Today, commercial and residential buildings account for approximately 30–40% of the total energy consumed globally [1]. To maintain ambient room temperature, air-conditioning systems alone consume $\sim$50% of the building’s total energy need [2]. In the architectural design of buildings, windows emerge as the main bottleneck to achieve thermal insulation. An energy-efficient design of passive windows requires selective transmission of visible and infrared radiation [3]. The visible radiation provides indoor illumination and visual comfort for the building occupants, whereas IR radiation causes radiative heating inside the room [4]. In the solar irradiance spectrum, almost 90% of the solar energy lies in the wavelength spectral regime between 400 nm and 1800 nm, consisting of visible (VIS; 400–750 nm) and infrared (IR; 750–1800 nm) radiation [5]. Therefore, for a typical hot weather condition, passive windows should have high transmission at wavelengths of 400–750 nm and low transmission at wavelengths of 750–1800 nm.

For designing passive windows, various techniques, such as thermochromic and photochromic effects, may be used [6–8]. Unfortunately, the dependence on external stimuli (heat and light) makes it challenging to control IR and visible transmission through these passive windows [2,8]. Nowadays, commercial passive windows are available in single-, double-, or triple-pane glass variants. Some of these glass variants may be coated with low-emissivity coatings to block the thermal blackbody radiation from nearby heated objects [9]. The single-pane glass variants are considered suitable for designing thin and lightweight windows. Unfortunately, most of these single-pane glass varieties show poor IR blocking capability. Therefore, the design of single-pane glass with significant IR blocking capabilities is strongly desired.

In the literature, Besteiro et al. showcased a different theoretical approach by introducing random-sized metallic nanocrystals inside transparent glass to efficiently block IR radiation [10]. Recently, a few passive glass varieties with low-emissivity coatings have been demonstrated to block the solar heat and provide some energy-saving [3,11,12]. Although some of these reported window coatings based on expensive noble metals, such as gold or silver, could match the figure of merit (FOM) used in industry [2,7], none were able to do so when relatively inexpensive alternative metals, such as copper, aluminum, or indium tin oxide, were used [10,12]. In the literature, many broadband absorbers are reported that can completely block infrared radiation, yet blocking infrared radiation while maintaining some visible transmission remains a challenge [13,14]. Therefore, a design of industry-standard passive windows based on relatively inexpensive materials that can efficiently block IR radiation while maintaining standard visible transmission remains largely in demand.

To bridge that gap, in this work, we report a new design of plasmonic “meta-glass” comprising a two-dimensional (2D) hexagonal array of tungsten nanorings placed on top of a silica glass substrate. Using the uniform-sized nanoring-based design, we achieve IR blocking of 80.3% in the wavelength range of 750–1800 nm, while maintaining an average visible transmission of 60%. We further improve the IR blocking efficiency up to 87.1% by introducing two distinct-sized nanorings. We chose tungsten because it has the highest melting point (3422°C), remarkable mechanical strength, and smallest thermal expansion coefficient among all known metals [15]. Unlike gold and silver, tungsten is CMOS compatible, which is crucial for reducing the fabrication cost using conventional lithography techniques for industrial-scale production. In addition to buildings, our plasmonic glass may find application in the sunroof of automobiles, windows of aircraft, sunglasses, and greenhouses, to name a few.

Figure 1 depicts a 3D schematic view of our nanoring-based plasmonic meta-glass along with its unit cell considered for simulation. This meta-glass comprises a 2D hexagonal array of tungsten nanorings printed on top of transparent silica glass. Note that all the dimensions are specified in the caption of Fig. 1. We use a wave optics module of a commercial finite element method (FEM) solver, COMSOL Multiphysics, for full-wave simulations. To capture all the structural details effectively, the maximum and minimum mesh element sizes are taken as 138 nm and $\sim$0.46 nm, respectively. Floquet boundary conditions are applied along the $x$ and $y$ directions to emulate a 2D periodic array of the unit cell.

 

Fig. 1. Nanoring-based plasmonic “meta-glass” for a typical hot weather condition in (a) 3D schematic view and (b) a 2D side view of the unit cell simulation model in the $xz$ plane. The design consists of a 2D hexagonal array of tungsten nanorings placed on top of silica glass. The incident radiation from port 1 is a plane wave of TM polarization traveling along the $z$-direction. A couple of perfectly matched layers (PML 1 and PML 2) are used to absorb undue reflections due to the confinement of the simulation domain. Dimensions in nm: $h_{\mathrm {PML}} = 400$; $h_{\mathrm {air}} = 800$; $h_{\mathrm {ring}} = 30$; and $h_{\mathrm {glass}} = 300$.

Download Full Size | PDF

We use plane wave excitation of TM polarization (electric field parallel to the $xz$ plane) at port 1, representing a “source” and port 2 representing a “destination.” The overall reflection (R) and transmission (T) in percent are calculated using the following two equations: $R = 100\times |S_{\mathrm {11}}|^2$ and $T = 100\times |S_{\mathrm {21}}|^2$, where $S_{\mathrm {11}}$ and $S_{\mathrm {21}}$ are the scattering (S) parameters [16]. These S parameters (for the two-port network system deployed here) are calculated numerically by the FEM solver over the solar spectral regime, i.e., at wavelengths of 400–1800 nm. Using the equation: $A = 100 - R - T$, the absorbance (A) in percent can be further calculated. To support the findings obtained using the FEM solver, we use a finite difference time domain (FDTD) solver. In the 3D FDTD solver, we apply periodic boundary conditions in both the $x$- and $y$-directions and apply perfectly matched layers (PML) along the $z$-direction. The simulation results obtained using the FEM and FDTD solvers are in good agreement. The optical constants for tungsten and silica glass are taken from the literature [17,18].

Design I: Uniform-sized nanorings. For our uniform-sized nanorings-based plasmonic glass design, the top view of the rectangular unit cell in the $xy$ plane is shown using the shaded region in Fig. 2(a). Using manual parameter optimization (discussed in the parametric analysis subsection), we optimize our design to achieve minimum transmission in the IR regime and maximum transmission in the visible regime. All the dimensions of the optimized design are provided in the figure caption. Figure 2(b) shows the numerically obtained transmission spectra using FEM and FDTD solvers over the solar spectral regime, which show a near-perfect match. We get an average transmission of 60.3% in the visible regime with a peak transmission ($T_{\mathrm {max}}$) of 68.2% around $\lambda _1$ = 600 nm. However, in the IR regime, we obtain an average transmission of only 19.7% with a transmission minimum ($T_{\mathrm {min}}$) of 3.6% around $\lambda _2$ = 1000 nm. Therefore, we could block 80.3% of IR radiation over the spectral range of 750–1800 nm using a uniform-sized nanoring-based design.

 

Fig. 2. Uniform-sized nanoring-based plasmonic glass: (a) top view of a unit cell, shown by the shaded region in the $xy$ plane. (b) Numerically calculated transmission (T) spectra using finite element method (FEM) and finite difference time domain (FDTD) solvers for the solar radiation spectrum. Top and side views of (c) normalized electric field distribution with arrow plots at $\lambda _1$ = 600 nm and $\lambda _2$ = 1000 nm. Note that the center and edge positions are highlighted using the white dashed line AOB. Dimensions in nm: inner radius of each ring, $R_{\mathrm {1}}$ = 105; outer radius of each ring, $R_{\mathrm {2}}$ = 145; width of each ring, t = 40; gap between the rings, g = $R_{\mathrm {2}}$/2; lattice constant, a = 2$R_{\mathrm {2}}$+g; period along the $y$-direction, L = 2acos30°; and period along the $x$-direction, W = 2asin30°.

Download Full Size | PDF

To discuss the underlying principle behind suppression in IR transmission, we show the normalized electric field distribution (top view and side views at the center and edge) with arrow plots in Fig. 2(c) corresponding to $T_{\mathrm {max}}$ obtained around $\lambda _1$ and $T_{\mathrm {min}}$ obtained near $\lambda _2$ [see Fig. 2(b)]. The top view at $\lambda _2$ indicates strong inter-ring electric field confinement as a result of dipolar charge separation (see side view at $\lambda _2$), indicating excitation of localized surface plasmons, which leads to significantly suppressed transmission around this resonant wavelength [19,20]. The suppression in the transmission spectrum is primarily because of the enhanced reflection around the resonant wavelength (see Fig. S1 of Supplement 1). However, at $\lambda _1$, the appearance of weak intra-ring electric field confinement (see the top view at $\lambda _1$) indicates coupling between opposite inner faces of each nanoring, allowing moderately high transmission through the gap between different nanoring structures [21].

To optimize our design, we study the effect of variation in the gap (g), width (t), thickness ($h_{\mathrm {ring}}$), and outer radius ($R_{\mathrm {2}}$) of each uniform-sized nanoring, as shown in Fig. 3. With a decrease in the gap among these nanorings between 150 nm and 25 nm, we see a slight redshift in the resonance wavelength and reduced transmission amplitude in the IR regime due to enhanced inter-ring coupling [21], as shown in Fig. 3(a). Next, we vary each nanoring width between 10 nm and 60 nm, keeping the outer radius constant. Figure 3(b) clearly shows a large blueshift in the resonance wavelength with increased nanoring width due to the reduced inner radius dimension. Figure 3(c) shows a similar trend of blueshift with an increase in nanoring thickness between 10 nm and 60 nm. Lastly, Fig. 3(d) shows a redshift in the resonance wavelength owing to an increase in ring dimension with an increase in the outer radius between 100 nm and 200 nm while keeping the nanoring width constant. Therefore, by carefully choosing the nanoring’s dimensions, our meta-glass can efficiently block infrared radiation.

 

Fig. 3. Numerically obtained contour plot for our design with uniform-sized nanorings for varying (a) gap between each nanoring, g, (b) width of each nanoring, t, (c) thickness of each nanoring, $h_{\mathrm {ring}}$, and (d) outer radius of each nanoring, $R_{\mathrm {2}}$. Here, the white dashed lines show the trends for the minimum transmission with changes in the above parameters.

Download Full Size | PDF

To have practical utility, the plasmonic meta-glass should be polarization- and angle-insensitive to a large extent. When the light of arbitrary polarization is incident from the top, it encounters a 2D array of nanorings on top of a planar silica glass substrate that is symmetrical along both the $x$ and $y$ orthogonal directions. Figure 4(a) shows a perfect match between the numerically calculated spectral response for both TM and TE polarization at a normal angle of incidence, and thus means that our design is polarization-insensitive. The spectral response at titled incidence angles for TM, TE, and unpolarized waves is provided in Fig. S2 of Supplement 1. At tilted angles, we observe that the transmission spectra in the IR regime remain unaffected, but there is a drop in transmission in the visible regime due to enhanced reflections. In Fig. 4(b), owing to the azimuthal symmetry of the nanorings, the transmission spectra remain unaffected by the change in polarization angle between 0 and 60 degrees, and thus means that our design is polarization angle-insensitive as well.

 

Fig. 4. Numerically obtained transmission spectra for our uniform-sized nanorings design considering (a) both TM and TE polarization at a normal angle of incidence and (b) contour plot for varying polarization angle between 0 and 60 degrees of the oblique angle of incidence.

Download Full Size | PDF

Design II: Two distinct-sized nanorings. For a uniform-sized nanorings-based design, we observe excitation of surface plasmon resonance at $\lambda _2$. If we could achieve multiple resonances, the entire IR regime can be blocked efficiently [10]. Therefore, we further investigate the spectral response by introducing two distinct-sized nanorings at the center and corner of the unit cell, depicted using a shaded region for choice of unit cells as A or B in Fig. 5(a). Figure 5(b) shows numerically obtained transmission spectra using the FEM and FDTD solvers at wavelengths of 400–1800 nm. We get an average transmission of 59.5% in the visible regime with a peak transmission of 68.7% around $\lambda _1$. However in the IR regime, we obtain an average transmission of only 12.9% with $T_{\mathrm {min}}$ as low as 5.0% at $\lambda _2$ and 6.7% at $\lambda _3$ = 1600 nm. Hence, we could block 87.1% of IR radiation over the spectral range of 750–1800 nm using two distinct-sized nanorings-based design.

 

Fig. 5. Two distinct-sized nanoring-based plasmonic glass depicting (a) choice of unit cells as A or B, shown by shaded region in the $xy$ plane. (b) Numerically calculated transmission (T) spectra using the FEM and FDTD solvers for the solar radiation spectrum. Top and side views of (c) normalized electric field distribution with arrow plots at $\lambda _1$ = 600 nm, $\lambda _2$ = 1000 nm, and $\lambda _3$ = 1600 nm, for choice of unit cell as A. Dimensions in nm: inner radius of corner ring, $r_{\mathrm {1}}$ = 100; outer radius of corner ring, $R_{\mathrm {1}}$ = 120; inner radius of center ring, $r_{\mathrm {2}}$ = 125; outer radius of center ring, $R_{\mathrm {2}}$ = 175; width of corner ring, $t_{1}$ = 20; width of center ring, $t_{2}$ = 50; gap between the rings, g = $R_{\mathrm {2}}$/2; lattice constant, a = 2$R_{\mathrm {2}}$+g; period along the $y$-direction, L = 2acos30°; and period along the $x$-direction, W = 2asin30°.

Download Full Size | PDF

We further study the normalized electric field with arrow plots for the choice of the unit cell as A in Fig. 5(c), corresponding to $T_{\mathrm {min}}$ obtained at $\lambda _2$ and $\lambda _3$, and $T_{\mathrm {max}}$ obtained near $\lambda _1$. The top view at $\lambda _2$ and $\lambda _3$ indicates strong confinement of the electric field due to dipolar charge separation (see side view at $\lambda _2$ for corner ring 1 and $\lambda _3$ for center ring 2). The reason can be attributed to the excitation of localized surface plasmons at $\lambda _2$ for the smaller ring 1 and $\lambda _3$ for the larger ring 2, leading to suppressed transmission in the entire IR regime [19,20]. Interestingly, closely spaced resonance wavelengths lead to more broadband and efficient IR radiation blocking than in uniform-sized nanoring-based plasmonic glass.

Figure of merit. We now introduce a couple of FOMs used in the industry: visible transmittance (VT) and infrared transmittance (IRT) [10]. VT and IRT represent the portion of visible and infrared radiation transmitted through a glass window over a particular wavelength range, respectively [3]. Mathematically, these can be expressed as

Table 1. FOM Comparative Overview with a Few Recently Reported IR Blocking Passive Glass Varieties^a,b

View Table

With the current state-of-the-art nanofabrication technology, the realization of our meta-glass varieties is indeed feasible. First, a polished silica glass substrate is spin-coated with a polymethylmethacrylate resist layer using the electron-beam (e-beam) technique [22]. When the resist is soft-baked, e-beam lithography may be used for exposure resist development. After that, a 30-nm tungsten layer is deposited using the e-beam evaporation technique [23]. After completing the lift-off process, the sample could be rinsed in acetone for a few hours to form nanorings. Considering CMOS compatibility of tungsten and hence, the feasibility of its industrial-scale production in semiconductor foundries, we can expect the overall cost of fabrication to come down eventually in the near future.

To conclude, we present a plasmonic “meta-glass” design that blocks up to $\sim$87% of IR radiation over the spectral range of 750–1800 nm while maintaining an average transmission of 60% in the visible regime. We achieve an infrared transmittance as low as 0.16 and a contrast ratio as high as 3.75, which indicate that our proposed plasmonic glass varieties outperform some of the recently reported window glass designs. Our meta-glass design features selective suppression in the transmission spectra by virtue of the localized surface plasmons in the IR regime. The simulation results obtained using the finite element method and finite difference time domain solvers are in good agreement. Our designs are low-cost, easy to fabricate, and polarization- and angle-insensitive to a large extent. These plasmonic meta-glass varieties based on relatively inexpensive materials could be useful in places with warm climates.