Deposition of an oxomanganese water oxidation catalyst on TiO₂ nanoparticles: computational modeling, assembly and characterization

Abstract

Inexpensive water oxidation catalysts are needed to develop photocatalytic solar cells that mimic photosynthesis and produce fuel from sunlight and water. This paper reports the successful attachment of a dinuclear di-µ-oxo manganese water oxidation catalyst [H₂O(terpy)Mn^III(µ-O)₂ Mn^IV(terpy)H₂O](NO₃)₃ (1, terpy = 2,2′:6′2″-terpyridine) onto TiO₂ nanoparticles (NPs) via direct adsorption, or in situ synthesis. The resulting surface complexes are characterized by EPR and UV-visible spectroscopy, electrochemical measurements and computational modeling. We conclude that the mixed-valence (III,IV) state of 1 attaches to near-amorphous TiO₂ NPs by substituting one of its water ligands by the TiO₂ NP, as suggested by low-temperature (7 K) EPR data. In contrast, the analogous attachment onto well-crystallized TiO₂ NPs leads to dimerization of 1 forming Mn(IV) tetramers on the TiO₂ surface as suggested by EPR spectroscopy and electrochemical studies.

---

Broader context

In Nature, the production of O₂ by oxidizing water is catalyzed by an Mn₄Ca inorganic unit in the oxygen-evolving complex of photosystem II. Heterogeneous arrays consisting of a light harvester, a water oxidation catalyst, and an electron acceptor have been considered a promising approach for artificial photosynthesis. In this study, we deposited a di-nuclear manganese water oxidation catalyst on the surfaces of TiO₂ nanoparticles, which have shown to be efficient electron acceptors. The synthesized materials were thoroughly characterized with a variety of techniques and computational modeling. Both experimental and theoretical studies indicate the successful surface immobilization of an oxomanganese water oxidation catalyst. This approach could be utilized to build an artificial photosynthetic system, in which the oxomanganese catalyst facilitates water oxidation and the TiO₂ surface collects photogenerated electrons for solar hydrogen production.

1. Introduction

Solar energy is a vast potential source of renewable energy but we still lack cheap, efficient and robust solar cells. For large-scale applications, it is important to avoid rare elements, such as ruthenium needed in Grätzel cells. Beyond electrical energy production, it should be possible to mimic photosynthesis and produce fuel from sunlight and water by incorporating a water oxidation catalyst into a solar cell.¹ In this paper, we report the attachment of a dimanganese water oxidation catalyst on TiO₂ nanoparticles (NPs). We describe the synthesis and characterization based on computational modeling, UV-visible and electron paramagnetic resonance (EPR) spectroscopy, and electrochemical measurements.

Oxide semiconductors have been widely explored as catalysts for H₂ production, based on solar water splitting.² In particular, TiO₂-based materials have been particularly attractive as photocatalysts,³ largely because the TiO₂ band edges bracket the redox potential for water splitting.⁴ However, TiO₂ has a relatively large band-gap and its direct photoactivation requires UV light (λ < 400 nm), accounting for only less than 4% of the natural sunlight spectrum.⁵ While it is possible to extend the absorption edges of TiO₂ into the visible-light region,^5–7 multiphoton activation would still be required to accumulate the energy needed for water splitting. Several other composite semiconductors are also visible light catalysts for water splitting,^8,9 but their efficiencies are still too low. Common challenges include the recombination of the photogenerated electron-hole pairs, and the usual weak photocatalytic activity of semiconductor surface sites.

Surface functionalization of TiO₂ NPs has been extensively investigated within the context of Grätzel cells as the most successful strategy for maximizing light-harvesting with wide band-gap semiconductors.^10,11 Dye-molecules attached to TiO₂ surfaces shift the absorption threshold to the visible light region and efficiently inject electrons into the semiconductor conduction band upon visible photoexcitation. However, the photo-generated holes (h⁺) are typically localized in the oxidized adsorbate molecules with a potential that is insufficient to oxidize water. Therefore, while dye-sensitized solar cells may be efficient for light harvesting and photon-to-electricity conversion, they do not necessarily lead to solar water splitting.

Water oxidation into O₂, protons and electrons is a four electron oxidation reaction,¹²

2H₂O → O₂ + 4H⁺ + 4e⁻,

often catalyzed by sacrificial electron acceptors capable of accumulating four oxidizing equivalents (h⁺'s). A variety of water oxidation catalysts have been synthesized and many of them have been coupled with chromophores in artificial photosynthetic assemblies.^13–17 Most of these catalysts are polynuclear transition metal coordination complexes of Mn, Ru, Ir and Co. For scalability, inexpensive metals such as Mn and Co are preferred over Ru and Ir. In Nature, water oxidation is catalyzed by an oxomanganese complex and involves an oxidation mechanism that we are beginning to understand at the molecular level.^18,19 Several Mn complexes have been reported to be water oxidation catalysts,^20,21 including our own di-µ-oxo Mn dimer [H₂O(terpy)Mn^III(µ-O)₂Mn^IV(terpy)H₂O](NO₃)₃ (1, terpy = 2,2′:6′2″-terpyridine) that is an efficient catalyst for O₂ evolution^22,23 when activated with a primary oxidant (e.g., oxone).

Complex 1 is thought to be biomimetic since some of its structural and mechanistic features are similar to those of the photosynthetic reaction center.^18,24–27 While the catalytic mechanism of water oxidation based on complex 1 is not yet established,²⁶ it is thought to involve the formation of a high-valent oxyl radical species Mn^IV–O˙ species according to the following reactions:

[H₂O(terpy)Mn^III(µ-O)₂Mn^IV(terpy)H₂O]³⁺ → [H₂O(terpy)Mn^IV(µ-O)₂Mn^IV(terpy)OH]³⁺ + H⁺ + e⁻

[H₂O(terpy)Mn^IV(µ-O)₂Mn^IV(terpy)OH]³⁺ → [H₂O(terpy)Mn^IV(µ-O)₂Mn^IV–O˙(terpy)]³⁺ + H⁺ + e⁻

[H₂O(terpy)Mn^IV(µ-O)₂Mn^IV–O˙(terpy)]³⁺ + H₂O → [Mn^II⋯Mn^III] + O₂ + 2H⁺

As indicated above in the last step of the postulated mechanism, the oxyl radical is susceptible to nucleophilic attack by a water molecule, forming a hydroperoxo intermediate that rapidly decomposes into O₂ upon deprotonation. As observed in ref. 22 and 23, one of the oxygen atoms in O₂ often comes from the primary oxidant (H₂SO₅), forming O¹⁶–O¹⁸ when reacting with H₂O¹⁸, although a minority component involves O¹⁸–O¹⁸. For heterogeneous applications, it is, therefore, important to investigate first whether complex 1 can be anchored to a sacrificial electron acceptor and whether O₂ evolution can be induced by activation of the resulting surface complex with a primary oxidant.

A few studies have already investigated the catalytic and photochemical properties of oxomanganese water oxidation catalysts on solid supports. Yagi and co-workers deposited complex 1 on clay materials and saw catalytic oxygen evolution from Ce⁴⁺ as the single-electron oxidant.^28,29 They suggested that complex 1 was autoxidized to the (IV,IV) state on kaolin, which is then capable of catalyzing single-electron water oxidation.²⁸ They later found that adsorption on clay suppressed the decomposition of 1 to MnO₄⁻ and suggested that water oxidation is facilitated by the cooperation of two equivalents of 1.²⁹ Recently, Weare et al.³⁰ reported a nanostructured assembly in which a related oxomanganese complex, [(bpy)₂Mn^III(µ-O)₂Mn^IV(bpy)₂](NO₃)₃ (bpy = 2,2′-bipyridine), was coupled to single Cr^VI charge-transfer chromophore in the channels of nanoporous silica. FT-Raman and EXAFS studies demonstrated that visible light induced electron transfer from the Mn^III(µ-O)₂Mn^IV core to Cr^VI forming Mn^IV(µ-O)₂Mn^IV and Cr^V.³⁰ In this study, we build upon earlier work^31,32 and we focus on the attachment of complex 1 onto TiO₂ NPs and on the characterization of the resulting surface complexes by computational studies, UV-visible and EPR spectroscopy, and electrochemical measurements.

2. Experimental procedures

All reagents and solvents were purchased from Aldrich and used without further purification. MilliQ water was used to make all of the aqueous solutions.

2.1. Materials synthesis

2.1.1. Synthesis of [(H₂O)(terpy)Mn^III(µ-O)₂Mn^IV(terpy)(H₂O)](NO₃)₃ (1). Following the method of Chen and co-workers,³³ the precursor of complex 1, Mn^II(terpy)Cl₂, was obtained by reaction of terpy with an excess of MnCl₂. Complex 1 was then obtained at 0 °C using oxone (2KHSO₅·KHSO₄·K₂SO₄) as the oxidant.³³

2.1.2. Synthesis of TiO₂ nanoparticles. A commercially available TiO₂ sample, Degussa P25, was used as received. TiO₂ NPs were also synthesized by a modified solvothermal method.³⁴ 10 ml of Ti(IV) isopropoxide was dissolved in 200 ml of dichloromethane in a flask under dry nitrogen. Six milliliters of 0.1 M HCl solution was then added dropwise to the solution. After stirring for 90 min at room temperature, the resulting mixture was refluxed for 24 h with stirring. The resulting NPs were recovered by solvent evaporation followed by washing with copious water and drying at room temperature. Some samples of the resulting TiO₂ powder were sintered at 450 °C to improve crystallinity. The synthetic materials with and without thermal treatment at 450 °C are denoted “D450” and “D70”, respectively.

2.1.3. Deposition of 1 on TiO₂. Complex 1 was deposited on the surface of TiO₂ NPs by one of two methods. In direct adsorption, 1.2 µmol of complex 1 and 20 mg of TiO₂ NPs were mixed in water at room temperature. In the in situ synthesis, 2.4 µmol Mn^II(terpy)Cl₂ was mixed with 20 mg TiO₂ in water before adding an oxone solution at room temperature. In both methods, the resulting mixture was stirred at room temperature for 90 min. Functionalized TiO₂ NPs were washed with 1.5 ml H₂O 3 times before recovery by centrifugation at 6000 rpm. Subsequent characterization and testing indicated that the two different preparative methods led to an essentially identical product (denoted “1–TiO₂”).

2.2. Materials characterization

2.2.1. Powder X-Ray diffraction (XRD). The XRD data were recorded on a Bruker-AXS D8 Focus diffractometer (λ = 1.5418 Å, Cu Kα radiation, step time 1 s) over the range of 20° < 2θ < 60°. The step sizes were 0.05° and 0.02° for P25 and synthetic NPs, respectively.

2.2.2. EPR spectroscopy. Perpendicular-mode EPR spectra were recorded on an X-band Bruker Biospin/ELEXSYS E500 spectrometer equipped with a SHQ cavity and an Oxford ESR-900 liquid helium cryostat. All spectra were collected on water-dispersed samples in 5 mm OD quartz EPR tubes. The samples were purged with nitrogen and frozen in liquid nitrogen prior to recording spectra at 7 K. A high-intensity illuminator (Fiber-Lite Series 180, Dolan-Jenner Industries, Inc.) equipped with a fiber optic light guide, which only transmits light with wavelengths greater than 425 nm, was used as the visible light source in photochemical studies.

2.2.3. UV-visible spectra. Thin films of TiO₂ NPs were prepared on microscopic slides by doctor-blading and were used to collect spectra with a Varian Cary 3 spectrophotometer in diffuse reflectance geometry.³¹ The baseline spectrum of the microscope slide, which only absorbs at wavelengths less than 350 nm, was always subtracted.

2.2.4. Electrochemical studies. Electrochemical measurements were made on an EG&G Princeton Applied Research Model 273 potentiostat/galvanostat using a standard three-electrode system. A platinum wire was used as the counter electrode, and a standard calomel electrode (SCE) was used as the reference electrode (NHE vs. SCE: +241 mV). The working electrodes were made from fluorine-doped tin oxide conducting glasses (FTO, Hartford Glass Co., Inc.), which were coated with TiO₂ NPs by doctor-blading. The TiO₂/FTO electrodes were sintered at 450 °C for 2 h and were washed with water immediately before running the electrochemical measurements. Experiments were carried out in unbuffered aqueous solutions containing 0.1 M KNO₃ as the supporting electrolyte.

2.2.5. Elemental analysis. All samples were analyzed on a Costech ECS 4010 elemental analyzer (Costech Analytical Technologies, Valencia, CA) coupled to a ThermoFinnigan DeltaPLUS Advantage stable isotope mass spectrometer (Thermo Scientific, Boca Raton, FL), with both controlled by the Thermo's Isodat 2.0 software. In a typical run, a dried TiO₂ sample (0.5–1.0 mg) was sealed in a tin capsule and loaded into the autosampler. Control experiments were run using pure TiO₂ materials and powdered complex 1, and the results were calibrated with the house standard (standard cocoa powder, 4.15% N; 48.7% C). The output of the elemental analyzer, the N₂ and the CO₂ from N and C on TiO₂ surfaces, was sampled by the mass spectrometer.

2.3. Oxygen evolution assay

Using Ce⁴⁺ as the oxidant, oxygen evolution from aqueous solutions of 1–TiO₂ dispersed in water was measured at 298 K by an oxygen probe (Clark electrode) connected to a YSI 5300A oxygen monitor with digital readout, interfaced to a computer for data logging.³⁵ The sample chamber was maintained at 25 °C with a circulating water bath. The instrument readout was calibrated against air-saturated distilled water kept stirred in the air-tight sample chamber under the Clark electrode. In a typical run, 3 µmol 1 and 50 mg TiO₂ were mixed in water and stirred at room temperature for 90 min. The resulting 1–TiO₂ was then washed with water to remove excess 1 and was dispersed in 4 ml water in the sample chamber. After allowing time for equilibration and ensuring a constant baseline reading, 50 µL of an aqueous solution of 2 M ceric ammonium nitrate was introduced into the sample chamber.

3. Computational analysis

This section describes the computational methodology, including the preparation of structural models, the description of the Quantum Mechanics/Molecular Mechanics (QM/MM) hybrid methods, the methodology applied for simulations of UV-visible spectra, and the calculation of binding enthalpies. The results and discussion, with an emphasis on mechanistic implications, are presented in Section 4.

3.1. Molecular models

3.1.1. TiO₂ anatase nanoparticles. Precursor structural models of TiO₂ anatase NPs were obtained by DFT geometry optimization, starting with the crystal structure of bulk anatase. Model NPs, composed of complex 1 attached onto the (101) surface of anatase, were described by 32 [TiO₂] units with dimensions 1.0 × 1.5 × 3.1 nm along the [−101], [010] and [101] directions, respectively. Periodic boundary conditions were imposed with a vacuum spacer between slabs, making negligible the interaction between distinct surfaces in the infinitely periodic model system. The surface dangling bonds were saturated with capping hydrogen atoms to quench the formation of surface states and avoid unphysical low coordination numbers. The DFT geometry relaxation was performed by using the Vienna Ab-initio Simulation Package (VASP/VAMP) implementing the PW91/GGA approximation in a plane-wave basis, and using ultrasoft Vanderbilt pseudopotentials to describe the ionic interactions. A wavefunction cutoff of 400 eV was used in all calculations as well as a single Gamma point k-point sampling, due to the large size of the supercell. The Kohn–Sham (KS) Hamiltonian was projected onto a plane-wave basis set and high-efficiency iterative methods were implemented to obtain the KS eigenstates and eigenvalues. Self-consistency was accelerated by means of efficient charge density mixing schemes.

3.1.2. TiO₂ anatase nanoparticles functionalized with 1. The functionalized TiO₂ model includes complex 1 attached onto a pristine TiO₂ surface. Molecular models of 1 are based on the X-ray atomic coordinates obtained from the Cambridge Crystallographic Data Center (CCDC) with reference code FIQFIU,²² and further optimized at the DFT/B3LYP level of theory (see Fig. S1†).³⁶

The reduced models included 1 adsorbed onto 15 [TiO₂] units and capped with terminal OH fragments, where the positions of the O atoms in the capping fragments were constrained to their corresponding positions in the 32 [TiO₂] unit precursor structure. To dock the complex to the surface, we assumed that one of the two water ligands is exchanged by the TiO₂ NP, leaving the Mn dimer immobilized on the surface by a single oxo bridge and hydrogen bonding interactions. The exchanged water molecule is stabilized by hydrogen bonding to the other water molecule in the complex. The orientation of the resulting complex adsorbate (Fig. 1), relative to the surface, was obtained by energy minimization at the DFT QM/MM level of theory.

Fig. 1 Left panel: DFT/QM-QMM model of the Mn complex 1 anchored to a TiO₂–NP. A water ligand is exchanged by the NP and remains hydrogen-bonded to the other water ligand to allow for calculations of binding enthalpies (see text for details). Color scheme: C (light-blue), H (white), Mn (purple), N (blue), O (red), Ti (gray). Right panel: Schematic representation of the 1–TiO₂.

3.2. QM/MM hybrid approach

QM/MM computations are based on the two-layer ONIOM electronic-embedding (EE) link-hydrogen atom approach³⁷ as implemented in Gaussian 03.³⁸ The ONIOM-EE method is applied by partitioning the system, according to a reduced high-level molecular domain (X layer) that includes complex 1, its terpy ligands and water molecules (the model system). The rest of the system (Y layer) includes the TiO₂ surface and is described at the molecular mechanics level. The ONIOM QM/MM methodology could be efficiently applied, as in previous studies of the multiple nuclear transition metal compounds, only after obtaining high-quality initial-guess states for the cluster of Mn ions (the model system) according to ligand field theory³⁹ as implemented in Jaguar 5.5.⁴⁰ The resulting combined approach exploits important capabilities of ONIOM, including both the link-hydrogen atom scheme for efficient and flexible definitions of QM layers and the possibility of modeling open-shell systems by performing unrestricted DFT (i.e. UB3LYP) calculations.

The total energy E of the system is obtained at the ONIOM-EE level from three independent calculations as follows,

E = E^{MM,X + Y} + E^QM,X − E^MM,X

where E^{MM,X + Y} is the energy of the complete system computed at the molecular-mechanics level of theory, while E^QM,X and E^MM,X correspond to the energy of the reduced system computed at the QM and MM levels of theory, respectively. Electrostatic interactions between the layers X and Y are included in the calculation of both E^QM,red and E^MM,red at the quantum mechanical and molecular mechanical levels, respectively. Therefore, the electrostatic interactions computed at the MM level in E^MM,red and E^MM,full cancel. The resulting QM/MM evaluation of the total energy at the ONIOM-EE level includes a quantum mechanical description of polarization of the reduced system due to the electrostatic influence of the surrounding environment.

The efficiency of the QM/MM calculations is optimized by using a combination of basis sets for the QM layer, including the lacvp basis set for Mn ions in order to consider non-relativistic electron core potentials, the 6-31G(d) basis set for O in order to include polarization functions on bridging oxides, and the 6-31G basis set for the N and 3-21G basis set for C and H atoms in the QM layer. Such a choice of basis set has been validated through extensive benchmark calculations on high-valent manganese complexes.³⁶ The molecular structure beyond the QM layer is described by the UFF MM force field.⁴¹ Charges of the molecular mechanics layer were computed by using the Charge Equilibration (QEq) method for molecular dynamics simulations,⁴² which is consistent with the UFF force field.

The model TiO₂ slab, with dimensions 1.5 nm × 1.5 nm × 0.4 nm, includes 113 atoms and corresponds to the smallest possible cluster that provides a converged value of the binding affinity of the complex. Its fully relaxed configuration is obtained by energy minimization at the ONIOM-EE (UHF B3LYP/lacvp (Mn),6-31G* (O),6-31G(N),3-21G (C and H) UFF level of theory, subject to the constraint of fixed OH capping fragments. The electronic state of the surface adsorbate complex involves high-spin anti-ferromagnetic couplings between Mn centers. The resulting optimized structures are analyzed and evaluated not only on the basis of the total energy of the system but also as compared to structural, electronic, and mechanistic features that should be consistent with experimental data.

3.3. Binding enthalpies

DFT QM/MM calculations of binding energies are based on the fully optimized configurations of the structural models (a), (b) and (c) shown in Scheme 1, modeled as described in Sections. 3.1 and 3.2. The binding enthalpy ΔH = E(a) − [E(b) + E(c)] is obtained by computing the energy difference between the energy of the whole system E(a) (i.e., the energy of 1 adsorbed on TiO₂) and the sum of the energies of 1 [i.e., E(c)] and TiO₂ [i.e., E(b)]. Complex 1 adsorbed on the TiO₂ surface includes an additional water molecule, as shown in Fig. 1, that serves to conserve the total number of atoms in the QM layers of reactants and products. Upon detachment of the complex from the surface, the additional water molecule forms an OH ligand to Mn^IV and the capping H of the TiOH fragment on the attachment site of the TiO₂ surface (see Scheme 1).

Scheme 1 Binding energies are calculated as the difference between the extrapolated energy of the whole system (a), which includes the TiO₂ surface, the terpy complex 1 and a substrate water molecule; minus an OH adsorbed on the surface (b); minus the complex with an H (c).

3.4. Simulations of UV-visible spectra

The simulations of UV-visible spectra are based on electronic structure calculations of the TiO₂ anatase nanoparticles functionalized with 1, as previously reported.³¹ The calculations are based on a tight-binding model Hamiltonian of the complete supercell 1–TiO₂, gained from the Extended-Hückel (EH) semiempirical approach for electronic structure calculations. The EH method requires a relatively small number of transferable parameters and is capable of providing semiquantitative descriptions of energy bands of elemental materials (including transition metals) as well as compound bulk materials in various phases. It is applicable to large extended systems and provides valuable qualitative insight on the role of chemical bonding. The method is, therefore, most suitable to develop a clear chemical picture of the nature of electronic transitions and the interfacial electron-transfer mechanism.

The EH Hamiltonian of the bare and functionalized TiO₂ is diagonalized in the basis of AOs by solving the generalized EH eigenvalue problem, HQ^q = E_qSQ^q, where H is the EH matrix, S is the atomic orbital overlap matrix, and Q^q are the expansion coefficients of the molecular orbital (MO) with eigenvalue E_q.

The oscillator strength f of the transition with energy E_q′ − E_q is computed by using the expression , where r̂ is the electronic position operator, [small nu, Greek, macron] is the wavenumber of the transition, c is the speed of light, m is the electron mass, h is Planck's constant, and 〈r|ψ_q〉 is the wavefunction of the MO with eigenvalue E_q in the coordinate representation. The oscillator strengths are used to assign relative weights to the transitions and the total absorbance is globally scaled to facilitate the comparison with the experimental results. The stick-spectrum of discrete transitions is convoluted with a Gaussian function to match the experimental inhomogeneous broadening.

4. Results and discussion

Well-crystallized TiO₂ materials are usually preferred in photocatalytic applications, over amorphous materials, as photogenerated electron-hole pairs tend to recombine at defect sites on an amorphous TiO₂ surface.⁴³ In this study, three TiO₂ materials with different degrees of crystallinity were analyzed as substrate supports for complex 1. P25, the gold standard for photocatalysts, was synthesized by a high-temperature (greater than 1200 °C) flame hydrolysis method.⁴⁴ X-ray diffraction data confirm P25 consists of ∼85% anatase and ∼15% rutile [Fig. 2(a)]. For the synthesized TiO₂ nanoparticles without further thermal treatment, a broad (101) diffraction in its XRD pattern [Fig. 2(c)] indicates that D70 nanoparticles are mainly anatase phase with a relatively low degree of crystallinity and have small particle sizes (TEM data are provided in Fig. S2†). Upon sintering at 450 °C, the anatase nanoparticles became better crystallized, as demonstrated by the narrowed (101) diffraction in Fig. 2(b) relative to that in Fig. 2(c). The BET surface areas of D70, D450, and P25 were measured to be 310, 85, and 50 m² g⁻¹, respectively.

Fig. 2 Powder XRD patterns of (a) P25, (b) D450, and (c) D70.

4.1. Structural characterization of 1 deposited on TiO₂ nanoparticles

EPR data. EPR spectroscopy serves as a highly sensitive technique to detect the existence of complex 1 on TiO₂ surfaces. The mixed-valent complex 1 shows a characteristic 16-line EPR spectrum [Fig. 3(d)], that results from the hyperfine coupling between an S = 1/2 electron spin with two nonequivalent ⁵⁵Mn nuclei.⁴⁵ In the EPR spectra of 1 adsorbed on crystallized TiO₂ materials (P25 and D450), the 16 lines are barely resolved, with relatively low intensity [see Fig. 3(a) and 3(b)]. Line overlapping and broadening are also significant. This implies that only a small quantity of 1 in the mixed-valence (III,IV) state exists on the surfaces of crystallized P25 and D450 nanoparticles. However, an intense 16-line signature is seen in the EPR spectrum of 1–D70, deposited via either direct adsorption or in situ synthesis (Fig. 3c), indicating that complex 1 containing a Mn^III(µ-O)₂Mn^IV core was anchored on the surface of near-amorphous D70 nanoparticles. The 16-line EPR spectrum of 1–D70 did not change significantly after washing 1–D70 with H₂O or solutions of 0.1 M NaNO₃ or Ca(NO₃)₂. Thus, complex 1 was irreversibly bound to D70 NPs. A close comparison between Fig. 3(c) and 3(d) revealed that the hyperfine coupling tensors for Mn^III (A₁) and Mn^IV (A₂) are significantly smaller for 1–D70 than for those of pure 1 (see Fig. S3†). This suggests that the Mn coordination environment was changed by binding to the surfaces of D70 NPs. As will be discussed in the DFT analysis, complex 1 may be bound to the TiO₂ surface via an oxo bridge Mn–O–Ti. The decrease in the hyperfine coupling tensors for 1 bound to the surface of D70 would then be explained by the replacement of a water molecule on the Mn(IV) center by an oxo group.

Fig. 3 EPR spectra of (a) 1–P25, (b) 1–D450, and (c) 1–D70. The EPR spectrum of complex 1 in a HOAc/NaOAc buffer solution (pH 4.5) is also shown in the figure (d). A trapped electron signal and an organic radical signal are also present in spectra (a) and (b), respectively. The spectrum (c) was scaled down to 1/10 of its original magnitude to allow a better comparison.

The photochemical behavior of 1–TiO₂ materials was studied with EPR spectroscopy. For each sample, EPR spectra were taken in dark and under illumination with visible light (>425 nm). A difference spectrum was obtained by subtracting the dark spectrum from the one under illumination. As shown in Fig. 4(a), a resonance corresponding to trapped electrons in the rutile lattice was seen for 1–P25 after illumination with visible light. This signal was not seen for bare P25 under the same experimental conditions. An electron signal identical to the one shown in Fig. 3(a) was also observed for 1–P25 prepared in complete dark conditions. Thus, the electrons may be generated in the dark and transferred to the deep trapping sites in P25. One possible process is the dimerization of 1 to form a Mn(IV) tetramer (2 Mn(III,IV) dimers form 1 Mn(IV) tetramer plus 2e⁻), as will be discussed later in the text. The lack of such trapping sites in D450 or D70 can explain the absence of an electron signal in the EPR spectra in Fig. 3(b) and 3(c), which were collected under the same conditions as the one in Fig. 3(a). Upon visible light illumination, electrons may also be liberated from the surface Mn complexes or residual, free terpy ligand on P25 and transferred to the trapping sites in P25, leading to the further increase in the rutile electron signal.

Fig. 4 Light-minus-dark difference EPR spectra measured at 7 K of (a) 1–P25, (b) 1–D450, and (c) 1–D70. A resonance at g ∼ 1.97, corresponding to trapped electrons in a rutile lattice, was seen for 1–P25.

DFT analysis. As can be inferred from the EPR data, shown in Fig. 3, the Mn coordination environments in the complex 1–D70 are different from those in the isolated complex 1. This difference is likely due to the slightly different magnetic environment of the Mn center directly bound to the TiO₂ surface via an oxo bridge Mn–O–Ti. Compared to well-crystallized nanoparticles, amorphous TiO₂ materials usually contain more surface hydroxyl groups (Ti–OH) that serve as anchoring sites for complex 1, as indicated by the QM/MM minimum energy configuration of complex 1 on the surface of TiO₂ anatase (see Fig. 1). In fact, our DFT QM/MM calculations predict that ΔH_b = −54 kcal mol⁻¹ is the binding enthalpy of 1 to the OH groups on the (101) surface of anatase.

The structural and electronic analysis of our QM and DFT-QM/MM models indicate that the attachment of 1 does not significantly affect the spin density of the Mn centers (i.e., the spin densities change less than 5% upon attachment, from 3.83 to 3.88 a.u. for Mn^III and from 2.57 to 2.53 a.u. for Mn^IV). Structural rearrangements, however, are more significant. One of the terpy ligands exhibits a slight distortion upon adsorption of the complex, leading to an increase in the Mn^III–N distances. The Jahn–Teller elongation axis of Mn^III does not change upon adsorption of the complex, but the Mn–N distances increase (see Table 1, third row). Also, replacing a water molecule by an oxo group, that is a better electron donor, decreases the Mn^IV–O distance (Table 1, second row) and stabilizes the Mn^IV center.

Table 1 Comparison of Mn–ligand distances for 1 and 1–TiO₂ as described by our QM and DFT-QM/MM models

	Atom–atom distances/Å
	1	1–TiO2
Mn^III-oxo	1.83, 1.91	1.82, 1.82
Mn^IV-oxo	1.73, 1.77	1.77, 1.84, 1.82
Mn^III–N	2.06, 2.21, 2.21	2.19, 2.28, 2.41
Mn^IV–N	2.00, 2.03, 2.03	2.02, 2.06, 2.06
Mn^III–O (of water)	2.22	2.11
Mn^IV–O (of water)	2.15	—

---

UV-visible data. The optical absorption spectra of the two systems are significantly different in the UV-visible region, providing valuable information regarding the electronic structure of the oxomanganese surface complex on different TiO₂ materials. The UV-visible spectrum of 1–D70 [Fig. 5(c)] resembles that of 1 in solution (see Fig. S4†).^22,33 This is consistent with the EPR results shown in Fig. 3(c), confirming the existence of a significant amount of 1 on the surfaces of D70 nanoparticles, and is also similar to the simulated UV-visible spectrum of 1–TiO₂ shown in Fig. 6, along with the experimental spectrum of 1–D70.

Fig. 5 UV-visible spectra of (a) 1–P25, (b) 1–D450, (c) 1–D70, and (d) bare D70 NPs.

Fig. 6 Simulated UV-visible difference spectrum (solid line) compared to the experimental difference spectrum (dashed line). The difference spectrum is the spectrum of 1–D70 [Fig. 5(c)] minus the spectrum of D70 [Fig. 5(d)]. The simulated spectrum was convoluted with gaussians with an FWHM of 60 nm to facilitate comparison.

Complex 1 exhibits two prominent features in the UV-visible absorption spectrum, including one at 450 nm and the other at 650 nm, as indicated by our computational analysis and the experimental spectrum (Fig. 6 and S4†). Upon adsorption to a TiO₂ nanoparticle, the two absorption features are blue shifted to 420 nm and 610 nm, respectively. The transition at 420 nm is assigned to a ligand-to-metal-charge-transfer (LMCT) transition from the terpyridyl ligands to the manganese oxo core, while the transition at 610 nm is assigned to a terpyridyl π–π* transition. Cross transitions from 1 to the TiO₂ surface are observed in the range 400–550 nm, although with smaller intensity.

The UV-visible spectrum of 1–D450 [Fig. 5(b)] is not very informative, but an intense and broad absorption around 450 nm is observed for 1–P25 [Fig. 5(a)], indicating the possible existence of an oxidized form of 1 containing a Mn^IV(µ-O)₂Mn^IV core, a Mn tetramer featuring a Mn^IV(µ-O)₂Mn^IV–O–Mn^IV(µ-O)₂Mn^IV unit (see S5†),⁴⁶ or even solid-state compounds of Mn⁴⁺ such as MnO₂.

Elemental analysis data. The elemental analysis of dried 1–TiO₂ samples suggests that the terpy ligand was intact on the surfaces of different TiO₂ nanoparticles, since the estimated N contents based on a N/C atomic ratio of 1 : 5 are in excellent agreement with experimental results (Table 2). The results shown in Table 2 further suggest that appreciable amounts of terpy-coordinated Mn complexes may exist on crystallized P25 and D450, although the EPR results in Fig. 3 ruled out the possibility of a significant amount of 1 remaining on the surfaces of these two crystallized TiO₂ materials.

Table 2 Content of C and N (weight percentages) in different 1–TiO₂ samples

TiO2	Experimental		Calculated^a
	C	N	N	1/µmol
a Estimation based on the experimental values of C content, making the assumption that the adsorbates remain in the di-µ-oxo dinuclear form with a Mn : N : C atomic ratio of 1 : 3 : 15 when attached on the surfaces of 1–TiO2 materials.
P25	0.09	0.020	0.021	0.05
D450	0.11	0.021	0.026	0.06
D70	0.23	0.057	0.055	0.13

---

O₂ evolution data. The 1–TiO₂ samples were also tested as catalysts for water oxidation using Ce⁴⁺ as the (single-electron) primary oxidant (see Fig. 7). Significant oxygen evolution is observed for 1–P25, while mild oxygen production was seen for 1–D450 and none for 1–D70, suggesting that the Mn surface complexes are different in the three materials. The O₂ uptake observed for 1–D70 and bare P25 NPs [Fig. 7(c) and 7(d)] is attributed to O₂ uptake by the Clark-type O₂ electrode under the acidic conditions of the assay. For 1–D70, the EPR results in Fig. 3(c) show that the mixed-valence Mn(III,IV) dimer is bound to the surface in high yield. However, previous work has shown that this species is unstable in acidic Ce⁴⁺ solutions.³⁵ Therefore, we conclude that the lack of O₂ evolution activity for 1–D70 is due to decomposition of 1. On the other hand, previous work has shown that a Mn(IV)-terpy tetramer, [Mn₄O₅(H₂O)₂(terpy)₄]⁶⁺ (2) with the connectivity of Mn^IV(µ-O)₂Mn^IV–O–Mn^IV(µ-O)₂Mn^IV, is formed from 1 under acidic conditions.^46,47 The observation of O₂ evolution activity for 1–P25 and 1–D450 [Fig. 7(a) and 7(b)] could be explained if 1 dimerizes on the TiO₂ surface to form the Mn(IV)-terpy tetramer. This would also account for the low yield of the mixed-valence Mn(III,IV) dimer EPR signal in these two samples since the Mn(IV) tetramer is EPR silent.

Fig. 7 O₂ evolution using Ce⁴⁺ as a single-electron oxidant. 1 was loaded on TiO₂ (50 mg) samples: (a) P25, (b) D450, and (c) D70. A control test was also done using (d) bare P25 NP's as the catalyst.

The loading of complex 2, the Mn(IV) tetramer, onto P25 would be ∼0.025 µmol, assuming complex 1 dimerizes on P25 (see Table 2). It can be inferred from Fig. 7(a) that ∼0.04 µmol O₂ was generated after 1000 s using functionalized P25 as the catalyst. This leads to a turnover number greater than 1 for O₂ evolution, suggesting that water oxidation on functionalized P25 is catalytic. Furthermore, the initial rate of O₂ evolution is estimated to be ∼0.04 nmol s⁻¹, comparable to that for complex 1 supported on clay at a loading of ∼0.3 µmol as reported by Yagi and co-workers.²⁸

4.3. Electrochemical characterization of 1–TiO₂

To understand the redox properties of 1–P25 and characterize the electrochemistry of 1 on different TiO₂ surfaces, we have performed cyclic voltammetry. TiO₂ electrodes were prepared by using standard techniques, doctor-blading TiO₂ suspensions on FTO conducting glass slides and subsequently sintering them at 450 °C, before use as the working electrode in a standard three-electrode configuration. The requirement for sintering the samples at elevated temperatures before the electrochemical measurements precluded electrochemical study of 1–D70.

The observed redox behavior of 1 on TiO₂ electrodes is similar to the redox activity previously reported for homogeneous Mn^III(µ-O)₂Mn^IV catalysts of water oxidation.⁴⁷ Our cyclic voltammograms, shown in Fig. 8, exhibit fairly well-resolved redox behavior, including a quasi-reversible redox couple centered at approximately 1 V, corresponding to Mn^III(µ-O)₂Mn^IV ↔ Mn^IV(µ–O)₂Mn^IV (P_a¹ and P_c¹).

Fig. 8 Cyclic voltammograms using different working electrodes: (a) P25/FTO, (b) D450/FTO, and (c) bare FTO. The solution contained 0.57 mM 1 in 0.1 M KNO₃. Redox processes are labeled including: oxidation of Mn^III(µ-O)₂Mn^IV (P_a¹), reduction of Mn^IV(µ-O)₂Mn^IV (P_c¹), reduction of the Mn(IV) tetramer (P_c² and P_c^2′).

Subsequent dimerization of Mn^IV(µ-O)₂Mn^IV can occur, forming the Mn(IV)-terpy tetramer (2).⁴⁷ Using the bare FTO electrode, the reduction of this tetrameric species gives a broad cathodic wave at ca. 850 mV (P_c²). This cathodic current corresponding to reduction of the Mn tetramer was not observed on the P25/FTO electrode and was barely noticeable on the D450/FTO electrode [see Fig. 8(a) and 8(b)]. However, a large cathodic current is seen at ca. 700 mV for both P25/FTO and D450/FTO, suggesting that the Mn tetramer can be stabilized on the surfaces of crystallized NPs (P25 and D450). Thus, we conclude that the reduction of the (IV,IV,IV,IV) species to the (III,IV,IV,IV) species occurs at a more reducing potential, indicating surface stabilization of the tetramer.

These results are consistent with the observation of O₂ evolution for 1–P25 and 1–D450 (see Fig. 7). The acidic conditions of Ce⁴⁺ solutions are known to give O₂ evolution for the Mn(IV) tetramer species 2.⁴⁸ Furthermore, the cooperative interaction of two equivalents of 1 has been previously suggested to contribute to the catalytic activity when using 1–mica as the catalyst.²⁹ A similar interaction is likely to be present in the current system, where dimeric oxomanganese catalysts on the surface of the nanoparticles additionally form tetrameric species.

5. Conclusions

We have assembled an oxomanganese water oxidation catalyst (complex 1) on the surfaces of near-amorphous TiO₂ nanoparticles. The 1–TiO₂ hybrid assemblies were characterized with a variety of techniques including computational modeling, EPR and UV-visible spectroscopies and electrochemistry. Quantum mechanics calculations suggest that 1 is anchored on the TiO₂ surface via an oxo bridge (Mn–O–Ti), formed upon substitution of a water ligand by the surface. The mixed-valence Mn(III,IV) state of the complex 1 is not the predominant form of the surface adsorbate complex for well-crystallized TiO₂ nanoparticles, probably due to formation of a Mn(IV) tetramer. Using Ce⁴⁺ as a primary oxidant, oxygen evolution was observed for 1–P25. However, the chemically-driven water oxidation was not observed for 1 in the mixed-valent form on near-amorphous TiO₂.

Our EPR measurements suggest that the formation/stabilization of a Mn(IV) tetramer from the mixed-valence Mn(III,IV) species 1 may be facilitated by irreversible electron injection into TiO₂. However, further work is required to fully characterize the resulting surface adsorbates. Work in progress includes X-ray studies and immobilization of complex 1 by covalently attaching the complex to the TiO₂ surface by using molecular linkers. The utilization of the injected electrons, trapped in 1–P25, and the activation of the surface catalysts with visible light will be reported elsewhere.

Acknowledgements

The authors acknowledge support from the Chemical Sciences, Geosciences, and Biosciences Division, Office of Basic Energy Sciences, Office of Science, U.S. Department of Energy (DE-FG02-07ER15909) and DOE supercomputer time from NERSC. NSF grant CHE-0215926 provided funds to purchase the ELEXSYS E500 EPR spectrometer and the NSF ECCS # 0404191 grant supported preliminary work. G.L. thanks Drs. Ping-yu Chen, Clyde Cady, Yunlong Gao, Siddhartha Das, Gerald Olack, Barry Piekos, Professor Charles Schmuttenmaer, Professor Gary Haller, Xiaoming Wang, and Gözde Ulaş for their assistance in various aspects of experiments and their helpful discussions.

References

1. A. Fujishima and K. Honda, Nature, 1972, 238, 37–38.
2. T. Bak, J. Nowotny, M. Rekas and C. C. Sorrell, Int. J. Hydrogen Energy, 2002, 27, 991–1022.
3. K. Rajeshwar, J. Appl. Electrochem., 2007, 37, 765–787.
4. A. L. Linsebigler, G. Q. Lu and J. T. Yates, Chem. Rev., 1995, 95, 735–758.
5. O. Carp, C. L. Huisman and A. Reller, Prog. Solid State Chem., 2004, 32, 33–177.
6. R. Asahi, T. Morikawa, T. Ohwaki, K. Aoki and Y. Taga, Science, 2001, 293, 269–271.
7. X. Chen and S. S. Mao, Chem. Rev., 2007, 107, 2891–2959.
8. K. Maeda, K. Teramura, D. L. Lu, T. Takata, N. Saito, Y. Inoue and K. Domen, Nature, 2006, 440, 295–295.
9. Z. G. Zou, J. H. Ye, K. Sayama and H. Arakawa, Nature, 2001, 414, 625–627.
10. B. O'Regan and M. Gratzel, Nature, 1991, 353, 737–740.
11. M. Gratzel, Nature, 2001, 414, 338–344.
12. F. Liu, J. J. Concepcion, J. W. Jurss, T. Cardolaccia, J. L. Templeton and T. J. Meyer, Inorg. Chem., 2008, 47, 1727–1752.
13. W. Ruttinger and G. C. Dismukes, Chem. Rev., 1997, 97, 1–24.
14. N. D. Morris and T. E. Mallouk, J. Am. Chem. Soc., 2002, 124, 11114–11121.
15. J. H. Alstrum-Acevedo, M. K. Brennaman and T. J. Meyer, Inorg. Chem., 2005, 44, 6802–6827.
16. W. Lubitz, E. J. Reijerse and J. Messinger, Energy Environ. Sci., 2008, 1, 15–31.
17. M. Borgstrom, N. Shaikh, O. Johansson, M. F. Anderlund, S. Styring, B. Akermark, A. Magnuson and L. Hammarstrom, J. Am. Chem. Soc., 2005, 127, 17504–17515.
18. E. M. Sproviero, J. A. Gascon, J. P. McEvoy, G. W. Brudvig and V. S. Batista, J. Am. Chem. Soc., 2008, 130, 3428–3442.
19. E. M. Sproviero, J. A. Gascon, J. P. McEvoy, G. W. Brudvig and V. S. Batista, J. Am. Chem. Soc., 2008, 130, 6728–6730.
20. A. K. Poulsen, A. Rompel and C. J. McKenzie, Angew. Chem., Int. Ed., 2005, 44, 6916–6920.
21. R. Brimblecombe, G. F. Swiegers, G. C. Dismukes and L. Spiccia, Angew. Chem., Int. Ed., 2008, 47, 7335–7338.
22. J. Limburg, J. S. Vrettos, L. M. Liable-Sands, A. L. Rheingold, R. H. Crabtree and G. W. Brudvig, Science, 1999, 283, 1524–1527.
23. J. Limburg, J. S. Vrettos, H. Y. Chen, J. C. de Paula, R. H. Crabtree and G. W. Brudvig, J. Am. Chem. Soc., 2001, 123, 423–430.
24. V. K. Yachandra, K. Sauer and M. P. Klein, Chem. Rev., 1996, 96, 2927–2950.
25. J. P. McEvoy and G. W. Brudvig, Chem. Rev., 2006, 106, 4455–4483.
26. R. Tagore, R. H. Crabtree and G. W. Brudvig, Inorg. Chem., 2008, 47, 1815–1823.
27. C. S. Mullins and V. L. Pecoraro, Coord. Chem. Rev., 2008, 252, 416–443.
28. M. Yagi and K. Narita, J. Am. Chem. Soc., 2004, 126, 8084–8085.
29. K. Narita, T. Kuwabara, K. Sone, K.-i. Shimizu and M. Yagi, J. Phys. Chem. B, 2006, 110, 23107–23114.
30. W. W. Weare, Y. Pushkar, V. K. Yachandra and H. Frei, J. Am. Chem. Soc., 2008, 130, 11355–11363.
31. S. G. Abuabara, C. W. Cady, J. B. Baxter, C. A. Schmuttenmaer, R. H. Crabtree, G. W. Brudvig and V. S. Batista, J. Phys. Chem. C, 2007, 111, 11982–11990.
32. W. R. McNamara, R. C. I. Snoeberger, G. Li, J. M. Schleicher, C. W. Cady, M. Poyatos, C. A. Schmuttenmaer, R. H. Crabtree, G. W. Brudvig and V. S. Batista, J. Am. Chem. Soc., 2008, 130, 14329–14338.
33. H. Y. Chen, R. Tagore, S. Das, C. Incarvito, J. W. Faller, R. H. Crabtree and G. W. Brudvig, Inorg. Chem., 2005, 44, 7661–7670.
34. G. Li and K. A. Gray, Chem. Mater., 2007, 19, 1143–1146.
35. R. Tagore, H. Chen, H. Zhang, R. H. Crabtree and G. W. Brudvig, Inorg. Chim. Acta, 2007, 360, 2983–2989.
36. E. M. Sproviero, J. A. Gascon, J. P. McEvoy, G. W. Brudvig and V. S. Batista, J. Inorg. Biochem., 2006, 100, 786–800.
37. S. Dapprich, I. Komaroni, K. S. Byun, K. Morokuma and M. J. Frisch, THEOCHEM, 1999, 461, 1–21.
38. M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, J. A. Montgomery Jr, T. Vreven, K. N. Kudin, J. C. Burant, J. M. Millam, S. S. Iyengar, J. Tomasi, V. Barone, B. Mennucci, M. Cossi, G. Scalmani, N. Rega, G. A. Petersson, H. Nakatsuji, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, M. Klene, X. Li, J. E. Knox, H. P. Hratchian, J. B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, P. Y. Ayala, K. Morokuma, G. A. Voth, P. Salvador, J. J. Dannenberg, V. G. Zakrzewski, S. Dapprich, A. D. Daniels, M. C. Strain, O. Farkas, D. K. Malick, A. D. Rabuck, K. Raghavachari, J. B. Foresman, J. V. Ortiz, Q. Cui, A. G. Baboul, S. Clifford, J. Cioslowski, B. B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaromi, R. L. Martin, D. J. Fox, T. Keith, M. A. Al-Laham, C. Y. Peng, A. Nanayakkara, M. Challacombe, P. M. W. Gill, B. Johnson, W. Chen, M. W. Wong, C. Gonzalez, J. A. Pople, Gaussian, Inc., Wallingford, CT, 2004.
39. (a) G. Vacek, J. K. Perry and J. M. Langlois, Chem. Phys. Lett., 1999, 310, 189–194; (b) E. M. Sproviero, J. A. Gascón, J. P. McEvoy, G. W. Brudvig and V. S. Batista, J. Chem. Theory Comput., 2006, 2, 1119–1134.
40. Jaguar 5.5; L. Schroedinger, Portland, OR, 2003.
41. A. K. Rappe, C. J. Casewit, K. S. Colwell, W. A. I. Goddard and W. M. Skiff, J. Am. Chem. Soc., 1992, 114, 10024–10035.
42. A. K. Rappe and W. A. I. Goddard, J. Phys. Chem., 1991, 95, 3358–3363.
43. G. Li and K. A. Gray, Chem. Phys., 2007, 339, 173–187.
44. A. Mills and S. LeHunte, J. Photochem. Photobiol., A, 1997, 108, 1–35.
45. S. R. Cooper, G. C. Dismukes, M. P. Klein and M. Calvin, J. Am. Chem. Soc., 1978, 100, 7248–7252.
46. H. Y. Chen, J. W. Faller, R. H. Crabtree and G. W. Brudvig, J. Am. Chem. Soc., 2004, 126, 7345–7349.
47. C. Baffert, S. Romain, A. Richardot, J. Lepretre, B. Lefebvre, A. Deronzier and M. Collomb, J. Am. Chem. Soc., 2005, 127, 13694–13704.
48. Y. Gao, C. W. Cady, R. H. Crabtree, G. W. Brudvig, manuscript in preparation.

---

Footnote

† Electronic supplementary information (ESI) available: Optimized structure of complex 1, TEM images of TiO₂ NPs, EPR spectra of 1-D70 and isolated 1, and UV-visible spectra of complexes 1 and 2. See DOI: 10.1039/b818708h

---