Plasmonic-enhanced lanthanide nanotransducers for mid-infrared detection

Abstract

Advanced mid-infrared (MIR) detection underpins a wide range of applications from chemical sensing to thermal imaging. Conventional MIR photodetectors rely on narrow-bandgap semiconductors that typically require cryogenic cooling, whereas uncooled alternatives suffer from inherently low detectivity. An emerging strategy addresses this trade-off by converting MIR photons into near-infrared or visible (NIR/VIS) emission using lanthanide-based nanotransducers, enabling room-temperature operation. However, practical implementation is hindered by pronounced nonradiative quenching, which limits conversion efficiency and suppresses MIR-induced ratiometric luminescence. Here, we report a plasmonic-enhanced, self-assembled MIR detection platform based on Nd³⁺/Yb³⁺ co-doped NaYF₄ nanoparticles coupled to gold nanospheres. This platform transduces MIR signals into NIR emission detectable by low-cost silicon photodetectors, achieving a room-temperature specific detectivity of 2×10⁹ Jones at 8 μm. Localized surface plasmon resonances generate strongly confined electromagnetic fields that concurrently amplify excitation and emission, delivering a seven-fold boost in specific detectivity relative to pristine nanoparticles. Combining scalable fabrication with silicon-readout compatibility, this plasmonic-assisted approach paves the way for high-performance MIR sensing, imaging, and spectroscopy.

Lanthanide nanotransducers can convert mid-infrared light into nearinfrared signals but suffer from weak emission. Here, authors couple Nd3 + /Yb3 + -doped nanoparticles with gold nanospheres to achieve silicon-readable detection with seven-fold improvement in specific detectivity relative to the pristine nanoparticles.

Introduction

Mid-infrared (MIR) photodetection underpins critical technologies in thermal countermeasures, molecular spectroscopy, medical diagnostics, and free-space optical communication^1–6. In particular, MIR spectroscopy provides label-free interrogation of protein structures and direct identification of gas molecules through their fundamental vibrational resonances.^7–9 A major goal in this field is the development of compact, low-noise, and cost-effective MIR detectors capable of room-temperature operation. Recent advances have leveraged two-dimensional materials coupled with engineered plasmonic metasurfaces, enabling broadband and polarization-sensitive MIR detection via the photothermoelectric effect under ambient conditions^10,11. Narrow-bandgap semiconductors, such as HgCdTe (MCT), also deliver high responsivity in optimized photodiode architectures^12–15. However, these approaches encounter intrinsic limitations, including material fragility, toxicity, and the need for cryogenic cooling, which impede scalability in next-generation MIR optoelectronic systems^16,17.

An alternative route exploits spectral conversion, whereby MIR photons are upconverted into near-infrared (NIR) signals that can be detected using low-cost silicon photodetectors. This strategy enables integration with silicon electronics and provides a promising route toward room-temperature MIR detection. Conventional upconversion techniques based on bulk nonlinear crystals require phase matching and high-power lasers, limiting their practicality^18,19. Plasmon-enhanced schemes, such as surface-enhanced Raman scattering, have enabled coherent MIR detection in nanosphere-on-resonator architectures by simultaneously enhancing MIR absorption and Raman scattering cross sections^20–22. Nonetheless, requirements for complex vertical assemblies, molecule-specific vibrational resonances, and the intrinsically low quantum efficiency of Raman scattering constrain their applicability, bandwidth, and efficiency, underscoring the demand for a broadband, scalable MIR detection platform. We recently demonstrated incoherent broadband MIR sensing (4–11 μm) by tracking ratiometric changes in NIR photoluminescence (PL) from Nd³⁺-doped NaYF₄ core-shell nanoparticles under MIR modulation²³. Besides, the lanthanide transducers are also applied to realize a mid- to long-wave infrared nanoscopy on a single particle level²⁴. Although this proof-of-concept established the feasibility of lanthanide-based spectral conversion for MIR detection, non-plasmonic lanthanide transducers remain constrained by weak MIR-induced ratiometric modulation, limited NIR quantum efficiency, high pump-power demand, and modest specific detectivity. These limitations motivate strategies that can simultaneously enhance excitation, emission, and energy transfer processes. In our recent work, we demonstrated a ligand-assisted strategy in which broadband mid-infrared vibrational absorption induces local photothermal modulation, thereby regulating temperature-sensitive energy-transfer/back-transfer pathways in Nd³⁺/Yb³⁺-doped NaYF₄ core–shell nanoparticles²⁵. Nevertheless, key figures of merit of the MIR response remain amenable to further improvement, and further improvements towards broader application scenarios are still demanded.

Here, we integrate NIR-resonant Au nanospheres with Nd³⁺/Yb³⁺ co-doped NaYF₄ nanocrystals to plasmonically amplify MIR-induced population redistribution and detector-level NIR readout. Yb³⁺ co-doping augments NIR emission via efficient energy transfer, while localized surface plasmon resonances amplify excitation via local-field enhancement and radiative decay through the Purcell effect^26–28. Optimized plasmon-nanocrystal coupling yields an over 140-fold Purcell enhancement and elevates simulated NIR emission quantum efficiency to 48%. Consequently, our system attains a responsivity of ~1000 V·W⁻¹ and a room-temperature specific detectivity of 2 × 10⁹ Jones at 8 μm, demonstrating substantially improved MIR sensing performance.

Results

Design and working principle of plasmonic-enhanced MIR detection with lanthanide nanotransducers

The design of the plasmonic-enhanced lanthanide nanoparticle system is illustrated in Fig. 1a. Hexagonal NaYF₄ nanoparticles co-doped with Nd³⁺ and Yb³⁺ ions were synthesized, and the corresponding TEM images of the core-only and core–shell structures are compared in Supplementary Fig. 1a, b. The nanoparticles were then drop-cast to form films for MIR response characterization. As shown in Supplementary Fig. 1c, the hexagonal phase is preserved after shell growth, with the slight peak shifts indicating a small average lattice contraction. Oleic acid (OA) serves as a surface-capping ligand, facilitating local heating under MIR irradiation. Gold nanospheres function as optical antennas, generating localized surface plasmon resonances that simultaneously enhance spontaneous emission rates and QE of nearby emitters.

Fig. 1. Theoretical analysis of plasmonic-enhanced mid-infrared (MIR) response in lanthanide transducers.

a A schematic of the operating mechanism of the MIR sensing system. A 740 nm CW laser excites Nd³⁺ ions from the ⁴I_9/2 ground state to the higher ⁴F_7/2 level. Forward and backward energy transfer between Nd³⁺ and Yb³⁺ results in dominant 980 nm emission from Yb³⁺. MIR irradiation facilitates backward energy transfer from Yb³⁺ to Nd³⁺, altering the 806/980 emission intensity ratio. Plasmonic resonances further amplify the excitation rate, spontaneous emission, energy-transfer efficiency and overall quantum efficiency, enabling sensitive and time-resolved MIR photoresponse. b Electric-field distribution around a gold nanosphere (~200 nm diameter) under plane-wave excitation at 740 nm. The black circle outlines the nanosphere contour. c Purcell factor profiles along the dashed arrow in (b), showing maximum Purcell enhancement at ~7 nm from the nanosphere surface. d Simulated quantum efficiency of the dipole emitter as a function of the emission wavelength with peak NIR QE approaching 48% near 800 nm. e Scanning electron microscopy (SEM) image of the self-assembled Au-nanoparticle hybrid nanostructure. Scale bar: 100 nm.

The detection platform employs continuous-wave (CW) excitation at 740 nm, promoting Nd³⁺ ions from the ⁴I_9/2 ground state to the ⁴F_7/2 level. Subsequent relaxation yields three primary emissions: 806 nm (⁴F_5/2 → ⁴I_9/2) and 866 nm (⁴F_3/2 → ⁴I_9/2) from Nd³⁺, and 980 nm (²F_5/2 → ²F_7/2) from Yb³⁺ (Fig. 1a). Energy transfer between Nd³⁺ and Yb³⁺ occurs through Förster resonance energy transfer (FRET), enabled by the close energetic match between the Nd^{3+ 4}F_3/2 state and the Yb^{3+ 2}F_5/2 state. The long radiative lifetime of Yb^{3+ 2}F_5/2 state, combined with rapid multiphonon-assisted nonradiative relaxation from Nd^{3+ 4}F_5/2 to ⁴F_3/2, establishes dominant 980 nm emission^29,30. Thus, the analysis centers on the 806 nm and 980 nm emission channels, whose intensity ratio provides a ratiometric metric for MIR detection.

Upon MIR irradiation, thermally induced perturbations altered population dynamics within the nanocrystals, leading to a significant quenching of Yb³⁺ emissions. Simultaneously, the 806 nm emission was markedly enhanced, which can be attributed to thermal-modulated Yb-to-Nd back-energy transfer. This produced a pronounced change in the 806/980 nm intensity ratio, enabling sensitive MIR readout. To further amplify this modulation, localized electromagnetic fields from resonant gold nanospheres were used to boost excitation rates, radiative decay rates, and energy transfer efficiency, thereby collectively enhancing MIR detection sensitivity. This step targets the transduction bottleneck rather than simply increasing steady-state PL.

To elucidate the plasmonic enhancement mechanism, a simplified theoretical model was developed and validated using 3D finite-difference time-domain (FDTD) simulations. A dipole emitter coupled to a gold nanosphere of varying diameters is modeled to optimize resonance conditions (Supplementary Fig. 4a). A 200-nm nanosphere exhibited a fundamental plasmon mode with maximal field confinement near the surface around 800 nm (Supplementary Fig. 5a, b). The simulated near-field intensity |E/E₀| is strongest in the near-surface region (Fig. 1b), and the calculated Purcell factor exceeds 140 at a 7-nm separation, consistent with the experimentally realized nanoparticle-antenna spacing (Fig. 1c). Radiative and nonradiative decay rates were further calculated by integrating power flow in near- and far-field monitors (Supplementary Fig. 4b). At a 10 nm dipole-nanosphere separation, nonradiative decay was markedly suppressed, yielding elevated QE of the dipole emitter across the 800–1000 nm range (Supplementary Fig. 2a and Fig. 1d). It is noticed that, at very short separations between the 200-nm gold nanosphere and the NaYF₄ nanoparticles, although the field enhancement is strongest, increased non-radiative loss/quenching suppresses the net radiative benefit. This leads to an optimum Purcell enhancement at a small but nonzero emitter–metal spacing.

To further clarify the origin of the plasmonic enhancement theoretically, we consider the Au nanosphere as an optical antenna that modifies both the excitation and emission processes of the lanthanide nanoparticles (Supplementary Note 1). In our model, the emitter is located at an emitter–metal separation of ~10 nm, consistent with the experimentally relevant nanoparticle–Au nanosphere spacing, which is chosen to avoid severe quenching. The localized plasmonic field enhances the excitation rate according to

$$\frac{{\gamma }_{\mathrm{ex}}}{{\gamma }_{\mathrm{ex}}^0}\approx {\mid \frac{{\mathbf{E}}_{\mathbf{loc}}}{{\mathbf{E}}_{\mathbf{0}}}\mid }^2$$ 1

thereby increasing the excited-state population of the lanthanide ions. Meanwhile, the Au nanosphere modifies the local density of optical states and enhances the spontaneous-emission dynamics through the Purcell effect,

$$F_{\mathrm{P}}=\frac{{\gamma }_{\mathrm{sp}}}{{\gamma }_{\mathrm{sp}}^0}$$ 2

which, together with the change in non-radiative decay rate, determines the quantum efficiency

$$\mathrm{QE}=\frac{{\gamma }_{\mathrm{rad}}}{{\gamma }_{\mathrm{rad}}+{\gamma }_{\mathrm{nr}}}$$ 3

At the enhancement positions, the enhancement of radiative decay outweighs additional non-radiative loss, leading to an overall increase in NIR emission, which can be approximated as

$$\mathrm{E}{\mathrm{F}}_{\mathrm{NIR}}\approx \frac{{\gamma }_{\mathrm{ex}}}{{\gamma }_{\mathrm{ex}}^0}\cdot \frac{\mathrm{QE}}{{\mathrm{QE}}_0}$$ 4

Beyond enhancing the steady-state luminescence, the plasmonic environment also amplifies the MIR-induced redistribution between the Yb³⁺-dominated 980 nm channel and the Nd³⁺-dominated 806 nm channel. Specifically, the enhanced local field increases the excited-state population available for interionic transfer, and the observed MIR response is consistent with an effectively enhanced Yb–Nd energy-transfer process in the plasmon-modified photonic environment. As a result, the thermally activated backward transfer under MIR irradiation produces a much larger variation in the ratiometric signal I₈₀₆/I₉₈₀, consistent with the experimentally observed enhancement of MIR modulation and the shortening of the 980 nm lifetime in the plasmon-coupled structure.

Plasmonic-enhanced photoluminescence measurements of core-shell composite structures

The fabrication and characterization of the plasmonic-enhanced Nd³⁺/Yb³⁺-doped NaYF₄ nanoparticles are shown in Supplementary Fig. 2b. Au nanospheres were coated with oleylamine (OAm), while lanthanide nanoparticle surfaces were capped with OA ligands. In cyclohexane, self-assembly of the Au-lanthanide nanoparticle hybrid was primarily driven by hydrocarbon-ligand-mediated van der Waals interactions. Both components possessed long C18 alkyl chains (OAm on Au nanospheres and OA on lanthanide nanoparticles), which deformed upon close approach, enabling partial interdigitation and extended alkyl-alkyl contact. These London-dispersion-dominated interactions, combined with core-core van der Waals attraction, reduce the interfacial free energy and stabilize the hetero-aggregated state^31–34. Additional TEM characterizations were performed to clarify the self-assembly process. Before introducing the ligand-mediated van der Waals assembly, simple mixing led to severe nanoparticle aggregation and highly non-uniform adsorption on the Au nanospheres (Supplementary Fig. 1d, e). By using two alkyl-chain-containing surface ligands (OAm on Au and OA on the lanthanide nanoparticles) together with optimized particle concentration, effective attachment of lanthanide nanoparticles onto the Au nanospheres was achieved (Supplementary Fig. 1f–o), consistent with the formation of a core–satellite hybrid architecture (Fig. 1e).

To evaluate MIR detection performance, we employed a custom transmission-mode fluorescence setup, in which lanthanide nanoparticle films were illuminated by collinear 740 nm CW excitation and 8 μm MIR beams incident from opposite sides of a BaF₂ substrate (Supplementary Fig. 6). Reference and plasmonic-enhanced samples with matched thicknesses were prepared by controlled drop-casting. Steady-state PL spectra showed that plasmonic coupling to gold nanospheres markedly boosts emission intensity, with the 980 nm band showing ~12-fold enhancement relative to pristine nanoparticles (Fig. 2a), consistent with FDTD predictions. This enhancement stemmed from increased excitation and radiative QE, described by the total PL enhancement factor

$${\mathrm{EF}}_{\mathrm{NIR}}=\frac{{\gamma }_{\mathrm{tot}}}{{\gamma }_{\mathrm{tot}}^0}=\frac{{\gamma }_{\mathrm{ex}}}{{\gamma }_{\mathrm{ex}}^0}\cdot \frac{\mathrm{QE}}{{\mathrm{QE}}_0}$$ 5

which was estimated from simulations to approach ~20, in good agreement with experimental values. When subjected to MIR irradiation (7.7 mW), the 806/980 nm intensity ratio increased dramatically from ~0.035–2.29 (Fig. 2b), accompanied by a clear decrease in the brightness of the 980 nm emission image recorded by the CMOS camera. These observations indicate strong MIR-induced modulation of the excited-state population dynamics.

Fig. 2. Ratiometric near-infrared (NIR) photoluminescence measurements under mid-infrared (MIR) irradiation in plasmonic-enhanced nanoparticle systems.

a A steady-state ratiometric photoluminescence (PL) spectra of plasmonic-enhanced and pristine nanoparticles under 740 nm excitation (integration time: 0.01 s), demonstrating marked emission intensification due to plasmonic coupling. b Ratiometric luminescence of the plasmonic-enhanced system measured with and without MIR radiation (8 μm, 7.7 mW), showing substantial MIR-induced modulation via gold nanosphere integration. Inset: The captured CMOS image of 980 nm emission from the nanoparticles, demonstrating much weaker 980 nm emission under MIR irradiation. The dark spots visible in the image originate from dust on the camera lens. c MIR power-dependent NIR emission intensities of plasmonic-enhanced nanoparticles (0.81 mW to 7.7 mW), showing progressive enhancement of the 806 nm emission and suppression of 980 nm emission. d Temperature-dependent PL spectra (270–150 K) of the nanoparticles measured with and without MIR irradiation (8 μm, 7.7 mW). e Transient PL decay curves of the 980 nm emission from nanoparticles with gold nanospheres under both MIR-off and MIR-on conditions (8 μm, 7.7 mW). f Temperature-dependent 980 nm emission intensities of plasmonic-enhanced nanoparticles measured from 30–180 °C.

MIR power-dependent measurements reveal linear scaling of the ratiometric PL response at low MIR intensities, followed by saturation at higher powers due to the finite MIR absorption cross-section (Supplementary Fig. 7a and Fig. 2c). Temperature-dependent studies from 270 K to 150 K showed a monotonic increase in 980 nm emission intensity (Fig. 2d), with robust ratiometric response across temperatures, confirming persistence of plasmonic resonance under cryogenic conditions^35,36. It is also observed that lower temperatures lead to a larger 980 nm emission intensity difference under the same MIR irradiation condition, suggesting that the present system may achieve enhanced MIR response in cooled environments (Supplementary Fig. 7b).

Time-resolved PL measurements revealed MIR-induced shortening of the 980 nm lifetime from 141– 49 μs, attributable to enhanced decay and accelerated Yb³⁺-to-Nd³⁺ back-energy transfer. Besides, compared with the lifetime of pristine nanoparticles with no plasmonic coupling (210 μs measured without MIR), the plasmonic coupling shortened the luminescence lifetime from 210–141 μs. Although lifetime shortening alone cannot distinguish radiative from non-radiative decay, its concurrence with the ~12-fold emission enhancement and the FDTD-predicted increases in radiative rate and quantum efficiency is consistent with a larger radiative contribution in the optimized coupling regime. In contrast, the Nd³⁺ 806 nm and 866 nm lifetimes increased under MIR illumination (Supplementary Fig. 8a, b), which is consistent with thermally modulated Yb–Nd population redistribution with enhanced back-transfer under local heating.

Thickness-controlled films fabricated via Langmuir–Blodgett (LB) assembly showed that MIR-induced PL modulation became pronounced beyond ~100 monolayers, reflecting cumulative absorption effects (Supplementary Fig. 9a–d). Independent temperature calibration using a hot plate (up to 200 °C; Supplementary Fig. 9e) and micro-thermocouple measurements under MIR illumination (Supplementary Fig. 9f) were in close agreement, supporting the predominantly thermal origin of MIR modulation (Fig. 2f). Across a broad NIR pump range (6.97–33.7 mW), the ratiometric PL response remained linear with MIR power (Supplementary Fig. 8c, d and Supplementary Fig. 10), providing a reliable sensing metric (Fig. 3b).

Fig. 3. Photodetector performance metrics: temporal response, sensitivity, noise, and detectivity.

a A schematic of the custom-built setup for transient photoluminescence (PL) and photodetection measurements. A silicon amplified detector equipped with bandpass filters was used to collect emissions at specific wavelengths. b Ratiometric 806/980 nm emission intensity ratio of the nanoparticles as a function of MIR power measured under different NIR pump powers. c Temporal photoresponse of the plasmonic-enhanced mid-infrared (MIR) photodetector, showing rise and decay times of 7.2 ms and 8.3 ms, respectively. d Low-frequency noise spectra of the MIR detection system measured with and without NIR pumping. Solid lines are plotted for guiding the eyes. e Photovoltage measurement results under MIR irradiation of different powers ranging from 0.27 mW to 5 mW. f MIR power-dependent responsivity and specific detectivity. Error bars represent the standard deviation of 5 independent measurements. g Comparison of Room-temperature specific detectivity (D*) between the plasmonic-enhanced system (~2 × 10⁹ Jones) and representative commercial long-wavelength MIR detectors^38–46.

Performance characterization of the plasmonic-enhanced MIR photodetection system

To quantitatively evaluate device performance, the spectrometer was replaced by a silicon-based amplified photodetector to directly monitor the 980 nm emission (Fig. 3a). This setup enabled direct temporal characterization of the MIR photoresponse, yielding rise and decay times of 7.2 ms and 8.3 ms, respectively, corresponding to a 3-dB cutoff frequency of ~60 Hz (Fig. 3c)³⁷. Such response speeds place the present platform in a practically relevant millisecond-scale regime for real-time MIR sensing and imaging-rate operation, although they are not intended for high-speed optical communication. Low-frequency noise spectra, acquired with and without NIR pumping (Fig. 3d), revealed a noise voltage of ~2.5 nV at 60 Hz, indicating a low intrinsic noise floor despite the additional pump-induced noise. The photovoltage difference (V_{ph(MIR off)} - V_{ph(MIR on)}) showed a clear dependence on incident MIR power at 8 μm (Fig. 3e), confirming device sensitivity. Wavelength-dependent MIR photoresponse was further demonstrated at three representative wavelengths of 3.7, 8, and 11 μm (Supplementary Fig. 11a, b), spanning the investigated MIR range. Among these wavelengths, the strongest response and the highest specific detectivity were obtained at 8 μm. From these measurements, the device exhibited a responsivity approaching ~1000 V·W⁻¹ under low MIR powers, demonstrating robust and stable signal transduction. Combined with the measured noise equivalent power (NEP), the resulting room-temperature specific detectivity (D*) approached 2 × 10⁹ Jones (Fig. 3f). The effective sensing area is defined by the ~60 μm NIR pump spot, fully covered by the ~200 μm MIR beam, and reflects the chosen excitation geometry rather than an intrinsic limitation of the film. The excitation density is an order of magnitude below previous work, making wide-field NIR excitation with spatially resolved Si/CMOS readout a realistic next step²⁵. To further verify the enhancement effect of plasmonic coupling on the MIR response, we measured the MIR detection performance of the core–shell nanoparticle sample without plasmonic enhancement under the same measurement conditions (Supplementary Fig. 12a, b). The high detectivity of the plasmonic-enhanced nanoparticles represents a several-fold improvement over the non-plasmonic lanthanide-transducer system (Table S1), and also exceeds the performance of several representative long-wavelength MIR detector technologies (Fig. 3g and Table S2)^38–46.

To highlight the application potential of the plasmonic-enhanced nanoparticles, we demonstrated a proof-of-concept MIR signal readout based on logic inversion. As shown in Fig. 4a, a modulated MIR laser was used as the input signal, with “off” and “on” states representing logic “0” and “1”, respectively, encoded according to ASCII standards (Fig. 4b). The MIR signals impinged on the nanoparticle film, where the ratiometric PL response optically inverted the input code. The resulting modulated NIR emission was detected by a silicon amplified photodetector interfaced with a source-measure unit (Fig. 4c). The recorded photovoltage displayed clean square-wave signals with negligible delay relative to the input modulation, demonstrating proof-of-concept logic-inverted MIR signal readout. The decoded ASCII characters were reconstructed on a computer (Fig. 4d), confirming reliable information transfer via the MIR channel.

Fig. 4. Proof-of-concept MIR signal readout based on logic inversion.

a Schematic of the system setup for logic-inverted MIR signal readout, incorporating a MIR quantum cascade laser (QCL), laser shutter, plasmonic-enhanced nanoparticle film, silicon amplified photodetector, and source-measurement unit. b Transmitted ASCII codes for “MIR NP” encoded using the on/off modulation of the incident MIR beam. c Received inverse-coded MIR signals obtained by monitoring the output photovoltage from the silicon amplified detector. d Reconstructed ASCII codes for “MIR NP” derived from the inverted photovoltage signals via source measurement unit processing.

Discussion

In summary, we have demonstrated a plasmonic-enhanced, self-assembled, low-noise, and low-cost MIR detection platform that operates at room temperature. Incorporation of NIR-resonant gold nanospheres simultaneously enhances the excitation and emission channels of Nd³⁺/Yb³⁺-doped nanoparticles, yielding a one-order-of-magnitude improvement in their MIR response. By spectrally monitoring the 980 nm emission, the detector achieves a responsivity of ~1000 V·W^-1 and rise/decay times of 7.2/8.3 ms at 8 μm. The resulting specific detectivity approaches 2 × 10⁹ Jones, positioning this approach as a promising room-temperature alternative to conventional MCT-based MIR detectors. Future efforts will focus on translating the present proof-of-concept platform toward application-relevant system architectures, including integrated excitation–readout modules, optimized plasmonic geometries for targeted MIR bands, and scalable film/device configurations compatible with practical sensing and imaging scenarios such as specific imaging frame-rate targets or pilot integration with a Si CMOS readout chip. More broadly, the present results establish plasmonic engineering as a tunable strategy for lanthanide-based MIR transducers. Future MIR-resonant or dual-band plasmonic structures may enable simultaneous control of MIR absorption and NIR emission, providing further opportunities to improve transduction efficiency, spectral selectivity, and system-level performance.

Methods

Materials

Sodium hydroxide (NaOH; >98%), ammonium fluoride (NH₄F; >98%), 1-octadecene (90%), oleic acid (90%), yttrium acetate hydrate (Y(CH₃CO₂)₃⋅xH₂O; 99.9%), neodymium acetate hydrate (Nd(CH₃CO₂)₃⋅xH₂O; 99.9%), ytterbium acetate hydrate (Yb(CH₃CO₂)₃⋅xH₂O; 99.9%), cyclohexane (>98%) and methanol (>99%) were all purchased from Sigma–Aldrich and used as received without further purification.

Preparation of plasmonic-enhanced nanoparticles

Nd³⁺/Yb³⁺-codoped nanocrystals were synthesized using a modified co-precipitation method. In a typical procedure, 4.0 mL of a 0.2 M aqueous solution containing Nd(CH₃CO₂)₃, Yb(CH₃CO₂)₃ and Y(CH₃CO₂)₃ at specific ratios 40:15:45 was added to a 50 mL flask containing 6.0 mL of oleic acid and 14.0 mL of 1-octadecene. The mixture was heated to 150 °C and held for 60 min to facilitate the formation of lanthanide-oleate complexes. After cooling to 50 °C, a methanol solution (12.0 mL) containing NH₄F (3.2 mmol) and NaOH (2.0 mmol) was added with stirring and maintained at 50 °C for an additional 60 min. The solution was then heated to 100 °C under vacuum for 20 min to remove volatile components. Subsequently, the mixture was raised to 290 °C and kept at this temperature for three hours in an inert nitrogen atmosphere. Once cooled to room temperature, the resulting nanocrystals were collected by centrifugation, washed multiple times with ethanol, and finally dispersed in 4.0 mL of cyclohexane for further use.

To synthesize core-shell nanocrystals, a shell precursor containing 0.8 mmol of Y(CH₃CO₂)₃ was prepared according to the previously outlined procedure. The synthesized core nanocrystals were then combined with the shell precursor and maintained at 80 °C for 40 min. Once the mixture was cooled to 50 °C, a methanol solution (12.0 mL) containing NH₄F (3.2 mmol) and NaOH (2.0 mmol) was introduced. After stirring for an additional 30 min at 50 °C, the reaction temperature was increased to 100 °C, where impurities were removed under vacuum, followed by further heating to 290 °C. After 2 h of reaction time, the resulting core-shell nanocrystals were washed with ethanol and subsequently dispersed in 4 mL of cyclohexane.

Gold nanospheres capped with oleylamine (OAm) ligands and dispersed in cyclohexane (0.1 mg/mL) were used for the self-assembly process. For a typical preparation, 2 mL of the Au nanosphere solution and 2 mL of the as-synthesized lanthanide nanoparticle dispersion were mixed and stirred at 600 rpm for 6 h to facilitate ligand-mediated self-assembly between the two components. After assembly, the mixed solution was purified by centrifugation-assisted washing with cyclohexane to remove excess unbound nanoparticles and residual impurities, and the purified hybrid structures were finally redispersed and stored in cyclohexane for subsequent characterization and measurements. For sample preparation, the purified solution was drop-cast onto a BaF₂ substrate and heated on a hot plate at 80 °C for 5 min to evaporate the solvent, thereby forming the plasmonic hybrid nanostructure on the substrate for further optical measurements.

Theoretical modeling

To investigate the structural design of the plasmonic-enhanced nanoparticles for MIR detection prior to fabrication, we conducted three-dimensional finite-difference time-domain (FDTD) simulations (using FDTD Solutions by Lumerical Solutions Inc.). These simulations were aimed at identifying the physical dimensions that produce a plasmon resonance mode aligned with the emission and excitation wavelengths of the nanoparticles. The constructed FDTD model consists of a dipole source with the wavelength corresponding to the emission wavelength of the nanoparticles, gold nanoparticles with different radii (10–200 nm) and air. The dipole source is placed near a gold nanosphere at a distance of 10 nm, coinciding with the radius of one nanoparticle. The radiative and non-radiative emissions rates are calculated by simulating the power collected from the far field and making a comparison with the initial dipole power. And the QE can then be derived from the simulated radiative emission rate and total spontaneous emission rate.

Optical measurements setup and data acquisition

A homemade fluorescence system setup was constructed to measure the MIR photoluminescence response of the plasmonic-enhanced nanoparticles. A 740 nm multimode laser (CNI-MDL-III) served as the excitation source, and its beam was directed by a dichroic long-pass filter with a 760 nm cut-off filter (HPD760, Newport). The light was then focused onto the nanoparticle film using a 40× objective lens (Nikon, NA 0.7). For the MIR source, a tunable quantum cascade laser (QCL) of 3.7/8/11 micrometers (Daylight Solutions, MIRcat) was used, with its beam precisely focused onto the film by a zinc selenide (ZnSe) lens with a 5 cm focal length (Thorlabs). In our experiments, a MIR beam spot of about 200 μm diameter is obtained by a ZnSe lens, and the diameter of a laser beam spot (740 nm) is around 60 μm. And the MIR power density ranges from 9.23–284 W/cm² (corresponds to MIR power ranging from 2.9–89.2 mW), while the NIR power density ranges from 246.6–1192 W/cm².

The lifetime of the nanoparticles was measured within the same system, where a Si amplified detector (Thorlabs PDA36A2) was used to collect NIR emissions. Band-pass filters centered at 805, 860, and 980 nm were applied to isolate specific emission bands. An oscilloscope (Tektronix TBS1000C) was connected to the Si amplified detector to capture the electrical signal, while a function generator provided the analog signal.

The luminescence of the film was recorded using a CCD spectrometer (Witec) with a 750 nm long-pass filter, and an integration time of 10 ms was used. Emission power was measured with a Si amplified detector (Thorlabs PDA36A2), set to a gain of 60 dB. A 950 nm long-pass filter was positioned before the detector to isolate Yb³⁺ emissions. The detector’s output voltage was connected to a high-sensitivity digital source meter (Keysight B2912A). MIR laser power was measured with a thermal power meter (Ophir, Nova Display-ROHS).

Author contributions

X.Z., C.W.W., and J.C. designed and fabricated the devices, performed the mid-infrared photodetection measurements, and analyzed the experimental data. J.C. synthesized and characterized the upconversion nanoparticles and contributed to the optical measurements and data analysis. M.D. and Y.L. assisted with experimental characterization and data analysis. L.L., X.L., and Q.J.W. conceived and supervised the project. X.Z., C.W.W., and J.C. prepared the initial manuscript with input from all authors. All authors discussed the results, reviewed the manuscript, and approved the final version.

Peer review

Peer review information

Nature Communications thanks Mitradeep Sarkar and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

This work was partially supported by the Singapore Agency for Science, Technology and Research (A*STAR) (M22K2c0080 (Q. J. W.), R23I0IR041 (Q. J. W.) and M23M2b0056 (Q. J. W.)), and National Research Foundation Singapore (Award No. NRF-CRP22-2019-0007 (Q. J. W.), NRF-CRP29-2022-0003 (Q. J. W.), and the mid-size center grant for National Center for Advanced Integrated Photonics NRF-MSG-2023-0002 (Q. J. W.)), the RIE2025 Manufacturing, Trade and Connectivity (MTC) Individual Research Grant (award no. M24N7c0092) (X. L.), the National Natural Science Foundation of China (62288102, 62375230) (L. L.), National Research Foundation Singapore (NRF-PA2025-NUS) (J. C.).

Data availability

The data that support the findings of this study are available within the paper, supplementary information and source data file. All data are available from the corresponding author upon request. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41467-026-76701-2.