Incandescent Light Bulbs Based on a Refractory Metasurface

Abstract

A thermal radiation light source, such as an incandescent light bulb, is considered a legacy light source with low luminous efficacy. However, it is an ideal energy source converting light with high efficiency from electric power to radiative power. In this work, we evaluate a thermal radiation light source and propose a new type of filament using a refractory metasurface to fabricate an efficient light bulb. We demonstrate visible-light spectral control using a refractory metasurface made of tantalum with an optical microcavity inserted into an incandescent light bulb. We use a nanoimprint method to fabricate the filament that is suitable for mass production. A 1.8 times enhancement of thermal radiation intensity is observed from the microcavity filament compared to the flat filament. Then, we demonstrate the thermal radiation control of the metasurface using a refractory plasmonic cavity made of hafnium nitride. A single narrow resonant peak is observed at the designed wavelength as well as the suppression of thermal radiation in wide mid-IR range under the condition of constant surface temperature.

Full text

photonics
hv
Article
Incandescent Light Bulbs Based on a Refractory
Metasurface †
Hirofumi Toyoda 1, Kazunari Kimino 1, Akihiro Kawano 1and Junichi Takahara 1, 2, *
1Graduate School of Engineering, Osaka University, Osaka 565-0871, Japan;
toyoda@ap.eng.osaka-u.ac.jp (H.T.); kimino@ap.eng.osaka-u.ac.jp (K.K.);
a.kawano@ap.eng.osaka-u.ac.jp (A.K.)
2Photonics Center, Graduate School of Engineering, Osaka University, Osaka 565-0871, Japan
*Correspondence: takahara@ap.eng.osaka-u.ac.jp; Tel.: +81-6-6879-8503
†Invited paper.
Received: 5 September 2019; Accepted: 9 October 2019; Published: 12 October 2019


Abstract:
A thermal radiation light source, such as an incandescent light bulb, is considered a legacy
light source with low luminous efficacy. However, it is an ideal energy source converting light
with high efficiency from electric power to radiative power. In this work, we evaluate a thermal
radiation light source and propose a new type of filament using a refractory metasurface to fabricate
an efficient light bulb. We demonstrate visible-light spectral control using a refractory metasurface
made of tantalum with an optical microcavity inserted into an incandescent light bulb. We use
a nanoimprint method to fabricate the filament that is suitable for mass production. A 1.8 times
enhancement of thermal radiation intensity is observed from the microcavity filament compared to
the flat filament. Then, we demonstrate the thermal radiation control of the metasurface using a
refractory plasmonic cavity made of hafnium nitride. A single narrow resonant peak is observed at
the designed wavelength as well as the suppression of thermal radiation in wide mid-IR range under
the condition of constant surface temperature.
Keywords:
incandescent light bulb; thermal radiation; refractory metal; microcavity; metamaterial;
metasurface; surface plasmon; infrared emitter; nanoimprint
1. Introduction
An incandescent filament shining in a transparent glass bulb is the origin of the beauty of lighting.
The presence of incandescent light can stimulate a brilliant human mind and exert a calming effect.
In addition, incandescent light displays a continuous spectrum of thermal radiation that is attractive
from the point of view of architecture, lighting design, and print color-matching. However, a thermal
radiation light source like an incandescent light bulb is a legacy light source, and in recent years, its
applications have decreased. This is because the luminous efficacy of an incandescent light bulb is only
15 lm
·
W
−1
, which is much lower than the emerging solid-state light sources, viz., the light-emitting
diode (LED) light bulbs. Most of the radiative power emitted from an incandescent light bulb is in
infrared (IR) light. Hence the luminous efficacy of an incandescent light bulb is low compared to an
LED light bulb because the luminous efficacy is limited to the narrow absorption band of the human
eye. Additionally, the lifetime of LED light bulbs is 40,000–50,000 h which is much longer than that of
incandescent light bulbs (1000–2000 h).
In contradiction to the trend favoring the use of LEDs over incandescent light bulbs, we propose
that a thermal radiation light source is an ideal, high-efficiency converter of electric input power to a
radiative output power source. The energy conversion efficiency of incandescent light bulbs is higher
than 90%, which is a result of the Joule heating of a filament [
1
,
2
]. Hence thermal radiation light
Photonics 2019,6, 105; doi:10.3390/photonics6040105 www.mdpi.com/journal/photonics

Photonics 2019,6, 105 2 of 20
sources have great potential for use as efficient light sources. Although the thermal radiation spectrum
obeys Planck’s law that depends on physical constants and temperature, its luminous efficacy can be
improved beyond that suggested by Planck’s law, if the thermal radiation spectra from a filament can
be controlled artificially and optimized to the absorption band of human eyes.
The basic concept for improving the efficiency of an incandescent light bulb by thermal radiation
control of a nanostructured filament was proposed by Waymouth in 1989 [
3
,
4
]. He reasoned that
IR radiation (longer wavelengths) can be suppressed by the cut-offeffect of a microcavity while
visible light (shorter wavelengths) is radiated into free-space, which is analogous to the operation of
a microwave waveguide. This idea is known as the Waymouth hypothesis, and the lamp is called
a “microcavity lamp.” Figure 1shows the concept of the microcavity lamp, where a cuboidal hole
array is formed on the surface of a tungsten (W) filament set into a bulb. Following the viewpoint of
modern photonics, such an artificial surface structure is called a “metasurface,” i.e., a two-dimensional
(2D) metamaterial.
Photonics 2019, 6, x FOR PEER REVIEW 2 of 20
spectrum obeys Planck’s law that depends on physical constants and temperature, its luminous
efficacy can be improved beyond that suggested by Planck’s law, if the thermal radiation spectra
from a filament can be controlled artificially and optimized to the absorption band of human eyes.
The basic concept for improving the efficiency of an incandescent light bulb by thermal radiation
control of a nanostructured filament was proposed by Waymouth in 1989 [3,4]. He reasoned that IR
radiation (longer wavelengths) can be suppressed by the cut-off effect of a microcavity while visible
light (shorter wavelengths) is radiated into free-space, which is analogous to the operation of a
microwave waveguide. This idea is known as the Waymouth hypothesis, and the lamp is called a
“microcavity lamp.” Figure 1 shows the concept of the microcavity lamp, where a cuboidal hole array
is formed on the surface of a tungsten (W) filament set into a bulb. Following the viewpoint of modern
photonics, such an artificial surface structure is called a “metasurface,” i.e., a two-dimensional (2D)
metamaterial.
Figure 1. The concept of a microcavity lamp: an incandescent light bulb with a microcavity array
filament acting as a refractory metasurface.
Independent of the Waymouth hypothesis, the modification of a thermal radiation spectrum
(deviation from Planck’s law) using a microstructured surface was demonstrated first in 1986 by a
deep silicon grating [5]. This was studied further using various types of micro and nanostructures
such as an open-end metallic microcavity [6,7], photonic crystal [8,9], plasmonic cavity [10,11], and
Mie resonators [12]. Additionally, narrow-band thermal radiation has been reported using surface
waves such as surface phonon polaritons [13], surface plasmon polaritons [14,15], spoof surface
plasmons [16], and Tamm plasmons [17]. In 2008, perfect absorbers based on metamaterials were
proposed [18,19]. In addition, thermal radiation emitters based on metasurfaces using metal-
dielectric-metal (MDM) structures were demonstrated in the infrared (IR) range [20,21]. In the last
decade, much research was done on perfect absorbers based on MDM metasurfaces, and their
applications such as thermal radiation emitters or gas sensing in the IR range [22–29]. There are many
papers about thermal radiation control based on metasurfaces in the IR range. A comprehensive list
of these efforts can be found in the literature [30,31].
In contrast to the IR range, there are a few studies about thermal radiation control in the visible
spectrum. This occurs because the melting points of typical plasmonic materials such as gold (Au)
and silver (Ag) are below 1500 K. Additionally, damage during the nanofabrication process induces
a decrease in the melting points of refractory metals due to defects. This reduces the durability of
nanostructures compared with bulk materials. It was reported that Tungsten microcavity structures
degrade the melting point to just above 1500 K, which was then held at 1400 K, less than one-half its
melting point [32]. In 2015, we fabricated a microcavity lamp utilizing nanoimprint technology and
demonstrated the enhancement of visible light using a microcavity filament [33,34]. In 2016, Ilic et al.
reported an efficient thermal light source that radiated only visible light (cut IR light emission from
Figure 1.
The concept of a microcavity lamp: an incandescent light bulb with a microcavity array
filament acting as a refractory metasurface.
Independent of the Waymouth hypothesis, the modification of a thermal radiation spectrum
(deviation from Planck’s law) using a microstructured surface was demonstrated first in 1986 by a deep
silicon grating [
5
]. This was studied further using various types of micro and nanostructures such
as an open-end metallic microcavity [
6
,
7
], photonic crystal [
8
,
9
], plasmonic cavity [
10
,
11
], and Mie
resonators [
12
]. Additionally, narrow-band thermal radiation has been reported using surface waves
such as surface phonon polaritons [
13
], surface plasmon polaritons [
14
,
15
], spoof surface plasmons [
16
],
and Tamm plasmons [
17
]. In 2008, perfect absorbers based on metamaterials were proposed [
18
,
19
].
In addition, thermal radiation emitters based on metasurfaces using metal-dielectric-metal (MDM)
structures were demonstrated in the infrared (IR) range [
20
,
21
]. In the last decade, much research
was done on perfect absorbers based on MDM metasurfaces, and their applications such as thermal
radiation emitters or gas sensing in the IR range [
22
–
29
]. There are many papers about thermal
radiation control based on metasurfaces in the IR range. A comprehensive list of these efforts can be
found in the literature [30,31].
In contrast to the IR range, there are a few studies about thermal radiation control in the visible
spectrum. This occurs because the melting points of typical plasmonic materials such as gold (Au)
and silver (Ag) are below 1500 K. Additionally, damage during the nanofabrication process induces
a decrease in the melting points of refractory metals due to defects. This reduces the durability of
nanostructures compared with bulk materials. It was reported that Tungsten microcavity structures
degrade the melting point to just above 1500 K, which was then held at 1400 K, less than one-half its
melting point [
32
]. In 2015, we fabricated a microcavity lamp utilizing nanoimprint technology and

Photonics 2019,6, 105 3 of 20
demonstrated the enhancement of visible light using a microcavity filament [
33
,
34
]. In 2016, Ilic et al.
reported an efficient thermal light source that radiated only visible light (cut IR light emission from the
filament) by directly sandwiching a dielectric multilayer filter [
35
]. However, it is still challenging to
construct an efficient incandescent light bulb with a filament controlled by a refractory metasurface.
In this paper, we review our recent work on thermal radiation control for incandescent light bulbs
based-on refractory metasurfaces. First, we demonstrate the spectral control of visible light using an
optical microcavity array fabricated on a filament inserted in a light bulb. Here, a nanoimprint method
is used to mass-produce the filaments. Then, to overcome the drawback of a microcavity, we introduce
a refractory metasurface based on a plasmonic cavity made from hafnium nitride (HfN), which is a
new kind of refractory plasmonic material. We demonstrate spectral control in the mid IR range using
this new refractory metasurface.
2. The Efficiency of Thermal Radiation Light Sources
Figure 2shows the power flow for typical commercial light sources: an incandescent light bulb
with and without inert gases (inert gases were not used in old incandescent lamps), a fluorescent lamp,
and an LED light bulb [
1
,
2
]. In Figure 2, we show the ratio (percentage) of output power to input
electric power. Here, all power flow data are taken from [36–40].
Photonics 2019, 6, x FOR PEER REVIEW 3 of 20
the filament) by directly sandwiching a dielectric multilayer filter [35]. However, it is still challenging
to construct an efficient incandescent light bulb with a filament controlled by a refractory metasurface.
In this paper, we review our recent work on thermal radiation control for incandescent light
bulbs based-on refractory metasurfaces. First, we demonstrate the spectral control of visible light
using an optical microcavity array fabricated on a filament inserted in a light bulb. Here, a
nanoimprint method is used to mass-produce the filaments. Then, to overcome the drawback of a
microcavity, we introduce a refractory metasurface based on a plasmonic cavity made from hafnium
nitride (HfN), which is a new kind of refractory plasmonic material. We demonstrate spectral control
in the mid IR range using this new refractory metasurface.
2. The Efficiency of Thermal Radiation Light Sources
Figure 2 shows the power flow for typical commercial light sources: an incandescent light bulb
with and without inert gases (inert gases were not used in old incandescent lamps), a fluorescent
lamp, and an LED light bulb [1,2]. In Figure 2, we show the ratio (percentage) of output power to
input electric power. Here, all power flow data are taken from [36–40].
Figure 2. The power flow ratio (percentage) of typical commercial light sources: (a) an incandescent
light bulb with inert gases (100 W) [36,37], (b) an incandescent light bulb without inert gases (10W)
[38], (c) a fluorescent lamp (40 W) [39], and (d) a light-emitting diode (LED) light bulb (blue LED +
yellow phosphor) [40].
For an LED light bulb, the conversion ratio from input power to visible light is 30–50% while it
is only 10% for an incandescent light bulb with inert gas. The LED light bulb is more efficient than
the incandescent light bulb. The energy loss of an LED light bulb is caused by various physical
processes such as wavelength-conversion losses, inner absorption, or non-radiative phonon
Figure 2.
The power flow ratio (percentage) of typical commercial light sources: (
a
) an incandescent
light bulb with inert gases (100 W) [
36
,
37
], (
b
) an incandescent light bulb without inert gases (10W) [
38
],
(
c
) a fluorescent lamp (40 W) [
39
], and (
d
) a light-emitting diode (LED) light bulb (blue LED +yellow
phosphor) [40].

Photonics 2019,6, 105 4 of 20
For an LED light bulb, the conversion ratio from input power to visible light is 30–50% while
it is only 10% for an incandescent light bulb with inert gas. The LED light bulb is more efficient
than the incandescent light bulb. The energy loss of an LED light bulb is caused by various physical
processes such as wavelength-conversion losses, inner absorption, or non-radiative phonon excitation,
resulting in the dissipation of energy to the environment around the bulb. However, considering the
conversion ratio from input power to total electromagnetic radiation, it is higher than 80% for the
incandescent light bulb. Additionally, it exceeds 90% (~94%) for an incandescent light bulb without
inert gas. This latter result is obtained because the loss of heat conduction from the filament to inert
gas is negligible. Thus, we can conclude that a thermal radiation light source is an ideal high-efficiency
energy converter from input electric power to output radiative power. If we suppress the IR light and
convert it to visible light, incandescent light bulbs can be recreated as an efficient light source.
3. The Basic Principle of Thermal Radiation Control by a Refractory Metasurface
There are two ways to suppress IR light from incandescent lamps: (i) use of an optical filter coated
on a bulb and (ii) thermal radiation control of a filament. For optical filters, dielectric multilayers
are used for short-pass (IR rejection) optical filters. Although short-pass optical filters are used for
commercially available halogen light bulbs, the shapes of blubs are limited to elliptical, and the
transparency of the bulb is reduced due to coloring of the dielectric multilayers, resulting in a reduction
of the beauty of incandescent light bulbs. In contrast, using thermal radiation control, we can modify
the thermal radiation spectrum for a filament directly by forming nanostructures on it. In this method,
IR light is suppressed, and visible light is enhanced from a filament directly, resulting in a significant
improvement in the luminous efficacy.
Spectral radiant intensity I
bb
(
λ
,T) [W
·
m
−2·
m
−1·
sr
−1
] of blackbody radiation per area and per solid
angle at temperature Tand wavelength λis given by Equation (1):
Ibb(λ,T)=2hc2
λ5
1
ehc/λkBT−1, (1)
where cis the speed of light, his the Planck constant, and k
B
is the Boltzmann constant. The thermal
radiation spectrum from a real surface can be calculated by the product of I
bb
(
λ
,T) and spectral
emissivity
ε
(
λ
). Hence, we can control the radiation spectrum artificially by specifying
ε
(
λ
). This is the
basic principle of thermal radiation control.
In a microcavity lamp,
ε
(
λ
) can be controlled by a microcavity array formed on the surface of a
refractory metal filament. Figure 3a shows a schematic view of a cuboidal hole microcavity array. Such
a cuboidal hole behaves as an open-end cavity for optical electromagnetic (EM) fields, and they are
confined inside the hole. Contrary to the Waymouth hypothesis, previous experimental studies in
the IR range demonstrated that a microcavity enhances thermal radiation at specific wavelengths by
resonance instead of suppression by the cut-offeffect [
6
,
7
]. The resonant wavelength of the microcavity
(Figure 3a) is given by Equation (2):
λ=2
qnx
a2+ny
a2+nz
2d2, (2)
where n
x
,n
y
=0, 1, 2, 3
. . .
are mode numbers of the x- or y- (horizontal) direction, respectively, and
nz=0, 1, 3, 5 . . .
is the mode number of the z- (vertical) direction, aand dare the width in the x-y
direction and the depth of the cuboidal cavity, respectively [5,6].

Photonics 2019,6, 105 5 of 20
Photonics 2019, 6, x FOR PEER REVIEW 5 of 20
Figure 3. Principle of thermal radiation control by a metasurface: (a) an array of a microcavity on a
refractory metal filament, (b) resonant modes for n = 1, 3, and 5 inside a microcavity with perfect
conductor walls, and (c) the thermal radiation spectrum can be controlled by the product of spectral
emissivity of the metasurface and Planck’s law.
Figure 3b shows typical resonant modes in the cavity. In principle, such resonant modes can
enhance absorption at the resonant wavelengths. According to Kirchhoff’s law, such resonant
absorption increases the emissivity of opaque materials in thermal radiation. On a metallic surface,
emissivity is very low at non-resonant wavelengths, as shown in Figure 3c, resulting in a steep
resonant enhancement of the spectral emissivity ε(λ). Hence, the total radiation spectrum can be
controlled by ε(λ).
4. Thermal Radiation Control by a Microcavity Array
4.1. Fabrication by Nanoimprint
To build a microcavity lamp, we fabricated microcavity array structures on a refractory metal
substrate, cut it into filament strips then inserted them into incandescent light bulbs. In the
nanofabrication process, we used a nanoimprint method to fabricate microcavity array patterns
looking forward to a mass-production process. The details of the fabrication process are shown in
Appendix A.
Figure 4a shows a photograph of a 20 × 20 mm polished tantalum (Ta) substrate with a thickness
of 100 µm, on which microcavity structures are formed. The structures were fabricated on a single
side or both sides of the substrate. Figure 4b shows a scanning ion microscope (SIM) image of the
structures. The pattern sizes of the mold are width a = 300 nm, depth d = ~200 nm, and period P = 600
nm. From Figure 4b it is confirmed that a 350-nm-squared cuboidal microcavity with P = 600 nm
formed on the substrate. The depth of the cavity is estimated to be ~280 nm by the slanted angle of
the SIM image at 30°. The measured depth is shallower than the designed depth of 500 nm. This is
because the depth is limited by the differences in the dry etching rate between the Cr mask and the
Ta substrate. After fabricating the pattern, the substrate was cut into strips (length: 20 mm, width:
500 µm) using a dicing saw. A single strip was placed into two holding stems to form a filament by
welding into a bulb made of Pyrex glass (borosilicate glass). Inert gases (75% Ar and 25% N
2
) were
put into the bulb. As a result, we prepared two types of light bulbs with a structure on both sides and
a single side. We also prepared a light bulb with a flat filament for reference.
Figure 3.
Principle of thermal radiation control by a metasurface: (
a
) an array of a microcavity on
a refractory metal filament, (
b
) resonant modes for n=1, 3, and 5 inside a microcavity with perfect
conductor walls, and (
c
) the thermal radiation spectrum can be controlled by the product of spectral
emissivity of the metasurface and Planck’s law.
Figure 3b shows typical resonant modes in the cavity. In principle, such resonant modes
can enhance absorption at the resonant wavelengths. According to Kirchhoff’s law, such resonant
absorption increases the emissivity of opaque materials in thermal radiation. On a metallic surface,
emissivity is very low at non-resonant wavelengths, as shown in Figure 3c, resulting in a steep resonant
enhancement of the spectral emissivity
ε
(
λ
). Hence, the total radiation spectrum can be controlled
by ε(λ).
4. Thermal Radiation Control by a Microcavity Array
4.1. Fabrication by Nanoimprint
To build a microcavity lamp, we fabricated microcavity array structures on a refractory metal
substrate, cut it into filament strips then inserted them into incandescent light bulbs. In the
nanofabrication process, we used a nanoimprint method to fabricate microcavity array patterns
looking forward to a mass-production process. The details of the fabrication process are shown in
Appendix A.
Figure 4a shows a photograph of a 20
×
20 mm polished tantalum (Ta) substrate with a thickness
of 100
µ
m, on which microcavity structures are formed. The structures were fabricated on a single side
or both sides of the substrate. Figure 4b shows a scanning ion microscope (SIM) image of the structures.
The pattern sizes of the mold are width a=300 nm, depth d=~200 nm, and period
P=600 nm
. From
Figure 4b it is confirmed that a 350-nm-squared cuboidal microcavity with P=600 nm formed on the
substrate. The depth of the cavity is estimated to be ~280 nm by the slanted angle of the SIM image at
30
◦
. The measured depth is shallower than the designed depth of 500 nm. This is because the depth
is limited by the differences in the dry etching rate between the Cr mask and the Ta substrate. After
fabricating the pattern, the substrate was cut into strips (length: 20 mm, width: 500
µ
m) using a dicing
saw. A single strip was placed into two holding stems to form a filament by welding into a bulb made
of Pyrex glass (borosilicate glass). Inert gases (75% Ar and 25% N
2
) were put into the bulb. As a result,

Photonics 2019,6, 105 6 of 20
we prepared two types of light bulbs with a structure on both sides and a single side. We also prepared
a light bulb with a flat filament for reference.
Photonics 2019, 6, x FOR PEER REVIEW 6 of 20
Figure 4. Microcavity filament: (a) 20-mm-squared Ta substrate and its split into strips using a dicing-
saw process and (b) scanning ion microscope (SIM) image of the microcavity. The horizontal scale bar
is 350 nm.
Figure 5a shows a photograph of a microcavity lamp prototype. The fact that rainbow colors are
seen on the filament shows that periodic structures are formed successfully on the filament. As shown
in Figure 5b, the light bulb was set into an E26 socket, and it emitted visible light from the filament
by connecting an electric power supply.
Figure 5. A prototype of the microcavity lamp: (a) turning off and (b) on.
4.2. Measurements
Measurements of thermal radiation spectra were performed using an integrating sphere for
collecting the total luminous flux. A light bulb with a microcavity filament was set into the integrating
sphere (LMS-200, Labsphere, Inc., North Sutton, NH, USA) with a diameter of 25 cm. The radiation
spectra were measured using a fiber multichannel spectrometer (QE65Pro, Ocean Optics, Inc., Largo,
FL, USA) over the wavelength range of 500–1100 nm. A voltage source was used to heat the filament
under a constant DC voltage of 1.5 V, where the two-terminal resistance of the light bulb was ~0.1 Ω
at room temperature. Next, the light bulb with a flat filament (without a microcavity) was measured
under the same conditions as the reference. All measurements were done under a constant electric
power of 7.9 W.
We note that the conditions for measuring thermal radiation spectra should be identical for all
samples. Two measurement conditions are standard: (i) constant temperature mode and (ii) constant
power mode. In constant temperature mode, the radiation spectra are measured, maintaining the
same filament temperature for all samples. In constant power mode, radiation spectra are measured
Figure 4.
Microcavity filament: (
a
) 20-mm-squared Ta substrate and its split into strips using a
dicing-saw process and (
b
) scanning ion microscope (SIM) image of the microcavity. The horizontal
scale bar is 350 nm.
Figure 5a shows a photograph of a microcavity lamp prototype. The fact that rainbow colors are
seen on the filament shows that periodic structures are formed successfully on the filament. As shown
in Figure 5b, the light bulb was set into an E26 socket, and it emitted visible light from the filament by
connecting an electric power supply.
Photonics 2019, 6, x FOR PEER REVIEW 6 of 20
Figure 4. Microcavity filament: (a) 20-mm-squared Ta substrate and its split into strips using a dicing-
saw process and (b) scanning ion microscope (SIM) image of the microcavity. The horizontal scale bar
is 350 nm.
Figure 5a shows a photograph of a microcavity lamp prototype. The fact that rainbow colors are
seen on the filament shows that periodic structures are formed successfully on the filament. As shown
in Figure 5b, the light bulb was set into an E26 socket, and it emitted visible light from the filament
by connecting an electric power supply.
Figure 5. A prototype of the microcavity lamp: (a) turning off and (b) on.
4.2. Measurements
Measurements of thermal radiation spectra were performed using an integrating sphere for
collecting the total luminous flux. A light bulb with a microcavity filament was set into the integrating
sphere (LMS-200, Labsphere, Inc., North Sutton, NH, USA) with a diameter of 25 cm. The radiation
spectra were measured using a fiber multichannel spectrometer (QE65Pro, Ocean Optics, Inc., Largo,
FL, USA) over the wavelength range of 500–1100 nm. A voltage source was used to heat the filament
under a constant DC voltage of 1.5 V, where the two-terminal resistance of the light bulb was ~0.1 Ω
at room temperature. Next, the light bulb with a flat filament (without a microcavity) was measured
under the same conditions as the reference. All measurements were done under a constant electric
power of 7.9 W.
We note that the conditions for measuring thermal radiation spectra should be identical for all
samples. Two measurement conditions are standard: (i) constant temperature mode and (ii) constant
power mode. In constant temperature mode, the radiation spectra are measured, maintaining the
same filament temperature for all samples. In constant power mode, radiation spectra are measured
Figure 5. A prototype of the microcavity lamp: (a) turning offand (b) on.
4.2. Measurements
Measurements of thermal radiation spectra were performed using an integrating sphere for
collecting the total luminous flux. A light bulb with a microcavity filament was set into the integrating
sphere (LMS-200, Labsphere, Inc., North Sutton, NH, USA) with a diameter of 25 cm. The radiation
spectra were measured using a fiber multichannel spectrometer (QE65Pro, Ocean Optics, Inc., Largo,
FL, USA) over the wavelength range of 500–1100 nm. A voltage source was used to heat the filament
under a constant DC voltage of 1.5 V, where the two-terminal resistance of the light bulb was ~0.1
Ω
at
room temperature. Next, the light bulb with a flat filament (without a microcavity) was measured
under the same conditions as the reference. All measurements were done under a constant electric
power of 7.9 W.

Photonics 2019,6, 105 7 of 20
We note that the conditions for measuring thermal radiation spectra should be identical for all
samples. Two measurement conditions are standard: (i) constant temperature mode and (ii) constant
power mode. In constant temperature mode, the radiation spectra are measured, maintaining the
same filament temperature for all samples. In constant power mode, radiation spectra are measured
maintaining constant electric power to heat a filament. Since it is difficult to directly measure the
temperature of a filament inside a light bulb, we used constant power mode here.
4.3. Results and Discussion
Figure 6shows the results of the thermal radiation spectra of the total flux from two light bulbs:
a light bulb with a two-sided microcavity filament (
Φc
(
λ
)) and one with a flat filament (
ΦF
(
λ
)) [
33
].
We can clearly see that the total flux of the microcavity filament is higher than the flat filament. This
suggests that the emissivity of the microcavity filament increases compared with the flat filament
because the temperature of both filaments was almost identical due to the use of the same input power.
Photonics 2019, 6, x FOR PEER REVIEW 7 of 20
maintaining constant electric power to heat a filament. Since it is difficult to directly measure the
temperature of a filament inside a light bulb, we used constant power mode here.
4.3. Results and Discussion
Figure 6 shows the results of the thermal radiation spectra of the total flux from two light bulbs:
a light bulb with a two-sided microcavity filament (Φ
c
(λ)) and one with a flat filament (Φ
F
(λ)) [33].
We can clearly see that the total flux of the microcavity filament is higher than the flat filament. This
suggests that the emissivity of the microcavity filament increases compared with the flat filament
because the temperature of both filaments was almost identical due to the use of the same input
power.
Figure 6. Thermal radiation spectra of the total flux from a microcavity surface (solid line) and flat
surface (dotted line). The ratio of total flux (solid red line) is also plotted, representing the
enhancement factor.
To analyze the enhancement mechanism, we plotted the enhancement factor defined as
Φ
c
(λ)/Φ
f
(λ), as shown in Figure 6. In the enhancement factor plot, we observe a single broad peak at
~700 nm. The enhancement factor physically means relative spectral emissivity, which is defined as
the ratio of the spectral emissivity of a microcavity surface to that of a flat surface: i.e., ε
c
(λ)/ε
f
(λ),
where ε
c
(λ) and ε
f
(λ) are spectral emissivities at λ of the microcavity and the flat surface, respectively.
4.4. Simulated Results
To analyze the enhancement effect quantitatively, we performed numerical calculations on the
spectral absorptivity for the microcavity filament versus the depth of the cavity using a commercially
available numerical simulator and the rigorous coupled-wave analysis (RCWA) method (Diffract
Mod, RSoft Inc.).
Figure 7a shows the spectral map, α(λ,d), of the calculated absorptivity versus the depth, d, of
the cavity. We see that α(λ,d) has a peak at ~600–900 nm with an α value of 0.9, which then decreases
to ~0.1 at λ > 1.0 µm. These peaks in absorptivity are attributed to the resonant modes inside a single
microcavity, and the rapid decrease is due to the cut-off effect in the cavity. However, no peak
structure is observed in the measured spectrum, Φ
c
(λ), as shown in Figure 6.
Figure 6.
Thermal radiation spectra of the total flux from a microcavity surface (solid line) and
flat surface (dotted line). The ratio of total flux (solid red line) is also plotted, representing the
enhancement factor.
To analyze the enhancement mechanism, we plotted the enhancement factor defined as
Φc
(
λ
)/
Φf
(
λ
),
as shown in Figure 6. In the enhancement factor plot, we observe a single broad peak at ~700 nm. The
enhancement factor physically means relative spectral emissivity, which is defined as the ratio of the
spectral emissivity of a microcavity surface to that of a flat surface: i.e.,
εc
(
λ
)/
εf
(
λ
), where
εc
(
λ
) and
εf(λ) are spectral emissivities at λof the microcavity and the flat surface, respectively.
4.4. Simulated Results
To analyze the enhancement effect quantitatively, we performed numerical calculations on the
spectral absorptivity for the microcavity filament versus the depth of the cavity using a commercially
available numerical simulator and the rigorous coupled-wave analysis (RCWA) method (Diffract Mod,
RSoft Inc.).
Figure 7a shows the spectral map,
α
(
λ
,d), of the calculated absorptivity versus the depth, d, of the
cavity. We see that
α
(
λ
,d) has a peak at ~600–900 nm with an
α
value of 0.9, which then decreases to

Photonics 2019,6, 105 8 of 20
~0.1 at
λ
>1.0
µ
m. These peaks in absorptivity are attributed to the resonant modes inside a single
microcavity, and the rapid decrease is due to the cut-offeffect in the cavity. However, no peak structure
is observed in the measured spectrum, Φc(λ), as shown in Figure 6.
Photonics 2019, 6, x FOR PEER REVIEW 8 of 20
Figure 7. Simulated spectral maps: (a) spectral absorptivity/emissivity of a Ta microcavity
metasurface to the depth of a microcavity with w = 350 nm and P = 600 nm and (b) relative spectral
absorptivity/emissivity for the flat surface of Ta.
To compare the simulation to the experimental results, we calculated the relative spectral
absorptivity, which is equal to the relative spectral emissivity, ε
c
(λ)/ε
f
(λ), from Kirchhoff’s law. By
taking the ratio of absorptivity to the flat surface, we obtain the relative spectral
absorptivity/emissivity map α(λ,d)/α(λ,0) shown in Figure 7b. Note that even a flat Ta surface has
moderate broad absorption of α = ~0.5 at λ < 0.6 µm. We see that absorption enhancement occurs at
~800 nm, and the relative absorptivity increases as the cavity depth increases, as shown in Figure 7b.
At a sample depth of d = 280 nm in Figure 4b, the peak position for the relative emissivity is at ~700–
900 nm (Figure 7b). This is consistent with the peak position of the experiment in Figure 6. Thus, the
broad peak observed for the relative emissivity is attributed to the microcavity effect.
If we can fabricate a sufficiently deep microcavity with d > 500 nm, we expect that the thermal
radiation spectrum will have a narrower resonant peak at ~850 nm and its relative absorptivity will
increase to five, as seen in Figure 7b. However, increasing the cavity depth further is difficult due to
the limit of the dry etching process using refractory metals. As a Ta substrate is a hard material
compared with Si, the etching contrast ratio between the resist mask and the Ta substrate is not
sufficient for deeper etching (see Appendix A). Besides, it is challenging to control the absorptivity
in the visible range because typical refractory metals such as Ta, Mo, and W are “dielectric” from the
negative value of the dielectric, resulting in absorption in the visible spectrum (see Appendix B) [41].
Actually, even a flat Ta surface (d = 0) has an absorption of α = ~0.5 at λ = 0.4–0.7 µm, as shown in
Figure 7a. This means that these metals are far from perfect conductors in the visible range. Hence, a
new kind of metasurface is needed beyond the performance of a microcavity array to enhance further
the emissivity in the visible spectrum. Plasmonic materials and its metasurfaces are needed beyond
conventional refractory metals to control the thermal radiation spectra in the visible range.
5. Thermal Radiation Control by a Refractory Plasmonic Metasurface
5.1. Thermal Radiation Control by Plasmonic Cavities
As described in Section 4, we achieved thermal radiation control using a microcavity filament in
the visible range. As a next step, we propose a new kind of filament using a plasmonic metasurface,
as illustrated in Figure 8. Figure 8 shows the concept of a plasmonic metasurface where thick
microcavities on the refractory metal are replaced by very thin MDM plasmonic cavities. A plasmonic
resonator is very thin (<<λ) compared with the wavelength while a microcavity needs a deep trench
structure on the order of the controlled optical wavelength (~λ). Since the thickness of the metasurface
Figure 7.
Simulated spectral maps: (
a
) spectral absorptivity/emissivity of a Ta microcavity metasurface
to the depth of a microcavity with w=350 nm and P=600 nm and (
b
) relative spectral
absorptivity/emissivity for the flat surface of Ta.
To compare the simulation to the experimental results, we calculated the relative spectral
absorptivity, which is equal to the relative spectral emissivity,
εc
(
λ
)/
εf
(
λ
), from Kirchhoff’s law. By taking
the ratio of absorptivity to the flat surface, we obtain the relative spectral absorptivity/emissivity map
α
(
λ
,d)/
α
(
λ
,0) shown in Figure 7b. Note that even a flat Ta surface has moderate broad absorption
of
α
=~0.5 at
λ
<0.6
µ
m. We see that absorption enhancement occurs at ~800 nm, and the relative
absorptivity increases as the cavity depth increases, as shown in Figure 7b. At a sample depth of
d=280 nm
in Figure 4b, the peak position for the relative emissivity is at ~700–900 nm (Figure 7b).
This is consistent with the peak position of the experiment in Figure 6. Thus, the broad peak observed
for the relative emissivity is attributed to the microcavity effect.
If we can fabricate a sufficiently deep microcavity with d>500 nm, we expect that the thermal
radiation spectrum will have a narrower resonant peak at ~850 nm and its relative absorptivity will
increase to five, as seen in Figure 7b. However, increasing the cavity depth further is difficult due to the
limit of the dry etching process using refractory metals. As a Ta substrate is a hard material compared
with Si, the etching contrast ratio between the resist mask and the Ta substrate is not sufficient for
deeper etching (see Appendix A). Besides, it is challenging to control the absorptivity in the visible
range because typical refractory metals such as Ta, Mo, and W are “dielectric” from the negative value
of the dielectric, resulting in absorption in the visible spectrum (see Appendix B) [
41
]. Actually, even a
flat Ta surface (d=0) has an absorption of
α
=~0.5 at
λ
=0.4–0.7
µ
m, as shown in Figure 7a. This
means that these metals are far from perfect conductors in the visible range. Hence, a new kind of
metasurface is needed beyond the performance of a microcavity array to enhance further the emissivity
in the visible spectrum. Plasmonic materials and its metasurfaces are needed beyond conventional
refractory metals to control the thermal radiation spectra in the visible range.
5. Thermal Radiation Control by a Refractory Plasmonic Metasurface
5.1. Thermal Radiation Control by Plasmonic Cavities
As described in Section 4, we achieved thermal radiation control using a microcavity filament in
the visible range. As a next step, we propose a new kind of filament using a plasmonic metasurface,
as illustrated in Figure 8. Figure 8shows the concept of a plasmonic metasurface where thick
microcavities on the refractory metal are replaced by very thin MDM plasmonic cavities. A plasmonic

Photonics 2019,6, 105 9 of 20
resonator is very thin (<<
λ
) compared with the wavelength while a microcavity needs a deep trench
structure on the order of the controlled optical wavelength (~
λ
). Since the thickness of the metasurface
is much smaller than the wavelength, the heat capacity is small and is compatible with a planar
fabrication process. However, the melting point of conventional plasmonic metals such as Ag and Au
are not high enough for thermal radiation control in the visible spectrum.
Photonics 2019, 6, x FOR PEER REVIEW 9 of 20
is much smaller than the wavelength, the heat capacity is small and is compatible with a planar
fabrication process. However, the melting point of conventional plasmonic metals such as Ag and
Au are not high enough for thermal radiation control in the visible spectrum.
Figure 8. Refractory metasurface (a) from a microcavity array to (b) a plasmonic cavity array.
In recent years, nitride ceramics such as titanium nitride (TiN) have been proposed and studied
as new plasmonic materials operating at higher temperatures (T > 1500 K [42–45]). Melting points of
typical plasmonic materials and nitride ceramics are summarized in Table 1. The melting points of
these materials are similar to conventional refractory metals, and the permittivity of these materials
is negative in the visible range. Hence, those are called “refractory plasmonic materials.”
Table 1. Refractory metals and refractory plasmonic materials in order of its melting point.
Material Melting Point (K) Permittivity in Visible Range
Ag 1235 ND
Au 1337 ND
SiO
2
1983 D
Mo 2896 D
HfO
2
3031 D
TiN 3203 ND
Ta 3290 D/ND
HfN 3607 ND
W 3695 D
1
ND: Negative Dielectric; D: Dielectric.
In this study, we used hafnium nitride (HfN) since the melting point of HfN is higher than that
of TiN and it is the same order as W. The most crucial property of nitride ceramics is that its
permittivity is negative in the visible range like the noble metals. Such a feature is useful for
plasmonic materials. The spectral permittivity of conventional and plasmonic refractory metals are
shown in Appendix C. If we realize plasmonic metasurfaces using plasmonic refractory materials
instead of noble metals, we can control the thermal radiation spectra more precisely and obtain higher
Q-value of the plasmonic cavity than that of the microcavity.
Figure 9 shows a schematic of a cross-sectional view of a refractory MDM metasurface, where
the Fabry–Pérot (FP) plasmonic resonator disk type based on HfN are arranged in a periodic array.
The diameter d, the period P of the resonator, the gap thickness in the dielectric layer, T
g
, and the top
metal layer (HfN) T
d
are shown in Figure 9. We note that the dielectric layer should be selected in
accordance with the operating temperature T as such that HfO
2
for T > 2000 K or SiO
2
for T < 2000 K.
Figure 8. Refractory metasurface (a) from a microcavity array to (b) a plasmonic cavity array.
In recent years, nitride ceramics such as titanium nitride (TiN) have been proposed and studied as
new plasmonic materials operating at higher temperatures (T>1500 K [
42
–
45
]). Melting points of
typical plasmonic materials and nitride ceramics are summarized in Table 1. The melting points of
these materials are similar to conventional refractory metals, and the permittivity of these materials is
negative in the visible range. Hence, those are called “refractory plasmonic materials.”
Table 1. Refractory metals and refractory plasmonic materials in order of its melting point.
Material Melting Point (K) Permittivity in Visible Range
Ag 1235 ND
Au 1337 ND
SiO21983 D
Mo 2896 D
HfO23031 D
TiN 3203 ND
Ta 3290 D/ND
HfN 3607 ND
W 3695 D
ND: Negative Dielectric; D: Dielectric.
In this study, we used hafnium nitride (HfN) since the melting point of HfN is higher than that of
TiN and it is the same order as W. The most crucial property of nitride ceramics is that its permittivity
is negative in the visible range like the noble metals. Such a feature is useful for plasmonic materials.
The spectral permittivity of conventional and plasmonic refractory metals are shown in Appendix C.
If we realize plasmonic metasurfaces using plasmonic refractory materials instead of noble metals, we
can control the thermal radiation spectra more precisely and obtain higher Q-value of the plasmonic
cavity than that of the microcavity.
Figure 9shows a schematic of a cross-sectional view of a refractory MDM metasurface, where
the Fabry–P
é
rot (FP) plasmonic resonator disk type based on HfN are arranged in a periodic array.
The diameter d, the period Pof the resonator, the gap thickness in the dielectric layer, Tg, and the top
metal layer (HfN) T
d
are shown in Figure 9. We note that the dielectric layer should be selected in
accordance with the operating temperature Tas such that HfO
2
for T>2000 K or SiO
2
for T<2000 K.

Photonics 2019,6, 105 10 of 20
Photonics 2019, 6, x FOR PEER REVIEW 10 of 20
Figure 9. A schematic and cross-sectional view of a metal-dielectric-metal (MDM) metasurface based
on hafnium nitride (HfN).
To confirm the efficiency of thermal radiation control by this refractory plasmonic metasurface,
we calculated the theoretical radiation spectrum of the MDM metasurface and compared it to a
blackbody surface under the condition that both radiation powers are identical, i.e., a constant power
mode as described in Section 4.2. Figure 10 shows the simulated results for the thermal radiation
spectrum obtained from the metasurface composed of HfN and HfO
2
at T = 2500 K (red line) with d
= 40 nm, P = 80 nm, T
g
= 60 nm, and T
d
= 20 nm. Here, we observe that the highest power radiated
from the metasurface is focused on the resonant peak at ~700 nm with a full-width half-maximum
(FWHM) value of 571 nm due to the plasmonic resonance in an FP resonator disk. From Figure 10,
the equivalent power from the metasurface at T = 2500 K corresponds to the power from a blackbody
at T = 1777 K. This means that radiated power from the metasurface at T = 2500 K equals that from
the blackbody at only T = 1777 K. According to the Stefan–Boltzmann law, the efficiency is improved
by a factor of (2500/1700)
4
= 3.9; i.e., the metasurface is 3.9 times more efficient than the blackbody
from the viewpoint of power consumption. Additionally, from Figure 10, the radiation intensity at
the plasmonic resonant wavelength is more than 10 times greater than that of the blackbody at T =
1777 K.
Figure 10. Simulated thermal radiation spectra in constant power mode: radiation spectra from the
MDM metasurface (red line) composed of HfN and HfO
2
at T = 2500 K with d = 40 nm, P = 80 nm, T
g
= 60 nm, T
d
= 20 nm, and the reference blackbody (blue line) at T = 1777 K.
5.2. Fabrication
To demonstrate the thermal radiation control by a refractory plasmonic metasurface, we
fabricated MDM metasurfaces based on HfN. In this study, we designed the metasurface to be a
perfect absorber in the mid-IR range (~4 µm) instead of in the visible as the first step towards the
Figure 9. A schematic and cross-sectional view of a metal-dielectric-metal (MDM) metasurface based
on hafnium nitride (HfN).
To confirm the efficiency of thermal radiation control by this refractory plasmonic metasurface, we
calculated the theoretical radiation spectrum of the MDM metasurface and compared it to a blackbody
surface under the condition that both radiation powers are identical, i.e., a constant power mode as
described in Section 4.2. Figure 10 shows the simulated results for the thermal radiation spectrum
obtained from the metasurface composed of HfN and HfO
2
at T=2500 K (red line) with d=40 nm,
P=80 nm
,T
g
=60 nm, and T
d
=20 nm. Here, we observe that the highest power radiated from the
metasurface is focused on the resonant peak at ~700 nm with a full-width half-maximum (FWHM)
value of 571 nm due to the plasmonic resonance in an FP resonator disk. From Figure 10, the equivalent
power from the metasurface at T=2500 K corresponds to the power from a blackbody at T=1777 K.
This means that radiated power from the metasurface at T=2500 K equals that from the blackbody
at only T=1777 K. According to the Stefan–Boltzmann law, the efficiency is improved by a factor of
(2500/1700)
4
=3.9; i.e., the metasurface is 3.9 times more efficient than the blackbody from the viewpoint
of power consumption. Additionally, from Figure 10, the radiation intensity at the plasmonic resonant
wavelength is more than 10 times greater than that of the blackbody at T=1777 K.
Photonics 2019, 6, x FOR PEER REVIEW 10 of 20
Figure 9. A schematic and cross-sectional view of a metal-dielectric-metal (MDM) metasurface based
on hafnium nitride (HfN).
To confirm the efficiency of thermal radiation control by this refractory plasmonic metasurface,
we calculated the theoretical radiation spectrum of the MDM metasurface and compared it to a
blackbody surface under the condition that both radiation powers are identical, i.e., a constant power
mode as described in Section 4.2. Figure 10 shows the simulated results for the thermal radiation
spectrum obtained from the metasurface composed of HfN and HfO
2
at T = 2500 K (red line) with d
= 40 nm, P = 80 nm, T
g
= 60 nm, and T
d
= 20 nm. Here, we observe that the highest power radiated
from the metasurface is focused on the resonant peak at ~700 nm with a full-width half-maximum
(FWHM) value of 571 nm due to the plasmonic resonance in an FP resonator disk. From Figure 10,
the equivalent power from the metasurface at T = 2500 K corresponds to the power from a blackbody
at T = 1777 K. This means that radiated power from the metasurface at T = 2500 K equals that from
the blackbody at only T = 1777 K. According to the Stefan–Boltzmann law, the efficiency is improved
by a factor of (2500/1700)
4
= 3.9; i.e., the metasurface is 3.9 times more efficient than the blackbody
from the viewpoint of power consumption. Additionally, from Figure 10, the radiation intensity at
the plasmonic resonant wavelength is more than 10 times greater than that of the blackbody at T =
1777 K.
Figure 10. Simulated thermal radiation spectra in constant power mode: radiation spectra from the
MDM metasurface (red line) composed of HfN and HfO
2
at T = 2500 K with d = 40 nm, P = 80 nm, T
g
= 60 nm, T
d
= 20 nm, and the reference blackbody (blue line) at T = 1777 K.
5.2. Fabrication
To demonstrate the thermal radiation control by a refractory plasmonic metasurface, we
fabricated MDM metasurfaces based on HfN. In this study, we designed the metasurface to be a
perfect absorber in the mid-IR range (~4 µm) instead of in the visible as the first step towards the
Figure 10.
Simulated thermal radiation spectra in constant power mode: radiation spectra from the
MDM metasurface (red line) composed of HfN and HfO
2
at T=2500 K with d=40 nm, P=80 nm,
Tg=60 nm, Td=20 nm, and the reference blackbody (blue line) at T=1777 K.
5.2. Fabrication
To demonstrate the thermal radiation control by a refractory plasmonic metasurface, we fabricated
MDM metasurfaces based on HfN. In this study, we designed the metasurface to be a perfect absorber
in the mid-IR range (~4
µ
m) instead of in the visible as the first step towards the fabrication of a

Photonics 2019,6, 105 11 of 20
“plasmonic” thermal radiation light source. To design and optimize the size parameters for a perfect
absorber operating at ~4
µ
m, we performed numerical simulations using the commercially available
finite-difference time-domain method (FDTD) software (Lumerical Inc., Vancouver, BC, Canada) for
the metasurface composed of HfN and SiO
2
as shown in Appendix F. From Figure A6, the designed
value of diameter d=1.2 µm was determined for achieving the absorption peak of 4 µm.
Figure 11 shows the MDM metasurface sample with d=1.14
µ
m, P=2.0
µ
m, T
g
=130 nm, and
T
d
=200 nm. The metasurface was fabricated on a 15
×
15 mm square quartz substrate using RF
sputtering and electron beam (EB) lithography. The details of the fabrication process are described
in Appendix D. The SEM image of meta-atoms is shown in Figure 11c. Additionally, we fabricated
a blackbody reference sample by spraying a blackbody spray (TA410KS, Ichinen TASCO Co., Ltd.,
Osaka, Japan) to the 15
×
15 mm square quartz substrate, as shown in Figure 11a. This blackbody
reference has an average absorptivity of α~0.989 at λ=3–10 µm.
Photonics 2019, 6, x FOR PEER REVIEW 11 of 20
fabrication of a “plasmonic” thermal radiation light source. To design and optimize the size
parameters for a perfect absorber operating at ~4 µm, we performed numerical simulations using the
commercially available finite-difference time-domain method (FDTD) software (Lumerical Inc.,
Vancouver, BC, Canada) for the metasurface composed of HfN and SiO
2
as shown in Appendix F.
From Figure A6, the designed value of diameter d = 1.2 µm was determined for achieving the
absorption peak of 4 µm.
Figure 11 shows the MDM metasurface sample with d = 1.14 µm, P = 2.0 µm, T
g
= 130 nm, and T
d
= 200 nm. The metasurface was fabricated on a 15 × 15 mm square quartz substrate using RF
sputtering and electron beam (EB) lithography. The details of the fabrication process are described in
Appendix D. The SEM image of meta-atoms is shown in Figure 11c. Additionally, we fabricated a
blackbody reference sample by spraying a blackbody spray (TA410KS, Ichinen TASCO Co., Ltd.,
Osaka, Japan) to the 15 × 15 mm square quartz substrate, as shown in Figure 11a. This blackbody
reference has an average absorptivity of α ~0.989 at λ = 3–10 µm.
Figure 11. Refractory MDM metasurface and blackbody reference: (a) a photograph of the blackbody
reference sample, (b) the MDM metasurface sample composed of HfN and SiO
2
on a quartz substrate
with d = 1.14 µm, P = 2.0 µm, T
g
= 130 nm, and T
d
= 200 nm. The patterned area is 10 × 10 mm, and (c)
SEM image of plasmonic resonators.
5.3. Measurements
Before we measured the thermal radiation, we measured the spectral reflectivity R(λ) of the
metasurface at λ = 3–12 µm using a confocal infrared microscope (HYPERION2000, Bruker Inc.,
Billerica, MA, USA) and a Fourier transform infrared (FTIR) spectrometer (VERTEX 70v, Bruker Inc.,
Billerica, MA, USA) at room temperature. IR light partially shielded by slits was focused on a sample
through an ×15 (NA: 0.4) Schwarzschild objective lens. The reflected light from the sample was
corrected through the objective using a detector and converted through a Fourier transform to
calculate reflectivity. Spectral absorptivity, A(λ), can be calculated by A(λ) = 1 − R(λ) if the sample is
opaque.
The thermal radiation spectra were measured using an FTIR spectrometer (FT/IR 6000, JASCO
Co., Tokyo, Japan) at λ = 3–12 µm. The setup of the thermal radiation measurement is shown in
Appendix E. To avoid oxidation of the surface, the sample was set in a vacuum chamber connected
to the FTIR spectrometer and heated on a ceramic heater. A DC power supply was used to control
the temperature. The temperature was measured by a thermocouple placed on the surface of the
sample. The radiation spectra were measured for both the metasurface and the blackbody sample at
573 K. Hence, all measurements were done under the constant temperature of 573 K.
5.4. Results and Discussion
The measured absorptivity spectrum of the MDM metasurface at room temperature is shown in
Figure 12a. We observed a single resonant peak at 4.11 µm. To identify the physical origin of the peak,
we calculated the spectral absorptivity (see the dashed line in Figure 12a) and field distribution by
the FDTD method. The measured spectrum is in good agreement with the simulated spectrum.
Figure 12b shows a cross-sectional view of the spatial distribution of the electric field normal to the
Figure 11.
Refractory MDM metasurface and blackbody reference: (
a
) a photograph of the blackbody
reference sample, (
b
) the MDM metasurface sample composed of HfN and SiO
2
on a quartz substrate
with d=1.14
µ
m, P=2.0
µ
m, T
g
=130 nm, and T
d
=200 nm. The patterned area is 10
×
10 mm, and
(c) SEM image of plasmonic resonators.
5.3. Measurements
Before we measured the thermal radiation, we measured the spectral reflectivity R(
λ
) of the
metasurface at
λ
=3–12
µ
m using a confocal infrared microscope (HYPERION2000, Bruker Inc.,
Billerica, MA, USA) and a Fourier transform infrared (FTIR) spectrometer (VERTEX 70v, Bruker Inc.,
Billerica, MA, USA) at room temperature. IR light partially shielded by slits was focused on a sample
through an
×
15 (NA: 0.4) Schwarzschild objective lens. The reflected light from the sample was
corrected through the objective using a detector and converted through a Fourier transform to calculate
reflectivity. Spectral absorptivity, A(λ), can be calculated by A(λ)=1−R(λ) if the sample is opaque.
The thermal radiation spectra were measured using an FTIR spectrometer (FT/IR 6000, JASCO
Co., Tokyo, Japan) at
λ
=3–12
µ
m. The setup of the thermal radiation measurement is shown in
Appendix E. To avoid oxidation of the surface, the sample was set in a vacuum chamber connected to
the FTIR spectrometer and heated on a ceramic heater. A DC power supply was used to control the
temperature. The temperature was measured by a thermocouple placed on the surface of the sample.
The radiation spectra were measured for both the metasurface and the blackbody sample at 573 K.
Hence, all measurements were done under the constant temperature of 573 K.
5.4. Results and Discussion
The measured absorptivity spectrum of the MDM metasurface at room temperature is shown
in Figure 12a. We observed a single resonant peak at 4.11
µ
m. To identify the physical origin of the
peak, we calculated the spectral absorptivity (see the dashed line in Figure 12a) and field distribution
by the FDTD method. The measured spectrum is in good agreement with the simulated spectrum.
Figure 12b shows a cross-sectional view of the spatial distribution of the electric field normal to the

Photonics 2019,6, 105 12 of 20
incident electric field at 4.11
µ
m around a meta-atom (plasmonic cavity). Here, we can confirm that
the gap plasmon is excited to an FP resonant mode between two metal layers. Hence, the peak in
absorptivity around 4
µ
m is attributed to the plasmonic resonance inside a single plasmonic cavity.
Note that the resonant peak position is robust against incident angle for both p- and s-polarizations
as shown in Appendix G. The measured FWHM of the peak (~2
µ
m) is higher than the simulated
value (~1.5
µ
m) while the peak position and the peak value are red-shifted slightly and decreased,
respectively. The difference in the FWHM is attributed to the unexpected loss increase in the real
materials. The difference in the peak value is probably due to the off-axial arrangement of the incident
light through the Schwarzschild objective lens of the infrared microscope. From this measurement,
we were able to confirm that the sample was correctly fabricated and operating as designed for a
perfect absorber.
Photonics 2019, 6, x FOR PEER REVIEW 12 of 20
incident electric field at 4.11 µm around a meta-atom (plasmonic cavity). Here, we can confirm that
the gap plasmon is excited to an FP resonant mode between two metal layers. Hence, the peak in
absorptivity around 4 µm is attributed to the plasmonic resonance inside a single plasmonic cavity.
Note that the resonant peak position is robust against incident angle for both p- and s-polarizations
as shown in Appendix G. The measured FWHM of the peak (~2 µm) is higher than the simulated
value (~1.5 µm) while the peak position and the peak value are red-shifted slightly and decreased,
respectively. The difference in the FWHM is attributed to the unexpected loss increase in the real
materials. The difference in the peak value is probably due to the off-axial arrangement of the incident
light through the Schwarzschild objective lens of the infrared microscope. From this measurement,
we were able to confirm that the sample was correctly fabricated and operating as designed for a
perfect absorber.
Figure 12. Absorptivity spectra and electric field distribution of the MDM metasurface composed of
HfN and SiO
2
with
P = 2.0 µm, d = 1.14 µm, T
g
= 130 nm, and T
d
= 200 nm: (a) measured (solid line) and
simulated (dashed line) absorptivity spectra at room temperature, and (b) normalized electric field
distribution around the meta-atom for the resonance at 4.11 µm.
Next, we performed a thermal radiation experiment. Figure 13a shows the thermal radiation
spectra at 573 K for the MDM metasurface and reference blackbody sample. We observe that the
radiation intensity is suppressed at λ > 5 µm compared with the blackbody level. Such suppression
is caused by the lower absorptivity (emissivity) at λ > 5 µm, as seen in Figure 12a. Additionally, we
can derive the spectral emissivity, ε (λ), of the metasurface from Figure 13a. From Kirchhoff’s law,
this must be equal to α (λ) shown in Figure 12a if the temperature of a sample is identical. Figure 13b
shows the measured spectral emissivity at 573 K of the MDM metasurface. The resonant peak value
of ε = ~1 at 4.1 µm with FWHM of ~3 µm is obtained from Figure 13a. This is consistent with the
simulated results for absorptivity in Figure 12a (see also the solid line in Figure 13b). These results
suggest that perfect absorption/emission occurred as designed, and the cavity loss was increased due
to the temperature increase. The FWHM of the measured peak actually is broader than the calculated
result of ~1.5 µm. This is evidence of the loss increase caused by the thermal effect.
Figure 12.
Absorptivity spectra and electric field distribution of the MDM metasurface composed of
HfN and SiO
2
with P=2.0
µ
m, d=1.14
µ
m, T
g
=130 nm, and T
d
=200 nm: (
a
) measured (solid line)
and simulated (dashed line) absorptivity spectra at room temperature, and (
b
) normalized electric field
distribution around the meta-atom for the resonance at 4.11 µm.
Next, we performed a thermal radiation experiment. Figure 13a shows the thermal radiation
spectra at 573 K for the MDM metasurface and reference blackbody sample. We observe that the
radiation intensity is suppressed at
λ
>5
µ
m compared with the blackbody level. Such suppression
is caused by the lower absorptivity (emissivity) at
λ
>5
µ
m, as seen in Figure 12a. Additionally, we
can derive the spectral emissivity,
ε
(
λ
), of the metasurface from Figure 13a. From Kirchhoff’s law,
this must be equal to
α
(
λ
) shown in Figure 12a if the temperature of a sample is identical. Figure 13b
shows the measured spectral emissivity at 573 K of the MDM metasurface. The resonant peak value
of
ε
=~1 at 4.1
µ
m with FWHM of ~3
µ
m is obtained from Figure 13a. This is consistent with the
simulated results for absorptivity in Figure 12a (see also the solid line in Figure 13b). These results
suggest that perfect absorption/emission occurred as designed, and the cavity loss was increased due
to the temperature increase. The FWHM of the measured peak actually is broader than the calculated
result of ~1.5 µm. This is evidence of the loss increase caused by the thermal effect.
Finally, we note that the measured radiation spectrum in Figure 13a is not an intrinsic radiation
spectrum, but it includes the transmission function of the optical system in the spectrometer (see
Appendix E). Hence, it is necessary to separate it out so we can estimate the intrinsic radiation spectrum
from the sample. Since we obtained
ε
(
λ
), as shown in Figure 13b, we can determine the intrinsic
radiation spectrum of the sample by calculating the product of
ε
(
λ
) and Planck’s law (Equation (1)).
Figure 14 shows the presumed spectrum of the intrinsic radiation as well as the blackbody radiation at
573 K. It is clearly observed that thermal radiation from the metasurface is significantly suppressed at
longer wavelength region at
λ
>5
µ
m while the surface temperature of the sample is 573 K. Here, we

Photonics 2019,6, 105 13 of 20
point out a crucial fact that the area under the spectral curve of the metasurface is much smaller than
that of the blackbody. This indicates that the radiative power from the metasurface is significantly
suppressed compared with the blackbody resulting in the achievement of an efficient IR emitter, i.e.,
we are able to heat a sample quite efficiently by a small amount of power. Such behavior is typical for
constant-temperature-mode measurements, which is different from the constant power mode.
Photonics 2019, 6, x FOR PEER REVIEW 13 of 20
Figure 13. Experimental thermal radiation spectra for the MDM metasurface: (a) thermal radiation
spectra for the MDM metasurface (red line) and blackbody reference sample (solid line) at 573 K. (b)
Experimental spectral emissivity at 573 K (red line) derived from (a) and simulated absorptivity at
room temperature (solid line).
Finally, we note that the measured radiation spectrum in Figure 13a is not an intrinsic radiation
spectrum, but it includes the transmission function of the optical system in the spectrometer (see
Appendix E). Hence, it is necessary to separate it out so we can estimate the intrinsic radiation
spectrum from the sample. Since we obtained ε (λ), as shown in Figure 13b, we can determine the
intrinsic radiation spectrum of the sample by calculating the product of ε (λ) and Planck’s law
(Equation (1)). Figure 14 shows the presumed spectrum of the intrinsic radiation as well as the
blackbody radiation at 573 K. It is clearly observed that thermal radiation from the metasurface is
significantly suppressed at longer wavelength region at λ > 5 µm while the surface temperature of
the sample is 573 K. Here, we point out a crucial fact that the area under the spectral curve of the
metasurface is much smaller than that of the blackbody. This indicates that the radiative power from
the metasurface is significantly suppressed compared with the blackbody resulting in the
achievement of an efficient IR emitter, i.e., we are able to heat a sample quite efficiently by a small
amount of power. Such behavior is typical for constant-temperature-mode measurements, which is
different from the constant power mode.
Figure 14. Calculated thermal radiation spectra at 573 K: The radiation spectrum from the MDM
metasurface (red line) is calculated from the measured emissivity shown in Figure 13b. The theoretical
blackbody radiation spectrum (Equation (1)) at 573 K (solid line) is plotted for reference.
Figure 13.
Experimental thermal radiation spectra for the MDM metasurface: (
a
) thermal radiation
spectra for the MDM metasurface (red line) and blackbody reference sample (solid line) at 573 K.
(
b
) Experimental spectral emissivity at 573 K (red line) derived from (
a
) and simulated absorptivity at
room temperature (solid line).
Photonics 2019, 6, x FOR PEER REVIEW 13 of 20
Figure 13. Experimental thermal radiation spectra for the MDM metasurface: (a) thermal radiation
spectra for the MDM metasurface (red line) and blackbody reference sample (solid line) at 573 K. (b)
Experimental spectral emissivity at 573 K (red line) derived from (a) and simulated absorptivity at
room temperature (solid line).
Finally, we note that the measured radiation spectrum in Figure 13a is not an intrinsic radiation
spectrum, but it includes the transmission function of the optical system in the spectrometer (see
Appendix E). Hence, it is necessary to separate it out so we can estimate the intrinsic radiation
spectrum from the sample. Since we obtained ε (λ), as shown in Figure 13b, we can determine the
intrinsic radiation spectrum of the sample by calculating the product of ε (λ) and Planck’s law
(Equation (1)). Figure 14 shows the presumed spectrum of the intrinsic radiation as well as the
blackbody radiation at 573 K. It is clearly observed that thermal radiation from the metasurface is
significantly suppressed at longer wavelength region at λ > 5 µm while the surface temperature of
the sample is 573 K. Here, we point out a crucial fact that the area under the spectral curve of the
metasurface is much smaller than that of the blackbody. This indicates that the radiative power from
the metasurface is significantly suppressed compared with the blackbody resulting in the
achievement of an efficient IR emitter, i.e., we are able to heat a sample quite efficiently by a small
amount of power. Such behavior is typical for constant-temperature-mode measurements, which is
different from the constant power mode.
Figure 14. Calculated thermal radiation spectra at 573 K: The radiation spectrum from the MDM
metasurface (red line) is calculated from the measured emissivity shown in Figure 13b. The theoretical
blackbody radiation spectrum (Equation (1)) at 573 K (solid line) is plotted for reference.
Figure 14.
Calculated thermal radiation spectra at 573 K: The radiation spectrum from the MDM
metasurface (red line) is calculated from the measured emissivity shown in Figure 13b. The theoretical
blackbody radiation spectrum (Equation (1)) at 573 K (solid line) is plotted for reference.
6. Conclusions
We fabricated a prototype of microcavity lamp by a nanoimprint method that is suitable for mass
production and demonstrated to control visible-light spectrum using a refractory metasurface made of
Ta with an optical microcavity implemented into an incandescent light bulb. It was confirmed that
thermal radiation intensity from the microcavity filament was increased 1.8 times compared to the
flat filament under the constant power input. Then, we fabricated and demonstrated the thermal
radiation control in mid-IR range by using an MDM plasmonic metasurface composed of a refractory
plasmonic cavity made of HfN. A single narrow resonant peak was observed at designed wavelength

Photonics 2019,6, 105 14 of 20
as well as the suppression of thermal radiation in wide mid-IR range under the condition of constant
surface temperature.
We revaluated a thermal radiation light source as an efficient light source from the perspective
of energy conversion. For a future energy-saving society, it is vital to reconsider thermal radiation
sources as energy-saving technology.
Author Contributions:
J.T. conceived the idea of incandescent light bulbs based on refractory metasurface. H.T.
and A.K. performed the numerical simulations. H.T. and K.K. performed the experiments. J.T. analyzed the
experimental data and wrote the initial draft of the manuscript. J.T. supervised the project. All the authors
discussed the results and contributed to the writing of the manuscript.
Funding:
This research was funded in part by the Photonics Advanced Research Center (PARC) from the Ministry
of Education, Culture, Sports, Science and Technology, Japan (MEXT) and the JSPS Core-to-Core Program,
and A. Advanced Research Networks (Advanced Nanophotonics in the Emerging Fields of Nano-imaging,
Spectroscopy, Nonlinear Optics, Plasmonics/Metamaterials, and Devices).
Acknowledgments:
We would like to thank Yosuke Ueba and Yusuke Nagasaki for useful discussions. A part of
this work was supported by the “Nanotechnology Platform Project (Nanotechnology Open Facilities in Osaka
University)” of the Ministry of Education, Culture, Sports, Science, and Technology, Japan [No.: F-17-OS-0011,
S-17-OS-0011].
Conflicts of Interest:
The authors declare no conflicts of interest. The funders had no role in the design of the
study, in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to
publish the results.
Appendix A
The microcavity filaments were fabricated by a nanoimprint process suitable for mass production
as described below. Figure A1 shows the fabrication process of a microcavity array onto a Ta substrate.
At first, a 3-inch Si wafer (thickness 380
µ
m) was prepared, and the electron beam (EB) resist (ZEP520A-7)
spin-coated to a thickness of 300 nm using a spinner for 60 s at 4000 rpm. It was then baked for 5 min at
180
◦
C. After that, a 20
×
20 mm square microcavity array pattern was drawn onto the EB resist using a
50-kV electron beam (EB) lithography system (F5112+VD01; ADVANTEST Co., Tokyo, Japan). After
developing the resist, the Si wafer was dry-etched by an inductively coupled plasma (ICP) etching
machine (EIS-700, ELIONIX Inc., Tokyo, Japan) to transfer the pattern to the substrate, resulting in
a Si master mold with 300
×
300 nm square holes, 300 nm wall width, and approximately 200 nm
depth. Then, the master mold was duplicated onto a photocrosslinkable resin film under heat and
high-pressure conditions, resulting in an intermediate resin membrane (IRM). These IRMs are used for
mass production in the future so the master mold can be preserved.
Next, we prepared polished Ta substrates (20
×
20 mm squares with a thickness of 100
µ
m) for
making filaments. A 30-nm-thickness Cr layer was deposited on the Ta substrate, and a photoresist
(MUR-XR2-150, Maruzen Petrochemical Co., Ltd., Tokyo, Japan) was spin-coated on the substrate to a
thickness of 215 nm by using a spinner at 3000 rpm. Then, the sample was placed in a vacuum chamber,
and the IRM placed on a quartz cylinder was pressed onto the resist under UV light irradiation,
resulting in stamping the pattern onto the resist residing on the substrate with the Cr layer. Since the
IRM and the cylinder are transparent, it can be used as a template to transfer a pattern to a photoresist
under UV irradiation.
After removing the IRM, the Ta substrate was dry-etched through the resist and the Cr mask film
using an electron cyclotron resonance (ECR) ion shower machine (EIS-200ER, ELIONIX Inc., Tokyo,
Japan) and ICP etching machine (EIS-700, ELIONIX Inc., Tokyo, Japan). The etching depths are 3.2 nm
for the Cr layer and 18 nm for the Ta substrate. Finally, the resist pattern was transferred to the Ta
substrate, as shown in Figure 4b. The resulting cavity size was wider than the mold (350
×
350 nm
square holes, 250-nm wall width, and ~280-nm depth). Note that such a widening effect was caused by
the dry-etching process and was calibrated by designing the EB lithography process.
We fabricated three kinds of Ta substrate: (i) the microcavity on a single side, (ii) the microcavity
on both sides, and (iii) a plane without patterning for reference. In the case of (i), we performed the

Photonics 2019,6, 105 15 of 20
process once. For (ii), we repeated the process twice. Finally, the Ta substrates were cut into narrow
500-µm strips as shown in Figure 4a.
Photonics 2019, 6, x FOR PEER REVIEW 15 of 20
After removing the IRM, the Ta substrate was dry-etched through the resist and the Cr mask
film using an electron cyclotron resonance (ECR) ion shower machine (EIS-200ER, ELIONIX Inc.,
Tokyo, Japan) and ICP etching machine (EIS-700, ELIONIX Inc., Tokyo, Japan). The etching depths
are 3.2 nm for the Cr layer and 18 nm for the Ta substrate. Finally, the resist pattern was transferred
to the Ta substrate, as shown in Figure 4b. The resulting cavity size was wider than the mold (350 ×
350 nm square holes, 250-nm wall width, and ~280-nm depth). Note that such a widening effect was
caused by the dry-etching process and was calibrated by designing the EB lithography process.
We fabricated three kinds of Ta substrate: (i) the microcavity on a single side, (ii) the microcavity
on both sides, and (iii) a plane without patterning for reference. In the case of (i), we performed the
process once. For (ii), we repeated the process twice. Finally, the Ta substrates were cut into narrow
500-µm strips as shown in Figure 4a.
Figure A1. Nanoimprint process for fabricating microcavity filaments.
Appendix B
Figure A2 shows the relative permittivity spectra of conventional refractory metals (W, Ta, and
Mo) [41]. The real part of the permittivity for W and Mo are both positive in the visible range (λ < 0.8
µm). The real part of the permittivity of Ta switches from negative to positive below 0.6 µm. The
imaginary part of the permittivity of Ta is ~1/2 that of W and Mo.
Figure A1. Nanoimprint process for fabricating microcavity filaments.
Appendix B
Figure A2 shows the relative permittivity spectra of conventional refractory metals (W, Ta, and
Mo) [
41
]. The real part of the permittivity for W and Mo are both positive in the visible range
(
λ<0.8 µm
). The real part of the permittivity of Ta switches from negative to positive below 0.6
µ
m.
The imaginary part of the permittivity of Ta is ~1/2 that of W and Mo.
Photonics 2019, 6, x FOR PEER REVIEW 16 of 20
Figure A2. Spectral relative permittivities of conventional refractory metals (W, Ta, and Mo): (a) real
and (b) imaginary part of the relative permittivity [41].
Appendix C
Figure A3 shows the relative permittivity spectra of conventional plasmonic metals (Au, Ag, and
Al) [41], and refractory plasmonic materials (TiN and HfN) [43]. The real part of the permittivity of
HfN is negative within the entire visible range and is very close to that of Au at λ < 0.6 µm. The
imaginary part of the permittivity of HfN is ~1/2 that of TiN in the visible range.
Figure A3. Spectral relative permittivity of conventional plasmonic metals (Au, Ag, and Al) [41] and
refractory plasmonic materials (TiN and HfN) [43]: (a) real and (b) imaginary part of the relative
permittivities.
Appendix D
Figure A4 shows the fabrication process for the refractory MDM metasurface. First, a HfN layer
and a 130-nm-thick SiO
2
layer were deposited on a quartz substrate (15 × 15 mm) using an RF
sputtering system (SVC-700LRF, SANYU Electron, Tokyo, Japan). In fabricating the HfN layer, we
used an HfN target (Toshima Manufacturing Co., Ltd., Saitama, Japan) under an Ar gas flow rate of
25 sccm at 2 × 10
−4
Pa. Next, hexamethyldisilazane (HMDS) was spin-coated using a spinner for 90 s
at 5000 rpm. Then, a photoresist (TSMR-8900) was spin-coated to a thickness of 700 nm using a
spinner for 90 s at 5000 rpm. The metasurface patterns were exposed using a mask-less UV
lithography system (DL-1000, Nanosystem Solutions Inc., Tokyo, Japan) then developed. Next, a 200-
Figure A2.
Spectral relative permittivities of conventional refractory metals (W, Ta, and Mo): (
a
) real
and (b) imaginary part of the relative permittivity [41].

Photonics 2019,6, 105 16 of 20
Appendix C
Figure A3 shows the relative permittivity spectra of conventional plasmonic metals (Au, Ag, and
Al) [
41
], and refractory plasmonic materials (TiN and HfN) [
43
]. The real part of the permittivity of HfN
is negative within the entire visible range and is very close to that of Au at
λ
<0.6
µ
m. The imaginary
part of the permittivity of HfN is ~1/2 that of TiN in the visible range.
Photonics 2019, 6, x FOR PEER REVIEW 16 of 20
Figure A2. Spectral relative permittivities of conventional refractory metals (W, Ta, and Mo): (a) real
and (b) imaginary part of the relative permittivity [41].
Appendix C
Figure A3 shows the relative permittivity spectra of conventional plasmonic metals (Au, Ag, and
Al) [41], and refractory plasmonic materials (TiN and HfN) [43]. The real part of the permittivity of
HfN is negative within the entire visible range and is very close to that of Au at λ < 0.6 µm. The
imaginary part of the permittivity of HfN is ~1/2 that of TiN in the visible range.
Figure A3. Spectral relative permittivity of conventional plasmonic metals (Au, Ag, and Al) [41] and
refractory plasmonic materials (TiN and HfN) [43]: (a) real and (b) imaginary part of the relative
permittivities.
Appendix D
Figure A4 shows the fabrication process for the refractory MDM metasurface. First, a HfN layer
and a 130-nm-thick SiO
2
layer were deposited on a quartz substrate (15 × 15 mm) using an RF
sputtering system (SVC-700LRF, SANYU Electron, Tokyo, Japan). In fabricating the HfN layer, we
used an HfN target (Toshima Manufacturing Co., Ltd., Saitama, Japan) under an Ar gas flow rate of
25 sccm at 2 × 10
−4
Pa. Next, hexamethyldisilazane (HMDS) was spin-coated using a spinner for 90 s
at 5000 rpm. Then, a photoresist (TSMR-8900) was spin-coated to a thickness of 700 nm using a
spinner for 90 s at 5000 rpm. The metasurface patterns were exposed using a mask-less UV
lithography system (DL-1000, Nanosystem Solutions Inc., Tokyo, Japan) then developed. Next, a 200-
Figure A3.
Spectral relative permittivity of conventional plasmonic metals (Au, Ag, and Al) [
41
]
and refractory plasmonic materials (TiN and HfN) [
43
]: (
a
) real and (
b
) imaginary part of the
relative permittivities.
Appendix D
Figure A4 shows the fabrication process for the refractory MDM metasurface. First, a HfN
layer and a 130-nm-thick SiO
2
layer were deposited on a quartz substrate (15
×
15 mm) using an RF
sputtering system (SVC-700LRF, SANYU Electron, Tokyo, Japan). In fabricating the HfN layer, we
used an HfN target (Toshima Manufacturing Co., Ltd., Saitama, Japan) under an Ar gas flow rate
of 25 sccm at 2
×
10
−4
Pa. Next, hexamethyldisilazane (HMDS) was spin-coated using a spinner
for 90 s at 5000 rpm. Then, a photoresist (TSMR-8900) was spin-coated to a thickness of 700 nm
using a spinner for 90 s at 5000 rpm. The metasurface patterns were exposed using a mask-less UV
lithography system (DL-1000, Nanosystem Solutions Inc., Tokyo, Japan) then developed. Next, a
200-nm-thick HfN layer was deposited by RF sputtering. Finally, the HfN layer was lifted offusing
N-methyl-2-pyrrolidone (NMP).
Photonics 2019, 6, x FOR PEER REVIEW 17 of 20
nm-thick HfN layer was deposited by RF sputtering. Finally, the HfN layer was lifted off using N-
methyl-2-pyrrolidone (NMP).
Figure A4. Fabricating the refractory MDM metasurface.
Appendix E
Figure A5 shows the experimental setup for measuring the thermal radiation spectrum. The
optical system was placed in a vacuum chamber connected to an FTIR spectrometer (FT/IR 6000,
JASCO Co., Tokyo, Japan) through a tunnel tube. The vacuum chamber and the FTIR were pumped
to 2.0 × 10
2
Pa and 1.4 × 10
2
Pa, respectively. The spectral resolution was set to 4 cm
−1
, and a DLATGS
detector was used for the measurement. The device is shown in Figure 12 and was placed on a micro-
ceramic heater (MS-1000, Sakaguchi E.H VOC Corp., Tokyo, Japan). The temperature of the sample
was measured by a K-type sheath thermocouple (T350251H, Sakaguchi E.H VOC Corp., Tokyo,
Japan) placed on the surface of the sample. The measurements were performed at 573 K.
Figure A5. Experimental setup for measuring the thermal radiation spectrum. The sample is set on a
ceramic heater in a vacuum chamber that is connected to the FTIR spectrometer through a tunnel
tube.
Appendix F
Figure A6 shows simulated spectral absorptivity to the diameter d of an MDM metasurface
composed of HfN and SiO
2
(see Figure 9) with P = 2.0 µm, T
g
= 130 nm, and T
d
= 200 nm. The peak
Figure A4. Fabricating the refractory MDM metasurface.

Photonics 2019,6, 105 17 of 20
Appendix E
Figure A5 shows the experimental setup for measuring the thermal radiation spectrum. The
optical system was placed in a vacuum chamber connected to an FTIR spectrometer (FT/IR 6000,
JASCO Co., Tokyo, Japan) through a tunnel tube. The vacuum chamber and the FTIR were pumped to
2.0 ×102Pa
and 1.4
×
10
2
Pa, respectively. The spectral resolution was set to 4 cm
−1
, and a DLATGS
detector was used for the measurement. The device is shown in Figure 12 and was placed on a
micro-ceramic heater (MS-1000, Sakaguchi E.H VOC Corp., Tokyo, Japan). The temperature of the
sample was measured by a K-type sheath thermocouple (T350251H, Sakaguchi E.H VOC Corp., Tokyo,
Japan) placed on the surface of the sample. The measurements were performed at 573 K.
Photonics 2019, 6, x FOR PEER REVIEW 17 of 20
nm-thick HfN layer was deposited by RF sputtering. Finally, the HfN layer was lifted off using N-
methyl-2-pyrrolidone (NMP).
Figure A4. Fabricating the refractory MDM metasurface.
Appendix E
Figure A5 shows the experimental setup for measuring the thermal radiation spectrum. The
optical system was placed in a vacuum chamber connected to an FTIR spectrometer (FT/IR 6000,
JASCO Co., Tokyo, Japan) through a tunnel tube. The vacuum chamber and the FTIR were pumped
to 2.0 × 10
2
Pa and 1.4 × 10
2
Pa, respectively. The spectral resolution was set to 4 cm
−1
, and a DLATGS
detector was used for the measurement. The device is shown in Figure 12 and was placed on a micro-
ceramic heater (MS-1000, Sakaguchi E.H VOC Corp., Tokyo, Japan). The temperature of the sample
was measured by a K-type sheath thermocouple (T350251H, Sakaguchi E.H VOC Corp., Tokyo,
Japan) placed on the surface of the sample. The measurements were performed at 573 K.
Figure A5. Experimental setup for measuring the thermal radiation spectrum. The sample is set on a
ceramic heater in a vacuum chamber that is connected to the FTIR spectrometer through a tunnel
tube.
Appendix F
Figure A6 shows simulated spectral absorptivity to the diameter d of an MDM metasurface
composed of HfN and SiO
2
(see Figure 9) with P = 2.0 µm, T
g
= 130 nm, and T
d
= 200 nm. The peak
Figure A5.
Experimental setup for measuring the thermal radiation spectrum. The sample is set on a
ceramic heater in a vacuum chamber that is connected to the FTIR spectrometer through a tunnel tube.
Appendix F
Figure A6 shows simulated spectral absorptivity to the diameter dof an MDM metasurface
composed of HfN and SiO
2
(see Figure 9) with P=2.0
µ
m, T
g
=130 nm, and T
d
=200 nm. The peak
position of absorption caused by gap plasmon mode in the circular cavity can be changed from 3.0 to
7.0 µm by changing d.
Photonics 2019, 6, x FOR PEER REVIEW 18 of 20
position of absorption caused by gap plasmon mode in the circular cavity can be changed from 3.0 to
7.0 µm by changing d.
Figure A6. Simulated spectral absorptivity/emissivity map to the diameter of an MDM metasurface
composed of HfN and SiO
2
with P = 2.0 µm, T
g
= 130 nm, and T
d
= 200 nm.
Appendix G
Figure A7 shows simulated spectral absorptivity to the incident angle to an MDM metasurface
composed of HfN and SiO
2
(see Figure 9). The single absorption peak caused by gap plasmon mode
in the circular cavity is observed at ~4.5 µm for both polarizations, which is not strongly dependent
on incident angle. The strong angle-dependent steep absorption is caused by diffraction at 2.5–4.0
µm only for p-polarization as shown in (a). Note that the designed value of d = 1.2 µm is slightly
greater than that of the experimental value (d = 1.14 µm).
Figure A7. Simulated spectral absorptivity/emissivity maps to the incident angle to an MDM
metasurface composed of HfN and SiO
2
with P = 2.0 µm, d = 1.2 µm, T
g
= 130 nm, and T
d
= 200 nm: (a)
p-polarization and (b) s-polarization.
References
1. Takahara, J.; Ueba, Y.; Nagatsuma, T. Thermal radiation control by microcavity and ecological incandescent
lamps. Jpn. J. Opt. 2010, 39, 482–488.
Figure A6.
Simulated spectral absorptivity/emissivity map to the diameter of an MDM metasurface
composed of HfN and SiO2with P=2.0 µm, Tg=130 nm, and Td=200 nm.

Photonics 2019,6, 105 18 of 20
Appendix G
Figure A7 shows simulated spectral absorptivity to the incident angle to an MDM metasurface
composed of HfN and SiO
2
(see Figure 9). The single absorption peak caused by gap plasmon mode in
the circular cavity is observed at ~4.5
µ
m for both polarizations, which is not strongly dependent on
incident angle. The strong angle-dependent steep absorption is caused by diffraction at 2.5–4.0
µ
m
only for p-polarization as shown in (a). Note that the designed value of d=1.2
µ
m is slightly greater
than that of the experimental value (d=1.14 µm).
Photonics 2019, 6, x FOR PEER REVIEW 18 of 20
position of absorption caused by gap plasmon mode in the circular cavity can be changed from 3.0 to
7.0 µm by changing d.
Figure A6. Simulated spectral absorptivity/emissivity map to the diameter of an MDM metasurface
composed of HfN and SiO
2
with P = 2.0 µm, T
g
= 130 nm, and T
d
= 200 nm.
Appendix G
Figure A7 shows simulated spectral absorptivity to the incident angle to an MDM metasurface
composed of HfN and SiO
2
(see Figure 9). The single absorption peak caused by gap plasmon mode
in the circular cavity is observed at ~4.5 µm for both polarizations, which is not strongly dependent
on incident angle. The strong angle-dependent steep absorption is caused by diffraction at 2.5–4.0
µm only for p-polarization as shown in (a). Note that the designed value of d = 1.2 µm is slightly
greater than that of the experimental value (d = 1.14 µm).
Figure A7. Simulated spectral absorptivity/emissivity maps to the incident angle to an MDM
metasurface composed of HfN and SiO
2
with P = 2.0 µm, d = 1.2 µm, T
g
= 130 nm, and T
d
= 200 nm: (a)
p-polarization and (b) s-polarization.
References
1. Takahara, J.; Ueba, Y.; Nagatsuma, T. Thermal radiation control by microcavity and ecological incandescent
lamps. Jpn. J. Opt. 2010, 39, 482–488.
Figure A7.
Simulated spectral absorptivity/emissivity maps to the incident angle to an MDM
metasurface composed of HfN and SiO
2
with P=2.0
µ
m, d=1.2
µ
m, T
g
=130 nm, and
Td=200 nm
:
(a) p-polarization and (b) s-polarization.
References
1.
Takahara, J.; Ueba, Y.; Nagatsuma, T. Thermal radiation control by microcavity and ecological incandescent
lamps. Jpn. J. Opt. 2010,39, 482–488.
2. Takahara, J.; Ueba, Y. Thermal Infrared Emitters by Plasmonic Metasurface. Proc. SPIE 2013,8818, 88180X.
3.
Waymouth, J.F. Where will the next generation of lamps come from? J. Illum. Energy Inst. Jpn.
1990
,74, 800.
[CrossRef]
4. Waymouth, J.F. Optical Light Source Devices. U.S. Patent 5,079,473, 7 January 1992.
5.
Hesketh, P.J.; Zemalr, J.N.; Gebhart, B. Organ pipe radiant modes of periodic micromachined silicon surface.
Nature 1986,324, 549–551. [CrossRef] [PubMed]
6.
Maruyama, S.; Kashiwa, T.; Yugami, H.; Esashi, M. Thermal radiation from two-dimensionally confined
modes in microcavities. Appl. Phys. Lett. 2001,79, 1393–1395. [CrossRef]
7.
Kusunoki, F.; Kohama, T.; Hiroshima, T.; Fukumoto, S.; Takahara, J.; Kobayashi, T. Narrow-Band Thermal
radiation with Low Directivity by Resonant Modes inside Tungsten Microcavities. Jpn. J. Appl. Phys.
2004
,
43, 5253–5258. [CrossRef]
8.
Lin, S.Y.; Fleming, J.G.; Chow, E.; Bur, J.; Choi, K.K.; Goldberg, A. Enhancement and suppression of thermal
emission by a three-dimensional photonic crystal. Phys. Rev. 2000,62, R2243–R2246. [CrossRef]
9.
Zoysa, M.D.; Asano, T.; Mochizuki, K.; Oskooi, A.; Inoue, T.; Noda, S. Conversion of broadband to narrowband
thermal emission through energy recycling. Nat. Photonics 2012,6, 535–539. [CrossRef]
10.
Ikeda, K.; Miyazaki, H.T.; Kasaya, T.; Yamamoto, K.; Inoue, Y.; Fujimura, K.; Kanakugi, T.; Okada, M.;
Hatada, K.; Kitagawa, S. Controlled thermal emission of polarized infrared waves from arrayed plasmon
nanocavities. Appl. Phys. Lett. 2008,92, 021117. [CrossRef]
11.
Miyazaki, H.; Ikeda, K.; Kasaya, T.; Yamamoto, K.; Inoue, Y.; Fujimura, K.; Kanakugi, T.; Okada, M.;
Hatade, K.; Kitagawa, S. Thermal emission of two-color polarized infrared waves from integrated plasmon
cavities. Appl. Phys. Lett. 2008,92, 141114. [CrossRef]

Photonics 2019,6, 105 19 of 20
12.
Schuller, J.A.; Taubner, T.; Brongersma, M.L. Optical antenna thermal emitters. Nat. Photonics
2009
,3, 658–661.
[CrossRef]
13.
Greffet, J.J.; Carminati, R.; Joulain, K.; Mulet, J.P.; Mainguy, S.; Chen, Y. Coherent emission of light by thermal
sources. Nature 2002,416, 61–62. [CrossRef] [PubMed]
14.
Kusunoki, F.J.; Kobayashi, T. Qualitative change of resonant peaks in thermal emission from periodic array
of microcavities. Electron. Lett. 2003,39, 23–24. [CrossRef]
15.
Biener, G.; Dahan, N.; Niv, A.; Kleiner, V.; Hasmana, E. Highly coherent thermal emission obtained by
plasmonic bandgap structures. Appl. Phys. Lett. 2008,92, 081913. [CrossRef]
16.
Ueba, Y.; Takahara, J.; Nagatsuma, T. Thermal radiation control in the terahertz region using the spoof surface
plasmon mode. Opt. Lett. 2011,36, 909–911. [CrossRef]
17.
Yang, Z.Y.; Ishii, S.; Yokoyama, T.; Dao, T.D.; Sun, M.-G.; Nagao, T.; Chen, K.-P. Tamm plasmon selective
thermal emitters. Opt. Lett. 2016,41, 4453–4456. [CrossRef]
18.
Landy, N.I.; Sajuyigbe, S.; Mock, J.J.; Smith, D.R.; Padilla, W.J. Perfect Metamaterial Absorber. Phys. Rev. Lett.
2008,100, 207402. [CrossRef]
19.
Watts, C.M.; Liu, X.; Padilla, W.J. Metamaterial Electromagnetic Wave Absorbers. Adv. Mater.
2012
,24, OP98.
[CrossRef]
20. Lee, B.J.; Wang, L.P.; Zhang, Z.M. Coherent thermal emission by excitation of magnetic polaritons between
periodic strips and a metallic film. Opt. Express 2008,16, 11328–11336. [CrossRef]
21.
Puscau, I.; Schaich, W.L. Narrow-band, tunable infrared emission from arrays of microstrip patches. Appl.
Phys. Lett. 2008,92, 233102. [CrossRef]
22.
Liu, X.; Starr, T.; Starr, A.F.; Padilla, W.J. Infrared Spatial and Frequency Selective Metamaterial with
Near-Unity Absorbance. Phys. Rev. Lett. 2010,104, 207403. [CrossRef] [PubMed]
23.
Liu, X.; Tyler, T.; Starr, T.; Starr, A.F.; Jokerst, N.M.; Padilla, W.J. Taming the Blackbody with Infrared
Metamaterials as Selective Thermal Emitters. Phys. Rev. Lett. 2011,107, 045901. [CrossRef] [PubMed]
24.
Nielsen, M.G.; Pors, A.; Albrektsen, O.; Bozhevolnyi, S.I. Efficient absorption of visible radiation by gap
plasmon resonators. Opt. Express 2012,20, 13311–13319. [CrossRef] [PubMed]
25.
Ueba, Y.; Takahara, J. Spectral control of thermal radiation by metasurface with a sprit-ring resonator. Appl.
Phys. Express 2012,5, 122001–122003. [CrossRef]
26.
Ito, K.; Toshiyoshi, H.; Iizuka, H. Densely-tiled metal-insulator-metal metamaterial resonators with
quasi-monochromatic thermal emission. Opt. Express 2016,24, 12803–12811. [CrossRef]
27.
Yokoyama, T.; Dao, T.D.; Chen, K.; Ishii, S.; Sugavaneshwar, R.P.; Kitajima, M.; Nagao, T. Spectrally Selective
Mid-Infrared Thermal Emission from Molybdenum Plasmonic Metamaterial Operated up to 1000
◦
C. Adv.
Opt. Mat. 2016,4, 1987. [CrossRef]
28.
Miyazaki, H.T.; Kasaya, T.; Iwanaga, M.; Choi, B.; Sugimoto, Y.; Sakoda, K. Dual-band infrared metasurface
thermal emitter for CO2sensing. Appl. Phys. Lett. 2014,105, 121107. [CrossRef]
29.
Matsuno, Y.; Sakurai, A. Perfect infrared absorber and emitter based on a large-area metasurface. Opt. Mat.
Express 2017,7, 618–626. [CrossRef]
30.
Ogawa, S.; Kimata, M. Metal-Insulator-Metal-Based Plasmonic Metamaterial Absorbers at Visible and
Infrared Wavelengths: A Review. Materials 2018,11, 458. [CrossRef]
31.
Baranov, D.G.; Xiao, Y.; Nechepurenko, I.A.; Krasnok, A.; Al
ù
, A.; Kats, M.A. Nanophotonic engineering of
far-field thermal emitters. Nat. Mater. 2019,18, 920–930. [CrossRef]
32.
Sai, H.; Kanamori, Y.; Yugami, H. High-temperature resistive surface grating for spectral control of thermal
radiation. Appl. Phys. Lett. 2003,82, 1685–1687. [CrossRef]
33.
Takahara, J. Renaissance of incandescent light bulbs through nanotechnology. Oyo Buturi
2015
,84, 564–567.
34.
Takahara, J.; Kimino, K. Visible Spectral Control of Incandescent Light Bulbs by Microcavity Filament. In
Proceedings of the Abstracts of 15th International Symposium on the Science and Technology of Lighting
(LS15), Kyoto, Japan, 22–27 May 2016. 23P-LL1.
35.
Ilic, O.; Bermel, P.; Chen, G.; Joannopoulos, J.D.; Celanovic, I.; Soljaˇci´c, M. Tailoring high-temperature
radiation and the resurrection of the incandescent source. Nat. Nanotech.
2016
,11, 320–324. [CrossRef]
[PubMed]
36.
Shimogaki, K.; Iwai, I.; Yoshiike, H. The Luminous Efficacy of an Incandescent Lamp Which has an Infrared
Reflective Film. J. Illum. Eng. Inst. Jpn. 1984,68, 72–76. [CrossRef]

Photonics 2019,6, 105 20 of 20
37.
The Illuminating Engineering Institute of Japan (IEI-J). Lighting Handbook (Compact Version); Ohmsha: Tokyo,
Japan, 2006; Volume 3 2-1-2.
38. IEI-J. Lighting Databook (New Edition); Ohmsha: Tokyo, Japan, 1968; Volume 3-1-2.
39. IEI-J. Lighting Handbook (Compact Version); Ohmsha: Tokyo, Japan, 2006; Volume 3 2-2-2.
40. IEI-J. 34th Textbook of Lighting Basic Course (New Edition); IEI-J: Tokyo, Japan, 2013; pp. 2–23.
41. Palik, E.D. Handbook of Optical Constants of Solids; Academic Press: San Diego, CA, USA, 1985.
42. Boltasseva, A.; Atwater, H.A. Low-Loss Plasmonic Materials. Science 2011,331, 290–291. [CrossRef]
43.
Naik, G.V.; Shalaev, V.M.; Boltasseva, A. Alternative Plasmonic Materials: Beyond Gold and Silver. Adv.
Mater. 2013,25, 3264–3294. [CrossRef]
44. Guler, U.; Boltasseva, A.; Shalaev, V.M. Refractory Plasmonics. Science 2014,344, 263–264. [CrossRef]
45.
Kumar, M.; Umezawa, N.; Ishii, S.; Nagao, T. Examining the Performance of Refractory Conductive Ceramics
as Plasmonic Materials: A Theoretical Approach. ACS Photonics 2015,3, 43–50. [CrossRef]
©
2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).