Introduction

Lead halide perovskites, with their outstanding carrier transport characteristics, high emission quantum yields, tunable bandgaps, and large absorption coefficients1,2,3,4, have attracted considerable interest for applications across a range of technologies from solar energy conversion1,2,5 to light-emitting diodes (LEDs)6. Recently the energy conversion and LED efficiencies of lead halid perovsktes reached 23.2% and 20.7%, respectively7,8. In perovskites, electrons and holes are confined to inorganic [PbX6]4− (X = Cl, Br, I) octahedral networks, which leads to enormous Coulomb interactions between electrons and holes and thereby, strong excitonic effects9,10. In low-dimensional perovskite structures, i.e., nanocrystals (NCs) or nanoplatelets (NPLs), the electron–hole interactions are further enlarged as a result of the increased spatial overlap between electrons and holes. For instance, the exciton-binding energy of CsPbBr3 bulk crystals is 40 meV11, while the value increases to 120 meV in NCs12. Furthermore, CsPbBr3 perovskite NCs exhibit high photoluminescence (PL) quantum yield of ~90%13. The large exciton-binding energy and high-quantum efficiencies of CsPbBr3 perovskite NCs hold the key to the development of excitonic and quantum devices due to their stable excitons at room temperature. Furthermore, excitons in perovskite NCs may be coupled with free carriers, forming trions14. A trion can decay to a single electron (hole) by the emission of circularly polarized light, and it can achieve population inversion and laser action. The trion state shows a lower threshold compared with the neutral state, which requires a higher excited state to achieve optical gain. This is advantageous to realize low-threshold lasers15.

A biexciton is formed by two free excitons in condensed exciton systems16, two photon absorption17, or excitation from the single exciton state to the biexciton state. Biexcitons are of fundamental interest and practical importance for quantum information and computation due to their overwhelming advantage for constructing of coherent combinations of quantum states. The coherent control of biexcitons via their four physically distinguishable quantum states can be applied to basic quantum operations, i.e. two-bit physical quantum and conditional quantum logic gates18,19,20. If fine structure splitting of the exciton state is smaller than the natural linewidth, the two indistinguishable radiative decay pathways of a biexciton can render a source of polarization entangled photon pairs. Furthermore, the nonlinear optical nature of the biexciton process can generate probabilistic number of entangled-photon pairs per excitation cycle21, which is similar to the standard entangled source parametric down conversion22 and four-wave mixing sources23. Moreover, light sources based on biexciton may achieve no blinking and high saturation intensity24,25. These intrinsic outstanding physical properties make biexciton interesting for inherent and coherent light sources including lasers, light emitting diodes, etc.26,27. Till now, biexcitons have been realized in semiconductor heterostructures28, NCs26, and two-dimensional semiconductors16,29, etc. Due to the small biexciton-binding energy and large Auger effect, biexcitonic effects have only been realized either under intense pumping by short pulsed lasers (~MW/cm2) or in cryogenic conditions, which severely limits their practical applications. Resonant excitations can increase the probability of biexciton transition. For example, in quantum wells, biexciton emission was observed under a continuous wave (CW) excitation beam with a power of ~70 mW/cm2, but only below liquid helium temperature due to the small biexciton-binding energy30,31. So far, CW excitation of biexciton at room temperature, which has fundamental practical significance is extremely challenging.

Plasmonic nanogap structures have been extensively explored in quantum electromagnetic dynamics due to their extraordinary capabilities for tailoring strong and weak light–matter interactions in the deep-subwavelength regime32,33,34. In the last decade, plasmonic nanogap have been widely used to enhance a variety of linear and nonlinear optical processes including emissions, hot carrier generation, Raman, high harmonic generation, etc.35,36,37,38. Near a metal structure, the increasing local density of states in a plasmonic nanostructure can shorten the fluorescence lifetime and greatly suppress Auger recombination, even to the point where Auger recombination is negligible39. When coupled to a metal structure, quantum dots (QDs) show super-Poissonian statistics of photon emissions, which is different from the behavior of QDs on quartz substrates that exhibit photon antibunching as single quantum emitters40. In this case, QDs exhibit strong photon bunching which leads to multiexciton emissions41. This provides a way to achieve biexciton emissions with a significantly lower pump energy, which consequently highly suppresses the biexciton Auger recombination with decreasing pump density42,43,44. The interactions between plasmons and QDs have been widely discussed41,45,46. Surface plasmons are regarded as continuous energy states while quantum dot excitations are discrete energy levels. The coupling between continuous and discrete energy states leads to Fano effect47,48. With increasing incident light intensity and enhanced coupling strength, the two-photon process will take the place of a single photon process which leads to nonlinear Fano effect45,49.

In this work, we demonstrate strong biexciton emissions from CsPbBr3 NPLs with continuous-wave excitation at room temperature, and the biexcitonic behavior is validated through experiments and simulations. By utilizing a Ag nanowire-Ag film nanogap structure, the exciton occupancy for biexciton emissions of CsPbBr3 NPLs is reduced ~106 times. The extremely large enhancement of biexciton emissions is attributed to the nonlinear Fano resonance between the biexciton and surface plasmon cavity modes. Our experimental results provide new pathways to develop high efficiency non-blinking single photon sources with spectral filtering of biexciton emission, entangled light sources, lasers, and high-efficiency LED based on biexciton states.

Results

Schematic of biexciton and plasmonic structure

Figure 1a shows the energy level diagrams for the excitation and decay of a single exciton. The single exciton bright state can decay to the ground state by emitting one photon, while the dark state decays to the ground state mainly by nonradiative processes. Two states are coupled by a spin–flip process. For perovskite NCs, the dark state exhibit microsecond-order decay times with a lower energy50. The dark state energy may be higher than that of the exciton due to the large Rashba spin-orbital effect51. Figure 1b shows the energy level diagrams of a biexciton. Biexciton is composed of two bright excitons and can decay to either of two single exciton states (bright state) accompanied by the emission of one photon32. The biexciton emission energy ħωxx is determined by the energy gap between the biexciton energy, Exx and single exciton emission Ex via ħωxx = Exx‒Ex. Therefore, when two excitons bind together to form one biexciton, the energy of the whole system decreases by Δxx = ħωx‒ħωxx, which is the so-called biexciton-binding energy. Figure 1c shows the two processes for the plasmon-enhanced biexciton state: NPLs absorb one electron–hole excitation to the exciton state (I) and then reach the biexciton state (II) by absorbing another one, while the electron–hole excitations in metal relax to the ground state. The plasmon excited by optical pumping decay by two following channels: (1) via transformation into photons and (2) via nonradiative decay into electron–hole excitations. The latter fall into two categories: intraband excitations and interband excitations52. As the dephasing time of perovskite is much longer than silver nanowire53, electron–hole excitations in plasmonic cavity decay and perovskite NPLs obtain the energy. The resonant effect between the plasmonic cavity and the perovskite biexciton energy extensively promote excitation cross-section of biexciton, which leads to an increased biexciton transition. The electron–hole excitations in metallic material loses energy via electron–phonon coupling on a picosecond time scale54. The Fano resonance can be set up within 10 fs which is much faster than electron–hole excitations decay time in plasmon55. Our designed structure is shown in Fig. 1d, which is a sandwiched structure of silver (Ag) nanowire/NPLs/Ag film. A thin layer of SiO2 and perovskite NPLs were coated onto Ag film step by step. SiO2 was used to control the gap width dG between the Ag nanowire and film and also prevent the perovskite NPLs from PL quenching nearby metal surface. To reduce the scattering and radiation losses in the plasmonic nanogap due to surface discontinuity, the growth condition of the SiO2/Ag film was optimized with a surface roughness ~1.1 nm (Supplementary Note 1). Figure 1e shows the schematic of the biexcitons of CsPbBr3 PLNs in the plasmonic nanogap. Perovskite NPLs (Supplementary Note 2) were deposited onto the SiO2 layer by solution-processed spin-coating methods. To isolate the perovskite NPLs with Ag nanowire, the 4-Methyl-1-acetoxycalix[6]arene (4M1AC6, 0.5% in chlorobenzene) polymer film was spin-coated onto the CsPbBr3 NPLs. The thickness of the perovskite NPL layer and polymer film was 13 nm, as measured by ellipsometry (Supplementary Note 3). A cross-section SEM image (Fig. 1f) shows that the perovskite NPL layer was uniform in scale of millimeters.

Fig. 1
figure 1

Schematic of biexciton and formation process. a Schematic diagram of exciton and biexciton transitions. Exciton has a bright state which can arrive in the ground state by radiation or non-radiation transitions. The corresponding dark state coupled though a spin–flip process arrives in the ground state mainly by non-radiation transitions. The dark state energy may be higher than that of the exciton due to the large Rashba spin-orbital effect. b The biexciton levels are formed by a four-level system: biexciton level, two bright exciton levels, and ground state. The two bright exciton levels are separated by fine structure splitting. The biexciton decays through a cascade process of either emitting two horizontally or vertically polarized photons. c Schematic diagram of biexciton formation process with nonlinear Fano effect. Nanoplatelets (NPLs) absorb one electron–hole excitation to the exciton state (I) and then reach the biexciton state (II) by absorbing another one, while the electron–hole excitations in metal relax to the ground state. d Schematic view of the structure. The cavity is composed of a Ag nanowire and Ag substrate separated by a SiO2 gap of 5 nm and perovskite nanoplatelets. e Schematic view of the exciton and biexciton in the cavity. The NPL is covered with 4-Methyl-1-acetoxycalix[6]arene molecular. f Cross-sectional scanning electron microscope (SEM) image of the sample. The scale bar is 20 nm

Full size image

Crystal structure of perovskite NPLs

Transmission electron microscopy (TEM) images of the CsPbBr3 NPLs are presented in Fig. 2a. The dispersed CsPbBr3 NPLs had an average lateral size of 30 nm. HRTEM images of the CsPbBr3 NPLs in Fig. 2b, combined with single photoluminance (PL) peak in solution (Supplementary Note 4), confirms that the CsPbBr3 NPLs possessed good crystallinity. The Fourier transform diffraction spot is illustrated at the upper right corner, which shows the perovskite NPLs cubic crystal structure with a 5.92 Å lattice constant. X-ray diffraction (XRD) patterns within the 2θ range of 10–50° were recorded at room temperature and are shown in Fig. 2c. The XRD pattern of the CsPbBr3 NPL show strong diffraction peaks, which were assigned to the pure cubic crystal structure13 consistent with the HRTEM Fourier transform diffraction spot. The thicknesses of the CsPbBr3 NPLs are shown in Fig. 2d with the atomic force microscopy (AFM) image on SiO2/Si substrate. A single CsPbBr3 NPL was characterized in Fig. 2e with a height ~4 nm and a lateral size ~30 nm corresponding to the TEM lateral size shown in Fig. 2b. Thickness statistics of the CsPbBr3 NPLs are shown in Fig. 2f with an average thickness ~3.8 nm and an exciton-binding energy of ~120 meV (Supplementary Note 5). Combined with the excitonic Bohr radius of 7 nm13, the NPLs exhibited a strong confinement in the z direction.

Fig. 2
figure 2

Crystal structure of perovskite nanoplatelets (NPLs). a Transmission electron microscopy (TEM) image of CsPbBr3 NPLs. The scale bar is 50 nm. b High-resolution transmission electron microscopy (HRTEM) images of CsPbBr3 NPLs and the Fourier transform diffraction spot. The scale bar is 5 nm. c X-ray diffraction (XRD) pattern of CsPbBr3 NPLs. d Atomic force microscope (AFM) image (500 nm × 500 nm) of CsPbBr3 NPLs on SiO2/Si substrate. The scale bar is 100 nm. e The single CsPbBr3 NPL shown in a with a height ~4 nm and a lateral size ~30 nm corresponding to TEM size shown in b. The scale bar is 10 nm. f Thickness statistics of CsPbBr3 NPLs which shows an average thickness ~3.8 nm

Full size image

Cavity-coupled perovskite NPLs fluorescence

To probe the biexciton effect in perovskite NPLs, PL of the perovskite NPLs on and off the plasmonic nanogaps was performed with a CW laser (Fig. 3, wavelength: 405 nm; spot diameter: 2 μm). The PL spectrum far from the plasmonic nanogap (P1, blue line) was assigned to single exciton recombination of CsPbBr356. It shows a single symmetric peak located at 500 nm, which corresponds to a free exciton when the excitation power P is 1.6 μW. In contrast, on the nanogap (P2, red line), a new peak arouse at 520 nm with an intensity much higher than that of single exciton emission. Although the nanowire is much smaller than the excitation spot diameter. Two obvious phenomena can be seen: (1) the strong biexciton emission is observed, which comes from the perovskite NPLs in the plasmonic cavity due to nonlinear Fano effect; (2) the exciton emission signal on silver nanowire is about three times (Fig. 3) than that off nanowire, which is attributed to the perovskite NPLs in the plasmonic cavity area due to strong field enhancement. These results suggest that the plasmonic nanocavities yield a strong effect in the biexciton emission process (the corresponding linear coordinate PL spectrum is shown in Supplementary Note 6). The energy difference between the single exciton (500 nm, 2.480 eV) and new emission peak (520 nm, 2.385 eV) was ~95 meV, which is almost the same as the biexciton-binding energy of CsPbBr3 quantum dot reported previously (~100 meV)27.

Fig. 3
figure 3

Spectral characteristics of cavity-coupled perovskite nanoplatelets (NPLs) fluorescence. Perovskite NPL off-cavity fluorescence spectrum (P1) shows a single symmetric exciton peak at 500 nm, while in the on-cavity spectrum (P2), a biexciton peak at 520 nm under continuous wave radiation emerged. The binding energy of the biexciton is about 95 meV. The inset is an scanning electronic microscopy (SEM) image of a cavity (scale bar 1 μm), illustrating the location of the incident laser spot for the on-cavity (P2) and off-cavity (P1) measurements. The error bar corresponds to the standard deviation of the measurements

Full size image

Power law of perovskite NPLs PL integral intensities

Under full thermal equilibrium conditions of the biexciton recombination process, the emission intensity of biexciton Ibx is proportional to the square of the single exciton emission intensity Iex57,58. And biexciton lifetime is about half of single exciton lifetime. Excitation power-dependent PL and time-resolved PL (TRPL) spectroscopy were conducted to confirm the occurrence of biexciton. Figure 4a, b shows the power-dependent PL spectra on and off the Ag nanowire on the Ag film, respectively. Figure 4c shows the integrated emission intensity of exciton (Iex) and biexciton (Ibx) as a function of the fluence intensity P. As the pumping power increased from 20 to 130 μW, the PL spectrum away from the plasmonic nanogap (Fig. 4a) showed one emission peak at 500 nm with the intensity I exhibiting a linear dependence on the pumping power P (Fig. 4c, purple), which is ascribed to single exciton recombination. However, on the plasmonic nanogap, a 520 nm peak due to biexciton emissions emerged, with the intensity Ibx growing more rapidly than that of the single exciton (Fig. 4b). Although the biexciton emission peak onset emerged when P > 100 mW/cm2 (0.5 μW, ~20 μm) (Supplementary Note 7), we performed emission spectroscopy with P > 500 W/cm2 (20 μW, ~2 μm) to lower the fitting error, when separating the power dependence of the biexciton emission from that of the single exciton emission. Figure 4c shows the integrated emission intensity of the exciton (Iex) and biexciton (Ibx) as a function of the fluence intensity. The power dependence of the biexciton emissions could be described adequately by a superlinear function with a power law of k = 1.83. The power law of the biexciton (1.83) was about 1.91 times greater than that for the single exciton (0.96), strongly suggesting the occurrence of biexciton emissions.

Fig. 4
figure 4

Photoluminescence spectra and integral intensities of perovskite nanoplatelets (NPLs) for different pump fluences. a PL spectroscopy of the nanogap only showed exciton recombination peaks as the pumping power increased from 20 to 130 μW. The emission intensity is linear with pumping power (c, purple). b The biexciton emission peak of 520 nm at high excitation fluence. With increasing pumping power, the biexciton, however, grows much faster than that of single exciton. The same color map is used to make a clear comparison. c A log–log scale was adopted to plot the integrated emission intensity strength, Ip, as a function of the exciton emission strength P. The purple line is a power-law fit Ip = Pk, with a power-law constant k1 = 0.96, while the red line k2 = 1.83. The ratio l = k2/k1, has a value of 1.91. This provides strong evidence that a shows exciton luminescence while b shows biexciton fluorescence. The error bar corresponds to the standard deviation of the measurements

Full size image

Lifetimes of excited states

Further, TRPL spectroscopy was conducted to probe the exciton and biexciton dynamics. Quantitatively under the condition of thermal equilibrium, the time evolution of the exciton and biexciton can be modeled with the transition function59 \(n_{{\mathrm{{bx}}}}\sim n_{{\mathrm{{bx}}}}^0{\mathrm{{e}}}^{ - 2t/\tau _{{\mathrm{{ex}}}}}\), where exchange interactions between excitons with different spin are ignored. In other words, the biexciton decay rate is indeed about twice that of single exciton, as suggested in bulk semiconductor60, QDs59, quantum wells61, and 2D semiconductors62. Figure 5a, b show the TRPL spectra of the CsPbBr3 NPLs on (a) and off (b) the plasmonic nanogap. To avoid any biexciton effects away from plasmonic nanogap under high pumping conditions, the excitation fluence of the femto-second pulsed laser was ~0.1 nJ/cm2, and thereby, the PL spectra showed only single exciton emission. The single exciton recombination curve could be well fit by a single exponential decay function with a time constant of τex = 1170 ± 10 ps. The TRPL spectroscopy of the biexciton emission was pumped with a power ~70 nJ/cm2 and could be well-fit by a single exponential decay curve, suggesting the emissions on the plasmonic nanogap were dominated by biexciton recombination. The lifetime of the biexciton τbx = 510 ± 5 ps, was half of single the exciton lifetime (1170 ± 10 ps), which is consistent with the transition properties of the biexciton states as discussed above. Therefore, we can conclude that the 520 nm peak arising on plasmonic nanogap was due to biexciton recombination.

Fig. 5
figure 5

Dynamics of excited states. a Time-resolved photoluminance of biexciton which can be seen as a single decay process shows that the lifetime is τbx = 510 ± 5 ps. b Single index decay of exciton and fit with a lifetime of τex = 1170 ± 10 ps represents the electron–hole recombination dominating the photoluminance. The lifetime of the exciton is approximately twice that of the biexciton. Here, the extracted ratio ~2.29 is consistent with the theoretical result. The error bar corresponds to the standard deviation of the measurements

Full size image

Fano effect of silver nanowire and perovskite NPLs

In the plasmonic nanogap, biexciton emission was observed when the pumping fluence exceeded 100 mW/cm2 (Supplementary Note 7). The estimated exciton occupancy 〈N〉~10−5 was much lower than 1, which indicates that the biexciton emission were not produced by exciton–exciton bounding in this nanogap region. However, the enhancement factor should be much larger than the local field enhancement factor (<100, Supplementary Note 8) in the plasmonic nanogap. This indicates that the enhanced biexciton emission was not mainly due to local field enhancement. Moreover, the Purcell effect of the NPLs was observed in the plasmon nanocavity and the factor is ~2.5 (Supplementary Note 9). However, it did not play an important role in biexciton formation process and can be negligible.

Biexciton forms mainly through the following three methods: (1) two photon absorption, (2) exciton–exciton bounding in condensed exciton environment, and (3) resonant excitation from a single exciton. The first two pathways required intense pumping, which is not feasible under CW excitation conditions. To confirm that the plasmon mode affected the biexciton between the Ag nanowire and Ag film, the emission polarization behavior was measured. As shown in Fig. 6a, the biexciton emission was strongest when the excitation polarization direction was parallel to the long axis of the nanowires (0°, 180°) and weakest when the excitation polarization direction was perpendicular to the long axis (90°, 270°). However, the emission intensity of the perovskite NPLs away from the Ag nanowire was almost the same (Supplementary Note 10). The polarization resolved emission properties of the 520 nm peak suggests the modulation of the Ag nanowire’s plasmonic mode which is transverse magnetic mode. Moreover, an asymmetric peak appeared with a zero detuning position around the perovskite biexciton resonant energy, resulting from Fano interference between biexciton resonance excitation and background continuum states excitation of surface plasmon modes as shown in Fig. 6b. The scattering spectra were fitted with Fano-line profiles, \(\sigma _{{\mathrm{sca}}}{\mathrm{ = A}}\left[ {\left( {q{\mathrm{ + \Omega }}} \right)^2{\mathrm{ + B}}} \right]{\mathrm{/}}\left( {{\mathrm{1 + }}\Omega ^2} \right)\), where q is the Fano parameter, \({\mathrm{\Omega = 2}}\left( {E - E_0} \right){\mathrm{/}}\Gamma\), \(\Gamma\) is the resonant width and E0 is the resonant energy, which is centered at the biexciton emission position63. The fitted Fano factors were −0.25 and −0.2, when the polarization was parallel and vertical to long axis of nanowire, respectively. The scattering lineshape was simulated by the finite-difference time-domain (FDTD) method, as shown in Fig. 6c. It is noticed that the nonlinear Fano resonance lineshape was only observed in the scattering spectra of the Ag nanowire–NPLs–Ag film system with strong biexciton emissions, which suggests that the exciton–plasmon energy transfer plays a vital role in the formation of biexciton. Plasmon resonant energy transfer can only occur when the plasmon peak overlaps with the biexciton states (Supplementary Note 11). Without energy transfer, the NPLs in the gap only show exciton emission (Supplementary Note 12). The nonlinear interaction witnessed as the Fano resonance contributed to the biexciton emission.

Fig. 6
figure 6

Fano effect of silver nanowire and perovskite nanoplatelets (NPLs). a Polarization of the biexciton shows the biexciton is along the Ag nanowire. b Plasmon resonant spectra of the nanogap shows a single symmetrical peak shown as a green dotted line. The plasmonic nanogap with perovskite NPLs shows a symmetrical peak. Red and blue dotted lines show the scattering spectra with parallel and vertical polarizations, respectively. c Finite-difference time-domain (FDTD) simulation results of b. The error bar corresponds to the standard deviation of the measurements

Full size image

The nonlinear Fano resonance suffers from the distance of metal structure with NPLs and external field. With a gap of 13 nm and external field of 0.1 W/cm2, the interaction can be viewed as linear interaction dominant, which can be seen as the threshold of biexcion emissions45,49. Due to the large binding energy of the CsPbBr3 NPLs, room temperature biexciton emissions off the nanogap could be excited only by fs laser pulse of amplifier (1 kHz, 80 fs) and the pumping fluence was on the magnitude of ~1011 W/cm2 with an estimated average NPL exciton occupancy 〈N〉~30 (Supplementary Note 13). The threshold in our system was much higher than 〈N〉~1, since a larger pump intensity leads to ionization64 and a faster Auger recombination rate with a very rapid growth of the carrier density65. To make sure the NPLs is not damaged, the increasing and decreasing power-dependent PL is measured (Supplementary Note 14). Moreover, the power-dependent PL is also pumped with fs laser pulses of oscillator (80 MHz, 100 fs), however, biexciton is not emerged until NPLs is damaged (Supplementary Note 15). Therefore, the exciton occupancy to generate biexciton in the plasmonic nanogap was reduced by ~106 times when comparing with fs laser pulse of amplifier.

Discussion

A resonant excitation process is proposed to explain the huge biexciton effect driven by surface plasmon cavity. As shown in Fig. 1b, since the single exciton–biexciton transition energy was nearly the surface plasmon energy, a nonlinear process with energies ω and ωʹ can be absorbed though resonant energy transfer from surface plasmons to NPLs, leading to the excitation of biexciton and annihilation of surface plasmons. The interference of the two excitation processes results in the nonlinear Fano resonance lineshape in the scattering spectra, where an asymmetric peak confirmed the process occurred. The surface plasmon-driven mechanism is similar to the biexciton generation by a resonant pump-probe in single carbon nanotube66. The highly populated energy of plasmon was confirmed by zero detuning position of scattering spectra, which is almost overlapped with the biexciton transition energy. The resonance effect extensively promoted the excitation cross-section of biexciton from exciton states.

In conclusion, we found clear evidence for the presence of high-efficiency biexciton generation in perovskite NPLs in a metallic nanostructure under CW pumping at room temperature. The observation of the biexciton was a consequence of the clear emission spectra besides exciton with a binding energy ~95 meV, which was verified by the activation energy. The fluorescence power law and lifetime provided further evidence of the biexciton emission. Moreover, the nonlinear Fano effects were confirmed by measuring the scattering spectrum. By fitting the spectrum with Fano parameters and FDTD simulations, we are sure that this demonstration of the four-particle complex in plasmonic nanostructure will give rise to new interesting effects, such like quantum logic gates, Bose–Einstein condensation of exciton, source of polarization-entangled photons, single-photon source and high-efficiency LEDs.

Methods

Sample preparation

The SiO2/silver substrate was made though magnetron sputtering with silver thickness of 50 nm and SiO2 thickness of 5 nm on a SiO2/Si substrate. CsPbBr3 NPLs were spin-coated on the SiO2/silver substrate at a speed of 3000 r/min. The 4-Methyl-1-acetoxycalix[6]arene (4M1AC6, 0.5% in chlorobenzene) was spin-coated on the CsPbBr3 NPLs at a speed of 5000 r/min. After waiting for about 30 min to allow the 4M1AC6 to dry, silver nanowires were spin-coated in isopropyl alcohol at a speed 2000 r/min.

SEM and TEM measurements

The samples were forced apart after liquid nitrogen treatment. The SEM equipment was a Merlin-61-53 with a working distance of 4.3 mm and a voltage of 5 kV. The perovskite NPLs for the TEM measurement were dropped onto TEM grids. High-resolution transmission electron microscopy (HRTEM) was performed with an FEI Tecnai F20 operated with an acceleration voltage of 200 kV.

Time-resolved PL spectroscopy

For time-resolved PL measurements, the excitation pulses (wavelength 400 nm) were doubled frequencies of a Coherent Mira 900 (120 fs, 800 nm, 76 MHz) and filtered by a 655 short-pass filter to generate 400 nm light. The backscattered signal was collected using a time correlated single photon counting (TCSPC, SPC-150) which has an ultimate temporal resolution of ∼40 ps. A 442 nm long-pass filter was placed before the optical fiber to filter out the residual 400 nm light.

PL measurements

The output from a 405 nm CW laser was circularly polarized by a quarter-wave plate, and focused on a sample by a microscope objective lens (×100, NA = 0.95, with spot size ~2 μm). The PL signal was then back collected by the same lens, and filtered by a long-pass filter before entering a spectrometer (PI Acton 2500i with a liquid nitrogen-cooled charge coupled device—CCD camera). The low-temperature PL was measured when the sample was in the liquid nitrogen refrigeration cryogenic instrument.

FDTD simulation

The FDTD simulations of the silver nanowire and QDs were simulated using the Lorentz model67. We modeled the CsPbBr3 nanoplatelet layer using a dielectric function as a single Lorentzian function. The dielectric constant for bulk CsPbBr3 is 3.868. The biexciton energy and plasmon resonant energy transfer were considered, while other high-order transitions were ignored. The biexciton linewidth was used as the transition linewidth. The metal dielectric function was treated as a Lorentz–Drude model.