Transparent conducting oxide nanotubes

Transparent conducting oxides (TCOs) combine electrical conductivity and optical transparency that plays an important role in modern optoelectronic applications [1, 2]. Moreover, the high chemical stability of oxides, ability to functionalize the surface easily, large surface areas for hollow nanotubes, and optimal alignment of energy states for charge extraction, makes TCO nanotubes an excellent platform for nanostructured light absorption/emission and catalytic applications. Recently, several groups have been developing Indium-tin-oxide (ITO) nanowires for nanostructured light emitting diodes and photovoltaic applications [3–5]. However, optimal band alignment of titanium dioxide (TiO₂) for charge injection/extraction with a wide range of semiconductors, and application as a photoelectrochemical, photocatalytic, and photovoltaic material [6–8] makes it an ideal candidate for TCO applications, for easy integration in wide variety of nanostructured devices. Moreover, due to it's low cost, abundance (application of ITO as TCO thin films [2] has resulted in scarcity of natural indium supply), high chemical stability, and low toxicity of anatase TiO₂, it can be an important TCO nanotube material [6–10]. Here, we demonstrate for the first time, TiO₂ TCO nanotubes using shallow Niobium (Nb) dopant, fabricated using a modified electrochemical anodization process. We report TCO nanotubes with high charge carrier concentrations (>3 × 10²¹ cm⁻³), large conductivity (1.54 × 10³ Scm⁻¹), high transmittance (up to 90% between 400–1000 nm) and good field-emission properties.

We chose Nb as the shallow electron dopant for anatase TiO₂ because of it's low ionization energy (4–30 meV [11–15]), high solubility [13], and similar ionic radius as Ti⁴⁺ (r(Ti⁴⁺) = 0.605 Å versus r(Nb⁵⁺) = 0.64 Å) [16]. Nb acts as a substitutional dopant for Ti and donates excess electron to the conduction band, forming strong hybridized 4d orbitals with the 3d orbitals of Ti [17]. Using epitaxial thin films of Nb-doped anatase TiO₂ (formed on SrTiO₃ or LaAlO₃ single crystalline substrates by pulsed laser deposition), it was recently shown that low electrical resistivity (2.1 × 10⁻⁴ Ωcm) and high internal transmittance (>90%, in the visible region) can be achieved [2, 13]. However, heavy doping of nanoscale structures has proven more challenging, due to large surface area, reduced dimensions for ejection of dopants, and high diffusivities [18–20]. A number of methods like post-growth ion implantation, thermal annealing, or using metal alloys as initial anodization material, have been tried for doping of TiO₂ nanotubes [21]. However, none of these methods has been able to achieve heavy doping levels (>10²¹ cm⁻³) required for TCO performance. Our new doping method using modified electrochemical anodization process, allows facile fabrication of undoped and doped widebandgap oxide nanotubes, with varying doping levels (from light doping to heavily doped nanotubes, >10²¹ cm⁻³).

TiO₂ nanotube growth was performed using anodization method [22–29]. Electrolyte consisted of solvent (ethylene glycol) with 1% ammonium fluoride (NH₄F), and 2% of water. 1'' × 1'' sized Titanium (Ti) sheets were for anodization and investigation by different characterization techniques. For the anodization process, Ti sheet and platinum (Pt) electrode were immersed in the electrolyte 2–3 cm apart. The growth rate of nanotubes in ethylene glycol was approximately ∼4 μm h^-1. All as-grown samples were amorphous and they were annealed at 500 °C for 1 h in air to convert to anatase phase. Doping to form TCO nanotubes was performed using modified electrochemical cell, with two sections separated by a porous membrane, and using alternating current cycles. The new modified electrochemical cell with Pt electrode is filled with electrolyte (described above) free of dopant precursor. The second section with Ti anodization sheet is filled with electrolyte that contains dopant precursor. The porous membrane is semi-permeable to the electrolyte, and serves as the physical barrier between two sections of the electrochemical cell (to prevent mixing), while still allowing flow of electrolyte to complete the cell.

For doping the TCO nanotubes with Nb, NbCl₅ was added as a dopant precursor in the electrolyte. We found a linear relationship between the amount of precursor added (NbCl₅ wt%) in electrolyte and the detected Nb dopant concentration in TCO nanotubes. For comparison, the undoped sample was also grown and studied along with doped samples. Anodization voltage, which determined the nanotube diameter [24, 27], was set to 30 V during growth for all samples. This was done to ensure similar nanotube diameter, thickness and other morphological parameters constant, except the dopant concentration in TCO nanotubes. This allowed us to study the effect of doping on TCO nanotubes properties. The nanotube diameter using this anodization process (30 V applied voltage) was ∼80 nm (figure 1(a)). The growth time, which determines the nanotube height, was also kept constant in subsequent measurements to 30 min; and the corresponding nanotube carpet height was ∼2 μm (figure 1(b)).

Zoom In Zoom Out Reset image size

The morphology of the grown TCO nanotubes was examined in field-emission scanning electron microscope FE-SEM JEOL 7401F. Compositional analysis were performed by energy dispersive x-ray spectroscopy (EDS) embedded in the same FE-SEM instrument. The EDS analysis were performed in different modes (2D mapping, point scan, etc) to test for uniformity and possible secondary phases (Niobium oxide Nb₂O₅, metallic Nb, etc). Quantification of Nb concentration in TCO nanotubes performed using INCA software (Oxford Instruments) integrated with FE-SEM, which used measured EDS spectra and a calibration table, to calculate the percentage of constituent elements. For Nb dopant, we utilized the Lα radiation (2.166 KeV) generated by the scanning electron beam. The crystal structure of doped and undoped TiO₂ nanotube samples were analyzed by x-ray diffraction (XRD) measurements using a Scintag XDS 2000 x-ray diffractometer. The measurements were performed using Cu Kα radiation at 45 kV and 40 mA.

I–V characterization was performed by two different ways: (1) from single nanotubes using current-sensing atomic force microscopy (CS-AFM) and (2) ensemble measurements using Keithley source meter (Keithley 2612A, Tektronix). CS-AFM I–V was used for resistivity measurements, while ensemble measurements were used for temperature dependent current–voltage characterization (I–V–T). The setup configuration for ensemble measurements was similar to that for CS-AFM (TCO nanotubes sandwiched between metal contacts). In CS-AFM measurements, ∼6 nm contact radius with a nanoscaled tip was used, while in case of ensemble measurements the contact radius of the metal probe was ∼100 μm.

I–V–T measurements were performed in temperature range 20 K–300 K, with 10 K temperatures steps, using a closed loop Helium Cryostat (ARS-202AE with ARS-2HW Helium compressor, Advanced Research Systems). The voltage used for these measurements was varied in the range −20 V to +20 V. Sheet resistance of nanotubes R was calculated from I–V measurements using Ohm's law $I=\frac{V}{R}$. Then, resistivity $\rho$ was calculated using equation $R=\rho \frac{L}{A}$, where L is nanotube length, and A—total nanotube contact cross-sectional area.

Reflection measurements were performed in the wavelength range 400–1000 nm using a home-built confocal optical microscope. The confocal setup used an inverted microscope (AXIO-Observer, Zeiss), series of confocal pinholes, a 400 μm fiber optical cable, and the collected UV-Vis absorption spectrum was measured on an Ocean Optics USB4000 spectrometer with resolution of ∼0.1 nm.

Scanning Tunneling Microscope (STM) characterization was done using a customized Molecular Imaging PicoScan 2500 setup (with PicoSPM II controller). An STM nosecone (N9533A series, Agilent Technologies) was used for scanning and spectroscopy using chemically etched Pt–Ir tips (80:20). The measurements were done at room temperature under atmospheric conditions. Tunneling junction parameters were set at tunneling currents ranging between 100 and 500 pA and sample bias voltage between −5 and +5 V. The tunneling current as a function of applied bias voltage (STS) was recorded at multiple positions on the sample. A modified Molecular Imaging PicoSPM setup was used for current sensing AFM (CS-AFM) measurements. The CS-AFM tips used were coated in-house using thermal evaporator with 5 nm of 99.99% Cr and 15 nm of 99.99% Au. Multiple current scans were taken at different bias voltage ranging from −5 to +5 V.

X-ray photoelectron spectra (XPS) of TiO₂ samples were obtained using a PHI 5600 x-ray photoelectron spectrometer. The nanotube powder samples were collected from the TiO₂ nanotube sample by scraping. Monochromatic Al Kα x-rays (1486.6 eV) were used for the XPS analysis. The pass energy was 93.9 eV and the step size was 0.400 eV. An electron beam neutralizer was employed at 17.8 mA. Data was collected with Auger Scan (RBD Enterprises, Bend, OR). XPS data was analyzed in CASA XPS (Casa Software, UK). Quantitative measurement of TiO₂ samples was performed with an ARL 3410+ inductively coupled optical emission spectrometer (ICP-OES).

Morphological characterization and elemental mapping of TCO nanotubes reveal uniform incorporation of Nb dopant in the TiO₂ nanotube matrix. Scanning electron micrographs (SEM) (figures 1(a), (b)) showed (a) nanotube diameter (top view), and (b) cross-section SEM images of TCO nanotubes grown during experiments. The hollow tubular structure of nanotubes is well preserved after doping (method described above). The nanotube diameter and height of all samples studied in this work was ∼80 nm and 2 μm, respectively (to eliminate geometrical effects). The anodization voltage used for their growth was 30 V and the growth time was 30 min. The wall thickness was ∼10–15 nm. All samples had anatase phase, that was confirmed from XRD patterns (figure 1(c)) and XPS spectrum (figure 1(d)). XRD pattern showed reflection peaks corresponding to anatase phase with predominant peaks (101) and (200) at 2θ = 25.3° and at 47.95°, respectively. XPS spectrum revealed anatase O1 s and Ti2p^3/2 peaks. The XRD pattern of doped samples did not change with doping level up to 11%, ruling out phase segregation and formation of other secondary phases even at high doping levels. In figure 2(a), XRD patterns for Nb doped TiO₂ nanotubes with doping level up to 11%, shows only reflection peaks corresponding to anatase TiO₂ with no peaks corresponding to any secondary phase (possibly due to formation of unintentional niobium oxide (Nb₂O₅) or other metallic Nb crystallites).

Zoom In Zoom Out Reset image size

EDS analysis confirmed presence of doped elements in TiO₂ nanotubes as shown in figures 2(b), (c). TCO nanotube samples were prepared with different amount of NbCl₅ in the electrolyte: 0.1%, 0.4%, 0.8%, and 1.2% shown in figures 2(b), (c). The corresponding Nb concentration in TCO nanotubes, quantified using EDS, were 1.37%, 3.8%, 7.8%, and 10.55%, respectively, presenting a linear relationship between the amount of incorporated dopants in TiO₂ nanotubes and the NbCl₅ precursor concentration in the electrolyte (figure 2(c)). The conductivity of Nb-doped TiO₂ nanotubes increased (resistivity decreased) with wt% NbCl₅ precursor (see discussion below), further confirming the incorporation of electronic Nb dopant into TiO₂ crystal. To further confirm uniform incorporation of Nb dopant along each nanotube, EDS point scans were taken from different locations on a individual Nb doped TiO₂ nanotube. The point scan was performed by focusing an electron beam onto the wall of single nanotube, and recording the corresponding EDS spectra. The results showed uniform distribution of Nb over the entire length of the nanotube (figure 3(a)).

Zoom In Zoom Out Reset image size

To further test the change in electronic density of states (DOS) of TiO₂ nanotubes due to incorporation of Nb dopants, and measure the resulting donor level created in the electronic spectra, we conducted STS analysis of the doped samples. STM characterization confirmed the presence of Nb in TiO₂ nanotubes and revealed new uniform shallow DOS near the conduction band of Nb-doped TiO₂ nanotubes (figure 3(b)). The Fermi level in TCO nanotubes is very close to the bottom of conduction band, indicating n-type conductivity (figure 3(b)). Comparison between the DOS for undoped TiO₂ and TCO nanotubes (figure 3(b)) also shows that Nb is likely a shallow donor, which does not produce any additional states within the bandgap of TiO₂ semiconductor (maintains transparency). Therefore, STS analysis confirmed incorporation of electronically active shallow Nb dopant, change in electronic DOS of doped TiO₂ nanotubes, and likely transparency of the doped TCO nanotubes.

Electronic properties of single TCO nanotubes were studied using CS-AFM. I–V characterization of single nanotubes was conducted, using a sharp conductive (gold-coated) AFM tip to make electrical contact with single hollow vertically aligned TiO₂ nanotubes and the open area of Ti sheet as the second contact to form complete circuit. Figure 4(a) illustrates the experimental configuration used for the CS-AFM measurements. Using I–V characteristics for Nb doped and undoped TiO₂ nanotubes, the resistivity, ρ, of individual nanotubes was calculated (using R = ΔV/ΔI, and $R=\frac{L\rho }{S}$, where L is length of nanotubes, S is contact area of the tip). The measured resistivity for 1% Nb doped nanotubes was found to be 2.3 × 10⁻³ Ωcm, which is two orders of magnitude smaller than the resistivity of undoped nanotubes 6 × 10⁻¹ Ωcm. The resistivity of the doped samples decreased with an increase of Nb concentration, and approached 6.5 × 10⁻⁴ Ωcm for 10% Nb doped TiO₂ nanotubes (figure 4(b)). The increase of conductivity (1/ρ) of Nb-doped TCO nanotubes was expected due to ionization of Nb dopants (activation energy 4–30 meV [11–15]) which populate the bottom of the conduction band and increase the charge carrier concentration (n).

Zoom In Zoom Out Reset image size

To further characterize the carrier concentration of the samples, we used CS-AFM I–V in the intermediate bias regime, where the reverse-biased Schottky barrier dominates the total current [30, 31]:

$$\begin{eqnarray}&&{\rm ln} (I)={\rm ln} (S)+e{\mkern 1mu} \left( \frac{1}{kT}-\frac{1}{{{E}_{0}}} \right)V+{\rm ln} {{J}_{s}},\end{eqnarray} \tag{ 1 }$$

where J_S is the current density through the Schottky barrier, S is the contact area associated with this barrier, E₀ is a parameter depending on carrier concentration n, e is electron charge, k is the Boltzmann constant, and J_s is a function of applied bias [30, 31]. The slope of ln(I) versus V is equal to $e{\mkern 1mu} (\frac{1}{kT}-\frac{1}{{{E}_{0}}})$, where ${{E}_{0}}={{E}_{00}}{\rm coth} ({{E}_{00}}/kT)$, and ${{E}_{00}}=(\hbar e/2){{(n/m*\varepsilon )}^{1/2}}$, h is the Planck's constant, ε₀ is vacuum permittivity, ε = 31 ε₀ is the dielectric constant for TiO₂, and m* = m₀ electron effective mass. This logarithmic plot of the current I as a function of the applied bias V gives approximately a straight line (figure 4(c)), the slope of which is equal to $e{\mkern 1mu} (\frac{1}{kT}-\frac{1}{{{E}_{0}}})$. From the slope, carrier (electron) concentration n can be calculated using expression for E₀₀. The carrier concentration for 1% Nb doped TiO₂ sample was 5 × 10¹⁹cm^-3 (figure 4(d)), which is about two orders of magnitude higher than the electron concentration 9 × 10¹⁷ cm⁻³ in nominally undoped sample (figure 2(c)). The carrier concentration, n, increased with Nb doping and reached 3.4 × 10²¹ cm⁻³ for 10.6% Nb concentration in TCO nanotubes. Using the n_e value of ∼3.4 × 10²¹ cm⁻³, we estimated that from the doped Nb atoms, ∼90% are electronically active and contribute an additional charge carrier to the electronic states of TCO nanotubes.

To understand the high degree of ionization of Nb dopant, we measured the temperature dependent I–V characteristics (I–V–T) of our TCO nanotubes. Using a Richardson plot ln(I/T²) versus 1/kT, we obtained the Nb donor ionization energy (E_a ) as 12 meV (from the high temperature region, figure 5(a)). Similar activation energy (18 meV) was also obtained from the slope of ln(σ) versus 1/kT plot, where σ is conductance I/V (figure 5(b)). This activation energy is very close to reported values of Nb donors in TiO₂, which ranges from 4 to 30 meV [12–15]. In order to rule out ionization due to application of bias (or hopping of charges between Nb centers), we also analyzed the activation energy as a function of applied voltage (from Richardson plot, figure 5(c)). The activation energy was found to be constant with applied bias, and hence occurs spontaneously due to available energy at room temperature. Therefore, the low activation energy (obtained from two separate methods [32]) and clear evidence of uniform Nb incorporation (lack of phase segregation or secondary phases) can explain the high degree of incorporation of charge carriers from doped Nb atoms in TCO nanotubes.

Zoom In Zoom Out Reset image size

Addition of Nb dopant in TCO nanotubes results in ionization of charge carriers and resulting increase in conductivity (figure 4(b)). To understand the charge transport in these nanotubes, we measured the change in resistivity (ρ = 1/σ) with temperature T (from 20 to 300 K) for an ensemble of TCO nanotubes. The ρ − T data showed two different dependences, pointing to a change in transport behavior. The resistivity of TCO nanotubes with low Nb doping (shown for 1 wt% Nb, figure 6(a)) decreased with temperature, indicating this sample showed typical semiconductor transport. However, for TCO nanotubes with 10 wt% Nb the resistivity monotonically increases with temperature (dρ/dT > 0) showing 'metal-like' behavior (or metallic conductivity). The resistivity for these TCO nanotubes decreased from 6.5 × 10⁻⁴ Ωcm at 300 K to 2.2 × 10⁻⁴ Ωcm at 20 K. This switch from semiconductor to metallic conduction indicates that charge transport in heavily-doped samples is limited mainly by electron–phonon or electron–electron scattering, which decreases at low temperatures leading to a decrease in TCO resistivity [11, 33]. At high temperatures (100–300 K) the ρ − T dependence is almost linear, similar to metals, and can be described by $\rho ={{\rho }_{0}}(1+\alpha \Delta T)$, where ρ₀ is the residual resistivity (∼1.76 × 10⁻⁵ Ωcm), α is the temperature coefficient (1.15 × 10⁻¹ K⁻¹ here), and ΔT is the temperature difference. The carrier concentration n was almost independent of temperature (shows a slight decrease, ∼10%, when the temperature is decreased from 300 K to 20 K, figure 6(b)). Using n, we also extracted the electron mobility μ using the relationship μ = 1/(neρ), where ρ is the resistivity of the TCO nanotubes. For highly doped samples (10 wt% Nb), while the resistivity decreases with temperature, the carrier mobility increases by a factor 3.5 within the same temperature range (from 2.9 to 10.3 cm² V−1s⁻¹, when temperature changes from 300 K to 20 K, figure 6(b)). This behavior is typical of semiconductor transport, whereas lower conduction electron scattering in heavily-doped TCO nanotubes causes the switch to metallic conductivity (since the charge carrier concentration stays almost the same due to shallow Nb donors, figure 6(b)).

Zoom In Zoom Out Reset image size

Besides the electrical properties, TCO materials are desirable for their light transmission. Optical transparency of metallic Nb doped nanotubes was evaluated using reflectivity measurements. The transmittance η was obtained from reflectance measurements using the following relation:

$$\begin{eqnarray}&&\eta =\left( 1-\frac{{{R}_{b}}-{{R}_{d}}}{{{R}_{s}}-{{R}_{d}}} \right)\times 100\%,\end{eqnarray} \tag{ 2 }$$

where R_b is measured signals from blank Ti sheet, R_s is the signal from the sample, and R_d is the background signal. The results, shown in figure 6(c), clearly demonstrate that η is above 90% for most of the visible wavelengths for TCO nanotubes. These results also compare well with ITO thin films which requires conductivities ∼10⁴ Scm⁻¹, while retaining >90% transparency, for TCO applications [2]. While nanotube films in this work were grown on non-transparent opaque Ti sheet, the grown nanotubes can be easily detached from Ti sheet and transferred onto transparent substrates as was shown in previous studies [34–36]. In addition, by deposition of Ti metal on a transparent substrate (e.g. glass) and subsequent anodization of metallic Ti films, completely transparent samples can be obtained [37, 38].

Zoom In Zoom Out Reset image size

Due to their high chemical stability and excellent charge conduction properties, we evaluated the field emission properties for these TCO nanotubes. The field emission measurements were performed in a vacuum chamber with a base pressure of 6.6 × 10⁻⁵ Pa, in a voltage range of 0–1000 V. The bias is applied to a copper grid placed at a distance 150 μm from the TCO nanotube surface [39, 40]. We found that Nb doped TCO nanotubes exhibited significantly higher field emission current, compared to undoped samples. Figure 7 compares field emission I–V characteristics of TCO nanotubes doped with 10% Nb (curve 1), 1% Nb (curve 2), and undoped sample (curve 3). The inset shows corresponding Fowler–Nordheim plots for the respective samples. Almost linear behavior of ln(I/E²) versus 1/E shows predominant field emission nature of the emitted electrons. Also, the field emission current increases with Nb doping (22 mA for 10 wt% Nb concentration compared to 1.8 mA for undoped tubes) due to decrease in field emission threshold and increase in electron carrier density. This observation of high field emission current from 10% Nb doped TiO₂ nanotubes (22 mA) is unprecedented for oxide nanotubes, and can be improved further by optimization of nanotube density (due to reduction of electric field screening effects) [40].

Recently several groups have tried doping TiO₂ nanotubes with Nb [28, 29]. Niobium doping of TiO₂ nanotubes using a fluoroniobate complex [28] during the anodization process simultaneously provides a source of the doping element and fluoride anions required for nanotube formation. While the negatively charged fluoroniobate complex allows insertion of Nb in the TiO₂ nanotube structure (positive bias), this method requires synthesis of negative charged complexes soluble in the electrolyte. This limits the flexibility of incorporating desired shallow metal dopants, as not every metallic dopant can have negative complex soluble in the electrolyte (usual single atom metal ion precursors are positively charged in electrolyte and repelled from positively biased Ti sheet). In contrast, our method works both with positively and negatively charged ions, and therefore is more easily applicable for formation of different TCO nanotubes. Moreover, while anodizing metallic Ti-Nb alloys [29] allows Nb doped TiO₂ nanotubes. the Nb doping was low (up to 0.5 at% N), limited by the Nb portion in Ti–Nb alloy (TCO applications require higher ∼10 wt% doping). Since the alloying method requires special preparation of Ti-metal alloys to grow nanotubes of given dopant concentration, the doping amount is limited by the amount of metal alloy that can be incorporated and requires special preparation of metal–alloy sheets, compared to facile incorporation of dopant metal precursor incorporation demonstrated here.

In conclusion, we demonstrate TCO nanotubes by doping TiO₂ nanotubes with shallow Nb donors. Nanotubes doped with 10 wt% Nb shows metal-like behavior, with resistivity decreasing from 6.5 × 10⁻⁴ Ωcm at T = 300 K (compared to 6.5 × 10⁻¹ Ωcm for nominally undoped nanotubes) to 2.2 × 10⁻⁴ Ωcm at T = 20 K. Optical properties, studied by reflectance measurements, showed light transmittance up to 90%, within wavelength range 400 nm–1000 nm range. Nb doping also improves the field emission properties of TCO nanotubes (reduction of field emission threshold and increase in emitter current). Therefore, these results can have important consequences for fabrication of porous membranes, made of hollow TCO nanotubes, for a variety of nanostructured photovoltaic, photodetector, photoelectrochemical and photocatalytic devices.

This work was supported by start-up funds from University of Colorado, and NSF CAREER Award (CBET-1351281).