COMPREHENSIVE REVIEW

Japanese Journal of Applied Physics 52 (2013) 010001 http://dx.doi.org/10.7567/JJAP.52.010001

Plasmonics: Future Outlooky Satoshi Kawata Department of Applied Physics, Graduate School of Engineering, Osaka University, Suita, Osaka 565-0871, Japan E-mail: [email protected] Received July 10, 2012; accepted August 24, 2012; published online December 12, 2012 Plasma resonance in metals exhibits some unique optical phenomena that occur on the surface of metal with nanostructures. The use of surface plasmons has been proposed in various fields, such as nanometer-resolution near-field optical microscopy, nanoscale optical circuits, singlemolecule detection, molecular sensors, cancer treatment, solar cells, lasers, and holography. The study of plasma resonance is called ‘‘plasmonics’’ and is expected as a new field of nanophotonics. In this report, I review the principles and limits of plasmonics and give a future outlook. # 2013 The Japan Society of Applied Physics

1. Introduction: Common Knowledge and Misconceptions of Plasmonics

I have already written about plasmonics in some review articles,1–7) while in this report I will give an explanation in a different manner from before. Although various characteristics and functions of plasmonics can be listed, I here consider that its essence is ‘‘slow light’’. Photons propagating on the surface of a metal interact with plasmons in the metal as a collective oscillation of free electrons, and hence the velocity (group velocity) of the photons on the metal surface is lower than that of ordinary propagating light in free space. As an extreme case, at the resonant frequency of surface plasmon, speed becomes zero and then photons remain at a fixed point on the metal surface and do not propagate. Another characteristic of plasmonics is ‘‘wavelength compression’’. The phase velocity also decreases on the metal surface, and the spot size of photons becomes much smaller than the wavelength. This essential characteristic of plasmonics is used for various applications,3,4) such as near-field optical microscopy,8–10) superlenses,11,12) and surface plasmon sensors.13–18) In the spectral range far enough from the plasmon resonance frequency, for example microwave and terahertz bands, metals are not any more plasmonic but almost conductive so that they do not exhibit slow-light characteristic nor wavelength compression. Field enhancement due to plasmonic electromagnetic effect does not appear in such spectral bands.7,19) ‘‘Plasmonics’’ refers to the optical nanoscience (nanophotonics) in the range limited to near-UV, visible, and near-IR ranges, those are near at the plasmon resonance frequency. The field enhancement at nanoparticles and aggregates is not an inherent property of plasmonics, but rather that of cavity effect as seen in Fabry–Pe´rot interferometers and laser resonators. A particle with an optimized length (or diameter) is more appropriate for the field enhancement than an extremely small particle. In addition, in the plasmonics range, particularly in UV and visible, the resonance effect due to the electron transfer and electron transition works for gaining Raman scattering efficiency by the chemical adsorption, when a metal particle is brought close enough to a Raman-active y

This is a translated version of the original paper which appeared in Oyo Buturi 80 (2011) 757 [in Japanese].

material. This is known as surface-enhanced Raman scattering (SERS).20,21) Surface plasmons refer to the quantum state of plasma waves in the surface-wave mode. When a metal (a solid state of plasma) has an interface with a dielectric, the surface plasmons near the interface have the component of electromagnetic waves and are quantized as surface plasmon polaritons (SPPs). The science of the interaction between SPPs and photons (electromagnetic fields) is defined as plasmonics. The electromagnetic component of SPPs exists in the vicinity of the metal surface as surface waves (nearfield light). The dispersion relation of SPPs is obtained by solving Maxwell’s equations of electromagnetism using dielectric functions of the metal and dielectric to satisfy the boundary conditions. It is hence treated as classical theory of electromagnetism. Although the dielectric function of a metal may be derived using Drude’s model for simplification, more precisely the values in the literature22) or measured spectroscopic data should be used because the absorption edge due to the interband transition for the metal, such as gold and silver, is in the UV and visible ranges.23) Figure 1 shows the dispersion relation for surface plasmons in silver as a typical plasmonic material. Dielectric medium facing on the silver surface is air or silica. This dispersion relation tells us all about the characteristics of plasmonic materials. The dispersion relation of light is linear between wavenumber k and angular frequency !, whereas that of surface plasmons is given by ! ksp ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ; c "m "d =ð"m þ "d Þ where "m and "d are dielectric constants of metal and dielectric, respectively. Wavenumber ksp of surface plasmon is always greater than the wavenumber k of the propagating light; in other words, the wavelength of surface plasmon is shorter than that of propagating light, and the group velocity d!=dksp of the surface plasmons is always lower than that of propagating light. Namely, the surface plasmon has a lower velocity and a shorter wavelength than ordinary propagating light, resulting as a nonradiative evanescent field (surface wave) on the metal surface. As the wavelength of SPP is shorter than that of ordinary light, SPP is used in super-resolution optical microscope. A near-field optical microscope with a metallic probe to excite SPPs provides a resolution exceeding beyond the limit of ordinary optical microscope using propagating photons.8)

010001-1

# 2013 The Japan Society of Applied Physics

COMPREHENSIVE REVIEW

Jpn. J. Appl. Phys. 52 (2013) 010001

9 8 7

Surface plasmons (silver-air) Propagating light (in silica)

Propagating light (in air) 300 ω P (325nm) ω SP (337nm) ω LSP (354nm) ω SP (357nm) ω LSP (396nm)

6 5

Surface plasmons (silver-silica)

4

of samples was necessarily used in the experiment of SPP imaging.24)

200

500 600 700 800 1000

Propagation

3

-- ++ -- ++ -- ++ --- ++ -- ++ -- ++ --

2

SPP

1

Wavelength [nm]

Angular frequency, ω [1015 Hz]

10

2000

Metal

0 0

10

20

30

40

50

60

70 6

-1

Real part of dielectric constant, Re{k sp} [10 m ]

(a) 10

Surface plasmons (silver-air)

8 7

Surface plasmons (silver-silica)

6

300

5

400

4

2

500 600 700 800 1000

1

2000

3

Wavelength [nm]

Angular frequency, ω [1015 Hz]

200 9

0 0

10

20

30

40

50

60

70 6

-1

Imaginary part of dielectric constant, Im{k sp} [10 m ]

(b) Fig. 1. (Color online) Dispersion relation of surface plasmons (silver) and light. (a) Real and (b) imaginary parts of dielecric function of surface plasmons. Red and blue lines represent the cases when the silver is in air and silica, respectively. The resonance frequencies of plasmons (!P ) in bulk silver, surface plasmons (!SP ), and localized surface plasmons (!LSP ) are shown (courtesy of Dr. Ichimura).

The dispersion curve of surface plasmon asymptotically approaches the value given based on the Drude’s free electron model as !p ! ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffi ; 1 þ "d where !p is the plasma resonance frequency. In the figure, curves for both in air and in silica double back in higher energy region than the light dispersion (straight lines for each) indicate that the incident angle of light at the interface is equal to the Brewster angle above the plasma resonance frequency. In such spectral range, silver becomes dielectric rather than metal and has a positive dielectric constant. The wavelength reduction is typically a factor of more than seven for silver (502 nm to 70 nm). When a sample is placed on the surface of silver, the surface acts as a microscope objective lens with a resolution equivalent to one-seventh of the wavelength. This is theoretically reasonable, and was experimentally demonstrated for the first time in 2005.24) A microscope using SPPs would be completely different from a conventional microscope that images a sample through objective lens from top or bottom, but images a sample with surface waves propagating along the twodimensional surface of metal. Scattering due to the roughness

2. ‘‘Thin-Film’’, ‘‘Wire’’, and ‘‘Dot’’ Mode Plasmons

For the practical application of SPPs, the absorption of light by metal would be the inevitable issue because the dielectric constant of metal has a non-negligible value in the imaginary part in visible region. This is an important issue to be considered in the discussion on the enhancement of electric field and the propagation length of plasmons. Recently, plasmonics has been practically applied to devices, due to the result of research achievements on the light absorption loss in structure. The most commonly used structure is a thin film [Fig. 2(a)]. When a metal film is sufficiently thin, two surface propagation modes develop, one of which forms a charge distribution antisymmetric with respect to the film. In this mode, the loss is small and the propagation length is long (so that it is called long-range mode). This was first reported by Fukui et al.25,26) Using this mode, the author has reported an ultrahigh-sensitivity surface plasmon resonance sensor in 1990 [Fig. 2(b)].27) Structures in which a dielectric thin film sandwiched between metals, shown in Fig. 2(c), have recently been frequently reported as plasmonic devices. There are two modes in such structures. One mode forms a large electric field within the dielectric film (but not in the metal) resulting in a small loss attributable to metal, and is thereby more suitable for the application to devices involving high electric field. However, in this configuration there is a possibility in which the waveguide mode of dielectric is excited rather than the propagation mode of surface plasmons (refer to Sect. 5.11 in Ref. 19). As a lower-dimensional structure than that of thin film, a thin-wire structure has been considered, as shown in Fig. 2(d). Takahara et al. has proposed the waveguide propagation mode of a metallic thin wire without any wavelength cutoff.28) Researchers of optical nano-circuits have become interested in this metallic wire structure, because this makes possible to send photons through nanometer metal (negative dielectrics). Figure 2(e) shows the dispersion relation for a silver nanowire obtained by Ditlbacher et al. in a waveguide experiment.29) The relation indicates that the effects of wavelength compression and group velocity reduction are more significant in wire plasmon mode (more accurately, rod plasmon mode) than those in surface plasmon mode. As increasing the dimensions of thin-wire structure to a small dot, plasmonics is reduced precisely to the Mie scattering problem [Fig. 2(f )]. When the size of metal structure is sufficiently smaller than the wavelength of light, the dipolar mode becomes dominant. The resonance frequency in this mode (localized mode) of plasmon is lower than that in the surface mode and is given by the following equation with Drude’s free-electron model, !p ! ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi : 1 þ 2"d Namely, the localized mode exists at a lower frequency (longer wavelength) than the surface mode. With increasing size of the metal structure, the mode further shifts to a lower frequency (longer wavelength) and higher-order modes will appear. This is called the size effect and is described later.

010001-2

# 2013 The Japan Society of Applied Physics

COMPREHENSIVE REVIEW

Jpn. J. Appl. Phys. 52 (2013) 010001

Propagation z --

++

--

++

++

--

++

--

Ez

(b) Incident light

θ

Prism Metal Metalic thin film Sample

Without dielectric Layer

SPP

θ

Short-range mode Propagation z --

++

--

++

--

++

--

++

Ez Metalic thin film

Prism Dielectric Metal Sample

(c)

Reflected light

Reflectance

(a) Long-range mode

Metal

Dielectric

--

++

--

++

--

++

--

++

z Ez

Metal

With dielectric layer

Incident angle (deg)

SPP

(e)

(d)

Propagation

(f) Nanosphere

Propagating light (glass) ire Propagating w light (air) no

Flat silver surface

Metal

r lve

++

na

--

Si

Fig. 2. (Color online) Surface-plasmonic structures with small loss. (a) Long- and short-range modes of metallic thin film, (b) structures for surface plasmon sensor with/without using long-range mode and the reflectance curves as functions of incident angle,27) (c) metal-dielectric-metal structure, (d) plasmon waveguide based on a metallic thin wire,28) (e) experimental result of dispersion relation for a nanowire,29) and (f ) excitation of localized plasmons at a metallic nanosphere.

(a)

(d) Light irradiation Sample Photoresist

PR Ag PMMA Cr

Substrate

Quartz

light

(b)

AFM observation

(c)

Fig. 3. (Color online) Superlens. (a) Structure of a superlens proposed by Fang et al.,12) (b) focused ion beam (FIB) image of the word ‘‘NANO’’ patterned

on Cr, (c) AFM image of characters recorded in a photoresist, (d) direct recording of near field distribution on a photoresist proposed by Kawata et al. in 1999.30)

3. Superlenses and Apertureless Near-Field Optical Microscopes (Tip-Enhanced Optical Microscopes)

As shown in the dispersion relation in Fig. 1, the velocity of surface plasmon becomes zero at the resonance frequency. Namely, light stops at the surface of a metal and its wavelength becomes infinitesimal, resulting in the formation of an optical nanospot without extensity (in fact, however, the spot broadens because due to the imaginary part of dielectric constant). Scientists conceived of using this phenomenon for

imaging an optical field distribution with the resolution much smaller than the wavelength of light. The principle of such imaging was proposed by Pendry in 200011) and experimentally demonstrated by Fang et al. in 200512) [Figs. 3(a)–3(c)]. Although the concept of superlens is fascinating, its implementation is impractical because the image is formed as a nonradiative optical field, which cannot be seen by human eyes or a camera. To solve this problem, Fang et al. recorded the image in the photoresist on the opposite side of metal thin film to the object [Fig. 3(a)] and read out the

010001-3

# 2013 The Japan Society of Applied Physics

COMPREHENSIVE REVIEW

Jpn. J. Appl. Phys. 52 (2013) 010001

(a)

(b)

(c)

Metallic nanoprobe Objective lens Scan Sample 100nm

(e) 70

200 nm

53 35 18

15 nm

0

19 99 20 00 20 01 20 02 20 03 20 04 20 05 20 06 20 07 20 08

(d)

Number of papers on TERS microscopy

Tip-enhanced near-field

09

20

Publication year Fig. 4. (Color online) Scattering near-field scanning optical microscope. Also called as apertureless near-field scanning optical microscope and tipenhanced near-field optical microscope. (a) Schematic of the principle, (b) SEM image of a silver nanoprobe, (c) schematic of principle of near-field optical microscopy with a laser-trapped metallic nanoparticle,29) (d) statistic of number of papers on TERS microscopy, and (e) image of a DNA network obtained using tip-enhanced CARS microscopy with spatial resolution of 15 nm.39)

recorded pattern using an atomic force microscope (AFM) [Fig. 3(c)].12) A better method is rather to record the optical field directly in a photoresist near a sample without a metal film. The direct recording of a near field using a photoresist was proposed and was experimentally demonstrated by Kawata et al. before Fang’s work [Fig. 3(d)].30) The essential problem of superlens lies in the fixed wavelength. Light does not stop but propagates for the wavelengths other than the resonance wavelength of surface plasmons (with pffiffiffi energy !p = 2), which makes images blurry. Compared to the fixed-wavelength imaging with superlens, the use of a deep- or extreme-UV light source or e-beam for microscopy may be more advantageous for high-resolution imaging. Nonetheless, the superlens is still a milestone in plasmonics as they can be used in microscopes with a resolution that is not affected by the diffraction limit. In 1992, authors developed an ultrahigh-resolution nearfield microscope that enables observation with objective lenses using a small metallic piece (probe) instead of a thin film [Figs. 4(a) and 4(b)].31) This small metallic piece, sufficiently shorter than the wavelength, is considered as a part of metallic thin film. With this structure, surface plasmons couple with propagating light, enabling the detection of signals in far field. The microscope can also be regarded as a superlens using the localized mode in Fig. 2(f ). We published a paper on the first-ever apertureless scanning near-field optical microscope in 1994,8) but did not receive much interest owing to the fact that near-field optical microscopes with aperture probes were attracting considerable attention of people at that time. A tungsten probe that was

electrolytically polished and coated with silver was initially used as the metallic nanoprobe, and the distance of the probe to the sample and scanning of the probe on the sample surface was controlled using scanning tunneling microscope (STM).32) We also developed a fluorescent microscope with a laser-trapped gold nanoparticle of 60 nm in diameter, realizing in-liquid measurement [Fig. 4(c)]. A positioning accuracy of 2 nm was achieved using this microscope with feedback control.33,34) In 1999, we developed a Raman scattering microscope using the above principle.35,36) It has been reported that Raman scattering is markedly enhanced when a molecule is placed on a metallic thin film.20,21) Scientists consider that this is largely due to the plasmonic effect of metal nanoparticles on enhancing electric field. Scanning the probe of our microscope is precisely equivalent to realizing SERS at every point of scanning. Although we suggested the theory underlying this phenomenon in 1994, it was first experimentally verified using the microscope in 1999. The microscope is called a tip-enhanced Raman scattering (TERS) microscope and has been reported by many researchers [Fig. 4(d)]. TERS microscopes have been applied to the analysis of a wide range of nanomaterials. Our group has observed and analyzed various nanomaterials, such as carbon nanotubes,37–40) molecules encapsulated into carbon nanotubes,41) fullerenes,42) graphene,38) DNA bases,37,44,45) strained-Si devices,46–48) and compound semiconductors.49) Figure 4(e) shows a Raman image of a DNA network44) that was obtained by detecting and imaging the coherent

010001-4

# 2013 The Japan Society of Applied Physics

COMPREHENSIVE REVIEW

Jpn. J. Appl. Phys. 52 (2013) 010001

(a) Half-wavelength antenna Microwave range

Plasmonic range λ

λ +

–

– +

L = λ SP /2 ~λ /2

ge

+

e

+

– –

Mi

+ – +

L = λ SP ~λ Zero current

v wa

u bliq

(o

ran

λ /2

cro

Plasmonic range λ

λ

nc

ei

L = λ SP /2 << λ /2

(b) One-wavelength antenna Microwave range

)

ce

n ide

+

+

– –

L = λ SP ~λ

L = λ SP << λ

Im

ag

(c)

ep

e

Intensity of transmitted light

no

Image plane

lan

p ga

na

50

nm

r

Ob

jec

fe ns

tp

ra et

ag

lac

e

Im

Single layer 3 layers

Magnifie d

7 layers 11 layers Broadband

500 600 700 Frequency [THz]

Object plane

Fig. 5. (Color online) Optical antennas for radio-wave (microwave) and light (visible and NIR). (a) Half-wavelength antenna (dipole antenna) and (b) onewavelength antenna. (c) Metallic nanorod arrays for nano-resolution imaging in color, transmission frequency bands, and geometrical magnification.56)

anti-stokes Raman scattering (CARS) of adenine molecules using two lasers and a silver nanoprobe. The spatial resolution was 15 nm. Although some TERS microscopes have been commercialized, they are still in the stage of research and development; few have contributed to nanomaterials market. The reason for this is their low reproducibility. An introduction of incompletely developed products to the market has had an adverse effect. In the heyday of near-field optical microscopes with aperture probes, a resolution practically impossible to obtain was sometimes ostentatiously reported and caused misunderstanding and confusion among users, resulting in their loss of credibility. I now hope that TERS microscopes with high reliability will become available by increasing the yield of the chip manufacturing process to a satisfactory level, which I believe will be realized in the very near future. 4. Differences between Plasmonic and Radio-Wave Antennas

Radio waves are electromagnetic waves that include light. The difference between radio waves and light only lies in their coherence and wavelength. Nano-wires or nanoantenna even for visible light can be fabricated using current nanotechnologies. There have been a number of reports on ‘‘optical antennas’’. So far as I remember, Pohl (an inventor

of near-field optical microscopes) and his group used the phrase ‘‘optical antennas’’ around 1995.50) Review reports on optical antenna have been published even recently.51,52) Antennas of various kinds used in the microwave band have been downsized and reported as plasmonic optical antennas. However, few scientists appear to recognize that plasmonic optical antennas exhibit functions different from those of microwave and radio wave antennas. I would like to discuss about the difference between two antennas as below. The most fundamental configuration of antenna is a dipole antenna shown in Fig. 5(a). The length of the dipole antenna is a half wavelength of transmitted and/or received radio wave. The alternative electric field is greatest at the two ends of the antenna (plus/minus signs are opposite), while the current becomes maximum at the center. The directionality of transmission/reception is the maximum in the direction perpendicular to the antenna wire. This antenna principle is common to all spectral range except that in UV/visible/near IR ranges the length of dipole antenna is much shorter than a half wavelength, typically 1/5 to 1/10. It is because that in such ranges the metal is no more conductive but rather plasmonic, so that the charge density wave (i.e., surface plasmons) is slower, as described in Sects. 1 and 2. An interesting feature of wavelength compression of plasmons in visible range is

010001-5

# 2013 The Japan Society of Applied Physics

COMPREHENSIVE REVIEW

Jpn. J. Appl. Phys. 52 (2013) 010001

found in the case of the one-wavelength antenna [Fig. 5(b)]. The electrical field at two ends of one-wavelength antenna should be equal to each other including signs, which means that the radio waves a do not induce current in this antenna or radiate from this antenna. The sign of the field at two ends of antenna should be opposite (positive and negative) as a transmitter or a receiver [see the left diagram in Fig. 5(a)]. As shown in the right diagram in Fig. 5(b), however, the phase difference generated between two ends of antenna due to the oblique incidence causes the current to flow, or radio wave is transmitted in the direction at a certain angle with the phase difference between the two ends of the antenna. The angular distribution of transmission/reception is different for wavelength. An optical antenna in visible range has a completely different from that of radiowave. As shown in the diagram in the middle of Fig. 5(b), the length of the optical antenna is equivalent to not one wavelength of light wave but that of plasmons, which is considerably shorter than the light wavelength. This one plasmon-wavelength antenna can neither radiate nor receive light at any angle. The visiblerange optical antenna can be fully invisible in any direction at certain wavelengths (integral multiples of the wavelength of the wire mode plasmons) and does not interact with the propagating light. Although such an antenna described above is invisible in any direction, it is still possible to induce current in antenna. In other words, a mode certainly exists. Such a mode is called subradiant mode,53) in which plasmons exist as a standing wave, although they are not radiated. At two ends of antenna in-phase electric fields are formed as near-field nanospots. When a molecule to be observed is brought close to the end of this antenna, a strong interaction between the molecule and plasmon polaritons occurs. If fluorescence or Raman scattering is induced with a wavelength differently from that of incident light, the light is radiated. Assuming that a probe of near-field optical microscopy is an antenna, it can be understood from the above explanation, the length of antenna determines the wavelength of light to be detected. This is also applicable to the photonic input/output to/from plasmonic nanocircuits. Length is the most important factor for antenna. Determining the length of antenna is equivalent to determining the cavity mode of laser. An optical antenna enters a forbidden mode when its length is an even multiple of a half wavelength of wire mode plasmons. For an antenna or metallic probe with the length of an even multiple of the half wavelength of wire mode plasmon, a cavity mode is excited, and therefore, plasmons resonate as standing waves within the antenna, enhancing the field at the tip. This is the mechanism of field enhancement in SERS. The field enhancement can be explained as the cavity effect of plasmons at a frequency equal to an even multiple of the half wavelength of plasmons, rather than wavelength compression effect of plasmons. This mechanism is seen as the cavity effect of Fabry–Pe´rot resonators and lasers. Since the metal is not a perfect conductor in visible range, SPPs and light absorption by the metal is not inevitable in addition to the field enhancement. At a given length of

antenna, the basic mode (dipole mode) shows the least absorption loss and the highest degree of field enhancement. Although plasmonic field enhancement has been considered to be applicable to solar cells, the high Q value for high enhancement narrows the spectral range loosing the usable spectral coverage. This trade-off relationship should be well taken cared of in practical applications. For gaining high field enhancement, it may be more effective to use transparent materials with high refractive index, which do not absorb light but effective in a narrow spectral band. The dielectrics and semiconductors (Si) can be downsized with cavity effect of enhancement if they have high refractive indices.54) Long antennas act as wire mode plasmon waveguides with multiple modes to transmit light with from one end to the other end of the antennas.29) When a multimode antenna is split and the segments are arranged in series with a small gap between them, broadband characteristics for a specific range (visible range) can be obtained. Moreover, a magnifier to see a nano-object can be built by bundling the antenna arrays in a tapered fashion, forming a cone with a dense inlet and coarse outlet. The spatial resolution and magnification of such a magnifier are determined as a function of the pitch of nanorods and the angle and the length of cone. We reported this magnifier under the name of a metallic nanolens in 2005 [Fig. 5(c)].55,56) 5. UV Plasmonics

As shown in the dispersion relation in Fig. 1, plasmonics is the optics or nanophotonics of near-UV, visible, and near IR. The dispersion curve for surface plasmons nearly meets that of light in terahertz and microwave bands, far from the plasma frequency, resulting in a greatly reduced plasmonic effect. Under such conditions, metals almost become complete conductors, and the velocity of the charge density wave propagating through the metals approaches that of light. The metals lose the near-field effect and wavelength compression effect, and act as reflectors, exhibiting characteristics similar to those of radio-wave antennas. On the contrary, in the energy range exceeding the plasma resonance frequency, metals become dielectrics and completely lose conductive and plasmonic effect. Plasmonics is not effective in UV, particularly deep-UV, range. However, photonics in UV will be a future key theme, as exemplified by UV microscopes, UV lithography, UV sterilization, and UV optical catalysis. It is highly expected that the plasmonics will be extended to the UV range. We have sought materials that exhibit the plasmonic effect in deep-UV, and found aluminum from among various candidates to carry out experiments on the plasmonic effect.57,58) Figure 6(a) shows the dielectric functions of metals in UV and visible. The resonance frequency of localized plasmons is given as the frequency at which the dielectric constant " is 2. The figure reveals that gold and silver exhibit the plasmonic effect in the visible range, whereas aluminum becomes a plasmonic material in the UV range. The imaginary part of the dielectric constant for aluminum is very large in the visible range, and the lifetime (propagation length) of plasmons is short; however, it is small in UV and deep-UV ranges, implying the existence of ideal plasmons. In addition, aluminum is a nontoxic and

010001-6

# 2013 The Japan Society of Applied Physics

COMPREHENSIVE REVIEW

Jpn. J. Appl. Phys. 52 (2013) 010001

because resonant Raman scattering occurs upon deep-UV excitation. Figure 6(b) shows the differences between the Raman spectra of biological cells under visible and deep-UV excitations.

Dielectric constant ε

(a)

5 0 -2 -5

Resonance

Au

Ag

6. Beyond Plasmonics

Plasmonic range

-10

Imaginary part of dielectric constant ε’’ Dielectric Constant ε

-15 0 100 200 300 400 500 600 700 Wavelength (nm) DUV UV NIR Vis 5 0 -2 -5

Resonance

Al

Rh

-10 Plasmonic range

4 3 2

1 0 0 100 200 300 400 500 600 700 Wavelength (nm)

DNA

DNA

cytochrome c cytochrome c peptide chain

amino acids DNA DNA amino acids DNA DNA

amino acids

Lipid

Excitation wavelength 532nm

amino acids glass cytochrome c cytochrome c peptide chain cytochrome c DNA

Raman scattering intensity (a.u.)

Excitation wavelength 244nm

amino acids DNA

(b)

1000 1200 1400 1600 Raman shift (cm-1) Fig. 6. (Color online) UV plasmonics. (a) Real and imaginary parts of dielectric functions (as functions of wavelength) of aluminum, rhodium, gold, and silver in UV and visible ranges. Aluminum behaves as a plamonic material in the deep-UV range.57) (b) Raman spectra of biological cells measured with deep-UV (244 nm) and visible (532 nm) excitation.58) DNA bases and amino acids are selectively detected owing to the resonance Raman effect upon deep-UV excitation.

noncombustible material with a very thin oxidized surface layer that can be used as a protective layer. DNA bases and amino acids, which are the most important biological molecules, exhibit characteristic Raman spectra, particularly upon deep-UV excitation. This is

As described in the previous sections, plasmonic field enhancement is equivalent to the cavity effect of nanoparticles and nanorods. Therefore, the dipole should have a physical size in order to be used as a cavity. The high value for Q of the cavity is necessary for the high enhancement of the field, while it narrows the spectral band. Many reports exaggerate the degree of field enhancement through the modeling of structures and finite-difference time domain (FDTD) calculation, but have rarely discussed on the issue of spectral range for the enhancement. The enhancement in a selective line or narrow spectral range is not practical for the applications including fluorescent, Raman scattering and absorption spectroscopy/imaging, but a high degree of enhancement in broad bands should be discussed. The excitation and fluorescent and Raman scattering wavelength bands of molecules have considerably wide ranges. Here, we focus on the limit of spatial resolution (or the minimum size of the hot spot) for plasmonics. It may have been believed that the resolution is given only by the physical size of particle and probe. As the size decreases, however, the number of atoms in metal decreases, thus plasmons, i.e., the collective oscillation of free electrons, are not excited. Excitation is still possible for a single-atom chain arranged in one-dimension or in two-dimension as a sheet. Cutting-edge research on this matter is being carried out by Nagao.59,60) We have reported some concepts that exceed the inherent limits of plasmonics while still using surface plasmons.45,61–63) One is the use of chemical effects. When a metallic nanoprobe and a sample are placed at a subnanometer separation, close to a contact state, the metal atom at the tip apex of the probe electronically interacts with the closest molecule of the sample, causing a change in the Raman spectrum, which is detected. The metal atom and the molecule instantaneously form a complex, changing the electron orbital of the molecule and the spectrum [Fig. 7(a)].63,64) Molecular-level spatial resolution can be theoretically obtained by monitoring the spectral change as the tip scans the sample. The orientation of molecule can also be detected through the spectral change due to the site of molecule adsorbed at metal atoms [Fig. 7(b)].45,64) The other concept is the use of force between a metallic nanoprobe and a sample. When a repulsive force is applied, a molecule of sample that first comes into contact with probe is distorted and changes its Raman spectrum, which is detected in the experiment [Fig. 8(a)].61) This method also requires a sub-nanometer distance control with a negative distance because of the repulsive force. Figure 8(b) shows an experimental diagram (left figure) and a one-dimensional transaxial image a carbon nanotube (right figure) obtained through the experiment.62) A silver-coated silicon cantilever was used as a probe and was scanned under a force of 2.4 nN. The figure reveals that the spatial resolution (full

010001-7

# 2013 The Japan Society of Applied Physics

COMPREHENSIVE REVIEW

m]

d [n

Me tal

com ple x

Ag

Adenine molecule

721 cm -1 Without interaction (far-field Raman)

d= 0.1

731 cm

d= 0.0

A peak emerges as a result of chemical enhancement

(a)

Metallic nanoprobe Pressurized area

Plasmon-enhanced field

-1

Frequency shift of Raman scattering

F

-1

Raman scattering frequency [cm ]

(b) (b)

F= 2.4 nN

Ag

750

F

F

740 Ag

730

Carbon nanotube

720 710 2

2.5

3

w

w+Dw

Sample molecule

680 700 720 740 760 780 Raman scattering frequency [cm-1]

w

w w

Pressure applied

Frequency shift [cm-1]

Ag nanoprobe

ω0 ω1 d= 155

10 8 6 FWHM:4 nm 4 2 0 0 10 20 30 Nanoprobe position [nm]

∞

(c)

F

Bond length between Ag atom and adenine molecule [Å]

Adenine nanocrystal

Fig. 7. (Color online) Exceeding beyond the limit of plasmonics using chemical adsorption. (a) Change in Raman spectra under the control of subnanometer distance between silver nanoprobe apex and adenine molecule.63) (b) Dependence of changes in Raman scattering shift in frequency as a function of distance between silver and adenine for different adsorption sites (calculation result obtained by density functional theory).64)

width at half-maximum) was 4 nm. The graph in the right part of Fig. 8(b) shows an image of a single single-walled carbon-nanotube, while it does not represent the Raman intensity distribution but spectral peak shift of G mode Raman scattering (1590 cm1 ). The peak shift distribution in Raman scattering represents the local strain distribution in the nanotube when tip scans the nanotube with a force. Figure 8(c) shows a one-dimensional image of an adenine molecular crystal. A spatial resolution of 4 nm was achieved as an edge response.62) The strain distribution for silicon devices was also measured. 7. A History and the Future

Plasmonics is not a new field and is 100 years old. I have written on the 50- and 100-year histories of plasmonics in review papers.7,19) In the foreword to the textbook written by Maier, Barnes ironically states, ‘‘You just have Maxwell’s equations, some material properties and some boundary conditions, all classical stuff — what’s new about that?’’65) He adds the follow-up remark ‘‘but surprises, adventure, the quest to understand — yes, we have all of those, and more’’. I think this nicely expresses the essence of science. The new and difficult are not always necessary. Rather than prestige and complexity, the innovation in science comes from the ideas that are pretty simple once their secret is known, I believe. Here I would like to add some remarkable contribution to plasmonics from Japan. The well-known long-range mode was reported by Fukui of Tokushima University for the first time in the world.25,26) Hayashi of Kobe University was the first to report the application of plasmonics to solar cells, to

50nm

F= 0.3 nN F

AFM image

Frequency [cm-1]

(a)

Raman scattering intensity [a.u]

Jpn. J. Appl. Phys. 52 (2013) 010001

730 728 726

4 nm

724 0 20 40 60 Nanoprobe position [nm]

Fig. 8. (Color online) Exceeding beyond the limits of plasmonics using mechanical perturbations.61,62) (a) Tip pressurizes a molecule. Results of imaging for (b) carbon nanotube and (c) adenine nanocrystal. A spatial resolution of 4 nm is achieved.

the best of our knowledge.66) Nanocircuits first appeared in a paper written by Takahara of Osaka University,28) and the authors were the first to develop a near-field optical microscope using a metallic tip and a tip-enhanced Raman microscope.8,35) Futamata, currently at Saitama University, has played a leading role in research on the enhancement of electric fields and the detection of single molecules.67,68) The plasmonics technology at the level of single-atom sequences and films developed by Nagao of the National Institute for Materials Science is unrivaled.59,60) Osawa, currently at Hokkaido University, has named the infrared absorption spectroscopy with a surface enhancement effect as surface-enhanced IR absorption spectroscopy (SEIRA).69) One can see the future when looking at the past. Note that the above leading achievements were obtained in fields that do not necessarily attract attention. Should focus on our own research without being overly swayed by fashionable research topics. In this report, there is no space to discuss plasmonic nanostructures. Technologies for manufacturing plasmonic nanostructures with various shapes, such as nanoshells, nanostars, nanoeggs, and nanocages, have been reported. In particular, nanoshells have been practically used because the mode with a peak in a long-wavelength band is effective for cancer treatment using near-IR light.70) The coupling of plasmons with propagating light and its applications have

010001-8

# 2013 The Japan Society of Applied Physics

COMPREHENSIVE REVIEW

Jpn. J. Appl. Phys. 52 (2013) 010001

19) S. Kawata and T. Ichimura: Optronics 29 (2010) 81 [in Japanese]. 20) A. Otto, I. Mrozek, H. Grabhorn, and W. Akemann: J. Phys.: Condens.

Table I. Plasmonics in the future.

Applications

Matter 4 (1992) 1143. 21) Surface-Enhanced Raman Scattering, ed. K. Kneipp, M. Moskovits, and H.

—Optical nanodevices, Optical nanocircuits —Spectroscopic nanoimaging (mainly Raman scattering)

Kneipp (Springer, Heidelberg, 2010).

Characterization and inspection of semiconductors, Analysis and evaluation of nanomaterials, Bio-imaging, Molecular imaging —Highly-sensitive highly-efficient optoelectronic devices (solar cells, light emitting diodes, lasers)

22) P. B. Johnson and R. W. Christy: Phys. Rev. B 6 (1972) 4370. 23) T. Okamoto and K. Kajikawa: Plasmonics — Fundamentals and Applica-

tions (Kodansha, Tokyo, 2010) [in Japanese]. 24) I. I. Smolyaninov, J. Elliott, A. V. Zayats, and C. C. Davis: Phys. Rev. Lett.

94 (2005) 057401.

25) M. Fukui, V. C. Y. So, and R. Normandin: Phys. Status Solidi B 91 (1979)

K61. 26) Y. Kuwamura, M. Fukui, and O. Tada: J. Phys. Soc. Jpn. 52 (1983) 2350. 27) K. Matsubara, S. Kawata, and S. Minami: Opt. Lett. 15 (1990) 75. 28) J. Takahara, S. Yamagishi, A. Morimoto, and T. Kobayashi: Opt. Lett. 22

—Highly-functional optical materials (optical catalysts) —Nano-photolithography, Nanofablication —Analytical sensors and medical diagnosis and therapy

(1997) 475.

(surface-plasmon sensors, DNA chips, biochips, cancer therapy)

29) H. Ditlbacher, A. Hohenau, D. Wagner, U. Kreibig, M. Rogers, F. Hofer,

F. R. Aussenegg, and J. R. Krenn: Phys. Rev. Lett. 95 (2005) 257403.

—Holography Breakthroughs

30) Y. Kawata, C. Egamia, O. Nakamura, O. Sugihara, N. Okamoto, M.

—Applications in deep-UV range

31) 32) 33) 34)

—Achieving resolutions of 1 and 0.1 nm —Development of nonlinear plasmonics

35)

been explained in detail in previous review articles and are not described here. Research on band-gap structures and their application to lasers has recently become more competitive because of stimulation from research on photonic crystals.71) Even structures with a high degree of field enhancement are useless if they strongly absorb light. The reduction of loss is the most important factor in the application of plasmonics to lasers, and our group has reported a structure that can realize this.72) Metamaterials are not mentioned in this review but are described in other review papers.73) The first report on a metamaterial with a negative refractive index had a striking impact on people, it cannot be decided at this moment whether the subsequent reports on a variety of structure studies have promoted the advancement of plasmonics or simply widen its range. Finally, the applications and expected future breakthroughs of plasmonics are summarized in Table I.

36) 37) 38) 39) 40) 41) 42) 43) 44) 45) 46) 47) 48)

1) Near-Field Optics and Surface Plasmon Polaritons, ed. S. Kawata

(Springer, Heidelberg, 2001).

49)

2) Nano Optics, ed. S. Kawata, M. Ohtsu, and M. Irie (Springer, Heidelberg,

2000) Chaps. 2 and 4. 3) Tip Enhancement, ed. S. Kawata and V. M. Shalaev (Elsevier, Amsterdam,

2007).

50) 51) 52)

4) Nanoplasmonics, ed. S. Kawata and H. Masuhara (Elsevier, Amsterdam,

2006). 5) S. Kawata, Y. Inouye, and P. Verma: Nat. Photonics 3 (2009) 388. 6) P. Verma, T. Ichimura, T. Yano, Y. Saito, and S. Kawata: Laser Photonics

Rev. 4 (2010) 548. 7) S. Kawata: to be published in Appl. Spectrosc. (2012). 8) Y. Inouye and S. Kawata: Opt. Lett. 19 (1994) 159. 9) P. Gleyzes, A. C. Boccara, and R. Bachelot: Ultramicroscopy 57 (1995)

318.

53) 54) 55) 56) 57) 58)

10) F. Zenhausern, Y. Martin, and H. K. Wickramashinge: Science 269 (1995) 11) 12) 13) 14) 15) 16) 17) 18)

1083. J. B. Pendry: Phys. Rev. Lett. 85 (2000) 3966. N. Fang, H. Lee, C. Sun, and X. Zhang: Science 308 (2005) 534. B. Liedberg, C. Nylander, and I. Lundstrom: Sens. Actuators 4 (1983) 299. K. Matsubara, S. Kawata, and S. Minami: Appl. Opt. 27 (1988) 1160. K. Matsubara, S. Kawata, and S. Minami: Appl. Spectrosc. 42 (1988) 1375. K. Matsubara, S. Kawata, and S. Minami: Opt. Lett. 15 (1990) 75. H. Kano and S. Kawata: Appl. Opt. 33 (1994) 5166. H. Kano and S. Kawata: Opt. Lett. 21 (1996) 1848.

59) 60) 61) 62) 63)

010001-9

Tsuchimori, and O. Watanabe: Opt. Commun. 161 (1999) 6. S. Kawata: Japan Patent No. 3196945 (filed in 1992, registered in 2001). S. Kawata and Y. Inouye: Ultramicroscopy 57 (1995) 313. S. Kawata, Y. Inouye, and T. Sugiura: Jpn. J. Appl. Phys. 33 (1994) L1725. T. Sugiura, T. Okada, Y. Inouye, O. Nakamura, and S. Kawata: Opt. Lett. 22 (1997) 1663. Y. Inouye, N. Hayazawa, K. Hayashi, Z. Sekkat, and S. Kawata: Proc. SPIE 3791 (1999) 40. N. Hayazawa, Y. Inouye, Z. Sekkat, and S. Kawata: Opt. Commun. 183 (2000) 333. N. Hayazawa, T. Yano, H. Watanabe, Y. Inouye, and S. Kawata: Chem. Phys. Lett. 376 (2003) 174. Y. Saito, N. Hayazawa, H. Kataura, K. Tsukagoshi, T. Murakami, Y. Inouye, and S. Kawata: Chem. Phys. Lett. 410 (2005) 136. T. Yano, P. Verma, S. Kawata, and Y. Inouye: Appl. Phys. Lett. 88 (2006) 093125. T. Yano, T. Ichimura, A. Taguchi, N. Hayazawa, P. Verma, Y. Inouye, and S. Kawata: Appl. Phys. Lett. 91 (2007) 121101. Y. Saito, K. Yanagi, N. Hayazawa, H. Ishitobi, A. Ono, H. Kataura, and S. Kawata: Jpn. J. Appl. Phys. 45 (2006) 9286. P. Verma, K. Yamada, H. Watanabe, Y. Inouye, and S. Kawata: Phys. Rev. B 73 (2006) 045416. Y. Saito, P. Verma, K. Masui, Y. Inouye, and S. Kawata: J. Raman Spectrosc. 40 (2009) 1434. T. Ichimura, N. Hayazawa, M. Hashimoto, Y. Inouye, and S. Kawata: Phys. Rev. Lett. 92 (2004) 220801. T. Ichimura, H. Watanabe, Y. Morita, P. Verma, S. Kawata, and Y. Inouye: J. Phys. Chem. C 111 (2007) 9460. Y. Saito, M. Motohashi, N. Hayazawa, M. Iyoki, and S. Kawata: Appl. Phys. Lett. 88 (2006) 143109. M. Motohashi, N. Hayazawa, A. Tarun, and S. Kawata: J. Appl. Phys. 103 (2008) 034309. A. Tarun, N. Hayazawa, and S. Kawata: Anal. Bioanal. Chem. 394 (2009) 1775. R. Matsui, P. Verma, T. Ichimura, Y. Inouye, and S. Kawata: Appl. Phys. Lett. 90 (2007) 061906. D. W. Pohl: Thin Solid Films 264 (1995) 250. L. Novotny and N. F. van Hulst: Nat. Photonics 5 (2011) 83. L. Novotny and B. Hecht: Principles of Nano-Optics (Cambridge University Press, Cambridge, U.K., 2011). P. Nordlander: Nat. Photonics 2 (2008) 387. L. Cao, P. Fan, E. S. Barnard, A. M. Brown, and M. L. Brongersma: Nano Lett. 10 (2010) 2649. A. Ono, J. Kato, and S. Kawata: Phys. Rev. Lett. 95 (2005) 267407. S. Kawata, A. Ono, and P. Verma: Nat. Photonics 2 (2008) 438. A. Taguchi, N. Hayazawa, K. Furusawa, H. Ishitobi, and S. Kawata: J. Raman Spectrosc. 40 (2009) 1324. Y. Kumamoto, A. Taguchi, N. I. Smith, and S. Kawata: Biomed. Opt. Express 2 (2011) 927. T. Nagao, S. Yaginuma, T. Inaoka, and T. Sakurai: Phys. Rev. Lett. 97 (2006) 116802. T. Nagao, T. Hildebrandt, M. Henzler, and S. Hasegawa: Phys. Rev. Lett. 86 (2001) 5747. T. Yano, Y. Inouye, and S. Kawata: Nano Lett. 6 (2006) 1269. T. Yano, P. Verma, Y. Saito, T. Ichimura, and S. Kawata: Nat. Photonics 3 (2009) 473. T. Ichimura, S. Fujii, P. Verma, T. Yano, Y. Inouye, and S. Kawata: Phys. Rev. Lett. 102 (2009) 186101.

# 2013 The Japan Society of Applied Physics

COMPREHENSIVE REVIEW

Jpn. J. Appl. Phys. 52 (2013) 010001 64) H. Watanabe, Y. Ishida, N. Hayazawa, Y. Inouye, and S. Kawata: Phys. 65) 66) 67) 68) 69) 70)

71) 72) 73)

Rev. B 69 (2004) 155418. S. A. Maier: Plasmonics: Fundamentals and Applications (Springer, Heidelberg, 2007). S. Hayashi, T. Yamada, and H. Kanamori: Opt. Commun. 36 (1981) 195. Y. Maruyama, M. Ishikawa, and M. Futamata: Chem. Lett. 30 (2001) 834. M. Futamata, Y. Maruyama, and M. Ishikawa: J. Phys. Chem. B 107 (2003) 7607. M. Osawa and M. Ikeda: J. Phys. Chem. 95 (1991) 9914. L. R. Hirsch, R. J. Stafford, J. A. Bankson, S. R. Sershen, B. Rivera, R. E. Price, J. D. Hazle, N. J. Halas, and J. L. West: Proc. Natl. Acad. Sci. U.S.A. 100 (2003) 13549. T. Okamoto, F. H’Dhili, and S. Kawata: Appl. Phys. Lett. 85 (2004) 3968. T. Okamoto, J. Simonen, and S. Kawata: Phys. Rev. B 77 (2008) 115425. For example, Tutorials in Metamaterials, ed. M. A. Noginov and V. A. Podolski (CRC Press, Boca Raton, FL, 2012).

Satoshi Kawata recieved his B. Sc, M. Sc, and Ph. D. all in Applied Physics from Osaka University in 1974, 1976, and 1979, respectively. Since 1993, he has been the Professor of Applied Physics. He has been the Chief Scientist at RIKEN from 2002 to 2012, and the Executive Director of the Photonics Center, Osaka University, since 2007. He has served as the President of the Spectroscopical Society of Japan, the Chair of International Council and a Board Member of Optical Society of America (OSA), Editor-in-Chief of Optics Communications, and now the vicePresident of the Japan Society of Applied Physics (JSAP). He has received various awards, such as the Medal with Purple Ribbon from the Emperor of Japan, the Spectroscopical Society of Japan Award, Leona Esaki Award, and the Japan IBM Science Award. Professor Kawata is a fellow of OSA, SPIE, IOP, and JSAP.

010001-10

# 2013 The Japan Society of Applied Physics