We introduce a novel approach to the seamless integration of plasmonic nanoparticle (NP) arrays into semiconductor layers and demonstrate their enhanced photoluminescence (PL) efficiency. Our approach utilizes focused ion beam-induced self-assembly of close-packed arrays of Ga NPs with tailorable NP diameters, followed by overgrowth of GaAs layers using molecular beam epitaxy. Using a combination of PL spectroscopy and electromagnetic computations, we identify a regime of Ga NP diameter and overgrown GaAs layer thickness where NP-array-enhanced absorption in GaAs leads to enhanced GaAs near-band-edge (NBE) PL efficiency, surpassing that of high-quality epitaxial GaAs layers. As the NP array depth and size are increased, the reduction in spontaneous emission rate overwhelms the NP-array-enhanced absorption, leading to a reduced NBE PL efficiency. This approach provides an opportunity to enhance the PL efficiency of a wide variety of semiconductor heterostructures.

Inorganic materials have been widely used for a variety of solid-state lighting applications.1–14 However, heat generation and non-directional light emission often lead to the need for additional thermal and optical components which complicate device packaging.1 Alternatively, plasmonic nanoparticle (NP) arrays have emerged as a promising approach to improve emission efficiency and provide flexibility in device design, especially if they could be selectively placed at specific interfaces within a device structure.1–14 To date, plasmonic materials research and device fabrication have focused nearly exclusively on silver and gold NP dispersions in two dimensions;7,8,12–14 these arrays exhibit localized surface plasmon resonances (LSPR) limited to the visible wavelength range. Furthermore, their fabrication has involved multi-step processes such as electron beam evaporation and lithography, typically limited to the front or back surface of a device structure. Meanwhile, alternative plasmonic nanomaterials including non-noble metals15–21 and metal oxides22,23 have been recently reported to exhibit widely tunable plasmon responses, without the need for costly multi-step fabrication methods. In particular, gallium (Ga) has been reported to produce plasmon responses tunable from the near infrared to the visible to the ultraviolet.15–21 Several research groups including our team recently reported a method to assemble Ga NP arrays with LSPR tunable from the infrared to the ultraviolet.16,17,24–26 Here, we introduce a novel approach to the seamless integration of non-noble metal NP arrays at buried interfaces within semiconductor layers and demonstrate their influence on the GaAs photoluminescence (PL) efficiency. This new approach provides an alternative opportunity to fabricate gain media containing buried plasmonic NP arrays, a 3D metamaterial geometry no yet demonstrated by any group.

This article is organized as follows. In Sec. II, we describe the methods for fabricating and characterizing the GaAs:Ga nanocomposites, including focused-ion-beam (FIB) irradiation, molecular-beam epitaxy (MBE) growth, spatially resolved PL, and finite-difference time-domain (FDTD) simulations. Investigations of the surface morphology and the microstructure of the nanocomposites and their influences on the photoluminescence emission are presented in Sec. III. A summary is presented in Sec. IV.

An overview of our nanofabrication process is shown in Fig. 1. In the first step, 2D close-packed Ga NP arrays are prepared via off-normal Ga+ FIB-irradiation in selected 20 × 20 μm2 regions of GaAs samples. The samples are subsequently transferred into the MBE chamber, where GaAs layers are overgrown atop the entire sample surface. Specifically, the 2D Ga NP arrays were prepared by raster scanning with off-normal incidence Ga+ irradiation, using an FEI Nova 200 Nanolab dual beam FIB system with 5 keV voltage, 0.23 nA current, 6.1 nm pitch, and 50 ns dwell time, as shown in Fig. 1(a). It has been reported that FIB irradiation at a normal angle of ion incidence (i.e., θion = 0°) leads to randomly distributed Ga NPs; meanwhile, as θion increases, due to the competition between mass-transport-induced supply and sputtering-induced loss of Ga, the NP sizes decrease and a close-packed hexagonal array of Ga NPs emerges.24,25 For this study, we used two extremes of Ga NP arrays, a random NP array with ⟨dGa⟩ = 66 nm and ⟨dGa-Ga⟩ = 108 nm (using FIB irradiation at θion = 30°) and a close-packed NP array with ⟨dGa⟩ = 40 nm and ⟨dGa-Ga⟩ = 72 nm (using FIB irradiation at θion = 52°). These two types of Ga NP arrays with varying ⟨dGa⟩ and ⟨dGa-Ga⟩ lead to similar fractional Ga NP surface coverages of 29.3% and 24.2%, respectively.

FIG. 1.

Overview of the FIB-MBE nanofabrication process for the integration of Ga NP arrays into GaAs layers and the microstructural analysis of the nanocomposite. (a) Off-normal Ga+ FIB irradiation is performed in selected 20 × 20 μm2 regions of GaAs samples. The FIB irradiation induces preferential sputtering of As atoms, leaving behind Ga-rich GaAs surfaces, from which Ga NP arrays are nucleated. (b) Following transfer into the MBE chamber, GaAs layers of various thicknesses are grown atop the entire sample surface. (c) The spotty rings in the in-situ RHEED pattern collected right after the overgrowth reveals the GaAs structure. (d) Cross-sectional BF TEM image showing FIB-patterned amorphous Ga NP array embedded between the single-crystal GaAs substrate and the MBE-overgrown polycrystalline GaAs layer. (e) The spotty rings (overlaid by dashed guidelines) in the SAED pattern, collected exclusively from the overgrown layer, reveal a polycrystalline ZB GaAs structure.

FIG. 1.

Overview of the FIB-MBE nanofabrication process for the integration of Ga NP arrays into GaAs layers and the microstructural analysis of the nanocomposite. (a) Off-normal Ga+ FIB irradiation is performed in selected 20 × 20 μm2 regions of GaAs samples. The FIB irradiation induces preferential sputtering of As atoms, leaving behind Ga-rich GaAs surfaces, from which Ga NP arrays are nucleated. (b) Following transfer into the MBE chamber, GaAs layers of various thicknesses are grown atop the entire sample surface. (c) The spotty rings in the in-situ RHEED pattern collected right after the overgrowth reveals the GaAs structure. (d) Cross-sectional BF TEM image showing FIB-patterned amorphous Ga NP array embedded between the single-crystal GaAs substrate and the MBE-overgrown polycrystalline GaAs layer. (e) The spotty rings (overlaid by dashed guidelines) in the SAED pattern, collected exclusively from the overgrown layer, reveal a polycrystalline ZB GaAs structure.

Close modal

Within 30 min of removal from the FIB chamber, each sample containing 20 × 20 μm2 regions with the NP arrays was introduced into the load-lock of a GEN II MBE chamber. Following a load-lock bakeout at 150 °C for 8 h, each sample was individually transferred to the MBE chamber and heated to 300 °C for 10 min under an As2 flux of 5.4 × 10−6 Torr, prior to overgrowth. Subsequently, 20 to 500 nm thick GaAs layers were grown at 300 °C, using a growth rate of 1 μm/h and a V/III beam-equivalent pressure ratio of 12, as shown in Fig. 1(b). At the end of each overgrowth, a reflection high-energy electron diffraction (RHEED) pattern was taken to assess the crystallinity of the overgrown layer. Following overgrowth, microstructures of the samples were examined using a JEOL 2010F transmission electron microscope (TEM) operating at 200 keV, and the surface morphology was examined using tapping-mode atomic force microscopy (AFM) with Si probes. The samples were mounted in a liquid-He flow cryostat operating at 10 K and spatially resolved PL measurements were performed in the regions with and without Ga NPs, using a HeNe laser with an excitation energy of 1.96 eV, an incident laser power of 0.38 mW, and a beam spot diameter of 5 μm. The penetration depth of the 1.96 eV incident laser in GaAs is 1/α = 1/(3.57 × 104 cm−1) = 280 nm, where α is the GaAs absorption coefficient at 1.96 eV.27 Since the laser penetration depth is significantly lower than the GaAs wafer thickness (∼600 μm), transmission through the samples is expected to be negligible.

To gain physical insight into the influence of the buried Ga NP array on the GaAs near-band-edge (NBE) PL emission, we compute the predicted absorptivity and spontaneous emission (SE) rates of the samples. We use finite-difference time-domain (FDTD) simulations to calculate the absorptivity. To determine the SE rates, we use LSPR energy (ELSPR) values from FDTD simulations, in conjunction with effective dielectric permittivities, εeff, based upon Maxwell-Garnett effective medium approximations.11 In our FDTD simulations, a broad-band (1 to 10 eV) plane wave is incident on a volume of a 25 nm-thick vacuum layer and 1 μm × 1 μm × 1 μm nanocomposite, consisting of a square array of hemispheroidal Ga NPs, with ⟨dGa⟩ of 40 nm or 66 nm, and inter-droplet spacing, ⟨dGa-Ga⟩, of 72 nm or 108 nm, buried at depths of 20, 50, 100, 200, and 500 nm. Using low temperature frequency-dependent complex permittivities of Ga26 and GaAs28,29 from the literature, we compute the electric field, E(x,y,z,ω) and the absorbed optical power per unit volume, Pabsω=12ω Im(n(x,y,z,ω))E(x,y,z,ω).2 To determine the total absorbed (injected) optical power, we then integrate the absorbed (injected) optical power per unit volume over the entire simulated space, with a 1 nm3 mesh size and perfectly matched layer boundary conditions, which allows attenuation without reflection at the boundary of the simulated space. The absorptance is defined as the ratio of the total absorbed optical power to the total injected power. To validate these simulations, we calculate and compare the reflectance of bulk GaAs to experimental data.30 The resulting energy-dependent reflectance spectrum of bulk GaAs at 300 K is similar to the measured reflectance reported by Philipp and Ehrenreich31 The FDTD simulations are also further validated by a comparison of εeff with that calculated using the Maxwell-Garnett effective medium approximation.32,33

To assess the crystallinity of the MBE-overgrown GaAs layers, we consider both in-situ RHEED and ex-situ cross-sectional TEM. A RHEED pattern, collected at the end of the overgrowth process, as shown in Fig. 1(c), contains spotty concentric rings which correspond to the {200}, {113}, {400}, {331}, and {115} planes of zincblende GaAs.34 A cross-sectional TEM image following 500 nm of MBE overgrowth of ⟨dGa⟩ = 40 nm Ga NP arrays, shown in Fig. 1(d), indicates that the FIB-patterned Ga NPs are embedded between the GaAs substrate and the MBE-overgrown GaAs layer. In the bright field (BF) TEM image, the high brightness of the Ga NPs (highlighted by dotted circles) in comparison to that of the GaAs substrate region suggests that the NPs are amorphous, consistent with literature reports.35 The selected-area electron diffraction (SAED) pattern exclusively collected from the overgrown layer, shown in Fig. 1(e), has spotty rings (overlaid by dashed guidelines) which correspond to the {111}, {220}, and {113} planes of ZB GaAs.34 

For epitaxial growth of GaAs, any deviation from III-V stoichiometry depends sensitively on the substrate temperature. Typically, a high concentration of excess arsenic in the form of As antisite defects, AsGa, is observed for GaAs growth at 200 °C, with a decrease in [AsGa] as the substrate temperature is increased.36 For our GaAs grown at 300 °C, we estimated the [AsGa] using an analysis of high-resolution x-ray rocking curve (HRXRC) measurements. [AsGa] is often determined from an increase in the lattice parameter of the low-temperature GaAs (LT-GaAs) layer according to Δa/aGaAs = 1.24 × 1024 × [AsGa], where aGaAs is the GaAs substrate lattice parameter, Δa is the difference between aGaAs and the LT-GaAs lattice parameter, and [AsGa] is in cm−3.37 Due to the absence of a distinct LT-GaAs diffraction peak, we estimate an upper bound for [AsGa] using the FWHM of the GaAs peak. Thus, for our overgrown GaAs layers, the [AsGa] values range from 1.2 × 1019 to 1.7 × 1019 cm−3, at least an order of magnitude lower concentration than standard LT-GaAs.36 

We now compare the AFM images of the surface morphologies for sample regions with and without buried Ga NP arrays, shown in Fig. 2. In all cases, corresponding fast Fourier transforms (FFTs) are shown as insets in each AFM image. In Figs. 2(a)–2(d), AFM images of the sample regions, without buried Ga NP arrays, are shown for overgrown GaAs layer thicknesses of (a) 0 nm, (b) 50 nm, (c) 100 nm, and (d) 500 nm. For the pristine GaAs without an overgrown layer shown in Fig. 2(a), a featureless surface is observed. For the overgrown layers without buried NP arrays, shown in Figs. 2(b)–2(d), the seemingly particle-like features are not associated with Ga NP arrays but rather due to the topography of the overgrown GaAs layer. Following overgrowth, the sizes of the particle-like features increase, as shown in Figs. 2(b)–2(d). Interestingly, for all overgrowth thicknesses, circular FFT patterns are observed, indicating an isotropic distribution of surface features.

FIG. 2.

Comparison of surface morphologies for sample regions with and without buried Ga NPs. Top-view AFM images of regions without buried Ga NP arrays, for overgrown GaAs layer thicknesses of (a) 0 nm, (b) 50 nm, (c) 100 nm, and (d) 500 nm; regions with ⟨dGa⟩ = 40 nm Ga NP arrays buried at depths of (e) 0 nm, (f) 50 nm, (g) 100 nm, and (h) 500 nm; regions with ⟨dGa⟩ = 66 nm Ga NP arrays buried at depths of (i) 0 nm, (j) 50 nm, (k) 100 nm, and (l) 500 nm. Corresponding fast Fourier transforms (FFTs) are shown as insets.

FIG. 2.

Comparison of surface morphologies for sample regions with and without buried Ga NPs. Top-view AFM images of regions without buried Ga NP arrays, for overgrown GaAs layer thicknesses of (a) 0 nm, (b) 50 nm, (c) 100 nm, and (d) 500 nm; regions with ⟨dGa⟩ = 40 nm Ga NP arrays buried at depths of (e) 0 nm, (f) 50 nm, (g) 100 nm, and (h) 500 nm; regions with ⟨dGa⟩ = 66 nm Ga NP arrays buried at depths of (i) 0 nm, (j) 50 nm, (k) 100 nm, and (l) 500 nm. Corresponding fast Fourier transforms (FFTs) are shown as insets.

Close modal

In Figs. 2(e)–2(h), AFM images of the sample regions with ⟨dGa⟩ = 40 nm buried Ga NP arrays are shown for overgrown layer thickness (i.e., NP array depths) of (e) 0 nm, (f) 50 nm, (g) 100 nm, and (h) 500 nm. Prior to overgrowth, the ⟨dGa⟩ = 40 nm Ga NP arrays, shown in Fig. 2(e), consist of close-packed arrays with ⟨dGa⟩ = 40 ± 6 nm, height, ⟨hGa⟩ = 8.0 ± 0.6 nm, and ⟨dGa-Ga⟩ = 72 ± 9 nm. The corresponding FFT inset in Fig. 2(e) exhibits a hexagonal spot pattern with a split center spot, suggesting the presence of a six-fold symmetry with a superimposed two-fold symmetry, consistent with the AFM image of elongated hexagonal arrays of NPs. Indeed, raster scanning of the ion beam induces Ga mass transport, eventually leading to the elongation of close-packed Ga NP arrays along the scan direction.38 Following 50 and 100 nm thick overgrowth of the NP arrays, similar elongated hexagonal arrays are observed, as shown in Figs. 2(f) and 2(g), with corresponding FFT insets consisting of hexagonal patterns with a split center spot. Finally, for NP array depths of 500 nm, the distribution of surface features becomes isotropic, as shown in the AFM image in Fig. 2(h), and confirmed by the corresponding circular FFT patterns in the inset.

Finally, in Figs. 2(i)–2(l), AFM images of the sample regions with ⟨dGa⟩ = 66 nm Ga NP arrays are shown for overgrown layer thickness (i.e., NP array depths) of (i) 0 nm, (j) 50 nm, (k) 100 nm, and (l) 500 nm. Prior to overgrowth, the ⟨dGa⟩ = 66 nm Ga NP arrays, shown in Fig. 2(i), consist of arrays with ⟨dGa⟩ = 66 ± 18 nm, ⟨hGa⟩ = 10 ± 3.3 nm, and ⟨dGa-Ga⟩ = 108 ± 12 nm. Following GaAs overgrowth, the array pattern becomes less well-ordered, as shown in Figs. 2(f) and 2(g). Indeed, the insets in Figs. 2(i)–2(l) reveal circular FFT patterns, indicating an isotropic distribution of surface features, independent of array depths ranging from 0 to 500 nm.

To take into account the influence of the surface morphology on the incident laser absorption and NBE emission, PL spectra were normalized by the integrated intensity of the absorbed laser light, i.e., the difference between the laser spectra and the surface-induced laser reflection.39,40 To consider the total reflected light at the GaAs surface, we estimate the light scattered out at the polycrystalline GaAs/air interface (i.e., diffuse reflection), assuming perfect transmission at the polycrystalline GaAs/single crystalline GaAs interface. Since the GaAs surfaces are not flat, as shown in the AFM images, we calculate the fraction of the light scattered by surface roughness (diffuse reflection) to the total reflected light consisting of the scattered light (diffuse reflection) plus the reflected light (specular reflection), defined as the total integrated scatter (TIS),39 

TIS=[1exp4πcosθσλ2],

using the root mean square (rms) roughness values (σ) obtained from the AFM images and the normal incident (θ = 90°) light wavelength (λ) of 633 nm. Figure 3 shows rms roughness (left) and TIS (right) as a function of overgrowth thickness, for regions without buried Ga NP arrays and regions with ⟨dGa⟩ = 40 nm and ⟨dGa⟩ = 66 nm buried Ga NP arrays. For each region, we use the measured reflection data and the calculated TIS values to estimate the total integrated reflected light at the GaAs surface.

FIG. 3.

NP depth dependence of root mean square (rms) roughness (left) and TIS (right) for regions without buried Ga NP arrays (black dashed line), regions with buried Ga NP arrays of ⟨dGa⟩ = 40 nm (green line), and regions with buried Ga NP arrays of ⟨dGa⟩ = 66 nm (blue line).

FIG. 3.

NP depth dependence of root mean square (rms) roughness (left) and TIS (right) for regions without buried Ga NP arrays (black dashed line), regions with buried Ga NP arrays of ⟨dGa⟩ = 40 nm (green line), and regions with buried Ga NP arrays of ⟨dGa⟩ = 66 nm (blue line).

Close modal

Figure 4 shows the overgrown layer thickness dependence of the GaAs PL intensities, for sample regions with and without buried NP arrays, shown as green/blue and black dotted lines, respectively, in comparison to that of a high-quality GaAs epilayer, shown as black solid lines. Within the range 1.475 to 1.525 eV, the GaAs NBE emission, including the band-to-band (BtB) emission at 1.515 eV and the donor-acceptor pair (DAP) emission at 1.491 eV, is apparent.40,41 For all NP array depths, the intensities of the DAP emissions are higher than those of the BtB emissions, both decreasing with increasing NP array depth. For ⟨dGa⟩ = 40 nm (66 nm), the DAP and BtB emissions are enhanced (reduced) in the presence of Ga NP arrays buried to depths ≤100 nm, as shown in Figs. 4(a) and 4(b). For NP array depths of 200 nm, the emission intensity is reduced for the sample region with Ga NP arrays, in comparison to that of the sample region without Ga NP arrays, as shown in Fig. 4(c). Finally, as shown in Fig. 4(d), for NP depth = 500 nm, the PL spectra from the sample regions with and without buried Ga NP arrays are similar, suggesting a negligible influence of NP arrays on PL emission. We note the PL intensities from the overgrown, polycrystalline GaAs layers without buried Ga NP arrays are lower than that from the high-quality, epitaxial GaAs layer, as expected.24,42 However, the PL intensity from the region with ⟨dGa⟩ = 40 nm NP arrays buried at 40 nm surpasses that of the high-quality, epitaxial GaAs layer, making these nanocomposites competitive. Since both the DAP and the BtB emission bands show a similar dependence on the NP array depth, we define the sum of the integrated areas under both emission bands as the “PL efficiency.” The PL enhancement ratio is then defined as the ratio of the PL efficiency from regions with buried NP arrays to that from regions without buried NP arrays. For arrays with ⟨dGa⟩ = 40 nm and NP array depths less than 100 nm, the PL enhancement ratio is 1.42. However, for NP array depths exceeding 100 nm, the PL enhancement ratio no longer exceeds 1.0. In addition, for arrays with ⟨dGa⟩ = 66 nm, the enhancement ratios do not exceed 1.0, revealing PL reduction in the presence of Ga NP arrays.

FIG. 4.

The NP array depth (overgrown layer thickness) dependence of the GaAs PL intensities for sample regions with and without buried NP arrays, in comparison to that of a high-quality epitaxial GaAs layer. The GaAs NBE emission, including BtB emission at 1.515 eV and DAP emission at 1.491 eV, is apparent. For NP arrays with ⟨dGa⟩ = 40 nm (66 nm), DAP and BtB emissions are enhanced (reduced) in the presence of Ga NP arrays with depths ≤100 nm, in some cases surpassing those of high-quality epitaxial GaAs layers. For a NP array depth of 200 nm, the emission intensity is reduced for the regions with buried Ga NP arrays, in comparison to regions without buried Ga NP arrays. For NP array depth = 500 nm, the PL spectra from the sample regions with and without buried Ga NP arrays are similar.

FIG. 4.

The NP array depth (overgrown layer thickness) dependence of the GaAs PL intensities for sample regions with and without buried NP arrays, in comparison to that of a high-quality epitaxial GaAs layer. The GaAs NBE emission, including BtB emission at 1.515 eV and DAP emission at 1.491 eV, is apparent. For NP arrays with ⟨dGa⟩ = 40 nm (66 nm), DAP and BtB emissions are enhanced (reduced) in the presence of Ga NP arrays with depths ≤100 nm, in some cases surpassing those of high-quality epitaxial GaAs layers. For a NP array depth of 200 nm, the emission intensity is reduced for the regions with buried Ga NP arrays, in comparison to regions without buried Ga NP arrays. For NP array depth = 500 nm, the PL spectra from the sample regions with and without buried Ga NP arrays are similar.

Close modal

To analyze the influence of the buried Ga NP arrays on the GaAs NBE PL emission, we quantify the Ga NP-array-induced emission enhancement and reduction. Since the predicted PL enhancement ratio is determined by the product of the absorptance enhancement ratio and the SE rate enhancement ratio, we now consider each ratio individually. Figure 5(a) shows the computed depth-dependent absorption (under 1.96 eV laser excitation) of GaAs with buried Ga NP arrays (⟨dGa⟩ = 40 nm and NP array depth = 100 nm). The absorption is visualized as colors where the intensity increases as the color changes from yellow to red. The depth-dependent absorptance of GaAs, with and without buried Ga NP arrays, is shown in Fig. 5(b) for ⟨dGa⟩ = 40 nm NP arrays at depths of 20 nm, 50 nm, 100 nm, 200 nm, and 500 nm. Since coupling of the incident radiation to the buried NP arrays is expected for the array depths less than the laser penetration depth (∼280 nm), absorption enhancement is observed for NP array depths of 20 nm, 50 nm, 100 nm, and 200 nm. However, minimal coupling of the incident radiation to the buried NP arrays is expected for the 500 nm NP array depths, leading to a negligible absorption.

FIG. 5.

(a) Simulated depth-dependent absorption (under 1.96 eV laser excitation) for GaAs with buried Ga NP arrays (⟨dGa⟩ = 40 nm and NP array depth = 100 nm). (b) The simulated depth-dependent absorption of GaAs, with and without buried Ga NP arrays (⟨dGa⟩ = 40 nm and NP array depths of 20 nm, 50 nm, 100 nm, 200 nm, and 500 nm). The simulated spectral absorptance of GaAs, with and without buried Ga NP arrays: (c) ⟨dGa⟩ = 40 nm and (d) ⟨dGa⟩ = 66 nm. (e) The plot of absorption enhancement ratio for Ga NP arrays (⟨dGa⟩ = 40 and 66 nm) as a function of NP array depth. For ⟨dGa⟩ = 40 nm NP arrays, with depths less than 100 nm, the absorptance enhancement ratio is 1.1–1.22. For NP arrays, with depths exceeding 200 nm, the absorption enhancement ratio decreases to near 1.0. The highest absorption enhancement ratio is predicted for “shallow” arrays of ⟨dGa⟩ = 40 nm, buried at a depth <100 nm.

FIG. 5.

(a) Simulated depth-dependent absorption (under 1.96 eV laser excitation) for GaAs with buried Ga NP arrays (⟨dGa⟩ = 40 nm and NP array depth = 100 nm). (b) The simulated depth-dependent absorption of GaAs, with and without buried Ga NP arrays (⟨dGa⟩ = 40 nm and NP array depths of 20 nm, 50 nm, 100 nm, 200 nm, and 500 nm). The simulated spectral absorptance of GaAs, with and without buried Ga NP arrays: (c) ⟨dGa⟩ = 40 nm and (d) ⟨dGa⟩ = 66 nm. (e) The plot of absorption enhancement ratio for Ga NP arrays (⟨dGa⟩ = 40 and 66 nm) as a function of NP array depth. For ⟨dGa⟩ = 40 nm NP arrays, with depths less than 100 nm, the absorptance enhancement ratio is 1.1–1.22. For NP arrays, with depths exceeding 200 nm, the absorption enhancement ratio decreases to near 1.0. The highest absorption enhancement ratio is predicted for “shallow” arrays of ⟨dGa⟩ = 40 nm, buried at a depth <100 nm.

Close modal

Figures 5(c) and 5(d) present the computed spectral absorptance of GaAs with NP arrays (both ⟨dGa⟩ = 40 nm and ⟨dGa⟩ = 66 nm) at depths of 0 nm, 20 nm, 50 nm, 100 nm, 200 nm, and 500 nm. In the energy range of 1 to 3 eV, the absorptance of the nanocomposites (αnanocomposite) is enhanced with respect to that of GaAs (αGaAs). The predicted LSPR-enhancement of absorption is determined by a convolution of the laser energy distribution with the predicted energy distribution of absorptance. As an approximation, we consider a delta-like laser energy distribution and calculate the integrated absorptance at the maximum of the laser energy distribution, 1.96 eV, both with and without Ga NP arrays, as αnanocomposite and αGaAs. We then define the absorption enhancement ratio as the ratio of αnanocomposite to αGaAs. Figure 5(e) shows the NP depth dependence of the absorptance enhancement ratio for ⟨dGa⟩ = 40 nm and ⟨dGa⟩ = 66 nm arrays, with NP array depths ranging from 0 to 500 nm. With increasing NP diameter and depth, the absorptance enhancement decreases, presumably due to a reduction in NP-array-induced E-field enhancement. For ⟨dGa⟩ = 40 nm NP arrays at depths less than 100 nm, the predicted absorptance enhancement ratio is 1.1–1.22, resulting in the enhancement of electron-hole pair (EHP) generation. However, for ⟨dGa⟩ = 40 nm NP arrays at depths exceeding 200 nm and ⟨dGa⟩ = 66 nm NP arrays at depths exceeding 50 nm, the absorptance enhancement ratio decreases to near 1.0, suggesting negligible NP-induced enhancement of EHP generation. The highest absorptance enhancement ratio is predicted for ⟨dGa⟩ = 40 nm at a depth <100 nm.

To compute the SE rate of GaAs with buried Ga NP arrays, we consider NP size- and depth-dependent ELSPR and εeff.32,33,43,44 We quantify ELSPR for NP arrays, at each depth, by fitting the energy-dependence of the GaAs absorptance in Figs. 5(c) and 5(d); the maximum likelihood absorptance is then attributed to the ELSPR, as shown in Fig. 6(a).40 For NP array depths below 100 nm, the LSPR depends on NP size, with ELSPR = 1.69–2.49 eV for ⟨dGa⟩ = 40 nm and ELSPR = 1.47–2.13 eV for ⟨dGa⟩ = 66 nm. Using the Maxwell-Garnett effective medium approximation,32,33εeff is calculated as a function of NP sizes and array depth. As shown in the inset of Fig. 6(a), as the NP volume fraction decreases with increasing NP depth, εeff approaches that of bulk GaAs. For each NP size and array depth, we calculate the resulting ratio of the SE rate of nanocomposite to that of bulk GaAs, i.e., the SE rate enhancement ratio, according to

FE(r)μEmax|μ|211+4Q2λincidentλ12,

where F is the Purcell factor,

F=34π2λ3n3QVmode,

λ/n is the wavelength within the material, and Q and Vmode are the quality factor and mode volume of the plasmonic structure, respectively. As shown in Fig. 6(b), for embedded NP arrays, the SE rate ratios are below 1, indicating a decrease in the presence of NP arrays. In addition, the SE enhancement ratio decreases with increasing NP size and array depth.

FIG. 6.

(a) Plot of computed ELSPR and εeff (inset) of the GaAs:Ga nanocomposites as a function of NP depth. The green and blue lines correspond to ⟨dGa⟩ = 40 and ⟨dGa⟩ = 66 nm, respectively. (b) A plot of SE rate enhancement ratio vs. NP array depth. The resulting enhancement ratios are below 1, suggesting that the SE rate is decreased in the presence of NP arrays.

FIG. 6.

(a) Plot of computed ELSPR and εeff (inset) of the GaAs:Ga nanocomposites as a function of NP depth. The green and blue lines correspond to ⟨dGa⟩ = 40 and ⟨dGa⟩ = 66 nm, respectively. (b) A plot of SE rate enhancement ratio vs. NP array depth. The resulting enhancement ratios are below 1, suggesting that the SE rate is decreased in the presence of NP arrays.

Close modal

Figure 7 shows the comparison between the measured and predicted GaAs PL enhancement ratios for arrays with (a) ⟨dGa⟩ = 40 nm and (b) 66 nm. Both the measured and predicted PL enhancement ratios decrease with increasing NP array depth up to 200 nm. In Fig. 7(a), for ⟨dGa⟩ = 40 nm arrays, the GaAs NBE emission enhancement is due to the NP-enhanced absorption [Fig. 5(e)], which overwhelms the NP-reduced SE rate for array depths up to 100 nm [Fig. 6(b)]. For NP array depths ≥100 nm, the emission is instead reduced due to the fact that the SE rate reduction overwhelms the enhanced absorption. On the other hand, for 500 nm array depths, the apparent PL enhancement of ∼1.0 is likely due to minimal coupling of the incident radiation to the buried NP arrays for a depth exceeding the laser penetration depth. Finally, in Fig. 7(b), for ⟨dGa⟩ = 66 nm arrays, the PL emission is reduced, independent of NP array depth, presumably due to the reduced SE rate and the insignificant absorption enhancement in this case.

FIG. 7.

A comparison between the measured and computed GaAs PL enhancement ratios vs. NP array depth for (c) ⟨dGa⟩ = 40 nm and (d) ⟨dGa⟩ = 66 nm. The PL efficiency is defined as the sum of the integrated areas under both the donor-acceptor pair (DAP) and Band-to-Band (BtB) emission bands. The PL enhancement ratios are then defined as the ratio of the integrated PL intensities for regions with and without buried Ga NP arrays. For ⟨dGa⟩ = 40 nm NP arrays buried at depths <100 nm, the PL enhancement ratio is 1.42, independent of NP array depth. For NP array depths exceeding 100 nm, the PL enhancement ratio is below or similar to 1.0. For ⟨dGa⟩ = 66 nm arrays, the PL enhancement ratios are similar to or less than 1.0, revealing PL reduction in the presence of buried Ga NP arrays.

FIG. 7.

A comparison between the measured and computed GaAs PL enhancement ratios vs. NP array depth for (c) ⟨dGa⟩ = 40 nm and (d) ⟨dGa⟩ = 66 nm. The PL efficiency is defined as the sum of the integrated areas under both the donor-acceptor pair (DAP) and Band-to-Band (BtB) emission bands. The PL enhancement ratios are then defined as the ratio of the integrated PL intensities for regions with and without buried Ga NP arrays. For ⟨dGa⟩ = 40 nm NP arrays buried at depths <100 nm, the PL enhancement ratio is 1.42, independent of NP array depth. For NP array depths exceeding 100 nm, the PL enhancement ratio is below or similar to 1.0. For ⟨dGa⟩ = 66 nm arrays, the PL enhancement ratios are similar to or less than 1.0, revealing PL reduction in the presence of buried Ga NP arrays.

Close modal

In summary, we have examined the formation of embedded Ga NP arrays and their influence on GaAs NBE PL efficiency. Embedded Ga NP arrays were fabricated using off-normal FIB irradiation on selected regions of GaAs substrates, followed by MBE overgrowth of GaAs layers, providing a seamless approach to embed non-noble metals during epitaxy. We consider the influence of embedded Ga NP arrays on the absorption and SE rate processes. We identify a regime of NP size and depth where NP-enhanced absorption overwhelms the reduction in SE rate in GaAs, leading to enhanced NBE emission from GaAs, which surpasses that from high-quality epitaxial GaAs layers. Although the overgrown GaAs is polycrystalline, efforts to improve the crystallinity of the overgrown GaAs layers are in progress. Furthermore, polycrystalline III-V compound semiconductors have been proposed for LEDs in large-area displays.45,46 Indeed, this new Ga NP plasmonics approach would enable polycrystalline gain media deposited on large-area substrates to maintain reasonable light-emitting characteristics. Thus, this approach provides an opportunity to enhance the PL efficiency from a variety of semiconductor heterostructures.

This work was supported by the National Science Foundation through the Materials Research Science and Engineering Center (MRSEC) at the University of Michigan, Grant No. DMR-1120923. We thank C. W. Reese and P. C. Ku for useful discussions.

1.
G.
Lozano
,
S. R. K.
Rodriguez
,
M. A.
Verschuuren
, and
J. G.
Rivas
,
Light: Sci. Appl.
5
,
e16080
(
2016
).
2.
K.
Okamoto
,
I.
Niki
,
A.
Shvartser
,
Y.
Narukawa
,
T.
Mukai
, and
A.
Scherer
,
Nat. Mater.
3
,
601
(
2004
).
3.
H. A.
Atwater
and
A.
Polman
,
Nat. Mater.
9
,
205
(
2010
).
4.
A. J.
Haes
,
S. L.
Zou
,
G. C.
Schatz
, and
R. P.
Van Duyne
,
J. Phys. Chem. B
108
,
6961
(
2004
).
5.
M. P.
Hanson
,
A. C.
Gossard
, and
E. R.
Brown
,
J. Appl. Phys.
102
,
043112
(
2007
).
6.
S.
Pillai
,
K. R.
Catchpole
,
T.
Trupke
, and
M. A.
Green
,
J. Appl. Phys.
101
,
093105
(
2007
).
7.
C. W.
Cheng
,
E. J.
Sie
,
B.
Liu
,
C. H. A.
Huan
,
T. C.
Sum
,
H. D.
Sun
, and
H. J.
Fan
,
Appl. Phys. Lett.
96
,
071107
(
2010
).
8.
H.
Naiki
,
A.
Masuhara
,
S.
Masuo
,
T.
Onodera
,
H.
Kasai
, and
H.
Oikawa
,
J. Phys. Chem. C
117
,
2455
(
2013
).
9.
H.
Mertens
,
J.
Verhoeven
,
A.
Polman
, and
F. D.
Tichelaar
,
Appl. Phys. Lett.
85
,
1317
(
2004
).
10.
G.
Sun
,
J. B.
Khurgin
, and
R. A.
Soref
,
Appl. Phys. Lett.
94
,
101103
(
2009
).
11.
P.
Spinelli
and
A.
Polman
,
Opt. Express
20
,
A641
(
2012
).
12.
Y. P.
Hsieh
,
C. T.
Liang
,
Y. F.
Chen
,
C. W.
Lai
, and
P. T.
Chou
,
Nanotechnology
18
,
415707
(
2007
).
13.
C. W.
Chen
,
C. H.
Wang
,
C. M.
Wei
, and
Y. F.
Chen
,
Appl. Phys. Lett.
94
,
071906
(
2009
).
14.
S. K.
Cushing
,
J.
Li
,
F.
Meng
,
T. R.
Senty
,
S.
Suri
,
M.
Zhi
,
M.
Li
,
A. D.
Bristow
, and
N.
Wu
,
J. Am. Chem. Soc.
134
,
15033
(
2012
).
15.
O.
Hunderi
and
R.
Ryberg
,
J. Phys. F: Met. Phys.
4
,
2096
(
1974
).
16.
P.
Wu
,
T.
Kim
,
A.
Brown
,
M.
Losurdo
,
G.
Bruno
, and
H.
Everitt
,
Appl. Phys. Lett.
90
,
103119
(
2007
).
17.
M.
Kang
,
T. W.
Saucer
,
M. V.
Warren
,
J. H.
Wu
,
H.
Sun
,
V.
Sih
, and
R. S.
Goldman
,
Appl. Phys. Lett.
101
,
081905
(
2012
).
18.
S. R. C.
Vivekchand
,
C. J.
Engel
,
S. M.
Lubin
,
M. G.
Blaber
,
W.
Zhou
,
J. Y.
Suh
,
G. C.
Schatz
, and
T. W.
Odom
,
Nano Lett.
12
,
4324
(
2012
).
19.
C.
Yi
,
T. H.
Kim
,
W.
Jiao
,
Y.
Yang
,
A.
Lazarides
,
K.
Hingerl
,
G.
Bruno
,
A.
Brown
, and
M.
Losurdo
,
Small
8
,
2721
(
2012
).
20.
Y.
Yang
,
J. M.
Callahan
,
T. H.
Kim
,
A. S.
Brown
, and
H. O.
Everitt
,
Nano Lett.
13
,
2837
(
2013
).
21.
M.
Yarema
,
M.
Worle
,
M. D.
Rossell
,
R.
Erni
,
R.
Caputo
,
L.
Protesescu
,
K. V.
Kravchyk
,
D. N.
Dirin
,
K.
Lienau
,
F.
Rohr
,
A.
Schilling
,
M.
Nachtegaal
, and
M. V.
Kovalenko
,
J. Am. Chem. Soc.
136
,
12422
(
2014
).
22.
G.
Garcia
,
R.
Buonsanti
,
E. L.
Runnerstrom
,
R. J.
Mendelsberg
,
A.
Llordes
,
A.
Anders
,
T. J.
Richardson
, and
D. J.
Milliron
,
Nano Lett.
11
,
4415
(
2011
).
23.
C.
Clavero
,
J. L.
Slack
, and
A.
Anders
,
J. Phys. D: Appl. Phys.
46
,
362001
(
2013
).
24.
M.
Kang
,
A. A.
Al-Heji
,
J.-E.
Lee
,
T. W.
Saucer
,
J. H.
Wu
,
L.
Zhao
,
A. L.
Katzenstein
,
D. L.
Sofferman
,
V.
Sih
, and
R. S.
Goldman
,
Appl. Phys. Lett.
103
,
101903
(
2013
).
25.
Q.
Wei
,
J.
Lian
,
W.
Lu
, and
L.
Wang
,
Phys. Rev. Lett.
100
,
076103
(
2008
).
26.
M. W.
Knight
,
T.
Coenen
,
Y.
Yang
,
B. J. M.
Brenny
,
M.
Losurdo
,
A. S.
Brown
,
H. O.
Everitt
, and
A.
Polman
,
ACS Nano
9
,
2049
(
2015
).
27.
H. C.
Casey
, Jr.
,
D. D.
Sell
, and
K. W.
Wecht
,
J. Appl. Phys.
46
,
250
(
1975
).
28.
M.
Scalora
,
M. A.
Vincenti
,
V.
Roppo
,
J. V.
Foreman
,
J. W.
Haus
,
N.
Akozbek
, and
M. J.
Bloemer
, in
IEEE/LEOS Conference
(
2009
), p.
132
.
29.
I.
Strzalkowski
,
S.
Joshi
, and
C. R.
Crowell
,
Appl. Phys. Lett.
28
,
350
(
1976
).
30.
S.
Jeon
, Ph. D. thesis,
University of Michigan
,
2016
.
31.
H. R.
Philipp
and
H.
Ehrenreich
,
Phys. Rev.
129
,
1550
(
1963
).
32.
G.
Xu
,
M.
Tazawa
,
P.
Jin
,
S.
Nakao
, and
K.
Yoshimura
,
Appl. Phys. Lett.
82
,
3811
(
2003
).
33.
X.
Liu
,
D.
Li
,
X.
Sun
,
Z.
Li
,
H.
Song
,
H.
Jiang
, and
Y.
Chen
,
Sci. Rep.
5
,
12555
(
2015
).
34.
K.
Nozawa
,
M.
Delville
,
H.
Ushiki
,
P.
Panizza
, and
J.
Delville
,
Phys. Rev. E
72
,
011404
(
2005
).
35.
M.
Kang
,
J. H.
Wu
,
S.
Huang
,
M. V.
Warren
,
Y.
Jiang
,
E. A.
Robb
, and
R. S.
Goldman
,
Appl. Phys. Lett.
101
,
082101
(
2012
).
36.
K. M.
Yu
,
M.
Kaminska
, and
Z.
Liliental-Weber
,
J. Appl. Phys.
72
,
2850
(
1992
).
37.
X.
Liu
,
A.
Prasad
,
J.
Nishio
,
E. R.
Weber
,
Z.
Liliental-Weber
, and
W.
Walukiewicz
,
Appl. Phys. Lett.
67
,
279
(
1995
).
38.
M.
Kang
,
J. H.
Wu
,
D. L.
Sofferman
,
I.
Beskin
,
H. Y.
Chen
,
K.
Thornton
, and
R. S.
Goldman
,
Appl. Phys. Lett.
103
,
072115
(
2013
).
39.
J. E.
Harvey
,
S.
Schroder
,
N.
Choi
, and
A.
Duparre
,
Opt. Eng.
51
,
013402
(
2012
).
40.
H. E.
Bennett
and
J. O.
Porteus
,
J. Opt. Soc. Am.
51
,
123
(
1961
).
41.
P. B.
Perry
and
A. E.
Blakeslee
,
J. Electron. Mater.
8
,
635
(
1979
).
42.
M.
Kang
, Ph. D. thesis,
University of Michigan
,
2014
.
43.
T. S.
Ferguson
,
J. Am. Stat. Assoc.
73
,
211
(
1978
).
44.
D.
Englund
,
D.
Fattal
,
E.
Waks
,
G.
Solomon
,
B.
Zhang
,
T.
Nakaoka
,
Y.
Arakawa
,
Y.
Yamamoto
, and
J.
Vuckovic
,
Phys. Rev. Lett.
95
,
013904
(
2005
).
45.
D. P.
Bour
,
N. M.
Nickel
,
C. G.
Van de Walle
,
M. S.
Kneissl
,
B. S.
Krusor
,
P.
Mei
, and
N. M.
Johnson
,
Appl. Phys. Lett.
76
,
2182
(
2000
).
46.
K.
Yamada
,
H.
Asahi
,
H.
Tampo
,
Y.
Imanishi
,
K.
Ohnishi
, and
K.
Asami
,
Appl. Phys. Lett.
78
,
2849
(
2001
).