Defect generation, d-d transition, and band gap reduction in Cu-doped TiO2 nanoparticles

TiO2 doped with Cu2+ initiates the formation of brookite phase along with anatase. Doping of Cu2+ introduces structural defects into TiO2. The direct evidence is the low intense and broad diffraction peaks. Raman peaks of doped TiO2 are also broad and are blueshifted. Pure TiO2 exhibits an absorption in the UV region, the position of which is shifted towards the visible region on incorporation of Cu into it. The visible absorption peaks arise due to the d-d transition of Cu2+ in the crystalline environment of TiO2. Incorporation of Cu2+ distorts the local structure of TiO2, resulting in the loss of octahedral symmetry surrounding Cu2+. The Jahn-Teller distortion splits the 2Eg and 2T2g state of Cu2+ into several d states. Interaction of light excites the electron from ground to several of the excited states and gives the visible absorption peaks in the framework of TiO2. These Cu2+d states and oxygen defects create band states, thereby favoring electronic transition to these levels and resulting in lowering of band gap of TiO2. A direct confirmation is the increase in the magnitude of Urbach energy with the reduction in the band gap of doped TiO2.


Background
TiO 2 is an important transition metal oxide material with a major application area in photocatalysis, where it shows its efficiency in the removal of unwanted impurities from water, air, etc. [1,2]. The main drawback of pure TiO 2 is its wide band gap (3 to 3.2 eV); because of which, it absorbs only the UV part of solar radiation that accounts for only 4% of the total solar radiation, leaving most of the visible portion of light [2,3]. This material can be made visible light-active either by introducing structural defects such as Ti 3+ and oxygen vacancies or by incorporation of nonmetals such as N, C, and S, and transition and rare earth ions [3][4][5][6][7][8][9][10]. These defects and dopants create sub-band states in the band gap of TiO 2 and shift the absorption edge of TiO 2 towards the visible region, thus enhancing the visible light photoactivity. Structural defects, generated on doping, can effectively tune the band structure and control the photoactivity. In Fe, Ce-codoped TiO 2 , it is reported that doping of these ions leads to the creation of grain boundary defects and results in the bending of the valence and conduction bands. These defects limit the electron mobility to the interface and prevent charge carrier recombination [11]. Cu is also an important dopant because it has high electronic conductivity, and it is cheap and highly available on the Earth's crust [12]. Cu-doped TiO 2 or CuO/TiO 2 composite serves as an efficient photocatalytic material in the decomposition of gas-phase alcohols, acid orange 88, methylene blue, etc. [13][14][15]. Cu doping can effectively reduce the wide band gap of TiO 2 by creating defects and d-band states of Cu in TiO 2 and can also act as active trap centers of electrons to reduce carrier recombination [3,16]. Other applications of Cu-doped TiO 2 are as anode materials in Li-ion batteries [12], CO sensing [17], NO and CO 2 reduction [18,19], etc. Cu-doped TiO 2 also shows activity in bacterial inactivation [20]. In this article, we have discussed the doping effect of Cu on the crystallinity of pure TiO 2 nanoparticles. We have examined how Cu extends the absorption edge of TiO 2 to the visible region and reduces the effective band gap of TiO 2 . The structural characterizations are done with X-ray diffraction, transmission electron microscopy, and Raman spectroscopy, while absorption spectra are analyzed in a UV-vis spectrometer in diffuse reflectance mode.
Corporation, St. Louis, MO, USA) to a mixture of 20 ml 2-propanol (Merck, Worli, Mumbai, India) and 10 ml of ethanol (Merck). The reaction mixture was stirred for 15 min, and then 1 ml of water was added to hydrolyze the isopropoxide chain. This was followed by dropwise addition of copper nitrate hexahydrate (Merck) solution with the requisite amount of Cu. The solution was stirred for 6 h, and after this time, the solution was transformed to gel. The gel was centrifuged in water and ethanol for five times and then dried in a vacuum oven at 80°C to get an amorphous Cu-doped TiO 2 nanopowder. The nanopowder was then annealed in air at 450°C for 4 h to obtain Cu-doped TiO 2 nanoparticles.
The X-ray diffraction (XRD) pattern of the samples was characterized using a Rigaku MiniFlex X-ray diffractometer (Rigaku, Tokyo, Japan) with CuKα radiation (λ = 1.54 Å) at a scanning angle between 20°and 70°. The high-resolution transmission electron microscope (TEM) images of the doped nanoparticles are obtained using a JEOL-JEM-2100 transmission electron microscope (JEOL Ltd., Tokyo, Japan) at an operating voltage of 200 kV. Energy dispersive X-ray (EDX) spectra of the samples were studied with a JEOL JSM (model 6390 LV) scanning electron microscope with an INCAx-Sight EDX detector (Oxford Instruments, Abingdon, Oxfordshire, UK). The Raman spectra of the samples were recorded using a Renishaw In-Via Raman spectrometer (Renishaw, Wottonunder-Edge, UK) at a resolution of 0.3 cm −1 . Electron paramagnetic resonance (EPR) spectra were obtained using a Bruker EMX 300 EPR spectrometer (Bruker BioSpin GmbH, Silberstreifen 4, Germany), Diffuse reflectance spectra (DRS) of the samples were obtained using a Shimadzu 2450 UV-vis spectrophotometer (Shimadzu Corporation, Kyoto, Japan) with BaSO 4 powder as the standard reference sample.

X-ray diffraction analysis
The diffraction pattern of pristine and Cu-doped TiO 2 nanoparticles is shown in Figure 1. The diffraction pattern corresponds to the tetragonal anatase phase of TiO 2 (JCPDS-782486). Indeed, a small fraction of brookite phase is also present in between 27°and 36°, and at 40°(JCPDS-761934). Anatase and brookite are the metastable phases of TiO 2 and usually formed at low temperature during solution-phase synthesis. Presence of brookite phase has also been observed by So et al. in TiO 2 nanoparticles [21]. They reported that the brookite phase completely disappeared at high annealing temperature. Increase in the brookite fraction with doping indicates inhibition of the formation of complete anatase phase due to incorporation of Cu on the lattice site or grain boundary. The crystallite size of the samples is calculated by using where d is the crystallite size, λ is the wavelength of X-ray radiation, β is full width at half maximum, and θ is the diffraction angle. The resultant sizes are 6, 5.5, 5, and 4 nm for pristine and 2%, 4%, and 6% Cu, respectively. The diffraction peaks are broad in all the samples. The widening of the XRD peaks indicates reduction in the grain size and increase in the fraction of the amorphous grain boundary, containing many of the structural defects. The diffraction peak intensity is lower in the doped samples than in the pristine one. The low intense diffraction peaks indicate degradation of structural quality or loss of crystallinity of TiO 2 on doping.

TEM and EDX analysis
The high-resolution transmission electron microscope images of 2%, 4%, and 6% Cu-doped TiO 2 nanoparticles are shown in Figure 2a,b,c. The particles are spherical in shape with some amount of agglomeration. The particle size distribution shows that the average numbers of particles have a size of 9, 8, and 8 nm, respectively, for 2%, 4%, and 6% Cu-doped TiO 2 . Therefore, doping does not affect the size of the nanoparticles. Figure 2d,e,f shows the EDX pattern of the 2%, 4%, and 6% Cu-doped TiO 2 nanoparticles, showing the presence of Cu, Ti, and oxygen in the nanoparticles. There is no peak of nitrogen in the EDX, indicating removal of nitrate impurities during centrifugation.

Raman spectroscopy study
Raman spectroscopy is a technique of importance for the understanding of the local structure changes on incorporation of dopant ions. The Raman spectra of our samples are shown in Figure 3. The Raman spectra of our samples correspond to the anatase phase of TiO 2 . We have not detected any secondary peaks related to Cu or its oxide phases. The intense peak at 149 cm −1 corresponds to the E g mode of anatase TiO 2 [22,23]. Other than this, two other low intense modes appear at 197 and 641 cm −1 , respectively. The B 1g mode occurs at 400 cm −1 , and the A 1g + B 1g mode appears at 530 cm −1 , respectively [23]. The significant observation in the Raman peaks of the samples is the broadness and shifting to a higher wavenumber with Cu loading. Any characteristic Raman vibration is associated with the Ti-O stretching, bending vibration [24]. The E g peak is associated with the symmetric stretching vibration of O-Ti-O in TiO 2 , the B 1g peak is due to the symmetric bending vibration of O-Ti-O, and the A 1g peak is the result of antisymmetric bending vibration of O-Ti-O [24]. The ionic size of Cu 2+ (0.73 Å) is larger than that of Ti 4+ (0.64 Å), and hence, doping of this ion will distort the lattice structure of TiO 2 ; since there is charge difference between Cu 2+ and Ti 4+ , doping of Cu generates oxygen vacancies in the lattice of TiO 2 to maintain the charge neutrality [3]. If doping occurs on the substitutional position on the Ti 4+ site, the Ti-O-Ti bond will be disturbed and a new Cu-O-Ti or Cu-O-Cu bond will be formed. Therefore, disturbance of the Ti-O-Ti bonds and the formation of new Cu-O bonds will affect the Raman-active modes and will result in the broadening and shifting of the peaks. Although Cu 2+ doping on the Ti 4+ site will affect the entire Raman-active modes, we have considered the intense E g peak to understand the doping effect. The E g peak is associated with the Cu-O-Cu stretching mode vibration, and on doping, the strength of this vibration lowers, since oxygen vacancy is formed nearby. The formation of oxygen vacancy nearby Cu is theoretically proved in the Cu-doped TiO 2 system [25]. Due to the generation of these oxygen vacancies, the lattice is contracted and the peak is shifted to a higher wavenumber. Perker and Siegel in their work reported that oxygen vacancies are responsible for the shifting and broadening of the Raman peak [26]. On the other hand, some people reported that quantum size effect has a role to play in the broadening and peak shifting [27]. From XRD and TEM, we have found that the size of the nanoparticles is in the nanoregime; therefore, phonon confinement will be prominent. Alongside, XRD also demonstrates that grain boundary defects are also generated. Therefore, both phonon confinement and structural defects may result in the shifting and broadening of the Raman E g peak. The oxygen vacancy generation  and lattice disruption of Cu-doped TiO 2 are shown in Figure 4a.

Electron paramagnetic resonance spectra analysis
Electron paramagnetic resonance (EPR) is an important tool to understand the valence state of the dopant and the coordination environment of the dopant in the framework of the host. The EPR spectra of 2% Cu-doped TiO 2 nanoparticles are shown in Figure 5. As shown in the spectra, the EPR spectra are asymmetric in shape and contain a signal corresponding to the presence of Cu 2+ in the distorted octahedral coordination of TiO 2 [15,28]. The intense peak is centered at g ⊥ = 2.08 followed by less intense quadruple signals at g II = 2.21, 2.28, 2.38, and 2.48, respectively, in the lower side of the magnetic field. The values of both g ⊥ and g II are greater than the g value of free electron g e = 2.0023. From the position of the above-mentioned g values, it can be said that Cu 2+ is coordinated in the octahedral coordination of TiO 2 and has substituted Ti 4+ on the lattice site [15]. The EPR peak is also broad which indicates the presence of dipolar interaction among neighboring Cu 2+ ions that leads to the increase in the width of the EPR peak.

Study of optical properties
The diffuse reflectance spectra of pure and doped TiO 2 are shown in Figure 6a. The method adopted to acquire the absorption spectra of powder samples in DRS mode is that of Kubelka-Munk. The equation for the Kubelka-Munk method is represented by F(R) = (1 − R) 2 /2R, where R is the reflectance and F(R) is the absorbance [29]. The F(R) curves of the samples are shown in Figure 6b.
Doped samples exhibit an absorption peak at 330 nm corresponding to the maximum absorption when electrons are excited from the valence to the conduction band. Doped samples show shifting in the absorption peak to the visible region. One absorption shifting occurs at just above 400 nm, and another big absorption hump appears in between 550 and 900 nm. In TiO 2 , the valence band (VB) is composed of O 2p states, and the conduction band (CB) is composed of Ti 3d states [30]. The 330-nm absorption peak is due to electronic transition from O 2p to Ti 3d. Doped samples contain an extended absorption edge above 400 nm and a broad absorption peak in between 550 and 900 nm. The first absorption between 400 and 500 nm appears as a result of interfacial charge transfer from the O 2p valence band to the Cu(II) state attached to TiO 2 [31][32][33][34][35][36][37]. These Cu(II) states may be present either as Cu(II) nanoclusters or in the form of amorphous oxide phase of CuO. Li et al. [15] demonstrated that Cu(II) does not show this absorption peak if it is present as CuO, but displays this absorption when it is present as Cu(II) ions attached to TiO 2 . However, Qiu et al. suggested that the absorption peak between 400 and 500 nm appears due to charge transfer from TiO 2 to CuO clusters in a system of CuO/TiO 2 nanocomposite [34]. Therefore, based on these observations, we can predict that the absorption appears due to charge transfer from O 2p to Cu(II) clusters or CuO amorphous phase. However, from EPR signals, it is found that copper is present as Cu(II) and is well placed in the octahedral coordination of TiO 2 . Moreover, the broadening of the EPR line suggests that Cu 2+ -Cu 2+ dipolar interaction is taking place and that this is possible when Cu 2+ is so closely associated possibly forming clusters. These may be nanoclusters of copper, since bulk copper clusters could have been detected in the diffraction pattern. Therefore, with the help of EPR results, we can suggest that the observed absorption band is due to charge transition from the valence band of TiO 2 to Cu(II) nanoclusters attached to TiO 2 . Apart from the absorption band between 400 and 500 nm, doped samples contain another absorption hump extending from 550 to 900 nm. The second absorption band is mainly due to the d-d transition of Cu 2+ in the crystalline environment of TiO 2 [31][32][33][34][35][36][37]. Cu 2+ has a d 9 electronic configuration, and in the pure octahedral coordination, the 2 D state of Cu 2+ is splitted into 2 E g ground state and 2 T 2g excited state with a single electronic transition. However, from the XRD and Raman spectra of the samples, it was found that doping of Cu 2+ generates oxygen vacancies which lie near Cu 2+ . Therefore, Cu 2+ is no longer in the pure octahedral (O h ) symmetry, and the symmetry is slightly distorted. A Jahn-Teller (J-T) distortion is taking place, changing the symmetry from O h to D 4h [35]. Due to the J-T effect, the ground state 2 E g is further splitted into 2 B 1g ground and 2 A g excited states, and the 2 T 2g is separated into 2 B 2g ground and 2 E g excited states [35,36]. Therefore, the possible optical transitions that may give rise to the above peaks are 2 B 1g → 2 A g , 2 B 1g → 2 B 2g , and 2 B 1g → 2 E g . In reports, the absorption at 450 and 900 nm and the broad band at 810 nm are assigned to 2 B 1g to 2 E g and 2 B 1g to 2 B 2g transitions, respectively [35][36][37][38]. This absorption peak cannot be shown by Cu + , since Cu + has completely filled 3d 10 configurations. Since 3d 10 is highly stable, photoexcitation will not release electrons, and therefore, no absorption will appear. The d-d electronic transition of Cu 2+ is shown in Figure 4b.
From the absorption spectra of the samples, it is understood that doping shifts the absorption edge of TiO 2 from the UV to visible region. Now, we can determine the effective reduction in the band gap (BG) of TiO 2 due to the incorporation of Cu 2+ ions. Figure 6c shows the band gap of all samples. For BG determination, [F(R)hυ] n is plotted against hυ. Since anatase TiO 2 is an indirect band gap semiconductor, the value of n = ½ (n = 2 for direct band gap) [29]. The line drawn on the linear part of [F(R)hυ] 1/2 vs. hυ curve at [F(R)hυ] 1/2 = 0 gives the band gap. From UV-vis spectroscopy, it is found that Cu 2+ forms sub-band states in the band gap of TiO 2 [39]. Along with Cu 2+ , oxygen defect band states are also formed in the band gap. In pure TiO 2 , the electronic transition occurs directly from VB to CB. However, on Cu doping, the electrons are not directly excited to CB since the unoccupied Cu 2+ s-d states and oxygen vacancies capture the electrons. The charged oxygen defect states formed in TiO 2 are F (two electrons), F + (single electron), and F ++ (devoid of electrons) [40]. The oxygen vacancy states that may capture electrons are F + and F ++ , respectively. Therefore, the sub-band states of Cu 2+ and oxygen defects are responsible for the reduction of effective band gap of TiO 2 nanoparticles. Sahu et al. [3] examined the red shift in the band gap of TiO 2 on Cu doping and examined that absorption edge shifting and band gap reduction are controlled by the surface of the nanoparticles, lattice strain, and vacancies.
We can determine the width of the defect bands formed as an intermediate state in the band gap of TiO 2 . These defect band states create a band tail extending from the lower of conduction band to deep down of band gap, and similarly, the defect states very near to the valence  band also smear the valence band edge deep inside the gap. Therefore, on both sides of the valence band maximum and conduction band minimum, an energy tail is formed. This defect tail is known as the Urbach tail, and the energy associated with this defect tail is referred to as Urbach energy. The equation for Urbach energy is , where α is the absorption coefficient, E is the photon energy, and E u is the Urbach energy [41,42]. The Urbach energy is calculated by plotting lnα vs. E. The reciprocal of the slopes of the linear portion, below the optical band gap, gives the value of E u . The Urbach energy of each sample is shown in Figure 7.
Absorbance coefficient α is proportional to F(R); hence, we can write ln[F(R)] vs. E. The Urbach energy of pure and 2%, 4%, and 6% Cu is 67, 259, 316, and 343 meV, respectively. Therefore, as band gap decreases, the magnitude of defect energy increases. This clearly supports our argument that sub-band states formed in between the valence and conduction bands result in the narrowing of the band gap. With doping level, the number of defect levels below the conduction band increases to such an extent that the band edge is shifted deep into the forbidden gap, thereby reducing the effective band gap of TiO 2 . The schematic of the formation of the Urbach tail and the relationship of the band gap with Urbach energy for different samples are shown in Figure 8a

Conclusion
Doping of Cu introduces structural defects and increases the fraction of brookite phase. The diffraction peaks are broad in the doped samples, indicating reduction in the grain size and increase in the amount of amorphous grain boundary. The small crystallite size and oxygen defects are responsible for the broadening and shifting of Raman peaks. Cu 2+ introduces d states in the mid-band gap of TiO 2 . The single electron in the d orbital undergoes Jahn-Teller distortion and gives rise to a broad absorption peak in the visible region. The d states of Cu 2+ and oxygen defects are responsible for the narrowing of the band gap of TiO 2 . These defects increase the magnitude of Urbach energy, and the magnitude of this energy increases as Cu loading increases.