1. Introduction

Plasmonic structures can strongly confine an electromagnetic field in a region smaller than the diffraction limit of light, which leads to the strong enhancement of various optical processes [1–5]. To achieve a large enhancement, it is important to couple a light-emitting material efficiently with the localized field. A gap-type plasmonic structure such as a bowtie nanoantenna [6] is especially good at concentrating an electric field within the gap region. In such gap structures, nanomaterials such as quantum dots and nanowires are more suitable as emitters than bulk materials because of their dimensions. The fabrication accuracy of plasmonic nanostructures has been greatly improved as a result of the progress made on electron beam (EB) lithography, but the appropriate positioning of the emitters has been a problem. Although optical tweezer [7] and atomic force microscopy (AFM) manipulation [8–10] techniques have been used to place nanoparticles on plasmonic structures, a simple dispersion technique has been mainly used to place nanoemitters on a fabricated nanostructure [6,11].

Here we focus on a semiconductor nanowire as a nanoemitter; it possesses various unique properties and has recently been extensively studied [12–17]. It has become possible to grow high-quality semiconductor nanowires with a diameter of less than 100 nm by using the vapor-liquid-solid (VLS) growth mode [18]. By tuning the nanowire growth sequence, it is possible to grow various functional structures such as multilayer heterostructures [19–21], quantum wells [22], quantum dots [16,23,24], core-shell structures [25], and p-n and p-i-n junctions [13–15,17,26,27], and various optical device applications, including lasers [12,22,28–32], detectors [15,33,34], wavelength converters [35], have been reported. In addition, it has been revealed that a nanowire’s unique subwavelength geometry results in an anomalous photonic density of states for polarization perpendicular to the wire, which leads to a strong field suppression effect [34,36] and the inhibition of spontaneous emission [37–39]. It is difficult to achieve strong light confinement in nanowires and efficient light emission from nanowires because of their subwavelength nature. Therefore, the combination of a semiconductor nanowire and a plasmonic structure constitutes a promising candidate for highly efficient ultrasmall optical devices. The combination of semiconductor nanowires and plasmonic modes has already been reported but only for structures where there is no precise positional control of the nanowires. The nanowires were simply dispersed on plasmonic structures [40] or simple metal films [30], or plasmonic structures were fabricated on dispersed nanowires [41,42]. However, we need to control the positions and orientations of both nanowires and plasmonic structures if we are to use such combinations for photonic integrated circuit applications.

In this study, we fabricate a combination consisting of a plasmonic bowtie nanoantenna fabricated by EB lithography and a VLS-grown semiconductor nanowire as illustrated in Fig. 1(a). We adopt a unique nanomanipulation technique to place a single subwavelength nanowire within the gap of a bowtie nanoantenna. That manipulation is performed with a nanomanipulator installed in a focused ion beam (FIB) system. The FIB shapes the probe tip of the nanomanipulator arbitrarily and directly. Our FIB system also has an electron beam column, which allows us to move nanomaterials while conducting in situ observation. Moreover, a gas injection system enables us to deposit metals on a small area with an electron beam. Thus, cutting, moving, and sticking are possible in our FIB system, which we call a “Nanofactory” [43]. The combination of these functions enables us to manipulate a subwavelength nanowire very precisely as described later. The bowtie nanoantenna consists of a pair of metal triangles known to possess a strongly confined mode within the gap, which can efficiently couple with free-space radiation modes. The photoluminescence (PL) from a nanoemitter placed in the gap can be enhanced. However, there has yet to be a report in which a semiconductor nanowire is placed on a bowtie nanoantenna with precise positional control and PL enhancement observed. A similar single hybrid structure has been reported byemploying randomly dispersed nanowires, and the enhancement of second harmonic generation has been investigated [40], but PL enhancement has not been reported as far as we know. We have undertaken detailed spatially resolved luminescence studies with cathodoluminescence (CL) and PL measurements for the fabricated samples. These studies show that when we excite a nanowire with a laser emitting at wavelengths around the plasmonic resonance of the nanoantenna, the light emission from a nanowire placed on a nanoantenna was greatly enhanced. We also show that this large enhancement is achieved by an interplay between the field enhancement induced by the nanoantenna and the breakdown of the field suppression in the nanowire.

 

Fig. 1 (a) Schematic structure of the sample and coordinate system for numerical calculations. The origin of the coordinate system is at the center of the nanoantenna on the substrate surface. X (Y) is the perpendicular (parallel) direction to the nanoantenna in plane. (b) Nanomanipulation process. We picked up a single nanowire with a diameter of 60 nm and a length of 7 μm and placed it in the nanoantenna gap. (c) SEM image of the fabricated sample. (d) Magnified SEM image of the nanoantenna. The stage was tilted by 45°.

Download Full Size | PDF

2. Results and discussion

2.1 Fabrication and manipulation

The sample studied here is a combination consisting of a gold bowtie nanoantenna and an InP nanowire [Fig. 1(a)]. The InP nanowires were grown on a Si substrate using the Au-free indium-particle-assisted VLS mode [44,45] with a metal-organic vapor phase epitaxy reactor. The diameter, length, and growth direction of the nanowire were 60 nm, 7 μm, and [112], respectively. The gold nanoantenna was fabricated on a thermally oxidized Si wafer (the SiO₂ was 1 μm thick) by EB lithography and the lift off technique. A 50-nm-thick Au layer was evaporated after the adhesion of a 0.6-nm Ti layer to the oxidized Si substrate. The nanoantenna was formed of a pair of equilateral triangles with a length of 100 nm and an 80-nm gap. These two parameters are important for characteristics of bowtie nanoantennas. The gap size was chosen to fit the nanowire diameter, and we determined the triangle size suitable for the excitation and emission wavelengths from the numerical simulation.

To realize a combined geometry, we used a nanomanipulator to place the nanowire directly in the nanoantenna gap in the chamber of a dual beam FIB system. The system is equipped with a FIB column, an electron beam column, a gas injection system, and a nanomanipulator. The nanomanipulator was controlled with a piezo actuator under in situ scanning electron microscope (SEM) observation [43]. Tungsten, which was deposited by electron beam assisted deposition, was used as a glue. To manipulate the nanowire, we first sharpened the probe tip of the nanomanipulator using in situ FIB milling so that it was smaller than 200 nm. We used the FIB only for sharpening the probe. Thus, neither the nanowire nor the nanoantenna was damaged by the FIB. Next, we brought the tip into contact with a single nanowire among the many nanowires on the substrate and deposited a small amount of tungsten onto the contact area. We then picked up the nanowire and placed it in the gap of the nanoantenna under in situ SEM observation [Fig. 1(b), see Visualization 1: movie showing a nanowire being picked up by the nanomanipulator (16x speed)]. After placing the nanowire in its final position, one end of the nanowire was anchored with a small amount of tungsten, and then the attached probe tip was removed by retracting the nanomanipulator. A SEM image of the fabricated sample is shown in Fig. 1(c), and the magnified figure [Fig. 1(d)] shows that we successfully placed the nanowire on the nanoantenna very precisely. We can place a nanowire at a target position with a success rate of about 50%. This manipulation technique enables us to position any nanowire in any place directly and precisely, even on 3D structures. When a nanoantenna is fabricated on a dispersed nanowire [41,42], the position and orientation of the nanowire-nanoantenna system is restricted by the initial random placement of the dispersed nanowires. In addition, extra nanowires are always left in other locations. Our method overcomes these problems and enables us to fabricate the nanowire-nanoantenna system at an arbitrary position and orientation. Thus, it will even be possible to employ more than one nanowire placed accurately in a single optical circuit, which would be very difficult to achieve with the previous fabrication method (where nanowires are first dispersed randomly, and then circuits are formed). We believe that on-demand hybrid nanomaterial-plasmonics structures will be an important and promising next step for nanophotonics research beyond random dispersion.

2.2 Numerical analysis of plasmonic modes in nanowire-nanoantenna system

Before presenting our experimental results, here we numerically analyze the plasmonic modes in our nanowire-nanoantenna system. We calculated the plasmonic properties of a gold nanoantenna with an InP nanowire by the finite element method (FEM). In the following calculation, we define the coordinate system with its origin at the center of the nanoantenna on the SiO₂ substrate surface [Fig. 1(a)], where X and Y are directions perpendicular and parallel to the nanoantenna in plane. The structural parameters were about the same as those of the fabricated sample (the radius of curvature of the triangle corner is 20 nm, the triangle is 100 nm long, the gap is 80 nm wide, the antenna is 50 nm thick). We assumed the permittivity of the SiO₂ substrate to be 2.1 and took the permittivities of gold and InP from previous studies [46,47].

First, we investigated the electric field intensity at the antenna gap for incident light polarized in the X and Y directions (E_⊥ excitation and E_// excitation) at various wavelengths, since our primary target for this study is to observe the plasmonic enhancement in the excitation process for light emission from nanowires. We estimated the electric field intensity at the center of the nanowire. Hereafter, E_a (E₀) represents the electric field intensity at (X, Y, Z) = (0, 0, d/2) with (without) the nanoantenna, where d is the nanowire diameter. The calculated results were obtained under a plane-wave excitation condition (incident direction: −Z), and the field intensity enhancement ratio was defined by (|E_a|/|E₀|)². We also calculated (|E_a|/|E₀|)² for the structure without nanowires, where we estimated the electric field intensity at (X, Y, Z) = (0, 0, 25 nm).

Figure 2 shows the wavelength dependence of (|E_a|/|E₀|)² in three cases: (i) without the nanowire for E_// excitation, (ii) with the nanowire for E_// excitation, and (iii) with the nanowire for E_⊥ excitation. The calculated result for the nanoantenna only (case (i)) shows a pronounced enhancement by a factor of 17 around 660 nm, and this peak corresponds to the fundamental parallel-polarized resonance of the bowtie nanoantenna. The enhancement factor in the bowtie nanoantenna was more than twice that in the single triangle, and the peak wavelength in the bowtie nanoantenna was about 10 nm longer than that in the single triangle. Therefore, our nanoantenna is not a structure consisting of two uncoupled triangles although the coupling between the two triangles is expected to be small because of its large gap. When we inserted nanowires with various d values into the nanoantenna (case (ii)), (|E_a|/|E₀|)² was increased as d increased. On the other hand, (|E_a|/|E₀|)² for E_⊥ excitation (case (iii)) mostlyoverlapped the gray line corresponding to (|E_a|/|E₀|)² = 1, which means that there is no enhancement. This indicates that the resonance in case (ii) is also a fundamental parallel-polarized resonance of the bowtie nanoantenna with the peak wavelength shifted slightly longer. The most significant finding is that the E_// field enhancement for the nanowire-nanoantenna system is larger than that for the nanoantenna alone. We will discuss the origin of this phenomenon later.

 

Fig. 2 Wavelength dependence of the electric field enhancement ratio (|E_a|/|E₀|)² calculated by FEM. E_a is the electric field at (X, Y, Z) = (0, 0, d/2) for the structure with the nanoantenna, and E₀ is that for the structure without the nanoantenna. d is the nanowire diameter. The gray line shows (|E_a|/|E₀|)² = 1.

Download Full Size | PDF

Next, we investigated the field distribution of the nanowire-nanoantenna system at a wavelength of 660 nm (around the resonance). Here, we used d = 50 nm to compare the field distributions of various structures at the same surface (Z = 25 nm). Figures 3(a) and 3(b) show the distributions of the electric field |E| for the structure with only the nanoantenna and only the nanowire, respectively, where the excitation polarization is E_//. Figure 3(a) shows that the electric field is concentrated at each corner of the nanoantenna, especially in the gap region. On the other hand, Fig. 3(b) shows that the E_// field is strongly suppressed inside the nanowire. This is the well-known field-suppressing effect in nanowires [34,36], which is induced by the continuity of the electric flux density across the boundary with a large refractive index difference. Although the continuity condition affects only the boundary region, when the nanowire diameter becomes much smaller than the wavelength of light, the photonic density of state of the entire nanowire is greatly reduced. In our study, the nanowire diameter is much smaller than the excitation wavelength, thus we consider the field suppression in the nanowire observed in Fig. 3(b) to be explained by this effect. The |E| distributions for the structure with the nanowire-nanoantenna system are shown in Figs. 3(c) and 3(e). The field distribution for E_// excitation [Fig. 3(c)] shows that the electric field is indeed greatly concentrated in the gap region of the nanoantenna. Figure 3(d) shows a cross-sectional plot of |E|², which corresponds to Fig. 3(c), along the X direction at (Y, Z) = (0, 25 nm). This clearly shows that the field intensity in the nanowire is strongly enhanced in the center of the nanoantenna (the approximate width of the enhanced region is 60 nm along the nanowire direction). On the other hand, the field distribution for E_⊥ excitation [Fig. 3(e)] shows that the nanoantenna does not enhance the E_⊥ field in the gap region, and the nanowire does not suppress the E_⊥ field inside the nanowire at all.

 

Fig. 3 (a)-(c) Distributions of electric field |E| at Z = 25 nm for E_// excitation. The electric fields were calculated for structures with (a) only a nanoantenna, (b) only a nanowire, and (c) the nanowire-nanoantenna system. The diameter of the nanowire d is 50 nm here, and the wavelength of the incident light is 660 nm. (d) Cross-sectional plot of |E|² along the X direction at (Y, Z) = (0, 25 nm) for the structure with the nanowire-nanoantenna system. This cross-sectional plot corresponds to (c). (e) Field distribution at Z = 25 nm for E_⊥ excitation. The electric field was calculated for the structure with the nanowire-nanoantenna system. d is 50 nm, and the wavelength of incident light is 660 nm.

Download Full Size | PDF

As we noted in Fig. 2, the field enhancement in the nanowire-nanoantenna coupled system is larger than in a system with only a nanoantenna. We believe that this is because in the nanowire-nanoantenna system, the enhancement is achieved not only by simple plasmonic enhancement, but also by the breakdown of the nanowire suppression effect. As described above, the photonic density of states for an E_// field in subwavelength nanowires is significantly reduced. However, when the same nanowire is placed in the antenna gap, this field suppression is relaxed. This means that there is an extra field enhancement for nanowires if we compare the field intensity with and without an antenna. Here we explain this mechanism in more detail. The suppression effect is realized by the E-field continuity condition at the abrupt nanowire-air boundary. This continuity condition is strongly affected when a metallic nanoantenna is placed in the vicinity of the nanowire-air boundary. The greater enhancement of the nanowire-nanoantenna system observed in Fig. 2 strongly indicates that it is in part due to the breakdown of the suppression effect. In other words, the interplay between the plasmonic enhancement for antennas and the field suppression effect for nanowires leads to the increased field enhancement. A combination of nanowires and metals in a somewhat similar configuration was also examined in Refs [42,48]. We guess that the enhancement in their experiments could also involve some degree of extra enhancement, but it was not discussed in those reports. Next, we investigated the emission characteristics of this interesting system fabricated by the nanomanipulation technique.

2.3 Emission characteristics of nanowire-nanoantenna system

First, we investigated a gold nanoantenna without a nanowire by using CL to confirm that there is a resonant plasmonic mode around 650 nm for the bowtie nanoantenna. It is well known that CL measurement enables us to observe the plasmonic mode directly [49,50]. We measured a nanoantenna without a nanowire at room temperature. We excited the plasmonicmode with an electron beam and observed the light emission from the excited area as shown in Fig. 4. We selected the wavelength using a band-pass filter with a center wavelength of 650 nm and detected the emission with a photomultiplier tube (PMT). In the CL mapping image [Fig. 4(b)], we observed a large emission at each corner of the nanoantenna. The observed mode is very similar to our numerical calculation shown in Fig. 3(a). On the other hand, we observed a uniform emission from the entire gold nanoantenna region using a filter with a center wavelength of 500 nm [Fig. 4(a)], which could be explained by the emission from the gold itself [51]. When we used a filter with a center wavelength of 800 nm [Fig. 4(c)], the observed emission was weak compared with that in Fig. 4(b), and the contrast was also low between the body and the corners. The quantum efficiency of PMT is dependent on wavelength at wavelengths between 500 and 800 nm, but the value at a wavelength of 800 nm is about two-thirds that at a wavelength of 500 nm. In addition, cathode radiant sensitivity is also dependent on wavelength and has its maximum value at a wavelength of 800 nm within the same wavelength region. Therefore, these results indicate that the nanoantenna has a resonant plasmonic mode at a wavelength of around 650 nm, which agrees well with the wavelength dependence of (|E_a|/|E₀|)² in Fig. 2.

 

Fig. 4 CL intensity mapping images around the nanoantenna measured at room temperature. The emission was filtered with a band-pass filter (the bandwidth was 50 nm). The center wavelengths of the filter were 500 (a), 650 (b), and 800 nm (c). We excited the plasmonic mode with an electron beam with an acceleration voltage of 15 kV and a current of 18 nA. Scale bar, 100 nm.

Download Full Size | PDF

Next, we investigated the PL characteristics of the fabricated nanowire-nanoantenna sample (d = 60 nm) shown in Fig. 1(c) at a temperature of 80 K. We mounted the sample on a scanner installed in a cryostat. An achromatic lens with a numerical aperture of 0.82 and a working distance of 0.4 mm, which was placed inside the chamber of the cryostat, focused light from laser diodes. The PL spectra were collected with a spectrometer and a charge-coupled device camera. In the PL measurement, we used another x-y coordinate system as shown in Fig. 5(a), where the antenna size is enlarged for clarity. We also defined the polarization parallel to the x (y) direction as E_⊥ (E_//). Figure 5(b) shows measured PL spectra for E_⊥ excitation when a continuous-wave (CW) laser operating at 636 nm was focused around the antenna gap (x ∼−0.3 μm) and on the nanowire far from the nanoantenna (x ∼2.2, 3.6 μm). The PL spectra are normalized at the maximum intensity. Note that the emission wavelengths are centered at about 875 nm, and the spectrum shapes are almost same at all three positions.

 

Fig. 5 (a) Coordinate system for PL measurement, where the antenna size is increased for clarity. x and y are the moving directions of the scanning stage, and we mounted the sample on the scanning stage with y parallel to the nanoantenna. The origin is set at the antenna position. (b) PL spectra measured for E_⊥ excitation. The sample temperature was 80 K, the excitation wavelength was 636 nm, and the excitation power was 160 μW. (c)-(f) Mapping images of normalized PL intensity I/I_R under various excitation and detection polarization settings: (c) E_⊥ (perpendicular to the nanoantenna) excitation and E_⊥ detection, (d) E_// (parallel to the nanoantenna) excitation and E_⊥ detection, (e) E_⊥ excitation and E_// detection, and (f) E_// excitation and E_// detection. The PL intensity is normalized by the intensity at a fixed reference position R (I_R) in each set of data. The excitation wavelength was 636 nm, and the excitation power was 160 μW. For this measurement, we used a band-pass filter with a bandwidth of 15 nm and a center wavelength of 875 nm to detect the emission from the nanowire. (g)-(j) Mapping images of I/I_R with excitation at a wavelength of 532 nm. The polarization of the excitation and the detection in (g)-(j) correspond to (c)-(f), respectively. The excitation power was 80 μW.

Download Full Size | PDF

Then, we performed PL intensity mapping of the nanowire-nanoantenna system to examine the enhancement effect of the nanoantenna. We used a set of CW lasers with wavelengths of 532 nm (far from the resonant wavelength of the nanoantenna), and 636 and 691 nm (near the resonance). The emission was detected with a Si avalanche photodetector (APD). In addition, we inserted another polarizer in front of the APD to select the emission polarization (E_⊥, E_//).

We first show the mapping of the PL intensity (I) excited by the laser with a wavelength of 636 nm. Figures 5(c)-5(f) show mapping images of normalized PL intensity I/I_R, where I_R is the intensity at a fixed reference position R that is sufficiently far from the antenna position, with various excitation polarization (P_ex) and emission polarization (P_em) settings: (c) (P_ex, P_em) = (E_⊥, E_⊥), (d) (P_ex, P_em) = (E_//, E_⊥), (e) (P_ex, P_em) = (E_⊥, E_//), and (f) (P_ex, P_em) = (E_//, E_//). The normalized PL intensities at the antenna position A (x = y = 0), I_A/I_R, in Figs. 5(c)-5(f) are 0.9, 1.5, 1.4, and 6.1, respectively. In Fig. 5(c), P_ex and P_em are E_⊥. The calculation results [Fig. 2] show that the E_⊥ field is not enhanced around the antenna gap by the plasmonic resonance. Thus, we could assume that Fig. 5(c) shows the emission from the bare nanowire. On the other hand, the E_// field is enhanced around the antenna gap, and thus the excitation and emission efficiencies are enhanced in Figs. 5(d) and 5(e), respectively. In Fig. 5(f), the PL intensity is strongly enhanced by the double enhancement in the excitation and emission efficiencies.

Figures 5(g)-5(j), on the other hand, show mapping images of I/I_R excited by a laser with a wavelength of 532 nm, where (P_ex, P_em) is (E_⊥, E_⊥), (E_//, E_⊥), (E_⊥, E_//), and (E_//, E_//), respectively. The I_A/I_R values in Figs. 5(g) and 5(h) are 0.9 and 0.8, respectively, and the mapping images are similar. This indicates that there is no enhancement of the excitation efficiency. These results are in good agreement with the calculation results [Fig. 2], where the electric field is not enhanced at wavelengths much shorter than the plasmonic mode. In Figs. 5 (i) and 5(j), I_A/I_R is 1.6, and the emission efficiency is still enhanced.

Since the I_A/I_R ratio could be affected by the arbitrariness of the reference position R, we next examine the enhancement ratio using a different procedure. We first show cross-sectional plots of I along the nanowire at 636-nm excitation. Figures 6(a)-6(d) show I along a white solid line drawn in Fig. 5(c) as a function of x, where (P_ex, P_em) is (E_⊥, E_⊥), (E_//, E_⊥), (E_⊥, E_//), and (E_//, E_//), respectively. Here, we label I in Figs. 6(a)-6(d) as I₀, I_ex, I_em, and I_ex-em, respectively (we show the E_// polarizations as subscripts). We then plotted the relative intensity I_α/I₀ (α = ex, em, ex-em) as shown in Figs. 6(e)-6(g). In Fig. 6(e), we show I_ex/I₀ at a wavelength of 532 nm (gray solid line) as a reference, and it is almost flat. This indicates that I_ex is not enhanced around the antenna position. The calculated results [Fig. 2] also show that there is no enhancement at a wavelength of 532 nm. On the other hand, I_ex/I₀ at a wavelength of 636 nm (red solid line) has a pronounced peak around the antenna position. If I_ex is not enhanced around the antenna position, I_ex/I₀ is flat on the nanowire as with 532-nm excitation. Thus, we draw an interpolated line of I_α/I₀ (a broken line). Here we assumed that I_α/I₀ is linear at −2.0 ≤ x ≤ 2.0 without the nanoantenna. The ratio of I_α/I₀ to the interpolated relative intensity (I_α/I₀)' at x = 0 is the enhancement ratio f_α at the antenna position: f_α = (I_α/I₀)/(I_α/I₀)'. The obtained excitation enhancement ratio f_ex [Fig. 6(e)], emission enhancement ratio f_em [Fig. 6(f)], and double enhancement ratio f_ex-em [Fig. 6(g)] are 1.6, 1.3, and 6.3, respectively. These values are in good agreement with I_A/I_R in Fig. 5. With the exception of the antenna region, I_ex/I₀ is about 0.15 at 636 nm excitation [Fig. 6(e)]. This ratio is much smaller than unity, which is due to the field-suppressing effect of the nanowires as mentioned above. In addition, I_ex/I₀ is closer to unity at 532-nm excitation, which agrees with fact that the field-suppressing effect should become smaller when the wire diameter divided by the wavelength becomes larger [52].

 

Fig. 6 (a)-(d) Cross-sectional plots of PL intensity I along the nanowire as a function of x (the cutting line is shown as a white solid line in Fig. 5(c)) at an excitation wavelength of 636 nm. I₀ (a), I_ex (b), I_em (c), and I_ex-em (d) correspond to Figs. 5(c)-5(f), respectively. (e)-(g) Relative PL intensity (I_ex/I₀, I_em/I₀, I_ex-em/I₀) at excitation wavelengths of 532 (gray solid line), 636 (red solid line), and 691 (black solid line) nm. The excitation power was 160 μW at an excitation wavelength of 691 nm.

Download Full Size | PDF

In our nanowire-nanoantenna system, the laser spot is much larger than the plasmonic mode. Although the emission is enhanced at the antenna position for the resonant polarization E_//, the outer region of the plasmonic mode within the laser spot gives the emission of the bare nanowire. Thus the enhancement ratio f_α is reduced from the intrinsic enhancement factor. Here we estimate the intrinsic enhancement factor from the observed enhancement ratio f_α. The diameter of the laser spot (l₀) is about 1.3 μm for 636-nm excitation, and the width of the plasmonic mode along the x-direction (l_a) is about 60 nm [Fig. 3(d)]. The intrinsic enhancement at the antenna position is given by the product of the excitation enhancement factor η_ex and the emission enhancement factor η_em. In the outer region, we assume that there is no enhancement. Thus, f_α and η_exη_em are connected by

In Fig. 6(e), f_ex is 1.6, and η_em should be unity. When using these values, η_ex is estimated to be 14. In Fig. 2, the calculated enhancement factor (|E_a|/|E₀|)² (case (ii)) is about 20 around the peak wavelength, which is fairly close to 14. In Fig. 6(f), on the other hand, η_ex should be unity. Then, η_em is estimated to be 7.6. In this work, we did not directly calculate the emission efficiency by numerical simulation, but it is well known that the transmitter efficiency is the same as the receiver efficiency in classical antennas. If the emission is the reverse process of the excitation in our nanowire-nanoantenna system, η_em is expressed by (|E_a|/|E₀|)². The calculated (|E_a|/|E₀|)² at a wavelength of 875 nm is around 10 in Fig. 2, which is close to 7.6. Then, we investigate the double enhancement factor η_ex-em in Fig. 6(g). In this case, η_exη_em is replaced by η_ex-em in Eq. (1), and η_ex-em estimated from f_ex-em is 110. This value is almost the same as the product of η_ex and η_em obtained separately in Figs. 6(e) and 6(f), which supports the validity of our estimation.

We also show I_α/I₀ at a wavelength of 691 nm (black solid line) in Figs. 6(e)-6(g). For 691-nm excitation, l₀ was 1.4 μm, and η_ex, η_em, and η_ex-em were 12, 3.6, and 68, respectively. These enhancement factors are slightly small when compared with those for 636-nm excitation. This indicates that 691 nm is further from the resonant mode compared with the value of 636 nm in our structure. In addition, I_ex-em for 532-nm excitation is shown in Fig. 6(g) as a gray solid line, and the estimated η_ex-em was 10 (l₀ was 0.9 μm). This enhancement is given by the emission enhancement because the excitation efficiency is not enhanced at a wavelength of 532 nm.

These results show that the PL from the subwavelength InP nanowire was enhanced at the gap of the plasmonic bowtie nanoantenna, and a very large field intensity enhancement of 110 was confirmed experimentally. The numerical simulation showed that the enhancement is caused by a combination of the field enhancement in the plasmonic resonance and the breakdown of the field suppression in the nanowire.

3. Conclusions

In this work, we developed an efficient way to place a subwavelength InP nanowire directly in the gap of a bowtie nanoantenna with a nanomanipulator installed in a FIB system. This method enables us to place a single nanowire in an arbitrary position by controlling the manipulator under in situ SEM observation. We have used this method to fabricate a nanowire-nanoantenna coupled system. Strong PL intensity enhancement was observed when the excitation laser irradiated the gap position and its wavelength was close to the plasmonic resonance of the nanoantenna, which was independently confirmed by a CL measurement. Quantitative studies of the absolute value of the enhancement revealed that the PL intensity of the nanowire in the polarization parallel to the nanoantenna was enhanced by a factor of 110. The overall observation agrees very well with a numerical simulation of the nanowire-nanoantenna coupled system by FEM method. This analysis clarified that the PL intensity of the bare nanowire when the polarization was perpendicular to the nanowire was strongly suppressed by the field suppression effect, but it was greatly enhanced when the nanowire was placed in the antenna gap. This enhancement is larger than the bare plasmonic enhancement for the nanoantenna. This difference can be accounted for by a breakdown in the field suppression effect as a result of the presence of the nanoantenna. That is, there is an interesting interplay between the plasmonic enhancement and the nanowire field suppression effects in our nanowire-nanoantenna coupled system, which leads to greater emission enhancement. This unique feature may be useful for artificially controlling the emission properties of semiconductor nanowires with a much greater contrast than the simple plasmonic effect. Considering the fact that various functionalities can be implemented in semiconductor nanowires, our coupled system may be promising for various optical devices where the light-matter interaction is greatly enhanced or suppressed by manipulating a nanowire-nanoantenna coupled system.