A metal grating on top of a light-emitting diode (LED) with a designed grating period for compensating the momentum mismatch can enhance the surface plasmon polariton (SPP) coupling effect with the quantum wells (QWs) to improve LED performance. Here, we demonstrate the experimental results showing that the induced localized surface plasmon (LSP) resonance on such a metal grating can dominate the QW coupling effect for improving LED performance, particularly when grating ridge height is large. The finding is illustrated by fabricating Ag gratings on single-QW, green-emitting LEDs of different p-type thicknesses with varied grating ridge height and width such that the distance between the grating ridge tip and the QW can be controlled. Reflection spectra of the Ag grating structures are measured and simulated to identify the SPP or LSP resonance behaviors at the QW emission wavelength. The measured results of LED performances show that in the LED samples under study, both SPP and LSP couplings can lead to significant enhancements of internal quantum efficiency and electroluminescence intensity. At the designated QW emission wavelength, with a grating period theoretically designed for momentum matching, the LSP coupling effect is stronger, when compared with SPP coupling.
© 2018 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
Various metal nanostructures have been fabricated on the top of a light-emitting diode (LED) for generating surface plasmon polariton (SPP) or localized surface plasmon (LSP) to couple with the quantum wells (QWs) of the LED for improving LED performances, including the enhancements of internal quantum efficiency (IQE) and electroluminescence (EL) intensity [1–6], the suppression of the efficiency droop effect [7–10], the increase of LED modulation bandwidth [8–10], the generation of polarized LED output [11–15], and the enhancement of light extraction efficiency . Also, theoretical/numerical studies have been undertaken to build models for predicting the behaviors of surface plasmon (SP) coupling with radiating dipoles [17–22]. Among different surface metal nanostructures, a metal grating on the top of an LED has been considered for achieving SP coupling because both SPP and LSP can be induced on such a structure for coupling with the QWs. Although the near-field SP coupling of a radiating dipole in a QW does not require a momentum compensation scheme, like plane-wave excitation of SPP, the grating nanostructure can help in enhancing the radiation of the SP-QW coupled energy. Besides, with a grating structure, the interference of the counter-propagating SPP modes at zero momentum can form a localized resonance feature for producing SP-QW coupling. The standing-wave nature results in the localized resonance behavior. Although its resonance behavior is similar to an LSP mode, its mode field distribution can cover the whole range of a grating period such that its coupling range in the lateral dimension can be larger than that of an LSP mode, whose mode field is strongly localized to a corner of a grating ridge/groove. The LSP coupling may become important when the grating ridge height is large. In such a structure, an LSP mode can be induced with its mode field distributed around the ridge tip and hence close to the QWs for strong SP coupling. Such a grating structure is particularly useful for an LED with a thick p-type layer.
Although SP-coupled LEDs with surface metal gratings have been widely demonstrated [11–16], the roles of SPP and LSP in such a device have never been well investigated. In particular, whether SPP or LSP coupling is more effective in improving LED performance is an important issue for optimizing the parameters of a surface metal grating structure, including grating period, metal ridge height, width, and shape. When grating ridge height is small (large), SPP (LSP) resonance is generally stronger than LSP (SPP) resonance. The design of a large or small grating ridge height depends on the thickness of the p-type layer in an LED. The p-type thickness is related to device resistance. Therefore, whether SPP or LSP coupling is preferred for effectively enhancing LED performances also depends on the LED structure. For matching the QW emission wavelength with an SPP or LSP feature, whichever is more effective in producing a coupling process, the relations between those grating parameters and the SPP and LSP resonance and coupling behaviors need to be well studied. In particular, the SP coupling effects under the conditions of fixing one of the grating-structure parameters while varying others need to be investigated.
In this paper, by fabricating Ag gratings on green-emitting LEDs of different p-type thicknesses with a fixed grating period, but varied metal ridge height and width, we can excite various SPP and LSP modes on the gratings for observing the different behaviors of SP coupling. By comparing the experimentally measured and numerically simulated results of reflection spectra, we can identify the features of SPP and LSP resonances. From the measurement results of LED performances, we find that both LSP and SPP couplings can result in significant improvements of LED performances, including the increases of IQE and EL intensity, and the reduction the efficiency droop effect. Among the samples under study with the designated QW emission wavelength and designed grating period, the SPP coupling effects are not as strong as those of LSP coupling. In section 2 of this paper, the designations, structures, and fabrication procedures of the LED samples under study are described. The optical characterization results, including internal quantum efficiency, time-resolved photoluminescence, and reflection, are presented in section 3. Then, the simulation results of reflection spectra and their comparisons with the corresponding experimental data are shown in section 4. Next, the LED performances are reported in section 5. More discussions about the results are made in section 6. Finally, the conclusions are drawn in chapter 7.
2. Sample designations and fabrication procedures
Three LED epitaxial structures (structures A-C) are prepared with metalorganic chemical vapor deposition on c-plane sapphire substrate. In each epitaxial structure, as schematically shown in Fig. 1, after a thin un-doped GaN buffer layer, an n-GaN layer of ~2 μm and a single InGaN/GaN QW structure (emission wavelength around 515 nm) with ~3 nm in well thickness (grown at 680 °C) and ~15 nm in the thickness of the lower barrier (grown at 830 °C) is deposited. The growth thickness of the upper barrier layer is ~20 nm. Before the growth of the p-AlGaN electron-blocking layer (EBL) of ~20% Al, an Mg pre-flow process is applied by pausing Ga and Al supplies with Mg flow at 220 sccm in Cp2Mg flow rate for 5 min [8, 23]. After the Mg pre-flow process, the p-AlGaN EBL of ~18 nm in thickness is grown with continuing Mg supply. Then, the Cp2Mg flow rate is increased to 280 sccm for the growth of a p-GaN layer. Finally, a 10-nm p+-GaN layer is deposited with 800 sccm in Cp2Mg flow rate. In LED epitaxial structures A-C, the p-GaN layer thicknesses are 50, 30, and 10 nm, respectively. The substrate temperature during the Mg pre-flow process and the growths of p-AlGaN, p-GaN, and p+-GaN layers is maintained at 970 °C. This temperature is high enough to thermally anneal the InGaN/GaN QW structure for changing its IQE [8, 24]. Because the p-GaN layer thickness varies among the three LED epitaxial structures, their high-temperature growth durations are different. The high-temperature durations of epitaxial structures A-C are 881, 803, and 726 sec, respectively. It is noted that during the Mg pre-flow process, the residual NH3, which provides us with the nitrogen source for growing GaN or InGaN, is dissolved to produce hydrogen. Hydrogen can back-etch the GaN upper quantum barrier during the Mg pre-flow process [8, 23, 25]. Therefore, the thickness of the upper quantum barrier is estimated to be ~15 nm after the complete growth of an LED epitaxial structure, i.e., GaN of ~5 nm in thickness is back-etched.
With each LED epitaxial structure, three kinds of surface nanostructure are fabricated for comparing the SP coupling effects of various Ag grating geometries. Without any Ag nanostructure at the top, the LEDs are designated as samples A-R, B-R, and C-R based on epitaxial structures A, B, and C, respectively. By depositing an Ag thin film of 150 nm in thickness on top of each epitaxial structure, we obtain LED samples A-C, B-C, and C-C. The grating structures are formed with electron-beam lithography and dry etching processes. After electron-beam writing to process a photoresist, the p-GaN layer is etched through an inductively coupled plasma reactive ion etching (ICP RIE) process to form grating grooves. Then, Ag is deposited on top to fill in the grooves for forming metal ridges and cover the whole top. The thickness of the Ag thin film covering the LED top is also ~150 nm. The grating period is fixed at Λ = 130 nm, which is designed for compensating momentum of SPP at the QW emission wavelength, i.e., 515 nm. Various grating groove depths and widths, i.e., Ag ridge heights and widths, respectively, are fabricated for comparing their SP coupling behaviors. In Table 1, we show the designations and grating parameters of various samples under study. The QW depths from the epitaxial tops of structures A-C are 93, 73, and 53 nm, respectively. Here, the Ag ridge height, h, ridge width, w, and the distance between the QW and Ag ridge tip, d, are defined in Fig. 1, which also schematically shows the structure of an LED sample with an Ag grating of Λ in period.
Figures 2(a) and 2(b) show the plan-view scanning electron microscopy (SEM) and tilted atomic force microscopy (AFM) images, respectively, of the grating structure before Ag deposition in sample B-1, which has the groove width of 43 nm and groove depth of 12 nm. Among those grating LED samples, samples A-1, B-2, C-1, C-2, and C-3 have about the same distance, d, between the Ag ridge tip and the QW. However, sample A-1, B-2, and C-1/C-2/C-3 have quite different ridge heights. Because the SP resonance can be dominated by LSP on a grating of a large ridge height, their comparisons in LED performance can provide us with the information about the dominating SP category (SPP or LSP) for QW coupling. Samples C-1, C-2, and C-3 have about the same ridge height and about the same distance between the Ag ridge tip and the QW, but different ridge widths. Their comparisons can provide us with the information about the effects of different ridge widths. Samples B-1 and C-2 have about the same ridge height and about the same ridge width, but quite different distances, d, between the Ag ridge tip and the QW. Their comparison can provide us with the information about the coupling effects of different distances between QW and metal nanostructure.
3. Optical characterization results
The IQEs and photoluminescence (PL) decay times of all the samples are measured for comparison. An IQE of a sample is obtained by taking the ratio of the integrated PL intensity at 300 K over that at 10 K. The temperature-dependent PL measurement is excited by a 406-nm laser at 6 mW in power. The PL decay time is obtained by exponentially fitting the PL intensity decay profile in a time-resolved PL measurement at room temperature. The time-resolved PL measurement is excited by the second-harmonic of a 780-nm femtosecond Ti:sapphire laser. Figure 3 shows the time-resolved PL decay profiles of all the samples under study. In columns 2 and 3 of Table 2, we show the IQEs and PL decay times, respectively, of the LED samples under study. The numbers inside the parentheses in column 2 of Table 2 show the IQE ratios of the samples with respect to their individual reference samples, i.e., samples A-R, B-R, and C-R. The IQEs of samples A-R, B-R, and C-R are 13.3, 14.6, and 16.0%, respectively. The increasing IQE with decreasing p-GaN thickness is due to the decreased high-temperature (970 °C) growth duration. The reduced thermal annealing duration of the QW leads to a higher IQE. In each epitaxial structure, by covering the LED with an Ag layer, the stronger reflection of the excitation laser increases the excitation intensity at the QW and hence enhances the IQE. A 6-8% enhancement of IQE can be observed in comparing sample A-C with A-R, sample B-C with B-R, or sample C-C with C-R. With grating structures at the tops of the LED samples, their IQEs are more or less further increased. The IQE enhancements of samples A-1, B-2, and C-3 are higher than 30%. The PL decay times of those samples are essentially inversely proportional to their IQEs. When IQE is significantly increased in a sample, its PL decay time is significantly reduced.
Polarization-dependent reflection spectra can provide us with the SP resonance behaviors of the formed Ag grating structures. It is noted that grating induced SP resonance can occur only when the excitation polarization has a component across the grating ridges, i.e., along the x-axis. Two incidence plane and polarization combinations can lead to such excitations, including the y-z-plane incidence with x-axis polarization and the x-z-plane incidence with x-z-plane polarization (see the definition of the coordinate system in Fig. 1). The former and later excitation conditions will be referred to as the “y-z excitation” and “x-z excitation”, respectively, for simplicity. Figures 4(a) and 4(b) [4(c) and 4(d)] show the reflection spectra of sample A-1 (B-1) under the y-z and x-z excitation conditions, respectively, when the incident angles are 15, 25, 45, and 65 degrees (incidence angle in air from the sapphire side). The incident angle is defined with respect to the z-axis. The incident angles of 15, 25, 45, and 65 degrees in air correspond to those of 6.2, 10.1, 17.1, and 22.2 degrees, respectively, in GaN, which are shown within the parentheses in the labels of Figs. 4(a)-4(d). Figures 5(a)-5(d) [6(a)-6(d)] show the results similar to Figs. 4(a)-4(d), respectively, for samples B-2 and C-1 (C-2 and C-3). With the x-z excitation of non-zero incident angle, propagating SPP can be excited besides LSP. With any incident angle in the y-z excitation and zero incident angle in the x-z excitation, counter-propagating SPP pairs can be excited besides LSP. A counter-propagating SPP pair, which forms a standing wave, behaves like an LSP mode. Propagating SPP modes can be observed only when the incident angle is nonzero under the x-z excitation. The depression features in Figs. 4-6 indicate the SP resonance modes, either SPP or LSP. The vertical dashed lines in those figures indicate the spectral location of the QW emission around 515 nm. The behaviors of reflection spectra of the LED samples under study will be further discussed together with the simulation results of plane-wave excitation described in the next section.
4. Simulation results and comparisons between experimental data
The grating period of 130 nm is chosen based on the evaluation shown in Fig. 7, in which the theoretical dispersion curve of SPP at a smooth Ag/GaN interface is depicted as the red curve. The two slant lines labeled by 90° correspond to the light lines. The horizontal dashed line indicates the QW emission wavelength at 515 nm. The angles shown at the top indicate the slant lines of different plane-wave incidence angles in GaN. The SPP dispersion curve of a flat interface shown as the red curve in Fig. 7 lies outside the light cone, which is formed by the two light lines, indicating that without a momentum compensation scheme, a plane wave cannot interact with SPP induced at the interface. The dispersion curve of an LSP mode is a horizontal line in a coordinate system like Fig. 7 . In other words, An LSP mode has a fixed energy with the wave vector covering the whole angle range such that part of its dispersion curve overlaps with the light cone for interacting with a light field, either near or far field. With the grating structure of Λ = 130 nm in period, the dispersion curve is left-shifted by 2π/Λ such that the new dispersion curve intersects with the vertical axis (0° incidence) at the photon energy of QW emission (515 nm in wavelength). Therefore, in such a grating structure, as the incidence angle increases, the −1-order SPP resonance feature red shifts, as indicated by the variation trend of the dot positions at the intersections of the shifted dispersion curve and the slant lines of different incidence angles. The intersections of the shifted dispersion curve with the slant lines right to the vertical axis correspond to + 1-order SPP modes. However, it is difficult to observe those SPP modes because they are close to the turning point and have high attenuation constants. At the intersection point between the shifted dispersion curve and the vertical axis, a counter-propagating SPP pair can be excited, whose interference produces a localized resonance feature.
Three-dimensional numerical simulations are conducted with the commercial software COMSOL, which is based on the finite-element method. For simulation, the computation domain is chosen to be a rectangular parallelepiped consisting of an upper half-space of Ag (including the ridges) and a lower half-space of GaN (excluding the ridges). To simulate the grating structure, the size of the computation domain along the x-axis is set to be a grating period, Λ, with the Bloch periodic boundary condition imposed at its two ends. The size of the computation domain along the y-axis is chosen to be 100 nm, also with the Bloch periodic boundary condition imposed at its two ends. The computation domain along the z-axis covers 100 nm in Ag and 200 nm in GaN. For simulating the infinite extensions in the + z- and -z-direction, two perfectly matched layers are placed at the top and bottom of the computation domain. By using COMSOL, the total electromagnetic field is numerically calculated with an incident plane wave. Then, we can compute the power absorbed in the Ag region. An absorption peak in spectrum corresponds to an SP resonance feature. The refractive index of GaN is fixed at 2.4. The wavelength-dependent dielectric constant of Ag is obtained from experiment reported in literature .
Figures 8(a) and 8(b) [8(c) and 8(d)] show the simulated absorption spectra with the grating geometry of sample A-1 (B-2) for the cases of y-z and x-z excitations, respectively, when plane waves of 0, 15, 22.5, 30, 45, 60, and 75 degrees in incident angle are applied from GaN. Figures 9(a) and 9(b) [9(c) and 9(d)] show the results similar to those in Figs. 8(a) and 8(b), respectively, with the grating geometry of sample C-1 (C-3). Figures 10(a1)-10(a6) show the simulated charge distributions at nT/6 in time (n = 0-5, respectively) in an electromagnetic oscillation period, T, for the feature around 565 nm with 0 degree in incidence angle of sample A-1 under the y-z excitation, as shown in Fig. 8(a). Here, blue and red colors correspond to the opposite charges. Figures 10(b1)-10(b6) show the results similar to Figs. 10(a1)-10(a6) for the feature around 665 nm with 45 degrees in incident angle of sample C-1 under the x-z excitation, as shown in Fig. 9(b). The local charge oscillation shown in Figs. 10(a1)-10(a6) shows the behavior of either an LSP mode or the feature of counter-propagating SPP modes. That shown in Figs. 10(b1)-10(b6) demonstrates the behavior of an SPP mode.
As shown in Fig. 9(c) for simulating sample C-3, with a small grating ridge height (11 nm) and a medium size for ridge width (65 nm), one can see two major peaks in the cases of different incident angles when the y-z excitation condition is applied. The short-wavelength peak around 500 nm corresponds to a mixed feature of the local resonance caused by counter-propagating SPPs and an LSP resonance mode. The long-wavelength peak corresponds to another LSP mode. When the x-z excitation condition is applied, as shown in Fig. 9(d), the extra peaks correspond to the generated SPP modes. As the grating ridge width is reduced to 20 nm in sample C-1, as shown in Figs. 9(a) and 9(b), the peak around 500 nm in Fig. 9(c) becomes a multi-peak broad feature below 500 nm. When the ridge height is increased to 34 nm and the ridge width is also quite small (39 nm) in sample B-2, as shown in Fig. 8(c), three local resonance features are generated. When the ridge height is further increased to 51 nm in sample A-1, as shown in Fig. 8(a), essentially two local resonance features can be observed. As shown in Figs. 8(b) and 8(d), the additions of SPP modes when the x-z excitation is applied lead to complicated SP resonance behaviors.
As mentioned earlier, the incident angles labeled in Figs. 4-6 refer to those in air and the incident angles labeled in Figs. 8 and 9 refer to those in GaN. The reflection spectra obtained from experiment in Figs. 4-6 correspond to simulation results between 0 and 22.2 degrees in Figs. 8 and 9. As shown in Fig. 4(a), with the y-z excitation, one can see only one SP resonance feature around 430 nm. This feature corresponds to the short-wavelength peak in Fig. 8(a). The uncertainties of several factors, including the ridge geometry and GaN and Ag dielectric constants, can make the resonance peaks not exactly matched between experiment and simulation. With the x-z excitation, one more resonance feature beyond 500 nm can be observed in Fig. 4(b). This feature corresponds to the LSP-SPP mixed peak cluster beyond 500 nm in Fig. 8(b). In this peak cluster, LSP resonance dominates and the major peak blue shifts with increasing incident angle that is consistent with the experimental result shown in Fig. 4(b). The long-wavelength LSP resonance feature in Fig. 8(a) is not seen in the experimental result. This is so because the charge oscillation of this localized resonance feature mainly distributes around the boundaries between a ridge and the connecting valleys, as shown in Figs. 10(a1)-10(a6). Under the condition of the y-z excitation, most scattered light from this corner cannot be collected in the specular-reflection direction. In our experiment, the reflection spectrum is obtained with specular-reflection measurement such that the absorption behavior of the long-wavelength localized resonance feature in Fig. 8(a) cannot be observed. The same argument can be applied to the comparisons between experimental results and simulation data for samples B-2, C-1, and C-3 under the y-z excitation. Under the x-z excitation in samples C-1 and C-3, the long-wavelength feature is dominated by SPP such that experimentally the reflectance minimum red-shifts with increasing incident angle. The same story can be applied to samples B-1 and C-2, for which simulation results are not shown. As long as the ridge height is small (~10 nm), SPP dominates the long-wavelength feature. It is expected that as long as there is an SP resonance feature at the QW emission wavelength, even though no reflectance depression can be seen there in Figs. 4-6, SP coupling can occur in an LED. Among the LED samples with gratings, the SP couplings with the QW in samples A-1 and B-2 are dominated by LSP, whose feature blue shifts with increasing angle. Meanwhile, the SP couplings with the QW in samples B-1, C-1, C-2, and C-3 are dominated by SPP, whose feature red shifts with increasing incident angle.
5. LED Performances
A lateral LED device of 300 μm x 300 μm in mesa size is fabricated based on each structure sample following the standard process procedures. EL intensities measured from both epitaxial and substrate sides are added together to give the output intensity of an LED device. Figure 11 shows the relation between injected current and applied voltage (I-V curves) of the samples under study. The insert shows the magnified portion in the voltage range of 2.5-4.5 V and current range of 10-20 mA. Here, one can see that the leakage current under reverse bias is negligibly small in each sample. The turn-on voltages are all around 3 V. The differential resistance levels of all the samples are listed in column 4 of Table 2. By decreasing the p-GaN thickness, the resistance is slightly increased due to the degradation of the current spreading effect in the p-type layer. By adding an Ag structure at the top, particularly fabricating an Ag grating on the top, the resistance is also slightly increased due to the relatively higher contact resistivity between Ag and p+-GaN, which is caused by the larger work function mismatch between Ag and p-GaN, when compared with that between Ni and p-GaN. During the process of grating fabrication, dry etching leads to GaN surface damage and hence even higher contact resistivity.
Figures 12(a1)-12(a3), 12(b1)-12(b4), and 12(c1)-12(c5) show the photographs of lit LEDs at 100 mA in injected current of the samples under study, as labeled. Because the Ag grating or film blocks the emitted light, the mesa portions of the devices are darker in all the samples except samples A-R, B-R, and C-R. Figure 13 shows the variations of EL intensities versus injected current (L-I curves) of all the samples under study. The EL intensities are normalized with respect to that of sample A-R at 100 mA in injected current. The normalized EL intensities of all the samples are listed in column 5 of Table 2. The numbers within the parentheses represent the normalized intensities with respect to those of individual reference samples in the three sample groups of different p-GaN thicknesses, i.e., samples A-R, B-R, and C-R. By comparing the EL intensity and IQE results, one can see the consistency of the variations between the two LED performance parameters. A sample with a higher IQE has a higher EL intensity. Those samples of high EL intensities and IQEs have certain SP resonance features at the QW emission wavelength. It is noted that a 2-3% increase of EL intensity corresponds to a 6-8% enhancement of IQE in the samples with top Ag thin films. However, with an Ag grating, the enhancement ratio of EL intensity is generally larger than that of IQE in the same sample. This result may indicate that the Ag grating can also enhance the light extraction efficiency. Figure 14 shows the variations of relative efficiency of all the samples with injected current. The efficiencies of all the samples are normalized with respect to that of sample C-3 at 10 mA. The efficiency droop range, which is defined as the decrease percentage of efficiency with respect to its maximum level of a sample, of each sample is shown in column 6 of Table 2. Here, one can see that with SP coupling, the efficiency droop range is reduced.
The reported results above are obtained based on the grating structure with the period (130 nm) designed to match the QW emission wavelength (515 nm) with SPP energy at zero momentum or vertical wave vector for maximizing the SPP coupling effect. In the samples of SPP-dominated coupling for improving LED performance (samples B-1, C-1, C-2, and C-3), the QW emission wavelength coincides with SPP depression minimum only when the incident angle is small, as shown in Figs. 4(d), 5(d), 6(b), and 6(d). In the samples of LSP-dominated coupling for improving LED performance (samples A-1 and B-2), the QW emission wavelength coincides with the LSP depression minimum when the incident angle is relatively larger. As mentioned earlier, at zero incident angle under either excitation condition, the resonance feature actually includes the localized resonance of counter-propagating SPP interference and an LSP mode. Therefore, the SP coupling results in samples B-1, C-1, C-2, and C-3 are also attributed to localized resonance feature, at least partially caused by an LSP mode. From these behaviors, it is difficult to determine whether the SPP- or LSP-dominated coupling can generally lead to a stronger coupling effect for improving LED performance. However, it is interesting to notice that with LSP-dominated coupling, the enhancement percentages of IQE and EL intensity with respect to those of the individual reference samples are generally higher than those with SPP-dominated coupling under the condition of about the same distance between Ag NPs and QW, d (among samples A-1, B-2, C-1, C-2, and C-3). This comparison result may imply that LSP-dominated coupling is more effective for LED performance improvement, when compared with SPP-dominated coupling. This speculation can be true because in SPP coupling, for a given QW emission wavelength, only a limited wave vector range of the dipole near field can match the SPP momentum and hence effectively couple with the SPP. In this situation, the coupling can be weaker, when compared with LSP coupling, in which no wave vector limitation applies. The other factor for the stronger LSP coupling effect is the near field decay in space of an LSP mode can be slower, when compared with the evanescent field decay of an SPP mode, such that the LSP mode field at a QW can be relatively stronger for stronger coupling. For LSP coupling in samples A-1 and B-2, the ridge heights, h, are larger than those of other samples. SPP modes can be clearly observed only when the ridge height is small. In an LED of a thick p-type layer, the grating ridge-height needs to be large such that the SP resonance field can be close to the QW for producing strong coupling.
It is noted that the controls of grating period, grating ridge height, width, and shape cannot be very accurate in fabricating a metal grating for achieving a designated SPP or LSP resonance wavelength. Between SPP and LSP, the spectral position of the localized resonance feature formed by counter-propagating SPP modes is determined by the grating period, which can be more easily controlled if an advanced lithography technique, such as electron-beam or nano-imprint lithography, is used. On the other hand, the spectral position of an LSP mode is strongly influenced by the grating ridge shape, which is formed through a dry etching process that is more difficult to control. Therefore, it is relatively easier to control the SPP coupling effects. Nevertheless, because of the broad-spectrum resonance nature of either SPP or LSP and the large emission spectral width of an InGaN/GaN QW, the inaccuracy tolerance range in matching the QW emission wavelength and an SPP or LSP feature is quite large.
In summary, we have demonstrated the SP coupling effects of Ag grating structures on the top of green-emitting single-QW LEDs of different p-type thicknesses. The grating period was designed for matching the QW emission wavelength with the SPP energy at zero momentum or vertical wave vector. Ag gratings of different ridge depths and widths were fabricated for exciting SP resonances dominated by either SPP or LSP and comparing their different SP-QW coupling behaviors. It was found that when the distance between grating ridge tip and the QW was about the same, at the QW emission wavelength, a grating structure of a large (small) ridge height led to LSP- (SPP-) dominated coupling and the larger (smaller) enhancements of IQE and EL intensity in LED performance. Under the conditions of the designated QW emission wavelength (515 nm) and grating period (130 nm), LSP coupling showed a more effective improvement of LED performance, when compared with SPP-dominated coupling. However, a general conclusion for this comparison requires more detailed investigation.
Ministry of Science and Technology (MOST), Taiwan, The Republic of China (grants of MOST 106-2221-E-002-163-MY3, MOST 105-2221-E-002-159-MY3, MOST 105-2622-E-002-012-CC2, and MOST 106-2221-E-002-162); US Air Force Office of Scientific Research (AFOSR) (grant AOARD-14-4105).
The authors declare that there are no conflicts of interest related to this article.
References and links
1. D. M. Yeh, C. F. Huang, C. Y. Chen, Y. C. Lu, and C. C. Yang, “Surface plasmon coupling effect in an InGaN/GaN single-quantum-well light-emitting diode,” Appl. Phys. Lett. 91(17), 171103 (2007). [CrossRef]
2. D. M. Yeh, C. F. Huang, C. Y. Chen, Y. C. Lu, and C. C. Yang, “Localized surface plasmon-induced emission enhancement of a green light-emitting diode,” Nanotechnology 19(34), 345201 (2008). [CrossRef] [PubMed]
3. 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(7), 1253–1257 (2008). [CrossRef]
4. C. Y. Cho, S. J. Lee, J. H. Song, S. H. Hong, S. M. Lee, Y. H. Cho, and S. J. Park, “Enhanced optical output power of green light-emitting diodes by surface plasmon of gold nanoparticles,” Appl. Phys. Lett. 98(5), 051106 (2011). [CrossRef]
5. C. Y. Cho, K. S. Kim, S. J. Lee, M. K. Kwon, H. Ko, S. T. Kim, G. Y. Jung, and S. J. Park, “Surface plasmon-enhanced light-emitting diodes with silver nanoparticles and SiO2 nano-disks embedded in p-GaN,” Appl. Phys. Lett. 99(4), 041107 (2011). [CrossRef]
6. Y. C. Lu, Y. S. Chen, F. J. Tsai, J. Y. Wang, C. H. Lin, C. Y. Chen, Y. W. 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(23), 233113 (2009). [CrossRef]
7. C. F. Lu, C. H. Liao, C. Y. Chen, C. Hsieh, Y. W. Kiang, and C. C. Yang, “Reduction of the efficiency droop effect of a light-emitting diode through surface plasmon coupling,” Appl. Phys. Lett. 96(26), 261104 (2010). [CrossRef]
8. C. Y. Su, C. H. Lin, Y. F. Yao, W. H. Liu, M. Y. Su, H. C. Chiang, M. C. Tsai, C. G. Tu, H. T. Chen, Y. W. Kiang, and C. C. Yang, “Dependencies of surface plasmon coupling effects on the p-GaN thickness of a thin-p-type light-emitting diode,” Opt. Express 25(18), 21526–21536 (2017). [CrossRef] [PubMed]
9. C. H. Lin, C. G. Tu, Y. F. Yao, S. H. Chen, C. Y. Su, H. T. Chen, Y. W. Kiang, and C. C. Yang, “High modulation bandwidth of a light-emitting diode with surface plasmon coupling,” IEEE Transact. Electron Dev. 63(10), 3989–3995 (2016). [CrossRef]
10. C. H. Lin, C. Y. Su, E. Zhu, Y. F. Yao, C. Hsieh, C. G. Tu, H. T. Chen, Y. W. Kiang, and C. C. Yang, “Modulation behaviors of surface plasmon coupled light-emitting diode,” Opt. Express 23(6), 8150–8161 (2015). [CrossRef] [PubMed]
11. K. C. Shen, C. Y. Chen, C. F. Huang, J. Y. Wang, Y. C. Lu, Y. W. Kiang, C. C. Yang, and Y. J. Yang, “Polarization dependent coupling of surface plasmon on a one-dimensional Ag grating with an InGaN/GaN dual-quantum-well structure,” Appl. Phys. Lett. 92(1), 013108 (2008). [CrossRef]
12. K. C. Shen, C. Y. Chen, H. L. Chen, C. F. Huang, Y. W. Kiang, C. C. Yang, and Y. J. Yang, “Enhanced and partially polarized output of a light-emitting diode with Its InGaN/GaN quantum well coupled with surface plasmons on a metal grating,” Appl. Phys. Lett. 93(23), 231111 (2008). [CrossRef]
13. K. C. Shen, C. H. Liao, Z. Y. Yu, J. Y. Wang, C. H. Lin, Y. W. Kiang, and C. C. Yang, “Effects of the intermediate SiO2 layer on polarized output of a light-emitting diode with surface plasmon coupling,” J. Appl. Phys. 108(11), 113101 (2010). [CrossRef]
14. H. Chen, H. Fu, Z. Lu, X. Huang, and Y. Zhao, “Optical properties of highly polarized InGaN light-emitting diodes modified by plasmonic metallic grating,” Opt. Express 24(10), A856–A867 (2016). [CrossRef] [PubMed]
15. G. Zhang, X. Guo, F. F. Ren, Y. Li, B. Liu, J. Ye, H. Ge, Z. Xie, R. Zhang, H. H. Tan, and C. Jagadish, “High-brightness polarized green InGaN/GaN light-emitting diode structure with Al-coated p-GaN grating,” ACS Photonics 3(10), 1912–1918 (2016). [CrossRef]
16. E. Homeyer, P. Mattila, J. Oksanen, T. Sadi, H. Nykanen, S. Suihkonen, C. Symonds, J. Tulkki, F. Tuomisto, M. Sopanen, and J. Bellessa, “Enhanced light extraction from InGaN/GaN quantum wells with silver gratings,” Appl. Phys. Lett. 102(8), 081110 (2013). [CrossRef]
17. G. Sun, J. B. Khurgin, and R. A. Soref, “Practicable enhancement of spontaneous emission using surface plasmons,” Appl. Phys. Lett. 90(11), 111107 (2007). [CrossRef]
18. J. B. Khurgin, G. Sun, and R. A. Soref, “Enhancement of luminescence efficiency using surface plasmon polaritons: Figures of merit,” J. Opt. Soc. Am. B 24(8), 1968–1980 (2007). [CrossRef]
19. G. Sun, J. B. Khurgin, and C. C. Yang, “Impact of high-order surface plasmon modes of metal nanoparticles on enhancement of optical emission,” Appl. Phys. Lett. 95(17), 171103 (2009). [CrossRef]
20. Y. Kuo, S. Y. Ting, C. H. Liao, J. J. Huang, C. Y. Chen, C. Hsieh, Y. C. Lu, C. Y. Chen, K. C. Shen, C. F. Lu, D. M. Yeh, J. Y. Wang, W. H. Chuang, Y. W. Kiang, and C. C. Yang, “Surface plasmon coupling with radiating dipole for enhancing the emission efficiency of a light-emitting diode,” Opt. Express 19(S4Suppl 4), A914–A929 (2011). [CrossRef] [PubMed]
21. G. Sun and J. B. Khurgin, “Plasmon enhancement of luminescence by metal nanoparticles,” IEEE J. Sel. Top. Quantum Electron. 17(1), 110–118 (2011). [CrossRef]
23. C. Y. Su, C. G. Tu, W. H. Liu, C. H. Lin, Y. F. Yao, H. T. Chen, Y. R. Wu, Y. W. Kiang, and C. C. Yang, “Enhancing the hole injection efficiency of a light-emitting diode by increasing Mg doping in the p-AlGaN electron blocking layer,” IEEE Trans. Electron Dev. 64(8), 3226–3233 (2017). [CrossRef]
24. C. Y. Chen, C. Hsieh, C. H. Liao, W. L. Chung, H. T. Chen, W. Cao, W. M. Chang, H. S. Chen, Y. F. Yao, S. Y. Ting, Y. W. Kiang, C. C. Yang, and X. Hu, “Effects of overgrown p-layer on the emission characteristics of the InGaN/GaN quantum wells in a high-indium light-emitting diode,” Opt. Express 20(10), 11321–11335 (2012). [CrossRef] [PubMed]
25. Y. T. Moon, Y. Fu, F. Yun, S. Dogan, M. Mikkelson, D. Johnstone, and H. Morkoç, “A study of GaN regrowth on the micro-facetted GaN template formed by in-situ thermal etching,” Phys. Status Solidi 202(5), 718–721 (2005). [CrossRef]
26. S. A. Maier, Plasmonics: Fundamentals and Applications (Springer, 2007).
27. E. D. Palik, Handbook of Optical Constants of Solids (Academic Press, 1991).