1. Introduction

Metallic structures at the subwavelength scale excite localized surface plasmons (LSPs) by coupling electromagnetic wave. These structures have attracted attention because the properties of LSPs, which confine, enhance, and absorb electromagnetic fields, can be used for a wide variety of applications, such as photothermal therapies [1], surface enhancement Raman spectroscopies [2], photothermal converters [3–5], and electromagnetically induced transparency window [6–9]. By tailoring the structure carefully, the desired optical response from the plasmonic nanostructure, which matches for these applications, can be obtained. A semishell, dielectric nanosphere coated partially with metal, is one of the excellent plasmonic nanostructures because of its large absorption cross-section; this symmetry-breaking structure has the intrinsic property that induces unidirectional strong scattering, which can be applied to a SPASER [10-11]. The semishells possess high tunability of their plasmon resonance, which is controlled in a wide range by designing the size of the dielectric sphere and shape and thickness of the metal shell [12–15], and feature fabrication simplicity using bottom-up process [12,16,17,34]. The noble metals Au and Ag are recognized as representative plasmonic materials because of their strong plasmon resonance. The wavelength range, wherein LSPs of Au and Ag can be excited, is limited in the visible and near-infrared regions because of the strong interband transition in the visible and UV regions. Using Al is a powerful solution to increase the working band of plasmonic nanostructure, because Al has narrow interband transition band at around 800 nm so that LSPs can be excited in visible and UV region. Combing the plasmon resonance tunability of semishells and the wide working band of Al, it is expected that a plasmonic nanostructure capable of tailoring its plasmon resonance in wavelengths ranging from UV to NIR can be achieved with a simple fabrication procedure and low cost. This allows manufacturing plasmonic nanostructures over a large area, which is necessary for solar energy harvesting systems.

When considering the use of an Al semishell for applications such as thermal emitters [18] and solar absorbers [17,19–22], which are expected to be used in high-temperature environments, high thermal stability is required to prevent degradation of the optical properties by melting. Here, two problems are expected. First, the melting point of Al (660 °C) is much lower than that of Au (1064 °C) and Ag (962 °C). The photothermal reshaping of Au and Ag nanostructures caused by heat generated from the Ohmic loss of plasmon resonance has been reported [23–25]. This reshaping of those metals is expected to begin at lower temperatures than their melting point because of the size effect at the nano-scale [26], which is also considered a problem for Al [27]. Thus, an Al semishell might have a much lower thermal stability than other plasmonic materials. Second, in addition to melting, Al shows an intrinsic ability to produce a self-limited oxide layer for protecting itself from further oxidation. The oxide layer growth is accelerated at high temperature, which shrinks the volume of Al remaining inside nanostructures, thereby changing their shape and optical properties [28,29].

In this paper, as is contrast to above predictions, we report that the Al semishells have high thermal stability, and their optical properties are retained at high temperatures. Such a new aspects of Al semishell is very useful for applications in high temperature environments, like a solar absorber. We observe the thermal stability of the structure of Al semishells by heating directly using an electrical furnace and taking SEM images. The absorption spectra are measured to evaluate the optical properties of the Al semishells with oxide layer growth. These prove that Al semishells have high thermal stability; structural deformation of Al shell was not observed at a heating temperature of 400 °C, and either no marked decrease in absorption efficiency or large shift in plasmon resonance occurred below 400 °C. The numerical absorption spectra of the Al semishell, including the oxide layer, were also obtained using finite element method to investigate the experimentally observed shift in plasmon resonance.

2. Experimental procedure

2.1 Fabrication methods and heating methods

Glass substrates cleaned using an ultrasonic bath with detergent solution were functionalized with a 1.1%(v/v) solution of N-(2-aminoethyl)-3-aminopropyltrimethoxysilane (JNC Sila-Ace S-320) in an ethanol/acetic acid mixture (95: 5 v/v) via immersion for 1 h. The immersed glass substrates were rinsed with ethanol, then immersed into a prepared suspension of SiO₂ spheres with diameters of 100 nm and surface functionalized with COOH (from micromod Partikeltechnologie, Germany). The density of SiO₂ sphere in the suspension was 0.2%(w/v). The suspension was stirred with a magnetic stirrer during immersion to increase the number of SiO₂ spheres colliding with the substrate surface. After immersion for 1 h, the substrates were carefully rinsed with pure water, followed by drying in air at room temperature (20 °C). The dried substrates were placed in a sputtering chamber (Quorum Q150T S) facing normal to the sputtering direction to form Al shells on SiO₂ spheres (Fig. 1). The deposition thickness was 10 nm, which was estimated from transmittance spectra of a simultaneously sputtered plain Al film using a transfer matrix method. Note that, in general, the self-limited oxide layer of a few nano-meter is formed in it, which is typically approximately 3 nm [28,29]. The fabricated samples were heated using an electrical furnace (AS ONE TMF-300N). The heating temperature was set as 100 °C, 200 °C, 300 °C, 350 °C, and 400 °C with heating durations of up to 40 h. The temperature reached the desired point at a rate of approximately 5 °C/min. The samples were individually heated at these temperatures.

 

Fig. 1 Image of Al semishells fabricated on glass substrate.

Download Full Size | PDF

2.2 Optical measurement and SEM images

Transmittance was measured using a NIHON BUNKO V-660 spectrophotometer with baseline subtraction. The baseline was measured with the same glass substrate used for the samples. A glass substrate coated with an Al film whose thickness was the same as the samples was used as a reference to measure the absorbance of samples. Absorbance was obtained using Beer’s law. SEM images were acquired using HITACHI S-4500 with acceleration voltage of 15 kV, and the observation angle was inclined at 0° and 45° from the perpendicular axis of the sample surface.

3. Results and discussion

3.1 Thermal stability of an Al semishell nanostructure

Images of the samples are shown in Fig. 2. The samples in the left column (Room temp) in Fig. 2 were not heated and were stored at room temperature. The top one in this column is a glass substrate sputtered with Al without SiO₂ spheres on the surface, and the bottom one has Al semishells on the left side of the substrate surface, which has a darker color compared with the right side. In the columns next to the column for room temperature, samples heated at 100 °C, 200 °C, 300 °C and 400 °C are included, moving from left to right. The time denoted below the samples is the heating time. In comparison with samples not heated and heated at 100 °C-300 °C, large differences in terms of the color of the surface are not observed. However, for the samples heated at 400 °C, the color drastically changed from blue to purple and pale pink as the heating time increased. The right side of the sample heated at 400 °C for 10 h became transparent because of complete oxidation [30].

 

Fig. 2 Image of samples after heating at 100 °C, 200 °C, 300 °C and 400 °C for 1, 5, and 10 h. The samples in the left column were stored at room temperature. Al semishells are on the left side of samples, which have darker color than the right side.

Download Full Size | PDF

Figure 3(a) shows an SEM image of the surface of the fabricated sample. Almost all semishells on the surface of the substrate were distributed randomly with separation from each other, but some of them had aggregated. The average density of semishells for all samples was $\text{14}.\text{25}/\mu m^2$. It should be noted that there are individual variabilities for density in samples, slightly differing from the average density, which can affect the optical response. The left side of Fig. 3(b) shows the SEM images of Au and Al semishells at room temperature. Although SEM images for Al semishell are not clear enough to identify the shape of Al covering SiO₂ core, we consider the Al deposit covers with the SiO₂ core partially, not completely, from point of view of the SEM observation for other metals, such as Au and Ag, and the numerical analysis. To compare the thermal stability of Al with another plasmonic material, Au and Al semishells were heated at high temperature. The SEM images of Au and Al semishells after heated are shown in right side of Fig. 3(b). The deformation of Au shells was observed at a heating temperature of 250 °C, which is much lower than the melting point of bulk Au (1046 °C). This is because of the size effect, which decrease the melting point [26]. For the Al semishells, although the melting point of Al (660 °C) is much lower than that of Au, Al shells did not deform when heated at over 250 °C for several hours. This might be attributed to the fact that Al has the intrinsic passivation properties, which result in an Al oxide layer (melting point: 2072 °C) covering the surface of Al to prevent further deterioration; and Al oxide barrier can protect Al shells from deformation.

 

Fig. 3 (a) SEM image of sample surface with adsorbed Al semishells composed of a 100 nm SiO₂ core and a 10-nm Al shell. The density of semishells is 14.25/μm². (b) SEM images of Au and Al semishells before and after heating. Both semishells are composed of 100-nm SiO₂ core and 10-nm metal shell. The heating temperature was 250 °C for Au semishells and 400 °C for Al semishells. The inset white scale bar corresponds to 100 nm.

Download Full Size | PDF

3.2 Optical properties of Al semi-shell

To analyze the effect of oxidation on the optical properties of Al semishells quantitatively, the increase and decrease of the absorption peaks are plotted as differences ΔA in terms of heating time in Fig. 4(a). Note that the absorption spectra, which we measured, include the absorption of the Al deposit on the glass substrate, however, the numerical analysis showed that the absorption peaks are not influenced by the Al deposit and are attributed only to plasmon resonances of the Al semishell. For heating temperature of 100 °C-350 °C, the absorption peak continues to increase for a heating time of 10 h, then starts to decrease from this point and stabilized after a heating time of 20 h. The curve for a heating temperature of 400 °C, however, drastically decreased. For 400 °C-20 h, absorption by Al semishells was not observed in the measured spectra. This is because the Al shell has been completely oxidized so that plasmon resonance cannot be excited. It has been reported that oxidation processing can be divided into two stages in terms of the oxide growth rate; the first stage is a fast growth rate, and the second stage is a slow oxidation growth rate [31]. The first stage is attributed to the initial thickness of the oxide film, the thickness of which increases with increasing temperature, and oxidation stops at a certain degree. At low temperature (<300 °C), oxidation stops progressing in this stage and further oxide film growth does not occur. The oxide film growth continues in the second stage with a low growth rate only at high temperature (>400 °C). In comparison with this report, our results for absorption peak difference are in good agreement as shown in Fig. 4(a); for heating temperature below 350 °C absorption peak stabilized, meaning oxide shell growth ceased in the first stage of oxidation, and for heating temperature above 400 °C drastically decreased because of oxidation progression in the second stage. Comparing this result to the literature [37], which investigated the effect of rapid thermal annealing to the structural and optical properties of Al nanoparticles, 400 °C can be considered as the threshold temperature of that the oxidation growth and the deterioration of optical properties is dominated for Al nanostructures.

 

Fig. 4 (Left side graph) Absorption peak difference ΔA and plasmon resonance shift Δλ are depicted. (Right side graph) (a) ΔA in terms of heating temperature and time. (b) ΔA in terms of Δλ. The red arrows indicate the direction increasing heating temperature and time. The absorption spectra of plots in the same dashed circles have the same characteristics with respect to the deformation of spectra.

Download Full Size | PDF

To estimate the optical properties of Al semishells with oxide shells whose thickness grows with increasing heating temperature and time, ΔA in terms of the amount of plasmon resonance shift Δλ are plotted in Fig. 4(b). The inset red arrows indicate the direction of increasing heating temperature and time. The plots can be separated into four groups because the absorption spectra of plots in the same group have analogous shapes. The representative spectra from each group are shown in Fig. 5. From these spectra, it can be observed that the fabricated Al semishells in this experiment have plasmon resonance at approximately 540 nm in the visible region. Figure 5(a) is the spectrum of the plot around the zero point for ΔA and Δλ, for which a slight redshift of plasmon resonance and increase of absorption peak can be observed. With increasing heating temperature and time, the plasmon resonance redshift and its amplitude increases further (Fig. 5(b)); then, the redshift and the amplitude begin to weaken (Fig. 5(c)); finally, there is a progressive blueshift and the amplitude decreases drastically (Fig. 5(d) and (e)). Note that the negative absorption might be attributed to smaller reflection of the sample than that of the reference, which is caused by the gradual refractive index change due to existence of SiO₂ spheres.

 

Fig. 5 Representative absorption spectra of Al semishells from each phase in Fig. 4(b).

Download Full Size | PDF

To determine the correlation between oxide shell growth and plasmon resonance in the Al semishells fabricated in this experiment, extinction spectra were obtained numerically using the finite element method by a commercial software package, COMSOL Multiphysics. It has been reported that the plasmon resonance of semishells is determined by the ratio of the core diameter to the shell thickness and the covering angle of the shell edge [33]. Taking this into account, Fig. 6(a) schematized the applied simulation model, involving oxide shell growth from two directions: shrinking of the Al shell thickness (t_Al) and narrowing of the covering angle of the Al shell (θ). This simulation model is composed of a core radius (r_core) of 100 nm and a shell thickness (t_shell) of 10 nm. The Al shell is covered by Al₂O₃ because of oxidation. The dielectric parameters of Al and Al₂O₃ are taken from Ref [35]. and [36], respectively, and the refractive index of the silica glass core is set to be n = 1.45, without an extinction coefficient. The extinction spectra for θ of 90°, 80°, 60°, 45°, 30° with varying t_Al of 8, 7, 6, 5, 4, and 2 nm were calculated, and the plasmon resonance peaks of these spectra are plotted in Fig. 6(b). From Fig. 6(b), the decrease in t_Al is attributed to the redshift of plasmon resonances, and a blueshift occurred with narrowing of θ. In the range θ from 90° to 70°, however, plasmon resonances showed a redshift with respect to 90°. It is revealed from numerical results that the direction of the shift in the plasmon resonance of Al semishells is determined by which directions of oxidation, determined by t_Al and θ, are dominant in the oxide shell growth.

 

Fig. 6 (a) Schematic of the simulation model of Al semishell with oxide shell. (b) Plots of the extinction peak of numerical extinction spectra of Al semishell. Black dashed lines indicate t_Al. The black circle indicates the point at which we assume the initial shape of Al semishells fabricated in our experiment.

Download Full Size | PDF

The inset green dashed line, indicating the average plasmon resonance of Al semishells before being heated in Fig. 5, is inserted in Fig. 6(b) to compare with experimental and numerical results and assume the process of oxide shell growth in Al semishells. In addition, the inset red and blue dashed lines in Fig. 6 indicate the plasmon resonances of Fig. 5(b) and (e), respectively. In the experimental results, the plasmon resonance of Al semishells showed a redshift from the initial green dashed line to the red dashed line with an increase in the absorption peak, followed by a blueshift to the blue dashed line with a drastic decrease in the absorption peak. In the literature [28,29], the measured thickness of the initial oxide layer of Al nanostructures at room temperature is approximately 3 nm. Comparing the literature results and our result, the initial shape of our semishells at room temperature is assumed from the point where the green dashed line intersects with the black dashed line of t_Al = 7 nm (Al₂O₃ shell thickness of 3 nm). This point is indicated by the black circle in Fig. 6(b). For the redshift with increasing absorption peak, it can be confirmed that the redshift from the green to red dashed lines is dominated by shrinking of t_Al; however, the reason for such the increase in the absorption peak cannot be understood from Fig. 6(b). The blueshift of the plasmon resonance and decrease in the absorption peak in the region between the red and blue dashed lines are strongly dominated by the decrease in covering angle; therefore, it can be assumed that the narrowing speed of the covering angle is faster than the shrinking of the shell thickness in this region, meaning that the direction of oxidation over the Al shell is not uniform. Considering a more realistic semishell structure than that used in the simulation model, the shell thickness is different over the shell, and the thickness around the edge is thinner than that on top of the shell [34]. For oxide shell growth, it has been reported that the thickness of the initial oxide shell formed on Al nanoparticles differs with their size, with smaller particles having thicker oxide shells, and the thickness is determined by the temperature surrounding them [32]. By taking into account these reports, faster oxide shell growth around the edge than on top of the shell can be considered for our fabricated Al semishells. This might cause the blueshift of the plasmon resonance and decrease in the absorption peak, which are observed in Fig. 5(b) and (c). The assumptions made above can describe well, except for the increase in the absorption peak, the factors that contribute to the shift in plasmon resonance in Al semishells observed in Fig. 5. But, the experimental confirmation is required to verify above assumption.

4. Conclusion

We investigated the thermal stability of Al semishells by heating them directly at high temperature and their optical properties by measuring their absorption spectra experimentally and numerically. The deformation of Al semishells after heating at 100-400 °C for several hours was not observed. Comparison the absorption spectra of Al semishells before and after heating, for heating temperatures of 100-350 °C, the spectral changes become stable as heating time increases; the absorption peak stops decreasing after a certain amount of heating time, but for 400 °C, the absorption peak decreases continuously. A shift in the plasmon resonance was also observed in the absorption spectra, which was described by simulation results; the narrowing of the covering angle of the Al shell edge and the shrinking of the Al shell thickness, caused by oxide shell growth, contribute to a blueshift and redshift, respectively. In our fabricated Al semishells, the shift in the plasmon resonance caused by the oxidation is dominated by the blueshift. From the numerical analysis, this might be attributed to that the narrowing of the covering angle is faster than the shrinking of the Al shell thickness. The investigations in study paper suggest that the desired optical response can be obtained from Al semishells in high-temperature environments by taking into account the oxide layer growth in Al shells when we determine the thickness of the Al shell. Al semishells are robust plasmonic nanostructures below 400 °C and can be a candidate as facile and economical plasmonic nanostructures with a variety of applications.