Nanoantenna Structure with Mid-Infrared Plasmonic Niobium-Doped Titanium Oxide

Among conductive oxide materials, niobium doped titanium dioxide has recently emerged as a stimulating and promising contestant for numerous applications. With carrier concentration tunability, high thermal stability, mechanical and environmental robustness, this is a material-of-choice for infrared plasmonics, which can substitute indium tin oxide (ITO). In this report, to illustrate great advantages of this material, we describe successful fabrication and characterization of niobium doped titanium oxide nanoantenna arrays aiming at surface-enhanced infrared absorption spectroscopy. The niobium doped titanium oxide film was deposited with co-sputtering method. Then the nanopatterned arrays were prepared by electron beam lithography combined with plasma etching and oxygen plasma ashing processes. The relative transmittance of the nanostrip and nanodisk antenna arrays was evaluated with Fourier transform infrared spectroscopy. Polarization dependence of surface plasmon resonances on incident light was examined confirming good agreements with calculations. Simulated spectra also present red-shift as length, width or diameter of the nanostructures increase, as predicted by classical antenna theory.

In this report, we demonstrate the use of TiO 2 :Nb in nanostrip and nanodisk antenna arrays for infrared plasmonic devices such as surface-enhanced infrared absorption spectroscopy (SEIRA). The nanopatterned arrays were fabricated using electron beam lithography combined with plasma etching and oxygen plasma ashing processes. The fabricated nanostrip and nanodisk antenna structures of TiO 2 :Nb were sequentially subjected to linearly polarized infrared light in relative transmittance measurement with Fourier transform infrared spectroscopy. Resulted plasmon resonances exposed polarization dependence and shifted to longer wavelengths as the length, width of nanostrips, or diameter of nanodisks expanded. Experiment and simulation results are in good concordance.

Materials and Methods
Fabrication process of niobium-doped titanium dioxide nanoantenna is demonstrated generally in Figure 1.
In this report, we demonstrate the use of TiO2:Nb in nanostrip and nanodisk antenna arrays for infrared plasmonic devices such as surface-enhanced infrared absorption spectroscopy (SEIRA). The nanopatterned arrays were fabricated using electron beam lithography combined with plasma etching and oxygen plasma ashing processes. The fabricated nanostrip and nanodisk antenna structures of TiO2:Nb were sequentially subjected to linearly polarized infrared light in relative transmittance measurement with Fourier transform infrared spectroscopy. Resulted plasmon resonances exposed polarization dependence and shifted to longer wavelengths as the length, width of nanostrips, or diameter of nanodisks expanded. Experiment and simulation results are in good concordance.

Materials and Methods
Fabrication process of niobium-doped titanium dioxide nanoantenna is demonstrated generally in Figure 1. Fabrication process of Nb-doped TiO2 nanoantenna (from left to right) consists of electron beam lithography, development, dry etching with SF6 gas, and oxygen plasma ashing. The electron resist layer (brown) was coated on TiO2:Nb thin film (blue), which was deposited on silicon substrate (grey). The remained electron resist layer (green) after lithography process helped to define the nanostrip and nanodisk of antenna structures.
At first, the Nb-doped TiO2 film was co-sputtered at room temperature with TiO2 ceramic and Nb metallic targets. Both silicon (001) and borosilicate glass were used simultaneously as sputtering substrates in every deposition. Radio frequency (RF) and direct current (DC) sputtering methods were used for TiO2 and Nb targets, respectively. The sputtering chamber (!-Miller, Shibaura Mechatronics Corporation, Yokohama, Japan) was initially evacuated down to 4.0 × 10 −5 Pa as base pressure, then 19 sccm of argon and 1 sccm of oxygen gas flow was introduced into the chamber to create working pressure of about 0.3 Pa. Sputtering power was set at 200 W (RF) and 20 W (DC) for TiO2 and Nb targets, respectively. After one hour of deposition, 80-nm-thick film was obtained. Then, the as-deposited films went through vacuum thermal annealing at 600 °C for 1 h. Optical properties of TiO2:Nb thin films were characterized with Ellipsometry (SENTECH, SE 850 DUV and SENDIRA, SENTECH Instruments GmbH, Berlin, Germany) from deep ultraviolet to far infrared region.
Consequently, the samples were spin coated with negative resist NBE-22A for electron beam lithography (at 2000 rounds per minute in 60 s) and baked with hot plate (110 °C in 5 min). Electron beam lithography (ELIONIX, ELS-7500EX, ELIONIX INC , Tokyo, Japan) was used to write desired pattern onto the resist layer ( Figure 1). After pattern writing process, samples were subjected to post-baking (110 °C in 5 min), developing with NMD-3 in 60 s and rinsing with iso-propyl alcohol (IPA), and finally blow-drying with nitrogen gas gun. At first, the Nb-doped TiO 2 film was co-sputtered at room temperature with TiO 2 ceramic and Nb metallic targets. Both silicon (001) and borosilicate glass were used simultaneously as sputtering substrates in every deposition. Radio frequency (RF) and direct current (DC) sputtering methods were used for TiO 2 and Nb targets, respectively. The sputtering chamber (!-Miller, Shibaura Mechatronics Corporation, Yokohama, Japan) was initially evacuated down to 4.0 × 10 −5 Pa as base pressure, then 19 sccm of argon and 1 sccm of oxygen gas flow was introduced into the chamber to create working pressure of about 0.3 Pa. Sputtering power was set at 200 W (RF) and 20 W (DC) for TiO 2 and Nb targets, respectively. After one hour of deposition, 80-nm-thick film was obtained. Then, the as-deposited films went through vacuum thermal annealing at 600 • C for 1 h. Optical properties of TiO 2 :Nb thin films were characterized with Ellipsometry (SENTECH, SE 850 DUV and SENDIRA, SENTECH Instruments GmbH, Berlin, Germany) from deep ultraviolet to far infrared region.
Consequently, the samples were spin coated with negative resist NBE-22A for electron beam lithography (at 2000 rounds per minute in 60 s) and baked with hot plate (110 • C in 5 min). Electron beam lithography (ELIONIX, ELS-7500EX, ELIONIX INC, Tokyo, Japan) was used to write desired pattern onto the resist layer ( Figure 1). After pattern writing process, samples were subjected to post-baking (110 • C in 5 min), developing with NMD-3 in 60 s and rinsing with iso-propyl alcohol (IPA), and finally blow-drying with nitrogen gas gun.
Reactive ion etching (ULVAC, CE-300I, SF 6 gas, 0.5 Pa, 100 W power) and oxygen plasma ashing (Mory PB-600, 300 W power in 15 min) steps were carried out to remove unnecessary surrounding TiO 2 :Nb nanostrips or nanodisks and wash away the remaining electron resist.
Shape and morphology of strip and disk nanostructures were investigated with scanning electron microscopy (SEM, Hitachi, SU-8400, Hitachi High-Technologies Corporation, Tokyo, Japan) and Atomic force microscope (AFM, Nanoscope 5, Bruker Corporation, Billerica, MA, USA). The silicon tips SI-DF20 (Hitachi High Technologies, Tokyo, Japan) were used at tapping mode.
Fourier transform infrared spectroscopy (FTIR, Thermo Scientific Nicolet iS50, Waltham, MA, USA) was used to assess the resonance features of nanostructures with different linear polarization of incident light, i.e., electric field vector was either parallel to length ( Reactive ion etching (ULVAC, CE-300I, SF6 gas, 0.5 Pa, 100 W power) and oxygen plasma ashing (Mory PB-600, 300 W power in 15 min) steps were carried out to remove unnecessary surrounding TiO2:Nb nanostrips or nanodisks and wash away the remaining electron resist.
Shape and morphology of strip and disk nanostructures were investigated with scanning electron microscopy (SEM, Hitachi, SU-8400, Hitachi High-Technologies Corporation, Tokyo, Japan) and Atomic force microscope (AFM, Nanoscope 5, Bruker Corporation, Billerica, MA, USA). The silicon tips SI-DF20 (Hitachi High Technologies, Tokyo, Japan) were used at tapping mode.
Fourier transform infrared spectroscopy (FTIR, Thermo Scientific Nicolet iS50, Waltham, MA, USA) was used to assess the resonance features of nanostructures with different linear polarization of incident light, i.e., electric field vector was either parallel to length nanostrips, as illustrated in Figure 2. Unpolarized incident light was used for nanodisks. Unpolarized incident infrared light (right side) passed through polarizer (P), aperture (A) and then exposed to nanostrip sample (S) before entering the detector (D). In case of nanodisk sample, polarizer (P) was removed and unpolarized infrared light was used.

Figure 2.
Transmission measurement setup with Fourier transform infrared spectrometer. Unpolarized incident infrared light (right side) passed through polarizer (P), aperture (A) and then exposed to nanostrip sample (S) before entering the detector (D). In case of nanodisk sample, polarizer (P) was removed and unpolarized infrared light was used.
Real part of the dielectric function (ε 1 ) exhibited a cross-over point in the mid-infrared region at approximately 8.5 µm. By increasing the Nb sputtering power, carrier concentration can be tuned, and Fermi level can easily be shifted toward conduction band. This in turn helps to adjust the cross-over point back and forth in the infrared region, whereas for conventional plasmonic metals such continuous tunability is not possible. Furthermore, this material also has great thermal durability, high surface smoothness, mechanical robustness, as well as photocatalytic activity and humid environment endurance. Combining altogether, TiO 2 :Nb film proves itself an excellent candidate for diverse infrared plasmonic applications. Real part of the dielectric function (ε1) exhibited a cross-over point in the mid-infrared region at approximately 8.5 µm. By increasing the Nb sputtering power, carrier concentration can be tuned, and Fermi level can easily be shifted toward conduction band. This in turn helps to adjust the cross-over point back and forth in the infrared region, whereas for conventional plasmonic metals such continuous tunability is not possible. Furthermore, this material also has great thermal durability, high surface smoothness, mechanical robustness, as well as photocatalytic activity and humid environment endurance. Combining altogether, TiO2:Nb film proves itself an excellent candidate for diverse infrared plasmonic applications.
To realize the plasmonic property of co-sputtered TiO2:Nb films in infrared region, nanostrips of different sizes were simulated with electromagnetic solver to check for the antenna resonance. The simulated parameters were transferred into lithographic patterning and fabrication, and then characterized by the IR spectroscopy. Figure 4a showed general designed nanostrip pattern used for Finite-Difference Time-Domain method (FDTD, Rsoft, Synopsis) simulation as well as electron beam lithography. In simulation, the lengths (L) of the strips were set from 650 to 750 nm while the width values (W) varied from 400 to 600 nm. To realize the plasmonic property of co-sputtered TiO 2 :Nb films in infrared region, nanostrips of different sizes were simulated with electromagnetic solver to check for the antenna resonance. The simulated parameters were transferred into lithographic patterning and fabrication, and then characterized by the IR spectroscopy. Figure 4a showed general designed nanostrip pattern used for Finite-Difference Time-Domain method (FDTD, Rsoft, Synopsis) simulation as well as electron beam lithography. In simulation, the lengths (L) of the strips were set from 650 to 750 nm while the width values (W) varied from 400 to 600 nm.  SEM and AFM images (Figure 4b,c) show periodic rectangular nanostrips, which are similar to intended design pattern (Figure 4a). The average strip surface roughness of 2.3 nm confirms that the etching and ashing processes did not cause considerable damage.
The smallest interval between two consecutive fabricated strips is approximately 300 nm. As illustrated in Figure 5d and Figure 6d, relative transmittance exhibits strong fundamental SEM and AFM images (Figure 4b,c) show periodic rectangular nanostrips, which are similar to intended design pattern (Figure 4a). The average strip surface roughness of 2.3 nm confirms that the etching and ashing processes did not cause considerable damage.
The smallest interval between two consecutive fabricated strips is approximately 300 nm. As illustrated in Figures 5d and 6d, relative transmittance exhibits strong fundamental antennalike resonances as polarized electric field vector is parallel to nanostrip length or width. Both measured and simulated spectra show good agreement. Calculated spectra demonstrate orderly shift of resonances toward longer wavelengths (black dash arrow) and reveal the improvement in quality factor of nanoantenna (sharper and deeper resonances) as length or width of nanostrips increases, as in the case of metal nanowires [32]. FDTD simulation results along x-y and x-z planes (Figure 5d,e and Figure 6d,e) also show strong electric field confinement and enhancement at the edges of nanostrips. This proves the antennalike dipole resonance of the structures.   General designed nanodisk pattern for FDTD calculation and electron beam lithography was shown in figure 7a. Diameters of nanodisks (D) were set from 600 to 800 nm.
In simulated results, systematic increase of resonance frequency with diameter (D) can be easily observed. Sharper resonances can be expected if disk diameter keeps expanding. Experimental relative transmittance spectrum is in consonance with calculated lines with strong fundamental antennalike resonances under any linear polarized infrared light parallel to the disk surface ( Figure  8d). Quality factor of these nanodisk antennae can be further improved when the diameter D increases since the surface plasmon resonance assumes more photonic nature and the inherent loss of the material becomes less. While resonant frequencies show good matching between simulation and experiment curves in x axis; there is some difference (~10% to 15%) in relative transmittance (y-axis). This can be explained by two factors: fabrication tolerance and etchant effect. Perfect cuboids were used in simulation, but the obtained ones were not perfectly flat at edges, as shown in Figure 4c,d. Furthermore, electron carrier concentration was decreased from 1.12 × 10 21 cm −3 for   In simulated results, systematic increase of resonance frequency with diameter (D) can be easily observed. Sharper resonances can be expected if disk diameter keeps expanding. Experimental relative transmittance spectrum is in consonance with calculated lines with strong fundamental antennalike resonances under any linear polarized infrared light parallel to the disk surface (Figure 8d). Quality factor of these nanodisk antennae can be further improved when the diameter D increases since the surface plasmon resonance assumes more photonic nature and the inherent loss of the material becomes less. While resonant frequencies show good matching between simulation and experiment curves in x axis; there is some difference (~10% to 15%) in relative transmittance (y-axis). This can be explained by two factors: fabrication tolerance and etchant effect. Perfect cuboids were used in simulation, but the obtained ones were not perfectly flat at edges, as shown in Figure 4c,d. Furthermore, electron carrier concentration was decreased from 1.12 × 10 21 cm −3 for pristine film to 6.4 × 10 20 cm −3 for remained one. SF 6 gas may introduce some electron-trapping fluorine ions on surface of nanostructure after etching.

Conclusion
We succeeded in designing and fabricating antenna arrays of nanostrips and nanodisks based on Nb doped TiO2. Relative infrared transmittance in calculated and experimental results are in good accordance. Spectral features exhibit red-shift in their peak position as the sizes of strips and disks increase. In the same tendency, quality factor of nanostrips and nanodisks can also be improved if structure dimension can be further extended.  Diameters of nanodisks (D) in simulation were set at 600, 700, and 800 nm, corresponding to blue, pink, and green curves. Measured and fitted lines were drawn in blue and red (D~750 nm) (c). Electric field distribution along x-y and x-z planes (d, e).

Conclusions
We succeeded in designing and fabricating antenna arrays of nanostrips and nanodisks based on Nb doped TiO 2 . Relative infrared transmittance in calculated and experimental results are in good accordance. Spectral features exhibit red-shift in their peak position as the sizes of strips and disks increase. In the same tendency, quality factor of nanostrips and nanodisks can also be improved if structure dimension can be further extended.