1. Introduction

CuAlO₂ is a typical p-type transparent oxide, which has a wide band-gap (>3eV) and room temperature photoluminescence through its UV near-band-edge emission due to the recombination of free excitons [1–4]. The delafossite structure in the class of CuMO₂ (M site accounts for trivalent ions) materials allows for chemical bond stretching from either Cu-O bond (xy-plane) or M-O bond (z-direction) and it is widely acknowledged that the electrical properties are limited by carrier compensation through local lattice relaxations [5, 6]. Substitution of the M site cations will cause the electronic density to shift towards different oxide ligands. Therefore, the introduction of divalent or trivalent dopants either as impurities or as substitutes into a layered CuAlO₂ crystal lattice may result in increased electrical conductivity [6]. As a result, reduced mass and band gap tuning usually occurs even though there is still lack of knowledge in interpretation of these doping mechanisms, as well as some inconstancy between engineered band-gaps and band-gap activation energies measured experimentally. It is also noteworthy that the ability of delafossite CuAlO₂ to emit UV near-band-edge emissions at room temperature along with its electrical properties and visible transmittance make it especially suitable for LED applications. The direct transition of carriers from valence band to conduction band and the resulting recombination of free excitons produces the luminescence in the UV range. Luminescent properties depend on the copper environment wherein an off-centered monovalent copper atom is linearly coordinated with an antiprismatic oxygen atom. The size of the M³⁺-cation leads to a stretching or relaxation of this linear Cu-O bond which consequently alters the electronic density and the photoluminescence spectrum. The idea of M³⁺ site doping with trivalent ions such as rare earth ions has been developed in recent years and the doping effects have been discussed from both morphological and electrical perspectives [7, 8]. The effects of oxygen intercalation resulting from the lattice distortions has been shown to contribute to observed variations in resistivity. Nevertheless, the effect of trivalent doping as observed through luminescent and other optical methods has not yet been reported.

In the present work, ceramic electrospun CuAlO₂ nanofibers have been prepared in their form on quartz substrates for luminescence characterizations. Near-band-edge emission at room temperature was identified and the dielectric constant could be estimated in the high frequency region. The effective mass, together with band-gap enhancement could be extracted from these transparent fiber coated samples to quantify the intercalation of yttrium affecting the host band-edge.

2. Experimental

Polyvinylpyrrolidone nanofibers have been electrospun from a precursor solution containing equal molar quantities of copper nitrate trihydrate (99-104%, Sigma-Aldrich) and aluminum nitrate nonahydrate (98.0-102.0%, Sigma-Aldrich). Yttrium nitrate hexahydrate (99.8%, Sigma-Aldirch) was added as dopant source in quantities sufficient to replace 0%, 1%, and 5% of the Al³⁺. The electrospun fibers were deposited on quartz substrate for later optical characterizations. Heat treatment of the nanofibers was then carried out at 900°C to transform the polymeric fibers into crystalline CuAlO₂ fibers. The as-spun nanofibers were held at the elevated temperatures for either 2, 4, or 5 hours in order to obtain ceramic nanofibers with different crystallite sizes (shown in Table 1).The average thickness of the nanofibers coated on quartz substrate was ~20µm. Photoluminescence studies (Horiba Spex FluoroLog Tau-3) were performed at room temperature using an excitation wavelength of 365nm. The CuAlO₂ fibers coated on quartz slide was directly measured. The slit was kept at 4/3 (ex/em) in order to normalize the spectra intensities in terms of wavelength among different samples. X-ray photoelectron spectroscopy (PHI Quantera) was used to examine the local chemical environment.

Table 1. Sample crystallite size under different dwelling time

View Table

3. Results and discussion

The fibrous fiber morphology was sustained after thermal annealing, which is shown in Figs. 1(a), 1(b) and 1(c).There was no evident difference in fiber diameter and coating thickness among samples with different doping levels. The XRD pattern (space group: R-3m) shows a prominent preferred orientation in the z-direction, which implied that the nucleation was initiated along the radial direction and the strong diffraction peaks were facilitated by the stacking fibrous layers. The Rietveld method was carried out in Bruker TOPAS and both the experimental and fitting deviation profiles were shown in Fig. 1(d).

 

Fig. 1 (a) and (b) SEM micrographs of CuAlO₂:1%Y fibers electrospun on quartz; (c) cross-section image indicating the fiber thickness ~20µm; (d) Experimental XRD pattern and fitting deviation profile. Inset shows the digital images of transparent fiber-coated quartz slides.

Download Full Size | PDF

Since the bare quartz substrate is not luminescent, the broad emission peaks shown in Fig. 2 belong to the direct transition of this wide band gap material. The PL spectra confirm that free excitonic emission can be expressed by wide band-gap oxides in which the bound excitons dominate only at low temperatures, and free excitonic emissions dominate at higher temperature (room temperature) due to strong localization of electron-hole pairs [9, 10]. Superlattice materials, in which the charge carriers are confined to two dimensional regions, are known to show a concentrated localization of electron-hole pairs. This results in enhanced electron-hole interactions which effect the optical properties of nanomaterials. In delafossite-CuAlO₂, the Cu-O bonds in the O-Cu-O dumbbell layers determine the electronic structure near the band-gap and lead to strong localization of excitons in the x-y plane as well as larger binding energy. Since the binding energy exceeded the room temperature thermal energy (kT≈0.025eV), room temperature PL emissions were presented. Figure 2(a) shows that the undoped sample exhibited a broad peak centered at 458nm. The sample doped with 1% Y³⁺ peaked at 439nm while the sample with 5% Y³⁺ content peaked at 423nm. A previous study which investigated delafossite-type oxides such as CuLaO₂ and CuYO₂ identified the source of these luminescent spectra [5]. The luminescent spectra consist of two adjacent emission bands from the transition between 4p_x,y to 3d_z²-4s, and 3d_z² + 4s to 3d_z²-4s. The 3d_z² + 4s to 3d_z²-4s transition originated from hybridization of the 3d_z² with 4s orbitals. Relaxation from the excited state causes the Cu-O bond to stretch, thereby decreasing the electronic density in the x-y plane and increasing the electronic density along the z axis. The introduction of yttrium did not modify the direct transition type. Instead, the degree to which the Cu-O bond was stretched by the introduction of the dopant resulted in the observed variations in peak location. The asymmetrical peak shape shown in the PL spectra was caused by an overlap of the two Gaussian peaks associated with the 3d_z² + 4s to 3d_z²-4s transition of both CuAlO₂ and CuYO₂. Increased yttrium content correlated with narrower emission intensity distribution for the two Gaussian peaks. As the covalency of the Cu-O bond increased from CuYO₂ to CuAlO₂, hybridization between the 3d_z² and 4s orbitals became more prolific with higher Cu-O binding energies. This lifted the non-bonding 3d_z²-4s orbital energy in order to compensate for the antibonding 3d_z² + 4s orbitals. This will increase the excitation energy from CuAlO₂ to CuYO₂. As a result, the deconvoluted CuYO₂ spectra (denoted as triangle in Fig. 2(a)) were shifted to a shorter wavelength. Additionally, the experimental binding energy for Cu2p_3/2 and Al2p (Fig. 2(b) and 2(c)) shifts to lower values at the higher dopant concentrations (932.5/74.4eV for CAO-0Y, 931.7/73.34eV for CAO-1Y and 930.9/73.21eV for CAO-5Y). This indicates a reduced binding energy around O-Cu-O and MO₆ after Y intercalation, which is consistent with the observed blueshift in PL spectra. Figure 2(d) confirms the Y doping into the lattice, and the peak intensity increases with an increase of the doping concentration.

 

Fig. 2 Photoluminescence emission spectra (a) of CuAlO₂ nanofibers annealed for 2hrs (solid line: experimental observation; dotted line: deconvolution of asymmetric peak due to yttrium intercalation). XPS spectra of Cu 2p (b), Al2p (c) and Y3d (d).

Download Full Size | PDF

The blueshift in the UV near-band-edge emission is also dependent on the size effect. This band-gap enhancement can be calculated from

where$\Delta E$represents the band-gap enhancement between bulk materials and nanostructured materials. Other variables include h, or Plank’s constant;${\mu }^{*}$, which is the reduced mass of the electrons and holes; $\rho ,$ the size of the crystallites; e, the electronic charge; ${\epsilon }_r$, the relative permittivity and ${\epsilon }_0$, the vacuum permittivity of free space. Equation (1) consists of particle-in-a-box quantum localization energy plus Coulomb energy. In order to determine the values on the right-hand-side of the Eq. (1), ${\mu }^{*}$and ${\epsilon }_r$are required for the CuAlO₂ electrospun fibers. Therefore UV-Visible transmittance and diffuse reflectance measurements were performed to estimate the value of the two terms from the transparent oxide samples. In an oxide insulator, the conducting electrons in high frequency ranges (near UV to visible) are dominant and the dielectric constant ${\epsilon }_r$ is therefore reduced to its real part. The ${\epsilon }_r$ can be calculated from the following equations:

Here, R is the reflectance data from the UV-visible diffuse reflectance measurement. Figure 3 depicts the results of the optical measurement in the range between 350 to 800nm. The upper limiting boundary value for dielectric constant could be calculated from Eq. (3), which is ~2.8 for undoped samples.

 

Fig. 3 Transmittance and reflectance measurements on the quartz substrate coated with CuAlO₂ fibers.

Download Full Size | PDF

The theoretical value of ${\mu }^{*}$ can then be estimated with Eq. (4) for the hydrogen-model binding energy (E_b) equation:

where $R^{\infty }$ is the Rydberg constant (13.6eV) and $m_0$ is the free electron mass. When considering the room temperature excitation of the thermal energy ~0.025eV as the boundary binding energy, the reduced mass, ${\mu }^{*}$, can be estimated to be 0.023$m_0$ for the undoped sample, 0.016$m_0$ for the sample with 1% Y³⁺, and 0.012$m_0$ for the sample with 5% Y³⁺. The reduced mass decreases with the increase of Y³⁺ dopant level, which is basically a cause of the decrease of the dielectric constant. Similarly, the XPS indicated an environment with reduced binding energy and a weaker localization of excitons in the two-dimensional confinement of Cu-O bonds, which could also lead to reduced mass decrease. These values can be considered to be the boundary values of the reduced mass for the purposes of Eq. (1). By substituting the boundary values of reduced mass ${\mu }^{*}$ and the dielectric constant ${\epsilon }_r$ into the Eq. (1), the theoretical band-gap enhancement as a function of crystallite size is plotted as solid lines in Fig. 4.The degree of band-gap enhancement also corresponds with the blueshift observed in the PL spectra. Further increasing the dopant level would shift the emission spectra to higher energies. Under the assumption of the quantum size effect expressed in Eq. (1), the experimental data was obtained from the differences between the bulk band-gaps and the PL spectra peak energies. The bulk band gap was measured as the optical direct band gap from the UV diffuse reflectance. The Kubelka-Munk equation was employed to calculate the bulk direct allowed band-gap for the nanofibers. The band-gap energies of the samples with 0%, 1% and 5% Y³⁺ dopant content were found to be 3.26ev, 3.42eV and 3.52eV. The increased band-gap which resulted from the introduction of yttrium is consistent with the theoretical model relating to the excitation energy to various Cu-O bond environments.

 

Fig. 4 Band-gap enhancement calculation based on theoretical estimation from UV-visible optical properties and the experimental data from both PL spectra and optical band-gap of nanofibers with different crystallite size. Inset plot shows the associated Burstein-Moss effect.

Download Full Size | PDF

It should be noted that in Fig. 4 the band-gap enhancement curves of the three samples which were projected based upon the size effect model did not match up with the experimental data points gathered from those samples at the different crystallite sizes. The undoped sample fit the theoretical curve well for crystallites below a crystallite size of 35nm. This would suggest that band-gap enhancement values could be calculated exclusively from crystallite size. However when a further blueshift was induced by the 3d⁹4p¹ to 3d¹⁰ transition in the CuYO₂ structure, the experimental data deviated from the quantum confinement estimation. We attribute this shift to the Burstein-Moss effect, which refers to the band gap widening in response to the change of carrier concentrations [11, 12]. The Burstein-Moss energy gap shift is shown in the inset of Fig. 4 and can be calculated by Eq. (5).

The carrier concentration in this equation is given by n. Assuming that CuAlO₂ and CuYO₂ have similar electron density maps [13], then the reduced mass can be the cause of the corresponding Burstein-Moss band-gap enhancement (shown in inset of Fig. 4), which exhibits a similar range of enhancement with the fitting discrepancies from quantum size model. The 4d orbitals formed by the Y³⁺ dopant at the lower edge of the conduction band caused the additional band-gap adjustments. The prominent coupling between the reduced mass and the Cu-O distance reveals that hole hopping could be enhanced augmented by the substitution of Y³⁺ into the M³⁺ site. Future work should be done to correlate the dopant concentration with the band gap enhancement.

4. Conclusion

In conclusion, the room-temperature photoluminescence of CuAlO₂ nanofibers was observed having a blueshift at elevated doping levels. A quantum size effect model was also evaluated to probe the fitting discrepancies associated with the yttrium intercalation and the binding energies of the Cu-O bond. The band-gap was further enhanced by the addition of a trivalent yttrium dopant. This caused a blue shift of the photoluminescence spectra, a phenomenon which has promising potential for the development of blue-violet photoluminescence. This investigation based on PL and optical characterization may provide useful guidance for the dopant enhanced transparent oxides with nanostructures.