Selective Plasmonic Enhancement of Electric- and ... - ACS Publications

Jul 20, 2016 - such as lasers and fiber-based telecommunications. ... band optical technologies and also as complex photoluminescence (PL) band deeply...
0 downloads 0 Views 2MB Size
Letter pubs.acs.org/NanoLett

Selective Plasmonic Enhancement of Electric- and Magnetic-Dipole Radiations of Er Ions Bongseok Choi,† Masanobu Iwanaga,*,† Yoshimasa Sugimoto,† Kazuaki Sakoda,†,‡ and Hideki T. Miyazaki† †

National Institute for Materials Science (NIMS), 1-1 Namiki, Tsukuba 305-0044, Japan Graduate School of Pure and Applied Sciences, Tsukuba University, 1-1-1 Tennodai, Tsukuba 305-8571, Japan



S Supporting Information *

ABSTRACT: Lanthanoid series are unique in atomic elements. One reason is because they have 4f electronic states forbidding electric-dipole (ED) transitions in vacuum and another reason is because they are very useful in current-day optical technologies such as lasers and fiber-based telecommunications. Trivalent Er ions are well-known as a key atomic element supporting 1.5 μm band optical technologies and also as complex photoluminescence (PL) band deeply mixing ED and magnetic-dipole (MD) transitions. Here we show large and selective enhancement of ED and MD radiations up to 83- and 26-fold for a reference bulk state, respectively, in experiments employing plasmonic nanocavity arrays. We achieved the marked PL enhancement by use of an optimal design for electromagnetic (EM) local density of states (LDOS) and by Er-ion doping in deep subwavelength precision. We moreover clarify the quantitative contribution of ED and MD radiations to the PL band, and the magnetic Purcell effect in the PL-decay temporal measurement. This study experimentally demonstrates a new scheme of EM-LDOS engineering in plasmon-enhanced photonics, which will be a key technique to develop loss-compensated and active plasmonic devices. KEYWORDS: Plasmon enhancement, electric dipole, magnetic dipole, Er ions, LDOS optimization, plasmon nanocavity

L

because MD transitions are generally a second-order effect like electric quadrupole transitions. Metamaterials revived the interest in magnetic components in EM waves, related to the effective negative refractive index.9,10 Indeed, magnetic hot spot was numerically studied in cavity configurations.11 There have been several experimental reports focusing on the enhancement of MD transitions in the lanthanoid ions such as Eu and Er;12−19 the experiments were based on simple multilayer structures (or mirror configurations20) and large EM LDOS have not been experimentally explored. Although a recent trial to incorporate a plasmonic resonator was carried out for prominent plasmon-enhanced effect for the MD transition in Eu,21 the result was not prominent. Plasmonics is expected to realize many types of enhancement effects. PL enhancement is one of the issues. To date, highly significant fluorescence-intensity enhancements of more than 1000-fold for relevant references have been reported using fluorescent molecules in several papers.22−29 The latest advance

anthanoid series are key atomic elements in present day optical technology. As is well-known, they are employed in commercially available lasers, doped fiber amplifiers, and so on. Of the lanthanoid series, Er ions are incorporated in optical fiber amplifiers, serving in daily high-speed optical telecommunications at the C band.1,2 It is also known that Er ions have complex electronic transitions in the PL band. In particular, the transition from 4I13/2 to 4I15/2 is responsible for the PL band at 1.5 μm; the transition is purely a MD transition in vacuum, which is determined by the change of spin state, ΔJ = 1 and ΔS = 0, in initial and final states where J and S denote total angular momentum and total spin, respectively.3 The transition is however modified in host matrices, associated with splitting of the initial and final states due to the crystalline fields and with weakly allowed ED transition because of the perturbation of the electronic states.4−7 The 1.5 μm PL band consequently is comprised of deeply mixed MD and ED components. Multipole radiation from the lanthanoid Eu ions was experimentally detected in 1941.8 Optical excitation of the MD transitions, however, attracted less interest in spectroscopy than the ED transitions that are dominant in most materials © XXXX American Chemical Society

Received: June 1, 2016 Revised: July 12, 2016

A

DOI: 10.1021/acs.nanolett.6b02200 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

Figure 1. Scheme of plasmonic nanocavity array. (a) Design and optical configuration. The array is one-dimensionally periodic along the x-axis. (b) Resonant EM-field intensity distributions at 1.53 μm normal incidence. (c) Upper panel: reflection (green line) and absorbance (black dashed line) spectra. Lower panel: electric- (red) and magnetic-field (blue) intensities at the positions A and B in b, respectively. (d) Section-view STEM images of Er-doped plasmonic nanocavities. The Er-doped layers (white layers) of 20 nm thickness are indicated by red arrows, corresponding to the depth positions A and B in b, respectively. Red horizontal bars indicate 200 nm.

1b is 0.733 μm, which is about two times shorter than the incident wavelength 1.53 μm. The standing-wave modes were experimentally examined in a single nanocavity.34 Figure 1c shows the numerically calculated reflectance (R) spectrum (green line) of the plasmonic nanocavity array depicted in Figure 1a. Light absorbance α is also shown with a black dashed line, which is given by α = 1 − R because transmission is far below the detection limit and diffraction is not induced under the present conditions of interest, which are 1.3 μm and longer wavelengths under the incidence of small numerical aperture less than 0.4. We note that diffraction appears at 1.168 μm and shorter wavelengths in the calculated R spectrum. Figure 1c also shows field enhancement of electric and magnetic fields (lower panel). The enhancement was defined by the ratio of field intensity |F|2 to that of the incident plane wave in SiO2, where F denotes electric or magnetic field. The field enhancements exceed 150 at the maxima. We note that the electric and magnetic intensities inside the nanocavities can be approximated by electric LDOS (E-LDOS) ρE and magnetic LDOS (M-LDOS) ρM, respectively. Indeed EM LDOS is generally expressed as

satisfying both significant enhancement and uniformity has been attained with employing high-emittance plasmo-photonic metasurfaces.28−30 The PL-intensity enhancement of Er ions by plasmons was also studied;31−33 however, the best result was at most 13-fold enhancement in grating structures based on metal−insulator−metal films.33 In the studies on Er ions, the ED and MD transitions were not discriminated. In this Letter, we report significant plasmonic enhancement of PL in the 1.5 μm band, especially selective enhancement of the MD and ED transitions in Er ions. Although the two components are deeply superimposed, we explicitly resolved the MD and ED components by the technique of EM-LDOS engineering with deep subwavelength precision. We also observed a magnetic Purcell effect in the time-domain measurement in addition to a Purcell effect for the ED transition. Figure 1a shows the concept of a plasmonic nanocavity array that is one-dimensionally periodic along the x-axis; the periodicity P was set to 800 nm. The structure is assumed to be infinitely long along the y-axis. Incident light of transversemagnetic (TM) polarization travels from the top and excites the resonant modes inside the nanocavity of narrow width W. The designed W was 100 nm and the depth D was set to 550 nm, based on precise numerical calculations. The numerical method is described in the Supporting Information (Section S1). The design of structural parameters (P, W, and D) was chosen to induce a resonant mode at 1.53 μm in the nanocavity array; the intensity of EM field distribution at the normal incidence is shown in Figure 1b, showing that a 3/4-wavelength standing wave is induced inside the nanocavity. A and B indicate the positions at which maxima of electric- and magnetic-field intensities appear, respectively. Obviously, the two field components are well separated, which is an advantage to employ the one-dimensional cavitiy and becomes relatively difficult to be realized in two-dimensional cavities. Note that the effective wavelength of the standing-wave mode in Figure

ρF (x , z , ω) =

∑ A kF |Fk (x , z , ω)|2 δ(ω − ω k ) i

i

i

i

(1)

where Fki denotes the electric or magnetic field vector that is responsible for the dipole transition, AFki is a constant, and the index i runs over the final states.35 Because the resonant modes in the nanocavity array are isolated in energy,36 a single mode, that is, the 3/4-standing-wave mode, contributes to the sum in eq 1 at the angular frequency ωrad of radiative transition in Er ions. Therefore, when we express the single mode as i = 0 ρF (x , z , ωrad) ∝ |Fk 0(x , z , ωrad)|2 B

(2) DOI: 10.1021/acs.nanolett.6b02200 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

Figure 2. Er PL spectra from the Er-doped plasmonic nanocavity array, induced at 0.523 μm. Resonant wavelengths of the nanocavity array, corresponding to the dips of reflectance R (black line, upper panels), were tuned on and off to the PL band of bulk Er:SiO2 (black thin lines, lower panels). (a) Tuned off. Green line denotes Er PL spectrum from the nanocavity array. (b,c) Tuned on. Magenta and blue lines are PL spectra from Er-doped layer at the position of E- and M-LDOS maxima, respectively; see the positions in Figure 1d. (d) Tuned off. Yellow line is Er PL spectrum from the nanocavity array.

corresponding to the low R is equivalent to high emittance due to reciprocity.37 Making use of this property, similar plasmonic nanocavity arrays were applied to thermal emitters.38 In the tuned-on cases in Figure 2b,c, the measured R spectra at the normal incidence are in good agreement with the numerically calculated R spectrum in Figure 1c; the Er-doped layers in the nanocavity arrays were set at the positions of ELDOS (∝ |Ex|2) and M-LDOS (∝ |Hy|2) maxima, respectively; the positions were precisely set at the E- and M-LDOSmaximum positions, as shown in Figure 1d. The PL spectra (magenta and blue lines) in Figure 2 are quite different in the spectral shapes and dominantly come from the ED and MD transitions, respectively, as examined later. Figure 3a shows electric-field intensity distribution at the excitation wavelength 0.523 μm. The wavelength is suitable for efficient excitation of Er ions from 4I15/2 to 2H11/2.7 On the other hand, the wavelength is off-resonant to the nanocavity array. Figure 3b,c shows PL spectra from the nanocavity array in the tuned-on conditions of Figure 2b,c, respectively. In comparison with the PL spectra from the bulk Er:SiO2 (black lines), net PL enhancement was estimated to be 83 and 26 folds at the E-LDOS and M-LDOS optimal positions, respectively. The net PL enhancement in the vertical axis was evaluated by normalizing the number of Er ions and the excitation power that reached the Er-doped layers in Figure 3a. These highly enhanced PL come directly from the large EM LDOS in the plasmonic nanocavity and from the high emittance efficiency. The present well-designed PL-emitting nanocavity arrays demonstrate that spatially resolved EM-LDOS engineering of deep subwavelength precision works well. Here we note the evaluation of the net PL enhancement. The excitation power is not measurable in the Er-embedded positions; therefore, the numerically calculated values were substituted. As is shown in Figure 3a, the excitation wavelength of 0.523 μm is off-resonant to the nanocavity and the incident wave hardly travels into it. In addition, Au itself has large

Thus, the EM LDOSs inside the nanocavity are approximately proportional to the field intensities in Figure 1b; more specifically, eq 2 is written as ρE ∝ |Ex|2 and ρH ∝ |Hy|2. Figure 1d shows section-view images of the fabricated plasmonic nanocavities, which were based on the design in Figure 1b. The images were taken with a scanning transmission electron microscope (STEM). Er-doped layers (white layers) inside the nanocavity are indicated by red arrows, which were precisely located at the designed positions. The thickness of the Er-doped layers was 20 nm and the density of Er ions was 0.37 atom % in the samples used in optical measurement. The depth positions correspond to the positions A and B in Figure 1b. Although the nanocavities were slightly tapered along the depth direction, the slight modification hardly affected our initial design and concept based on the standing-wave mode. The nanofabrication procedure was conducted through thin-film deposition and electron-beam lithography, described in the Supporting Information (Section S2). Figure 2 shows PL spectra of Er ions doped in the plasmonic nanocavity array. The resonances of the nanocavity array were tuned on and off to the 1.5 μm PL band by changing the width W at the half depth. From Figure 2a−d (upper panels), the width W was finely modified from W = 80−130 nm. Accordingly, the resonance of the nanocavity array was shifted from 1.60 to 1.50 μm, moving across the Er PL band at 1.5 μm in bulk Er:SiO2 (black thin lines), which is shown for reference. In addition, the depth position T of the Er-doped layers, defined in Figure 1a, was changed from T = 365−10 nm in Figure 2a−d (lower panels). Measured PL spectra from the Erdoped nanocavity array are shown with colored lines. The excitation wavelength was 0.523 μm, suited to excite Er ions; the excitation light was emitted by a continuous-wave laser. We note that the PL intensities in Figure 2 were normalized at the peak. In the tuned-off cases in Figure 2a,d, PL spectra exhibit broad distribution at the vicinity of nanocavity-resonant wavelengths because the emission efficiency increases at the low-R wavelengths; note that the high absorbance α C

DOI: 10.1021/acs.nanolett.6b02200 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

Figure 4. (a) Decomposition of PL spectrum of bulk Er:SiO2 (gray) into ED (magenta) and MD (navy) components. (b,c) Top panels: electric- (red) and magnetic-field (blue) enhancement factors, βE and βM, respectively; the measured R spectra (black) are shown again for reference. Bottom panels: Measured raw PL spectra (solid lines) from the nanocavity arrays are fitted in b,c, using the ED- and MDcomponent PL spectra (red dashed lines), respectively.

Figure 4a shows PL spectrum of bulk Er:SiO2 (gray line) induced at 0.523 μm, which is quite similar in shape to the typical PL spectrum of Er-doped silica1 and the resolved ED (magenta line) and MD (navy line) components. Each resolved component was extracted by the analysis for PL spectra in mirror configurations15,20 where Er-doped SiO2 layers of 20 nm thickness were set at several distances from gold mirrors. In the mirror configurations, ED and MD decay rates can be determined analytically; therefore the ratio of ED and MD transitions in the PL band is also evaluated analytically. We thus decomposed the PL band into the ED and MD components in Figure 4a. The detailed extraction procedure is described in the Supporting Information (Section S3). Here we examine whether the spectral decomposition in Figure 4a is consistent with the ED- and MD-enhanced PL spectra in Figure 3b,c, respectively. PL spectra from the Erdoped nanocavity array are shown with colored solid lines in the lower panels of Figure 4b,c. We first express the PL spectra of ED and MD components in Figure 4a as IED and IMD, respectively. We next define field enhancement factors βE and βM by βE = |Ex,cavity|2/|Ex,SiO2|2 and βM = |Hy,cavity|2/|Hy,SiO2|2, respectively. The subscript cavity means EM fields inside the nanocavity, and SiO2 denotes EM fields in uniform SiO2 film. The numerically calculated factors βE and βM are plotted with red and blue lines in the upper panels of Figure 4b,c, where the measured R spectra (black curves) are also shown for reference

Figure 3. PL enhancement under E- and M-LDOS maxima in the nanocavity array. (a) Electric-field intensity distribution at the incidence of 0.523 μm. Arrows indicate the positions where |Ex|2 and |Hy|2 take maxima at the 3/4 standing-wave mode in Figure 1b. (b,c) Enhanced PL spectra of the Er ions doped at the E- and MLDOS maxima with magenta and blue lines, respectively. For comparison, PL spectrum from bulk Er:SiO2 is also shown with black lines, enlarged by 10 times for clarity. The net PL enhancement was evaluated by normalizing the number of Er ions and the excitation power at the Er-doped positions.

absorption coefficient at the wavelength.39 In the situation, the periodic structure is a corrugated Au film rather than a resonant nanocavity array, mostly inducing larger optical loss than the corresponding simulation. Thus, we do not have any sign to suggest a smaller optical loss than the simulation. It is consequently likely that we used a larger excitation power in evaluating the net PL enhancement and had the smaller values than the actual ones. Note that we present the estimated values and do not provide values based on a rigorous theory. D

DOI: 10.1021/acs.nanolett.6b02200 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

denote radiative and nonradiative, respectively. The observed tot tot tot ratio was Γtot cED/ΓbED = 5.2 and ΓcMD/ΓbMD = 3.5. Related to the enhancement of the spontaneous emission rate in Figure 5, Purcell factor is a well-known index to characterize the phenomenon.40 Strictly, the original Purcell factor is limited to ideal two-level radiative transitions; rad therefore, the ratio Γrad cEMD/ΓbEMD represents the Purcell factor. Although the nonradiative decay rate in bulk Er:SiO2 has not been reported, the quantum-theory calculation for radiative transition in Er:LaF3 showed a fairly good agreement with the measured decay rate,6 suggesting that the 1.5 μm PL band is not substantially affected by the nonradiative decay. It is thus rad likely that the relation of Γtot bEMD ≈ ΓbEMD holds. On the other hand, the interpretation for Γtot cEMD is not straightforward because the nonradiative decay rate Γnr cEMD in the nanocavity is undetermined. Indeed, in some plasmon-incorporated systems, definite PL quenching and significant increase in nonradiative decay rate were observed.41,42 Considering the present case, the net PL intensities increase by tens of folds, as shown in Figure 3. It is therefore unlikely that the nonradiative decay rate Γnr cEMD becomes dominant in the nanocavity arrays though we cannot conclude here that the Γnr cEMD is negligible. Thus, although it is tot indefinitive that the observed ratio Γtot cEMD/ΓbEMD is identical to the original Purcell factor, it is probable that the Purcell factors for the ED and MD transitions increases in this experiment. We note that the observed PL intensities are enhanced not only by the Purcell factors but also by the high emittance of the nanocavity array. This situation is similar to the fluorescenceenhanced case.28 Er ions are now working in optical-fiber amplifiers of far more than a meter length. In contrast, the amount of Er ions employed in the plasmonic nanocavity was very small because large EM-LDOS are available in the subwavelength dimensions. This is a distinct difference from conventional Er-doped systems in dielectric. Modulation of 1.5 μm PL band of Er ions was once tested in a one-dimensional photonic-crystal (PhC) cavity.43 The PL of Er ions efficiently came out through narrow transmission window of the PhC. In comparison with the present plasmonic nanocavity array, the EM LDOS in PhC cavity was not so large. Consequently, the Purcell effect was relatively smaller than this study, being limited to 20% modulation at most. Besides, the ED and MD transitions were not discriminated in the PhCcavity experiment. The 1.5 μm band PL enhancement of Er ions was attempted in two-dimensional PhC slabs based on silicon-on-insulator substrates.44 In accordance with the photonic band diagram, the PL was detected in an enhanced manner. The integrated intensity was at most 35-fold enhanced at a particular emission angle; the sensitive angle dependence may be a disadvantage in actual application. In the study, the ED and MD transitions were not observed separately. In conclusion, we have shown the EM-LDOS engineering in the plasmonic nanocavity array and achieved several-10-fold large and selective enhancement of PL coming from the ED and MD transitions in Er ions. Moreover, in addition to the ED-transition Purcell effect, the MD-transition Purcell effect was definitely observed in the decay-time measurement. Because a known drawback of plasmonic devices is the optical loss, we expect one of the key applications of the EM-LDOS engineering to plasmon-enhanced photonics is compensation of optical loss in plasmonic devices, which will enable ultrafast responses and comprise far smaller elements than other

and exhibit good agreement in the resonant wavelengths. We note that the βE and βM were evaluated by averaging the EMfield intensities on the Er-doped layers in the nanocavities. Compared to the field enhancement in Figure 1c, the βM is reduced due to the tapered structure of the nanocavities whereas the βE is less affected; the two factors are directly related to the PL enhancement in Figure 3. Let us define reconstructed PL spectra IPL by IPL(ω) = βE(ω)IED(ω) + βM (ω)IMD(ω)

(3)

Note that the PL spectra in Figure 4b,c are normalized at the peaks. In Figure 4b, a relation of βE ≫ βM is satisfied, and in Figure 4c a relation of βM ≫ βE holds. For example, at 1.53 μm, βE/βM = 189.3 in Figure 4b and βM/βE = 186.6 in Figure 4c, meaning that the smaller EM components are hardly included, less than 0.6%. In the previous mirror configurations,19 the smaller EM components were included at about 10%. Thus, the reproduced PL spectra (red dashed lines) in Figure 4b,c using eq 3 indicate that the PL spectra of the Er ions located at the Eand M-LDOS maximum positions dominantly originate from the ED and MD transitions, respectively. Although the measured PL spectrum at E-LDOS maximum condition in Figure 4b is a little broader than IPL, this may come from (i) the deviation of resonant wavelength of the nanocavity array from the Er PL peak and (ii) the deviation of the calculated βE from the actual one. Overall, it has became evident that the PL enhancements in Figure 3b,c represent individually enhanced ED and MD transitions. Figure 5 shows PL decay measured at 1.53 μm with 0.07 μm width, induced by 0.523 μm laser pulses with microseconds

Figure 5. Measured PL decay at 1.53 μm. Black dots: Er ions are in bulk Er:SiO2. Magenta and blue dots: Er ions are in the layer located at the E- and M-LDOS maximum positions of the nanocavity, respectively; see Figure 1b,d.

sharp edge. Black dots represent the PL decay of bulk Er:SiO2; the decay time is approximate 7.8 ms. Magenta and blue dots denote the decay of PL emitted from the Er-doped nanocavity array in the conditions of E- and M-LDOS maxima, respectively (Figure 1d); the decay time is approximate 1.5 and 2.2 ms, respectively. Clearly, the decay time becomes shorter by the Eand M-LDOS enhancement, respectively. The observed decay tot time reflects total decay rates Γtot bEMD and ΓcEMD where the subscripts bEMD and cEMD denote bulk ED or MD transitions and cavity ED or MD transitions, respectively. rad nr The total decay rates are written as Γtot bEMD = ΓbEMD + ΓbEMD and tot rad nr ΓcEMD = ΓcEMD + ΓcEMD where the superscripts rad and nr E

DOI: 10.1021/acs.nanolett.6b02200 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters photonic devices based only on optical fibers and Si photonics. Also, from a practical point of view, the operation wavelengths of 1.5 μm band are an advantage, compatible with the current optical telecommunication technologies.



(17) Taminiau, T. H.; Karaveli, S.; van Hulst, N. F.; Zia, R. Nat. Commun. 2012, 3, 979. (18) Karaveli, S.; Weinstein, A. J.; Zia, R. Nano Lett. 2013, 13, 2264− 2269. (19) Li, D.; Jiang, M.; Cueff, S.; Dodson, C. M.; Karaveli, S.; Zia, R. Phys. Rev. B: Condens. Matter Mater. Phys. 2014, 89, 161409. (20) Chance, R. R.; Prock, A.; Silbey, R. Adv. Chem. Phys. 1978, 37, 1−65. (21) Hussain, R.; Kruk, S. S.; Bonner, C. E.; Noginov, M. A.; Staude, I.; Kivshar, Y. S.; Noginova, N.; Neshev, D. N. Opt. Lett. 2015, 40, 1659−1662. (22) Kinkhabwala, A.; Yu, Z.; Fan, S.; Avlasevich, Y.; Müllen, K.; Moerner, W. E. Nat. Photonics 2009, 3, 654−657. (23) Schmelzeisen, M.; Zhao, Y.; Klapper, M.; Müllen, K.; Kreiter, M. ACS Nano 2010, 4, 3309−3317. (24) Fu, C.-C.; Ossato, G.; Long, M.; Digman, M. A.; Gopinathan, A.; Lee, L. P.; Gratton, E.; Khine, M. Appl. Phys. Lett. 2010, 97, 203101. (25) Zhang, W.; Ding, F.; Li, W.-D.; Wang, Y.; Hu, J.; Chou, S. Y. Nanotechnology 2012, 23, 225301. (26) Zhou, L.; Ding, F.; Chen, H.; Ding, W.; Zhang, W.; Chou, S. Y. Anal. Chem. 2012, 84, 4489−4495. (27) Punj, D.; Mivelle, M.; Moparthi, S. B.; van Zanten, T. S.; Rigneault, H.; van Hulst, N. F.; García-Parajó, M. F.; Wenger, J. Nat. Nanotechnol. 2013, 8, 512−516. (28) Choi, B.; Iwanaga, M.; Miyazaki, H. T.; Sugimoto, Y.; Ohtake, A.; Sakoda, K. Chem. Commun. 2015, 51, 11470−11473. (29) Iwanaga, M.; Choi, B.; Miyazaki, H. T.; Sugimoto, Y. Nanoscale 2016, 8, 11099−11107. (30) Iwanaga, M.; Choi, B. Nano Lett. 2015, 15, 1904−1910. (31) Managaki, N.; Fujii, M.; Nakamura, T.; Usui, Y.; Hayashi, S. Appl. Phys. Lett. 2006, 88, 042101. (32) Mertens, H.; Polman, A. Appl. Phys. Lett. 2006, 89, 211107. (33) Gong, Y.; Yerci, S.; Li, R.; Dal Negro, L.; Vučković, J. Opt. Express 2009, 17, 20642−20650. (34) Miyazaki, H. T.; Kurokawa, Y. Phys. Rev. Lett. 2006, 96, 097401. (35) Novotny, L.; Hecet, B. Principles of Nano-Optics, 2nd ed.; Cambridge University Press: New York, 2012. (36) Hooper, I. R.; Sambles, J. R. Phys. Rev. B: Condens. Matter Mater. Phys. 2002, 65, 165432. (37) Greffet, J.-J.; Nieto-Vesperinas, M. J. Opt. Soc. Am. A 1998, 15, 2735−2744. (38) Miyazaki, H. T.; Ikeda, K.; Kasaya, T.; Yamamoto, K.; Inoue, Y.; Fujimura, K.; Kanakugi, T.; Okada, M.; Hatade, K.; Kitagawa, S. Appl. Phys. Lett. 2008, 92, 141114. (39) Rakić, A. D.; Djurušić, A. B.; Elazar, J. M.; Majewski, M. L. Appl. Opt. 1998, 37, 5271−5283. (40) Purcell, E. M. Phys. Rev. 1946, 69, 681. (41) Anger, P.; Bharadwaj, P.; Novotny, L. Phys. Rev. Lett. 2006, 96, 113002. (42) Galloway, C. M.; Etchegoin, P. G.; Le Ru, E. C. Phys. Rev. Lett. 2009, 103, 063003. (43) Vredenberg, A. M.; Hunt, N. E. J.; Schubert, E. F.; Jacobson, D. C.; Poate, J. M.; Zydzik, G. J. Phys. Rev. Lett. 1993, 71, 517−520. (44) Galli, M.; Politi, A.; Belotti, M.; Gerace, D.; Liscidini, M.; Patrini, M.; Andreani, L. C.; Miritello, M.; Irrera, A.; Priolo, F.; Chen, Y. Appl. Phys. Lett. 2006, 88, 251114.

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.nanolett.6b02200. Detailed descriptions for the sample preparation, the analysis of ED and MD components in the PL spectra of bulk Er:SiO2, and the numerical method.(PDF)



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Present Address

(B.C.) Materials and Devices Advanced Research Institute, LG Electronics, Seoul, South-Korea. Author Contributions

M.I., K.S., and H.T.M. initiated this study and arranged the whole plan. B.C. mainly fabricated the plasmonic nanocavity arrays and measured the optical responses with assistance by Y.S. and H.T.M. The experimental data were analyzed by B.C., H.T.M., and M.I. All the authors contributed to the discussions and writing of the manuscript. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This study was supported by a Grant-in-Aid for Scientific Research on Innovative Areas (Grant 22109007) from the Japanese Ministry of Education, Culture, Sports, Science, and Technology and by the third midterm project “Innovative Photonic Materials” in NIMS. We thank N. Ikeda, M. Ochiai, T. Kasaya, and K. Mitsuishi (NIMS) for technical supports and the late Professor M. Hangyo (Osaka University) and T. Ochiai (NIMS) for discussions. Also, this work was partially supported by Nano-Integration Foundry, Low Carbon Research Network, Materials Analysis Station, and MANA Foundry in NIMS.



REFERENCES

(1) Miniscalco, W. J. J. Lightwave Technol. 1991, 9, 234−250. (2) Richardson, D. J.; Nilsson, J.; Clarkson, W. A. J. Opt. Soc. Am. B 2010, 27, B63−B92. (3) Bransden, B. H.; Joachain, C. J. Physics of Atoms and Molecules, 2nd ed.; Prentice Hall: Harlow, 2003. (4) Ofelt, G. S. J. Chem. Phys. 1962, 37, 511−520. (5) Judd, B. R. Phys. Rev. 1962, 127, 750−761. (6) Weber, M. J. Phys. Rev. 1967, 157, 262−272. (7) Polman, A. J. Appl. Phys. 1997, 82, 1−39. (8) Freed, S.; Weissman, S. I. Phys. Rev. 1941, 60, 440−442. (9) Pendry, J. B.; Smith, D. R. Phys. Today 2004, 57 (6), 37−43. (10) Shalaev, V. M. Nat. Photonics 2007, 1, 41−48. (11) Ameling, R.; Giessen, H. Nano Lett. 2010, 10, 4394−4398. (12) Thommen, Q.; Mandel, P. Opt. Lett. 2006, 31, 1803−1805. (13) Noginova, N.; Zhu, G.; Mavy, M.; Noginov, M. A. J. Appl. Phys. 2008, 103, 07E901. (14) Noginova, N.; Barnakov, Y.; Li, H.; Noginov, M. A. Opt. Express 2009, 17, 10767−10772. (15) Karaveli, S.; Zia, R. Opt. Lett. 2010, 35, 3318−3320. (16) Karaveli, S.; Zia, R. Phys. Rev. Lett. 2011, 106, 193004. F

DOI: 10.1021/acs.nanolett.6b02200 Nano Lett. XXXX, XXX, XXX−XXX