Preparation of Anodic Aluminum Oxide Masks with Size-Controlled Pores for 2D Plasmonic Nanodot Arrays

Anodic aluminum oxide (AAO) layers with nano-sized pores have been used as shadow masks for the fabrication of two-dimensional (2D) metal nanodot arrays (NDAs). However, the localized surface plasmon resonance (LSPR) of size-controlled NDAs fabricated using AAO masks has not been much studied. In this study, we report on the improved preparation method and utilization of an ultrathin AAO mask for the fabrication of 2D plasmonic metal NDAs. The greatest challenge in preparing an AAO mask is to control the pore diameter and to make it reproducible. AAO masks with size-controlled pores were reproducibly prepared using a two-step chemical wet etching method. Ag NDAs with different dot diameters (42, 60, and 80 nm) and Ag, Cu, and Au NDAs with dot a diameter of 80± 5 nm were fabricated on indium tin oxide glass substrates using AAO masks. The wavelengths corresponding to LSPR of 2D metal NDAs were investigated using ultraviolet–visible spectroscopy. Our results show that AAO masks with tunable pores can be used as shadow masks for the fabrication of 2D plasmonic NDAs.


Introduction
Fabrication techniques for two-dimensional (2D) plasmonic nanodot arrays (NDAs) have attracted significant attention in versatile plasmonic applications for the enhancement of the performance of optoelectronic devices and for the highly sensitive detection of biological and chemical sensors [1][2][3]. The localized surface plasmon resonance (LSPR) of metal nanoparticles is sensitive to parameters such as shape, size, composition, and the surrounding environment [4][5][6]. Along with the development of nanostructure manufacturing technology, research on various plasmonic applications based on the LSPR of metal nanostructures has been proceeding rapidly [7][8][9][10][11][12]. Plasmonic nanosensors have been utilized to detect chemical and biomolecular binding on metal surfaces by measuring extremely small wavelength shifts in LSPR owing to a change in the refractive index [7][8][9]. Plasmonic properties of metal nanoparticles have been investigated to improve the performance of optoelectronic devices such as organic light-emitting diodes (OLEDs), solar cells, and light-emitting diodes [10][11][12].
In addition, the LSPR features depending on the shape and size of metal nanostructures have a significant influence on the characterizations of optoelectronic devices such as the change of emission color in OLEDs, the conversion efficiency of solar energy, and the activity of catalyst [13][14][15]. Specifically, metal nanoparticle arrays exhibit much higher electromagnetic field enhancements than those of singleinsulated nanoparticles and are easy to reproducibly produce in the desired region [16,17]. Thus, manufacturing techniques that can reproduce a metal nanoparticle array on a substrate are important for optimizing the performance in plasmonic applications.
2D nanoparticle arrays such as Au disk, Ag disk, and ellipsoidal Au nanoparticles with different lattice constants have been achieved using electron beam lithography [17,18]. This technique facilitates the control of the geometric pattern of nanoparticles, but it has some drawbacks such as high cost and large-area patterning. Owing to these limitations, 2D metal nanoparticle arrays have been fabricated by nonlithographic techniques using self-organized nanostructured materials as a template [4,[8][9][10][11][19][20][21][22][23][24][25][26][27]. Among the materials, a self-organized anodic aluminum oxide (AAO) layer with uniform through holes of nanometer dimensions has been widely used as a shadow mask to fabricate 2D NDAs of various materials [20][21][22][23][24][25][26][27]. However, the plasmonic properties of metal NDAs produced using AAO masks have not been reported widely. Several studies have reported the LSPR of Ag or Au NDAs with height variations by varying the amount of metal deposited using AAO masks with similar pore diameters [24][25][26]. However, the LSPR of metal NDAs with different diameters of similar height has not yet reported due to the difficulty of controlling the pore diameter of the AAO mask. The most challenges in preparing the AAO mask are to completely remove the barrier layer at the bottom and control the pore diameter. In order to obtain an ultrathin AAO mask, the removal of the barrier layer and pore widening were performed by chemical etching in 5% H 3 PO 4 or by Ar + milling [24][25][26][27][28][29][30]. The Ar + ion milling method for the barrier layer removal is expensive and time-intensive [28]. Removing the barrier layer through long-term chemical etching is difficult to control the size of the pores because the AAO mask tends to break or crack [29]. In this study, we demonstrate an improved preparation method of AAO masks with adjustable pore diameters for the fabrication of size-controlled 2D NDAs. AAO masks were prepared by a two-step chemical wet etching method as reported previously [24]. The pore diameters of AAO masks were adjusted by controlling the dipping time in the second chemical wet etching process. Ag NDAs with different dot diameters and Ag, Cu, and Au NDAs with similar dot diameters were fabricated on indium tin oxide (ITO) substrates using AAO masks with sizecontrolled pores. The chemical structures of metal NDAs were examined by X-ray photoelectron spectroscopy (XPS). The plasmonic properties of metal NDAs were investigated using ultraviolet-visible spectroscopy.

Experiments
2.1. Preparation of AAO Mask. The surface of the aluminum foil was pretreated using electrolytic polishing at 20 V for 60 s in a solution of perchloric acid and ethanol (HClO 4 : C 2 H 6 O = 1 : 5 (in volume ratio)). The aluminum foil inserted in a sample holder made of Teflon was connected to the anode, and the platinum electrode was connected to the cathode. Figure 1(a) shows the schematic diagrams of the equipment configuration for the anodization method and the structure of the AAO layer formed on the aluminum foil by the anodization method. Figure 1(b) shows the schematic diagrams of the two-step anodization process. The first anodization was performed in a 0.3 M oxalic acid   The second anodization was performed for 240 s under the same conditions as those of the first anodization. The AAO layer was etched slightly via immersion for 600 s in 5 wt% phosphoric acid at 30°C following the second anodization. The surface of the AAO layer was subsequently coated with a solution containing a mixture of nitrocellulose and polyester resin in butyl acetate, ethyl acetate, and isopropyl alcohol. The remaining aluminum substrate was removed using a saturated mercuric chloride (HgCl 2 ) solution for approximately 4 h. The pore diameters of AAO masks were adjusted by controlling the dipping time in the second chemical wet etching process. Finally, the coating layer on the AAO surface was dissolved in acetone. Subsequently, the AAO mask was filtered using a filter paper.

Fabrication of Plasmonic NDAs.
The AAO mask was placed on an ITO substrate, which was subsequently placed on a sample holder in the evaporator chamber. Metal was deposited on the surface of the ITO substrate covered with the AAO mask by using an electron beam evaporator (ULVAC EBX-1000). Under a vacuum of~10 −4 Torr, metals such as Ag, Cu, and Au were deposited at a deposition rate of 0.1 nm/s for 500 s. Under these deposition conditions, the thickness of the metal film deposited on the ITO substrate without an AAO mask was approximately 50 nm. After the deposition of each metal, the mask on the ITO substrate was selectively dissolved in 1 M sodium hydroxide (NaOH) for several minutes and rinsed several times with distilled water. Thus, Ag, Cu, and Au NDAs were formed on the ITO substrate.

Characterizations.
The morphologies of the AAO mask and metal NDAs were observed using a field-emission scanning electron microscope (FE-SEM; Hitachi S-4700) and atomic force microscope (AFM; Park Systems XE-100). The chemical compositions and electronic structures of the metal NDAs were confirmed using XPS (PHI 5000 Versa Probe, ULVAC-PHI Inc.). The XPS measurements were performed using microfocused (100 μm × 100 μm, 25 W, 15 kV) monochromatic Al Kα radiation with an energy of 1486.6 eV. The instrument was calibrated to provide a C1s binding energy of 284.6 eV. The pressure inside the analyzer was maintained at 5 × 10 −10 Torr. The plasmonic properties of the metal NDAs were measured using a UV-Vis spectrometer (Perkin Elmer Lambda 18) in a wavelength range of 270-900 nm. The sample was placed behind a collimator lens and on a position-controlled stage, and the extinction measurement was recorded using unpolarized light with a probe beam size of approximately 4 mm 2 .

Results and Discussion
3.1. Preparation of AAO Mask. The AAO layer was prepared from an aluminum foil by the two-step anodization process reported by Masuda and Satoh [22]. After the second anodization, the AAO layer was immersed once in the phosphoric acid solution, and then the pore diameter of the AAO mask was controlled by varying the immersion time in the second etching process. The AAO mask with size-controlled pores was prepared by a two-step wet etching process, which was designed with the hint in the two-step anodization method. Figure 2 shows schematic illustrations and FE-SEM images of the pore opening process of the AAO layer using the two-step wet etching method. Figure 2 After the first wet etching, the surface of the AAO layer was coated with the coating layer, and the remaining aluminum substrate was removed using a saturated HgCl 2 solution. In the second wet etching process, the AAO layer was immersed again in 5 wt% H 3 PO 4 solution for 690 s. Figure 2(d) shows the FE-SEM image of the AAO mask after the removal of the protecting layer in acetone. As shown in Figure 2(d), it was confirmed that the barrier AAO layer was completely dissolved. The pores were widened uniformly, and the average diameter of the pores was approximately 60 nm (±5 nm).

Metal NDAs Using AAO Masks.
Ultrathin, large-area AAO masks could be prepared using a two-step wet etching process. Au was deposited on an ITO substrate covered with an AAO mask for 500 s at a deposition rate of 0.1 nm/s. Figure 3(a) shows the FE-SEM image of Au-deposited AAO mask and Au NDA revealed from which the AAO mask was partially stripped with a carbon tape. Figure 3(b) shows the FE-SEM image of Au NDA after removing the AAO mask with a carbon tape. Figure 3(c) shows the interface between the Au-deposited mask surface and the Au NDA after the AAO mask is partially removed. Figure 3(d) shows that Au clusters were deposited on the top surface of the AAO mask as a replica of the pore pattern of the mask. Under   Figure 5(a), with an average height of 20 ± 5 nm corresponding to the dots on the line. The mean diameter of Ag NDA formed on the ITO substrate was 45 ± 5 nm. Figure 6(b) shows the crosssectional line profile along the red line in the AFM topology image of the curve, with an average height of 22 ± 5 nm corresponding to the vertices of two points on this line. The mean diameter of Ag NDA was 60 ± 5 nm. In the curved AFM topology image, the cross-sectional line profile along the red line is shown in Figure 5(c), with an average height of 18 ± 4 nm corresponding to the two vertices of the point. The average diameter of Ag NDA was 80 ± 5 nm. Considering the variation in height at the nanolevel of the surface of the coated ITO substrate, it can be observed that the results were produced at approximately the same height. The dot heights of Ag NDA depend on the aspect ratio of the AAO mask and the amount of the deposited metal [20,26]. Thus, it can be observed that AAO masks were manufactured under similar conditions. Figures 6(a)-6(c) show the FE-SEM images of Ag, Cu, and Au NDAs with similar diameters that were fabricated on ITO substrates using AAO masks with a pore diameter of 80 ± 5 nm. The average diameter of the metal NDAs was 80 ± 5 nm as a replica of the AAO mask. The chemical composition and bonding states of Ag, Cu, and Au NDAs were studied using XPS. The binding energies were referenced to the C1s line, which was at 284.6 eV from carbon. As shown in Figure 6(d), the XPS spectra obtained from Ag NDA were centered at 368.2 and 374.2 eV. These values are consistent with the Ag 3d 5/2 and Ag 3d 3/2 core-level binding energies for the Ag nanoparticles appearing at 368.1 and 374.1 eV [31]. The separation value between the Ag 3d 5/2 and Ag 3d 3/2 transitions exhibits a spin orbital splitting of 6.0 eV, which is a characteristic of metallic silver [32]. Figure 6(e) shows the XPS spectra obtained from Cu NDA. The XPS spectra of Cu 2p 3/2 at 932.4 eV and 2p 1/2 at 952.4 eV confirm the presence of metallic copper [33,34]. Otherwise, the two dominant peaks were Cu 2p 1/2 at 953.9 eV with an intense satellite at 962.5 eV and 2p 3/2 at 934.4 eV with an intense satellite at 941.2 eV, which are the peaks corresponding to CuO and copper hydroxide (Cu(OH) 2 ). Copper hydroxide usually displays the Cu 2p 3/2 spectrum in the range of 934 to 936 eV [34,35]. The values of the Cu 2p 3/2 binding energies for Cu metals, CuO, Cu(OH) 2

Journal of Nanomaterials
These results indicate that copper hydroxide was formed on the surfaces of the Cu NDA during the process of dissolution of the AAO mask in aqua solution. As shown in Figure 6(f), the Au 4f 7/2 and Au 4f 5/2 core-level binding energies for Au NDA were observed at 84.1 and 87.9 eV. The binding energy of Au 4f 7/2 for metallic gold was observed at 84.0 eV [37]. The binding energies of Au 4f 7/2 and Au 4f 5/2 for the Au NDA were observed at 84.7 and 88.3 eV, respectively [16]. The slight difference in the XPS peak position may be due to the changes in binding energy because of the particle shape or local chemical composition on the sample surface [37]. Thus, the material components of plasmonic NDAs were confirmed using XPS measurement.

Plasmonic
Properties of Metal NDAs. The extinction spectrum, which is the sum of the absorption and scattering spectra, of metal NDA was measured in transmission mode using UV-Vis spectroscopy. The optical extinction has a maximum at the plasmon resonant frequency, which corresponds to the LSPR peak of noble metal nanoparticles appearing in the visible wavelength range [6]. Figure 7(a) shows the LSPR spectra of Ag NDAs with different dot diameters in the period of 105 nm. The LSPR spectrum of the Ag NDA with a diameter of 45 ± 5 nm indicates the maximum peak at 498 nm. The spectra of the Ag NDAs with diameters 60 ± 5 nm and 80 ± 5 nm indicate the maximum peaks at 516 nm and 533 nm. As shown in Figure 7(a), the LSPR peak position is observed at different positions depending on the Ag dot size, shifting to the red region as the Ag nanodot diameter increases. The LSPR intensity of an Ag NDA with larger dots is higher than that with smaller dots. The LSPR peak broadening for Ag NDA on the substrate becomes larger as the diameter of the nanodots increased. The linewidth and the amplitude of the LSPR peaks depend on the size distribution and the distances between the particles [38,39]. Our results are similar to those of the previous 5 Journal of Nanomaterials studies in which the LSPR wavelength shifts to red and the peak broadens as the particle size increases [4,18].   Journal of Nanomaterials of the Cu and Au NDAs, the maximum peaks were shown at 648 nm and 673 nm, respectively. The LSPR peak of the Ag NDAs was observed at a shorter wavelength than those of the Cu and Au NDAs. This result is consistent with the trends of peak positions of the LSPR from Ag and Au nanoparticles [40]. The LSPR of metal NDAs prepared using AAO masks has mainly focused on Ag or Au NDAs of different heights with similar pore diameters [24][25][26]. Au NDAs of different heights (25,40,60, and 80 nm) with a similar dot diameter of 55 nm were fabricated on quartz plates, and the LSPR peaks are located at 700, 650, 620, and 585 nm [24].   7 Journal of Nanomaterials peaks are observed at 433, 452, and 502 nm. The peak positions of the Ag NDAs with σ = 0 88, 1.56, and 2.44 shift to higher wavelengths as the aspect ratio increases [25]. The calculated peak broadening is smaller than the experimental width of the LSPR peak for the Ag NDA on the glass substrates. This is due to the dispersion of geometrical dimensions of the nanodots in the actual samples [25,38]. The LSPR characteristics of metal NDAs are significantly affected by the substrate, metal deposition conditions, and the aspect ratio of the AAO mask [6,20]. In addition, the LSPR in conductive metal oxide nanoparticles and in vacancy-doped semiconductor quantum dots is strongly influenced by the free electron carrier density [41,42]. Since LSPR wavelengths are affected by various factors such as the nanoparticle size, composition, and substrate, free electron carrier density, more experiments should be systematically conducted to explore the LSPR in many material systems. To do this, reproducible manufacturing techniques of AAO masks are important. In the conventional wet etching process, the immersion time required for the pore opening of the AAO layer was difficult to control well owing to a slight difference in the thickness of the AAO barrier layer [29]. The etching rate may vary depending on the diffusion rate of the acid and depending on the shape of the surface, and hence even though there is a very small difference in height, the shape of the pores is significantly affected by etching for a long time [30]. A conventional chemical wet etching method for a long time results in the pore opening of the alumina barrier layer, but the pore wall of the AAO layer is sometimes collapsed. By using a two-step wet etch process, it was possible to achieve the pore opening in a short time without breaking the pore wall of the AAO layer, and it was possible to stably and reproducibly prepare AAO masks with size-controlled pores.

Conclusions
We demonstrated reproducible fabrication and plasmonic properties of size-controlled plasmonic NDAs using AAO masks. Large-area nanoporous alumina masks were achieved by a two-step wet etching process. The pore diameter of the AAO mask can be reproducibly controlled in the range of the hexagonal cell size by varying the immersing time of the second chemical wet etching. The dot diameter of the metal nanodot array depends on the AAO mask used as the evaporation mask. The LSPR spectra of Ag NDAs with different sizes (42 nm, 60 nm, and 80 nm) showed the maximum peak at 498 nm, 516 nm, and 533 nm and shifted to the red region as the dot diameter increased. The LSPR spectra for the Ag, Cu, and Au NDAs with a similar diameter of approximately 80 ± 5 nm indicated the maximum peaks at 533 nm, 648 nm, and 673 nm. These AAO masks could be utilized as a versatile mask to manufacture various 2D plasmonic NDAs.

Data Availability
All the data supporting the results are shown in the paper and are available from the corresponding author upon request.

Conflicts of Interest
There are no conflicts to declare.