( = 366 nm). The light intensity was estimated as I = 2.7 × 10−6 mol photon s−1 from data given by the supplier. A set of dilutions without catalyst was also exposed to the UV light as control. Finalized the period of irradiation, the samples were centrifuged during 1 min at 3000 min−1 . The supernatant concentration of dyes was determined with a Spectronic 20 UV–Vis spectrophotometer at 449 (AO) and 450 nm (AF).

C.E. Zubieta et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 386 (2011) 71–78

73

Fig. 1. The surfaces studied in anatase and rutile.

2.4. Theoretical studies The computational simulation was based on the method ASED of electronic delocalization and atomic superposition with the extended Hückel theory [24]. The systems properties were computed from approximate wave functions through the inclusion of experimental data. The adsorption energy Eads was computed with the following equation and minimized: Eads = Edye/titania − Edye − Etitania

(4)

where Edye is the energy of the isolated dye, Etitania the energy of the isolated TiO2 surface and Edye/titania the energy of the dye adsorbed on the titania surface. The atomic coordinates of anatase and rutile were obtained from literature [25,26], and finite cluster models were employed. The TiO2 surfaces studied were the (0 0 1) plane of anatase and the planes (0 0 1) and (1 0 0) of rutile, which are shown in Fig. 1. These planes were selected because it is expected that they will be predominantly exposed by cleavage [27]. Different adsorption sites were studied for each surface. Two orientations of the molecular plane of dyes with respect to the titania surface: horizontal or parallel, and vertical or perpendicular to the TiO2 surface (see Fig. 2). The equilibrium height z and the rotation angle (˛) around the (0 0 1) plane direction were also studied (see Fig. 2).

The adsorption sites chosen for the studied dyes on the (0 0 1) plane of anatase may be seen in Fig. 3. The sites studied on rutile surface are shown in Fig. 4. 3. Results 3.1. Characterization of the titania materials The synthesized titania XRD diffraction plot (not shown) displayed only two narrow peaks at 2 = 27◦ and 35.9◦ , indicating that the material is composed of pure rutile [28] (see Fig. 1 in the Supplementary Material (SM)). The specific surface area was determined by the Brunauer–Emmet–Teller (BET) isotherm. The adsorption– desorption isotherm was the typical Type II (see Fig. 5), which is related to a monolayer–multilayer adsorption that is generally associated non-porous or mesoporous materials [29]. The t plots (Fig. 6) show the behavior associated with materials having slit-shaped pores [30,31]. The analysis of the nitrogen adsorption isotherm gave the following results: BET surface area ABET = 3.83 m2 g−1 , mean pore radius 11.68 nm, mean pore volume 0.0112 cm3 g−1 . The SEM microphotographs (Fig. 7) showed that the material seems to be a sponge with large pores. The higher magnification obtained by TEM showed some lamellae forming slit-shaped micropores (not shown).

Fig. 2. The orientations of the molecular plane of dyes with respect to the titania surface.

74

C.E. Zubieta et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 386 (2011) 71–78

25

V (cm 3/g)

20

15

anatase rutile

10

5

0

0

5

10

t (A)

15

20

25

Fig. 5. t-Plots of anatase and rutile. Fig. 3. The adsorption sites for the studied dyes on the (0 0 1) plane of anatase.

Anatase was also studied by the same methods. Figs. 5 and 6 also show the adsorption isotherm and t-plot of anatase, which are compatible with slit-shaped pores too. This microstructure was confirmed by TEM (not shown). SEM photomicrograph (Fig. 7) shows that the material is formed by an agglomeration of nearly spherical particles having an average diameter of 100 nm. From these plots the ABET = 9.91 m2 g−1 , the mean pore radius is 10.07 nm

Fig. 4. The sites studied on the rutile surface.

and the mean pore volume is 0.0250 cm3 g−1 . Following the definition of IUPAC, both materials are mesoporous (pore radius between 1 and 25 nm). 3.2. Adsorption from aqueous solution Fig. 8 shows as examples the adsorption isotherms of AO and AF on anatase and rutile at T = 45 ◦ C.

Fig. 6. SEM photomicrograph of anatase (above) and the synthesized titania (below).

C.E. Zubieta et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 386 (2011) 71–78

75

Table 1 Adsorption parameters of the tested dyes on rutile and anatase. AO on anatase TiO2 Anatase ◦

25 C 35 ◦ C 45 ◦ C

Langmuir

Freundlich 2

qmon (mmol/g)

KL (L/mmol)

r

n

KF

r2

0.0179 0.0126 0.0136

3.41 8.16 6.45

0.8902 0.9888 0.9897

0.783 1.51 1.42

0.0981 0.0263 0.0273

0.9841 0.9823 0.9853

AO on rutile TiO2

Langmuir

Freundlich 2

Rutile

qmon (mmol/g)

KL (L/mmol)

r

n

KF

r2

25 ◦ C 35 ◦ C 45 ◦ C

0.127 1.04 0.0570

0.569 0.0636 1.196

0.9920 0.9911 0.9958

1.05 0.943 1.12

0.0614 0.0812 0.0474

0.9938 0.9968 0.9936

AF on anatase TiO2

Langmuir

Anatase

qmon (mmol/g)

KL (L/mmol)

r2

Freundlich n

KF

r2

25 ◦ C 35 ◦ C 45 ◦ C

0.0230 0.420 0.0228

3.76 0.191 3.23

0.9931 0.9988 0.9994

1.10 1.01 0.901

0.057 0.077 0.12

0.9949 0.9983 0.9999

AF on rutile

Fig. 7. Adsorption isotherms of acridine orange and acriflavine on rutile (above) and on anatase (below) at T = 45 ◦ C.

The parameters obtained from the fitting of adsorption data with the isotherms of Langmuir and Freundlich are presented in Table 1. 3.3. Photocatalytic degradation results Fig. 9 shows the photodegradation of AO on anatase and rutile. Without a catalyst there was no degradation. The maximum degra-

TiO2

Langmuir

Rutile

qmon (mmol/g)

KL (L/mmol)

r2

n

KF

r2

0.319 0.159 0.0337

0.196 0.431 2.04

0.9958 0.9931 0.9984

1.01 1.02 1.05

0.0598 0.0643 0.0538

0.9928 0.9889 0.9868

◦

25 C 35 ◦ C 45 ◦ C

Freundlich

dation reached was 97%, in 6.75 h on anatase, and 67% in 8 h on rutile. Fig. 9 shows the results for AF. In this case the irradiation without catalyst produced a degradation of 35%. With anatase the degradation was 93% in about 7 h, while with rutile it was 82% in 6.5 h. 3.4. Molecular simulation results The relative adsorption energies (with respect to the maximum adsorption energy) computed for AO adsorbed on the (0 0 1) plane of anatase are shown as an example in Fig. 10. It is evident that the preferential adsorption of AO on the (0 0 1) plane of anatase is in

Fig. 8. Degradation of AO (above) and AF (below) on anatase and rutile.

Fig. 9. The relative adsorption energies (with respect to the maximum adsorption energy) computed for AO adsorbed on the (0 0 1) plane of anatase. Numbers indicate the adsorption sites, H: horizontal, V: vertical orientation of dye molecule.

76

C.E. Zubieta et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 386 (2011) 71–78

˚ Fig. 10. The preferential adsorption of AO on the (0 0 1) plane of anatase in site 3 with horizontal orientation, with z = 1.6 A.

˚ Fig. 10 shows the site 3 with horizontal orientation, with z = 1.6 A. as an example the representation of this orientation. Fig. 2 in the supplementary Material (SM) shows the energetic diagram of AO adsorption on the (0 0 1) plane of rutile, showing that the preferential adsorption occurs on site 2 with vertical orientation, z = 1.4 A˚ and ˛ = 10◦ . Fig. 3 in the SM shows the representation of the adsorbed molecule. Accordingly Fig. 4 in the SM, the adsorption of AF on the (0 0 1) plane of rutile is horizontal on sites 1 or 2, with z ≈ 1.9 A˚ and ˛ = 21◦ (site 2). The representation is shown in Fig. 5 in SM. Fig. 6 in the SM shows the energy curves for the adsorption of AO on the (1 1 0) plane of rutile, where it is seen that the preferential adsorption is in site 2 in horizontal orientation with z ≈ 1 A˚ and ˛ = 4◦ . However, when AO orientation is vertical the preferential adsorption is in site 4 (Fig. 7 in the SM). However, the absolute adsorption energies indicate that AO adsorbs preferentially on site 2 than site 4, whose representation is shown in Fig. 8 in the SM. The study of the adsorption of AF on the (0 0 1) plane of anatase gave the results shown in Fig. 9 in the SM. It is seen that AF adsorbs preferentially in sites 1 or 2 and horizontal orientation, ˚ ˛ = −8◦ (site 3). Fig. 10 in the SM shows the orientation with z = 1.5 A, of the molecule on the adsorbent surface. When the adsorption of AF is studied on both crystalline planes of rutile, it is seen that on the (0 0 1) plane of rutile the dye adsorbs ˚ ˛ = 0◦ (Fig. 4 in the SM), while horizontal on site 1 with z ≈ 0 A, when oriented vertically the preferential adsorption is in site 4. Fig. 11 in the SM shows the orientation of the adsorbed molecules on the crystalline plane. Fig. 12 in the SM shows the adsorption energy of AF on the (1 1 0) plane of rutile with horizontal orientation, and Fig. 13 in the SM that of the AF on the same plane and in vertical orientation. Comparing the absolute Eads values, it may be concluded that the adsorption on site 1 is energetically favored. This orientation is shown in Fig. 14 in the SM.

Table 2 Values obtained for the more favorable adsorption situations for both dyes on both titania phases, and the experimental adsorption enthalpies. The sign of Eads in kJ/mol was changed to facilitate the comparison with Hads . Dye

AF AF AF AO AO AO

Adsorbent/plane

Anatase/0 0 1 Rutile/0 0 1 Rutile/1 1 0 Anatase/0 0 1 Rutile/0 0 1 Rutile/1 1 0

Dye orientation

h h h h v h

Site

3 2 1 3 2 2

Hads

Eads eV

kJ/mol

kJ/mol

140 119 218 211 141 284

−13,500 −11,400 −21,000 −20,000 −13,600 −27,400

−112 −101 −101 −175 −5470 −5470

Table 2 collects the values obtained for the more favorable adsorption situations for both dyes on both titania phases, together with the experimental adsorption enthalpies. It may be concluded that the dye that shows the maximum interaction capacity with all the tested surfaces is AO. The more active surface of rutile is the (1 1 0) plane, which shows the largest amount of exposed oxygen atoms. 4. Discussion Comparing the qmon values obtained from Langmuir equation the monolayer is more compact on rutile than for anatase for both dyes. This is in agreement with the higher Eads values obtained in both cases by computational modeling. It also may be noted that the amount of adsorbed AF is larger than that of AO on both adsorbents. This is not observed in the computational results, where the Eads values on both titania structures are lower for AF than for AO. However, it must be taken into account that Eads was computed under vacuum, whereas the experimental adsorption experiments were carried out in an aqueous medium. Then, some discrepancies are expectable. The value of 1/n for AO on anatase (1.28) is significantly higher than that on rutile (0.952) that indicates that the interaction on anatase is more favorable. The difference is not significant for AF: 0.923 on anatase, 0.989 on rutile (all values at 25 ◦ C). Taking into account the specific area of both adsorbents (9.92 m2 g−1 for anatase, 3.83 m2 g−1 for rutile), the area occupied per dye molecule on the TiO2 materials at 25 ◦ C on anatase is amon = 71.5 A˚ 2 /molecule for AF and amon = 91.8 A˚ 2 /molecule for AO. On rutile, amon = 2 A˚ 2 /molecule for AF and amon = 5 A˚ 2 /molecule for AO. The computational computations indicate that the orientation of both dyes on anatase is horizontal, whereas on rutile the vertical orientation has an Eads value only slightly inferior that that of the horizontal one. Then the vertical orientation can participate in a significant proportion in the adsorption of dyes on rutile on planes (0 0 1), besides the horizontal adsorption on planes (1 1 0). Moreover, the very low value of amon in rutile may be partially caused by the inclusion of molecules inside the pores, not adsorbed but simply trapped. Rutile has a bicontinuous structure with macropores with a diameter of about 6 ␮m (see Fig. 6). Anyway, despite the amon values are low, the general tendency is compatible with the computational results. The fitting of the experimental data on rutile is better for Freundlich equation than for that of Langmuir, which supports the interpretation that the adsorbent/dye interface is heterogeneous, while in anatase there is no significant difference between both isotherms, suggesting that the adsorbent interface is prob-

C.E. Zubieta et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 386 (2011) 71–78

ably homogeneous. There is no assurance that the derivation of the Freundlich equation is unique; consequently, if the data fit the equation, it is only likely, but not proved that the surface is heterogeneous [32]. Since in many cases the Langmuir equation gives adequate results in systems in which surface heterogeneity is known to be present [22] the fitting of the data by both models with similar correlation coefficients is possible. Then, this only observation does not assure the conclusion, but together with the computational information reinforces the interpretation that rutile has two almost equivalent different locations of adsorption. Direct comparison between computed Eads and experimental Hads is known that it is not possible. Moreover, Eads values were computed in vacuum, and Hads in aqueous solution; here the hydration of dyes and adsorbent must be taken into account. Therefore, only the relative tendencies are comparable. To facilitate the comparison, the sign of Eads expressed in kJ/mol was changed in Table 2. For AO on Anatase Hads = −175 kJ/mol and Eads = −20,400 kJ/mol, while for AF on the same adsorbent Hads = −112 kJ/mol and Eads = −13,500 kJ/mol. Then both methods (experimental and computational) indicate that the adsorption energy on anatase of AF is higher than that of AO. The inverse occurs in the adsorption on the 0 0 1 plane of rutile, i.e., the adsorption energy of AF is lower than that of AO. When the adsorption energies for the adsorption of AO on the planes 0 0 1 of anatase and rutile are compared, Eads and Hads show the same tendency: both values are lower for rutile than for anatase. The same conclusion can be obtained for AF. Then, the tendencies obtained from experimental and computational heats are the same. The cause of the stronger adsorption of AO compared with that of AF may be due to the –N(CH3 )2 groups of AO, which are stronger bases than the –NH2 , and then the interaction with the acidic groups of TiO2 . An indication of this interaction is that the adsorbent surface/adsorbed molecule distance (z) is smaller for AO on rutile ˚ than for AF (horizontal (horizontal position, (1 1 0) plane: 0.9 A) ˚ On the other hand, both molecules position, (0 0 1) plane: 1.95 A). ˚ on the (0 0 1) plane of anatase in show the same distance, 1.5 A, horizontal position. A comparison between the other positions and adsorption locations leads to the same general conclusion. The results of photocatalysis experiments confirmed that the activity of anatase is higher than that of rutile, and this activity depends on the contaminant nature. The higher photocatalytic activity of anatase may be due to surface porosity, band width and the presence of hydroxyl groups [33]. The specific area of anatase was larger than that of rutile (9.92 m2 g−1 and 3.83 m2 g−1 , respectively). The larger specific surface gives a larger amount of active sites and lower recombination of holes and electrons, which is the step determining the velocity of photodegradation. The process also depends on the amount of hydroxyl groups in the surface, which is in turn related to the specific area. However, the optimal band width is the most important factor in the photocatalytic activity. The generation of electron–hole pairs depends on this factor. For anatase the band width is 3.2 eV, whereas that of rutile is 3.0 eV, and this makes the recombination of the electron–holes pairs more difficult in anatase. The computational analysis has shown that the active surfaces for adsorption for anatase have a higher Ti4+ /O2− ratio than those of rutile. Since the metal generates the electron–hole pairs this may be the cause of the higher catalytic capacity of anatase when compared with rutile. On the other hand, the adsorption capacity seems to depend on the oxygen proportion on the surface, which favors rutile in comparison with anatase, in spite of the higher specific area of the latter. 5. Concluding remarks Experimental results indicate that the adsorption capacity of rutile is higher than that of anatase, despite of the higher specific

77

area of the latter adsorbent. Data also suggest that the adsorption surface of rutile is heterogeneous. The computational study supports these experimental findings. In spite of the fact that they are not directly comparable among them, both the computed Eads and the experimental Hads values indicate the same tendencies when the adsorption of each dye on the different adsorbents and both dyes on the same adsorbent. The adsorption of acridine orange (AO) on both adsorbents is higher than that of acriflavine (AF). This may be caused by the by the higher basic character of –N(CH3 )2 of AO groups compared with that of –NH2 of AF, which interacts with the oxygen atoms at the adsorbent surface. Computational simulation confirms that the main adsorbent surfaces of rutile have a higher proportion of oxygen atoms than that of anatase, and that AO equilibrium distance between the adsorbed molecule and the adsorbent surface is smaller for AO than for AF. Another possible factor is the macroporosity of rutile which may retain dye molecules trapped in the interior of pores. The photodegradation capacity of anatase is higher than that of rutile. This may be related to the higher Ti4+ /O2− ratio in the anatase adsorption surface than that in the rutile surfaces, as revealed by the computational simulation. Acknowledgements This work was financed by a grant of the Universidad Nacional del Sur and PICT # 560 and 656 of the Agencia Nacional de Promoción Científica y Tecnológica (ANPCyT). CEZ, EG and ILC have fellowships of the Consejo Nacional de Investigaciones Científicas y Técnicas de la República Argentina (CONICET). PVM and AJ are researchers of CONICET. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.colsurfa.2011.06.030. References [1] Y. Xie, F. Chen, J. He, J. Zhao, H. Wang, Photoassisted degradation of dyes in the presence of Fe3 + and H2 O2 under visible irradiation , J. Photochem. Photobiol. A: Chem. 136 (2000) 235–240. [2] Part A27. Triarylmethane and diarylmethane dyes, Ullmann’s Encyclopedia of Industrial Chemistry , 6th ed., Wiley-VCH, New York, 2001. [3] O. Ligrini, E. Oliveros, A. Braun, Photochemical processes for water treatment , Chem. Rev. 93 (1993) 671–698. [4] H. Zollinger, in: H.F. Eblel, C.D. Brenzinger (Eds.), Colour Chemistry, VCH, New York, 1987 (chapter 16). [5] C.E. Searle, Chemical Carcinogenesis ACS Monograph , American Chemical Society, Washington, DC, 1976. [6] C.T. Helmes, C.C. Sigman, Z.A. Fund, M.K. Thompson, M.K. Voeltz, M. Makie, A study of azo and nitro dyes for the selection of candidates for carcinogen bioassay , J. Environ. Sci. Health A 19 (1984) 97–231. [7] M. Boeninger, Carcinogenecity and Metabolism of Azo Dyes, Especially those Derived from Benzidine , DHHS (NIOSH), July 1980, Publication No. 80-119. [8] J.J. Roxon, A.J. Ryan, S.E. Wright, Reduction of water soluble azo dyes by intestinal bacteria , Food Cosmet. Toxicol. 5 (1967) 367–369. [9] M. Faisal, M.A. Tariq, M. Muneer, Photocatalysed degradation of two selected dyes in UV irradiated aqueous suspensions of titania , Dyes Pigments 72 (2007) 233–239. [10] S. Chen, G. Cao, Study on the photocatalytic reduction of dichromate and photocatalytic oxidation of dichlorvos , Chemosphere 60 (2005) 1308–1315. [11] C.C. Chen, C.S. Lu, Y.C. Chung, Photocatalytic degradation of ethyl violet in aqueous solution mediated by TiO2 suspensions , J. Photochem. Photobiol. A: Chem. 181 (2006) 120–125. [12] I.K. Konstantinou, T.A. Albanis, TiO2 -assisted photocatalytic degradation of azo dyes in aqueous solution: kinetic and mechanistic investigations: a review , Appl. Catal. B: Environ. 49 (2004) 1–14. [13] H. Kyung, J. Lee, W. Choi, Simultaneous and synergistic conversion of dyes and heavy metal ions in aqueous TiO2 suspensions under visiblelight illumination , Environ. Sci. Technol. 39 (2005) 2376–2382. [14] H. Mestankova, J. Krysa, J. Jirkovsky, G. Mailhot, M. Bolte, The influence of Fe(III) speciation on supported TiO2 efficiency: example of monuron photocatalytic degradation , Appl. Catal. B: Environ. 58 (2005) 185–191.

78

C.E. Zubieta et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 386 (2011) 71–78

[15] C. Chen, X. Li, J. Zhao, H. Hidaka, N. Serpone, Effect of transition metal ions on the TiO2 -assisted photodegradation of dyes under visible irradiation: a probe for the interfacial electron transfer process and reaction mechanism , J. Phys. Chem. B 106 (2002) 318–324. [16] N. Watanabe, S. Horikoshi, A. Kawasaki, H. Hidaka, N. Serpone, Formation of refractory ring-expanded triazine intermediates during the photocatalyzed mineralization of the endocrine disruptor amitrole and related triazole derivatives at UV-irradiated TiO2 /H2 O interfaces , Environ. Sci. Technol. 39 (2005) 2320–2326. [17] N.M. Dimitrijevic, Z.V. Saponjic, B.M. Rabatic, T. Rajh, Assembly and charge transfer in hybrid TiO2 architectures using biotineavidin as a connector , J. Am. Chem. Soc. 127 (2005) 1344–1345. [18] C.C. Chen, C.S. Lu, F.D. Mai, C.S. Weng, Photooxidative N-de-ethylation of anionic triarylmethane dye (sulfan blue) in titanium dioxide dispersions under UV irradiation , J. Hazard. Mater. B 137 (2006) 1600–1607. [19] D.F. Duxbury, The photochemistry and photophysics of triphenylmethane dyes in solid and liquid media , Chem. Rev. 93 (1993) 381–433. [20] M.R. Hoffman, S.T. Martin, W. Choi, D.W. Bahnermann, Environmental applications of semiconductor photocatalysis , Chem. Rev. 95 (1995) 69–96. [21] A. Fujishima, T.N. Rao, D.A. Tryk, Titanium dioxide photocatalysis , J. Photochem. Photobiol. C: Photochem. Rev. 1 (2000) 1–21. [22] P.C. Hemenz, R. Rajagopalan, Principles of Colloid and Surface Chemistry , 3rd ed., Dekker, New York, 1997, pp. 336–338. [23] A.W. Adamson, A.P. Gast, Physical Chemistry of Surfaces , 6th ed., Wiley, New York, 1997, pp. 336–338.

[24] A.B. Anderson, Derivation of the extended Hückel method with corrections: one electron molecular orbital theory for energy level and structure determinations , J. Chem. Phys. 62 (1975) 1187–1188. [25] P.J. Cromer, K. Herrington, The structures of anatase and rutile , J. Am. Chem. Soc. 77 (1955) 4708–4709. [26] S.C. Abrahams, J.C. Bernstein, Rutile normal probability plot analysis and accurate measurement of crystal structure , Chem. Phys. 55 (1971) 3206–3211. [27] A. Davydov, N. Sheppard, Molecular Spectroscopy of Oxide Catalyst Surfaces , John Wiley and Sons, New York, 2003. [28] H. Zhuang, J.F. Banfield, Understanding pholymorphic phase transformation behavior during growth of nanocrystalline aggregates: insights from TiO2 , J. Phys. Chem. B 104 (2000) 3481–3487. [29] K.S.W. Sing, Reporting physisorption data for gas/solid systems with special reference to the determination of surface area and porosity , Pure Appl. Chem. 57 (4) (1985) 603–619. [30] J.H. de Boer, B.C. Lippens, B.G. Linsen, J.C.P. Broeckhoff, The t-curve of multimolecular N2 -adsorption , J. Colloid Interface Sci. 21 (1966) 405–414. [31] D.M. Smith, G.W. Scherer, J.M. Anderson, Shrinkage during drying of silica gel , J. Non-Cryst. Solids 188 (1995) 191–206. [32] A.W. Adamson, A.P. Gast, Physical Chemistry of Surfaces , 6th ed., Wiley, New York, 1998, pp. 393–394. [33] R.J. Tayade, P.K. Surolia, R.G. Kulkarni, R.V. Jasra, Photocatalytic degradation of dyes and organic contaminants in water using nanocrystalline anatase and rutile TiO2 , Sci. Technol. Adv. Mater. 8 (2007) 455–462.