Performance enhancement of GaN-based light emitting diodes by the interaction with localized surface plasmons

Author's Accepted Manuscript

Performance enhancement of GaN-based light emitting diodes by the interaction with Localized surface plasmons In-Hwan Lee, Lee-Woon Jang, Alexander Y. Polyakov

www.elsevier.com/nanoenergy

PII: DOI: Reference:

S2211-2855(14)20316-8 http://dx.doi.org/10.1016/j.nanoen.2015.01.050 NANOEN715

To appear in:

Nano Energy

Received date: 5 September 2014 Revised date: 15 December 2014 Accepted date: 30 January 2015 Cite this article as: In-Hwan Lee, Lee-Woon Jang, Alexander Y. Polyakov, Performance enhancement of GaN-based light emitting diodes by the interaction with Localized surface plasmons, Nano Energy, http://dx.doi.org/10.1016/j. nanoen.2015.01.050 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Performance enhancement of GaN-based light emitting diodes by the interaction with localized surface plasmons In-Hwan Lee1,*, Lee-Woon Jang1, and Alexander Y. Polyakov1,2 1

School of Advanced Materials Engineering and Research Center for Advanced Materials Development, Chonbuk National University, Jeonju 561-756, Korea 2

National University of Science and Technology MISiS, Moscow, Russia

Abstract The effects of localized surface plasmons (LSPs) on the enhancement of photoluminescence and electroluminescence efficiency of GaN-based light-emitting diode (LED) structures are reviewed. It is shown that the LSPs formed by metal nanoparticles (NPs) or by local thickness variations of the metal films can contribute to the enhancement of light emitted by LED structures under optical or electrical excitation. The problems of choosing a suitable metal for such LSPs in the given spectral range are discussed. Various approaches to incorporating the LSP NP films into the LED structures are analyzed. The issues specific to different spectral ranges (blue, green, UV) are described. The wavelength range for which the application of LSP particles can be useful depends on the metal used, the shape and dimensions of the particles, and on their density and distribution. Roughly, as a first order approximation, it is concluded that Ag is in general the best material for the blue spectral region, Au is best for the green region, and Al is most suitable for the UV region. It is demonstrated that core/shell Ag/SiO2 NPs for the blue region have serious advantages in terms of stability over Ag NPs. For blue LEDs it is found that the most important factor is the suppression of the efficiency droop with increasing LED output power. For green and
 

UV regions the most important factor is the strong influence of the non-radiative recombination channels. In addition, for the UV LEDs the increased light extraction efficiency is also an important issue. It is shown how LSP NPs can provide efficient means to combat these problems.

Key words: localized surface plasmons, light emitting diodes, metal nanoparticles, external quantum efficiency

*Corresponding author: [email protected]

I. Introduction Surface plasmons (SPs) are collective oscillations of conducting electrons near the surface of metals. Such oscillations occur upon interaction with light and produce resonant absorption and scattering of light known as plasmon resonances. These resonances occur at specific frequencies of electromagnetic field whose frequency in the bulk, ωp, is a function of the density of conducting electrons and their effective mass, but for electrons near the surface depends also on the dimensions and shape of the metallic particle and the refractive index of the surrounding media (see, e.g. Ref. [1-4]). The strength of the electric field in the vicinity of SPs can be very high, which has a profound effect on absorption, scattering and spontaneous light emission of molecules, quantum wells (QWs) or quantum dots (QDs) placed in close proximity to SPs. When the dimensions of metallic particles become small this leads to the formation of another type of collective excitation called localized surface  

plasmons (LSPs) (Naturally, any surface irregularity, such as thickness variations induced by roughening or intentionally by lithography, will also lead to the formation of LSPs). For particles whose dimensions are small compared to the light wavelength the resonance frequency becomes a strong function of the dimensions, shape, and local environment of the nanoparticles (NP). The interference of electric field produced by several closely spaced LSP NPs can produce enormous concentration of local electric field that can strongly enhance the scattering, absorption or emission of molecules, QWs, QDs placed in the “right” location. For metallic particles whose dimensions in one direction are much longer than the wavelength the result could be effective propagation of light in this direction. Thus, an enormous field of research and practical applications called plasmonics is opening with the adoption of metallic nanostructures (see e.g. Ref. [1-4]). The completely new phenomena related to huge increase of molecular fluorescence or Raman scattering in the vicinity of metallic NPs, strong improvement of quantum efficiency of photovoltaic cells owing to the LSP-related concentration of electric field, formation of microantennae and microwaveguides based on LSP NPs have been discovered and studied. These studies have resulted in practical applications of metallic nanostructure LSPs in sensors measuring locally the refractive index of the gases and liquids and thus their composition, in extraordinary increase of sensitivity of fluorescence (so called surface enhanced fluorescence) and Raman scattering (surface enhanced Raman scattering) techniques making it possible to detect individual molecules, in development of new characterization techniques in biology and medicine allowing to detect and localize e.g. cancer tumors, and in many other exciting phenomena and applications. At the same time, a great progress has been achieved in developing techniques of  

fabrication and assembly of metallic NPs of desired dimensions and shape. This is a vast and rapidly developing field that we would not endeavor to even partially cover within the space of the present paper. The reader is referred to several excellent recent reviews covering different aspects of the problems and providing a good introduction to the field and supplying references to important original papers [1-8]. What we will concentrate on instead will be the physical phenomena related to one particular aspect of LSPs interaction with matter, namely, LSP phenomena in semiconductor light emitting diodes (LEDs), and, even more narrowly, LSP phenomena as related to performance of QW LEDs based on group III-Nitrides. Such LEDs have come recently to the forefront of research and development aimed at creation of new extremely efficient solid-state light sources for visible-UV spectral range (see e.g. [9]). The material basis of III-Nitride LEDs is constituted by solid solutions of InAlGaN and respective QWs and QDs. All these materials have direct bandgaps spanning the spectral range from near infrared (0.6 eV or 2000 nm for InN) to deep ultraviolet (DUV) (6.2 eV or 200 nm for AlN). These solid solutions have hexagonal wurtzite structure and can be grown by a variety of methods such as metallorganic chemical vapor deposition (MOCVD), molecular beam epitaxy (MBE), and hydride vapor phase epitaxy on various substrates of sapphire, SiC, Si, GaN, AlN, etc (see e.g. Ref. [10]). The main features of this material system are as follows: high dislocation density on the order of 106-109 cm-2 caused by lattice mismatch and thermal expansion coefficient difference between the substrates and the layers, high in-plane strain, strong piezoelectric polarization fields, problems with In solubility in InGaN solid solutions for In mole fractions exceeding ~30% causing serious deterioration of crystalline quality [11], reduction of the internal quantum efficiency (IQE) with increasing injection current  

(efficiency droop), degradation of crystalline quality of Al-rich AlGaN ternary solutions, problems with obtaining good p-type conductivity for such ternaries, low light extraction efficiency related to a small critical angle for total internal reflection, and so forth (see e.g. Ref. [12-15]). Despite all these problems very efficient blue LEDs based on multi-quantum-well (MQW) GaN/InGaN LEDs and white LEDs with yellow phosphor converters have been built and are believed to be on the way to supplant the existing incandescent and fluorescent electric lamps in general lighting applications [16]. High Al mole fraction AlGaN QW LEDs seem to be the only realistic option for fabricating solid-state light sources for DUV spectral region (see e.g. Ref. [17, 18]). Green LEDs and laser diodes (LDs) based on In-rich GaN/InGaN MQWs are rapidly progressing in quality and seem to be bridging the so-called “green gap” in solid-state light sources (see e.g. Ref. [19]). So what can be the role of LSP-mediated improvements for these different spectral regions? First of all, LSPs with resonance energy close to the photon energy of nearby QWs enhance spontaneous emission in the QW via the effect of strong electric field of the LSP. The LSPs have a very high density of photonic states. Both these factors create an alternative energy transfer route for charge carriers recombining in the QW (see the next section). For LSPs, this energy is transferred into emission of photons and therefore increases the IQE by suppressing the non-radiative recombination channels. Secondly, the additional scattering on LSP NPs can increase the light extraction efficiency (LEE). The expected consequences are rather different for blue LEDs and high-efficiency white LEDs based on them, for UV and DUV LEDs based on AlGaN QW LEDs with high Al mole fraction, and for green-red InGaN QW LEDs and LDs with high In mole fraction. In  

blue LEDs, the external quantum efficiency (EQE) has to be increased by another 20-30% to make them really competitive with the existing high-efficiency compact fluorescent lamps. At the same time, the cost of devices should be further driven down, e.g. by growth on cheap large area Si substrates (see e.g. Ref. [16, 20]). The IQE of the best blue LEDs grown on sapphire or GaN is close to 100 % at low driving currents and the main concern is the so called efficiency droop at high driving current caused by, as it has been convincingly demonstrated, the increasing contribution of non-radiative Auger recombination channel (see e.g. Ref. [21]), although current spilling out from the QWs and some other factors could also be at play. The non-radiative Auger recombination channel, as it is hoped, can be effectively suppressed by the LSP coupling, although direct evidence of that has yet to be strengthened. At the same time, for structures grown on Si, the IQE is not as high as for those grown on sapphire or GaN, presumably due to the presence of deep traps serving as non-radiative recombination channels (see e.g. [22]). In this respect, LED-mediated improvements of IQE can bring the efficiency of cheaper LEDs on Si close to the IQE of the best structures and contribute to improved cost efficiency. Together with expected improvements of LEE via LSP in blue LEDs this could greatly facilitate the proliferation of GaN/InGaN white LEDs in the general lighting applications. For UV LEDs with Al mole fractions below ~0.3, the IQE of the structures remains quite high, but the LEE drops dramatically compared to the blue LEDs because of the increasing contribution of the transverse magnetic (TM)-polarized light versus transverse electric (TE)-polarized light (see e.g. [15]) with Al mole fractions. Propagating mainly in the horizontal direction, TM-polarized light undergoes strong effects of total internal reflection due to the large incident angle on the interface of an LED chip, resulting in LEE decrease.  

Here in UV LEDs the main contribution of LSPs to the improved EQE could be in the domain of improved LEE. In DUV LEDs the IQE decreases dramatically both due to generation of deep traps (oxygen DX-centers, Ga and N vacancies and their complexes with C and Mg, see e.g. Ref. [23-25]) and to the increased level depth (activation energy) of Mg acceptors (see e.g. Ref. [13, 14]). Alongside with that the LEE continues to decrease owing to the increased contribution of TM-polarized light when moving towards AlN. Here LSP interactions can be instrumental in increasing both the IQE and the LEE of the structures. Additionally, one can hope that very strong fringe electric field associated with LSPs will be instrumental in increasing the fraction of ionized Mg due to the decrease of ionization energy of impurity in the presence of a strong electric field. In a large electric field, the hole requires a less thermal energy to get into the valence band because part of this energy comes from being pulled by the electric field. This can help in improving the hole injection efficiency and decreasing the series resistance effects. Finally, for green and yellow LEDs the structural quality of In-rich GaN/InGaN QWs deteriorates considerably resulting in the appearance of deep non-radiative traps decreasing the IQE at low injection levels. In addition, the impact of polarization fields becomes stronger in these highly strained QW structures which results in strong quantum-confined Stark effect (QCSE) causing the spatial separation of electrons and holes in QWs and lowering the luminescence efficiency (see e.g. Ref. [26]). Again, it can be hoped that the LSP-QW coupling facilitating radiative recombination could alleviate the QCSE effect. In what follows we discuss which metals show the best promise for use in LSPs in different spectral regions, outline the basic physics of the observed LSP-QW interaction,  

describe experimental methods of characterization for the LSP-QW coupling efficiency, present briefly the fabrication methods of LSP NPs of various metals, discuss the relative advantages of using core-shell NPs instead of simple metallic NPs, discuss published experimental and theoretical results for various LED structures, and finally outline the main approaches to practical fabrication of LSP LED structures for different spectral regions. Although, as clear from what was said above, the highly efficient blue LED structures stand to benefit the least from the LSP-mediated enhancement, the main bulk of experimental material has been obtained for these structures because they are most practically important, because the technology of fabrication of Ag NPs used in such LSP LEDs is the most advanced and because the physical principles are easier to demonstrate in the system with reasonably well controlled and understood behavior. The main advantages of LSPs, however, will be undoubtedly found for less advanced UV LEDs and green-red LEDs.

II. Basic physics underlying LSP interaction with QWs in various spectral regions The first question to be asked is what metals are good as candidates for LSP materials in the near infrared to UV spectral ranges. Semi-classical Drude theory provides reasonably accurate answers. Respective Maxwell equations allow exact analytical solutions in the case of spherers of arbitrary dimensions (Mie theory) or rods with dimensions much lower than the wavelength of incident light (Gans theory, good up-to-date summaries of both theories can be found in recent reviews [2-5,27,28]). Qualitatively it follows that good plasmonic materials for the given spectral range are the ones that have a strongly negative real part εr and a small positive imaginary part εi of the complex dielectric function (ε=εr+iεi). The  

extinction cross section of the metal NPs with radius R placed in the medium with the real part of the dielectric constant equal to εm and interacting with light with the wavelength of λ is proportional to (R3εm3/2/λ){εi/[(εr+2εm)2+εi2]} [5]. The resonant condition is then given by εr=−2εm, which, in air (εm=1), turns into εr=−2. This gives a reasonable estimate of the spectral position of the LSP resonance, while the quality factor QLSP=|εr/εi| provides a reasonably accurate measure of the width and strength of the resonance, as discussed e.g. in a recent paper [29]. Fig. 1(a), (b) taken from Ref. [29] gives the examples of the spectral dependences of dielectric functions for good near-UV (Ag) and UV (Mg, Rh, Al, Ga, In) LSP materials. Fig. 2(a) shows the estimated positions of LSP resonances for various metals based on the known spectral dependences of the dielectric functions. Fig. 2(b) displays the magnitudes of the maximum quality factors QLSPmax characterizing in a certain way the strength of the LSP resonance for various metals [29]. It can be seen that these crude estimates mark Au as a good LSP material for the visible photon energies range near 2 eV, Ag as a very good material for visible and the near-UV range (up to ~3 eV), Mg and Al as quite good LSP materials for DUV spectral range, with Ga and In also being a possibility. This qualitative approach is, of course, very imprecise. It does not provide the measure of the localized electric field E in the vicinity of the LSP structure, while the |E|2 is, in fact, a very important characteristic of the LSP interaction efficiency with the QW light source. To obtain these characteristics for NPs of arbitrary shape and to get the results in the more realistic settings of the NPs possibly consisting of a core comprised of one material and shell (or shells) comprised of a dielectric or another metal, one has to numerically solve respective equations. The three main algorithms developed for solving such problems are

 

the discreet dipole approximation (DDA), finite difference time domain method (FDTD), and finite element method (FEM). A brief description of the methods, references to original work describing the algorithms, analysis of comparative merits of these computational techniques can be found for example in excellent reviews [2-5]. (See also descriptions of other modeling approaches in Ref. [27, 28]). The authors of Ref. [29] used the DDA method to calculate the spectral dependence of the extinction cross section, the maximal squared fringe electric field |E|2max and the averaged squared field for spheres or semispheres of various metals placed on sapphire substrate in the normal light incidence geometry. They also analyzed the influence of changing the NPs radius and of native oxide formation on the surface of NPs (the latter effect particularly important for such easily oxidized metals as Ag, Al, Mg). The important conclusions are that, not unexpectedly, Au is found to be a good LSP material for the photon energy near 2 eV, Ag is good for energies close to 3 eV, Mg is suitable for near UV range of 3-4 eV, Al and Ga perform well in the DUV region near 4-5 eV. Increasing the sphere diameter shifts the resonance position to lower energy which is intuitively quite understandable. Interestingly, in many cases the spectral position of the maximum in the near-field |E|2 dependence is measurably shifted to lower energy compared to the spectral maximum in the far-field extinction cross section spectrum. One has to bear that in mind when analyzing the matching conditions for the LSP resonance and the peak QW emission energy (in most cases the decisions are made based on the far-field data, but, of course, it is the near-field data that really matters). The formation of native oxide leads to two oppositely directed effects: the blue shift of the LSP peak owing to the decreased diameter of the NP and the red shift caused by the
 

emergence of the oxide film with the refractive index higher than 1 [2, 29]. This means that, in order to achieve stable luminescent characteristics of QWs coupled with LSPs, one has to think about passivation of the metal NP surface. We will return to this issue when analyzing the performance of Ag NPs versus core/shell Ag/SiO2 NPs. The results obtained in Ref. [29] suggest that transition metals are not good LSP materials because of the low magnitude of the near field |E|2 and low quality factor of the resonance. This is understandable on the basis of the semi-classical considerations taking into account strong additional damping due to interband and intraband absorption in these metals [1-5, 29]. For Au, Ag, Al this interband absorption is also a factor. It restricts the Ag use to photon energies below 3-3.5 eV, the Au use to the photon energies below 2-2.5 eV. For Al, fortunately, respective absorption is near 1.6 eV and does not contribute strongly for the DUV region where the material is to be used. Interestingly, the authors of Ref. [29] find a good correlation between the magnitude of the near-field resonance and the magnitude of imaginary part of the dielectric constant εi. All in all, it appears that the choice of “good” LSP metals is rather limited to mostly Au for the red-green region, Ag for blue and nearUV, and Al for DUV spectral regions. However, other metals could also be considered, despite the respective LSP resonance being weak, if these metals can be comfortably intergrated into the LED fabrication process (Mg, Pt, Pd, Cr for the UV region are possible examples, see Ref. [29]). One has to also consider possible higher-order multiple interactions and the effects of interparticle interaction that can generally shift the resonance to longer wavelength, so that, for example, Ag can be successfully used as a mediator in the LSP enhancement of performance of green LEDs (see below section VII). The basic physics of interaction between the LSP NPs and the GaN/InGaN QWs has

 

been analyzed in some detail in Ref. [30, 31]. The authors treated the case of thin Ag layer deposited on GaN/InGaN QW structure, i.e. the SP rather than the LSP resonance situation. They pointed out that the decay time of spontaneous emission from the QW into the SP mode for emission energies close to the SP resonance energy was greatly affected by the presence of the SP fringe electric field and by the increased photonic density of states of the SP, compared to the emission into empty space. As a result, the reciprocal decay time of the SP turned out to be proportional to the product of the SP photonic density of states ρ(ω) and the squared module of the SP electric field magnitude |E(a)|2×ρ(ω), where ω is the circular frequency of light, a is the distance from the surface to the QW, and E(a) is the SP electric field strength at the depth corresponding to the QW. The density of photonic states is very high near the LSP resonance. The decay time was also found to decrease with increasing the Ag film thickness [30] Thus, in the QW coupled with SP, an additional fast decay mechanism emerges in the metal surface via rapid energy transfer from QW into SP. This mechanism competes against the standard radiative recombination and non-radiative recombination paths in the QW, but it can easily outperform both standard relaxation mechanisms because the SP decay is very fast. The energy transfer efficiency from the standard recombination to the SP mode, so called Purcell factor Fsp, is Fsp=1+τ0/τ1

(1)

where 1/τ0=1/τr+1/τnr, 1/τ1=1/τr+1/τnr+1/τSP, and τr, τnr, and τSP are, respectively, the decay times for radiative recombination, non-radiative recombination, and SP recombination [30, 31]. The authors of Ref. [30] have measured the decay times of the GaN/InGaN QW with and without a thin (8 nm or 6 nm) Ag film. The QW was located at 12 nm from the surface
 

coated with Ag film. Without the Ag film the photoluminescence (PL) decay was characterized by a single exponent with the decay time slightly varying in the vicinity of the QW peak as measured by time resolved PL (TRPL) spectra. Ag deposition led to a strong decrease of the short relaxation time and to the appearance of a long tail in relaxation. The short decay time was a strong function of the Ag layer thickness and was thus associated with the QW coupling to the SP resonance. The spectral dependences of the Fsp values for the Ag films with thicknesses of 6 nm and 8 nm are compared to the modeling results in Fig. 3 taken from Ref. [30]. The agreement between the model and the experiment is quite reasonable and demonstrates a very efficient transfer of the energy of recombining electronhole pairs into the SP state (Fsp values on the order of 50-60). Mind that the authors of Ref. [30, 31] were dealing with planar Ag films forming a SP resonance rather than a LSP resonance. Thus, although the energy of electron-hole pairs recombining in the QW was very efficiently converted into the SP excitation, this did not result in the external increase of the photon intensity. Instead, the excitation energy was transformed into the energy of phonons in the Ag film. Correspondingly, the PL intensity of the QW structure with Ag film, when corrected for the absorption in Ag, was lower for the QW PL band in the vicinity of the SP resonance, while the spectrum in other spectral regions was not seriously affected (see Fig. 4 taken from Ref. [31]). In order to extract the energy of the surface plasmon in the form of photons, one has to switch from the planar metallic film producing the SP resonance to nanostructured metallic film with characteristic features much smaller than the wavelength of light, i.e. to the LSP resonance conditions (see e.g. Ref. [32, 33]). While the results presented in Ref. [30, 31] consistently demonstrate and model the SP coupling, the works reported by Okamoto et al.
 

[34, 35] were among the first demonstrations of the LSP resonance coupling with GaN/InGaN QWs producing external photons. The authors capitalized on the roughness of relatively thick (~50 nm) Ag, Au or Al thin films deposited on the surface of blue GaN/InGaN QW structures. This roughness made it possible to convert the energy of SP excitation into photons exiting the metallic film, as illustrated in Fig. 5 taken from Ref. [35]. The presence of roughness profile with characteristic features smaller than the wavelength of light was confirmed by atomic force microscope and scanning electron microscope (SEM) imaging of the metal surface [34, 35]. The non-uniformity was also induced by lithographic patterning of the Ag films. For blue GaN/InGaN QW structures with the QW located 10 nm below the surface the authors observed a strong enhancement of PL at room temperature for Ag, less strong enhancement for Al, and no enhancement for Au, which was explained by good energy matching of the Ag LSP resonance to the QW emission peak and less efficient non-resonant matching for the Al LSP [34, 35]. For Au, the LSP peak was too weak at ~2.8 eV to provide any efficient coupling. The authors measured the PL decay times for metal-covered and metal-free GaN/InGaN QW samples. From that they calculated the Purcell factors for the enhancement, as described above, and obtained a reasonable agreement with the Purcell factors and their spectral dependence reported in Ref. [30, 31]. Concerning the works in Ref. [30, 31], mind that, because the QW structures showed very weak contribution of the non-radiative recombination and because almost the entire energy of electron-hole pairs radiatively recombining in the QW region was converted to the SP excitation and lost to the external PL intensity, the ratio of PL intensities in the QW structure without the Ag film and with the Ag film was in that case close to the Purcell factor determined from Eq. (1), but this is not the case generally. For LSP structures in
 

which the LSP energy can be converted into the energy of external photons, one can have a very high Purcell factor indicating a strong coupling between the QW and the LSP states, but the ratio of respective PL intensities can be much lower than the Purcell factor and determined by the relative contribution of the non-radiative recombination channel in the QW. Measurements of the temperature dependence of PL intensity allow to determine the IQE of the structures at room temperature under the assumption that at low temperatures (typically 4-10K) the contribution of non-radiative recombination is very small (excitons do not get to the non-radiative recombination sites) and the IQE is equal to 100%. Then the room temperature IQE can be calculated as the ratio of PL at room temperature and the low temperature. It is shown in Ref. [34, 35] that the Purcell factor at room temperature in that case can be estimated as (1-IQE)/(1-IQE*), where IQE* is the internal quantum efficiency in the presence of LSP (naturally, this expression follows from the general expression of Purcell factor via respective decay rates). From these measurements it was deduced that the IQE value of GaN/InGaN QWs could be increased by several times owing to the suppression of the non-radiative recombination channel by effective energy transfer to the LSP and conversion of this energy to photons [34, 35]. These papers have provided a great boost to the work on plasmonic enhancement of GaN/InGaN QW LEDs. However, they have also drawn attention to the two major points in such research: First, that a much tighter control over the conversion of the LSP energy into the external flow of photons rather than into ohmic losses in the metal is needed. And second, that serious alteration would be necessary in LED fabrication technology in order to practically use the LSP enhancement. The latter is due to the very limited extent of the
 

fringe electric field of LSP. For continuous metal films and SP resonance, the distance Z (penetration depth) from the metal film to the point at which the SP electric field falls down to 1/e of the initial value is given by the expression, Z=λ/2π{[εr(GaN)εr(metal)]/εr(metal)2}1/2, where εr(GaN) and εr(metal) are the real parts of the dielectric constants of GaN and metal, respectively, λ is the wavelength of light [30, 32, 34, 35]. For a typical blue LED with the photon energy of 2.7-2.9 eV, the penetration depth is estimated to be 40-47 nm for Ag and 77 nm for Al [30, 32, 34, 35]. Meanwhile, the top part of a typical GaN/InGaN QW LED should consist of at least about 100-nm-thick p-GaN contact layer following a p-AlGaN electron blocking layer of ~10 nm in thickness (see e.g. Ref. [36]). At such thicknesses the effect of SP enhancement will be totally lost, as demonstrated for continuous Ag and Al films [34, 35]. The work reported in Ref. [37, 38] suggests that the decrease of the non-radiative energy loss in SP via leakage through the ohmic contact could be expected if the Ag layer is deposited on a thin dielectric layer, such as SiO2 or SiN. Moreover, under certain conditions, a wave-guiding action in the SP/dielectric/GaN can ensue. It has also been pointed out that in such a fixture it is essential that the ohmic contact to the p-GaN layer is electrically insulated from the SP Ag layer on SiO2. An alternative approach is provided by using true metallic NPs or core/shell nanostructures to obtain the localized surface plasmon resonance. In that case LSP particles of desired shape and composition can be synthesized and deposited on the surface or formed by lithography (see next section). The existing modeling methods allow in principle to obtain the electric field distribution for NPs of arbitrary shape and composition and to predict the LSP resonance spectrum and the amount of coupling to the QW emission. For example, Fig. 6 taken from our paper [39] shows the electric field distribution for Ag/SiO2
 

core/shell NPs on top of GaN/InGaN blue QW structure as calculated by 3-D FDTD modeling. The upper panel presents the electric field distribution for clusters of 1, 3, and 7 NPs (NPs) with the Ag core of 60 nm in diameter and the SiO2 shell of 20 nm in thickness for the illumination from below by the 2.7 eV light of the GaN/InGaN QW (respectively, Fig. 6(a), (b), and (c)). The lower panel (Fig. 6(d)) displays the LSP electric field decay in the GaN/InGaN structure for clusters with 1, 3, 7 NPs. The inset in the figure presents the calculated extinction spectrum of the seven NP clusters. It can be seen that the electric field distribution is highly nonuniform, with the field being concentrated near the NP and forming hot spots in the GaN spacer under the NPs and between the NPs. The LSP field intensity increases dramatically with the increased cluster size, while the penetration depth increases from 8 nm for 1 NP cluster to 18 nm for 7 NPs cluster. These values are considerably smaller than for continuous Ag films and the electric field at 100 nm from the surface is about an order of magnitude weaker than at the surface. The decreased penetration depth is partly due to the presence of the SiO2 shell [39]. Note, however, that the electric field decay curve is not a simple exponent and has a long tail extending rather deeply into GaN. This low penetration depth creates serious complications even for the LEDs with a thin top GaN layer. Indeed, the best performance of blue GaN/InGaN QW LEDs is achieved for multiple, usually 3-5, QWs with typical QW width of 2-3 nm and typical GaN barrier width of about 10 nm (see e.g. Ref. [9]). Hence, even for the very thin top GaN layer the LSP field distribution and the PL enhancement thereof will be very different for the uppermost and the lowermost QWs. Experimental studies performed by us for the MQW GaN/InGaN LEDs with Ag/SiO2 LSP NPs on top and the backside PL excitation showed that the main contribution to the LSP enhancement is provided by the topmost QW [40]. Therefore, to
 

capitalize on the LSP enhancement of the MQW GaN/InGaN LEDs serious alterations have to be incorporated into the LED fabrication process in order to bring the metallic NPs closer to the active LED region without compromising the LED performance. We will discuss below several approaches proposed in the literature in order to solve the problem, but it has to be said outright that the optimal practical solution has yet to be found.

III. Preparation of LSP NPs for different spectral regions Nowadays, LSP NPs of different composition, different dimensions, and different shape can be prepared by a variety of methods. The NP fabrication techniques are in the most advanced state for Ag and Au NPs due to the long history of the use of these metals in photonics (see e.g. Ref. [2]) and to their wide use in various plasmonic applications (see e.g. Ref. [5]). A very good review of different fabrication approaches can be found in Ref. [5]. Briefly, these methods can be separated in several main branches: 1) chemical synthesis in solution (for Ag NPs the versions of the method include, among others, citrate reduction, silver mirror reaction, polyol process), 2) seed-mediated synthesis, 3) template mediated synthesis, 4) lithography methods including standard lithography, electron beam lithography, focused ion beam lithography, nanosphere lithography. Chemical synthesis can result, depending on the chosen method, in differently shaped NPs (quasi-spheres, nanocubes, bipyramids) with a wide range of dimensions from several nanometers to hundreds of nanometers and a degree of control over the shape and dimensions varying for different versions of the methods [5]. The NPs thus obtained are usually single crystalline. Lithographic methods using electron beam or ion beam are usually too time-consuming and costly to be practically applied in LSP-mediated LEDs. Rather, in this case either relatively
 

large particles produced by standard lithography or smaller particles produced by nanosphere lithography are employed. The structures thus obtained are polycrystalline, which could be a disadvantage in LSP applications due to stronger losses in metallic layers. The field of NP fabrication is extremely vast and we would not even try to fully cover it here. Instead, we present below several successful examples of metallic NP fabrication for use in LSP-enhanced LEDs for various spectral ranges. Ag, as pointed out above, is the best LSP material for performance enhancement in blue LEDs. The spherical silver NPs of approximately 30 nm in diameter can be prepared by sol-gel method as follows. First, a 500 ml beaker is filled with 180 ml of aqueous solution including cetyl trimethyl ammonium bromide (0.145 g) under vigorous magnetic stirring. Next, the prepared aqueous solution of silver nitrate (0.1 M, 10 ml) is added to the mixed solution. And then, 20 ml of ascorbic acid in aqueous solution is added to the mixture solution slowly within 5min. After the mixture has been further stirred for 10 min, sodium hydroxide (0.1 M) is added to accelerate the chemical reaction, and the pH of the mixed solution is set at about 5. Fig. 7 presents the transmission electron microscope (TEM) image of the Ag NPs thus produced [41]. In order to form core/shell Ag/SiO2 NPs, 50 ml of ethanol and a certain amount of tetraethoxysilane (TEOS) should be added into the abovementioned silver colloids. The solution should be stirred for three more hours at room temperature. The TEM images of the resultant Ag/SiO2 core/shell structures are shown in Fig. 8 [42]. The radius of the Ag cores is close to 30 nm and can be varied by changing the amount of silver nitrate in the solution, and the thickness of the SiO2 shell can be effectively controlled in the 2 nm-20 nm range by the amount of TEOS added to the solution. Fig. 8 shows typical core/shell NPs prepared using this technique.
 

In both cases of Ag and Ag/SiO2 NPs, the coating of NPs can be done by drop casting on the surface of the MQW structure and drying at 100oC. The areal density of the NPs is in the range of 1-4×1010 cm-2 [41]. Such NPs produce effective LSP resonances near 2.7-2.9 eV. The resonance frequency increases with decreasing the Ag core diameter and decreases with increasing the SiO2 shell thickness, as predicted by theory [42] (Fig. 9 taken from Ref. [42] illustrates the LSP resonance frequency dependence on the SiO2 shell thickness). Such synthesized NPs have been widely used in experiments aimed at improving the efficiency of blue LED structures that we will describe in the next section. There are different ways to deposit such NPs on the surface of GaN/InGaN QW structures. Compared with the drop-casting and drying, spin-coating with subsequent drying could be more appropriate for uniform deposition on large area wafers. In the next section we will also discuss possible applications of synthesized Ag or core/shell NPs in conjunction with nanopillar QW LED structures. In that case one would like to have the NPs deposited on the sidewalls of nanopillars, but not on top of them. In achieving that the socalled stamping method [43] would be useful. Another approach to the formation of Ag NPs is by depositing a thin film of Ag and annealing it. This approach has been applied in several papers with reasonably good results. For example, in Ref. [36] a thin Ag layer was deposited on n-GaN and annealed at 750oC for 10 minutes. The starting film showed clustering of Ag with the in-plane particles dimensions of 250±50 nm and thickness of 8±4 nm. After annealing the dimensions increased respectively to 450±50 nm and 15±5 nm and the areal density of NPs was around 6.4×107 cm-2. These NPs were successfully used in overgrowth experiment of the active QW region on the n-GaN buffer layer with NPs (to be discussed in the next section).  

In Ref. [44] Ag NP were produced near the active MQW region of the blue LED structure by deposition of 0.3 nm-thick or 0.6 nm-thick Ag films and overgrowing them with 50-nm p-GaN layer at 800oC with subsequent regrowth with thick (0.15 µm) p-GaN contact layer at 970oC in an attempt to bring the Ag LSP NPs as close as possible to the MQW region. Fig. 10 shows the TEM image of the MQW region and the Ag NPs located very close to it, at about 30 nm. It was found that, irrespective of the thickness of starting Ag layer the Ag film broke into spheroid NPs with in-plane dimensions of 50± 15 nm, and height of 40±10 nm. The effect of the varying starting thickness was in the increased areal density of NPs with increasing the starting Ag layer thickness (from 8×108 cm-2 to 4×109 cm-2 for the starting Ag layer thickness of 0.3 nm and 0.6 nm, respectively). A similar approach was used to produce small Au NPs with characteristic dimensions of 80± 20 nm and 8± 5 nm and density of 2×108 cm-2 built inside the p-GaN film close to the GaN/InGaN QW by annealing a 0.2-nm-thick Au film deposited on thin p-GaN [45]. In an attempt to improve the structural quality and electrical performance of the overgrown p-GaN contact layer with the Ag NPs incorporated into it close to the MQW region, the authors of Ref. [46] covered the Ag NPs by 40 nm-thick SiO2 discs fabricated by photolithography as shown in Fig. 11. (The Ag NPs in that case were produced by annealing the Ag layer at 500oC). For green LEDs Au is the best SP or LSP material, as mentioned earlier. Most of the work on SP-enhanced green GaN/InGaN LEDs was performed using continuous Au layers roughened by annealing (see Ref. [47]). In some cases the annealed layer was further structured by standard photolithography [48]. Ag layers roughened by annealing was also used in some cases to promote the LSP-mediated performance enhancement of green LEDs [48, 49], although for Ag NPs the LSP coupling is non-resonant and thus should be less 
 

efficient than coupling with well spectrally tuned Au NPs. The possibility to use such outof-tune NPs for the LSP enhancement is based on the LSP resonances being generally quite broad and showing extended LSP enhancement tails into the non-resonant energies. In principle, there are no serious obstacles to using single-crystalline Au or core/shell Au/SiO2 NPs prepared by different versions of sol-gel synthesis. For example, in Ref. [50] it has been demonstrated that Au shell NPs can be prepared by seed-mediated electroless of Au on silica spheres. The Au/SiO2 core/shell NPs can then be fabricated by adding TEOS to the growth solution, as in the case of Ag/SiO2 core/shell structures described above. The Au/SiO2 NPs thus produced showed intense LSP resonance whose peak energy could be easily regulated in the 500-600 nm wavelength by varying the Au diameter and extended up to ~800 nm by increasing the SiO2 shell thickness by increasing the TEOS concentration [50, 51]. For UV region down to ~320 nm, Ag is still a reasonably good LSP material [52]. Pd has also been used in the near-UV region of ~400 nm [53]. For 380-400 nm near-UV LEDs, Ag and Pt NPs produced by the annealing of continuous layer have been used to form LSP structures [54]. Al, however, is a better option for this near-UV range and seems to be the best solution for the DUV spectral region. The simplest way to produce Al NPs is the roughening of continuous Al layers by annealing [55]. In Ref. [56] a more advanced photolithographic approach was used to form LSP structures to enhance the performance of DUV AlGaN LEDs with peak wavelength near 350 nm. In this approach, 90-nm-deep and 5-µm-wide flat holes with 10 µm pitch were produced in the p-AlGaN top layer having ohmic contact metallization and Al slabs were then deposited into the holes to form the LSP resonance close to the active QW region, as shown in Fig. 12.  

Much smaller Al LSP features can be produced by nanosphere lithography using a colloidal monolayer self-assembled hexagonal mask formed by polystyrene beads with the starting diameter of 300-400 nm. Al is deposited by vacuum evaporation through the openings between the beads. By exposing the mask to microwave heating the polystyrene beads diameter could be controllably increased thus decreasing the size of the openings. Fig. 13 shows the observed Al NPs patterns in this technique as a function of microwave heating time. The shape of the NPs changed from triangular pyramids to discs and the dimensions gradually decreased from about 80 nm to 50 nm, with the spectral position of the LSP resonance wavelength shifting respectively from 340 nm to 270 nm (see Fig. 13 and 14 taken from Ref. [57]). As we see, there exists a wide range of approaches to fabrication of NPs for use in LSP enhancement of LED performance. The questions to be further answered are: 1) which techniques produce a better performance in terms of luminescence efficiency enhancement and what generally are the requirements to the LSP structures in terms of high efficiency of conversion of the LSP energy into the externally propagating photons, 2) what the relative stability of the effect is for different types of NPs, 3) which techniques are more easily incorporated into the LED fabrication process, 4) what the best positions to put the NPs in are to obtain the strongest LSP effect. We will consider at least some of these issues in the next section when analyzing the LSP effects for blue LEDs.

 

IV. LSP effects in blue LEDs IV.1. LSP/QW coupling efficiency and quantum efficiency of blue LEDs It is instructive to begin this analysis of the LSP effects in blue LEDs by presenting the EQE of LEDs as the product of the LEE and the IQE, EQE=LEE×IQE.

(2)

Okamoto et al. [34] analyze the changes introduced into the IQE value by coupling to the SP or LSP film. This altered value of IQE (IQE*) is expressed via the decay rates for radiative (kr) and non-radiative (knr) transitions, that in the presence of SP or LSP coupling (kSP), and the efficiency of converting the SP or LSP energy into the energy of photons (Cphext) IQE*=(kr+Cphext×kSP)/(kr+knr+kSP).

(3)

For high conversion efficiency of the SP mode energy into the energy of photons, i.e. when Cphext=1, Eq. (3) becomes simply the ratio of respective decay rates with and without nonradiative recombination. At that the Purcell factor determined as the ratio of decay rates with and without LSP can be written as [34] Fsp=(kr+knr+kSP)/(kr+knr) =(1-IQE)/(1-IQE*)§τ0/τSP. (4) The PL enhancement, when one does not take into account possible influence of LSPs on light extraction efficiency, is simply the ratio of respective IQE values. For the samples studied in Ref.[34] the room temperature IQE value determined from the ratio of PL intensities at 4.2K and 300K was about 6% without the Ag layer and about 40% with Ag layer (see the respective temperature dependences in Fig. 15). The Purcell factor at 470 nm (the QW PL peak position) was determined to be about 1.6 from Eq. (4) and the ratio of PL intensities with and without Ag should have been 6.8 if no contribution from the LEE factor in Eq. (2) was expected. Instead, the  

observed PL intensity increase was 14 (see Fig. 16 taken from Ref. [34]). The authors did not discuss this discrepancy, but to us it suggests that either the determination of room temperature IQE from the ratio of PL intensities is not quite accurate or the LEE is in fact changing as a result of Ag film deposition. It should be noted that Okamoto et al. [34] have not considered the several potentially important factors, such as contribution of the PL emission into the upper halfspace and possible increase of the injection level in the QW due to the light reflected from the metal layer. It is difficult to accurately estimate the magnitude of these contributions, but they could be substantial. The spectral dependence of Purcell factor measured in Ref. [34] indicated that the value should greatly increase (to about 40) when approaching the SP resonance energy about 440 nm for the Ag film used in Ref. [34]. It was also observed that increasing the separation between the QW and the Ag (or Al) film decreased the PL enhancement factor by about an order of magnitude (see Fig. 17), presumably owing to the strong decrease in the magnitude of the SP fringe electric field and respective decrease in the Purcell factor. Several things are in order for discussion here. First, how justified is the assumption that the LSP energy conversion efficiency into photons is close to 1, how this efficiency changes for different types of LSP filmes (continuous layers broken down into fragmented NPs, films of that type with imposed gratings pattern, synthesized NPs of different size and concentration). Second, how important is the absolute magnitude of the Purcell factors, how they actually vary with distance. Third, what will be the impact of the LSP effect in high-quality blue LED structures whose IQE are already close to 80-90%. The data presented in Ref. [34] and also those discussed in the previous section suggest that even nominally continuous films are broken down into separate NPs in order to produce outward  

PL enhancement. In Ref. [34] the characteristic in-plane dimensions of NPs in the nominally 40nm-thick Ag film were around 200 nm and it was clearly demonstrated that the waveguiding grating with the 200-nm-wide and 100-nm-opening stripes strongly increased the luminescence intensity. Increasing the grating period beyond the wavelength of light tends to decrease the PL enhancement [48]. For synthesized NPs and for smaller metal nanodots produced by annealing of continuous films of various thickness, the conversion efficiency can in principle be determined from modeling the scattering cross section (e.g. by 3-D FDTD), but systematic results directly applicable to the kinds of NPs used for LSP coupling in LEDs are very scarce in the literature (we present an example of such treatmet below in section VII). From general considerations (see e.g. Ref. [5]) the extinction cross section of the NP increases with increased NP diameter and, for NPs smaller than ~30 nm, absorption cross section is higher than the scattering cross section [58, 59]. Another factor is the density of NPs. Because the range of the LSP fringe electric field is very limited the active area of LSP enhancement covered by the NP fringe field is of the order of the NP in-plane dimensions. This puts serious requirements on the density of NPs necessary to achieve a high fill-factor in the LSP enhancement experiments. Fig. 18 shows that the PL enhancement ratio greatly increases with the increased areal density of synthesized Ag and Ag/SiO2 NPs with characteristic radius of Ag core about 30 nm and the SiO2 shell thickness of about 20 nm. (We present below in section VII an example of calculations for the effects of the NP dimension and spacing on overall enhancement efficiency.) All in all, it would seem that compact NPs with relatively large diameter are more favorable for efficient extraction of light from the LSP mode, but there exist other considerations that make the situation less  

straightforward. One of the considerations is the one regarding dissipation losses in the NPs. It has been noted that the efficient conversion of the LSP energy into the energy of photons can be achieved only if the dissipation losses in the LSP film are much lower than the non-radiative recombination losses in the QW [37, 38, 49, 60]. One of the ways to reduce the dissipation losses in continuous metal films (Ag, Au) is, as already mentioned, to introduce a thin (8-30 nm) dielectric layer between the metal and GaN [37, 38, 49, 60]. This is illustrated by Fig. 19 showing that in the presence of dielectric layer the PL enhancement is much stronger than without it. The effect is by no means negligible and is at least in part explained by the fact that a considerable portion of the plasmon energy is imparted not in the metal, but in the dielectric ajacent to it. This has been demonstrated by modeling the electric field distribution in Ref. [61]. Interestingly, in the core/shell Ag/SiO2 synthesized NPs the field distribution is qualitatively similar (see Fig. 6) and one expects considerable decrease of dissipation losses. An additional bonus here is that the Ag core in such NPs is single crystalline and lower losses can be anticipated in them than in polycrystalline metal films prepared by evaporation or sputtering. Mind also that it has been demonstrated in Ref. [38] that having the ohmic contacts to p- or nGaN directly contacting the LSP layer leads to a substantial increase in the LSP energy dissipation losses. In continuous metal films this means introducing additional fabrication steps to separate the LSP layer from ohmic contacts and current spreading metallization. It also necessarily limits the useful area of the LSP enhanced performance [37, 38, 49]. For core/shell Ag/SiO2 NPs with dielectric SiO2 shell one does not need to worry about that. Consider now the factors determining the relaxation time in the presence of SP or LSP and the Purcell factor describing quantitatively the efficiency of the QW light pumping into the SP

 

mode. The theory applicable to the SP case was developed in Ref. [30, 31]. It gives for the reciprocal decay time of the SP mode the expression 1/τSP~|pE(a)|2 ×k×(dk/dω),

(5)

where p is the dipole moment of the QW emitter, ω is the circular frequency, k is the SP wavevector, E(a) is the SP electric field strength at the depth corresponding to QW. As mentioned already, the electric field magnitude exponentially decreases with distance to the QW and characteristic length for the Ag SP on blue GaN/InGaN QW is about 40 nm. The spectral shape of the resonance is characterized by the sharp edge on the high energy side and a long tail on the low energy side where the Purcell factor is much lower than in maximum, but weakly dependent on photon energy (see Fig. 3). As demonstrated by Okamoto et al. [34] the SP resonance energy for thin Ag film is considerably blue shifted compared to the peak energy of typical blue GaN/InGaN QW LED. As a result, the Purcell factor is substantially lower than the peak value. With continuous films, the SP resonance peak position is relatively fixed. With Ag NPs the flexibility in energy adjustment is higher, and in principle a better fit to the QW emission peak can be achieved, for example, in core/shell Ag/SiO2 NPs by increasing the thickness of the SiO2 core (see Fig. 9). There are, however, natural limitations here related to the increased distance of the Ag core from the QW and corresponding decrease in the electric field strength. Generally, synthesized Ag NPs show higher Purcell factors near 2.6 eV QW energy than continuous thin Ag films. For example, Fig. 20 shows the decay times at the 2.6 eV peak QW line for the sample without the Ag NP layer and with the Ag NP layer (see Ref. [41]. The Purcell factor determined from the ratio of decay times was in that case equal to 7.2, i.e. considerably higher than for the offresonance Purcell factor of the continuous Ag layer in Ref. [34, 35] (Fsp=1.6). This increase in   

Purcell factor most likely is due to lower dissipation losses in the single crystalline NPs. Introduction of SiO2 shell in Ag/SiO2 NPs increases the distance from the Ag core to the QW and thus should decrease the Purcell factor according to Eq. (5). This is indeed observed as demonstrated in Fig. 21 (see also Ref. [42]). It should be noted that experimental PL decay curves often show two or more exponential decay regions, as in Fig. 20. Generally, the shorter decay times have a stronger dependence on coupling to the Ag or Ag/SiO2 NPs. Following the suggestions of Ref. [30, 31] it was the ratio of these short decay times that was taken to calculate the Purcell factors above. However, the starting values of the short decay times can vary significantly from sample to sample, apparently depending on the contribution of the non-radiative decay channel. For example, for QW samples studied in Ref. [42] the starting short decay time was 21.6 ns, whereas for samples studied in Ref. [41] the starting short decay time was only 1.4 ns, obviously owing to the stronger impact of the non-radiative recombination channel decreasing the starting IQE. The amount of increase in PL intensity produced by coupling with LSPs naturally depends on the starting IQE value and is reasonably accurately described by Eqs. (2)-(4). But for the same starting IQE the enhancement for the Ag/SiO2 NPs always follows the trend demonstrated in Fig. 22 (also see Ref. [42]). Despite the lower enhancement efficiency for the Ag/SiO2 core/shell NPs there exist other advantages that offset it, as will be discussed below. Thus, we see that the QW-LSP coupling efficiency does strongly decrease with increasing distance to the QW which presents a problem in practical use of the LSP enhancement effect in GaN/InGaN QW LEDs. However, as already mentioned, introducing a thin dielectric layer between the metal film and GaN can result in considerable enhancement of PL intensity even for large thicknesses of the GaN cap layer in the GaN/InGaN QWs. Their reasoning is illustrated by   

Fig. 23 taken from Ref. [61]. The figure shows the dispersion curves for Ag-on-SiO2 SPs on GaN/InGaN QWs and respective energy dissipation rates (the inset of the figure) as obtained by finite element modeling. It can be seen that, with the SiO2 thickness of about 30 nm, the dispersion curve approaches the light curve in GaN so that GaN acquires the waveguiding properties and supports the propagating wave in the Ag/SiO2/GaN/QW/GaN/sapphire resonator, the propagating surface plasmon polariton (SPP) [61]. Even though the extent of the SP fringe field remains limited such a structure can support a measurable enhancement of the QW PL intensity, but this effect is more in the realm of improving the light extraction efficiency in SPQW structures. These phenomena will be discussed in some detail later. In order that the LSP layer is effectively brought into close proximity to the QW region, in one approach, the p-type GaN contact layer is locally etched down and the Ag NPs are placed in the pockets of such partially etched nanostructure. The Ag NPs could be either pre-synthesized or could be produced by Ag evaporation and annealing. In Ref. [62] the latter version of the technique was used. An irregular (nanopillar-shaped when seen cross-sectionally) 150-nm-thick p-GaN contact layer on top of GaN/InGaN MQW structure was produced by partial dry etching through an etching mask formed by rapid thermal annealing of a thin Ni layer deposited on pGaN. Such annealing resulted in breaking down of the Ni film into dense metal islands with characteristic dimensions of ~100 nm and areal density of ~109 cm-2. Etching down the p-GaN contact layer through this nanomask almost down to the QW region produced the necessary pockets for positioning the Ag NPs. The latter were prepared by deposition of Ag films with various thickness and annealing them. It was noticed that the characteristic dimensions of the Ag NPs varied strongly depending on the starting Ag layer thickness. PL measurements performed on test QW structures with 10 nm undoped GaN cap showed a strong dependence of the LSP PL  

enhancement on the Ag NPs dimensions, with the optimal Ag NP size being close to ~100 nm. When such Ag NPs were prepared on the 150-nm-thick p-GaN contact layer partially etched down to about 10 nm from the QW region (see Fig. 24), a strong increase by 80% in PL intensity and by 58% in electroluminescence (EL) intensity was observed. The amount of PL enhancement was close to that observed in the structures with 10 nm GaN cap, and the Purcell factor for 2.68 eV peak QW luminescence (as determined from the measured PL decay rates) was 1.54, i.e. close to the typical values of the Purcell factor in Ag LSPs. As far as we are aware, this was one of the first successful demonstrations of the LSP enhancement in blue GaN/InGaN MQW structures with the p-GaN thickness close to the required thickness in real LEDs. The results can most probably be improved more if the pre-synthesized Ag or Ag/SiO2 NPs with a higher Purcell factors were used instead of Ag NP prepared by annealing.

IV.2. LSP effects in nanopillar blue LEDs The discussion above proves, among other things, the importance of bringing the LSP layer as close as possible to the active QW region. One of the ways to do so is by producing nanopillar (or nanorod) QW structures, as grown or fabricated e.g. by a deep dry etching of the LED structure down below the QW region, and putting the Ag NPs on the sidewall of the nanopillars thus bringing the LSP NPs in very close contact with the active QW region. This approach was realized in our work [43]. The nanopillar LED structures were prepared by dry etching of pGaN/MQW(GaN/InGaN) planar LEDs using the Ni etching mask. The typical nanopillar diameter was about 200 nm, the areal density was 5-10×108 cm-2, the height was 500 nm, thus exposing the entire MQW region to the LSP coupling at the nanopillar sidewalls. The sol-gel synthesized Ag NPs with diameter close to 30 nm and Ag/SiO2 NPs with Ag core diameter of 30 
 

nm and SiO2 shell thickness of 20 nm were deposited on the sidewalls of and between the nanopillars by drop casting, as illustrated by Fig. 25. It was observed that the deposition of Ag/SiO2 NPs led to 1.7 times increase of the PL intensity. However, the deposition of Ag NPs resulted in a decrease of the PL intensity (see Fig. 26), although the measurements of the decay times for the QW structure without and with Ag or Ag/SiO2 NPs (Fig. 27) indicated a stronger coupling of Ag NPs to the QW (faster decay time and higher Purcell factor) compared to the Ag/SiO2 core/shell NPs, as would be expected. The observed discrepancy was attributed to the strong LSP electric field caused by the very close proximity of Ag NPs to the QW. This is believed to have caused strong tunneling between the closely packed Ag NPs leading to high dissipative losses and decreased conversion efficiency of Ag LSP into external photons, even though the coupling to the LSP mode was quite strong. This was not an issue with core/shell NPs (and is not an issue for Ag NP placed in the pockets of p-GaN film in Fig. 24). It is interesting that the PL enhancement for the Ag/SiO2 NPs is so strong in Fig. 26. Considering that the diameter of the nanopillars is about 200 nm and the LSP field penetrates into the nanopillars by only about 20 nm as follows from Fig. 6 (d), only the outer ring of the QW region should be affected by the LSP coupling. One explanation is that the actual increase of the PL in this area is very strong due to a good coupling to LSP. The other possibility that we tend to favor is that the PL efficiency in nanopillars drops substantially for the regions close to the sidewall surface because of the defects introduced by dry etching (see e.g. Ref. [63, 64]). Coupling to the LSP enhances the PL efficiency in this near-surface defective region and thus contributes to the overall increase in the PL intensity. Naturally, the nanopillar LED structures can also be prepared by growth using MBE or MOCVD (see e.g. Ref. [65, 66]). The general effects of the LSP enhancement in such structures  

should be similar to the effects in the nanopillar structures produced by dry etching, but weaker PL enhancement factors are expected for the grown nanopillar structures because the starting PL intensity should be higher owing to the absence of dislocations in the grown nanopillars as opposed to those produced by dry etching [66]. However, the surface band bending in such structures and respective surface recombination velocity can still be strong and this recombination will be effectively suppressed by the LSP coupling.

IV.3. Stability of the LSP NPs One more factor to take into account, when considering the use of different types of LSP NPs in PL and EL enhancement experiments, is the stability of NPs with respect to exposure to ambient and device-operating conditions. One can predict agglomeration and oxidation of NPs at the surface along with interdiffusion between the NPs and the semiconductor. This is particularly important for the prevailing operation conditions of high-power LEDs that can typically lead to the surface temperature of the LEDs up to ~150oC [67]. In that sense the Ag/SiO2 core shell NPs have strong advantages over Ag NPs. Respective comparative measurements were performed by us in Ref. [41]. Ag and Ag/SiO2 NPs were deposited on the surface of blue GaN/InGaN MQW LEDs heated on a hot plate to 100oC, and the PL evolution with time was monitored for these structures. These conditions were supposed to model conditions during continuous operation of high-power LEDs. Fig. 28 shows the schematics of the experiment and the variations of the PL intensity with time observed for the two structures for time durations up to 400 minutes. It can be seen that, for the LED coated with Ag/SiO2 NPs, the PL intensity slightly (by about 10-15%) decreased after 60 min or so and then stabilized. In contrast, in Ag NP case the PL intensity continuously decreased with time and after 425 minutes it was only 45% of the initial value.  

SEM and TEM observations showed that, for Ag NPs, a strong agglomeration occurred and increased the NPs dimensions to almost 1 µm. This led to a change in the LSP resonance position and to increase in the energy dissipation in the LSPs. X-ray photoemission spectroscopy indicated a marked increase in the intensity of the lines attributed to the Ag oxide. In contrast, the Ag/SiO2 samples showed virtually no agglomeration and no change in the photoemission spectra. This higher stability was attributed to the presence of the chemically and thermally stable SiO2 shell effectively protecting the NPs from changing their composition and structure. This enhanced stability is a serious argument in favor of using Ag/SiO2 core/shell NPs instead of the Ag NPs, even though one has to sacrifice somewhat the coupling efficiency.

IV.4. Estimating LSP effects in high-quality blue LEDs So far we have been mostly talking about LSP effects in blue LEDs with relatively low IQE. However, real advanced high-power blue LEDs and white LEDs based on them have the IQE values exceeding 80%. Clearly, the expected increase in PL and EL efficiency produced by the LSP coupling in such structures cannot be as strong as for the structures with IQE on the order of 10-20%. After all, the IQE cannot become higher than 100%, although, for LSP structures at the surface, the energy imparted to the LSP mode and converted to photons directly goes into the external photons flux circumventing the total internal reflection limitations. The authors of Ref. [68] assessed the impact of SP coupling on EQE for a GaN/InGaN QW LED structure with the starting IQE of 80% and EQE of 33%. The authors used the model very similar to the one developed by Okamoto et al. [34]. They assumed quite realistic values of Purcell factors of 2 or 10, and obtained the SP enhanced EQE values of 38% for Purcell factor of 2 and 42% for Purcell factor of 10 (see Fig. 29). While these expected increases are not as high  

as the ones shown above for LED structures with inferior quality, they certainly are not negligible. But more importantly, the authors also analyzed the possible effects of SP coupling on the amount of EQE droop with increasing driving current. This droop, as mentioned above, is attributed to various reasons, such as carriers spilling out from the QW [67], non-radiative recombination defects in the QW [69, 70] and QCSE effect [71]. But the most prominent among them seems to be the Auger recombination [72, 73]. The authors of Ref. [68] measured experimentally the EQE dependence of their MQW structures on the driving current (see the reference sample curve in Fig. 29) and fitted it with the well-known phenomenological, so called ABC model describing the reciprocal lifetime dependence on carrier concentration (n) as the sum of the contributions from non-radiative recombination (A), radiative recombination (B×n), and Auger recombination (C×n2), where A, B, C are respective constants [74]. The A, B, C constants determined that way were then used to calculate the efficiency droop for the case of the SP coupling and the results for the two values of Purcell factors are presented in Fig. 29. It can be seen that SP coupling quite measurably suppresses the efficiency droop. The droop suppression becomes stronger with increasing the Purcell factor. The authors also estimated the possible contribution of the SP coupling to the EQE increase for various efficiencies of the SP energy conversion into the external photons. As already mentioned above, for SP layers at the surface of the QW structure, the energy imparted to the SP mode and converted to photons provides a direct gain in the light extraction efficiency. The authors introduced the dimensionless parameter r=Cextph/LEE (the notation is the same as in Eqs. (1)-(4)) to characterize this effect. Fig. 30 displays the results of their modeling with two different values of r=0.7 and r=1.3 and the Purcell factor of 2. This indicates that, by using efficient SP structures with a high SP energy conversion into light, both the efficiency at low driving current can be improved and the efficiency droop  

can be decreased. These results are of great importance and show the high potential of SP or LSP effects for improving performance of even very high-quality blue LED structures, particularly if stable LSP NPs providing high Purcell factors and good LSP energy conversion to light, such as the synthesized Ag/SiO2 core/shell NPs, are used.

IV.5. Methods to incorporate LSP enhancement in practical LED structures Now let us turn to discussing practical means of introducing the LSP enhancement into real blue LED devices. The key issues here are: (a) designing the ways to bring the metal film as close as possible to the QW region without compromising the LED performance, (b) decreasing dissipation losses and increasing the conversion efficiency of LSP energy into light, (c) providing for stable performance enhancement, (d) doing all that in ways that are cost-efficient and compatible with device fabrication routines adopted by the industry. With regards to item (a) it is useful to consider how close is close enough. We have seen that for continuous Ag films the characteristic penetration depth of the SP fringe film into GaN is close to 40 nm, for Ag films on SiO2 or for core/shell Ag/SiO2 films this penetration depth is closer to 20 nm. Beyond these distances between the SP or LSP film and the QW, the electric field decreases rapidly and the coupling loses efficiency. Standard GaN/InGaN QW LEDs consist of several microns thick n-type GaN film, one or several (usually 3-5) QWs with the GaN barrier thickness of around 10 nm and the InGaN QW thicknes of about 2-3 nm, the electron blocking p-AlGaN layer of about 20 nm in thickness, and the contact p-GaN film. The latter has to be thick enough not to be depleted by the built-in p-n junction voltage and thick enough to allow good current spreading. So the lower limit of the p-GaN film thickness acceptable is given by the depletion width at 0V. In Ref. [36], the depletion width of the top p-GaN film at 0V was  

estimated to be 76 nm. This value was arrived at by taking the hole concentration in p-GaN as close to 1017 cm-3. It would seem, however, that this estimate is overly conservative. The depletion width in p-GaN is, in fact, determined by the density of uncompensated Mg acceptors, rather than the hole density [75-77]. The former is typically about 1019 cm-3, which would give an order of magnitude lower depletion width. And part of this depletion width will be, in fact, located in the p-AlGaN electron blocking layer. So the question is how thin one can get the pGaN cap without increasing too strongly the series resistance and without running into problems with devices AC performance (the latter is due to the dual nature of Mg that is a shallow p-type dopant, but also a rather deep center [75-77]). In blue LEDs that are currently in use the p-GaN film thickness is about 0.15 µm or higher, but perhaps, for the LSP enhancement applications, the question has to be revisited. As for item (b) above, it would seem that having SP films on SiO2, with sub-wavelength grating imposed or using Ag/SiO2 core/shell NPs should produce the best results for LSP energy conversion to light. Having the Ag NPs covered by SiO2 should, as discussed above, have beneficial effects on stability of performance and should also be conducive to improving the crystalline quality and the electrical performance in structures that use overgrowth (relevant to item (c) above). And we’ll discuss item (d) after having surveyed the actual results of LSP LEDs device work. In designing practical LSP enhanced LEDs several approaches have been tried with reasonable success. The first and the most straightforward approach uses the SPs produced by deposition of continuous Ag films on top of the ready LED structure, a) with a thin SiO2 or SiN film beneath the Ag layer to decrease the dissipation losses, b) with electrical insulation of the SP film from the ohmic contact and the current spreading metallization, and c) with the sub 

wavelength period grating imposed on the SP film. The thickness of the p-GaN contact film used in this approach is taken to be below the penetration depth of the SP plasmon electric field into GaN. Typically, the thickness is taken as 20-70 nm. This version of SP enhanced LED has been realized in Ref. [37, 38, 49]. (It could be noted that useful enhancement owing to the waveguiding action in the Ag/SiO2/p-GaN/QW/n-GaN/sapphire structure forming the deeply penetrating SPP state (see Fig. 23) has also been demonstrated for thick p-GaN contact layers [61].) Fig. 31 displays a typical LED design, and Fig. 32 shows the typical ohmic contacts pads and the SP grating design. A considerable EL efficiency increase of 25-50% was demonstrated for such LEDs [37, 38, 49, 60] and the efficiency droop was alleviated at the same time [78]. However, the thickness and design of p-GaN cap layer in this technique has to be further optimized to minimize the series resistance. For example, p-AlGaN/GaN modulation doped superlattices with high in-plane conductivity could be used to cut down the series resistance (see e.g. Ref. [79]). It is also necessary to study the stability of the SP enhancement under the operation conditions. Ag oxidation could be an issue. Using Ag/SiO2 NPs instead of continuous Ag film could be quite advantageous both in terms of enhanced stability and improved overall LSP efficiency and effective LSP energy conversion into photons. One of the clear advantages of the described approach is that the SP structure is on top and thus directly contributes to the enhancement of light extraction efficiency. The second approach involves using partial etching off of the p-GaN layer and placing Ag or Ag/SiO2 NPs in the pockets of this profiled structure as shown schematically in Fig. 24 [12]. Typically the nanoholes in the p-GaN are prepared by a local dry etching through various etching masks. The partially etched nanoholes thus formed are filled with Ag or Ag/SiO2 NPs and the whole p-GaN top surface is deposited with the indium tin oxide (ITO) contact layer [80]. With   

the LSP-active in-plane cross section of Ag/SiO2 NPs close to 60-100 nm (for Ag/SiO2, 40 nm Ag and 20 nm SiO2; for Ag, from 40 nm for synthesized NPs to 100 nm for annealed nanodots) and the areal density of NPs close to 1010 cm-2, the total area effectively covered by LSP enhancement can amount to about 30-80% which is a fairly respectable result. (The quoted areal density of NPs is quite realistic as seen in Fig. 18.) The EL output increase attained in this method can be as high as 1.6 times, but the turn-on voltage of the LEDs tends to somewhat increase due to a decreased effective thickness of p-GaN causing enhanced series resistance (in Ref. [62] it increased by 0.5 V). Using Ag/SiO2 NPs instead of Ag NPs in this technique should be advantageous because of the good electrical insulation of the NPs from the ITO contact, even despite some loss in coupling efficiency. One should also consider the improved stability of Ag/SiO2 NPs. In principle, this method should additionally provide a certain improvement of the light extraction efficiency, because the LSP layer is on top of GaN, which circumvents the problem of the total internal reflection and extracting light from GaN. But the light should still be extracted via ITO, so the benefits on that count are limited. The method described is quite promising, but technologically rather challenging, mostly because of the difficulty in getting the distance between the bottom of the hole and the QW uniformly and reproducibly low. The third approach described in Ref. [36, 81] involves building-in LSP NPs on n-GaN beneath the QW layer followed by overgrowing the QW active film and top p-GaN contact layer. Two versions of this approach have been described. In Ref. [36] Ag NPs prepared by annealing of Ag film were deposited on n-GaN contact layer and overgrown by MOCVD to form the MQW layer and the rest of the LED structure (see schematic representation in Fig. 33). The inplane NP dimensions were close to 450 nm, the areal density close to 6.4×107 cm-2, so the total effective area coverage for the Ag NPs was below 10%. The authors observed the room   

temperature IQE to increase from 29.7% to 45.7%, the EL output at 100 mA to increase by 1.3 times as the result of the Ag NP incorporation. The Purcell factor was estimated to be equal to 1.75. The observed EL and IQE increase and the determined Purcell factor seem quite reasonable given the low effective area of the LSP enhancement. The authors claim that the crystalline quality of the LSP LEDs was not inferior to the crystalline quality of the reference LEDs, but the turn-on voltage was slightly increased (from 3.85 V to 3.9 V). Another version of the same approach is described in Ref. [81]. In this version the growth of n-GaN contact film by MOCVD was interrupted, the sample was taken out and subjected to 30 minutes photoelectrochemical etching in 0.06M aqueous solution of KOH at anodic voltage of 2V under Xe lamp illumination. As a result the top portion of the layer was etched down by about 300 nm everywhere except in the region around dislocations, forming GaN nanoneedles of 10-30 nm in diameter [82]. A 50-nm-thick Ag layer was deposited on that surface and annealed at 750oC to form a dense layer of Ag NPs with diameter around 100 nm. The nanoneedle layer with Ag NPs embedded in it was then put again into the MOCVD reactor and overgrown by 50 nm of n-GaN at somewhat reduced (compared to standard) temperature to suppress strong evaporation of Ag from the NPs. After that the MQW structure and the p-GaN cap layer were grown at standard growth temperatures (see the schematic representation of the process in Fig. 34). As a result the dislocation density in the structure was reduced from the starting 1010 cm-2 to 109 cm-2, presumably owing to the epitaxial lateral overgrowth from the nanoneedles. The luminescence efficiency increased by 2.6 times as a result of incorporation of the Ag NPs in close proximity to the QW layer. It should be noted that, although both versions of the method produce useful results, the method is technologically extremely challenging because the most crucial part of the structure, the QW region, is to be grown on n-GaN film with Ag NPs  

embedded in it and the thickness to merge the NPs is very limited. This requires very careful optimization. In the fourth approach the Ag NPs are built into the p-GaN very close above the QW by depositing a thin Ag layer and annealing it. The schematic representation of the structure is shown in Fig. 35. Two versions of the process have been described. In the first the overgrowth is performed directly over the small Ag NPs produced by annealing the Ag layer [36]. In the second additional SiO2 micro-discs are placed to cover the NPs from above [45]. For both versions considerable enhancement of the IQE was oserved. For example, the IQE from 6% increased to 13% with Ag NPs for the areal density 8×108 cm-2 and to 22% for the density of 4×109 cm-2. The turn-on voltage of the devices was virtually not affected by the introduction of Ag NPs, but the series resistance measurably increased and so did the reverse current (see Fig. 36). The authors associated the observed increase in leakage with the enhanced density of nanotubes in the film that could be related to the difficulty in overgrowing the Ag NPs. But when looking at Fig. 10 one wonders if Ag-related defects in the immediate vicinity of the QWs could not also contribute. However, for Ag NPs covered from above with SiO2 discs (Fig. 11) the excessive leakage problem was greatly alleviated (Fig. 36), although the increased series resistance was still a problem, as it should be, considering a serious disturbance of the current frow by the NPs with the discs (see Fig. 11). The alleviation of the excessive leakage problem with the introduction of SiO2 discs was associated with the decreased density of nanotubes, which is most likely related to more favorable conditions for lateral overgrowth over the SiO2 discs compared to trying to directly overgrow the Ag NPs. Whatever the explanation, the increase in the IQE for Ag NPs with SiO2 microdiscs was again very considerable: the IQE increased from 12% to 38% and the EL light output at 20 mA increased by 72%. The authors 
 

also pointed out that the SiO2 microdiscs, in addition to somewhat suppressing the excessive leakage, were also contributing to the increased light extraction efficiency due to strong light scattering. As in other techniques, it would seem that using Ag/SiO2 core/shell NPs instead of Ag NPs or Ag NPs covered by SiO2 microdiscs should bring further improvements in crystalline quality via facilitating more efficient lateral overgrowth. They could also prevent QW contamination with Ag related defects, and further increase light scattering. The techniques described in Ref. [36, 46] look quite promising, and we will see later that the approach is reasonably universal and applicable to green LEDs with Au NPs [45] and UV LEDs with Pd and Pt NPs [53, 54]. But the method, as other techniques already described above, is very technologically challenging because it involves careful optimization in order to overgrow the NPs without causing serious degradation of crystalline quality. It should be noted that, for nanopillar QW LEDs introduction of the LSP enhancement is quite straightforward and could seriously contribute to alleviating problems with band bending and surface recombination at the sidewalls without causing any additional difficulties in device fabrication, particularly if the core/shell Ag/SiO2 NPs with highly insulating SiO2 shells are used. For the nanopillar LEDs with the p-n junction plane parallel to the growth plane (so called axial QW) the effect will be limited because of the limited portion of the nanopillar cross section that is affected by the LSP interaction. However, recently a new class of nanopillar LEDs has come into being, the so-called core/shell QW LEDs (see e.g. Ref. [83, 84]). In such LEDs the p-n junction is formed along the nanopillar growth axis. In that case the effect of the LSP enhancement could be quite considerable. The question about the number of QWs optimal for maximizing the LSP effect also needs to be briefly discussed. As pointed out earlier, it is quite common to use several QWs (usually 3-5)  

to obtain the highest performance. But it is intuitively clear and has been demonstrated experimentally [40] and by theoretical modeling [85] that the QW most efficiently coupled to the LSP is the one nearest to the NPs layer. Therefore, the best effect is expected for single QW structures. To conclude this discussion: we have seen that so far no universal ready-to-go method of incorporating the LSP enhancement technique into the fabrication process of high-power blue LEDs has been developed. The easiest for implementation seems to be the approaches described in Ref. [37, 38, 49, 60, 61, 62], involving only additional post-growth procedures such as deposition of Ag layers with underlying SiO2 films and grating-patterning of the SP structures or partial thinning of p-GaN and deposition of Ag NPs in the nanoindentations thus produced. In the first approach the main stumbling block is getting the p-GaN cap that is thin enough, but doing so without increasing the series resistance. In the second approach the problem of consistency and reproducibility in recess etching of p-GaN to the exactly required depth and doing it in order not to bring along the unacceptable increase in series resistance is critically important. All techniques involving overgrowth of Ag NPs placed near the QW region are technologically very challenging. It seems that techniques in which the LSP NPs are embedded in the p-GaN cap and overgrown [36, 46] should be considerably easier in implementation than the techniques relying on overgrowth of the QW active region over the n-GaN layer with NPs embedded in it. All techniques can be realized using the standard procedures to be found in the standard toolbox of a laboratory fabricating LEDs: overgrowth by MOCVD, dry etching, deposition of metal and dielectric layers, annealing. At the same time, all the methods to fabricate LSP enhanced LEDs can benefit substantially from using synthesized Ag or core/shell  

Ag/SiO2 NPs of specified dimensions and composition. The technology necessary here is not to be found in a typical LED-fabrication outfit. For university groups or company R&D divisions it is conceivable to develop such a technology in-house by hiring a group of qualified chemical engineers or by setting up collaborations with outside research groups. The fabrication of blue LEDs and white LEDs based on them is a multi-billion-dollars industry nowadays. In order to fit in a new technology into it this technology should meet industrial standards of quality control, reliability, and reproducibility, which mean that the synthesized NPs to be used in the process have to be commercially available with required properties and at reasonable cost. The ground work for that has already been done: the materials science of NPs is well developed thanks to multiple plasmonic applications in other fields, there do exist companies that are fabricating NPs on commercial basis. After all, phosphores used in white LEDs are a routine part of the white LED fabrication process. It should also be noted that developing predictive theoretical models of LSP LEDs is also vital in really getting the field going in practical applications.

V. LSP effects in green LEDs Green LEDs are prepared in the GaN/InGaN materials system by increasing the In concentration in the QWs. Because of limited In solubility in InGaN one runs into problems with the formation of QD-like In composition fluctuations, resulting in phase segregation into In-rich and In-poor region, for GaN/InGaN QW structures. In blue GaN/InGaN the typical dimensions of the In-rich phase are well below a micron and much shorter than the diffusion length of charge carriers. Hence, the recombination occurs preferentially in the In-rich phase and the contribution of non-radiative recombination centers within these regions is very limited. The non-radiative

 

centers only contribute to decreasing the IQE via hampering the carriers transport along the QWs to the In-rich phases. In contrast, for even more In-rich QWs in green, yellow, and red LEDs the dimensions of the In-rich phases strongly increase. For green LEDs they are typically over a micron and thus much larger than the diffusion length of charge carriers. Therefore, the radiative recombination in these In-rich fluctuation regions should be reduced. On the other hand, higher In composition increases the dislocation density due to a higher strain caused by the mismatch in lattice parameters. At the same time, a higher density of point defects is created within the QWs upon InGaN partial decomposition and upon movement of misfit dislocations. Both these factors greatly increase the density of non-radiative recombination centers and decrease the IQE of the green (and longer wavelength) LEDs (see e.g. the excellent discussions in Ref. [86]). It should also be kept in mind that additional strain in green LEDs causes a great increase in the strength of piezoelectric polarization field and QCSE thereof, leading to the spatial separation of electron and hole wave functions in the QW and a strong decrease of the IQE [87]. As a result the IQE of the green LEDs is typically about an order of magnitude lower than for blue LEDs. Because the SP or LSP couplings are quite efficient in suppressing the detrimental effects of non-radiative recombination centers on IQE of LEDs, this coupling should have a much stronger impact on the performance of green LEDs compared to the blue LEDs. The amount of the expected effects can be estimated by modeling. As discussed in section IV.4, the authors of Ref. [68] have done calculations of the SP enhancement efficiency for green LED structures using the same approach as the one employed to generate Figs. 29, 30. The result of their estimates for green LEDs is illustrated by Fig. 37. Comparing the data in Fig. 37 with the results shown in Figs. 29, 30 one can see that a much stronger improvement is expected for green LEDs, both in  

terms of the EQE at low driving current and in terms of expected efficiency droop. These predictions are on the same order as the results observed for low-efficiency blue LEDs with the high contribution of the non-radiative recombination channel. The effect at high driving currents can be still further enhanced by taking into account the polarization field effects. As already pointed out, the strong QCSE in green LEDs has a profound negative effect on their performance [87]. Because the strength of the built-in piezoelectric field can be screened by the carriers injected into the QW, the increase in the injection level leads to a measurable enhancement of the IQE and a considerable blue shift of emission peak position. If Ag is used to produce the surface plasmon resonance in green LEDs the Purcell factor is on the low value slope in its spectral dependence (see Fig. 3). Thus, by increasing the driving current in EL experiments or by pumping the laser power in PL experiments one not only increases the EL or PL intensity, but also blue-shifts the emission peak position closer to the true Ag SP resonance, which causes additional efficiency enhancement as illustrated by Fig. 38. The data in Ref. [88] were obtained on blue LEDs, but the effect for green LEDs with Au SPs should be even stronger, although this has not as yet been proven experimentally. It is also worth noting that in Ref. [89] the authors considered the impact of the image charge electric field of an Au layer on top of a GaN/InGaN QW on the potential profile in the QW. It was discovered that the piezoelectric field and related QCSE could be strongly affected by the image charge force, even when it is orders of magnitude weaker than the piezoelectric field. Both effects studied in Ref. [88, 89] certainly warrant further investigation. One more issue related to the impact of piezoelectric field and QCSE is the use of non-polar GaN/InGaN QW structures in which the polarization field is in the QW plane and therefore does not cause the separation of the electron-hole wave function and the respective decrease in the  

IQE (see e.g. Ref. [87, 90]). This should have a profound effect on the efficiency of GaN/InGaN QW LEDs. Unfortunately, growth in non-polar directions causes serious problems with structural performance of GaN and related compounds because such growth produces a high density of staking faults with associated partial dislocations and accompanying point defects [9096]. The only ways to radically decrease the density of such defects are the growth on highquality non-polar GaN substrates cut from high-quality polar GaN boules [97] or the careful optimization of the epitaxial lateral overgrowth techniques [92-94]. Both approaches are timeconsuming and are costly. In standard epitaxial growth the non-polar GaN films or GaN/InGaN QWs, the density of deep non-radiative recombination traps is quite high. In our work [49] we studied electrical and optical properties and effects of LSP coupling to Ag NPs for nonpolar a-plane (11-20) GaN/InGaN QW structures grown by MOCVD on r-plane (1-102) sapphire substrate. This study was carried out for the purpose of proof-of-concept demonstration of the QW-LSP coupling in which the GaN capping layer on the QW was only 10 nm thick. Admittance spectra and deep level transient spectroscopy (DLTS) measurements on these structures revealed the presence of a very high density of the 0.45 eV electron traps and 1 eV electron traps in or near the QW region (see Fig. 39 (a), (b)). Both traps were quite active as non-radiative recombination centers, leading to low QW PL intensity (see Fig. 40). Deposition of the Ag NPs intensified the PL emission by 1.5 times and shifted the QW peak from 490 nm to 481 nm, the latter due to the coupling to the Ag LSP with the peak near 450 nm. The QW peak in the figure is at 490 nm, i.e. cyan rather than true green at 520-550 nm. However, the same phenomena can be simply foreseen for green nonpolar LED structures with Au LSPs, although the shorter wavelength shift of the peak is not anticipated. Rather, the peak should move deeper into the green owing to the spectral position of  

the Au LSP resonance. Note that the magnitude of the observed change in Fig. 40 is not far away from that predicted by modeling in Fig. 38. Consider now the results of the practical applications of LSP coupling to enhancing the performance of green GaN/InGaN QW LEDs. As with the blue LEDs discussed above in section IV.5 the methods used can be divided into LSP enhancement experiments in which the LSP layer is deposited on top of the green QW LED structure with suitably thin p-GaN top layer and methods in which the LSP NPs are embedded into p-GaN cap layer and overgrown. The first types of experiments were described in Ref. [47, 48, 61, 78]. In Ref. [48] Ag SP layer was deposited on SQW GaN/InGaN LED structure with PL emission at 550 nm and 70nm-thick p-GaN cap layer. Owing to the presence of a long tail in the Purcell factor of Ag that extends into the green region, 150% increase in EL intensity was observed. The authors of Ref. [78] used a 515 nm green LED structure with 60-nm-thick p-GaN cap layer and studied the effects of the geometry of the Ni/Au current spreading layer and that of the grating imposed in the Ag SP layer on the EQE and efficiency droop (see the schematics of the gratings in Fig. 32). It was noticed that the SP coupling increased the EQE value by about 1.5 times, and that the peak current density of the EQE-current curve could be very substantially increased from 50 to 75 A/cm2 when decreasing the period of the current spreading grating and the Ag film grating. In Ref. [61] for the 525 nm green QW LED structure with 15-nm-thick GaN cap layer, considerable (about 3 times) PL enhancement was achieved by introducing a thin (8-20 nm) SiO2 layer underneath the Ag film. Some increase in PL could also be observed even for a much thicker GaN cap layer (100 nm) owing to the waveguiding properties of the SP structure with the 30-nmthick SiO2 film. The PL enhancement of a 535 nm green SQW with 10 nm GaN cap layer was studied [47]. A 50-nm-thick Au SP film was deposited on top of the GaN cap without a grating   

or with a grating having a certain period (see Fig. 41). Without the grating prepared in the Au film and with the grating period larger than the wavelength of light, the PL enhancement was close to 2, but for short grating period it could be increased to 7, which graphically demonstrates the importance of the optimized grating in extracting light from the SP-coupled structure with continuous metal film. In the second approach, the Au NPs were formed in a thin p-GaN layer adjacent to the MQW region [46] as it was done with Ag NPs (see Fig. 10 and Fig. 33). The growth of p-GaN film on the 525 nm MQW was interrupted after deposition of 20 nm film. The sample was taken out of the growth chamber, 0.2-nm-thick Au film was sputtered on its top, and annealed for 3 minutes at 800oC to form Au NPs with the in-plane dimensions of 80±20 nm, height of 8±5 nm and areal density of ~3×109 cm-2. The sample was then reloaded to the MOCVD reactor, and 150-nm-thick p-GaN layer was overgrown to complete the SP LED structure. The PL enhancement ratio observed was around 30 near the peak value (Fig. 42). The density of nanopits/nanotubes (believed to be responsible for the excessive leakage current [36,42]) in the structure with Au NPs was virtually the same as in the structure without Au NPs (~3×108 cm-2), most likely due to the thermal and chemical stability of Au as opposed to Ag. Note that in Ag case the density of nanopits was about an order of magnitude higher than for the reference structure [36]. Correspondingly, the increase in reverse current caused by embedding Au into the p-GaN was very moderate and much lower than for embedded Ag NPs (see Fig. 43). Still, some increase was observed and it could be attributed to Au-related defect centers introduced into the QW region. Recall that a similar reverse leakage decrease linked to the decrease of nanopits density was observed after embedding of the Ag NPs covered by SiO2 nanodisks [45]; here again, the residual increase in reverse leakage should be ascribed to the Ag-related deep traps   

introduced in the QW. In this sense, the use of Au/SiO2 core/shell NPs instead of Au NPs should be quite conducive in decreasing the leakage current.

VI. UV LEDs Some of the problems encountered in UV LEDs are similar to the ones we described above for green LEDs. These are the low internal quantum efficiency due to the high density of nonradiative recombination centers and the negative consequences of the strong polarization fields and QCSE related to them (see e.g. Ref. [17, 18, 98-100]). Although much has been done in recent years to alleviate these problems [19,101-103], there is certainly room for improvement, and we will show in what follows how these improvements can be brought about by LSP coupling. Low LEE is perhaps the most important of the problems specific to UV LEDs. The LEE of deep UV LEDs is limited by strong light absorption in the p-GaN contact layer and total internal reflection, and by the unique anisotropic optical polarization property in AlGaN QWs with high Al composition. LSP resonance is capable of alleviating the problem via introducing additional scattering and waveguiding actions along the LSP pattern that both help to extract the light trapped in GaN. In Ref. [104], Au NPs were formed on top of the fully assembled blue LED structure with a thick ITO p-type contact layer. The increase in EL efficiency of up to 1.5 times was observed and was convincingly demonstrated to be due to the improved light extraction efficiency. However, for UV LEDs that employ AlxGa1-xN/Al yGa 1-yN QWs rather than GaN/InGaN QW, a specific problem arises. In GaN/InGaN QWs the optical transitions in the QW take place between the conduction band and the heavy hole/light hole (HH/LH) band. This light is polarized normal to the [0001] c-axis (TE mode) and propagates along the c-axis. For that the light extraction  

efficiency is determined simply by the total internal reflection angle [105]. In AlGaN/AlGaN QWs, with the increase in the Al mole fraction XAl>0.25, the recombination between the conduction band and the crystal field split hole band (CH) becomes more and more important. At XAl=0.66 the cross-over occurs between the HH and the CH bands and the recombination involving the latter band becomes predominant. This light wave is polarized parallel to the c-axis (TM mode) (see e.g. Ref. [103, 105-107]) and is virtually impossible to extract via the (0001) planes. As a result, the LEE of UV LEDs constantly decreases with increasing the Al mole fraction (decreasing the emission wavelength), and for the wavelength region of 280-320 nm (UV-B) the EQE is only about 2% [106, 108]. Several routes have been tried in that respect with varying degrees of success [109, 110]. But it would seem that the QW coupling to LSP or SP layer at the surface offers one of the best solutions. The authors in Ref. [52] note that the TM light mode propagating in the direction normal to the c-axis and trapped in p-GaN can interact efficiently with the SP formed by the deposition of a thin metal film (SP-TM wave coupling). As in Fig. 44 illustrating the principle, if the peak energies of the two modes are well tuned the TM light trapped in p-GaN can be efficiently converted into the SP mode and extracted in the form of external photons. For Al SP film this effect becomes stronger when the Al content xAl in the QW is higher because the TM mode contribution in the QW increases with increasing xAl and the energy matching between the SP resonance and the TM wave improves with increasing xAl. It was observed that the relative LEE enhancement increased from 1 for the wavelength of 365 nm, to 1.23 for 311 nm, 1.59 for 294 nm, and 2.36 times for the wavelength of 280 nm [52]. A 217% enhancement in peak PL intensity at 294 nm was observed as well. It is noteworthy that in their experiment the enhanced


 

LEE is the sole source of the PL yield increase because the p-GaN (separating Al SP and QW) was 90nm thick so the IQE increase was not likely. The other important cause of low LEE in AlGaN UV LEDs is the strong light absorption in p-GaN cap layer. The efficiency of p-type doping in p-AlGaN is known to decrease dramatically when moving from GaN to AlN. This happens because with increasing xAl the level depth of Mg acceptors in p-AlGaN increases strongly and the compensation ratio of p-AlGaN by donor complexes involving native defects and Mg rapidly increases [13]. For instance, the depth of Mg acceptors in GaN is close to 0.16 eV, but it starts to gradually increase for xAl>0.1 and reaches the value of 0.5 eV for AlN; the hole concentration that can be achieved in p-AlGaN decreases accordingly. Hence, for Al-rich DUV LEDs it is not practical to use p-AlGaN as the hole injecting layer. Instead, the common practice is to have a heavily Mg doped p-GaN contact layer on a short-period p-AlGaN/uid-GaN superlattice (see e.g. Ref. [1-4]). This provides good current spreading due to selective doping effect in uid-GaN by holes coming from acceptors in pAlGaN. It accommodates a considerable lattice mismatch between GaN and AlGaN and makes possible efficient vertical tunneling and acceptable hole injection efficiency into the AlGaN QW [1-4]. However, the light generated in the AlGaN QW is strongly absorbed in the top p-GaN layer, thus reducing the LEE. One of the ways to alleviate the problem is to use nanopillar structures in which the light can escape through the sidewalls [111]. It should be noted here that the LSP arrangements of the type illustrated by Fig. 24, 25, i.e. with locally thinned p-GaN top layer or nanopillar structure with etching extending down to n-AlGaN below the QW are ideally suitable for increasing the LEE of DUV AlGaN LEDs, both by capitalizing on the formation of the nanopillar structure per se, as in Ref. [111], and on IQE and LEE enhancement owing to the LSP coupling.  

Let us now turn to the results of more conventional studies of LSP-mediated enhancement of UV LED performance based on the IQE improvement by suppressing the non-radiative recombination channel in the active LED region. All major contributions in this field were using optimized LSP introduction techniques in which the metal layer is placed as close as possible to the QW region. In Ref. [56] the authors explored the Al SP coupling with a 346 nm UV LED structure grown by MOCVD on Si(111) substrate. The schematic of the structure is represented in Fig. 45. The LSP coupling was provided by Al microdisks incorporated into the pattern with 5 µm in diameter and 10 µm in pitch that were produced by photolithography and dry etching (see also the etch pattern in Fig. 12). The Al NPs penetrated through p-GaN and partly through the pAlGaN (XAl=0.15) cap contact layers almost down to the p-AlGaN (XAl=0.3) electron blocking layer. From comparing the PL intensity measured at 10K and 300K on the structures without and with Al microdisks the authors calculated the IQE to increase by 45% as a result of LSP coupling. They observed a very strong dependence of the enhancement efficiency on the distance between the Al layer and the MQW region, which suggests the leading role of the SP coupling in the observed effect. Mind that the areal fraction taken by the Al microdisks in Ref. [56] was relatively low (0.25), therefore, the enhancement ratio can be substantially increased by using a more dense etching pattern. In Ref. [54], the authors incorporated Ag or Pt LSP into ~400 nm emitting near UV LEDs using the same scheme as already described for blue (Fig. 10 and Fig. 35) and green LEDs. Namely, the LSP particles were incorporated into the top p-GaN film very close to the QW. The LSPs were produced by deposition of Ag or Pt films on top of 20-nm-thick p-GaN and thermal annealing. The metal NPs were then overgrown by p-GaN to form the complete LED structure. Fig. 46 shows the schematic representation of the LED design, the cross section of the LED  

structure, and the plan-view of the Ag and Pt NPs produced by the annealing. The characteristic dimensions and densities were 35-75 nm and 2×109 cm-2 for Ag NPs, 10-40 nm and 4×1010 cm-2 for Pt NPs, respectively. The authors observed the EL intensity enhancement at 20 mA of 20.1% for Ag NPs and 57.9% for Pt NPs. The Purcell factors of the LSPs coupled with the QW could be estimated from the observed changes in PL decay times: 4.26 ns without NPs, 3.77 ns with Ag NPs (Fsp=2.1), 3.68 ns for Pt NPs (Fsp=2.2). The IQE increase caused by the introduction of LSP NPs and determined from the ratios of the PL peak intensities at 300K and 10K was 24.9% in the case of Ag and 44.7% in the case of Pt. The higher enhancements in the case of Pt NPs are due to the higher NP density and the better energy match between the LSP absorbance and the QW peak (see Fig. 47). Meanwhile, one can envisage using Pd NPs, as described in Ref. [53]. For Pd nanocubes synthesized by sol-gel method, the LSP resonance peak can be shifted between 320 nm (for cubes with the size of 25 nm) to 380 nm (for the cubes with size of 50 nm) depending on NP size. It was observed that the nanocubes of 25 and 50 nm in size oxidized in air to form Pd/PdO core−shell structures, but the 8-nm nanocubes were stable in air. This means that Pd NPs, if carefully prepared, are quite suitable for use of LSP-mediated enhancement in UV LEDs LSP resonance peaked near 280 nm that is very interesting for use in DUV LEDs was observed for sol-gel synthesized In/SiO2 core/shell NPs with In cores of ~80 nm in diameter and SiO2 shells of 12 nm in thickness [112]. Such NPs would probably be of interest only for designs in which they are deposited at the very last stage of processing on top of the UV LED structures due to thermal stability concern. For schemes involving overgrowth of NPs the In-based core/shell NPs will be most likely unstable because of the low melting point of In.

 

VII. Theoretical considerations regarding the impact of NP diameter, density, and distribution on the luminescence wavelength and enhancement Finally, it is instructive to understand how changing the NP diameter, density and distribution will affect the enhancement of light emitted by the QW active region of LEDs. Individual NPs can only enhance the radiation emitted by a small volume of the QW located under the NP. Thus, to usefully employ the LSP related enhancement one has to create an ensemble of relatively closely packed NPs in close proximity to the QW region. Intuitively, it would appear that for too sparsely placed NPs the effect will be much weaker than the enhancement expected for a single NP. For high density of NPs the luminescence enhancement should be approaching the efficiency for a single NP. A quantitative treatment of the problem was developed by Khurgin et al. in Ref. [113]. Using the effective mode volume concept the authors showed that, for a single spherical Ag NP, the luminescence enhancement that can be achieved shows a maximum for a certain value of the NP radius a and that the overall enhancement depends on the QW luminescence efficiency without the NP, ηrad: the higher the starting efficiency the lower the enhancement ratio. When substituting the single NP with the NP array characterised by the interparticle distance R one still gets the enhancement Farray that is very much more pronounced for the lower starting efficiency (see Fig. 48 taken from Ref. [113]), the effect still is characterized by a maximum as a function of the particle radius a and can be maximized by choosing the optimal interparticles distance Ropt different for different sphere radii (the dependence is also shown in the figure; it can be seen that, for optimal radius around 15 nm, the optimal spacing between the spheres should be around 60 nm). The data in Fig. 48 are shown for the distance d between the NP layer and the QW equal to 10 nm. The dependence of the optical enhancement Fopt on the starting radiation efficiency for three different d values is shown  

in Fig. 49. As expected, the degree of enhancement increases with decreasing the distance between the QW and the NPs, decreases for increasing the starting luminescence efficiency, and tends to 1 for the starting efficiency approaching 1. The general conclusion of these calculations is that, for the NP array, the enhancement of luminescence efficiency is several times lower than the one calculated for single particles, that the efficiency improvement becomes more pronounced for the lower starting efficiency, and that the interparticle distance should be on the order of 60-100 nm. These calculations clearly show that, for blue LEDs with IQE~100%, the effect of LSP enhancement will be only marginal, as expected based on the discussion in the previous sections. The LSP enhancement will be much more important for blue LEDs with lower IQE. For these latter the spacing between the NPs, i.e. the NPs density should be at least 1010 cm2

, in general agreement with experimental observations (see e.g. Fig. 18). At the same time, the

results of Ref [113] suggest that, for less efficient green and UV LEDs, the effect of LSP enhancement should be more pronounced than for highly efficient blue LEDs. The effects of the NP array density and the NP diameter on the extinction spectra of such arrays was studied theoretically in several papers, one of the most recent being Ref. [114] where the reader can find several useful references to earlier work. The results of DDA [5-8] calculations of the extinction spectra for Ag NPs with the diameter of 8 nm as a function of interparticle spacing are presented in Fig. 50. It can be seen that as the distance between the NPs decreases the calculated extinction spectra broaden and shift to longer wavelengths. If one fixes the distance between the NPs and varies the NP diameter one observes the shift of the peak wavelength to longer values with increasing the NP diameter (see Fig. 51). The authors also performed experimental measurements of the extinction spectra of the Ag clusters deposited from vapor phase as a function of deposition time and film coverage. TEM measurements  

allowed to determine the distribution of the particle dimensions and spacing for the given coverage. A very reasonable agreement of experimental results with the calculated dependences in Fig. 50-51 was observed if one considered that the main factor in these dependences was the fraction of closely spaced pairs (CSPs) with the fixed diameter and varying spacing between the pairs or the fraction of CSPs with the fixed spacing but varying diameter. The explanation offered in Ref. [114] was that, for CSPs, a strong confinement of the LSP dipole electric field leading to strong near field coupling occurs and such coupling determines the optical properties of the LSP spectra produced by the ensemble of NPs. For the lowest spacing between the NPs the experimental spectra were considerably red-shifted compared to the model spectra, presumably due to the contribution of the higher order multipole interactions [114]. The predicted and observed behavior in Ref. [114] illustrates that the spectral range in which the metallic NPs can be used for the enhancement of LED performance can be quite broad. For example, Ag NPs can be successfully used not only in the blue, but also in the green spectral range depending on the particle size and spacing. The dependence on the interparticle spacing in Fig. 51 is in line with the calculations of the enhancement factor performed in Ref. [113].

VIII. Conclusions We have seen above that SPs formed by continuous metal films or LSPs formed by metal NPs can provide a very efficient recombination channel in GaN-based QW LED structures if the distance between the SP/LSP layer and the QW is low enough, typically below ~50 nm. Under these conditions the energy of electron-hole pairs recombining in the QW can be effectively redirected from the non-radiative recombination channel into the SP/LSP excitation and then converted into the external photons flux. The re-directing efficiency of the QW recombination  

energy into the SP/LSP energy (the Purcell factor) is determined by the structure of the metal film (metal type, thickness, and roughness for SPs; NP size, shape, and composition for LSPs), by the distance to the QW, and by the closeness of the QW energy to the SP/LSP resonance energy. The conversion of the SP/LSP energy into external photons flux increases with increased roughness of the metal films, and by the introduction of the subwavelength grating for SPs, and by the size, shape, and composition of the NPs in LSPs. The conversion efficiency is also increased by minimizing the dissipation losses in metal films or NPs by using insulating films with low refractive index to separate the NP one from another or to separate metal from underlying semiconductor. It can also be enhanced by using crystalline NPs produced by chemical synthesis as opposed to polycrystalline NPs prepared by deposition and annealing of metal films. With high Purcell factors, low dissipation losses, and high-yield of photons in SP/LSP, remarkable improvements in LED performance can be achieved owing to the suppression of the non-radiative recombination channel in QWs. The best SP/LSP materials for different spectral regions are Au for the red-green region, Ag for the green-blue-near UV region, and Al for the UV region. At that, the use of core/shell NPs with metal core and dielectric shell, such as Ag/SiO2 or Au/SiO2 provides pivotal advantages in terms of the high stability of optical and structural properties and suppressed contamination of the LED layers by metal-related complexes (the latter is particularly important in the schemes employing overgrowth of NPs). For different spectral regions the amount of improvements introduced by SP/LSP coupling and the main focus of research efforts are somewhat different. In high-quality blue LEDs with IQEs approaching ~100% the main contributions of SP/LSP coupling are the suppression of the efficiency droop and the increase in the light extraction efficiency. For green LEDs the focus is   

mainly on increasing the IQE that is very low because of the strong impact of non-radiative recombination caused by high defect density in these LEDs. This factor remains extremely important in UV LEDs. However, for the UV LED devices the improvement of the LEE, very low in general because of the unique anisotropic optical polarization property and the strong light absorption in p-GaN capping layer, is similarly or even more important, and SP/LSP coupling has been shown to be capable of tackling this problem. A wide variety of methods of integrating SP/LSP coupling into the actual LED structures has been described. The most straightforward way is, of course, simple deposition of the metal layers on top of the ready LED structures. This approach can be easily built into standard LED fabrication procedures and thus is very attractive from the standpoint of easy implementation and good compatibility with existing technologies. Its main stumbling block is the optimization of the p-GaN cap design in such a way as to produce good coupling between the SP/LSP layer and the QW without compromising the series resistance and the injection efficiency. The approach in which the metal NPs are placed into the pockets of the LED structure, either shallow, in which the p-GaN cap is partially thinned down, or deep, where the etchings goes all the way down to n-GaN layer beyond the p-n junction, is also quite attractive in terms of compatibility with routine LED fabrication techniques because it involves only post-growth processing. The problems here are the serious technological challenges related to the required high precision control over the size and etch damages of the nanopillars, and the necessity of careful optimization of the p-GaN roughening in order not to increase the series resistance and the turn-on voltage beyond acceptable limits. The process designs in which the metal NPs are embedded into the LED structure very close to the QW by placing them either underneath the QW active layer or above the QW active layer   

into the p-GaN film have produced very promising results, but are technologically extremely challenging. The version in which the NPs are embedded into p-GaN above the QW region seems less demanding in that respect and has been demonstrated to be suitable for LEDs in the blue, green and UV spectral regions. The main issue here is to minimize the defect formation caused by the overgrowth of metal NPs and the contamination of the QW region with metalrelated complexes. In that respect the use of core/shell NPs of Ag/SiO2 or Au/SiO2 types seems quite promising. It is obvious that the field will benefit tremendously by development of consistent predictive LSP/QW interaction models and by the emergence of commercially available and economically feasible sources of metal NPs directly applicable for implementing into the LED fabrication process as is the case with phosphors or QD color converters. The potential of the light redirecting and extraction using this approaches for waveguiding, focusing, and beam splitting has not been as yet exploited to its full capacity. But all in all, it is clear that the performance enhancement in GaN-based LEDs by using SP/LSP coupling has already become a very rapidly developing and promising area of plasmonics.

Acknowledgments I.-H Lee would like to acknowledge financial support from National Research Foundation of Korea (NRF) funded by Ministry of Science, ICT & Future Planning (2013R1A2A2A07067688 and 2010-0019626). AYP gratefully acknowledges the support by the Ministry of Education and Science of the Russian Federation in the framework of Increase Competitiveness Program of NUST «MISiS» (ʋ Ʉ2-2014-055). We would also like to thank our many years collaborators in the field, Dr. Jong-Hyeob Baek (KOPTI, Gwanju, Korea), Prof Jin-Kyu Yang (Kongju National  

University, Chonan, Korea), Dr. Nikolay B. Smirnov (Institute of Rare Metals, Moscow, Russia). The authors would like to thank Prof. Min-Ki Kwon at Chosun University for having read the manuscript and for helpful discussions and suggestions. REFERENCES [1] M. Stockman, Opt. Express 19 (2011) 22029-22106. [2] S.A. Maier, H.A. Atwater, J. Appl. Phys. 98 (2005) 011101. [3] S.M. Morton, D.W. Silverstein, L. Jensen, Chem. Rev. 111 (2011) 3962-3994. [4] G.V. Hartland, Chem. Rev. 111 (2011) 3858-3887. [5] M. Rycenga, C.M. Cobley, J. Zeng, W. Li, C.H. Moran, Q. Zhang, D. Qin, Y. Xia, Chem. Rev. 111 (2011) 3669-3712. [6] K.M. Mayer, J.H. Hafner, Chem. Rev. 111 (2011) 3828-3857. [7] V. Giannini, A.I.F. Domínguez, S.C. Heck, S.A. Maier, Chem. Rev. 111 (2011) 38883912. [8] Y. Takahashi, T. Tatsuma, Appl. Phys. Lett. 99 (2011) 182110. [9] E.F. Schubert, Light-Emitting Diodes, Second ed., Cambridge University Press, Cambridge, UK, 2006. [10] A. Denis, G. Goglio, G. Demazeau, Mater. Sci. Eng. R 50 (2006) 167-194 [11] W. Walukiewicz, J.W. Ager III, K.M. Yu, Z. Liliental-Weber, J. Wu, S.X. Li, R.E. Jones, J.D. Denlinger, J. Phys. D: Appl. Phys. 39 (2006) R83. [12] J. Li, K.B. Nam, M.L. Nakarmi, J.Y. Lin, H.X. Jiang, Appl. Phys. Lett. 81 (2002) 3365. [13] M.L. Nakarmi, N. Nepal, J.Y. Lin, H.X. Jiang, Appl. Phys. Lett. 94 (2009) 091903. [14] M.L. Nakarmi, N. Nepal, C. Ugolini, T.M. Altahtamouni, J.Y. Lin, H.X. Jiang, Appl. Phys. Lett. 89 (2006) 152120. 
 

[15] H.Y. Ryu, I.G. Choi, H.S. Choi, J.I. Shim, Appl. Phys. Express 6 (2013) 062101. [16] R. Mueller-Mach, G. Mueller, M.R. Krames, H.A. Hoppe, F. Stadler, W. Schnick, T. Juestel, P. Schmidt, Phys. Stat. Sol. (a) 202 (2005) 1727-1732. [17] M. Kneissl, T. Kolbe, C. Chua, V. Kueller, N. Lobo, J. Stellmach, A. Knauer, H. Rodriguez, S. Einfeldt, Z. Yang, N. M. Johnson, M.Weyers, Semicond. Sci. Technol. 26 (2011) 014036. [18] A. Khan, K. Balakrishnan, T. Katona, Nat. Photonics 2 (2008) 77-84. [19] H. Zhao, G. Liu, J. Zhang, J.D. Poplawsky, V. Dierolf, N. Tansu, Opt. Express 19(S4), (2011) A992. [20] B. Zhang, Y. Liu, Chin. Sci. Bull. 59 (2014) 1251-1275. [21] Y.L. Li, Y.R. Huang, Y.H. Lai, Appl. Phys. Lett. 91 (2007) 181113. [22] M. Meneghini, M. la Grassa, S. Vaccari, B. Galler, R. Zeisel, P. Drechsel, B. Hahn, G. Meneghesso, E. Zanoni, Appl. Phys. Lett. 104 (2014) 113505. [23] N.T. Son, M. Bickermann, E. Janzen, Appl. Phys. Lett. 98 (2011) 092104. [24] B.E. Gaddy, Z. Bryan, I. Bryan, R. Kirste, J. Xie, R. Dalmau, B. Moody, Y. Kumagai, T. Nagashima, Y. Kubota, T. Kinoshita, A. Koukitu, Z. Sitar, R. Collazo, D.L. Irving, Appl. Phys. Lett. 103 (2013) 161901. [25] N. Nepal, K.B. Nam, M.L. Nakarmi, J.Y. Lin, H.X. Jiang, J.M. Zavada, R.G. Wilson, Appl. Phys. Lett. 84 (2004) 1090. [26] S. Nakamura, Science 281 (1998) 956-961. [27] J.W. Liaw, Eng. Anal. Bound. Elem. 30 (2006) 734-745. [28] K.L. Kelly, E. Coronado, L.L. Zhao, G.C. Schatz, J. Phys. Chem. B 107 (2003) 668-677. [29] J.M. Sanz, D. Oritz, R. Alcaraz de la Osa, J.M. Saiz, F. Gonzalez, A.S. Brown, M. Losurdo, H.O. Everitt, H. Moreno, J. Phys. Chem. C117 (2013) 19606-19615.  

[30] A. Neogi, C.W. Lee, H.O. Everitt, T. Kuroda, A. Takeuchi, E. Yablonovitch, Phys. Rev. B66 (2002) 153305. [31] I. Gontijo, M. Boroditsky, E. Yablonovitch, S. Keller, U.K. Mishra, S.P. DenBaars, Phys. Rev. B60 (1999) 11564-11567. [32] N.E. Hecker, R.A. Hopfel, N. Sawaki, T. Maier, G. Strasser, Appl. Phys. Lett. 75 (1998) 1577-1579. [33] W.L. Barnes, J. Lightwave Technol. 17 (1999) 2170. [34] K. Okamoto, I. Niki, A. Shwartser, Y. Narukawa, T. Mukai, A. Scherer, Nat. Mater 3 (2004) 601. [35] K. Okamoto, I. Niki, A. Scherer, Y. Narukawa, T. Mukai, Y. Kawakami, Appl. Phys. Lett. 87 (2005) 071102. [36] M.K. Kwon, J.Y. Kim, B.H. Kim, I.K. Park, C.Y. Cho, C.C. Byeon, S.J. Park, Adv. Mater. 20 (2008) 1253. [37] D.M. Yeh, C.F. Huang, C.Y. Chen, Y.C. Lu, C.C. Yang, Appl. Phys. Lett. 91 (2007) 171103. [38] D.M. Yeh, C.F. Huang, Y.C. Lu, C.Y. Chen, T.Y. Tang, J.J. Huang, K.C. Shen, Y.J. Yang, C.C. Yang, Appl. Phys. Lett. 91 (2007) 063121. [39] L.W. Jang, D.W. Jeon, T. Sahoo, D.S. Jo, J.W. Ju, S.J. Lee, J.H. Baek, J.K. Yang, J.H. Song, A.Y. Polyakov, I.H. Lee, Opt. Express 20 (2012) 2116-2123. [40] L.W. Jang, D.W. Jeon, J.W. Jeon, M. Kim, M.K. Kim, A.Y. Polyakov, J.W. Ju, S.J. Lee, J.H. Baek, M.S. Jeong, Y.H. Kim, I.H. Lee, J. Electrochem Soc. 159 (2012) H522-H524. [41] L.W. Jang, D.W. Jeon, M. Kim, J.W. Jeon, A.Y. Polyakov, J.W. Ju, S.J. Lee, J.H. Baek, J.K. Yang, I.H. Lee, Adv. Funct. Mater. 22 (2012) 2728-2734.  

[42] L.W. Jang, T. Sahoo, D.W. Jeon, M. Kim, J.W. Jeon, D.S. Jo, M.K. Kim, Y.T. Yu, A.Y. Polyakov, I.H. Lee, Appl. Phys. Lett. 99 (2011) 251114. [43] L.W. Jang, D.W. Jeon, T. Sahoo, A.Y. Polyakov, B. Saravanakumar, Y.T. Yu, Y.H. Cho, J.K. Yang , I.H. Lee, J. Mater. Chem. 22 (2012) 21749-21753. [44] C.Y. Cho, M.K. Kwon, S.J. Lee, S.H. Han, J.W. Kang, S.E. Kang, D.Y. Lee, S.J. Park, Nanotechnology 21 (2010) 205201. [45] C.Y. Cho, K.S. Kim, S.J. Lee, M.K. Kwon, H.Ko, S.T. Kim, G.Y. Jung, S.J. Park, Appl. Phys. Lett. 99 (2011) 041107. [46] C.Y. Cho, S.J. Lee, J.H. Song, S.H. Hong, S.M. Lee, Y.H. Cho, S.J. Park, Appl. Phys. Lett. 98 (2011) 051106. [47] K. Okamoto, I. Niki, A. Shvartser, G. Maltezos, Y. Narukawa, T. Mukai, Y. Kawakami, A. Scherer, Phys. Stat. Sol. (a) 204 (2007) 2103-2107. [48] D.M. Yeh, C.F. Huang, C.Y. Chen, Y.C. Lu, C.C. Yang, Nanotechnology 19 (2008) 345201. [49] A.Y. Polyakov, L.W. Jang, D.S. Jo, I.H. Lee, N.B. Smirnov, A.V. Govorkov, E.A. Kozhukhova, K.H. Baik, S.M. Hwang, J. Appl. Phys. 111 (2012) 033103. [50] R. Bardhan, N.K. Grady, N.J. Halas, Small 4 (2008) 1716-1722. [51] Y. Lu, Y. Yin, Z.Y. Li, Y. Xia, Nano Lett. 2 (2002) 785-788. [52] N. Gao, K. Huang, J. Li, S. Li, X. Yang, J. Kang, Sci. Rep. 2 (2012) 816. [53] Y. Xiong, J. Chen, B. Wiley, Y. Xia, Y. Yin, Z.Y. Li, Nano Lett. 5 (2005) 1237-1242. [54] S.H. Hong, C.Y. Cho, S.J. Lee, S.Y. Yim, W. Lim, S.T. Kim, S.J. Park, Opt. Express 21 (2013) 3138-3144. [55] A. Ono, M. Kikawada, R. Akimoto, W. Inami, Y. Kawata, Opt. Express 21 (2013) 17447.  

[56] C.Y. Cho, Y. Zhang, E. Cicek, B. Rahnema, Y. Bai, R. McClintock, M. Razeghi, Appl. Phys. Lett. 102 (2013) 211110. [57] A. Taguchi, Y. Saito, K. Watanabe, S. Yijian, S. Kawata, Appl. Phys. Lett. 101 (2012) 081110. [58] M.A. Dijk, A.L. Tchebotareva, M. Orrit, M. Lippitz, S. Berciaud, D. Lasne, L. Cognet, B. Lounis, Phys. Chem. Chem. Phys. 8 (2006) 3486. [59] A. Tcherniak, J.W. Ha, S. Dominguez-Medina, L.S. Slaughter, S. Link, Nano Lett. 10 (2010) 1398. [60] X. Miao, I. Brener, T.S. Luk, J. Opt. Soc. Am. B 27 (2010) 1561-1569. [61] Y.C. Lu, Y.S. Chen, F.J. Tsai, J.Y. Wang, C.H. Lin, C.Y. Chen, Y.W. Kiang, C.C. Yang, Appl. Phys. Lett. 94 (2009) 233113. [62] L.W. Jang, J.W. Ju, J.W. Jeon, D.W. Jeon, J.H. Choi, S.J. Lee, S.R. Jeon, J.H. Baek, E. Sari, H.V. Demir, H.D. Yoon, S.M. Hwang, I.H. Lee, Proc. SPIE Vol. 7945 (2011) 794511-1. [63] A.Y. Polyakov, D.W. Jeon, N.B. Smirnov, A.V. Govorkov, E.A. Kozhukhova, E.B. Yakimov, I.H. Lee, J. Appl. Phys. 112 (2012) 073112 [64] A.Y. Polyakov, D.W. Jeon, A.V. Govorkov, N.B. Smirnov, V.N. Sokolov, E.A. Kozhukhova, E.B. Yakimov, I.H. Lee, J. Alloy. Compd. 554 (2013) 258-263 [65] J. Kamimura, K. Kishino, A. Kikuchi, Phys. Stat. Sol. R 6 (2012) 123 [66] C. Hahn, Z. Zhang, A. Fu, C.H. Wu, Y.J. Hwang, D.J. Gargas, P. Yang, ACS Nano 5 (2011) 3970. [67] M.H. Kim, M.F. Schubert, Q. Dai, J.K. Kim, E.F. Schubert, J. Piprek, Y. Park, Appl. Phys. Lett. 91 (2007) 183507. [68] W. Yang, Y. He, L. Liu, X. Hu, Appl. Phys. Lett. 102 (2013) 241111.  

[69] J. Hader, J.V. Moloney, S.W. Koch, Appl. Phys. Lett. 99 (2011) 181127. [70] X. Shao, H. Lu, D. Chen, Z. Xie, R. Zhang, Y. Zheng, Appl. Phys. Lett. 95 (2009) 163504. [71] Y.B. Tao, Z.Z. Chen, F.F. Zhang, C.Y. Jia, S. L. Qi, T.J. Yu, X.N. Kang, Z.J. Yang, L.P. You, D.P. Yu, G.Y. Zhang, J. Appl. Phys. 107 (2010) 103529. [72] Y.C. Shen, G.O. Mueller, S. Watanabe, N.F. Gardner, A. Munkholm, M.R. Krames, Appl. Phys. Lett. 91 (2007) 141101. [73] E. Kioupakis, P. Rinke, K.T. Delaney, C.G. Van de Walle, Appl. Phys. Lett. 98 (2011) 161107. [74] Q. Dai, Q. Shan, J. Wang, S. Chhajed, J. Cho, E.F. Schubert, M.H. Crawford, D.D. Koleske, M.H. Kim, Y. Park, Appl. Phys. Lett. 97 (2010) 133507. [75] M. Schmeits, N.D. Nguyen, M. Germain, J. Appl. Phys. 89 (2001) 1890. [76] D.J. Kim, J. Appl. Phys. 88 (2000) 1929. [77] D.J. Kim, D.Y. Ryu, N. Bojarczuk, J. Karasinski, S. Guha, S. H. Lee, J.H. Lee, J. Appl. Phys. 88 (2000) 2564. [78] C.F. Lu, C.H. Liao, C.Y. Chen, C. Hsieh, Y.W. Kiang, C.C. Yang, Appl. Phys. Lett. 96 (2010) 261104. [79] A.Y. Polyakov, N.B. Smirnov, A.V. Govorkov, A.V. Osinsky, P.E. Norris, S.J. Pearton, J. Van Hove, A. Wowchak, P. Chow, Appl. Phys. Lett. 79 (2001) 4372-4374. [80] J.H. Sung, J.S. Jang, B.S. Kim, C.H. Choi, M.W. Lee, S.G. Lee, S.G. Park, E.H. Lee, B.H. O, Appl. Phys. Lett. 96 (2010) 261105. [81] L.W. Jang, J.W. Ju, D.W. Jeon, J.W. Park, A.Y. Polyakov, S.J. Lee, J.H. Baek, S.M. Lee, Y.H. Cho, I.H. Lee, Opt. Express 20 (2012) 6036-6040.

 

[82] L.W. Jang, D.W. Jeon, A.Y. Polyakov, A.V. Govorkov, V.N. Sokolov, N.B. Smirnov, T.H. Chung, H.S. Cho, J.H. Yun, K.D. Shcherbatchev, J.H. Baek, J.W. Lee, I.H. Lee, J. Alloy. Compd. 589 (2014) 507-512. [83] F. Qian, Y. Li, S. Gradecak, D. Wang, C.J. barrelet, C.M. Lieber, Nano Lett. 4 (2004) 19751979. [84] C.M. Lieber, Z.L. Wang, MRS Bull. 32 (2007) 99-105. [85] Y. Li, R. Zhang, Z. Xie, B. Liu, P. Chen, G. Zhang, T. Tao, Z. Zhuang, L. Zhi, Appl. Phys. Lett. 114 (2013) 113104. [86] K. Okamoto, Y. Kawakami, IEEE J. Sel. Top. Quantum Electron. 15 (2009) 1199–1209. [87] E. Kioupakis, Q. Yan, C.G. Van de Walle, Appl. Phys. Lett. 101 (2012) 231107. [88] C.Y. Chen, Y.C. Lu, D.M. Yeh, C.C. Yang, Appl. Phys. Lett. 90 (2007) 183114. [89] K.G. Gryczynski, P.R. Vemuri, I. Watson, A. Neogi, Appl. Phys. Lett. 99 (2011) 121905. [90] T. Takeuchi, S. Sota, M. Katsuragawa, M. Komori, H. Takeuchi, H. Amano, I. Akasaki, Jpn. J. Appl. Phys. 36 (1997) L382. [91] M. Funato, Y. Kawakami, MRS Bull. 34 (2009) 334. [92] B.M. Imer, F. Wu, S.P. DenBaars, J.S. Speck, Appl. Phys. Lett. 88 (2006) 061908. [93] T. Kawashima, T. Nagai, D. Iida, A. Miura, Y. Okadome, Y. Tsuchiya, M. Iwaya, S. Kamiyama, H. Amano, I. Akasaki, Phys. Status Sol. (b) 244 (2007) 1848. [94] A.Y. Polyakov, N.B. Smirnov, A.V. Govorkov, A.V. Markov, T.G. Yugova , E.A. Petrova, H. Amano, T. Kawashima, K. D. Scherbatchev, V.T. Bublik, J. Appl. Phys. 105 (2009) 063708. [95] A.Y. Polyakov, N.B. Smirnov, A.V. Govorkov, A.V. Markov, Q. Sun, Y. Zhang, C.D. Yerino, T.S. Ko, I.H. Lee, J. Han, Mater. Sci. Eng. B 166 (2010) 220-224.

 

[96] A.Y. Polyakov, N.B. Smirnov, A.V. Govorkov, H. Amano, S.J. Pearton, I.H. Lee, Q. Sun, J. Han, S.Y. Karpov, Appl. Phys. Lett. 98 (2011) 072104. [97] D.F. Feezell, M.C. Schmidt, S.P. DenBaars, S. Nakamura, MRS Bull. 34 (2009) 318. [98] M.S. Shur, R. Gaska, IEEE Trans. Electron. Devices 57 (2010) 12-25. [99] H. Hirayama, Proc. SPIE 7617 (2010) 76171G. [100] M. Shatalov, J. Yang, W. Sun, R. Kennedy, R. Gaska, K. Liu, M. Shur, G. Tamulaitis, J. Appl. Phys. 105 (2009) 073103. [101] J. Ryou, P. Yoder, J. Liu, IEEE J. Sel. Top. Quantum Electron. 15 (2009) 1080-1091. [102] R.M. Farrell, E.C. Young, F.Wu, S.P. DenBaars, J.S. Speck, J. Seimicond. Technol. Sci. 27 (2012) 024001. [103] T.N. Oder, K.H. Kim, J.Y. Lin, H.X. Jiang, Appl. Phys. Lett. 84 (2004) 466. [104] J.H. Sung, B.S. Kim, C.H. Choi, M.W. Lee, S.G. Lee, S.G. Park, E.H. Lee, B.H. O, Microelectron. Eng. 86 (2009) 1120-1123. [105] K.B. Nam, J. Li, M.L. Nakarmi, J.Y. Lin, H.X. Jiang, Appl. Phys. Lett. 84 (2004) 5264. [106] H. Kawanishi, M. Senuma, M. Yamamoto, E. Niikura, T. Nukui, Appl. Phys. Lett. 89 (2006) 081121. [107] J. Zhang, H. Zhao, N. Tansu, Appl. Phys. Lett. 97 (2010) 111105. [108] M. Khizar, Z.Y. Fan, K.H. Kim, J.Y. Lin, H.X. Jiang, Appl. Phys. Lett. 86 (2005) 173504. [109] D.S. Liu, T.W. Lin, B.W. Huang, F.S. Juang, P.H. Lei, C.Z. Hu, Appl. Phys. Lett. 94 (2009) 143502. [110] T. Fujii, Y. Gao, R. Sharma, E.L. Hu, S.P. DenBaars, S. Nakamura, Appl. Phys. Lett. 84 (2004) 855. [111] H.Y. Ryu, Nanoscale Res. Lett. 9 (2014) 58.   

[112] F. Magnan, J. Gagnon, F.G. Fontaine, D. Boudreau, Chem. Comm. 49 (2013) 9299-9301. [113] J.B. Khurgin, G. Sun, R.A. Soref, Appl. Phys. 93 (2008) 021120. [114] Y. Gong, Y. Zhou, L. He, B. Xie, F. Song, M. Han, G. Wang, Eur. Phys. J. D (2013) 67:87. Figure captions  Fig. 1 (Color online) Spectral dependence of real and imaginary parts of dielectric function for several metals; the horizontal red line corresponds to εr=-2 (After Ref. [29] J. M. Sanz, D. Oritz, R. Alcaraz de la Osa, J. M. Saiz, F. Gonzalez, A. S. Brown, M. Losurdo, H. O. Everitt, and H. Moreno, UV Plasmonic Behavior of Various Metal Nanoparticles in the Near- and Far-Field Regimes: Geometry and Substrate Effects, J. Phys. Chem. C117, 19606-19615 (2013), Fig. 2 (a), (b), reproduced with permission, copyright: the American Chemical Society, 2013)

Fig. 2 (Color online) (a) The positions of the LSP resonance estimated from the Frolich energy calculated from the spectral dependence of the bulk dielectric functions of various metals; (b) the spectral position and estimated magnitude of the maximum quality factor QLSPmax estimated from the spectral dependence of the bulk dielectric function (After Ref. [29] J. M. Sanz, D. Oritz, R. Alcaraz de la Osa, J. M. Saiz, F. Gonzalez, A. S. Brown, M. Losurdo, H. O. Everitt, and H. Moreno, UV Plasmonic Behavior of Various Metal Nanoparticles in the Near- and Far-Field Regimes: Geometry and Substrate Effects, J. Phys. Chem. C117, 19606-19615 (2013), Fig. 3 (a),(b), reproduced with permission, copyright: the American Chemical Society, 2013)   

Fig. 3 Measured and calculated Purcell factors Fsp for Ag film on GaN/InGaN QW structure, TRPL-P1 (solid triangles) corresponds to the experimental spectrum for the Ag layer thickness of 8 nm, TRPL-P2 (solid circles) –to the thickness of 6 nm (after Ref. [30] A. Neogi, C.-W. Lee, H. O. Everitt, T. Kuroda, A. Takeuchi, and E. Yablonovitch, Enhancement of spontaneous recombination rate in a quantum well by resonant surface plasmon coupling, Phys. Rev. B66, 153305 (2002), Fig. 3 (a), reproduced with permission, copyright: the American Physical Society, 2002)

Fig. 4 Photoluminescence (PL) spectrum for uncoated GaN/InGaN QW structure (curve A), PL spectrum corrected for the absorption attenuation due to the 8 nm-thick Ag film (curve B), PL spectrum measured for the Ag coated (thickness 8 nm) structure (curve C) (After Ref. [31] I. Gontijo, M. Boroditsky, E. Yablonovitch, S. Keller, U.K. Mishra, and S.P. DenBaars, Coupling of InGaN quantum well photoluminescence to silver surface plasmons, Phys. Rev. B60(16), 11564-11567 (1999), Fig. 1(b), reproduced with permission, copyright: the American Physical Society, 1999)

Fig. 5 (Color online) Schematic representation of optical transitions and decay rates in QW structure with roughened planar metal layer (after Ref. [35] K. Okamoto, I. Niki, A. Scherer, Y. Narukawa, T. Mukai, and Y. Kawakami, Surface plasmon enhanced spontaneous emission rate of InGaN/GaN quantum wells probed by time-resolved photoluminescence spectroscopy, Appl. Phys. Lett. 87, 071102 (2005), Fig. 3, reproduced with permission, copyright: American Institute of Physics, 2005)  

Fig. 6 (Color online) Electric field distribution of the dipolar SP mode with different source radiation: (a) single Ag/SiO2-NP, (b) three Ag/SiO2-NP, (c) seven Ag/SiO2-NP and (d) resonance intensity of Ag/SiO2-NPs along the z-axis. The Fig. 6(d) inset shows radiation spectra of Ag/SiO2-NPs from FDTD simulation (after Ref. [39] Lee-Woon Jang, Dae-Woo Jeon, Trilochan Sahoo, Dong-Seob Jo, Jin-Woo Ju, Seung-Jae Lee, Jong-Hyeob Baek, JinKyu Yang, Jung-Hoon Song, Alexander Y. Polyakov, and In-Hwan Lee, Localized surface plasmon enhanced quantum efficiency of InGaN/GaN quantum wells by Ag/SiO2 nanoparticles, Opt. Express 20(3), 2116-2123 (2012), Fig. 3; reproduced by permission, copyright: Optical Society of America, 2012)

Fig. 7 TEM images of Ag NPs and Ag/SiO2 core/shell NPs (after Ref. [41] Lee-Woon Jang , Dae-Woo Jeon , Myoung Kim , Ju-Won Jeon , Alexander Y. Polyakov ,Jin-Woo Ju , SeungJae Lee , Jong-Hyeob Baek , Jin-Kyu Yang , and In-Hwan Lee, Investigation of Optical and Structural Stability of Localized Surface Plasmon Mediated Light-Emitting Diodes by Ag and Ag/SiO2 Nanoparticles, Adv. Funct. Mater. 22, 2728-2734 (2012); Fig. 2 (a), (b); Reproduced with permission, copyright: Wiley-VCH Verlag GmbH&Co, 2012)

Fig. 8 TEM images of Ag/SiO2 core/shell NPs with different shell thicknesses (A, B, C), selective electron diffraction image demonstrating single crystalline nature of Ag core (D) (after Ref. [42] Lee-Woon Jang, Trilochan Sahoo, Dae-Woo Jeon, Myoung Kim, Ju-Won Jeon, Dong-Seob Jo, Min-Kyu Kim, Yeon-Tae Yu, Alexander Y. Polyakov, and In-Hwan Lee, Quantum efficiency control of InGaN/GaN multi-quantum-well structures using 
 

Ag/SiO2 core-shell nanoparticles, Appl. Phys. Lett. 99, 251114 (2011), Fig. 1, reproduced with permission, copyright: American Institute of Physics, 2012)

Fig. 9 (Color online) Absorbance spectra of Ag/SiO2 core/shell NPs with shell thicknesses of 2 nm (curve A), 5 nm (curve B), 20 nm (curve C) (after Ref. [42] Lee-Woon Jang, Trilochan Sahoo, Dae-Woo Jeon, Myoung Kim, Ju-Won Jeon, Dong-Seob Jo, Min-Kyu Kim, Yeon-Tae Yu, Alexander Y. Polyakov, and In-Hwan Lee, Quantum efficiency control of InGaN/GaN multi-quantum-well structures using Ag/SiO2 core-shell nanoparticles, Appl. Phys. Lett. 99, 251114 (2011), Fig. 2, reproduced with permission, copyright: American Institute of Physics, 2012)

Fig. 10 TEM image of Ag NPs formed in the p-GaN contact layer close to the MQW region by annealing (after Ref. [44] Chu-Young Cho, Min-KiKwon, Sang-Jun Lee, Sang-Heon Han, Jang-Won Kang, Se-EunKang, Dong-Yul Lee and Seong-Ju Park, Surface plasmonenhanced light-emitting diodes using silver nanoparticles embedded in p-GaN, Nanotechnology 21 (2010) 205201, Fig. 1 (b); reproduced by permission, copyright: IOP, 2010)

Fig. 11 (Color online) (a) schematic drawing of MQW GaN/InGaN LED with a layer of Ag NPs and SiO2 nanodiscs built into the p-GaN layer close to the MQW region, (b) plan-view SEM images of SiO2 nanodiscs on top of Ag NPs (After Ref. [45] Chu-Young Cho, Ki Seok Kim, Sang-Jun Lee, Min-Ki Kwon, Hyungduk Ko, Sung-Tae Kim, Gun-Young Jung, and Seong-Ju Park, Surface plasmon-enhanced light-emitting diodes with silver nanoparticles  

and SiO2 nano-disks embedded in p-GaN, Appl. Phys. Lett. 99, 041107 (2011), Fig. 1(a), (b); reproduced with permission, copyright: American Institute of Physics, 2011)

Fig. 12 Al discs prepared by photolithography on UV AlGaN LEDs (after Ref. [56] ChuYoung Cho, Yinjun Zhang, Erdem Cicek, Benjamin Rahnema, Yanbo Bai, Ryan McClintock, and Manijeh Razeghi, Surface plasmon enhanced light emission from AlGaNbased ultraviolet light-emitting diodes grown on Si (111), Appl. Phys. Lett. 102, 211110 (2013), Fig. 1(d), reproduced with permission, copyright: American Institute of Physics, 2013)

Fig. 13 Variation of the NSL fabricated Al NPs size and shape upon microwave heating of the PS monolayer mask increasing the diameter of polystyrene beads; the times of microwave heating are shown on each figure (after Ref. [57] Atsushi Taguchi, Yuika Saito, Koichi Watanabe, Song Yijian, and Satoshi Kawata, Tailoring plasmon resonances in the deep-ultraviolet by size-tunable fabrication of aluminum nanostructures, Appl. Phys. Lett. 101, 081110 (2012), Fig. 2 (a), (b), (c), (d); reproduced with permission, copyright: American Institute of Physics, 2012)

Fig. 14 (Color online) The change in the spectral position of LSP resonance in Al NPs prepared by NSL as a function NPs lateral size: comparison of experiment and modeling (after Ref. [57] Atsushi Taguchi, Yuika Saito, Koichi Watanabe, Song Yijian, and Satoshi Kawata, Tailoring plasmon resonances in the deep-ultraviolet by size-tunable fabrication of aluminum nanostructures, Appl. Phys. Lett. 101, 081110 (2012), Fig. 3(b), reproduced with  

permission, copyright: American Institute of Physics, 2012)

Fig. 15 (Color online) The temperature dependence of PL intensity for blue GaN/InGaN QW LED structure without and with Ag or Al SP layers (after Ref. [34] K. Okamoto, I. Niki, A. Shwartser, Y. Narukawa, T. Mukai, and A. Shere,. Surface-plasmon-enhanced light emitters based on InGaN quantum wells. Nat. Mater 3, 601 (2004), Fig. 4 (a), Copyright: Nature Publishing Group, 2004)

Fig. 16 (Color online) The room temperature PL spectra taken for blue GaN/InGaN QW LED structures without and with Ag, Al or Au SP films (after Ref. [34] K. Okamoto, I. Niki, A. Shwartser, Y. Narukawa, T. Mukai, and A. Shere,. Surface-plasmon-enhanced light emitters based on InGaN quantum wells. Nature materials 3, 601 (2004), Fig. 1 (b), Copyright: Nature Publishing Group, 2004)

Fig. 17 (Color online) The dependence of PL enhancement ratio for Ag, Al, Au on QW LEDs on the thickness of the spacer layer between the metal film and the QW (after Ref. [34] K. Okamoto, I. Niki, A. Shwartser, Y. Narukawa, T. Mukai, and A. Shere,. Surfaceplasmon-enhanced light emitters based on InGaN quantum wells. Nature materials 3, 601 (2004), Fig. 2 ( c), Copyright: Nature Publishing Group, 2004)

Fig. 18 (Color online) The dependendence of the PL enhancement ratio for Ag and Ag/SiO2 NPs on the areal density of NPs (After Ref. [41] Lee-Woon Jang , Dae-Woo Jeon , Myoung Kim , Ju-Won Jeon , Alexander Y. Polyakov, Jin-Woo Ju , Seung-Jae Lee , Jong-Hyeob  

Baek , Jin-Kyu Yang , and In-Hwan Lee, Investigation of Optical and Structural Stability of Localized Surface Plasmon Mediated Light-Emitting Diodes by Ag and Ag/SiO2 Nanoparticles, Adv. Funct. Mater. 2012, 22, 2728–2734 (Fig. 3(a); Copyright: Wiley-VCH Verlag 2012)

Fig. 19 (Color online) PL spectra of blue GaN/InGaN QW structures without and with Ag, Ni/Ag, SiN/Ag films (After Ref. [38] Dong-Ming Yeh, Chi-Feng Huang, Yen-Cheng Lu, Cheng-Yen Chen, Tsung-Yi Tang, Jeng-Jie Huang, Kun-Ching Shen, Ying-Jay Yang, and C. C. Yang, Surface plasmon leakage in its coupling with an InGaN/GaN quantum well through an Ohmic contact, Appl. Phys. Lett. 91, 063121 (2007) , Fig. 2, Copyright: AIP 2007)

Fig. 20 (Color online) PL decay times for MQW structures without (reference sample) and with Ag or Ag/SiO2 NPs (After Ref. [41] Lee-Woon Jang, Dae-Woo Jeon, Myoung Kim, Ju-Won Jeon, Alexander Y. Polyakov, Jin-Woo Ju, Seung-Jae Lee, Jong-Hyeob Baek , JinKyu Yang, and In-Hwan Lee, Investigation of Optical and Structural Stability of Localized Surface Plasmon Mediated Light-Emitting Diodes by Ag and Ag/SiO2 Nanoparticles, Adv. Funct. Mater. 2012, 22, 2728–2734 (Fig. 3(c))

Fig. 21 Dependence of Purcell factor calculated from the ratio of decay rates on the thickness of the SiO2 shell

Fig. 22 Dependence of PL enhancement factor on the thickness of the SiO2 shell for  

Ag/SiO2 core/shell NPs

Fig. 23 (Color online) Dispersion curves of the cases of various SiO2 thicknesses, including 0, 8, 15, 20, 30 nm, and infinity Ag/SiO2 evaluated based on the four-layer structure. The dashed curve represents the light lines in GaN and SiO2. The inset shows the SPP dissipation rates under various conditions (After Ref. [61] Yen-Cheng Lu, Yung-Sheng Chen, Fu-Ji Tsai, Jyh-Yang Wang, Cheng-Hung Lin,Cheng-Yen Chen, Yean-Woei Kiang, and C. C. Yang, Improving emission enhancement in surface plasmon coupling with an InGaN/GaN quantum well by inserting a dielectric layer of low refractive index between metal and semiconductor, Appl. Phys. Lett. 94, 233113 (2009) Fig. 4; Copyright: AIP 2009

Fig. 24 The schematics of the LSP structure formed by locally thinning down the p-GaN cap in the GaN/InGaN MQW structure (After Ref. [62] Lee-Woon Jang, Jin-Woo Ju, Ju-Won Jeon, Dae-Woo Jeon, Jung-Hun Choi, Seung-Jae Lee, Seong-Ran Jeon, Jong-Hyeob Baek, E. Sari, H. V. Demir, Hyung-Do Yoon, Sung-Min Hwang, and In-Hwan Lee, Enhanced optical characteristics of light emitting diodes by surface Plasmon of Ag nanostructures, Proc. of SPIE Vol. 7945 794511-1 (2011), Fig. 1(b); Copyright: SPIE 2011)

Fig. 25 (Color online) The schematic representation and SEM imaging of the nanopillar formed by dry etching in planar p-GaN/MQW GaN/InGaN MQW structures (Figures (a), (b)), of nanopillar structure with deposited Ag NPs (Figures (c ), (d)) and with Ag/SiO2 NPs (Figures (e), (f)) (After Ref. [43] Lee-Woon Jang, Dae-Woo Jeon, Trilochan Sahoo, Alexander Y. Polyakov, Balasubramaniam Saravanakumar, Yeon-Tae Yu, Yong-Hoon Cho,  

Jin-Kyu Yang, and In-Hwan Lee, Energy coupling processes in InGaN/GaN nanopillar light emitting diodes embedded with Ag and Ag/SiO2 nanoparticles, J. Mater. Chem., 2012, 22, 21749), Fig. 2(a), (b), (c), (d), (e ), (f)) Copyright: Royal Society of Chemistry, 2012)

Fig. 26 (Color online) PL changes induced in GaN/InGaN MQW nanopillar structures by deposition of Ag and Ag/SiO2 NPs (After Ref. [43] Lee-Woon Jang, Dae-Woo Jeon, Trilochan Sahoo, Alexander Y. Polyakov, Balasubramaniam Saravanakumar, Yeon-Tae Yu, Yong-Hoon Cho, Jin-Kyu Yang, and In-Hwan Lee, Energy coupling processes in InGaN/GaN nanopillar light emitting diodes embedded with Ag and Ag/SiO2 nanoparticles, J. Mater. Chem., 2012, 22, 21749), Fig. 3; Copyright: Royal Society of Chemistry, 2012)

Fig. 27 (Color online) PL decay curves in GaN/InGaN MQW nanopillar structures without and with Ag or Ag/SiO2 NPs (After Ref. [43] Lee-Woon Jang, Dae-Woo Jeon, Trilochan Sahoo, Alexander Y. Polyakov, Balasubramaniam Saravanakumar, Yeon-Tae Yu, YongHoon Cho, Jin-Kyu Yang, and In-Hwan Lee, Energy coupling processes in InGaN/GaN nanopillar light emitting diodes embedded with Ag and Ag/SiO2 nanoparticles, J. Mater. Chem., 2012, 22, 21749), Fig. 4; Copyright: Royal Society of Chemistry, 2012)

Fig. 28 (Color online) PL intensity evolution with time for Ag/MQW LED or Ag/SiO2/MQW LED structures heated to 100 oC in air; also shown is the schematics of the experimental setup (after Ref. [41] Lee-Woon Jang , Dae-Woo Jeon , Myoung Kim , JuWon Jeon , Alexander Y. Polyakov, Jin-Woo Ju , Seung-Jae Lee , Jong-Hyeob Baek , JinKyu Yang , and In-Hwan Lee, Investigation of Optical and Structural Stability of Localized  

Surface Plasmon Mediated Light-Emitting Diodes by Ag and Ag/SiO2 Nanoparticles, Adv. Funct. Mater. 2012, 22, 2728–2734; Fig. 5; Copyright: Wiley-VCH, 2012

Fig. 29 (Color online) Calculated EQE dependence on driving current for a blue QW LED structure without (reference sample) and with SP layer with Purcell factors assumed to be equal to 2 and 10 (After Ref. [68] Wei Yang, Yongfa He, Lei Liu, and Xiaodong Hu, Practicable alleviation of efficiency droop effect using surface plasmon coupling in GaNbased light emitting diodes, Appl. Phys. Lett. 102, 241111 (2013); Fig. 2(b); Copyright: AIP Publishing LLC, 2013)

Fig. 30 (Color online) Effect of varying the efficiency of LSP energy conversion into light (varying the r value in the figure) on the stating IQE values and the efficiency droop in blue GaN/InGaN QW LEDs(After Ref. [68] Wei Yang, Yongfa He, Lei Liu, and Xiaodong Hu, Practicable alleviation of efficiency droop effect using surface plasmon coupling in GaNbased light emitting diodes, Appl. Phys. Lett. 102, 241111 (2013); Fig. 3; Copyright: AIP Publishing LLC, 2013)

Fig. 31 Design of the SP-enhanced LED structure, showing the mesa structure with the pGaN/p-AlGaN/single quantum well (SQW) stack, the Ag SP layer sitting on SiN dielectric layer, the current spreading layer (CSL), Ni/Au ohmic contact to p-GaN, Ti/Al/Ti/Au ohmic contact to n-GaN (after Ref. [37] Dong-Ming Yeh, Chi-Feng Huang, Cheng-Yen Chen, Yen-Cheng Lu, and C. C. Yang, Surface plasmon coupling effect in an InGaN/GaN single  

quantum-well light-emitting diode, Appl. Phys. Lett. 91, 171103 (2007), Fig. 1, Copyright: AIP 2007)  Fig. 32 (Color online) Ohmic contacts pads and waveguiding grating design on SP enhanced LED (after Ref. [78] Chih-Feng Lu, Che-Hao Liao, Chih-Yen Chen, Chieh Hsieh, YeanWoei Kiang, and C. C. Yang, Reduction in the efficiency droop effect of a light-emitting diode through surface plasmon coupling, Appl. Phys. Lett. 96, 261104 (2010), Fig. 1, Copyright: AIP 2010)

Fig. 33 (Color online) Schematics of the Ag NP LSPs incorporation into GaN/InGaN MQW LED structure below the MQW region (After Ref. [36] M. K. Kwon, J. Y. Kim, B. H. Kim, I. K. Park, C. Y. Cho, C. C. Byeon, and S. J. Park, Surface-plasmon-enhanced light-emitting diodes, Adv. Mater. 20, 1253 (2008), Fig. 1, Copyright Wiley-VCH, 2008)

Fig. 34 (Color online) The schematic representation of the process flow in fabrication of LSP LEDs with Ag NPs embedded below the QW region and respective SEM images (after Ref. [81] Lee-Woon Jang, Jin-Woo Ju, Dae-Woo Jeon, Jae-Woo Park, A. Y. Polyakov, Seung-jae Lee, Jong-Hyeob Baek, Song-Mei Lee, Yong-Hoon Cho, and In-Hwan Lee, Enhanced light output of InGaN/GaN blue light emitting diodes with Ag nano-particles embedded in nano-needle layer, Opt. Express 20(6), 6036-6040 (2012), Fig. 1, Copyright Optical Society of America (OSA) 2012) 

  

Fig. 35 (Color online) Schematics of LED structure with Ag NPs embedded in p-GaN (After Ref. [44] Chu-Young Cho, Min-Ki Kwon, Sang-Jun Lee, Sang-Heon Han, Jang-Won Kang, SeEun Kang, Dong-Yul Lee and Seong-Ju Park, Surface plasmon-enhanced light-emitting diodes using silver nanoparticles embedded in p-GaN, Nanotechnology 21 (2010) 205201, Fig. 1(a), Copyright: IOP Publishing Ltd., 2010)

Fig. 36 (Color online) I-V characteristics of LEDs without and with Ag nanoparticles or Ag/SiO2 microdiscs nanoparticles; the inset shows the blow-up of the reverse current region (After Ref. [45] Chu-Young Cho, Ki Seok Kim, Sang-Jun Lee, Min-Ki Kwon, Hyungduk Ko, Sung-Tae Kim, Gun-Young Jung, and Seong-Ju Park, Surface plasmon-enhanced light-emitting diodes with silver nanoparticles and SiO2 nano-disks embedded in p-GaN, Appl. Phys. Lett. 99, 041107 (2011), Fig. 4(a), Copyright: AIP 2011)

Fig. 37 (Color online) Calculated effect of SP coupling on EQE and EQE droop in green LED structures (After Ref. [68] Wei Yang, Yongfa He, Lei Liu, and Xiaodong Hu, Practicable alleviation of efficiency droop effect using surface plasmon coupling in GaN-based light emitting diodes, Appl. Phys. Lett. 102, 241111 (2013), Fig. 4, Copyright: AIP, 2013)

Fig. 38 (Color online) The combined effects of increased laser pumping power and Ag SP coupling on the blue shift and intensity of the QW peak for GaN/InGaN QW LED (After Ref. [88] Cheng-Yen Chen, Yen-Cheng Lu, Dong-Ming Yeh, and C. C. Yang, Influence of the quantum-confined Stark effect in an InGaN/GaN quantum well on its coupling with surface

 

plasmon for light emission enhancement, Appl. Phys. Lett. 90, 183114 (2007), Fig. 1; Copyright: AIP, 2007)

Fig. 39 (Color online) (a) Admittance spectra of non-polar GaN/InGaN QW taken at various frequencies shown near respective curves; (b) DLTS spectrum obtained for this QW structure for reverse bias -1V and forward bias pulse of 1V for time windows 500 ms/ 5000 ms

Fig. 40 (Color online) PL spectra of the GaN/InGaN non-polar QW without and with Ag NPs

Fig. 41 Schematics of the Au grating on top of the green GaN/InGaN QW LED structure and respective PL enhancement as a function of the grating period (After Ref. [47] K. Okamoto, I. Niki, A. Shwartser, G. Maltezos, Y. Narukawa, T. Mukai, Y. Kawakami, and A. Sherer, Surface plasmon enhanced bright light emission from InGaN/GaN, Phys. Stat. Sol. (a) 204(6), 2103-2107 (2007), Fig. 3; Copyright: Wiley-VCH Verlag, 2007)

Fig. 42 (Color online) The PL enhancement in green GaN/InGaN MQW LEDs with Au NP embedded into p-GaN close to the QW; the inset shows the extinction cross section of the Au NPs on GaN (After Ref. [46] Chu-Young Cho, Sang-Jun Lee, Jung-Hoon Song, Sang-Hyun Hong, Song-Mae Lee, Yong-Hoon Cho, and Seong-Ju Park, Enhanced optical output power of green light-emitting diodes by surface plasmon of gold nanoparticles, Appl. Phys. Lett. 98, 051106 (2011), Fig. 2(b), Copyright: AIP 2011)


 

Fig. 43 (Color online) IV characteristics of the green GaN/InGaN MQW LEds with and without Au NPs embedded into p-GaN; the inset shows the blow-up of the reverse current region (After Ref. [46] Chu-Young Cho, Sang-Jun Lee, Jung-Hoon Song, Sang-Hyun Hong, Song-Mae Lee, Yong-Hoon Cho, and Seong-Ju Park, Enhanced optical output power of green light-emitting diodes by surface plasmon of gold nanoparticles, Appl. Phys. Lett. 98, 051106 (2011), Fig. 3, Copyright: AIP 2011)

Fig. 44 (Color online) A schematic illustration of the SP enhanced deep-UV LED (After Ref. [52] Na Gao, Kai Huang, Jinchai Li, Shuping Li, Xu Yang, Junyong Kang, Surface-plasmonenhanced deep-UV light emitting diodes based on AlGaN multi-quantum wells, Sci. Rep. 2, 816 (2012), Fig. 1, Copyright: Nature publishing group 2012)

Fig. 45 (Color online) Schematic diagram of the SP-enhanced AlGaN-based UV LED structure with embedded Al layer (After Ref. [56] C. Y. Cho, Y. Zhang, E. Cicek, B. Rahnema, Y. Bai, R. McClintock, M. Razeghi, Appl. Phys. Lett. 102, 211110 (2013), Fig. 1 (a), Copyright: AIP 2013)

Fig. 46 (Color online) Schematic diagram of the LSP-enhanced NUV-LEDs with metal NPs in p-GaN layer. SEM images of (b) Ag NPs and (c) Pt NPs on the 20 nm-thick p-GaN spacer layer after a rapid thermal annealing process (After Sang-Hyun Hong, Chu-Young Cho, Sang-Jun Lee, Sang-Youp Yim, Wantae Lim, Sung-Tae Kim, and Seong-Ju Park, Opt. Express 21, 31383144 (2013), Fig. 1, Copyright: OSA 2013)

 

Fig. 47 (Color online) PL spectra at 10 K and 300 K of the NUV-LEDs; (a) without metal NPs, (b) with Ag NPs, and (c) with Pt NPs. (d) PL enhancement ratio of PL intensity of the NUVLEDs with Ag and Pt NPs to that of the NUV-LED without metal NPs. The inset shows the absorbance of Ag and Pt NPs as a function of wavelength (After Sang-Hyun Hong, Chu-Young Cho, Sang-Jun Lee, Sang-Youp Yim, Wantae Lim, Sung-Tae Kim, and Seong-Ju Park, Opt. Express 21, 3138-3144 (2013), Fig. 2, Copyright: OSA 2013)

Fig. 48 Dependence of the luminescence enhancement by LSPs produced by Ag nanospheres array with the the sphere radius a and the distance between the spheres R as a function of a for three different values of the starting QW luminescence efficiency hrad and the distance between the QW and the Ag NPs layer d=10 nm; also shown is the optimal spacing between the spheres Ropt (After Ref. [113], Fig. 3, Copyright: American Institute of Physics 2008)

Fig. 49 Dependence of LSP enhancement by the array of Ag NPs as a function of the starting radiation efficiency hrad for three different values of the spacing between the the QW and the Ag NPs d (After Ref. [113], Fig. 4, Copyright: American Institute of Physics 2008)

Fig. 50 (Color online) Calculated dependence of the extinction spectra of Ag NP film on NP spacing for the fixed NP diameter of 8 nm (After Ref. [114], Fig. 4, Copyright: EDP Sciences, Societ`a Italiana di Fisica, Springer-Verlag 2013)

 

Fig. 51 (Color online) Calculated dependence of the extinction spectra of Ag NP film on NP diameter for the fixed NP spacing of 10 nm (after Ref. [114], Fig. 7, Copyright: EDP Sciences, Societ`a Italiana di Fisica, Springer-Verlag 2013)

Figures 

Fig. 1 (Color online) Spectral dependence of real and imaginary parts of dielectric function for several metals; the horizontal red line corresponds to εr=-2 (After Ref. [29] J. M. Sanz, D. Oritz, R. Alcaraz de la Osa, J. M. Saiz, F. Gonzalez, A. S. Brown, M. Losurdo, H. O.  

Everitt, and H. Moreno, UV Plasmonic Behavior of Various Metal Nanoparticles in the Near- and Far-Field Regimes: Geometry and Substrate Effects, J. Phys. Chem. C117, 19606-19615 (2013), Fig. 2 (a), (b), reproduced with permission, copyright: the American Chemical Society, 2013)

Fig. 2 (Color online) (a) The positions of the LSP resonance estimated from the Frolich energy calculated from the spectral dependence of the bulk dielectric functions of various metals; (b) the spectral position and estimated magnitude of the maximum quality factor QLSPmax estimated from the spectral dependence of the bulk dielectric function (After Ref. [29] J. M. Sanz, D. Oritz, R. Alcaraz de la Osa, J. M. Saiz, F. Gonzalez, A. S. Brown, M. Losurdo, H. O. Everitt, and H. Moreno, UV Plasmonic Behavior of Various Metal Nanoparticles in the Near- and Far-Field Regimes: Geometry and Substrate Effects, J. Phys. Chem. C117, 19606-19615 (2013), Fig. 3 (a),(b), reproduced with permission, copyright: the American Chemical Society, 2013)

 

Fig. 3 Measured and calculated Purcell factors Fsp for Ag film on GaN/InGaN QW structure, TRPL-P1 (solid triangles) corresponds to the experimental spectrum for the Ag layer thickness of 8 nm, TRPL-P2 (solid circles) –to the thickness of 6 nm (after Ref. [30] A. Neogi, C.-W. Lee, H. O. Everitt, T. Kuroda, A. Takeuchi, and E. Yablonovitch, Enhancement of spontaneous recombination rate in a quantum well by resonant surface plasmon coupling, Phys. Rev. B66, 153305 (2002), Fig. 3 (a), reproduced with permission, copyright: the American Physical Society, 2002)

 

Fig. 4 Photoluminescence (PL) spectrum for uncoated GaN/InGaN QW structure (curve A), PL spectrum corrected for the absorption attenuation due to the 8 nm-thick Ag film (curve B), PL spectrum measured for the Ag coated (thickness 8 nm) structure (curve C) (After Ref. [31] I. Gontijo, M. Boroditsky, E. Yablonovitch, S. Keller, U.K. Mishra, and S.P. DenBaars, Coupling of InGaN quantum well photoluminescence to silver surface plasmons, Phys. Rev. B60(16), 11564-11567 (1999), Fig. 1(b), reproduced with permission, copyright: the American Physical Society, 1999)

 

Fig. 5 (Color online) Schematic representation of optical transitions and decay rates in QW structure with roughened planar metal layer (after Ref. [35] K. Okamoto, I. Niki, A. Scherer, Y. Narukawa, T. Mukai, and Y. Kawakami, Surface plasmon enhanced spontaneous emission rate of InGaN/GaN quantum wells probed by time-resolved photoluminescence spectroscopy, Appl. Phys. Lett. 87, 071102 (2005), Fig. 3, reproduced with permission, copyright: American Institute of Physics, 2005)

 

Fig. 6 (Color online) Electric field distribution of the dipolar SP mode after source radiation. (a) single Ag/SiO2-NP, (b) three Ag/SiO2-NP, (c) seven Ag/SiO2-NP and (d) resonance intensity of Ag/SiO2-NPs along the z-axis. The Fig. 6(d) inset shows radiation spectra of Ag/SiO2-NPs from FDTD simulation (after Ref. [39] Lee-Woon Jang, Dae-Woo Jeon, Trilochan Sahoo, Dong-Seob Jo, Jin-Woo Ju, Seung-Jae Lee, Jong-Hyeob Baek, Jin-Kyu Yang, Jung-Hoon Song, Alexander Y. Polyakov, and In-Hwan Lee, Localized surface plasmon enhanced quantum efficiency of InGaN/GaN quantum wells by Ag/SiO2 nanoparticles, Opt. Express 20(3), 2116-2123 (2012), Fig. 3; reproduced by permission, copyright: Optical Society of America, 2012)  

Fig. 7 TEM images of Ag NPs and Ag/SiO2 core/shell NPs (after Ref. [41] Lee-Woon Jang , DaeWoo Jeon , Myoung Kim , Ju-Won Jeon , Alexander Y. Polyakov ,Jin-Woo Ju , Seung-Jae Lee , Jong-Hyeob Baek , Jin-Kyu Yang , and In-Hwan Lee, Investigation of Optical and Structural Stability of Localized Surface Plasmon Mediated Light-Emitting Diodes by Ag and Ag/SiO2 Nanoparticles, Adv. Funct. Mater. 22, 2728-2734 (2012); Fig. 2 (a), (b); Reproduced with permission, copyright: Wiley-VCH Verlag GmbH&Co, 2012)

 

Fig. 8 TEM images of Ag/SiO2 core/shell NPs with different shell thicknesses (A, B, C), selective electron diffraction image demonstrating single crystalline nature of Ag core (D) (after Ref. [42] Lee-Woon Jang, Trilochan Sahoo, Dae-Woo Jeon, Myoung Kim, Ju-Won Jeon, Dong-Seob Jo, Min-Kyu Kim, Yeon-Tae Yu, Alexander Y. Polyakov, and In-Hwan Lee, Quantum efficiency control of InGaN/GaN multi-quantum-well structures using Ag/SiO2 core-shell nanoparticles, Appl. Phys. Lett. 99, 251114 (2011), Fig. 1, reproduced with permission, copyright: American Institute of Physics, 2012)


 

Fig. 9 (Color online) Absorbance spectra of Ag/SiO2 core/shell NPs with shell thicknesses of 2 nm (curve A), 5 nm (curve B), 20 nm (curve C) (after Ref. [42] Lee-Woon Jang, Trilochan Sahoo, Dae-Woo Jeon, Myoung Kim, Ju-Won Jeon, Dong-Seob Jo, Min-Kyu Kim, Yeon-Tae Yu, Alexander Y. Polyakov, and In-Hwan Lee, Quantum efficiency control of InGaN/GaN multi-quantum-well structures using Ag/SiO2 core-shell nanoparticles, Appl. Phys. Lett. 99, 251114 (2011), Fig. 2, reproduced with permission, copyright: American Institute of Physics, 2012)

 

Fig. 10 TEM image of Ag NPs formed in the p-GaN contact layer close to the MQW region by annealing (after Ref. [44] Chu-Young Cho, Min-KiKwon, Sang-Jun Lee, Sang-Heon Han, Jang-Won Kang, Se-EunKang, Dong-Yul Lee and Seong-Ju Park, Surface plasmonenhanced light-emitting diodes using silver nanoparticles embedded in p-GaN, Nanotechnology 21 (2010) 205201, Fig. 1 (b); reproduced by permission, copyright: IOP, 2010)

 

Fig. 11 (Color online) (a) schematic drawing of MQW GaN/InGaN LED with a layer of Ag NPs and SiO2 nanodiscs built into the p-GaN layer close to the MQW region, (b) plan-view SEM images of SiO2 nanodiscs on top of Ag NPs (After Ref. [45] Chu-Young Cho, Ki Seok Kim, Sang-Jun Lee, Min-Ki Kwon, Hyungduk Ko, Sung-Tae Kim, Gun-Young Jung, and Seong-Ju Park, Surface plasmon-enhanced light-emitting diodes with silver nanoparticles and SiO2 nano-disks embedded in p-GaN, Appl. Phys. Lett. 99, 041107 (2011), Fig. 1(a), (b); reproduced with permission, copyright: American Institute of Physics, 2011)

 

Fig. 12 Al discs prepared by photolithography on UV AlGaN LEDs (after Ref. [56] ChuYoung Cho, Yinjun Zhang, Erdem Cicek, Benjamin Rahnema, Yanbo Bai, Ryan McClintock, and Manijeh Razeghi, Surface plasmon enhanced light emission from AlGaNbased ultraviolet light-emitting diodes grown on Si (111), Appl. Phys. Lett. 102, 211110 (2013), Fig. 1(d), reproduced with permission, copyright: American Institute of Physics, 2013)

 

Fig. 13 Variation of the NSL fabricated Al NPs size and shape upon microwave heating of the PS monolayer mask increasing the diameter of polystyrene beads; the times of microwave heating are shown on each figure (after Ref. [57] Atsushi Taguchi, Yuika Saito, Koichi Watanabe, Song Yijian, and Satoshi Kawata, Tailoring plasmon resonances in the deep-ultraviolet by size-tunable fabrication of aluminum nanostructures, Appl. Phys. Lett. 101, 081110 (2012), Fig. 2 (a), (b), (c), (d); reproduced with permission, copyright: American Institute of Physics, 2012)

 

Fig. 14 (Color online) The change in the spectral position of LSP resonance in Al NPs prepared by NSL as a function NPs lateral size: comparison of experiment and modeling (after Ref. [57] Atsushi Taguchi, Yuika Saito, Koichi Watanabe, Song Yijian, and Satoshi Kawata, Tailoring plasmon resonances in the deep-ultraviolet by size-tunable fabrication of aluminum nanostructures, Appl. Phys. Lett. 101, 081110 (2012), Fig. 3(b), reproduced with permission, copyright: American Institute of Physics, 2012)

 

Fig. 15 (Color online) The temperature dependence of PL intensity for blue GaN/InGaN QW LED structure without and with Ag or Al SP layers (after Ref. [34] K. Okamoto, I. Niki, A. Shwartser, Y. Narukawa, T. Mukai, and A. Shere,. Surface-plasmon-enhanced light emitters based on InGaN quantum wells. Nat. Mater 3, 601 (2004), Fig. 4 (a), Copyright: Nature Publishing Group, 2004)

 

Fig. 16 (Color online) The room temperature PL spectra taken for blue GaN/InGaN QW LED structures without and with Ag, Al or Au SP films (after Ref. [34] K. Okamoto, I. Niki, A. Shwartser, Y. Narukawa, T. Mukai, and A. Shere,. Surface-plasmon-enhanced light emitters based on InGaN quantum wells. Nature materials 3, 601 (2004), Fig. 1 (b), Copyright: Nature Publishing Group, 2004)

 

Fig. 17 (Color online) The dependence of PL enhancement ratio for Ag, Al, Au on QW LEDs on the thickness of the spacer layer between the metal film and the QW (after Ref. [34] K. Okamoto, I. Niki, A. Shwartser, Y. Narukawa, T. Mukai, and A. Shere,. Surfaceplasmon-enhanced light emitters based on InGaN quantum wells. Nature materials 3, 601 (2004), Fig. 2 ( c), Copyright: Nature Publishing Group, 2004)


 

Fig. 18 (Color online) The dependendence of the PL enhancement ratio for Ag and Ag/SiO2 NPs on the areal density of NPs (After Ref. [41] Lee-Woon Jang , Dae-Woo Jeon , Myoung Kim , Ju-Won Jeon , Alexander Y. Polyakov, Jin-Woo Ju , Seung-Jae Lee , Jong-Hyeob Baek , Jin-Kyu Yang , and In-Hwan Lee, Investigation of Optical and Structural Stability of Localized Surface Plasmon Mediated Light-Emitting Diodes by Ag and Ag/SiO2 Nanoparticles, Adv. Funct. Mater. 2012, 22, 2728–2734 (Fig. 3(a); Copyright: Wiley-VCH Verlag 2012)



 

Fig. 19 (Color online) PL spectra of blue GaN/InGaN QW structures without and with Ag, Ni/Ag, SiN/Ag films (After Ref. [38] Dong-Ming Yeh, Chi-Feng Huang, Yen-Cheng Lu, ChengYen Chen, Tsung-Yi Tang, Jeng-Jie Huang, Kun-Ching Shen, Ying-Jay Yang, and C. C. Yang, Surface plasmon leakage in its coupling with an InGaN/GaN quantum well through an Ohmic contact, Appl. Phys. Lett. 91, 063121 (2007) , Fig. 2, Copyright: AIP 2007)


 

Photon count (arb. units)

Ref Ag/SiO2

10000

Ag

0

2

4

6

8

10

Time (ns) Fig. 20 (Color online) PL decay times for MQW structures without (reference sample) and with Ag or Ag/SiO2 NPs (After Ref. [41] Lee-Woon Jang, Dae-Woo Jeon, Myoung Kim, Ju-Won Jeon, Alexander Y. Polyakov, Jin-Woo Ju, Seung-Jae Lee, Jong-Hyeob Baek , JinKyu Yang, and In-Hwan Lee, Investigation of Optical and Structural Stability of Localized Surface Plasmon Mediated Light-Emitting Diodes by Ag and Ag/SiO2 Nanoparticles, Adv. Funct. Mater. 2012, 22, 2728–2734 (Fig. 3(c))


 

Purcell factor (Arb. units)

8 7 6 5 4 3 2 1 0

0

5

10

15

20

SiO2 shell thickness (nm) Fig. 21 Dependence of Purcell factor calculated from the ratio of decay rates on the thickness of the SiO2 shell


 

PL enhancement (Arb. units)

3.5 3.0 2.5 2.0 1.5 1.0

0

5

10

15

20

SiO2 shell thickness (nm) Fig. 22 Dependence of PL enhancement factor on the thickness of the SiO2 shell for Ag/SiO2 core/shell NPs


 

Fig. 23 (Color online) Dispersion curves of the cases of various SiO2 thicknesses, including 0, 8, 15, 20, 30 nm, and infinity Ag/SiO2 evaluated based on the four-layer structure. The dashed curve represents the light lines in GaN and SiO2. The inset shows the SPP dissipation rates under various conditions (After Ref. [61] Yen-Cheng Lu, Yung-Sheng Chen, Fu-Ji Tsai, Jyh-Yang Wang, Cheng-Hung Lin,Cheng-Yen Chen, Yean-Woei Kiang, and C. C. Yang, Improving emission enhancement in surface plasmon coupling with an InGaN/GaN quantum well by inserting a dielectric layer of low refractive index between metal and semiconductor, Appl. Phys. Lett. 94, 233113 (2009) Fig. 4; Copyright: AIP 2009


 



Fig. 24 The schematics of the LSP structure formed by locally thinning down the p-GaN cap in the GaN/InGaN MQW structure (After Ref. [62] Lee-Woon Jang, Jin-Woo Ju, Ju-Won Jeon, Dae-Woo Jeon, Jung-Hun Choi, Seung-Jae Lee, Seong-Ran Jeon, Jong-Hyeob Baek, E. Sari, H. V. Demir, Hyung-Do Yoon, Sung-Min Hwang, and In-Hwan Lee, Enhanced optical characteristics of light emitting diodes by surface Plasmon of Ag nanostructures, Proc. of SPIE Vol. 7945 794511-1 (2011), Fig. 1(b); Copyright: SPIE 2011)


 

Fig. 25 (Color online) The schematic representation and SEM imaging of the nanopillar formed by dry etching in planar p-GaN/MQW GaN/InGaN MQW structures (Figures (a), (b)), of nanopillar structure with deposited Ag NPs (Figures (c ), (d)) and with Ag/SiO2 NPs (Figures (e), (f)) (After Ref. [43] Lee-Woon Jang, Dae-Woo Jeon, Trilochan Sahoo, Alexander Y. Polyakov, Balasubramaniam Saravanakumar, Yeon-Tae Yu, Yong-Hoon Cho, Jin-Kyu Yang, and In-Hwan Lee, Energy coupling processes in InGaN/GaN nanopillar light emitting diodes embedded with Ag and Ag/SiO2 nanoparticles, J. Mater. Chem., 2012, 22, 21749), Fig. 2(a), (b), (c), (d), (e ), (f)) Copyright: Royal Society of Chemistry, 2012)
  

Fig. 26 (Color online) PL changes induced in GaN/InGaN MQW nanopillar structures by deposition of Ag and Ag/SiO2 NPs (After Ref. [43] Lee-Woon Jang, Dae-Woo Jeon, Trilochan Sahoo, Alexander Y. Polyakov, Balasubramaniam Saravanakumar, Yeon-Tae Yu, Yong-Hoon Cho, Jin-Kyu Yang, and In-Hwan Lee, Energy coupling processes in InGaN/GaN nanopillar light emitting diodes embedded with Ag and Ag/SiO2 nanoparticles, J. Mater. Chem., 2012, 22, 21749), Fig. 3; Copyright: Royal Society of Chemistry, 2012)


  

Fig. 27 (Color online) PL decay curves in GaN/InGaN MQW nanopillar structures without and with Ag or Ag/SiO2 NPs (After Ref. [43] Lee-Woon Jang, Dae-Woo Jeon, Trilochan Sahoo, Alexander Y. Polyakov, Balasubramaniam Saravanakumar, Yeon-Tae Yu, YongHoon Cho, Jin-Kyu Yang, and In-Hwan Lee, Energy coupling processes in InGaN/GaN nanopillar light emitting diodes embedded with Ag and Ag/SiO2 nanoparticles, J. Mater. Chem., 2012, 22, 21749), Fig. 4; Copyright: Royal Society of Chemistry, 2012)



 

Fig. 28 (Color online) PL intensity evolution with time for Ag/MQW LED or Ag/SiO2/MQW LED structures heated to 100 oC in air; also shown is the schematics of the experimental setup (after Ref. [41] Lee-Woon Jang , Dae-Woo Jeon , Myoung Kim , JuWon Jeon , Alexander Y. Polyakov, Jin-Woo Ju , Seung-Jae Lee , Jong-Hyeob Baek , JinKyu Yang , and In-Hwan Lee, Investigation of Optical and Structural Stability of Localized Surface Plasmon Mediated Light-Emitting Diodes by Ag and Ag/SiO2 Nanoparticles, Adv. Funct. Mater. 2012, 22, 2728–2734; Fig. 5; Copyright: Wiley-VCH, 2012




 

Fig. 29 (Color online) Calculated EQE dependence on driving current for a blue QW LED structure without (reference sample) and with SP layer with Purcell factors assumed to be equal to 2 and 10 (After Ref. [68] Wei Yang, Yongfa He, Lei Liu, and Xiaodong Hu, Practicable alleviation of efficiency droop effect using surface plasmon coupling in GaNbased light emitting diodes, Appl. Phys. Lett. 102, 241111 (2013); Fig. 2(b); Copyright: AIP Publishing LLC, 2013)



 

Fig. 30 (Color online) Effect of varying the efficiency of LSP energy conversion into light (varying the r value in the figure) on the stating IQE values and the efficiency droop in blue GaN/InGaN QW LEDs(After Ref. [68] Wei Yang, Yongfa He, Lei Liu, and Xiaodong Hu, Practicable alleviation of efficiency droop effect using surface plasmon coupling in GaNbased light emitting diodes, Appl. Phys. Lett. 102, 241111 (2013); Fig. 3; Copyright: AIP Publishing LLC, 2013)



 

Fig. 31 Design of the SP-enhanced LED structure, showing the mesa structure with the pGaN/p-AlGaN/single quantum well (SQW) stack, the Ag SP layer sitting on SiN dielectric layer, the current spreading layer (CSL), Ni/Au ohmic contact to p-GaN, Ti/Al/Ti/Au ohmic contact to n-GaN (after Ref. [37] Dong-Ming Yeh, Chi-Feng Huang, Cheng-Yen Chen, Yen-Cheng Lu, and C. C. Yang, Surface plasmon coupling effect in an InGaN/GaN singlequantum-well light-emitting diode, Appl. Phys. Lett. 91, 171103 (2007), Fig. 1, Copyright: AIP 2007)



 

 

Fig. 32 (Color online) Ohmic contacts pads and waveguiding grating design on SP enhanced LED (after Ref. [78] Chih-Feng Lu, Che-Hao Liao, Chih-Yen Chen, Chieh Hsieh, YeanWoei Kiang, and C. C. Yang, Reduction in the efficiency droop effect of a light-emitting diode through surface plasmon coupling, Appl. Phys. Lett. 96, 261104 (2010), Fig. 1, Copyright: AIP 2010)



 



Fig. 33 (Color online) Schematics of the Ag NP LSPs incorporation into GaN/InGaN MQW LED structure below the MQW region (After Ref. [36] M. K. Kwon, J. Y. Kim, B. H. Kim, I. K. Park, C. Y. Cho, C. C. Byeon, and S. J. Park, Surface-plasmon-enhanced light-emitting diodes, Adv. Mater. 20, 1253 (2008), Fig. 1, Copyright Wiley-VCH, 2008)



Fig. 34 (Color online) The schematic representation of the process flow in fabrication of LSP LEDs with Ag NPs embedded below the QW region and respective SEM images (after Ref. [81]

 

Lee-Woon Jang, Jin-Woo Ju, Dae-Woo Jeon, Jae-Woo Park, A. Y. Polyakov, Seung-jae Lee, Jong-Hyeob Baek, Song-Mei Lee, Yong-Hoon Cho, and In-Hwan Lee, Enhanced light output of InGaN/GaN blue light emitting diodes with Ag nano-particles embedded in nano-needle layer, Opt. Express 20(6), 6036-6040 (2012), Fig. 1, Copyright Optical Society of America (OSA) 2012)



 Fig. 35 (Color online) Schematics of LED structure with Ag NPs embedded in p-GaN (After Ref. [44] Chu-Young Cho, Min-Ki Kwon, Sang-Jun Lee, Sang-Heon Han, Jang-Won Kang, SeEun Kang, Dong-Yul Lee and Seong-Ju Park, Surface plasmon-enhanced light-emitting diodes using silver nanoparticles embedded in p-GaN, Nanotechnology 21 (2010) 205201, Fig. 1(a), Copyright: IOP Publishing Ltd., 2010)



 

 Fig. 36 (Color online) I-V characteristics of LEDs without and with Ag nanoparticles or Ag/SiO2 microdiscs nanoparticles; the inset shows the blow-up of the reverse current region (After Ref. [45] Chu-Young Cho, Ki Seok Kim, Sang-Jun Lee, Min-Ki Kwon, Hyungduk Ko, Sung-Tae Kim, Gun-Young Jung, and Seong-Ju Park, Surface plasmon-enhanced light-emitting diodes with silver nanoparticles and SiO2 nano-disks embedded in p-GaN, Appl. Phys. Lett. 99, 041107 (2011), Fig. 4(a), Copyright: AIP 2011)



 



Fig. 37 (Color online) Calculated effect of SP coupling on EQE and EQE droop in green LED structures (After Ref. [68] Wei Yang, Yongfa He, Lei Liu, and Xiaodong Hu, Practicable alleviation of efficiency droop effect using surface plasmon coupling in GaN-based light emitting diodes, Appl. Phys. Lett. 102, 241111 (2013), Fig. 4, Copyright: AIP, 2013)



 



Fig. 38 (Color online) The combined effects of increased laser pumping power and Ag SP coupling on the blue shift and intensity of the QW peak for GaN/InGaN QW LED (After Ref. [88] Cheng-Yen Chen, Yen-Cheng Lu, Dong-Ming Yeh, and C. C. Yang, Influence of the quantum-confined Stark effect in an InGaN/GaN quantum well on its coupling with surface plasmon for light emission enhancement, Appl. Phys. Lett. 90, 183114 (2007), Fig. 1; Copyright: AIP, 2007)


 

Capacitance (pF)

1750

(a)

1500

10 kHz 1 kHz

1250

0.1 kHz

1000

100 kHz

750

Ec-0.41 eV

G/ω (pF)

400

200

0 50

100

150

200

250

300

350

400

450

Temperature (K)

DLTS signal (Arb. units)

0.8

(b)

Ec-1 eV

0.6

0.4

0.2

0.0

100

150

200

250

300

350

400

Temperature (K) Fig. 39 (Color online) (a) Admittance spectra of non-polar GaN/InGaN QW taken at various frequencies shown near respective curves; (b) DLTS spectrum obtained for this QW structure for reverse bias -1V and forward bias pulse of 1V for time windows 500 ms/ 5000 ms



 

PL intensity (arb. units)

40000

Ag coated sample Control sample

30000

20000

10000

0

450

500

550

600

Wavelength (nm) Fig. 40 (Color online) PL spectra of the GaN/InGaN non-polar QW without and with Ag NPs



Fig. 41 Schematics of the Au grating on top of the green GaN/InGaN QW LED structure and respective PL enhancement as a function of the grating period (After Ref. [47] K. Okamoto, I. Niki, A. Shwartser, G. Maltezos, Y. Narukawa, T. Mukai, Y. Kawakami, and A. Sherer, Surface


 

plasmon enhanced bright light emission from InGaN/GaN, Phys. Stat. Sol. (a) 204(6), 2103-2107 (2007), Fig. 3; Copyright: Wiley-VCH Verlag, 2007)



Fig. 42 (Color online) The PL enhancement in green GaN/InGaN MQW LEDs with Au NP embedded into p-GaN close to the QW; the inset shows the extinction cross section of the Au NPs on GaN (After Ref. [46] Chu-Young Cho, Sang-Jun Lee, Jung-Hoon Song, Sang-Hyun Hong, Song-Mae Lee, Yong-Hoon Cho, and Seong-Ju Park, Enhanced optical output power of green light-emitting diodes by surface plasmon of gold nanoparticles, Appl. Phys. Lett. 98, 051106 (2011), Fig. 2(b), Copyright: AIP 2011)


 



Fig. 43 (Color online) IV characteristics of the green GaN/InGaN MQW LEds with and without Au NPs embedded into p-GaN; the inset shows the blow-up of the reverse current region (After Ref. [46] Chu-Young Cho, Sang-Jun Lee, Jung-Hoon Song, Sang-Hyun Hong, Song-Mae Lee, Yong-Hoon Cho, and Seong-Ju Park, Enhanced optical output power of green light-emitting diodes by surface plasmon of gold nanoparticles, Appl. Phys. Lett. 98, 051106 (2011), Fig. 3, Copyright: AIP 2011)


 

 

Fig. 44 (Color online) A schematic illustration of the SP enhanced deep-UV LED (After Ref. [52] Na Gao, Kai Huang, Jinchai Li, Shuping Li, Xu Yang, Junyong Kang, Surface-plasmonenhanced deep-UV light emitting diodes based on AlGaN multi-quantum wells, Sci. Rep. 2, 816 (2012), Fig. 1, Copyright: Nature publishing group 2012)




 



Fig. 45 (Color online) Schematic diagram of the SP-enhanced AlGaN-based UV LED structure with embedded Al layer (After Ref. [56] C. Y. Cho, Y. Zhang, E. Cicek, B. Rahnema, Y. Bai, R. McClintock, M. Razeghi, Appl. Phys. Lett. 102, 211110 (2013), Fig. 1 (a), Copyright: AIP 2013)


 



Fig. 46 (Color online) Schematic diagram of the LSP-enhanced NUV-LEDs with metal NPs in pGaN layer. SEM images of (b) Ag NPs and (c) Pt NPs on the 20 nm-thick p-GaN spacer layer after a rapid thermal annealing process (After Sang-Hyun Hong, Chu-Young Cho, Sang-Jun Lee, Sang-Youp Yim, Wantae Lim, Sung-Tae Kim, and Seong-Ju Park, Opt. Express 21, 31383144 (2013), Fig. 1, Copyright: OSA 2013)


 



Fig. 47 (Color online) PL spectra at 10 K and 300 K of the NUV-LEDs; (a) without metal NPs, (b) with Ag NPs, and (c) with Pt NPs. (d) PL enhancement ratio of PL intensity of the NUVLEDs with Ag and Pt NPs to that of the NUV-LED without metal NPs. The inset shows the absorbance of Ag and Pt NPs as a function of wavelength (After Sang-Hyun Hong, Chu-Young Cho, Sang-Jun Lee, Sang-Youp Yim, Wantae Lim, Sung-Tae Kim, and Seong-Ju Park, Opt. Express 21, 3138-3144 (2013), Fig. 2, Copyright: OSA 2013) 


  



Fig. 48 Dependence of the luminescence enhancement by LSPs produced by Ag nanospheres

array with the the sphere radius a and the distance between the spheres R as a function of a for three different values of the starting QW luminescence efficiency hrad and the distance between the QW and the Ag NPs layer d=10 nm; also shown is the optimal spacing between the spheres Ropt (After Ref. [113], Fig. 3, Copyright: American Institute of Physics 2008)


  



Fig. 49 Dependence of LSP enhancement by the array of Ag NPs as a function of the starting

radiation efficiency hrad for three different values of the spacing between the the QW and the Ag NPs d (After Ref. [113], Fig. 4, Copyright: American Institute of Physics 2008)


 

Fig. 50 (Color online) Calculated dependence of the extinction spectra of Ag NP film on NP

spacing for the fixed NP diameter of 8 nm (After Ref. [114], Fig. 4, Copyright: EDP Sciences, Societ`a Italiana di Fisica, Springer-Verlag 2013)



 

Fig. 51 (Color online) Calculated dependence of the extinction spectra of Ag NP film on NP

diameter for the fixed NP spacing of 10 nm (after Ref. [114], Fig. 7, Copyright: EDP Sciences, Societ`a Italiana di Fisica, Springer-Verlag 2013)




 

Highlights

• Enhanced performances of InGaN/GaN LEDs by localized surface plasmons are reviewed

• Issues of choosing suitable metal nanoparticles in the given spectral range are discussed

• Practical issues specific to different spectral ranges (blue, green, UV) are described

• Ag is the best material for blue, Au is best for green, and Al is most suitable for the UV region

• Core/shell Ag/SiO2 nanoparticles have pivotal advantages in stability over Ag nanoparticles 


 

In-Hwan Lee received a Ph.D. degree in materials science and engineering from Korea University, Korea, in 1997. During 1997–1999, he was a postdoctoral fellow at the Northwestern University. He joined Samsung Advanced Institute of Technology, where he led an epitaxial team and developed InGaN/GaN violet LDs. Since 2002, he has been a faculty member in the School of Advanced Materials Engineering, Chonbuk National University, Korea. With the sabbatical grant from LG foundation, he was at Yale University during 2008–2009. His current research focuses on the development of nanotechnologyinspired novel optoelectronic devices including LEDs, photovoltaic devices, and sensors. He has authored or coauthored over 180 peer-reviewed research articles in major scientific journals, and presented over 50 invited seminars and talks around the world, and holds over 20 patents at various stages of the process.

Dr. Lee-Woon Jang is a post-doc fellow at Department of Mechanical Engineering at Texas Tech University. His research focuses on detecting hazardous chemical complexes by capillary electrophoresis and microfluidic system. During Ph. D. study at Chonbuk National University, he worked on localized surface plasmon mediated light-emitting diodes for enhancement in emission efficiency. His research interests are in the convergence and applications of capillary electrophoresis, nanoparticle synthesis and light-emitting diodes.

Dr. Alexander Y. Polyakov was born in Moscow in 1951, graduated from the Semiconductor Materials Science Department of Moscow Institute of Steel and Alloys in 1973, got the equivalent of PhD in Semiconductor Devices Engineering in 1982. Worked on various aspects of semiconductor physics including radiation defects, defects in conducting and semi-insulating III-V materials, hydrogen passivation effects in semi-conductors, growth and properties of heterojunctions, quantum wells, superlattices in III-V and II-VI materials, defects in wide-bandgap materials and devices, properties of localized surface plasmons and nanopillar structures in III-Nitrides. He is the author or co-author of more than 300 papers and more than 100 talks at conferences, several monographs, multiple invited chapters in books on various aspects of semiconductor materials science and many invited talks at international conferences. He permanently works in Moscow, but spent extended periods of time doing research in the USA and recently in Korea.