Computational investigation of the adsorption and photocleavage of chlorobenzene on anatase TiO2 surfaces

Chemical Physics 353 (2008) 93–103

Contents lists available at ScienceDirect

Chemical Physics journal homepage: www.elsevier.com/locate/chemphys

Computational investigation of the adsorption and photocleavage of chlorobenzene on anatase TiO2 surfaces Hilal S. Wahab a,*, Thomas Bredow b, Salah M. Aliwi c a

University of Technology, Department of Applied Sciences, P.O. Box 35319, Baghdad, Iraq Universität Bonn, Institut für Physikalische und Theoretische Chemie, Bonn, Germany c Ministry of Higher Education and Scientific Research, Baghdad, Iraq b

a r t i c l e

i n f o

Article history: Received 25 April 2008 Accepted 30 July 2008 Available online 5 August 2008 Keywords: Titanium oxide Adsorption Photocleavage Chlorobenzene Semiempirical method Model calculations

a b s t r a c t The adsorption and photocleavage of chlorobenzene (CB) molecule on the anatase TiO2 (0 0 1), (1 0 0) and (0 1 0) surfaces are studied with semiempirical SCF MO method, MSINDO. The surfaces have been modeled with two saturated clusters Ti21O58H32 and Ti36O90H36. The dissociative perpendicular adsorption of CB on TiO2 (0 1 0) and (1 0 0) surfaces revealed comparable stabilities and much higher than on the (0 0 1) surface. The aromatic ring cleavage by atomic oxygen, singlet oxygen and superoxide anion molecules has been investigated computationally and relevant mechanisms are proposed. Molecular dynamics (MD) simulations have been implemented for the adsorption models and the early stages of photocleavage mechanisms. The oxygen-type chemistry is involved actively in the water mediated photocleavage step upon excitation. The O 2 radical anion dependant ring opening mechanism, through the dioxetane intermediate, is thermodynamically the most favourable. Ó 2008 Elsevier B.V. All rights reserved.

1. Introduction Chloroaromatics are widely distributed organic pollutants in the environment. Their presence in air and aquatic environments stems principally from effluents of pesticides, plasticizers, preservatives, dyes, and disinfectants manufactures [1,2]. Furthermore, they have been detected in chlorinated and surface water [3], in urban air and automobile exhausts [4], and in emissions from municipal waste incinerators [5]. These materials show low biodegradability, and are classified as high priority pollutants due to their toxicity, persistence and carcinogenicity [6,7]. Moreover, the chlorination disinfection method of the drinking water may increase their toxicity through generation of other haloaromatics as disinfection by products, which are equally or more toxic than the parent compounds [8,9]. It is now generally accepted that the degradation of the organic moieties occurs on the surface of the photocatalyst [10,11]. It is therefore logical to expect that the adsorption of the organic substrate is an important factor for the highly efficient detoxification process [12]. Moreover, several investigators have observed that there is a clear relationship between substrate adsorbability and photocatalytic degradation [13,14].

* Corresponding author. Tel.: +964 7801 620 707. E-mail addresses: [email protected], [email protected] (H.S. Wahab). 0301-0104/$ - see front matter Ó 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.chemphys.2008.07.017

In this work, chlorobenzene (CB) was chosen as a model compound due to its environmental impact as a common recalcitrant contaminant and the relatively simple molecular structure which facilitates a deeper understanding of the photoreaction mechanisms. The photocatalysis of chlorinated aromatics using TiO2 particles have become, over the last three decades, the most attractive and commonly employed method [15]. Nevertheless, the ring opening step, which is the key mechanism step, is still experimentally under debate. The ambiguities concerning the mechanistic pathways have often led to conflicting remarks [16]. A theoretical approach is therefore, strongly motivated for predicting the non-aromatic and acyclic intermediates. We are aware of only a few studies in which any ring opened products are discussed [17–20]. Some investigators have reported that the reaction of oxygen, as superoxide or singlet oxygen, with aromatics in photocatalytic degradation process results in different intermediate products like; dioxetane [20,21], endoperoxide [18,22], and benzene oxide or epoxide [10,23] prior to the ring opening step. On the other hand, some researchers [17–19,24,25] have suggested that the ring opening reactions are induced by a single electron transfer rather than hydroxyl-type chemistry. Accordingly, one of the main objectives of this study, in addition to the adsorption modes of CB molecule, is to scrutinize its ring opening mechanism on anatase TiO2 surface through oxygen mediated attack.

94

H.S. Wahab et al. / Chemical Physics 353 (2008) 93–103

2. Computational method The quantum chemical model calculations are performed with the semiempirical molecular orbital method MSINDO [26,27]. This method is a modification of the semiempirical SCF MO method SINDO1 [28], which is based on the INDO approximation. The new parameterization of MSINDO for first-row transition metals from Sc to Zn [29] offers a highly improved accuracy for structure and energy of transition metal compounds. The transition metal atoms are described by a pseudominimal Slater basis set (3d, 4s, 4p), the first row elements by a (2s, 2p) basis set and the Al to Cl by (3s, 3p) basis set and 3d for polarization. The combination of reliable accuracy for structure and energy and the speed of computation for large systems make the semiempirical MSINDO a useful tool for the present study. Molecular Dynamics (MD) has been used for the studying of the dynamics of certain cluster-substrate (TiO2–C6H5Cl) models, in which the statistical average of the kinetic energy of the system becomes constant when the simulation has reached constant temperature. Constant temperature dynamics are performed using Nosé– Hoover-Chain thermostat. The adsorption energies (Eads) are computed from the binding energies (EB) of TiO2 cluster, substrate (chlorobenzene) and cluster-substrate systems for all the fully optimized geometries, to determine their stability relative to each other, using the following formula [30];

Eads ¼ Ecluster-substrate  Ecluster  Esubstrate B B B

ð1Þ

A negative value for Eads therefore, designates a stabilization of the cluster-substrate system due to adsorption. We have used the Ti21O58H32 and Ti36O90H36 clusters for modeling of anatase (0 0 1), (0 1 0) and (1 0 0) surfaces, which are devoted for the adsorption studies.

3. Results and discussion 3.1. Adsorption geometries Fig. 1 depicts the parallel adsorption geometry of CB molecule on the anatase TiO2 (0 0 1), (1 0 0) and (0 1 0) surfaces. In the case of (0 0 1) surface, it is noticed that the CB molecule moves away from the surface during optimization. Whereas, it approaches the surface in the case of (1 0 0) and (0 1 0) surfaces. This counter-intuitive phenomenon, in the case of (0 0 1) surface, could be ascribed either to the surface morphology of projected twofold oxygen ions which repulse the chloride ion, or the cluster Ti21O58H32 is not large enough in order to describe the relaxation of the parallel adsorption realistically. The drawbacks of some minimal surface models have also been reported previously [31]. Also from Fig. 1, it is obvious that the distance of chloride ion from the Ti5C ion of the surface is more than 3 Å, i.e., the possibility of real bond formation is unrealistic. Hence, one can envisage this sort of adsorption as physisorption rather than chemisorption due to the weak attraction forces of the Van der Waals type [32]. The same trend was observed with perpendicular adsorption through chloride ion, which is illustrated in Fig. 2. The strong electrostatic force between the chloride ion and the coordinatively unsaturated surface titanium ion, induces the attraction of CB molecule toward the catalyst surface, resulting in its adsorption through Cl–Ti interaction, at a distance of less than 3 Å (Fig. 2 /b and /c). This computational outcome is consistent with the experimental results of Tanaka and Saha [33], in their photocatalytic degradation study of 2,4,6-trichlorophenol (TCP). They observed a significant amount of adsorption at relatively low pH (3.1–4.2), where the undissociated TCP was predominant. Besides, they reported that the adsorption of TCP is

mainly ascribed to the electrostatic attraction of the negatively charged sites of TCP (chloride ions) by the high positively charged surface at low pH values. Table 1 shows the calculated adsorption energies, Eads, for the parallel and perpendicular adsorption geometries seen in Figs. 1 and 2, respectively. The adsorption process is exothermic [32] and accordingly, all the Eads values in Table 1 are negative in sign. Results from Table 1 indicate that: (a) the adsorption energy is influenced by the substrate geometry; (b) lower substrate–surface interaction energy corresponds to closer approach of CB molecule to the surface; (c) the CB molecule is adsorbed on the anatase TiO2 surfaces in parallel and perpendicular modes, but with different Eads energies. The perpendicular adsorption mode through chloride ion is energetically more favoured in comparison with parallel adsorption geometry. Substantial experimental research work has been reported on aromatic ring adsorption geometry. Wu et al. [34] have studied the adsorption of benzene and iodobenzene on TiO2 powder. They observed, through FTIR study, parallel adsorption geometry. The flat adsorption of benzene molecule on the surface was also suggested by other researchers [33,35]. Moreover, Lichtenberger and Amiridis [36] have reported the parallel adsorption of benzene and chlorinated benzenes over TiO2 and V2O5/TiO2 catalysts surfaces. In contrast, other investigators [37,38] have reported the perpendicular orientation of the aromatic ring relative to the surface in their studies of benzoic acid and chlorobenzoic acid adsorption on titania. Furthermore, Tahiri et al. [37] have concluded that the ionic interaction of a carboxylic group with the OH groups of titania would induce a preferential orientation of the molecule with the plane of the aromatic ring perpendicular to the surface. Ogawa [38] also reported similar conclusion. In addition, Street et al. [39] conducted adsorption studies of benzene on MgO using temperature programmed desorption and high resolution electron energy loss spectroscopy (HREELS). They concluded that benzene chemisorbs nondissociatively and the ring plane is essentially parallel to the catalyst surface at low surface coverage, but becomes perpendicular to the surface at higher coverage. Accordingly, our computational findings in the present work regarding the stability of both parallel and perpendicular adsorption geometries of the aromatic plane, agree well with the experimental results. However, the perpendicular adsorption mode is energetically and geometrically more preferable. 3.2. Dissociative adsorption Upon optimization, the CB molecule is oriented with the chloride ion pointing towards the surface titanium ion, representing the experimental adsorption in the dark (Fig. 2 /b and /c). Actually, there are variety of speculations about possible reactions which can occur after light absorption (excitation) at titanium dioxide particles both in air and aqueous suspensions. The cluster-substrate system has been excited using Restricted Hartree Fock (RHF)-Configuration Interaction for Singles (CIS) 30  30 calculation. Bond order difference calculations for the system before and after excitation have shown a decrease in C–Cl bond order (0.106) and an increase in Cl–Ti bond order (+0.165). This clearly indicates the possibility of C–Cl bond breakage and Cl–Ti bond formation during the excitation. Lichtenberger and Amiridis [36] have reported that C–Cl bond is weaker than C–H bonds, and hence more prone to be attacked particularly by nucleophiles. Furthermore, Chen et al. [40] have stated that upon direct excitation of monosubstituted halobenzenes, a specific heterolytic scission of carbon–halogen bond takes place. The potential energy curves (S0, ground and S1, excited states) are determined by linear interpolation of a reaction coordinate R, between the reactant and the product. The distance of the chloride ion of CB molecule from the titanium atom of the (0 1 0) surface is

H.S. Wahab et al. / Chemical Physics 353 (2008) 93–103

95

Fig. 1. Parallel adsorption of CB molecule on anatase TiO2; (a) (0 0 1); (b) (1 0 0); (c) (0 1 0) surfaces; (a, b and c) represent initial structures; (/a, /b and /c) represent optimized structures. Ti (large light), H (small light), O (small dark), C (medium dark), Cl (large dark). Dashed lines represent distances.

the reaction coordinate for the detachment of chloride. This reaction coordinate was measured at a step size of 0.1 Å starting from R = 2.3 Å. For each interpolation step, all the system was fully optimized excluding those atoms, which are involved in the reaction using cartesian select geometrical optimization parameter. For these optimized geometries, the RHF CIS 30  30 calculation was performed in order to estimate total energies for each interpolation step. For an estimate of the relative probability of a certain reaction, the activation energy plays a special role. The potential curve of the S0 ground state in Fig. 3 shows a barrier of 82 kJ/mol (0.85 eV) at R = 1.8 Å, which indicates that the thermal reaction is unlikely. While, the barrier on the excited S1 surface is only 27 kJ/mol (0.28 eV) and this would allow a reaction on the excited singlet surface. Alternatively, the S0 ground state barrier can be overcome after excitation to S1 and subsequent Internal Conversion (IC) from S1 to a vibrationally excited state of S0. In organic chemistry, this is called a hot ground state reaction [41]. The S0/S1 activation energy ratio is > 3. This confirms clearly the advantages of the photochemical reactions over thermal reactions. The present calculation has elucidated the first step of a series of complex processes in photocatalysis of CB molecule in air and solution. Lichtenberger and Amiridis [36] concluded that the first step in the oxidation of chlorinated benzenes is their dissociative adsorption on a transition metal oxide site via chloride abstraction. They experimentally determined the activation energy for this step as 96 kJ/mol. The ground state activation energy for the first step process found in the present work (82 kJ/mol) is in a good agreement with the experimental value.

Fig. 4a shows the simulation of the phenyl ring adsorption, after chloride ion detachment, on the surface titanium ion. Upon optimization, the phenyl ring detached from surface titanium and was bound to the surface oxygen ion (Fig. 4b) with a C–O bond length of 1.521 Å. This indicates that the adsorption of phenyl ring on surface titanium ion is energetically not favorable. In contrast, the perpendicular adsorption of phenyl ring on surface lattice oxygen ions, as it is illustrated in Fig. 4b, has resulted in adsorption energies, Eads, of 74, 151 and 157 kJ/mol with (0 0 1), (1 0 0) and (0 1 0) surfaces of anatase TiO2, respectively. Based on the above computational findings, one may propose that some sort of nucleophilic mechanism is taking place onto the surface leading to the formation of an adsorbed surface phenoxy species. The nucleophilic nature mechanism suggested in the present work agrees well with the following experimental interpretations. Wu et al. [34] have concluded that the generation of phenoxy species in photo irradiation of benzene on TiO2 in the absence of water or oxygen indicates that the photoreaction can be initiated by the hole capture of adsorbed benzene, forming benzene radical cation ( C6 Hþ 5 ). This species may then interact with surface lattice oxygen ions to produce phenoxy species. Lichtenberger and Amiridis [36] have also reported, for the adsorption of 1,3-dichlorobenzene on pure titania at 250 °C, formation of surface phenolate species. Accordingly, they concluded that the first step of the mechanism is believed to have a nucleophilic substitution nature as indicated by the in situ FTIR results. Finocchio et al. [42] have also proposed the involvement of nucleophilic oxygen species (i.e., O2 ions on oxidized cationic sites) in the mechanisms of the oxidation of

96

H.S. Wahab et al. / Chemical Physics 353 (2008) 93–103

Fig. 2. Perpendicular adsorption of CB molecule through chloride ion on anatase TiO2; (a) (0 0 1); (b) (1 0 0); (c) (0 1 0) surfaces; (a, b and c) represent initial structures; (/a, /b and /c) represent optimized structures. Ti (large light), H (small light), O (small dark), C (medium dark), Cl (large dark). Dashed lines represent distances.

Table 1 Values of adsorption energies, Eads (kJ/mol) for different adsorption modes of CB molecule on anatase TiO2 surfaces Mode

Parallel Perpendicular through chloride ion

Surface 001

100

010

Figure

13 37

52 121

55 128

1 2

hydrocarbons over MgCr2O4 catalyst. Our proposed nucleophilic attack is also consistent with conclusions of Van den Brink et al. [43] who reported that C–Cl bond scission is the first step of oxidation of CB molecule over Pt/c-Al2O3 and takes place even at room temperature. For further confirmation of the earlier suggested mechanism, we employed a MD simulation for the adsorption model depicted

in Fig. 4a. The temperature of the MD simulation was 300 K and the time step used was one femtosecond (1 fs). The simulation was started from the non-optimized geometry to facilitate the crossing of any energy barriers. During the simulation, phenyl ring migrated and was bound to the surface lattice oxygen ion. The adsorption structure after two picoseconds (2 ps) was similar to the structure seen in Fig. 4b. The vibrational density of states (VDOS) of the selected atoms (carbon, hydrogen and the candidate surface oxygen) in the system were obtained by calculating the Fourier transform of their velocity auto correlation function (VACF) available in MSINDO. The estimated VDOS for the selected atoms is shown in Fig. 5a. The spectra shown in Fig. 5a include several bands that can be assigned to a surface phenoxy species. In particular bands located at 1576, 1474 and 1514 cm1 can be assigned to the C–C and C@C stretching vibrations, bands at 1244 and 1160 cm1 to the C–H in-plane bending vibrations, bands at 3038 and 3068 cm1 for the C–H stretching vibrations and a band at

H.S. Wahab et al. / Chemical Physics 353 (2008) 93–103

97

Fig. 3. Potential curves for chloride ion detachment.

Fig. 5. The VDOS of adsorption geometry; (a) aromatic carbon and hydrogen atoms, and the candidate surface oxygen atom; (b) carbon and oxygen atoms involved in C–O bond.

It is clear that the C–O bond exists in the system. VDOS of C–O bond was calculated separately from the selected atoms. The stretching vibration band of 1274 cm1 is in an excellent agreement with literature results of 1278 cm1 [36] and 1276 cm1 [34]. Hence, it is inferred that the adsorption of CB molecule on TiO2 precedes through the formation of a surface phenoxy species. The adsorption model seen in Fig. 4b was excited on RHF CIS 30  30 level. The excitation energy has been determined to be 3.52 eV, which is lower than the excitation energies of both the cluster and the substrate separately. This indicates that the substrate is involved strongly in the system through bonding with cluster surface due to the adsorption. 3.3. Plausible photodegradation mechanism pathways of chlorobenzene Fig. 4. (a) Simulation of phenyl ring adsorption on surface titanium ion (initial structure); (b) phenyl ring adsorption on surface lattice oxygen ion (minimum energy structure). Ti (large light), H (small light), O (small dark), C (medium dark), Cl (large dark). Dashed lines represent distances.

1274 cm1 is assigned to the C–O stretching vibration of an adsorbed phenoxy species. Similar bands located at 1256, 1478, 3032 and 3068 cm1 were observed during the adsorption of benzene on TiO2 [34]. Also, from FTIR experiments of catalytic oxidation of chlorinated benzene on TiO2 [36], analogous bands located at 1468, 1575, 1442, 1252, 1158 and 1278 cm1 have been obtained. The estimated VDOS for the C–O bond is shown in Fig. 5b.

All species observed during the adsorption of CB molecule onto the titania are ring structures, suggesting that the molecule adsorbs without any bond breakage in the aromatic ring. Hence, it is expected that the bond breakage in the aromatic ring requires the presence of oxygen and to some extent water molecules. The photocatalytic degradation studies for aromatic [44] and haloaromatic [45,46] molecules which identify the partially oxidized intermediates prior to the ring opening step, are most likely photocatalytic transformation rather than photocatalytic degradation. Because of the carcinogenic activity and cytotoxic nature of the aromatic and haloaromatic compounds [23], it is vital to reveal

98

H.S. Wahab et al. / Chemical Physics 353 (2008) 93–103

the ring opening step, which is the key mechanism step for the photocatalytic degradation [19,24]. Moreover, it is not always necessary to drive the reaction all the way through to CO2, since the ring opening process converts toxic chlorinated aromatic compounds into biodegradable species such as aldehydes and carboxylic acids [47]. In this study, the mechanism of cleavage of the aromatic ring has been given the attention, focusing on different oxygen species, which have been previously proposed [25] as key reactants during the ring opening in the degradation process. The bond order changes in the adsorbed substrate (Fig. 6) upon S0 ? S1 excitations are given in Table 2. Briefly, all C–C bonds are influenced by excitation. The decrease in bond order of the C@C bonds are greater than of C–C bonds. This indicates that the probability of the addition reactions is dominant. While, there is a positive increase in the bond order of C–O bond, which refers implicitly to the stronger binding of the phenyl ring to the surface at this stage. The geometry changes, upon electronic excitation of CB molecule, have also been observed by other investigators [48,49]. They reported that all aromatic ring C–C bonds expand proportionally due to the excitation.

Table 2 Bond order changes followed S0 ? S1 excitation for the adsorption model seen in Fig. 6

3.3.1. Possible mechanism for atomic oxygen dependent ring opening The role of three types of oxygen species, which exhibited a high adsorbability on anatase TiO2 surfaces [50], atomic (O), singlet (1O2) and superoxide radical anion (O 2 ) in the ring opening step, has been investigated computationally in this work. It is known that two types of mechanisms [51] play a role in heterogeneous catalysis: the Langmuir–Hinshelwood mechanism (LH) [19], according to which the reactants are adsorbed from the gas or liquid phase on the surface of the catalyst and react there to the product, or the Eley–Rideal mechanism (ER) [52], where it is assumed that one reactant is adsorbed and the second reactant reacts from the gas or liquid phase with the adsorbed reactant. As a result, experimentally, two different reactions of the same compound might show different responses to light intensity, catalyst loading and electron accepting additives [53]. The majority of experimentalists [10,12–14,36] observed that there is a clear relationship between the adsorbability and photocatalysis, and consequently it has been suggested that preadsorption (i.e., LH model) is a prerequisite for highly efficient detoxification process. Accordingly, the LH mechanism has been chosen as a basis in this study. Moreover, since all the reactions are taking place on the surface, all the reactants and products are kept adsorbed to the surface. In addition, the reaction sequence implied two steps, first oxygen then water interactions. Atomic oxygen is generally considered as an efficient oxidant with oxidation potential of 2.42 V [54]. Furthermore, the high electron density in aromatic rings induces the electrophilic radicals for interaction.

For further confirmation, the potential energy curves (S0 ground and S1 excited states) are determined. The C5@C6 bond length was fixed at 1.42 Å; because this bond exhibited the lowest bond order upon excitation as it is presented in Table 2. The length of the forming C–O bond is the reaction coordinate, R, which was measured at a step size of 0.10 Å starting from 2.10 Å. For each step, the system has been optimized under constrained conditions. For these optimized geometries, the CIS 30  30 calculations have been carried out aiming to estimate the total energy for each step. The potential curve for the ground state, S0, in Fig. 7 exhibits a barrier of 76 kJ/mol (0.79 eV) at R = 1.80 Å, which indicates that the thermal reaction is unlikely. Whereas, the barrier on the excited S1 surface is only 23.6 kJ/mol (0.25 eV), which agrees well with the experimental data of 20.2 kJ/mol [55] and 21 kJ/mol [56] for the addition reaction of O (3P) with C6H6. The above outcomes are in accordance with the previously reported literature data. Barckholtz et al. [57] have concluded, in their study of monocyclic aromatic hydrocarbon reaction with

Fig. 6. The model of the adsorbed phenyl ring at the surface lattice oxygen ion (O1).

Fig. 7. Potential curves for atomic oxygen addition reaction.

Bond

D bond order

Bond

D bond order

C1@C2 C2–C3 C3@C4 C4–C5

0.155 0.062 0.150 0.066

C5@C6 C6–C1 C1–O1

0.161 0.088 +0.031

Eq. (2) reveals the MD simulation results for the formation of what is called benzene oxide or benzene epoxide in the literature. The simulation was performed at 300 K and for two picoseconds (2 ps). The formation of structure 2 indicates that the benzene oxide intermediate molecule is stable for the first 2 ps of the reaction. O + O(ads) Structure 1

ð2Þ

hν, TiO2 Structure 2

99

H.S. Wahab et al. / Chemical Physics 353 (2008) 93–103

atomic H and O using DFT that the radical addition channel is preferred at 298 K and the H-atom abstraction channel becomes dominant at high temperatures. However, Ghigo and Tonachini [23,30] have reported, in their theoretical investigation on some troposphere oxidation channels of phenol with atmospheric O2, the formation of benzene oxide as one of the products. Another theoretical study on toluene [58], carried out at the DFT level of theory, has focused on the identification of plausible reaction intermediates, explored the epoxide structure and found it to be rather stable. While, some experimentalists [25,59] have reported the formation of benzene epoxide as an intermediate in the oxidation of aromatic compounds. However, Walling and Johnson [59] have stated that there is no evidence for its actual formation, because it should immediately re-open due to hydrolysis. In the same spirit, Wu et al. [34] have stated that the photoproducts extracted from the TiO2 catalyst after the photo process, may not reflect the actual species that are formed on the surface during the photo illumination. The hydrolysis of benzene oxide has also been modeled in this work. Fig. 8b shows the product of simultaneous interaction of atomic oxygen with the phenyl ring and attacking of the formed epoxide by water molecule which was initially adsorbed at surface lattice oxygen (O2C) via hydrogen bonding. Fig. 8a represents the simulation of the excited state geometry in which the created atomic oxygen is adsorbed on the surface titanium ion and C5@C6 bond-length is 1.42 Å. The full optimization of this geometry has lead to the final structure seen in Fig. 8b. The hydrolysis of epoxide in either acidic or basic media produces glycols [60]. The geometry shown in Fig. 8b has been excited on CIS 30  30 level and the bond order calculation has shown a decrease in the bond order of C–C bond (0.049) and in one of the C–O bonds of the epoxide (0.084), whereas, the bond order of the other C–O bond of epoxide is increased (+0.013). Accordingly, the following reaction is anticipated;

H HO HO

ð3Þ H

hν, TiO2

+ O(ads) + H2O(ads)

Structure 3

It has been reported [25,61] that the epoxide intermediate, under UV illumination, is subjected to immediate hydrolysis, rearrangement, the opening of the ring and formation of a bifunctional species as follows; O

OH O

C + H2 O

hν

OH

ð4Þ

CH2OH

The bond order calculations of the excited diol intermediate have shown an increase in the bond order of one of the C–O bonds (+0.105), which is most likely attributed to the change in hybridization from sp3 to sp2, and a decrease in the bond order of C–C bond of the diol (0.199). According to the forgoing findings, the reaction (4), therefore, should proceed through the diol intermediate formation step (reaction (3)) prior to the ring cleavage as illustrated in Eq. (5):

H

O

O

H

H-C H2C OH hν

H-O + O(ads) + H2O(ads)

hν TiO2

H

TiO2

Structure 4

ð5Þ Johnson et al. [22] have also observed this type of ring cleavage in their study about the atmospheric photo oxidation of toluene, which resulted in a dialdehyde ring cleaved product. For further evidence for the proposed ring opening mechanism pathway in this study, the energy of reaction, Er, was computed for the intermediates and the final ring cleaved product, from the total energy, ET, values presented in Table 3. The calculated Er for the reactions (2) and (3) and the overall reaction (5) have been determined to be 71, 176 and 362 kJ/mol, respectively. Despite all the reactions are exoergic, but there is a clear distinction between Table 3 Calculated total energies, ET, for the reactants, intermediates and products of ring opening paths using O, 1O2 and O 2 species

Fig. 8. Reaction of water molecule with benzene oxide; (a) initial structure; (b) final structure. Ti (large light), H (small light), O (small dark), C (medium dark). Dashed lines represent distances; dotted line represents hydrogen bond.

Molecule

ET (Hartree)

DE (kJ/mol)

Location

Ti36O90H36 C6H5Cl Cl H2O Atomic O 1 O2 O 2 Structure 1 Structure 2 Structure 3 Structure 4 Structure 5 Structure 6 Structure 7 Structure 8 Structure 9 Structure 10 Structure 11 Structure 12 Structure 13 Structure 14

1578.220 52.268 13.968 17.018 15.656 31.453 31.380 1616.540 1632.223 1649.281 1649.352 1648.070 1648.081 1648.066 1648.095 1648.108 1648.091 1616.371 1647.931 1648.061 1648.089

– – – – – – – –

– – – – – – – Eq. (2) Eq. (2) Eq. (3) Eq. (5) Fig. 12 Fig. 12 Fig. 12 Fig. 12 Fig. 12 Fig. 12 Fig. 13 Fig. 13 Fig. 13 Fig. 13

71 176 362 202 231 192 66 71 66 – 473 341 74

100

H.S. Wahab et al. / Chemical Physics 353 (2008) 93–103

the stabilities of benzene epoxide on the one hand, and the diol and the product on the other hand. The latter are more stable and thermodynamically favoured. This finding justifies the immediate hydrolysis of the benzene epoxide experimentally into the bifunctional, aldehyde and carboxylic acid, product. 3.3.2. Possible mechanisms for O2-dependent ring opening Some investigators [62,63] have studied the reaction of 1O2 with benzene and chlorobenzene. Dilmeghani and Zahir [62] have observed, experimentally, the formation of photo oxygenated intermediates, whereas, Bobrowski et al. [63] concluded, according to the ab initio calculations on the hypothetical addition of 1O2 to benzene, the formation of 1,4-endoperoxide when the benzene ring bears electron donating substituents. Moreover, the 1,2 chemical addition of 1O2 to xanthene methoxy vinyl pyrene [64] and carotene [65] have also revealed the formation of endo- and exoperoxides which are known as dioxetanes. The relatively long lifetime of 1O2, because of the return to the ground state is spin forbidden [66], extends its presence on the excited surface and consequently enhances the reaction possibility with the substrates on the surface. Accordingly, three types of dioxetanes (Fig. 9) have been modeled and studied. The molecular dynamic simulations for the geometries shown in Figs. 9a, b and c at 300 K and for 2000 fs (2000 steps) have resulted in similar structures. This indicates that these dioxetane intermediates are stable for the first 2 ps of the reaction. Furthermore, the computed heats of formation (obtained at the MSINDO level) of these three geometries have presented exoergic values. For further confirmation, the VDOS of the selected atoms involved in the dioxetane model has been estimated and compared with the VDOS of hydrogen peroxide (H2O2). The spectrum shown in Fig. 10a includes two bands located at 1116 and 1198 cm1 that can be assigned to the C–O and O–O stretching vibrations, respectively. The band located at 1198 cm1 in Fig. 10b is also assigned to the O–O stretching vibration in the dioxetane structure. For the sake of comparison, the band located at 1236 cm1 in Fig. 10c which belongs to the stretching vibration of O–O bond in the simulated H2O2 model confirms the presence of the peroxide conformation in the structure of the dioxetane intermediate which is suggested in this study. The simulation of hydrolysis for the structure seen in Fig. 11a with H2O molecule did not show, during optimization of the model, any interaction with water molecule. The hydrogen bonded water molecule to the surface, detached from the surface and migrated away from the adsorbed dioxetane as it is seen in Fig. 11b. For scrutinization purpose, the MD simulation, at 300 K using 2000 steps, has been performed for the initial structure in Fig. 11. The outcome of the MD simulation has also exhibited the migration of the H2O molecule away from the adsorbed dioxetane. Accordingly, one can predict that the cleavage of the dioxetane does not follow the water hydrolysis path. Andino et al. [67] have

O O O O

O

Fig. 9. The modeled dioxetanes: (a) ortho; (b) meta; (c) para.

O

Fig. 10. The VDOS of the dioxetane and H2O2 molecules; (a) carbon and oxygen atoms; (b) only oxygen atom of peroxo group in dioxetane; (c) hydrogen peroxide molecule as reference mode.

concluded, in their theoretical atmospheric photo oxidation study for toluene that the only path for the fragmentation of bicyclic oxy radicals (endo-peroxides) is via favourable b-scission reactions. These fragmentation reactions yield eventually dicarbonyl products. Furthermore, Muneer et al. [21] have observed, in the photocatalysis of 2-naphthoic acid on TiO2 in CH3CN/H2O mixture, the cleavage of the exo-peroxide naphthalene dioxetane into dialdehydes under UV light and in the presence of atmospheric oxygen. For comparison purpose with other researchers’ observations [21,67], RHF CIS 30  30 excitation has been performed for the geometry presented in Fig. 9a as a sample model. The bond order calculations have exhibited a decrease in the bond order of the

101

H.S. Wahab et al. / Chemical Physics 353 (2008) 93–103

O

1

O2(ads)

hν, TiO2 ΔE = - 192 kJ/mol

hν, TiO2 ΔE = - 231 kJ/mol

hν, TiO2 ΔE = - 202 kJ/mol

+

O O O O

Structure 7 hν, TiO2 ΔE = - 66 kJ/mol

Fig. 11. Simulation of the dioxetane hydrolysis; (a) initial structure; (b) final structure. Ti (large light), H (small light), O (small dark), C (medium dark). Dashed lines represent distances; dotted line represents hydrogen bond.

O

H

O

C O–O and C–C bonds, with values of 0.169 and 0.154, respectively. Hence, this finding is in a fair agreement with the observations of other computational [67] and experimental [21] studies. For further scrutinization, and to rationalize the formation of dicarbonyls from the cleavage of the dioxetane under UV illumination, the localization approach of Wheland [68] was tested. According to this approach, the energy of the intermediate determines the path of the reaction. Furthermore, the primary means of differentiating between the models is the calculated reaction energy, Er [69]. The total energies, ET, of the intermediate dioxetane and the products calculated in this study are tabulated in Table 3. As can be seen from the values given in Table 3, the structure 6 is situated 231 kJ/mol below the reactants, as compared to 202 kJ/ mol for structure 5 and 192 kJ/mol for structure 7. Moreover, the highest exothermic character, as it is seen from Fig. 12 belongs to structure 9, i.e., most negative reaction energy. This is somewhat self confirmation for the data shown in Table 2 which presents the maximum decrease in the bond order, upon excitation, belongs to C5@C6 bond. This consequently, promotes the addition reaction of singlet oxygen with aromatic pollutants. The use of superoxide radical anion (O 2 ) is another element of the present research strategy. In many heterogeneous photocatalytic reactions, the adsorbed oxygen serves as a trap for the photogenerated conduction electron, resulting in a highly active O 2 species [70]. This radical therefore, can attack organic molecules and adsorbed intermediates [12]. Because of its nucleophilic character it is suggested to preferentially interact with organic cationic radical substrates (S+) [21,71] forming a highly reactive peroxyl radical [25,67]. Furthermore, Chen et al. [24] have concluded that the reaction between O 2 and the organic cationic radical species may be responsible for the complete mineralization of the dichlorophenol pollutant.

Structure 6

hν, TiO2 ΔE = - 66 kJ/mol

Structure 5

hν, TiO2 ΔE = - 71 kJ/mol

O

C

H

H O=C O=C

C

H

C O

H O

Structure 8

Structure 9

Structure 10

Er = - 268 kJ/mol

Er = - 302 kJ/mol

Er = - 256 kJ/mol

1

Fig. 12. Proposed mechanistic steps for the O2-dependent ring opening process.

Despite the controversy on the importance of the role played by the holes produced in the valence band, upon illumination, evidences support the idea that the hole (h+) is a powerful oxidizing specimen responsible for the photo oxidation of many organic compounds [12,19]. The surface adsorbed organic moieties as elecþ tron donors [72], and the highly oxidative hVB ðEox ¼ 2:8 VÞ [73] are the main partners of the single electron transfer mechanism. This mechanism is also known in the literature as charge transfer complex (CTC) mechanism, in which the adsorbed organic substrate (S) þ is oxidized directly by hVB as follows: þ

S þ hVB ! Sþ

ð6Þ

Tunesi and Anderson [74] have reported that this mechanism is more efficient than the free radical attack when charge transfer occurs at low solution pH. Furthermore, Li et al. [17,18] suggested that the ring opening reactions can be induced by single electron transfer rather than hydroxyl-type chemistry. The direct oxidation of the photo generated holes for the adsorbed phenyl ring at the surface lattice oxygen ion has been modeled as the intermediate cyclohexadienyl radical cation (C6 Hþ 5 ).

102

H.S. Wahab et al. / Chemical Physics 353 (2008) 93–103

The outcome of the ROHF SCF optimization calculation for this cationic radical accompanied with a negative binding energy value and an adsorption energy of 36.3 kJ/mol. This indicates that this step is exoergic, and also is stable on TiO2 surface during photo excitation. Similar phenyl cationic radical has been observed by Wu et al. [34] in the photo irradiation of benzene on TiO2, and by Seo et al. [15], who discovered that pure polycyclic hydrocarbon arenes, which are physically adsorbed, form 1:1 and 2:1 charge transfer complexes with dry TiO2 surfaces. The facts that the steps following the formation of Sþ are debatable [23] and that the field is thus open to speculations, justify the undertaking of theoretical investigations. Cermenati et al. [71] have reported that the oppositely charged radical ions, S+ and O 2 , are expected to react at a diffusion controlled rate. The regioselectivity is expected to be determined by the spin density on the organic radical. Accordingly, relatively stabilized intermediates are formed in this way, and may include diradicals, zwitterions or dioxetane, and the latter can be cleaved directly. Pichat [20] has supported the above results, in the photocatalytic degradation study for several aromatic pollutants in water. He concluded that the plausible pathway may be the one electron transfer from the organic molecule to the photo excited TiO2 and the subsequent reaction of the resulting cation radical with the nucleophilic species, O 2 , to form the dioxetane, which further degrades. The trapping of the resulted substrate radical cation by O 2 to form the corresponding peroxyl radical was modeled for three possible sites of attack (ortho-, meta- and para-positions of the surface attached carbon). The peroxyl radical adducts were subjected to geometry optimization calculations using the parameter NVIB = 4 FULL, to obtain their heats of formation. The corresponding heats of formation were compared, in which the two positions, paraand meta- have shown identical values (38,984 kJ/mol), whereas the ortho-position presented a value of 39,594 kJ/mol for it’s heat of formation. As a result, the ortho-adduct structure was chosen for subsequent energy calculations. Table 3 shows the ET values for the reaction path seen in Fig. 13. The exothermic value for Er (887 kJ/ mol) indicates that the ring cleavage mechanism, following the path of cationic radical interaction with O 2 , is energetically possible.

Fig. 13. Proposed mechanism for the O 2 -dependent ring opening process.

The comparison among the proposed mechanisms reveals that the O 2 path exhibited the highest exothermicity relatively to the 1 O2 and O paths. This finding of the O 2 remarkable activity is in accordance with the experimental results [21,24,25,71]. However, the contributions of 1O2 and O cannot be excluded from reaction mechanism considerations, because the distribution profiles of the ring opening intermediates show that oxygen is highly needed for photocatalysis processes. 4. Conclusions The CB molecule is adsorbed on the anatase TiO2 surfaces in parallel and perpendicular modes with different stability. The perpendicular dissociative adsorption of the CB molecule via chloride ion abstraction, resulting in a phenoxy species, is energetically more favoured. The characterization of the photo oxidation intermediate products is crucial for the selection of the organic pollutants photo degradation pathway. The details of the ring opening reaction mechanism remain somewhat speculative and species other than OH radical, such as atomic oxygen, singlet oxygen and superoxide radical anion, are involved actively in the photo cleavage step. The O 2 ring opening path through the dioxetane intermediate is thermodynamically the most favourable. Acknowledgements One of the authors (H.S. Wahab) warmly thanks the Theoretical Chemistry Institute/Hannover University for giving the opportunity to work in the institute, and the Iraqi Ministry of Higher Education for the partial financial support. The constructive and useful comments of the referees are also acknowledged. References [1] C. Tai, G. Jiang, Chemosphere 59 (2005) 321. [2] S. Lathasree, A.N. Rao, B.S. Sankar, V. Sadasivam, K. Rengaraj, J. Mol. Catal. A 223 (2004) 101. [3] Y. Peng, S.Y. Xie, M. Chen, Y.Q. Feng, L.J. Yu, R.B. Huang, L.S. Zheng, J. Chromatogr., A 1016 (2003) 61. [4] P. Haglund, I. Alsberg, A. Bergman, B. Jansson, Chemosphere 16 (1987) 2441. [5] H.P. Nirmaier, E. Fischer, A. Meyer, G. Henze, J. Chromatogr., A 730 (1996) 169. [6] X. Xu, H. Zhou, P. He, D. Wang, Chemosphere 58 (2005) 1135. [7] K.V. Murthy, P.M. Patterson, M.A. Kean, J. Mol. Catal. A 225 (2005) 149. [8] H. Komulainen, Toxicology 198 (2004) 239. [9] M. Bedner, W.A. Maccrehan, G.R. Helz, Environ. Sci. Technol. 38 (2004) 1753. [10] M.R. Hoffmann, S.T. Martin, W. Choi, D.W. Bahnemann, Chem. Rev. 95 (1995) 69. [11] E. Carpio, P. Zuniga, S. Ponce, J. Solis, J. Rodriguez, W. Estrada, J. Mol. Catal. A 228 (2005) 293. [12] M.A. Fox, M.T. Dulay, Chem. Rev. 93 (1993) 341. [13] V. Subramanian, V.G. Pangarkar, A.A. Beenackers, Clean Prod. Proc. 2 (2000) 149. [14] K. Tanaka, K. Padarmpole, T. Hisanaga, Water Res. 34 (2000) 327. [15] Y.S. Seo, C. Lee, K.H. Lee, K.B. Yoon, Angew. Chem., Int. Ed. 44 (2005) 910. [16] A.R. Nicolaescu, O. Wiest, P.V. Kamat, J. Phys. Chem. A 109 (2005) 2822. [17] X. Li, J.W. Cubbage, W.S. Jenks, J. Org. Chem. 64 (1999) 8525. [18] X. Li, J.W. Cubbage, T.A. Tetzlaff, W.S. Jenks, J. Org. Chem. 64 (1999) 8509. [19] X. Li, J.W. Cubbage, W.S. Jenks, J. Photochem. Photobiol. A 143 (2001) 69. [20] P. Pichat, Water Sci. Technol. 35 (1997) 73. [21] M. Muneer, M. Qamar, D. Bahnemann, J. Mol. Catal. A 234 (2005) 151. [22] D. Johnson, M.E. Jenkin, K. Wirtz, M.M. Reviejo, Environ. Chem. 1 (2004) 150. [23] G. Ghigo, G. Tonachini, J. Am. Chem. Soc. 121 (1999) 8366. [24] C. Chen, P. Lei, H. Ji, W. Ma, J. Zhao, Environ. Sci. Technol. 38 (2004) 329. [25] Y. Wang, C.S. Hong, Water Res. 34 (2000) 2791. [26] B. Ahlswede, K. Jug, J. Comput. Chem. 20 (1999) 563. [27] B. Ahlswede, K. Jug, J. Comput. Chem. 20 (1999) 572. [28] D.N. Nanda, K. Jug, Theor. Chim. Acta 57 (1980) 95. [29] T. Bredow, G. Geudtner, K. Jug, J. Comput. Chem. 22 (2001) 861. [30] T. Homann, T. Bredow, K. Jug, Surf. Sci. 515 (2002) 205. [31] T. Bredow, K. Jug, Surf. Sci. 327 (1995) 398. [32] N. Serpone, E. Pelizzetti, Photocatalysis, Fundamentals and Applications, John Wiley & Sons, New York, 1989. [33] S. Tanaka, U.K. Saha, Water Sci. Technol. 30 (1994) 47.

H.S. Wahab et al. / Chemical Physics 353 (2008) 93–103 [34] W.C. Wu, L.f. Liao, C.F. Lien, J.L. Lin, Phys. Chem. Chem. Phys. 3 (2001) 4456. [35] F. Hatayama, T. Ohno, T. Maruoka, H. Miyata, React. Kinet. Catal. Lett. 45 (1991) 265. [36] J. Lichtenberger, M.D. Amiridis, J. Catal. 223 (2004) 296. [37] H. Tahiri, Y.A. Ichou, J.M. Herrman, J. Photochem. Photobiol. A 114 (1998) 219. [38] H. Ogawa, J. Phys. Org. Chem. 4 (2004) 346. [39] S.C. Street, Q. Guo, C. Xu, D.W. Goodman, J. Phys. Chem. 100 (1996) 17599. [40] J. Chen, W.J. Peijnenburg, X. Quan, Y. Zhao, D. Xue, F. Yang, Chemosphere 37 (1998) 1169. [41] T. Bredow, K. Jug, J. Phys. Chem. 99 (1995) 285. [42] E. Finocchia, G. Busca, V. Lorenzelli, R.J. Willey, J. Catal. 151 (1995) 204. [43] R.W. van den Brink, V. Jorg, R. Louw, P. Maggi, P. Mulder, Catal. Lett. 71 (2001) 15. [44] K. Chhor, J.F. Bocquet, C.C. Justin, Mater. Chem. Phys. 86 (2004) 123. [45] J. Peller, O. Wiest, P.V. Kamat, J. Phys. Chem. A 108 (2004) 10925. [46] Y. Song-hu, L. Xiao-hua, J. Hazard Mater. B 118 (2005) 85. [47] T.A. Dahl, in: G.R. Helz, R.G. Zepp, D.G. Crosby (Eds.), Aquatic and Surface Photochemistry, Lewis Pub. CRC Press, Ann Arbor, 1994. [48] P. Imhof, D. Krügler, R. Brause, K. Kleinermanns, J. Chem. Phys. 121 (2004) 2598. [49] C. Ratzer, J. Küpper, D. Spangenberg, M. Schmitt, Chem. Phys. Lett. 283 (2002) 153. [50] H.S. Wahab, Computational Study of the Induced Photocatalytic Degradation of the Chlorinated Aromatic Compounds on Nanocrystalline Titanium Dioxide Particles, Ph.D. Thesis, College of Science, Chemistry Department, AlMustansiriya University, Baghdad, 2006. [51] K. Jug, T. Homann, T. Bredow, J. Phys. Chem. A 108 (2004) 2966. [52] T. Homann, T. Bredow, K. Jug, Surf. Sci. 515 (2002) 205. [53] A.V. Emeline, V. Ryabchuk, N. Serpone, J. Photochem. Photobiol. A 133 (2000) 89.

[54] [55] [56] [57] [58] [59] [60] [61] [62] [63] [64] [65] [66] [67] [68] [69] [70] [71] [72] [73] [74]

103

O. Legrini, E. Oliveros, A.M. Braun, Chem. Rev. 93 (1993) 671. J.M. Nicovich, C.A. Gump, A.R. Ravishankara, J. Phys. Chem. 86 (1982) 1684. T. Ko, G.Y. Adusei, A. Fontijn, J. Phys. Chem. 95 (1991) 8745. C. Barckholtz, T.A. Barckholtz, C.M. Hadad, J. Phys. Chem. A 105 (2001) 140. L.J. Bartolotti, O.E. Edney, Chem. Phys. Lett. 245 (1995) 119. C. Walling, R.A. Johnson, J. Am. Chem. Soc. 97 (1975) 363. J. March, Advanced Organic Chemistry-Reactions, Mechanisms and Structure, fourth ed., John Wiley & Sons, New York, 1992. C.J. Chien, M.J. Charles, K.G. Sexton, H.E. Jeffries, J. Phys. Chem. B 101 (1998) 2650. M. Dilmeghani, O. Zahir, J. Environ. Qual. 30 (2001) 2062. M. Bobrowski, A. Liwo, S. Oldziej, D. Jeziorek, T. Ossowski, J. Am. Chem. Soc. 122 (2000) 8112. K. Lang, J. Mosinger, D.M. Wagnerova, Chem. Listy 99 (2005) 211. M. Garavelli, F. Bernardi, M. Olivucci, M. Robb, J. Am. Chem. Soc. 120 (1998) 10210. A.A. Abdel-Shafi, D.R. Worrall, J. Photochem. Photobiol. A 172 (2005) 170. J.M. Andino, J.N. Smith, R.C. Flagan, W.A. Goddard, J.H. Seinfeld, J. Phys. Chem. 100 (1996) 10967. N. San, A. Hatipoglu, G. Koçtürk, Z. Çinar, J. Photochem. Photobiol. A 146 (2002) 189. M. Steveson, T. Bredow, A. Gerson, Phys. Chem. Chem. Phys. 4 (2002) 358. D.E. Park, J.L. Zhang, K. Ikeue, H. Yamashita, M. Anpo, J. Catal. 185 (1999) 114. L. Cermenati, P. Pichat, C. Guillard, A. Albini, J. Phys. Chem. B 101 (1997) 2650. C. Kormann, D. Bahnemann, M.R. Hoffman, Environ. Sci. Technol. 22 (1988) 798. S. Irmak, E. Kusvuran, O. Erbatur, Appl. Catal. B 54 (2004) 85. S. Tunesi, M. Anderson, J. Phys. Chem. 95 (1991) 3399.