pccp

View Article Online View Journal

PCCP Accepted Manuscript

This article can be cited before page numbers have been issued, to do this please use: R. G. Freitas, F. W. Lucas, M. A. Santanna, R. Mendes, A. J. Terezo, G. Souza, L. H. Mascaro and E. C. Pereira, Phys. Chem. Chem. Phys., 2016, DOI: 10.1039/C6CP03319A.

This is an Accepted Manuscript, which has been through the Royal Society of Chemistry peer review process and has been accepted for publication. Accepted Manuscripts are published online shortly after acceptance, before technical editing, formatting and proof reading. Using this free service, authors can make their results available to the community, in citable form, before we publish the edited article. We will replace this Accepted Manuscript with the edited and formatted Advance Article as soon as it is available. You can find more information about Accepted Manuscripts in the Information for Authors. Please note that technical editing may introduce minor changes to the text and/or graphics, which may alter content. The journal’s standard Terms & Conditions and the Ethical guidelines still apply. In no event shall the Royal Society of Chemistry be held responsible for any errors or omissions in this Accepted Manuscript or any consequences arising from the use of any information it contains.

www.rsc.org/pccp

Page 1 of 32

Physical Chemistry Chemical Physics View Article Online

DOI: 10.1039/C6CP03319A

An experimental and theoretical study on the electronic and structural properties

R. G. Freitasa,b, F. W. S. Lucasb, M. A. Santannab, R. A. Mendesa, A. J. Terezoa, G. L. C. de Souzaa,c, L. H. Mascarob, E. C. Pereirab*

a

Department of Chemistry, Federal University of Mato Grosso, Laboratório

Computacional de Materiais, 78060-900, Cuiaba, MT, Brazil b

Department of Chemistry,

Federal

University of

São Carlos,

Laboratório

Interdisciplinar de Eletroquímica e Cerâmica, P. O. Box 676, 13560-970 São Carlos, SP, Brazil c

Institute of Exact Sciences and Technology, Federal University of Amazonas, CEP

69100-000, Itacoatiara, AM, Brazil

* Corresponding author: Tel.: +55 16 3351 9309 E-mail address: [email protected]

Physical Chemistry Chemical Physics Accepted Manuscript

Published on 02 September 2016. Downloaded by Cornell University Library on 06/09/2016 03:52:02.

of CdSe@TiO2 nanotubes arrays

Physical Chemistry Chemical Physics

Page 2 of 32 View Article Online

DOI: 10.1039/C6CP03319A

In this work, the effects of the structural (crystallite size, stress) and electronic


parameters (band gap, lifetime) on the photoelectrocatalysis and electron transport over CdSe electrodeposited inside TiO2-nanotubes (CdSe@TiO2NT) were investigated. Density functional theory (DFT) calculations of TiO2 were used to elucidate the electronic band structure and to correlate with experimental values. CdSe was grown by pulsed electrodeposition into previous and late thermal-treated TiO2NT (Sample-PTT and Sample-LTT, respectively) TiO2NT without blocking the nanotube’s entrance. The Rietveld refinement method was used to obtain information from crystallographic data of each photoelectrode. The lattice strains calculated from Rietveld analysis were 0.472 and 0.540, and the average volume of the TiO2-anatase unit cell increased from 133.235(0) Å3 to 136.950(6) Å3 for Sample-PTT and Sample-LTT, respectively. The Sample-PTT exhibited higher experimental electron lifetime, larger than 1.0 order of magnitude compared to Sample-LTT photoanodes. The band structures and DOS obtained by computational modelling showed theoretical band gap values of 2.54eV and 2.75 eV, which were close to the experimental values. All studies evidenced the strong dependence of the morphology on the CdSe@TiO2 electronic properties, and, consequently, on its photoelectrochemical activity in water splitting.

Keywords: TiO2 nanotubes, CdSe, electron lifetime, band structure, density of states


Abstract

Page 3 of 32


DOI: 10.1039/C6CP03319A

Researchers all over the world are interested in hydrogen gas as the most


promising energy resource to overcome the energy crisis. In this sense, since Fujima and Honda1 reported the photoelectrochemical water splitting using TiO2 electrode under sunlight, the photocatalytic decomposition of water has been regarded as one of the potential approaches to overcome this society need. The key to a hydrogen economy is to find a pathway to efficiently and inexpensively generate hydrogen from a renewable source, that is, without the use of fossil fuel. To reach the purpose, a photoelectrochemical (PEC) system that combines the harvesting of solar energy with water electrolysis is able to generate hydrogen and oxygen, which can be used in different electrochemical devices to be converted into energy. There are several semiconductors studied as active materials in PEC systems for water splitting. However, the major challenge for solar production of hydrogen lies in finding a material with the following properties: i) chemical stability, ii) narrow band gap and iii) appropriated band edge positions. Pure TiO2 is not efficient enough for this process when used under visible light, due to its wide band gap (around 3.2 eV), which requires an excitation wavelength of less than 388 nm. Therefore, extensive efforts have been made to enhance its photoactivity. In this sense, TiO2 nanotubes (TiO2NTs) emerge as a promising class of semiconductors able to combine the above-mentioned desired features. Moreover, compared to traditional TiO2 nanocrystals-based photoelectrodes, the tubular structure can facilitate separation of the photoexcited charges, leading to higher charge collection efficiencies2. In addition, any shift in the optical response from UV to the visible range spectrum will have positive effect on the photocatalytic efficiency of TiO23. Semiconductor quantum dots such as CdSe, which has tunable band gaps, offer new opportunities for harvesting light energy in the visible and near infrared regions of solar light4,5. Published papers have demonstrated the ability of chemically and electrochemically deposited CdS and CdSe nanostructures on TiO26,7,8 to generate photocurrent under visible light radiation. Different semiconductors such as CdS9, PbS10, Bi2S311, CdSe12 and InP13, which absorb light in the visible spectrum, can also to be used as sensitizers, as they are able to transfer electrons to large band gap semiconductors such as TiO2. For example, literature data have shown that the charge injection from excited CdSe nanocrystals into


1 - Introduction



DOI: 10.1039/C6CP03319A

TiO2 can be tuned by controlling CdSe particle size14. Decreasing the particle diameter charge injection rate. Other possibility to tuning the charge injection is to fill the


TiO2NT´s tube with CdSe using a multiple-pulse electrochemical technique15. This method is promising due to its low cost, and its ability to coat large areas with good control of the materials’ quality15. Tong

et

al.16

studied

the

photoelectrochemical

properties

of

CdSe

electrodeposited inside TiO2NT (CdSe@TiO2NT). The electrodeposition procedure was carried out using cyclic voltammetry with Cd2+ and Se2- ions in the solution. It was possible to observe CdSe deposition into TiO2NT, which, in this case, blocked the nanotubes’ entrance. Takahashi et al.3 also studied CdSe deposition into TiO2NT by chemical bath deposition (CBD). Although the authors observed enhanced photoelecrochemical properties, the CBD technique did not yield homogeneous CdSe deposition in the TiO2NT once the nanotubes were sealed the tube´s mouth and has a non homogeneous distribution over the surface. It is well known that Ti anodization in a solution containing F- ions leads to the formation of TiO2nanotubes, TiO2NT, array17. At the end of anodization, TiO2NT are amorphous. The electronic and structural properties of the TiO2NT, where the CdSe will be electrodeposited, play a key role in the photoelectrochemical activity of the binary compound, and, to the best of our knowledge, no paper in the literature has described either the effect of CdSe filling crystalline or amorphous TiO2NT, or the electronic band structure of such materials. Considering the above described information, the aim of this work is to investigate how structural (crystallite size, stress) and electronic (band gap, lifetime) parameters affect photoelectrocatalysis and electron transport over CdSe@TiO2NT. Furthermore, a theoretical calculation using density functional theory (DFT) was carried out to elucidate the band structure and correlate the theoretical band gap with experimental values.


of CdSe from 7.5 to 2.8 nm resulted in anincrease of 3 orders of magnitude in the

Page 5 of 32


DOI: 10.1039/C6CP03319A

2 - Experimental


TiO2NT arrays were prepared by anodizing titanium foils (99.8% Alfa Aesar) with an exposed area of 1.0 cm2 in an electrolyte composed of ethylene glycol (Synth; 85 vol. %), water purified by reverse osmosis (10 vol. %), and NH4F (Sigma Aldrich; 5 vol. %). Prior to each anodization, Ti samples were ultrasonicated in acetone, rinsed with distilled water, and dried in an N2 stream. The process was performed at 20 oC using a standard two-electrode cell with the Ti foil as anode and a platinum foil (4.0 cm2) as cathode. The temperature was kept constant using a thermostatically controlled water bath (Quimis-Q214S). The reaction was conducted using a two-step procedure: i) the voltage was swept from the open circuit potential until 20V using a sweep rate of 0.1 V s-1, and ii) maintained at 20V for 120 minutes. Both steps were carried out using a Keithley-2410 power source. After the anodization, the samples were rinsed with distilled water and thermally treated at 450oC for 120 minutes to eliminate any water, improve its mechanical stability, and to obtain the TiO2-anatase crystalline structure phase. The samples were subsequently cooled at a rate of 5 oC min-1. Electrodeposition of CdSe details into TiO2NT can be found in Supplementary Information (SI).

2.2 - Electrode Characterization

The TiO2NTs morphological characterization was carried out using a Supra 35 Zeiss Field Emission Scanning Electron Microscope. The optical absorption spectra were obtained in the diffuse reflectance mode using a UV-Vis spectrometer (Cary 5G), UV-Vis DRS. The X-ray diffraction (XRD) patterns were obtained using a Siemens diffractometer model D-5000 with CuKα radiation (λ=1.5406 Å). To obtain the microstructural data of the TiO2NT electrodes, the Rietveld refinement18 was performed using the General Structure Analysis System (GSAS) program19 suite with the EXPGUI interface20. The original Rietveld formulation and many of its successors21 treat the diffraction line width as a smooth function of the d-spacing of the diffraction angle (2θ),


2.1 - Electrode Preparation



DOI: 10.1039/C6CP03319A

whereas many peaks of interest near 2θ have very different widths. Hence, in this work,

data. In this method, it was considered that the diffraction peaks widths are not a smooth


function of d, meaning that they might arise from anisotropic broadening of the crystallite size or from a particular pattern of defects (e.g. stacking faults). Finally, the bi-dimensional model for the crystallite size described by Larson &von Dreele19 was used to account for the crystallite anisotropy. The electron density maps were calculated following the method described in detail previously23. In brief, a point (x, y, z) of the crystallite cell with volume (V) was calculated by the Fourier series using the structural factors F(h, k, l):

ρ ( x, y , z ) = V −1 ∑ ∑∑ F (h, k , l ) exp[2π i (hx, ky , lz )] h

k

(1)

l

where (x, y, z) represents a vector (r) of real space, with one vector space (a, b, c) and another vector (h, k, l), which represent the coordinates of one vector from the reciprocal space with base (a*, b*, c*), i.e. they are the coordinates of the diffraction plane given by Bragg’s Law. The electron density distribution in the base plane slice 1 with Z=0 (i.e. face ab), with a set (hkl) as the projection plane (001) were calculated using the GSAS program19 from XRD data for both TiO2NT samples. The lifetime experiments were performed in a three-electrode configuration using an Ag/AgCl/Cl-(sat.

KCl)

as reference electrode, a counter electrode of platinum

wire, and CdSe filling TiO2NTs electrodes as the photoanode. The measurements were performed in 0.5M Na2SO4 solution and illuminated with a xenon lamp irradiation (Newport Oriel Instruments 66881 QTH model). All experiments were conducted at 25oC. The working electrodes are henceforth referred to as: i)Sample-PTT: in this case, TiO2NT arrays were prepared by anodization and thermally treated. After the calcination procedure, CdSe was electrodeposited inside previously thermally treated (PTT) TiO2NTs. A second working electrode is referred as ii)Sample-LTT: in this case, TiO2NT arrays were prepared by anodization, and soon after, CdSe was electrodeposited inside TiO2NTs. Finally, the CdSe@TiO2NT photoelectrode was later thermally treated (LTT). In the case of Sample-LTT, the CdSe@TiO2NT photoelectrodes were sealed in a glass tube in vacuum conditions, and thermal treatment


the peak profile function developed by Stephens et al.22 was used to fit the experimental

Page 7 of 32


DOI: 10.1039/C6CP03319A

was carried out. This procedure prevents the Cd and Se from converting to their oxide

2.3-Computational Methods

Based on the results of TiO2-anatase CIF (crystallographic information file) and the results obtained from Rietveld refinement, the band structure and density of states (DOS) were calculated. Crystallographic data, without any relaxation and neutral charge, were used in order to compute the previously mentioned properties. All calculation were performed using density functional theory (DFT)24,25 within the plane wave-pseudopotential framework as implemented in the QuantumESPRESSO (QE)26 package within the generalized gradient approximation (GGA), which is well reported in the literature as an exchange-correlation functional suitable for TiO2 crystalline structure and solid state systems27,28,29,30,31,32,33,34,35,36. Electron ion interaction were described by Perdew-Burke-Ernzerhof (PBE) exchange-correlation functional37 with Vanderbilt ultra soft pseudopotential. The ultra soft pseudopotential includes explicitly electrons from Ti 3s, 3p, 3d and 4s and O 2s, 2p shells in the calculations. We perform the optimization convergence test of the system, and planewave basis set cutoffs for the smooth part of the wave functions was 500 Ry. Test calculation revealed that increasing cutoff energy did not affect the results. Electronic structure calculations were performed using a 4x4x2 Monkhorst-Pack k-points grid to sample the Brillouin zone, which are sufficient to provide satisfactory accuracy of the DFT calculations. Test calculation also revealed that increasing k-points did not affect the results. Full optimization of the (a, b, c) lattice parameters as well as the (x, y, z) internal coordinates was conducted, and any significant difference compared to atomic positions obtained from Rietveld refinement was achieved. It should be emphasized that the displacements, which generate the theoretical models represented in this work, were based on previous experimental results which occurred on the bulk of our sample. The theoretical simulations do not reflect the exact reality of experimental samples but do provide an interesting scheme which sheds light on the effects of structural deformation on the electronic structure. All molecular modeling were performed in an Intel® Core™ i7-6950X Extreme Edition, 32GB DDR4 2133MHz workstation under Ubuntu 14.04.1 LTS OS.



species.



DOI: 10.1039/C6CP03319A

In order to obtain more information about the morphology, FESEM images were


measured for both CdSe@TiO2NT photoelectrodes,and they are shown in Figure 1.The average tube diameter reached was near 44±5nm and 42±8nm for Sample-PTT and Sample-LTT photoelectrodes, respectively. Although Figure 1a presents a typical morphology related to TiO2NT obtained at 20 V38, the CdSe electrodeposition into amorphous TiO2NTs changes its well-ordered morphology, as observed in Figure 1b.

Insert Figure 1

Several authors have reported the CdSe deposition into TiO2NT using different techniques39,40,41. However, all of them obtained CdSe blocking the TiO2NT entrance. From a PEC point of view, such morphology decreases the efficiency of photoelectrodes, since TiO2NT filled electrolyte should increase the electrochemical surface area. Nevertheless, using multiple potential steps as the electrochemical synthesis procedure, it is possible to deposit CdSe into TiO2NT to create CdSe@TiO2photoelectrodes without blocking tube´s entrance. In order to confirm the presence of CdSe in the TiO2NT, energy-dispersive X-ray spectroscopy (EDX) was carried out. As observed in Figures 1c and 1d, it is possible to confirm the deposition of CdSe@TiO2NT photoelectrodes. The band gap of the CdSe@TiO2NT was calculated from UV-Vis DRS data in order to investigate the optical absorption properties. The optical band gap should obey the following Wood-Tauc equation42: (αhv)s = hv - Eg

(2)

where α is the absorption coefficient, and hv and Eg are photon energy and optical band gap energy, respectively. The value used for s was 2.0 since TiO2 was classified as an indirect gap semiconductor, as described further herein using DFT. Eg values were then determined by extrapolation of the linear portion of the (αhv)s curve versus the photon energy (hv) to (αhv)s= 0. For TiO2, the electronic properties of the relevant energy


3 - Results and discussion

Page 9 of 32


DOI: 10.1039/C6CP03319A

levels that form the band edges, and thus define the band gap, are considered to be the

representative of the conduction band (CB) edge, whereas the full O2p states,


particularly the nonbonding pπ states, define the valence band (VB). Both anatase and rutile show this general distribution of states43,44. Additionally, the linear dependence of (αhv)2.0 on hv indicates that TiO2 is essentially an indirect-transition-type semiconductor. The straight-line portion of the curve, when extrapolated to zero, gives the optical band gap (Eg).

Insert Figure 2

It is possible to notice that the band gap values obtained were 2.43 and 2.70 eV for Sample-PTT and Sample-LTT, respectively. Therefore, the electrodeposition of CdSe inside TiO2NT previously thermally treated led to a decrease in band gap values. It could be related to vertically ordered arrays obtained for such TiO2NTs, since the bandgap is morphology dependent. In Figure 1, it is possible to observe that the morphology of Sample-PTT is uniform (the nanotubes are more well-ordered) than that of the Sample-LTT, which can lead to higher transport electron efficiency45. In addition, TiO2 has an indirect optical band gap for anatase (3.2 eV) and for rutile (3.0 eV) phases. The amorphous material is reported to have a mobility gap of about 3.2-3.5 eV44. Then, band gap values calculated for Sample-PTT and Sample-LTT were intermediate values between TiO244 and CdSe (2.0eV)46 ones. Based on these band bap values, morphology and the EDX data presented in Figure 1, it is possible to propose that the engineering design of CdSe@TiO2NT were obtained. According to Grimes et al.47, a slight blue shift in the band gap value for Sample-PTT and Sample-LTT might be due to a quantization effect in the nanotubular structure47. The degree of lattice distortion is likely to be relatively high for nanotube array films, thus causing aggregation of vacancies acting as trap states along the nanotube´s walls, leading to a decrease in the band-to-band transition energy. Structural parameters such as lattice stress and crystallite size can be calculated using X-ray diffraction. From Figure 3, which presents XRD data, it was concluded that Sample-PTT and Sample-LTT present TiO2-anatase and CdSe-wurtzite crystallography phase. It was characterized by the diffraction peaks at 2θ = 25.281 (101), 2θ = 36.946


Ti3d states and the O2p levels. The lowest empty energy levels are Tidxy, which are



DOI: 10.1039/C6CP03319A

(103), 2θ = 37.800 (004), 2θ = 48.049 (200) (JCPDS 21-1272) for TiO2-anatase and

88-2346) for CdSe-wurtzite crystallography phases, respectively. The TiO2NT self-


ordered film was very thin, since titanium peaks, rather than TiO2-anatase peaks, were observed (JCPDS 44-1294) at higher intensities. Also, small contributions of the TiO2rutile crystalline phase at 2θ = 27.446 (110) and 2θ = 36.085 (101) (JCPDS 21-1276) were observed.

Insert Figure 3

The results obtained by the Rietveld refinement method indicated good agreement between the observed XRD patterns and the theoretical results. Moreover, the difference between the experimentally observed XRD profile patterns and the theoretically calculated data was close to zero in the scale of intensity, as illustrated by a line (Yobserved – Ycalculated). The 2D XRD multiplots (residuals) for Sample-PTT and Sample-LTT areindicatedin Figures 3a and 3b, respectively. The Rietveld method is a least-squares refinement procedure, where the experimental step-scanned values are adapted to the calculated values. The quality of the structural refinement data is acceptable when the weighted index (Rwp) is Rwp