Plasmonic metamaterials for multi-effect enhancement: bridging optical, electronic, thermal, and acoustic domains

Abstract

Plasmonic metamaterials integrate subwavelength structural design with nanoscale field–matter interactions to deliver functionalities beyond intrinsic materials. Progress in nanofabrication and hybrid material integration increasingly enables coupled optical, electronic, thermal, and acoustic responses within a unified multi-effect framework. This review summarizes key mechanisms, design rules, and representative applications from sensing and photodetection to catalysis, integrated circuits, imaging/therapy, and information functions. It further outlines near-term opportunities toward integrative, programmable plasmonic systems.

Introduction

Surface plasmons, the collective oscillations of free electrons at metal–dielectric interfaces, provide an exceptional means of manipulating light at the nanoscale through surface plasmon polaritons (SPPs) and localized surface plasmon resonances (LSPRs)¹. In recent years, the emergence of plasmonic metamaterials has extended this concept from naturally occurring interfaces to artificially engineered media composed of subwavelength resonators². By precisely tailoring their geometry and spatial arrangement, these metamaterials enable unprecedented control over optical dispersion, impedance, and the local density of optical states, thereby modulating the phase, amplitude, and polarization of light in a deterministic manner³. This transformation, from materials that host plasmons to materials constructed from plasmonic building blocks, has redefined nanoscale light management. It has laid the foundation for compact, integrable, and highly tunable platforms now widely used in sensing, spectroscopy, imaging, energy harvesting, and information processing⁴.

The development of plasmonic metamaterials is closely tied to the enhancement effects that arise from the interaction of plasmons with different materials. For instance, when plasmons interact with biomolecules like proteins or DNA, nearfield enhancement effects, thermal effects, and nonlinear effects can be observed⁵. This enhancement effect is widely used in biosensing, disease diagnosis, and treatment. When plasmons are combined with energy materials such as perovskites, the charge transfer induced by plasmons can promote the separation and migration of electron-hole pairs, thereby enhancing the photoconversion efficiency⁶. This mechanism has important applications in solar cells and photocatalysis. When plasmons are combined with graphene, a plasmon-photocurrent effect can be established between the excitation of plasmons and the photoresponse of graphene. This effect can achieve highly controllable manipulation of light by controlling the excitation conditions of plasmons and the electronic state of graphene⁷. This phenomenon has wide applications in electronic devices, optical communication, and photodetection. The in-depth study and application of these enhancement effects are of great significance in promoting the development of optics, nanotechnology, biomedical science, and energy science.

Earlier studies on plasmonic devices largely focused on enhancing a single physical effect. For instance, strong near-field confinement was used for surface-enhanced Raman and fluorescence spectroscopy⁸, hot-electron generation improved photocatalytic reactions⁹, and acoustic modulation was employed for dynamic spectral tuning. Recent progress, however, has shifted toward intentional multi-effect integration, where multiple physical processes are synergistically coupled within the same plasmonic metamaterial¹. In parallel, several comprehensive reviews have discussed multi-domain couplings in such systems, including the control of light, heat, and vibrations in plasmonic and phononic platforms and plasmomechanical systems that integrate optical and mechanical interactions^10,11. By co-localizing optical fields, charge carriers, heat, and mechanical strain within nanoscale volumes, these systems achieve complex loops that link optical, electronic, thermal, and acoustic responses. These coupled dynamics enable a range of autonomous and adaptive functionalities. For instance, integrative plasmon systems empower sensors to combine optical and photoacoustic readout methods¹². Catalytic systems also benefit, as plasmon could leverage both hot carriers and nanoscale temperature gradients to achieve superior reaction selectivity¹³. Furthermore, the integration of reconfigurable materials, such as phase-change compounds¹⁴, and microelectromechanical system (MEMS) actuators provides active tunability¹⁵. Consequently, the research focus has evolved from optimizing isolated figures of merit to co-designing stable, low cross-talk, and reconfigurable multi-effect plasmonic systems.

In this review, we provide a comprehensive overview of the multi-effects in plasmonic metamaterials and their applications (Fig. 1). We begin in Section “Plasmonic nanomaterials, fabrication, and effects” by describing the types of plasmonic nanomaterials and fabrication techniques, then detail the fundamental enhancement effects in the optical, electronic, thermal, and acoustic realms. In Section “Applications of plasmonic enhancements”, we discuss how these effects are harnessed in a broad range of applications – from advanced sensors and light emitters to cancer therapy and data storage. Our emphasis is on making these concepts accessible to researchers in nanophotonics and applied physics, with a focus on current applications and emerging trends. Finally, in Section “Discussion”, we outline the future perspectives and challenges, including new materials and cutting-edge technologies (quantum plasmonics, 5G communications, etc.) that will shape the next generation of integrative plasmonic systems. By highlighting both the fundamental effects and their synergies, we aim to underscore how plasmonic metamaterials serve as a versatile nexus connecting multiple physical domains for innovative technological solutions.

Fig. 1: Schematic of plasmonic enhancement pathways from device to application.

The central circle depicts a canonical plasmonic resonator used to excite surface plasmons. The surrounding panels illustrate representative enhancement mechanisms—optical field confinement, hot-carrier generation, photothermal heating, and acousto-plasmonic modulation. The outer ring summarizes application domains elevated by these effects, including bio/chemical sensing, light emission, photodetection, catalysis, integrated plasmonic circuits, imaging and therapy, and information-oriented functions. SEIRA surface-enhanced infrared absorption. SERS surface-enhanced Raman scattering. SEF surface-enhanced fluorescence.

Plasmonic nanomaterials, fabrication, and effects

Plasmonic nanomaterials

Plasmonic nanomaterials refer to the metal (or conductive) nanostructures that support surface plasmons^16,17,18. The core components for exciting plasmons are typically metallic structures, and the plasmon discussed in this article refers to localized surface plasmon polaritons (LSPP). From the material side, there are four types of plasmon excitations: metals, semiconductor nanocrystals, 2D materials, and conductive polymers. Common metals used for supporting plasmons include gold, silver, and aluminum in various nanostructures, such as nanoparticles, nanorods, nanowires, nanocubes, and nanocross (Fig. 2a)¹⁹. These metals have free electrons that can collectively oscillate in response to incident light, giving rise to the phenomenon known as plasmon resonance. The choice of the metal and the nanostructure design play crucial roles in tailoring the plasmonic properties that endow plasmons with attractive characteristics. It exhibits distinctive properties and advantages absent in natural materials, such as structure-dependent tunable performance and frequency, broadband adaptability, and customizable surface enhancement effects. In addition to metals, Semiconductor nanocrystals, like indium tin oxide (ITO), can also support surface plasmon polaritons through the interaction of free electrons in ITO with incident light (Fig. 2b)²⁰. Unlike noble metals such as gold and silver, ITO has a lower density of free electrons, resulting in a plasmon frequency much lower than that of gold and silver. ITO’s plasmon frequency falls within the near-infrared region²¹, making it well-suited for transparent plasmonic applications in the visible range²². 2D materials (Fig. 2c)²³, like graphene, are also good candidates for plasmon generation, which is often referred to as graphene plasmons. In graphene, plasmons are associated with the collective oscillations of electrons on its surface. The charge carriers in graphene are massless Dirac fermions, and the plasmons in graphene are characterized by a linear dispersion relation. They exhibit tunable plasmon frequencies through electrostatic gating or chemical doping. Besides, conductive polymers can exhibit plasmonic properties, and one notable example is polypyrrole (PPy) (Fig. 2d)²⁴. PPy is a class of polymer materials with electrical conductivity derived from π bonds in their molecular structure. These π bonds can form a conjugated system, allowing electrons to move along the molecular chain. Conductive polymers can generate plasmons, but they differ from metal plasmons. Metal plasmons result from the collective oscillation of free electrons on the metal surface under light excitation, while conductive polymer plasmons arise from the collective oscillation of π electrons on the molecular chain under light excitation.

Fig. 2: Materials, fabrication, and mechanism of plasmon polaritons.

a–d Plasmonic nanomaterials including metallic nanostructures¹⁹, semiconductor nanocrystals²⁰, 2D materials²³, and conductive polymers²⁴. e Plasmonic nanostructure fabrication, including bottom-up fabrication and top-down fabrication. Reproduced with permission¹⁹. Copyright 2022 John Wiley and Sons. Reproduced with permission²⁰. Copyright 2009 American Chemical Society. Reproduced with permission²³. Copyright 2019 John Wiley and Sons. Reproduced with permission²⁴. Copyright 2018 Springer Nature Limited.

From the perspective of geometric dimensions, plasmonic nanostructures are generally divided into four categories, namely Zero-dimensional (0D), One-dimensional (1D), Two-dimensional (2D), and three-dimensional (3D). The common 0D metamaterial is nanoparticles²⁵, including spherical or polyhedral metal nanoparticles (e.g., Au nanospheres, nanocubes, nanostars) that support localized surface plasmon resonances (LSPRs) where the electron cloud collectively oscillates around the particle^25,26. These LSPRs are tunable by size, shape, and dielectric environment. For example, Au nanospheres (~50 nm) typically resonate in the green (~520 nm), whereas elongated Au nanorods have two resonances (transverse and longitudinal) that can be tuned from visible to NIR by adjusting aspect ratio. Nanoparticle assemblies can create “hotspots” (tiny gaps with concentrated fields), boosting local fields immensely (we will discuss in 2.3.1). 0D plasmonic nanoparticles are widely used as building blocks in sensing and imaging because of their strong scattering and ease of functionalization.

The common 1D metamaterials are nanowires and nanorods^27,28. By extending a plasmonic structure in one axis, one obtains nanowires or nanorods. These support not only LSPRs (e.g., the end-to-end mode of a nanorod) but also propagating plasmon modes along their length. Metallic nanowires can channel plasmonic energy as surface plasmon polaritons (SPPs), effectively acting as sub-diffraction optical waveguides. For instance, a silver nanowire can transport plasmons excited at one end to the other end, guiding light below the free-space diffraction limit. Nanowires have a high aspect ratio and can couple multiple plasmon modes (longitudinal, transverse), making them useful as interconnects in plasmonic circuits and as nanoantennas²⁹. Their tips often concentrate charge, yielding intense fields useful for sensing. However, ohmic losses in long wires can damp propagation, a challenge for long-distance plasmon transport.

The common 2D metamaterials are plasmonic metasurfaces^30,31. In 2D plasmonic structures, metal nanostructures are arranged in a planar periodic array (or lattice). Examples include arrays of nano-disks, nanoantennas, nanoholes, or gratings on a surface. Collectively, these arrays support delocalized modes such as surface lattice resonances (SLRs) and guided plasmonic modes, in addition to the individual nanoparticle LSPRs^32,33. Because of diffractive coupling between elements, 2D plasmonic crystals can exhibit sharp and intense resonances (Fano resonances, etc.) that are much narrower than those of an isolated particle. By engineering the lattice constant and particle geometry, one can tune the resonances across visible, IR, and even THz regimes³⁴. Metasurfaces leverage this to achieve specific optical functions, such as planar lenses, holograms, and structural color generation. Metamaterials, broadly speaking, are extensions of this concept into 3D (stacked layers of plasmonic structures) or high-density 2D patterns that achieve effective optical constants not found in nature (such as negative refractive index or epsilon-near-zero behavior). While early plasmonic metamaterials pursued negative index cloaks, modern research often focuses on metasurfaces for flat optics and multi-modal functionality.

3D plasmonic metamaterials extend the concept of planar metasurfaces into volumetric architectures, where metallic and dielectric components are arranged periodically or quasi-periodically in all three spatial dimensions^35,36. Typical examples include multilayer “fishnet” stacks, nanowire arrays, inverse opal structures, and gyroid networks. These 3D configurations allow simultaneous control of electric and magnetic responses throughout the volume, enabling unique dispersion properties such as hyperbolic or epsilon-near-zero behavior, negative refraction, and enhanced photonic density of states³⁷. As a result, 3D metamaterials support strong light–matter interactions not limited to surfaces but distributed within the bulk, offering opportunities for broadband absorption, thermal emission control, enhanced nonlinear processes, and volumetric sensing.

Fabrication

Advancements in nanofabrication technology have diversified the fabrication methods for plasmonic devices, categorized as bottom-up and top-down fabrication (Fig. 2e)³⁸. Bottom-up fabrication methods build nanostructures from atomic or nanoparticle building blocks, often relying on chemical and physical self-assembly processes³⁹. A prime example is the colloidal synthesis of metal nanoparticles, where well-developed protocols exist for synthesizing monodisperse Au nanospheres, rods, and stars by controlling reaction conditions⁴⁰. These colloids can then self-organize into ordered arrays (e.g., by drop-casting or controlled evaporation) or be directed into position via templates. Recent advances in DNA nanotechnology have greatly enhanced bottom-up plasmonic fabrication⁵: DNA strands can be designed to selectively bind to nanoparticles and arrange them into precise architectures with nanometer-scale control. Using DNA origami, researchers have assembled complex plasmonic clusters and even dynamic structures that respond to stimuli (e.g., DNA-linked particles that change configuration with pH)⁴¹. Bottom-up methods also include seed-mediated growth (growing larger nanostructures from smaller seeds), nanosphere lithography (using colloidal beads as masks), and chemical vapor deposition (CVD) for growing materials like graphene or nanowires. A notable achievement in bottom-up fabrication is the creation of 3D plasmonic architectures⁴². For instance, guiding nanoparticle assembly in tiny liquid droplets has produced 3D nanoparticle clusters for volumetric plasmonic hotspots. Bottom-up methods are generally parallel and scalable, yielding large quantities of nanostructures at low cost. However, they can struggle with polydispersity (size variations) and precise positioning; ensuring uniform nanogaps <5 nm or integrating colloids onto a chip at exact sites remains challenging.

Top-down fabrication methods adapt methods from the semiconductor industry to carve or print plasmonic patterns with designed shapes⁴³. The workhorse is lithography. In photolithography, UV light is projected through a mask onto a photoresist-covered substrate to define microscale patterns; modern immersion and deep-UV photolithography can achieve feature sizes down to ~50–100 nm, but traditional photolithography struggles below the optical diffraction limit. For truly nanoscale metamaterials, electron-beam lithography (EBL) is widely used⁴⁴. EBL can write arbitrary patterns (e.g., periodic antenna arrays, arbitrary metasurface pixels) by scanning a focused electron beam on a resist, achieving resolution <20 nm. It offers excellent precision—for example, one can fabricate nanogap antennas with ~10 nm gaps—albeit at the cost of low throughput and high expense (EBL is serial). Nanoimprint lithography is another top-down method where a prefabricated mold with nanoscale features is stamped into a resist or polymer, replicating patterns across a large area in one step⁴⁵. This is promising for the mass production of plasmonic chips. After lithography (which defines a pattern), metal is added (by evaporation or sputtering), and a lift-off process leaves behind the metal nanostructures in the lithographically defined shape. Top-down approaches excel in pattern accuracy and integration: one can design complex circuits of plasmonic waveguides, resonators, and align them with other on-chip components (photodetectors, microfluidic channels, etc.). They also produce very smooth surfaces and fine features if done in advanced facilities. The challenges for top-down are primarily achieving <5 nm gaps and atomically smooth surfaces—even EBL has limits in gap resolution, and metal evaporation can introduce sidewall roughness. Such imperfections can dampen plasmonic performance (increasing losses, variability). Additionally, top-down methods can be costly and typically require cleanroom processes⁴⁶.

Plasmonic acousto-optic-electric-thermal enhancement effects

Localized surface plasmons concentrate electromagnetic energy into nanoscale volumes, leading to several enhancement effects that couple to different physical domains. Here we outline the primary plasmonic enhancement mechanisms in: (2.3.1) Optical field enhancement, (2.3.2) Electronic (hot-carrier) generation, (2.3.3) Thermal (photothermal) effects, and (2.3.4) Acoustic (plasmon-phonon) phenomena. These effects often occur simultaneously in a plasmonic system—for example, a gold nanoparticle illuminated by a laser will both amplify the local optical field and produce hot electrons and heat. Understanding each effect individually is crucial to designing plasmonic metamaterials that exploit multiple of them.

Optical enhancement effects

One of the most celebrated properties of plasmonic nanostructures is their ability to enhance electromagnetic fields in their vicinity. When a metal nanoparticle or nanogap is at plasmon resonance, it can concentrate the incident light into an intensely localized near-field (orders of magnitude stronger than the excitation field)⁴⁷. This near-field enhancement is the basis of surface-enhanced optics like SERS, SEIRA, and SEF⁴⁸. Optical enhancement effects include near-field enhancement and far-field enhancement. Near-field enhancement is based on the resonance phenomenon of the plasmon. In metallic structures, localized surface plasmon resonance (LSPR) is observed (Fig. 3a, c), where the collective oscillation of surface electrons results in a resonance peak in the absorption or scattering spectrum. The reported work identifies specific characteristics^49,50. First, the resonance frequency in LSPR is highly dependent on the antenna size. As the size changes, the LSPR wavelength shifts accordingly (Fig. 3b, d). Different shapes exhibit unique LSPR characteristics. For instance, nanospheres show a single resonance peak, while Asymmetrical cross-shaped structures have multiple resonances due to their anisotropic nature⁵¹. Besides, the LSPR properties can be tuned by altering the composition of the nanoparticle or by using core-shell structures. Different metals and dielectric coatings influence the resonance characteristics. LSPR in metallic nanoparticles can be described using Mie theory, which provides a mathematical framework for calculating the interaction between electromagnetic waves and spherical particles. The Mie theory solution for the scattering cross-section C_sca(ω) of a spherical particle, such as a nanoparticle, is commonly used to characterize LSPR. The formula for the extinction cross-section is given by⁵²:

$${C}_{sca}(\omega )=\frac{{k}^{4}}{6\pi }|\alpha (\omega )|$$

(1)

where The complex polarizability (α) is a function of the particle size, shape, and material properties. In quasi-static approximation, the scalar potential in the Laplace formula is employed to govern the isotropic polarizability of spherical nanoparticles with specific permittivity (ε₁) and radius (a) bounded by a medium with specific permittivity (ε₂), as given by

$$\alpha =4\pi {a}^{3}\frac{{\varepsilon }_{1}-{\varepsilon }_{2}}{{\varepsilon }_{1}+2{\varepsilon }_{2}}$$

(2)

where we can conclude that the LSPR wavelength is affected by factors such as nanoparticle size, shape, and the dielectric environment. Therefore, it is often used as a sensing platform.

Fig. 3: Mechanism of plasmonic resonances in nanoparticles and nanorod antennas.

a Schematic diagram of the resonances effect in nanoparticles. b Demension-dependent characteristics of plasmonic resonances in nanoparticles²²⁸. Reproduced with permission²²⁸. Copyright 2019 Springer Nature Limited. c Resonance effect in nanorod antennas. d Dimension-dependent characteristics in nanorod antennas²²⁹. Insets: Electric field distribution. Reproduced with permission²²⁹. Copyright 2009 National Academy of Sciences.

LSPR in metallic structures is not only characterized by its near-field enhancement phenomenon but also by the significant enhancement in the far field⁵³. The far-field refers to a region sufficiently distant from a light source or an optical system, where the properties of the optical field can be approximated using plane waves. The far-field includes scattering, reflection, and transmission. The electromagnetic field generated by plasmon excitation not only forms a near-field near the metal surface but can also influence regions far from the surface by propagating to the far-field. This near-field enhancement effect results in an enhanced electric field observed in the far-field, providing favorable conditions for various optical applications. For instance, in surface-enhanced Raman scattering (SERS), far-field enhancement effects can boost the scattering signal, enabling highly sensitive molecular detection. The SERS enhancement factor is calculated by⁵⁴

$${G}_{SERS}=\frac{{I}_{SERS}({\omega }_{R})}{{I}_{CR}({\omega }_{R})}$$

(3)

where I_SERS(ω_R) and I_CR(ω_R) represent the total Raman intensity of SERS and conventional Raman, respectively. I_SERS(ω_R) is dictated by the induced dipole p_m(ω_R, r_m) at the Raman scattering frequency (ω_R) and the induced dipole of the plasmonic antennas, denoted as p_A(ω_R, r_A). The determination of p_m(ω_R, r_m) inges on the electric field strength during the excitation of the target molecules and the derivatives of Raman polarizability, that is^55,56,57,

$${{\boldsymbol{p}}}_{m}({\omega }_{R},{{\boldsymbol{r}}}_{m})={{\boldsymbol{\alpha }}}_{m}^{I}({\omega }_{R},{\omega }_{0})\cdot {\boldsymbol{E}}{\boldsymbol{L}}({\omega }_{0},{{\boldsymbol{r}}}_{m})$$

(4)

$${\boldsymbol{E}}{\boldsymbol{L}}({\omega }_{0},{{\boldsymbol{r}}}_{m})={g}_{1}({\omega }_{0},{{\boldsymbol{r}}}_{m}){{\boldsymbol{E}}}_{R}({\omega }_{0})$$

(5)

where EL(ω₀, r_m) and E_R(ω₀) denote the local and incident electric field strength at the position r_m, corresponding to the presence or absence of plasmonic antennas aimed at enhancing molecular signals. The term g₁(ω₀, r_m) signifies the enhancement factor of the incident electric field strength. The Raman polarizability derivative, α^I_m(ω_R, ω₀), is contingent upon the incident light frequency ω₀ and the Raman scattering frequency ω_R. To achieve heightened electric field strength for molecule detection, augmentation of either incident light intensity E_R(ω₀) or local electric field intensity EL(ω₀, r_m) is feasible. However, elevated incident laser light intensity can induce surface damage to the analyte, constraining sensitivity to a specific threshold. Enhancing g₁(ω₀, r_m) through plasmonic antennas offers a strategic approach to significantly amplify local photon-molecule interaction power near the antenna with reduced laser power—a pivotal advantage of SERS. The local excitation of the antenna at r_A by the adjacent point dipolar source p_m(ω_R, r_m) further influences its local electric field intensity at r_A, that is,

$${\boldsymbol{E}}{\boldsymbol{L}}({\omega }_{R},{{\boldsymbol{r}}}_{A})={C}_{2}\cdot {{\boldsymbol{p}}}_{m}({\omega }_{R},{{\boldsymbol{r}}}_{m})$$

(6)

where C₂ is dictated by the spatial relationship between plasmonic antennas and molecules. Subsequently, under the induced dipole approximation, the induced dipole of the antennas, denoted as p_A(ω_R, r_A) is calculated by

$${{\boldsymbol{p}}}_{A}({\omega }_{R},{{\boldsymbol{r}}}_{A})={{\boldsymbol{\alpha }}}_{A}({\omega }_{R})\cdot {\boldsymbol{E}}{\boldsymbol{L}}({\omega }_{R},{{\boldsymbol{r}}}_{A})={C}_{2}{{\boldsymbol{\alpha }}}_{A}({\omega }_{R}){{\boldsymbol{p}}}_{m}({\omega }_{R},{{\boldsymbol{r}}}_{m})$$

(7)

Subsequently, the additive local source p(ω_R) esponsible for eliciting far-field signals can be mathematically formulated as follows

$${\boldsymbol{p}}({\omega }_{R})={{\boldsymbol{p}}}_{m}({\omega }_{R},{{\boldsymbol{r}}}_{m})+{{\boldsymbol{p}}}_{A}({\omega }_{R},{{\boldsymbol{r}}}_{A})$$

(8)

Substituting Eq. (7) into Eq. (8) yields the expression for p(ω_R):

$${\boldsymbol{p}}({\omega }_{R})={{\boldsymbol{p}}}_{m}({\omega }_{R},{{\boldsymbol{r}}}_{m})(1+{C}_{2}{\alpha }_{A}({\omega }_{R}))={{\boldsymbol{p}}}_{m}({\omega }_{R},{{\boldsymbol{r}}}_{m})\cdot {g}_{2}({\omega }_{R},{{\boldsymbol{r}}}_{A})$$

(9)

Substituting Eqs. (4) and (5) into Eq. (8) yields:

$${\boldsymbol{p}}({\omega }_{R})={g}_{1}({\omega }_{0},{{\boldsymbol{r}}}_{A})\times [{g}_{2}({\omega }_{R},{{\boldsymbol{r}}}_{A}){\alpha }_{m}^{1}({\omega }_{R},{\omega }_{0})\cdot {{\boldsymbol{E}}}_{R}({\omega }_{0})]$$

(10)

The total Raman intensity I_SERS(ω_R) is directly proportional to ${|{\boldsymbol{p}}({\omega }_{R})|}^{2}$, that is,

$${I}_{SERS}\propto {|{g}_{1}({\omega }_{0},{{\boldsymbol{r}}}_{A})|}^{2}\cdot {|{g}_{2}({\omega }_{R},{{\boldsymbol{r}}}_{A}){{\boldsymbol{\alpha }}}_{m}^{1}({\omega }_{R},{\omega }_{0})\cdot {{\boldsymbol{E}}}_{R}({\omega }_{0})|}^{2}$$

(11)

In the context of conventional Raman spectroscopy, the total Raman intensity can be represented as:

$${I}_{CR}\propto {|{{\boldsymbol{\alpha }}}_{m}^{1}({\omega }_{R},{\omega }_{0})\cdot {{\boldsymbol{E}}}_{R}({\omega }_{0})|}^{2}$$

(12)

By substituting Eqs. (11) and (12) into Eq. (3), the SERS enhancement factor can be derived as

$${G}_{SERS}={|{g}_{1}({\omega }_{0},{{\boldsymbol{r}}}_{m})|}^{2}\cdot {|{g}_{2}({\omega }_{R},{{\boldsymbol{r}}}_{A})|}^{2}={\left|\frac{{\boldsymbol{E}}{\boldsymbol{L}}({\omega }_{0})}{{{\boldsymbol{E}}}_{R}({\omega }_{0})}\right|}^{2}\cdot {\left|\frac{{\boldsymbol{E}}{\boldsymbol{S}}({\omega }_{R})}{{\boldsymbol{S}}({\omega }_{R})}\right|}^{2}$$

(13)

where ES(ω_R) and S(ω_R) represent the enhanced and conventional Raman scattering, respectively. According to Eq. (13), enhancing the enhancement factor hinges on amplifying the scattering signal intensity. Influential factors include metal selection, surface morphology, laser wavelength, molecule adsorption, environmental conditions, and surface modification.

Electronic enhancement effect

Electronic enhancement effects mean the hot electron generation and transfer process for the energy conversion. Figure 4 illustrates this process in plasmonic nanostructures, including surface plasmon decay and hot-electron generation, plasmonic energy conversion, and hot-electron injection¹³. There are two possible ways that plasmons can decay after being excited by light in metal nanoparticles (Fig. 4a). One way is radiative decay, which emits photons with the same frequency as the plasmonic resonance. The other way is non-radiative decay, which transfers energy to electrons in the metal, creating hot electrons that can be injected into a neighboring semiconductor. The Schottky barrier is the difference between the work function of the metal and the electron affinity of the semiconductor. Only hot electrons with enough energy to overcome the Schottky barrier can be injected into the semiconductor, while the rest of the electrons remain in the metal. As shown in Fig. 4b, the hot electrons have a broad distribution of energies above the Fermi energy of the metal, and some of them can overcome the Schottky barrier and enter the conduction band of the semiconductor. Then, these hot electrons are injected into a neighboring semiconductor, forming a Schottky junction (Fig. 4c), where they can be collected as photocurrent or used for photocatalysis. After the formation of a metal-semiconductor Schottky junction, the generated holes are transported to the electrode by an electron donor solution or a hole-transporting material (HTM) that needs to be in contact with the nanostructure to maintain charge balance and current. Plasmonic energy conversion is a promising alternative to conventional solar cells, as it can potentially achieve high efficiencies, broad spectral absorption, and low fabrication costs⁵⁸. In contrast, conventional solar cells use the photoelectric effect to generate current. In this process, photons are absorbed by a semiconductor material, which excites electrons from the valence band to the conduction band, creating a flow of current. The efficiency of conventional solar cells is limited by several factors, such as the bandgap of the semiconductor material, the recombination of electron-hole pairs, and the reflection and transmission of light.

Fig. 4: Hot-electron generation and transfer in nanoparticles and nanorod antennas.

a Plasmons can undergo radiative decay through the emission of photons or non-radiative decay by exciting hot electrons¹³. b Plasmonic energy conversion involves the excitation of electrons from occupied energy levels to levels above the Fermi energy. DOS: parabolic density of states. c Hot electrons can be introduced into a semiconductor by establishing a Schottky barrier with the plasmonic nanostructure. Reproduced with permission¹³. Copyright 2014 Springer Nature Limited.

Beyond energy conversion, hot electrons strongly influence photocurrent intensity, which is exploited in plasmonic photodetectors. For instance, the ability of plasmons to support ultrafast oscillations, can lead to rapid response times in photodetectors, making them promising for high-speed optoelectronic applications. The plasmon charge transfer introduced above is a mechanism of the photodetection effect⁵⁹. Moreover, plasmons can synergistically enable photodetection when integrated with other phenomena, such as their combination with the bulk photovoltaic effect (BPVE)⁶⁰. BPVE refers to a nonlinear optical phenomenon that occurs in non-centrosymmetric materials, where the breaking of spatial inversion symmetry leads to a photovoltaic response. We take a metal antenna–graphene heterostructure as a representative example to illustrate the plasmon-integrated BPVE effect, as shown in Fig. 5⁶⁰. The metal nanoantenna comprises non-centrosymmetric metallic structures as meta-atoms atop graphene flakes (Fig. 5a). Mechanistic studies involve utilizing a back gate to modulate the doping level and consequently the Fermi level of graphene. When subjected to uniform illumination and under zero external bias (V_d = V_g = 0 V), photocarriers generated exhibit real-space shifts, determined by the direction and magnitude of the shift currents. The shift currents are determined by the plasmonic-enhanced local field. Figure 5b shows the near-field distribution and predicted vectorial photocurrent. The near-field distribution and intensity of the photocurrent are determined by the polarization and intensity of incident light (Fig. 5c). Notably, the photo-detection effect is a broad concept used to describe the process in which materials respond to light and convert it into an electrical signal. Within this category, the plasmonic photoelectric effect is a mechanism, particularly in the case of surface plasmon resonance. However, there are also other photoelectric effects, such as photoemission and internal photoeffect, which may not involve plasmons.

Fig. 5: Plasmon-integrated photocurrent interference for photodetector.

a Illustration of the plasmon-integrated bulk photovoltaic effect for photodetectors⁶⁰. b Plasmon-dependent near-field distribution and predicted vectorial photocurrent (I_ph). c Polar plot of measured I_ph. Reproduced with permission⁶⁰. Copyright 2020 Springer Nature Limited.

Thermal enhancement effects

Plasmonic resonances in metal nanostructures upon interaction with light can result in localized energy absorption, leading to enhanced thermal effects⁶¹. It is a process of nonradiative relaxation through electron–electron collisions or electron–lattice phonon coupling, leading to the absorption of light by the nanoparticle⁶². This effect finds applications in nanophotonics, opto-thermal conversion, sensing technologies, and medical fields⁶³. The photothermal characteristics of metal nanoparticles are illustrated in Fig. 6, depicting the underlying principle. Upon excitation with resonant photons, metal nanostructures experience rapid non-equilibrium heating due to the photoexcitation of the electron gas (Fig. 6a, b). The initial electronic excitation is swiftly followed by subpicosecond relaxation through electron–electron scattering, causing a rapid surge in the metal’s surface temperature (Fig. 6c). This rapid heating is succeeded by cooling to equilibrium through energy exchange between electrons and lattice phonons (Fig. 6d). Within the first few hundred picoseconds post-excitation, the lattice cools via phonon–phonon coupling, dissipating heat into the surrounding medium (Fig. 6e). The fundamental basis of the photothermal process has been extensively explored through ultrafast dynamics. These thermal effects have found wide applications, including the conversion of light to heat for photothermal therapy and photothermal gene release in cancer cells. The temperature distribution of plasmonic nanoparticles is shown in Fig. 6f. The nanoparticles exhibit uniform surface temperatures, with a rapid decrease observed in temperature beyond the surface. A boundary element method can be used to theoretically analyze the temperature distribution of nanoparticles, the temperature change can be expressed as⁶⁴

$$\Delta T(r)\approx \frac{p}{4\pi {k}_{m}r}$$

(14)

where P represents the power absorbed by a single nanoparticle with thermal conductivity k_m, and r is the distance from the center of the sphere. For non-spherical nanoparticles, the determination of temperature increase lacks a straightforward analytical expression, necessitating reliance on numerical simulations for a comprehensive discussion.

Fig. 6: Plasmon-enhanced thermal effects.

Schematic illustrating the photothermal conversion principle through plasmonic nanostructures. a Photoexcitation of the electron gas. b Rapid heating. c Rapid increase in the surface temperature. d Cooling back to equilibrium. e Heat dissipation into surrounding media. Reproduced with permission²³⁰. Copyright 2014 Royal Society of Chemistry. f Temperature distribution of plasmonic nanoparticles with varying dimensions. Reproduced with permission²³¹. Copyright 2020 Springer Nature Limited.

Acoustic plasmon effects

Graphene plasmons (GPs) are electromagnetic fields coupled to the oscillations of charge carriers, and they have garnered significant attention due to their short wavelengths, strong field confinement, and electrical tunability⁶⁵. These distinctive characteristics position GPs as promising tools for manipulating electromagnetic waves at the nanometer scale, enabling applications in highly integrated sensitive spectroscopy, lasers, and detectors⁶⁶. A noteworthy aspect is that when the graphene sheet is in close proximity to a metallic surface, the GPs combine with their mirror image, forming an electromagnetic mode that involves anti-phase charge oscillations in both the graphene and the metal. The advantage of acoustic graphene plasmons (AGPs) is the confinement of their modes within dielectric spacers. In contrast to regular GPs, whose energy scales with the square root of their momentum, this newly hybridized mode, referred to as AGPs, follows a linear energy versus momentum dispersion. Therefore, ohmic losses in graphene are expected to hinder the propagation of AGP to a lesser extent. On the other hand, larger AGP wave vectors require intermediate structures to alleviate the momentum mismatch under far-field excitation. AGP has been observed experimentally in both the infrared and terahertz frequency bands^67,68.

Examining the field distribution of AGPs facilitates comprehension of this mode, attainable through coupling with a scattering-type scanning near-field optical microscope (s-SNOM) integrated with an atomic force microscope (AFM), as depicted in Fig. 7⁶⁹. The intrinsic reason is that although the AGP fields primarily exist within the dielectric spacer, their evanescent components exhibit non-zero amplitudes beyond the graphene layer (Fig. 7a). The vertical component of the electric field for the mode, Ez, enters the open space above the structure, allowing effective coupling to the AFM tip in s-SNOM, essentially acting as a z-oriented electric dipole. The penetration depth (De) of Ez above graphene is defined as the distance corresponding to the 1/e attenuation of the field amplitude (Fig. 7b). By solving Maxwell’s equations within a multilayer configuration, the unique solution for the AGP eigenmode supported by the graphene structure can be determined. The z-component of the electric field, as depicted in the exponentially decaying profile (Fig. 7b), predominantly resides in the gap between graphene and the metal, exhibiting a rapid decay upon departing from the graphene surface. Besides, AGPs offer a diverse range of applications, generating ultra-strong confined and enhanced electric fields between the two materials. These fields can be confined in-plane to match their equivalent free-space wavelength. Furthermore, AGPs contribute to the heightened sensitivity of infrared molecular vibrational spectroscopy, reaching down to ångström-thick material layers. The concentration of fields by AGPs also facilitates the enhancement of nonlinear processes, including second-harmonic generation and four-wave mixing. This establishes AGPs as a platform for exploring and exploiting enhanced nonlinear light–matter interactions.

Fig. 7: Acoustic plasmons in graphene.

a Schematic illustrating the acoustic graphene plasmon (APG) inside the layered structure⁶⁹. b The z-component of the AGP electric field Ez, where the AFM tip couples to the AGP for the observation of APG mode. Reproduced with permission⁶⁹. Copyright 2021 Springer Nature Limited.

Photoacoustic enhancement effects

The photoacoustic (PA) effect begins with the absorption of light by a material, initiating a process that results in the generation of acoustic waves (Fig. 8a)⁷⁰. When a material absorbs light, it undergoes a rapid increase in temperature due to the conversion of optical energy into heat. This localized temperature rise induces a transient thermal expansion in the material. As a consequence, pressure waves or acoustic waves are generated as the material expands and contracts. These acoustic waves propagate through the material and can be detected as ultrasound signals. The amplitude and frequency characteristics of these signals provide valuable information about the optical absorption properties of the material. This phenomenon is particularly useful in imaging applications, where the spatial resolution of ultrasound is combined with the contrast capabilities of optical absorption, forming the basis for techniques such as photoacoustic imaging in various fields such as biomedical imaging, cancer detection, and materials characterization. The material that produces a signal different from the background in PA imaging is called a contrast agent. Plasmonic nanoparticles such as gold nanorods and nanostars are excellent candidates for PA contrast agents because of their high absorption cross-section, relative inertness, high stability, and tunability⁷¹. The PA response of plasmonic nanoparticles is characterized by the thermoelastic expansion model, wherein light absorption induces a temperature rise, resulting in thermoelastic expansion (Fig. 8b)⁷². This conversion of optical energy to pressure waves (P₀) relies on key factors: absorption coefficient (μ_a), thermal expansion coefficient (β), specific heat capacity (C_p) of the material, speed of sound (c) in the medium, and illumination fluence (F). Specifically, the pressure difference generated due to thermal expansion is given below:

$${p}_{0}=\left(\frac{\beta {c}^{2}}{{C}_{{\rm{p}}}}\right){\mu }_{{\rm{a}}}F=\Gamma A$$

(14)

Fig. 8: Plasmon-enhanced photoacoustic effects.

a Schematic illustrating the photoacoustic effect⁷⁰. Reproduced with permission⁷⁰. Copyright 2020 Springer Nature Limited. b Plasmonic enhancement of the photoacoustic effect⁷². Reproduced with permission⁷². Copyright 2020 American Chemical Society. c Photoacoustic images enhanced by plasmons with different configurations⁷³. Reproduced with permission⁷³. Copyright 2015 American Chemical Society.

The equation incorporates β, c, C_p, μ_a, and F, denoted in Κ^–1, m/s, J/(K kg), cm^–1, and J/cm², respectively. This expression follows the standard thermoelastic expansion model used in photoacoustic imaging⁷². Γ is the Grüneisen parameter characterizing the thermoacoustic conversion efficiency, and it can be expressed by

$$\Gamma =\left(\frac{\beta {c}^{2}}{{C}_{{\rm{p}}}}\right)$$

(15)

A is the local energy deposition density in J/cm³ and can be written by

$$A={\mu }_{{\rm{a}}}F$$

(16)

Researchers have optimized PA signal generation from plasmonic nanoparticles by adjusting key parameters outlined in the equations above. While illumination fluence (F) can be independently controlled, factors like the speed of sound (c) are intrinsic to tissue types and challenging to externally manipulate. Nevertheless, tuning the absorption coefficient (μ_a), thermal expansion coefficient (β), and specific heat capacity of the absorber (C_p) is achievable by modifying nanoparticle characteristics like shape, material, and solvent. These equations are applicable not only to plasmonic nanostructures but also to all absorbing molecules, including endogenous hemoglobin and melanin. For instance, enhancing plasmonic photothermal conversion efficiency and optimizing material heat transfer properties via doping can significantly amplify the photoacoustic response (Fig. 8c)⁷³.

Applications of plasmonic enhancements

Having established the four primary enhancement channels—optical, electronic, thermal, and acoustic—we now map them onto application-level functions. In practice, plasmonic metamaterials act as compact “transduction cells” that co-localize fields, carriers, heat, and strain within nanogaps, antennas, metasurfaces, waveguides, and metal-backed cavities. Geometry, materials (Au/Ag/Al, doped semiconductors, 2D layers), and biasing (electrical, thermal, acoustic) provide orthogonal knobs to tune resonance energy, quality factor, impedance, and coupling to emitters or analytes. This enables distinct but interoperable readouts: SERS and SEIRA for label-free fingerprints; metal-enhanced fluorescence for serology; hot-electron, photogating, and bolometric pathways for detection; thermo-optic and electro-optic effects for modulation; and acousto-plasmonic control for dynamic reconfiguration. Across the following subsections, we proceed from sensing and emission to detection, catalysis, and circuit-level integration, then to biophotonic imaging/therapy and, finally, to quantum- and information-oriented applications.

Optical enhancements for bio/chemical sensors

Plasmonic metamaterials have become workhorses for optical transduction in chemical and biosensing. By concentrating electromagnetic fields and tailoring radiative and non-radiative decay pathways, they enable complementary readouts that raise sensitivity and specificity: surface-enhanced Raman scattering (SERS)^54,74,75,76, surface-enhanced infrared absorption (SEIRA)^{19,77,78,79,80,81}, and surface-enhanced fluorescence (SEF)⁸². These effects are especially valuable in bio/chemical contexts, where signals are weak, backgrounds are complex, and relevant interactions occur at surfaces and interfaces.

SERS for interfacial chemistry

Surfaces and interfaces govern key steps in heterogeneous catalysis, electrochemistry, and photo(electro)chemistry, yet their molecular-scale evolution is difficult to observe directly. SERS provides the required surface sensitivity and chemical selectivity^83,84,85,86. For example, Yu et al. reported a depth-sensitive, plasmon-enhanced Raman method to probe both the nanostructure and composition of the solid-electrolyte interphase (SEI) on lithium metal anodes (Fig. 9a)⁸⁷. The study captured the dynamics of SEI formation and aging in operando, offering molecular-level guidance for engineering more stable, higher-performance battery interfaces. Beyond energy storage, the combination of high spatial resolution and sub-nanometer hot spots in SERS substrates enables nanoscale chemical mapping and even single-molecule optomechanical analyses⁸⁸.

Fig. 9: Plasmonic metamaterials for bio/chemical sensors.

a SERS-based wearable sensor for the biomolecular fingerprint detection⁸⁷. Reproduced with permission⁸⁷. Copyright 2021 American Association for the Advancement of Science. b SEF-based biosensing microarray for the detection of cancer biomarkers⁹¹. Reproduced with permission⁹¹. Copyright 2023 Elsevier. c SEIRA-based biosensors for the hyperspectral imaging of severe acute respiratory syndrome coronavirus (SARS-CoV) spike proteins in saliva⁹⁹. Workflow: (1) sampling and immobilization, (2) spectral measurement, (3) image processing using multimodal deep neural network model, and (4) prediction. Reproduced with permission⁹⁹. Copyright 2024 American Association for the Advancement of Science.

SEF for serological assays

Plasmonic metamaterials also enhance fluorescence by boosting excitation fields and improving radiative efficiency^89,90. Operating in the near-infrared (NIR) further suppresses tissue autofluorescence and scattering, improving signal-to-noise ratio in complex biofluids. As a representative case (Fig. 9b)⁹¹, a gold nanoisland substrate (AuNIS) with a gap-rich morphology was engineered to amplify the NIR emission of a symmetric heptamethine dye. The platform delivered >200-fold enhancement and underpinned a multiplexed enhanced-fluorescence microarray immunoassay (eFMIA). Arrays were patterned by non-contact piezo-driven micro-dispensing (PDMD) to quantify soluble PD-L1 and ICAM-1 from liquid biopsies. AuNIS was grown in situ on poly-L-lysine–treated polystyrene via seed-mediated nucleation and hydroxylamine reduction, after which PDMD deposited capture probes and controls. Mechanistically, performance tracked the distribution of inter-island gaps and the dye–metal separation; quantitative image analysis identified an optimal mean gap of ~14.3 nm for NIR enhancement. Analytically, the assay showed linear calibration for ICAM-1 and PD-L1 (R² ≈ 0.97 and 0.99) with limits of detection of ~0.8 and ~4.8 pg mL⁻¹, respectively, coefficients of variation ≤15%, and improved low-concentration performance compared with conventional enzyme-linked immunosorbent assay (ELISA)⁹². These results illustrate how NIR-tuned plasmonic metasurfaces can deliver sensitive, reproducible, and multiplexed serological readouts.

SEIRA for label-free fingerprinting

Where fluorescence labels are undesirable, SEIRA leverages mid-infrared “fingerprints” to provide chemical specificity^93,94,95, but conventional IR suffers from weak absorption. Plasmonic nanoantennas concentrate the mid-IR field and raise effective absorption cross-sections^78,96,97,98. Zhou and co-workers developed a dual plasmon–phonon nanoantenna platform on CaF₂, combining asymmetric Au and SiO₂ antennas, to image severe acute respiratory syndrome coronavirus (SARS-CoV) spike proteins directly in saliva (Fig. 9c)⁹⁹. Hyperspectral datacubes acquired on the platform were analyzed by a multimodal deep neural network trained on 80% of spectra and tested on 20%, enabling rapid (~2.3 × 10⁵ spectra per second) and accurate (~93%) de-overlapping of two spectrally similar spike proteins. Although demonstrated on liquid samples, the approach highlights how plasmon-enhanced IR spectroscopy, coupled with robust data analytics, can resolve subtle biochemical signatures relevant to infectious disease monitoring.

Collectively, these examples show how plasmonic metamaterials translate fundamental optical-enhancement effects (SERS, MEF, and SEIRA) into practical sensing modalities across electrochemical interfaces, serology, and label-free mid-IR fingerprinting. The common theme is control of the near field and emitter–metal geometry to raise signal strength while preserving fidelity, enabling sensitive, rapid, and potentially multiplexed bio/chemical analyses.

Optical enhancements for light emitters

Light emitters play a pivotal role in a multitude of applications across various fields, owing to their ability to generate and release electromagnetic radiation in the form of light^{100,101,102,103}. Their significance lies in the diverse range of purposes they serve, from everyday technologies to cutting-edge scientific advancements. In fields like imaging and microscopy, fluorescent light emitters, such as quantum dots, enable precise visualization of biological structures at the nanoscale¹⁰⁴. Additionally, light-based therapies, such as photodynamic therapy, utilize specific emitters for the targeted treatment of various medical conditions. One compelling example of the impact of light emitters is the light-emitting diode (LED). Since 1993, InGaN LEDs have undergone advancements and entered the commercial market¹⁰⁵. However, their realization as solid-state substitutes for light bulbs has been hindered by limitations in light-emission efficiencies. Plasmon-enhanced light emitters have received much attention because plasmons can enhance the absorption of light in LED. For instance, Okamoto and colleagues introduced large photoluminescence (PL) increases from InGaN/GaN quantum well (QW) material coated with metal layers (Fig. 10a)¹⁰⁶. In their experimental configuration (Fig. 10a), QW emission is photoexcited and detected from the back of the substrate by refining the lower surface of InGaN samples grown on sapphire substrates. According to the luminescence images of the uncoated GaN surface, the Ag-coated GaN, and the grating structure on GaN, the GaN with Ag grating has a greater luminescence enhancement effect (Fig. 10b). It indicates that the enhancements of plasmons on emitters mainly come from two aspects: one is the surface plasmon coupling between the metal layer and the QWs, which increases the spontaneous radiation rate and internal quantum efficiency (IQE) of the QWs; the other is the surface roughness and size of the metal layer and grating structure, which increase the scattering of surface plasmons to photons and the extraction efficiency of luminescence. The study of emission efficiencies and Purcell enhancement factors for emitters in various configurations reveals that optimal performance is achieved when the peak positions of InGaN quantum wells align with the plasmons generated by silver layers (Fig. 10c). This alignment results in enhanced radiative efficiency (η) approaching 100% across the entire luminescence spectrum. Geometric tuning of plasmons is attainable through the fabrication of nanostructures.

Fig. 10: Plasmon-enhanced light emitters.

Plasmonic light emitters based on quantum wells¹⁰⁶. a Structure and excitation/emission configuration of InGaN quantum wells. b SEM and microluminescence images of the uncoated GaN surface, 50 nm Ag film evaporated on GaN, and the Ag grating structure on GaN. c Wavelength-dependent emission efficiencies (η_int^*(ω)) and Purcell enhancement factor (F_p(ω)) of GaN quantum wells in different configurations. Reproduced with permission¹⁰⁶. Copyright 2004 Springer Nature Limited. d Plasmonic light emitters based on perovskite¹⁰⁷. Preparation procedures of coaxial core/shell (CS) nanowire (NW) heterojunction architecture and the corresponding top-view SEM images. e Schematic diagram of the heterostructure emitters. f Band alignment of the heterostructure. g Electroluminescence (EL) spectra of the plasmonic emitters with Au NPs decoration and the reference device without Au NPs. h Time dependence (60 h) of the emission intensity and current density of the plasmonic emitters at a fixed bias voltage. i EL performance of the plasmonic emitters and the reference emitters as a function of storage time in an ambient environment. Reproduced with permission¹⁰⁷. Copyright 2018 John Wiley and Sons.

This mechanism is also applicable to perovskite-based light emitters. Shi and colleagues reported the enhanced performance and stability of perovskite light-emitting diodes (PeLEDs) by combining plasmons and coaxial core/shell nanostructure configuration (Fig. 10d)¹⁰⁷. This configuration provides a large interface area, a strong carrier confinement, and an effective surface passivation for the perovskite QDs, improving the carrier recombination efficiency and the device stability. The PeLEDs are fabricated using CsPbBr3 quantum dots (QDs) and ZnO nanowires (NWs) as the active layer and the carrier injector, respectively (Fig. 10e). Plasmonic Au NPs are embedded into the device and optimized the thickness of the MgZnO spacer layer to achieve efficient exciton-plasmon coupling and emission enhancement (Fig. 10f). The MgZnO spacer layer serves as an electron-injection and hole-blocking layer, as well as a dielectric layer to prevent the nonradiative energy transfer and quenching of the emission. The plasmonic PeLEDs exhibited a luminance of 10206 cd/m², an external quantum efficiency of 4.626%, and a current efficiency of 8.736 cd/A, which is strong than that of LEDs without plasmons (Fig. 10g). Additionally, the plasmonic PeLEDs were demonstrated a improved operation stability compared with the reference PeLEDs (Fig. 10h). This study establishes that the mechanisms contributing to electroluminescence enhancement are linked to heightened rates of spontaneous emission and improved internal quantum efficiency, resulting from resonant coupling between excitons in perovskite quantum dots and plasmons in gold nanoparticles—a phenomenon observed similarly in plasmonic quantum well emitters.

Thermal enhancements for nanofribration, therapy and nanotweezers

The thermal effect of plasmons finds diverse applications, including nanofabrication, thermal therapy, and nanotweezers^108,109,110. The mechanism of plasmonic thermal nanofabrication is that laser pulses produce transient local heating on the metasurface, causing the metal nanodisks to melt and deform, as demonstrated in Fig. 11a by Zhu and colleagues¹¹¹. Depending on the energy density of the laser pulse, different surface morphologies can be created, which alter the plasmonic resonance frequency and color (Fig. 11b). By controlling the laser parameters (such as power, spot size, frequency, etc.) and position, arbitrary color patterns can be printed on the metasurface (Fig. 11c, d). This method achieves a printing speed of 1 nanosecond per pixel, a resolution of 127,000 DPI, and a power consumption of 0.3 nJ. Importantly, this method enables a resolution that surpasses the diffraction limit imposed by conventional optics. The diffraction limit traditionally restricts spatial resolution in visual applications to a quarter of a micrometer. Plasmonic laser printing overcomes this limitation through several mechanisms. Firstly, photon heating is observed only in the vicinity of the focal spot, and the size of the melted material can be minimized due to the Gaussian probability distribution of photon fluence density. This results in a critical exposure energy threshold that initiates the melting reaction, excluding low-intensity wings and reducing the size of the melted material. One critical aspect is the application of surface plasmons, which possess two key properties: subwavelength light confinement and intense field enhancement^112,113. By using a plasmonic super-lens, sub-diffraction-limited imaging with a 60 nm half-pitch resolution has been achieved. In the context of laser-induced color printing, the resolution is pushed to the sub-diffraction scale by leveraging plasmonic thermal reshaping. In this process, the electric field is confined and enhanced at the plasmonic metasurface. This localization of the electric field helps in concentrating light intensity, leading to localized heating at the intended particle. This mechanism is crucial for achieving higher spatial resolution, better stability (as only a laser fluctuation affecting the nearest neighboring unit cell is relevant), and lower power requirements. The latter is particularly notable, as only 1% of the original laser power is needed due to the 100-fold enhancement of the electric field intensity. This combination of factors enables plasmonic laser printing to achieve resolutions beyond the diffraction limit imposed by conventional optics.

Fig. 11: Plasmon-enhanced thermal applications.

Plasmon-enhanced color laser printing¹¹¹. a Schematic illustrations of plasmonic laser printing. b Measured spectra of the printed metasurface. c Plasmonic laser printing has a resolution that exceeds the diffraction limit. d Printed images in different color schemes. Reproduced with permission¹¹¹. Copyright 2016 Springer Nature Limited. e Plasmonic electrothermal nanotweezer¹¹⁸. Schematic of the hybrid plasmonic nanotweezer. f Experimentally measured radial velocity vector plot of microfluidic flow induced around the plasmonic nanostructure. Reproduced with permission¹¹⁸. Copyright 2016 Springer Nature Limited.

Plasmonic thermal enhancement holds significant promise for tumor therapy^114,115,116. From a treatment mechanism standpoint, the enhancement of plasmonic nanoantennas with prolonged in vivo circulation times, elevated absorption coefficients per unit weight, and narrower absorption spectra would facilitate more effective tumor permeation post-administration. This, in turn, would amplify photothermal contrast between antennas and normal tissue, enabling enhanced tumor treatment efficacy at lower laser intensities or greater in vivo depths. For instance, Maltzahn and colleagues synthesized polyethylene glycol (PEG)–protected gold nanorods and demonstrated their ability to destroy all irradiated human xenograft tumors in mice¹¹⁷. PEG gold nanorods show prolonged stability in biological media (>1000 h), whereas traditional cetyltrimethylammonium bromide (CTAB)-coated nanorods precipitated over time. In this experiment, they used nude mice with bilateral human MDA-MB-435 tumors and injected them with either PEG nanorods or saline. After the nanoparticles cleared from the bloodstream, the right flank of each mouse was exposed to NIR irradiation for 5 min (810 nm, 2 W/cm²). Tumor growth was monitored over time. Within 10 days, all the PEG-nanorod–targeted tumors that were irradiated disappeared, while other tumors, including those exposed to laser after saline injection, continued to grow. To assess the survival benefit, mice with a single tumor were divided into four groups: PEG-nanorods + laser, PEG-nanorods − laser, saline + laser, and saline − laser. By day 20 after treatment, mice in the PEG-nanorods + laser group showed only a minor scar with no tumor regrowth, while all other surviving mice had thriving tumors (Fig. 11i). The results suggest that the PEG-nanorods combined with NIR irradiation effectively destroyed tumors and provided a survival benefit compared to other treatments.

Plasmonic nanoantennas offer the capability to generate intensely localized and enhanced electromagnetic fields, enabling the efficient trapping and manipulation of nanoscale objects. This surpasses the limitations of conventional diffraction-limited optical tweezers. Recently, Ndukaife and colleagues introduced an optofluidic mechanism to induce fluidic transport and on-demand particle delivery (Fig. 11e). This approach involves the integration of photo-induced heating from a singular plasmonic nanoantenna with an applied AC electric field within a particle suspension. Termed electrothermoplasmonic (ETP) flow, this method generates a microfluidic flow, capturing suspended particles and efficiently transporting them towards the illuminated nanoantenna as needed¹¹⁸. The mechanism for generating on-demand optofluidic flow involves localized heating of a fluid by illuminated plasmonic nanoantennas (Fig. 11e). This heating creates a local gradient in the electrical properties of the fluid (Fig. 11f), specifically in permittivity and electrical conductivity. When an AC electric field is applied, it induces an electrical body force per unit volume in the fluid due to these gradients. This electrical body force causes fluid motion, leading to a drag force on suspended particles that transports them toward the hotspots created by the plasmonic nanoantennas, where they get trapped. This approach allows for controlled fluid motion triggered by laser illumination of the plasmonic nanoantenna and the application of an AC electric field. The on-demand nature of this optofluidic flow is facilitated by the ability to switch the AC field. Consequently, the separation of roles between the plasmonic trapping force and fluidic drag is achieved, enabling precise control over particle transport in the system.

Acoustic enhancements for ultrasensitive sensing and quantum applications

AGPs can generate an exceptionally confined out-of-plane electric field within the gap between graphene and the metal substrate^{119,120,121,122}. This tight confinement, while promising for sensing applications leveraging the SEIRA effect, presents a significant challenge due to a pronounced momentum mismatch with free-space light, resulting in suboptimal coupling efficiency. In the context of SEIRA sensing, a highly efficient far-field coupling scheme becomes imperative to discern the subtle vibrational signatures of thin-film analytes amidst background noise. Lee and colleagues address this challenge by demonstrating a graphene acoustic plasmon resonator that overcomes the momentum mismatch, achieving nearly perfect absorption (94%) of incident mid-infrared light and delivering robust Ångström-thick film sensing performance (Fig. 12a)⁶⁸. The graphene acoustic plasmon resonator consists of a continuous graphene layer and a metal ribbon array separated by a nanometer-scale gap (Fig. 12b). The key advantage of this mechanism is that it uses GPs as an intermediary to enhance the coupling efficiency of AGPs via plasmon conversion processes (κ₀₂κ₂₁e₂ and κ₀₁κ₁₂e₁), instead of relying solely on scattering-mediated coupling processes (κ₀₁ and κ₀₂), where κ is a coupling term and e is electric field-related term. This mechanism eliminates the fundamental trade-off between field confinement and coupling efficiency, enabling the acoustic plasmon resonator to maintain high-efficiency resonance absorption even at reduced gap sizes, while scattering-mediated resonators rapidly lose efficiency as the gap size decreases. The sensing enhancement mechanism is based on the coupling between graphene and a quarter-wavelength cavity (Fig. 12c). The cavity is a structure that prevents the reflection of light at a certain wavelength, called the critical wavelength, by creating a phase difference of π between the incident and reflected waves. The authors demonstrate experimentally that the AGP resonator can achieve up to 94% absorption of incident light, and that the absorption bandwidth can vary between 52% and 10% of the central wavelength. The authors also demonstrate that the AGP resonator can detect submonolayer protein films and ångström-thick SiO₂ films by measuring the absorption spectra and observing the plasmon–phonon coupling effects (Fig. 12d).

Fig. 12: Acoustic plasmons for ultrasensitive sensing and quantum applications.

a Acoustic plasmons in graphene for ultrasensitive infrared spectroscopy⁶⁸. The architectural configuration of the acoustic plasmon resonator, illustrating coupling pathways to plasmon modes under normal incidence of a TM polarized plane wave. Red and green arrows signify AGPs and conventional GPs, respectively. The inset displays typical plasmon dispersions for AGPs and GPs in the free-standing scenario. b Schematic of the device structure. c The mechanism for plasmonic absorption enhancement using the quarter-wavelength condition. d Ångström-thick film sensing using the AGP sensor. Reproduced with permission⁶⁸. Copyright 2019 Springer Nature Limited. e Graphene acoustic plasmons for revealing quantum surface-response of metals¹²³. Schematics of a dielectric-graphene–dielectric–metal heterostructure. The zoom-in view presents surface-response functions, namely d_⊥ and d_∥, alongside corresponding microscopic parameters defining the metal surface: equilibrium electronic density, n₀(z), and induced charge density, ρ_ind(z). f Employing the spectral shifting of AGPs to extract the quantum surface-response of metals. Reproduced with permission¹²³. Copyright 2021 Springer Nature Limited.

Another interesting application of AGPs is to detect quantum surface-response of metals. Gonçalves and colleagues developed a dielectric–graphene–dielectric–metal (GDM) heterostructure consisting of a continuous graphene layer sandwiched between two dielectric layers and a metallic substrate (Fig. 12e)¹²³. The thickness of the dielectric spacer controls the separation between graphene and the metal substrate. The GDM device can simultaneously considers the quantum nonlocal effects of graphene and the metal substrate, and find that the quantum surface response of the metal can be described by the so-called Feibelman d-parameters, which reflect the distribution of charge density and tangential current density on the metal surface. Furthermore, the quantum surface response of the metal can cause a redshift or blueshift of the AGP spectrum, depending on the sign of the d-parameter, and this frequency shift does not come with a significant increase in attenuation (Fig. 12f). This scheme enables experimental extraction of the low-frequency quantum surface-response of metals by tracking the AGP spectral shift, which cannot be achieved by traditional metal plasmons. Significantly, this method attains sub-angstrom precision in examining the stationary quantum response of metals, playing a pivotal role in enhancing optical design and tailoring chemical processes on metal surfaces.

Electronic enhancements for catalytic chemistry

Catalysis plays a pivotal role in modern chemistry and industry, serving as a critical enabler for accelerating chemical transformations and enhancing reaction selectivity^9,124. By providing an alternative reaction pathway with a lower activation energy, catalysts facilitate faster reaction rates, allowing for more sustainable and efficient production processes. Additionally, catalysts often contribute to the reduction of energy consumption and waste generation, making them indispensable in the development of environmentally friendly and economically viable chemical processes. Plasmonic metamaterials introduce a unique and valuable dimension to catalysis. The strong coupling of metal nanoparticles with specific photon energies during LSPR enhances light absorption, thereby initiating and promoting catalytic processes^{6,125,126,127,128}. This heightened light-matter interaction leads to increased electron excitation and localized electromagnetic field enhancement on the metal surface, contributing to improved catalytic activity. Besides, the assistance provided by plasmonic effects in catalysis is particularly crucial for reactions conducted at low temperatures and under ambient conditions. By harnessing the optical excitation of collective electronic resonances of metals, plasmonic metamaterials enable catalysis to occur under milder conditions, ensuring the longevity of active sites and minimizing the potential for high-temperature side reactions. Next, we introduce the enhancement mechanism, exemplifying it with plasmon-induced H₂ dissociation. As we introduced in the electronic enhancement effect section above, hot electrons will be generated on the surface of AuNPs under the excitation of light (Fig. 13a)¹²⁹. In catalysis, the initial stage involves H2 physically adsorbing near the Au surface for an extended period at equilibrium conditions (Fig. 13b). Given the low binding energy and adhesion coefficient of H₂ on gold, which prevents adsorption on pure gold surfaces, a common approach to enhance H₂ retention is by loading gold nanoparticles onto TiO₂. Following resonant excitation of a plasmon, a subset of hot electrons transitions to H₂^σ−, thereby initiating H₂ dissociation (Fig. 13c, d). This mechanism is viable for both photocatalytic and electrocatalytic reactions.

Fig. 13: Plasmon-enhanced catalytic reactions.

The scheme of plasmon-driven catalytic reactions on Au NP surface¹²⁹. a Hot electrons on the surface. b–d Mechanistic representation of H₂ dissociation on the AuNP surface. Reproduced with permission¹²⁹. Copyright 2013 American Chemical Society. e Plasmon-enhanced light-controlled photocatalytic reactions¹³⁰. Synthesis of the plasmonic catalysts. f Scheme for the phenylacetylene hydrogenation reaction leading to the formation of styrene and ethylbenzene. g, h Conversion percentages for the phenylacethylene hydrogenation performed in the dark and under visible light excitation. Reproduced with permission¹³⁰. Copyright 2018 American Chemical Society. i Plasmon-enhanced electrocatalytic reaction¹³¹. Schematic apparatus for in situ electrochemical scattering measurement in solution under a dark-field microscope. j TEM image of Au–MoS₂ hybrids. k UV–vis spectra of various nanoparticles. l HER polarization curves obtained on several nanoparticles as indicated. m Schematic energy level diagram illuminating hot electrons injection and change of MoS₂ Fermi level. n Different plausible hot electrons transfer pathways likely to occur during plasmon excitation. Reproduced with permission¹³¹. Copyright 2015 American Chemical Society.

Photocatalytic reactions

Quiroz and colleagues demonstrated the controllable modulation of reaction selectivity in sequential photocatalytic reactions through plasmon-induced effects under visible light irradiation (Fig. 13e)¹³⁰. Obviously, light irradiation enhances the conversion rate of phenylacetylene, primarily attributed to the excitation of plasmon (Fig. 13f). High electric field intensities of plasmons play a crucial role in facilitating rapid charge carrier transfer from plasmonic nanoparticles. These intensified fields release energy through radiative scattering or nonradiative excitation (absorption) in metal nanoparticles. Larger electric field intensities correspond to higher energy transfer rates. In core–shell and nanorattle multimetallic architectures, plasmon-excited energy can efficiently dissipate through absorption in metal shells, thereby enhancing the extraction probability of energetic charge carriers. Besides, the material of plasmons is crucial, alongside the electric field strength. Low phenylacetylene conversions (<10%) were observed for Au/SiO₂ and AgAu/SiO₂, in darkness or under light, which suggests that neither Au nor Ag exhibits high conversion activity in these conditions (Fig. 13g). In contrast, Pt demonstrated excellent catalytic activity, leading to accelerated conversion. Consequently, Au@AgPt nanorattles with plasmon extraction exhibited the most effective catalytic performance in the reaction (Fig. 13h).

Electrocatalytic reaction

Shi and colleagues demonstrated that the Au nanorod/MoS2 heterostructure, under plasmon excitation, surpasses conventional catalysts (Fig. 13i–k)¹³¹. This heterostructure exhibits superior catalytic activity compared to bare MoS₂, attributed to electron transfer from Au to MoS₂, enhancing H₂ adsorption on S sites. Illumination with an 808 nm laser further boosts catalytic activity, reducing the onset potential by 60 mV (Fig. 13l). Tafel and Arrhenius plots illustrate plasmon introduction accelerating kinetics and lowering the activation barrier of the hydrogen evolution reaction (HER). Dark-field microscopy directly observes electron transfer, supporting a proposed plasmon-assisted mechanism. The catalytic enhancement is attributed to hot electrons from Au nanorod’s nonradiative decay transferring to MoS2’s conduction band. The study suggests further improvement using alcohol to consume hot holes in Au nanorod, introducing the concept of plasmon-assisted electrocatalysis and systematically investigating hot electrons’ impact on catalytic reactions (Fig. 13m, n). Additionally, plasmon-enhanced catalytic reactions were observed in systems such as Au/N-doped carbon materials, MoO3−x, and TiN-graphene, emphasizing the role of plasmon in enhancing catalytic activity, particularly for nonprecious electrocatalysts.

Electronic enhancements for detectors

Detectors play a crucial role in various fields, serving as essential components for capturing and measuring signals or events^{132,133,134,135}. Their primary purpose is to convert different forms of input, such as light, radiation, or particles, into measurable output signals. Plasmonic detectors leverage the unique properties of surface plasmons to enhance their performance, particularly in scenarios where conventional detectors may face limitations. One notable advantage of plasmonic detectors lies in their capability to operate without conventional lenses. Traditional detectors often rely on complex optical systems, including filters, to focus and capture light. In contrast, plasmonic detectors can exploit the interaction between surface plasmons and incident light without the need for traditional lenses. This unique feature enables the development of filterless detectors, simplifying the design and reducing the overall complexity of optical systems. For instance, Wei and colleagues reported a nanoantenna-mediated semimetal photodetector exhibiting polarization ratio (PR) values that span the entire numerical range from positive (unipolar regime) to negative (bipolar regime), encompassing the entire spectrum from 1 to ∞ and −∞ to −1 (Fig. 14a)⁷. The detector consists of an array of plasmonic nanoantennas on a graphene transistor, which is placed on a silicon substrate with a SiO₂ layer (Fig. 14a). The nanoantennas have different orientation angles and can be made of different metals to create heterogeneous structures. The detector works at zero drain–source bias under uniform illumination because the vectorial and non-local photoresponse in graphene is a semimetal with zero bandgap. The photoresponse is driven by the local photocurrents generated at the metal–graphene interfaces, where the incident light is concentrated by the plasmonic resonance of the nanoantennas (Fig. 14b, c). The photocurrents are dependent on the polarization and orientation of the electric field of the light, as well as the geometry and material of the nanoantennas. Therefore, the detector can achieve a transition between unipolar and bipolar polarization dependence by tuning the orientation angles of the nanoantennas (Fig. 14d, e). In the unipolar regime, the photocurrents are sign-maintaining, meaning that they have the same sign regardless of the polarization angle of the light. In the bipolar regime, the photocurrents are sign-flipping, meaning that they change sign when the polarization angle of the light is rotated by 90°. The transition point corresponds to a large PR, which is the ratio of the maximum and minimum polarization-dependent photoresponse. Moreover, the detector can operate as a self-contained balanced polarization detector, which can measure small polarization-angle perturbation with high sensitivity and low noise. It exhibits a polarization-angle sensitivity of 10.4 μV per degree and a noise density of 0.24 μV Hz−1/2 at an illumination power of 350 μW. Consequently, the noise-equivalent polarization-angle rotation is approximately 0.02° Hz−1/2 (Fig. 14f).

Fig. 14: Plasmon-enhanced detectors.

a Plasmonic mid-infrared polarization detectors⁷. Schematic of the plasmonic photodetector, comprising a graphene transistor adorned with an array of metallic nanoantennas. b Calculated polarization ratio (PR) as a mapping of orientation angle (θ₁, θ₂) values. c Orientation-dependent photoresponse in tapered nanoantennas on graphene. d Calculation of the overall polarization-dependent photoresponse J(θ₁, θ₁, φ). e Photodetector responses for three distinct devices characterized by varying (θ1, θ2) values. f The polarization dependence showing a polarization-angle sensitivity. Reproduced with permission⁷. Copyright 2021 Springer Nature Limited. g Plasmonic silicon-aluminum-nanodisk-based color-sensitive photodetectors²³². Schematic illustration of the color-sensitive detector. h Absorptance spectrum for nanodisk array with a diameter D = 120 nm in Si. i Optical microscope reflection image of the device. j Measured photocurrent as as a function of D under the illumination laser with wavelengths of 488 nm (blue), 532 nm (green), and 633 nm (red). Reproduced with permission²³². Copyright 2022 American Association for the Advancement of Science.

Furthermore, plasmons can contribute to detector miniaturization¹³⁶. Traditional digital camera sensors employ color filters on photodiodes for color selectivity, but the separate placement of color filters and photosensitive silicon layers leads to optical cross-talk, imposing constraints on minimum pixel size. Ho and colleagues reported the design and manufacturing of color-sensitive photodetectors based on silicon-aluminum nanostructures that can achieve submicron-sized pixels and minimal optical crosstalk (Fig. 14g). The detectors consist of silicon nanoposts with aluminum disks on top and aluminum film between the posts (Fig. 14g). The diameter of the silicon nanodisks is varied to achieve different color sensitivity for red (633 nm), green (532 nm), and blue (488 nm). The aluminum layer has two functions: it supports a hybrid plasmon-Mie resonance to enhance light absorption at the selected wavelength, and it forms a Schottky barrier with the p-doped silicon substrate to separate the generated electron-hole pairs. When light illuminates the hybrid nanostructures, it excites a magnetic dipole resonance in the silicon nanodisks, which leads to strong electric field confinement and photon absorption. The absorbed photons generate electron-hole pairs, which are then separated by the internal electric field of the Schottky barrier. The electrons move to the aluminum film, while the holes move to the silicon substrate, creating a photocurrent that can be measured by external electrodes. The photocurrent is proportional to the intensity and wavelength of the incident light, thus enabling color-sensitive photodetection (Fig. 14h). The photocurrent from the nanostructured color detector varies with the nanodisk diameter and the wavelength of the incident light, indicating the wavelength-dependent photoresponse of the nanodisks (Fig. 14i, j). In summary, plasmonic detectors, distinguished by their sensitivity to both light polarization and intensity, present a valuable alternative to conventional counterparts, affording distinctive functionalities.

Integrative enhancements for integrated plasmonic circuits

Continued scaling of microelectronic devices has delivered exponential gains in performance, but also exposed physical and economic limits of top-down silicon technology. One route beyond these limits is to process information in the optical domain. Purely dielectric photonics, however, is constrained by the diffraction limit, which hinders dense on-chip routing and subwavelength active functions. Plasmonic nanostructures address this gap by binding light to free-electron oscillations at metal–dielectric interfaces, confining fields to deep-subwavelength volumes and strengthening light–matter interaction^{137,138,139,140}. In practice, integrated plasmonic circuits are emerging as hybrid platforms that couple dielectric waveguides for low-loss transport with plasmonic elements for compact sources, modulators, switches, detectors, and logic primitives (Fig. 15a)¹⁴¹. The goal is to combine optical bandwidth with nanoscale footprints, electrical addressability, and low switching energy.

Fig. 15: Integrated plasmonic circuits.

a Overview of plasmonic integrated circuit components featuring high functionality on small wafer footprints¹⁴¹. Reproduced with permission¹⁴¹. Copyright 2012 Springer Nature Limited. b Plasmonic logic gates as basic components of digital circuits¹⁴⁵. TEM and SEM image of the device. c Photoluminescence microscopy images using four distinct combinations of polarized inputs. The polarization directions of the two laser beams are denoted by red arrows. The values 1 and 0 were assigned based on the intensities of the scattered surface plasmon polaritons. d, e A fundamental logic gate featuring dual inputs and a single output signal is constructed. The optical logic operations, encompassing AND and OR gates, are succinctly summarized, along with the scattering spectra of plasmons at the O-terminal under excitations denoted in (c). T1 and T2 denote the thresholds for the OR and AND gates, respectively. Reproduced with permission¹⁴⁵. Copyright 2012 John Wiley and Sons. f Plasmonic laser as a nanolight source¹⁴⁹. Schematic representation of the laser. g Working mechanism. Reproduced with permission¹⁴⁹. Copyright 2013 American Chemical Society. h Plasmonic modulator as an adapter¹⁵⁴. Schematic representation of the modulator. i Lumped element model of the modulator. Reproduced with permission¹⁵⁴. Copyright 2014 Springer Nature Limited. j Plasmonic detector as an interface to an electric circuit¹⁶⁰. Schematic representation of the detector. k Spectral responses of the detectors. Reproduced with permission¹⁶⁰. Copyright 2009 Springer Nature Limited. l Plasmonic switcher¹⁶³. A schematic illustration depicting a plasmonic switch based on a pixelated nano-­opto-­electro-­mechanical grating that exploits electromechanically induced symmetry breaking for precise, pixel-­level control of grating coupling strength. m Each grating pixel is individually controlled by distinct electrical biases, allowing programmable coupling strengths. n Square wave driving signal and corresponding optical response at λ = 10 nm. The steady-­state optical switching contrast between “ON” and “OFF” already exceeds 99%. a.u.: arbitrary units. Reproduced with permission¹⁶³. Copyright 2025 American Association for the Advancement of Science.

Plasmonic logic gates

Optical logic operations can be realized by converting changes in excitation conditions into controlled variations of plasmonic output^142,143,144. Fang and co-workers demonstrated a gate based on organic/metal nanowire heterojunctions (Fig. 15b)¹⁴⁵. Organic nanowires were selectively grown at the termini of Ag nanowires by physical vapor transport (Fig. 15b), forming a junction where exciton–polaritons in the organic segment couple to surface plasmons in the metal. Rotating the incident polarization adjusts absorption in the organic nanowire, thereby tuning the exciton–polariton population and, in turn, the intensity of plasmons launched into the metal wire. Logical inputs (I1, I2) are encoded by the polarization states; the output (O) is defined by the scattered plasmon intensity at the metal wire end (Fig. 15c). Using a simple intensity threshold (e.g., 0.4 a.u.) yields OR and AND behaviors without additional circuitry (Fig. 15d, e). This scheme highlights a key advantage of plasmonic logic: compact, directly addressable elements that map optical state to electrical or optical readout within a single nanoscale junction.

Plasmonic light sources

Light sources for on-chip photonics must be both tunable and miniaturized^146,147,148. Liu and colleagues realized wavelength tuning in a single CdS nanowire laser by exploiting a plasmon-enhanced Burstein–Moss (BM) effect (Fig. 15f)¹⁴⁹. The device places a CdS nanowire on a thin SiO₂ spacer above a Au film (Fig. 15f). Adjusting the SiO₂ thickness modulates plasmon–exciton coupling strength, which in turn affects carrier filling and the apparent band-edge shift associated with the BM effect (Fig. 15g). At room temperature, the lasing wavelength was tuned from 504 nm to 483 nm by varying the spacer thickness, accompanied by increased photoluminescence intensity and faster decay—signatures of enhanced radiative rates due to plasmonic coupling. This approach illustrates how plasmonic environments can provide electrical- or geometry-based knobs for compact, integrable, and spectrally agile sources.

Plasmonic modulators

On-chip modulators must be aggressively miniaturized without sacrificing bandwidth. Resonant designs already achieve small footprints (down to ~78 µm²)¹⁵⁰, and plasmonic confinement can shrink active lengths further by enhancing the overlap between optical fields and the drive medium^151,152,153. Melikyan and colleagues demonstrated a plasmonic phase modulator (PPM) that encodes information in the phase of a surface-plasmon polariton at 40 Gbit s⁻¹ with a total device length of only 29 µm (Fig. 15h)¹⁵⁴. The PPM integrates seamlessly with silicon photonics through 15° metal tapers that convert between silicon nanowire modes and plasmonic modes (Fig. 15h). The central phase section comprises two metal electrodes separated by a nanometer-scale vertical slot filled with a nonlinear organic electro-optic polymer. Applying a voltage across the electrodes produces a Pockels-effect index change in the slot, which modulates the SPP phase; a second taper reconverts the plasmon back to a dielectric waveguide mode. With appropriate lumped-element loading (Fig. 15i), the device operates over a ~ 120 nm window centered at 1550 nm, exhibits a flat modulation response to at least 65 GHz, supports 40-Gb/s line rates, and maintains stable operation up to ~85 °C. This architecture highlights the core advantage of plasmonic modulators: strong field confinement yields large phase shifts in extremely short interaction lengths, enabling compact, CMOS-addressable phase control that can be co-packaged with low-loss dielectric routing. Remaining challenges include ohmic loss and thermal management in the metal slot, but the demonstrated bandwidth–footprint combination illustrates a practical pathway to dense, high-speed modulation in integrated plasmonic circuits^155,156.

Plasmonic detectors

Detector design in integrated plasmonic circuits faces two linked challenges: propagation loss and mode integration^157,158,159. Ohmic damping in metal films limits SPP propagation to micrometer–millimeter scales, while the modal mismatch between tightly confined plasmonic fields and conventional photonic/electronic interfaces complicates efficient readout. A representative solution couples a metal–insulator–metal (MIM) waveguide to an inherently fast nanoslit metal–semiconductor–metal (MSM) photodetector, as demonstrated by Neutens and colleagues (Fig. 15j)¹⁶⁰. In this architecture, the MSM nanoslit serves simultaneously as the electrical contact and the near-field coupler, converting the guided plasmon into a photocurrent within a deep-subwavelength interaction region. The device exhibits high signal-to-noise ratio and sub-nanosecond response, making it suitable for high-bandwidth links. Polarization- and wavelength-dependent measurements confirm plasmonic operation, with extracted e⁻¹ propagation lengths ranging from ~3.5 µm at 660 nm to ~9.5 µm at 870 nm (Fig. 15k). This co-designed waveguide–detector approach mitigates coupling loss, preserves compactness, and provides a practical path to plasmon-compatible receivers on chip.

Plasmonic switchers

Plasmonic optical switchers route light from one input to selected outputs by exploiting surface plasmon resonances to confine and modulate fields at deeply subwavelength scales^161,162. Compared with purely dielectric devices, plasmonic platforms offer stronger light–matter interaction, compact footprints, fast response, material and bias tunability, and straightforward co-integration with electronics. These features enable dense, low-energy, and reconfigurable on-chip signal routing for high-speed links, filtering, and computing. For instance, a recent NOEM grating exemplifies these advantages (Fig. 15l)¹⁶³. The device introduces electromechanically induced symmetry breaking to toggle between subwavelength and Bragg regimes, providing large, bias-controlled changes in coupling. Pixel-level actuation gives each unit cell an independently tunable coupling coefficient, allowing programmable spectra (Fig. 15m). Target responses are synthesized with a layer-peeling (inverse-scattering) algorithm, and superposed periodicities form multilayer moiré gratings for complex spectral control. The result is an ultracompact, reconfigurable spectral switch with key metrics: switching time <10 ns (Fig. 15n), extinction >100 dB, energy ~1.1 pJ per event, footprint ~0.007 mm², >99.5% fabrication yield, and stable operation over ~2 × 10¹² cycles.

Integrative enhancements for scattering and fluorescence bioimaging

In vitro and in vivo bioimaging play crucial roles in biological research, medical diagnostics, and treatment monitoring. In vitro bioimaging involves the examination of biological samples outside the living organism, such as cells or tissues in laboratory settings. In vivo bioimaging, on the other hand, focuses on imaging within living organisms to study biological processes and monitor diseases in real-time. Contrast agents, as substances that enhance the contrast of biomaterials in medical imaging, are important for achieving high-sensitivity and high-resolution bioimaging. Plasmonic nanoantennas exhibit significant potential as dual-function contrast agents, seamlessly integrating near-field enhancement for heightened imaging intensity and resolution, alongside thermal enhancement effects^{164,165,166,167}. The former enhances imaging capabilities, while the latter facilitates thermal treatment upon identification of diseased tissue through imaging. Consider the exemplary application of plasmon-enhanced in vivo bioimaging, exemplified by Huang et al.¹⁶⁸. In their study, they showcased the merits of plasmon-enhanced shortwave infrared fluorescence imaging probes for ovarian cancer fluorescence imaging. These probes exhibited attributes such as high brightness, deep tissue penetration, low tissue scattering, and minimal tissue autofluorescence (Fig. 16a). Three variants of imaging probes, distinguished by surface coatings (PSS-terminated, mPEG-terminated, and FA-terminated), were synthesized for comparative analysis. The PSS-terminated probes served as a reference (Fig. 16a), the mPEG-terminated probes were designed to minimize nonspecific interactions in the biological environment (Fig. 16b), and the FA-terminated probes were tailored for active targeting of ovarian cancers with overexpressed folate receptors (Fig. 16c). In in vivo imaging of ovarian cancer, the probes effectively detected individual disseminated tumor nodules and tumor nodules on peritoneal organs, with minimal accumulation in normal tissue. This plasmon-enhanced imaging technique holds potential for aiding fluorescence imaging-guided surgery and enhancing survival rates among ovarian cancer patients.

Fig. 16: Plasmon-enhanced in vitro and in vivo optical bioimaging.

a AuNRs-enhanced in vivo imaging of ovarian cancer¹⁶⁸. Imaging results using sodium polystyrene sulfonate (PSS)-terminated AuNRs. b mPEG-terminated AuNRs. c folic acid (FA)-terminated AuNRs. Reproduced with permission¹⁶⁸. Copyright 2021 John Wiley and Sons. d Typical dark-field microscopy (DFM) and spectroscopy setup. Reproduced with permission¹⁷⁰. Copyright 2012 Royal Society of Chemistry. Typical dark-field microscopy (DFM) and spectroscopy setup. e DFM images of S. aureus. (f) E. coli. (g) S13’ phage-immobilized SiO₂@AuNPs. h S. aureus labeled with S13’ phage-immobilized SiO₂@AuNPs. Reproduced with permission¹⁷¹. Copyright 2019 American Chemical Society.

In addition to imaging large volumes of in vivo tissue, plasmon can also enhance in vitro imaging of micron-sized cells¹⁶⁹. A representative example is the high selectivity of plasmonic enhancement in cell imaging demonstrated by Imai et al. (Fig. 16d)¹⁷⁰. The plasmonic SiO₂@AuNPs are immobilized by S13’ phage, which binds only to specific bacteriophages. The nanoprobes bind to the target bacteria and enhance their light scattering intensity, allowing for easy visualization and quantification. This method has high selectivity because the nanoprobes are designed to bind only to specific bacteriophages, which in turn bind only to specific bacteria¹. This ensures that the nanoprobes only bind to the target bacteria and not to other bacteria or substances that may be present in the sample. The specificity of the method is further enhanced by the use of dark-field microscopy (DFM, Fig. 16d), which allows for the visualization of the nanoprobes and the target bacteria with high contrast and resolution. As demonstrated in four different configurations (Fig. 16e–h)¹⁷¹, the phage-immobilized SiO₂@AuNPs can selectively bind to S. aureus and enhance its light scattering contrast, while E. coli remains unaffected. In addition to the high selectivity, the imaging via the interactions between plasmonic NPs and living cells are influenced by factors such as surface chemistry¹⁷², plasmon coupling¹⁷³, and nanoparticle morphology¹⁷⁴. Various modifications in capping materials enable NPs to exhibit different internalization and localization dynamics in living cells. For instance, a study by Sobral-Filho and colleagues used AuNPs coated with anti-Thy1.1 to distinguish circulating tumor cells from healthy cells in blood samples¹⁷⁵. The coupling of NPs, altering the inter-particle distance, also impacts the interaction with living cells. Researchers, such as Qian et al.¹⁷³, leveraged miRNA-21 to regulate the exocytosis inhibition of AuNPs in living cells, introducing the concept of controllable aggregation-induced exocytosis inhibition. The morphology of NPs plays a crucial role in their imaging and delivery capabilities within living cells by influencing surface chemistry, charge, and functionalization properties¹⁷⁶. Studies have shown that certain nanoparticle shapes, like Au nanorod and Au nanocage, are better suited for imaging, while AuNS is more effective for cellular internalization and biomolecule delivery systems¹⁷⁷.

The optical and thermal enhancement effects of plasmons allow targeted treatment after biological imaging. A successful marriage of these imaging and therapeutic modalities would yield bioimaging-guided on-demand cancer treatments with real-time tumor visualization, high controllability, low side-effects, and excellent therapeutic efficacy. An “all-in-one” NIR-II phototheranostic and imaging agent, dual-plasmonic Au@Cu_2−xSe core-shell nanocrystals (dpGCS NCs), for hypoxia-tolerant chemodynamic-photocatalytic-photothermal cancer therapy were demonstrated by Shan et al. (Fig. 17a)¹⁷⁸. dpGCS NCs are synthesized by a two-step approach, in which Au NPs are coated with an amorphous Se layer and then converted into crystalline Cu_2−xSe in the presence of Cu²⁺ and L-ascorbic acid. dpGCS NCs can produce ‧OH radicals via Fenton-like reaction between Cu²⁺ and H₂O₂ in the tumor microenvironment, which can be enhanced by NIR-II laser irradiation. The NIR-II laser also induces plasmonic hole-induced photocatalytic reaction to generate extra ‧OH radicals. The ‧OH radicals can cause oxidative damage to cancer cells and deplete glutathione, a major antioxidant in the tumor microenvironment. Moreover, the NIR-II laser also triggers the PT effect of dpGCS NCs, which can cause hyperthermia and ablation of cancer cells. Thus, dpGCS NCs enable a synergistic combination therapy for cancer. Compared with CT imaging (Fig. 17b), NIR-II imaging achieves less attenuation of the imaging signal in the NIR-II biowindow than that in the NIR-I biowindow after penetrating the same tissue depth. Besides, NIR-II imaging can use a higher power density of the laser (1.0 W cm⁻² for 1064 nm) than NIR-I imaging (0.33 W cm⁻² for 808 nm), which can improve the imaging quality and contrast. Then, NIR-II imaging can avoid the overheating and photodamage of the tissues caused by the absorption and scattering of the visible or NIR-I light. In the therapy experiment (Fig. 17c), the mice treated with RGD-dpGCS NCs plus NIR-II laser have the highest survival rate, reaching 100% at day 35. The mice treated with RGD-dpGCS NCs alone have a survival rate of about 40% at day 35, which is higher than the control groups. The mice treated with PBS alone or PBS plus NIR-II laser have similar survival rates, which are less than 20% at day 35. While the described method using dpGCS NCs for cancer therapy and imaging shows promising results, it has some limitations. One potential drawback is the reliance on Fenton-like reactions, which involve metal ions and hydrogen peroxide, leading to the generation of hydroxyl radicals. This approach may raise concerns about potential off-target effects and toxicity due to the involvement of reactive oxygen species. Moreover, the use of plasmonic effects induced by NIR-II laser irradiation may introduce challenges related to tissue penetration and potential damage, despite efforts to optimize power density.

Fig. 17: Imaging-guided plasmonic photothermal therapy.

a Schematic illustration of the therapeutic mechanism for plasmonic phototheranostics triggered by a single NIR-II laser (1000–1700 nm wavelength)¹⁷⁸. b Comparison of NIR-II imaging and CT imaging. c The Kaplan–Meier survival curve of U87-MG tumor-bearing mice (n = 5) after various treatments. As observed, the plasmonic phototheranostics is effective. Reproduced with permission¹⁷⁸. Copyright 2022 John Wiley and Sons.

Integrative enhancements for photoacoustic bioimaging and imaging-guided therapy

The utilization of strategies harnessing photoacoustic effects represents a promising avenue in bioimaging¹⁷⁹. Photoacoustic imaging capitalizes on the robust optical absorption of both endogenous molecules (e.g., hemoglobin) and exogenous contrast agents (e.g., plasmonic nanoparticles)^180,181. This heightened absorption yields robust photoacoustic signals, facilitating exceptional imaging contrast even for structures located at considerable depths. In contrast to conventional optical imaging, which contends with light scattering issues in tissues (e.g., DFM), photoacoustic imaging relies on ultrasound detection. Ultrasound waves experience minimal scattering in biological tissues, enabling the detection of photoacoustic signals at significant depths. This characteristic proves advantageous for imaging structures situated deep within biological tissues¹⁸². Notably, upon introducing nanoparticles into biological tissues and subjecting them to laser irradiation, they enhance the efficient conversion of absorbed light energy into localized heat. Consequently, this augmentation amplifies the photoacoustic signals, resulting in enhanced contrast and spatial resolution in bioimaging. Furthermore, Yim et al. demonstrated that the photoacoustic signals can be further improved by modifying plasmonic nanoparticles (Fig. 18a)¹⁸³. They coated plasmonic nanoparticles with polydopamine (PDA), a synthetic melanin, to form AuNP@PDAs (Fig. 18a). The authors demonstrated that AuNP@PDAs showed a threefold higher PA signal than pristine GNRs due to the increased optical absorption, cross-sectional area, and thermal confinement (Fig. 18b). Moreover, the PA signal of AuNP@PDAs only decreased by 29% during the 5 min of laser illumination in the NIR-II window, while significant attenuation (77%) was observed for AuNPs. The AuNP@PDAs maintained 87% of their original PA signal in vivo even after 10 min of laser illumination. The PDA coating also improved the biocompatibility and stability of AuNPs in various biological media.

Fig. 18: Plasmon-enhanced photoacoustic bioimaging.

a In vivo photoacoustic imaging enhanced by plasmonic nanorod–melanin hybrids¹⁸³. Schematic of the gold nanoparticles (AuNPs) and nanorod–melanin hybrids coated with polydopamine (AuNP-PDA). b Comparison of in vivo photoacoustic imaging using AuNPs and AuNP-PDA hybrids as contrast agents. Reproduced with permission¹⁸³. Copyright 2021 American Chemical Society. c Dual ratiometric SERS and photoacoustic imaging of inflammation -related hydrogen peroxide⁷¹. The schematic diagram depicting the utilization of a nanoprobe for SERS imaging and ratiometric PA imaging in the assessment of lipopolysaccharide (LPS)-induced peritonitis. d Ratiometric SERS intensity (I2228/I1418) changes over time. e Photoacoustic imaging at wavelengths of 750 nm and 1250 nm, accompanied by ratiometric photoacoustic imaging, with a scale bar set at 5 mm. Reproduced with permission⁷¹. Copyright 2021 John Wiley and Sons.

Plasmonic nanoparticles play a pivotal role not only in enhancing cellular and tissue imaging but also in facilitating the photoacoustic imaging of transient small molecules, particularly those with brief lifespans such as oxidative species^184,185. Li et al. showcased the applicability of modified plasmonic nanoparticles for H₂O₂ imaging in peritonitis, illustrating heightened sensitivity, accuracy, and real-time in vivo feedback, particularly when synergistically employed with SERS (Fig. 18c)⁷¹. The nanostructure consists of a nanogapped AuNP coated with a mesoporous silica shell loaded with horseradish peroxidase (HRP) and 2,2′-azino-bis(3-ethylbenzothiazoline-6-sulfonic acid) (ABTS), and decorated with gold nanoparticles (AuNPs) modified with 4-mercaptobenzoboric acid and d-(+)-galactose. The nanostructure responds to H₂O₂ by dissociating the AuNPs, which changes the SERS and PA signals of the nanostructure in a ratiometric manner. The nanoprobe is injected into the peritoneal cavity of mice with LPS-induced inflammation, where H₂O₂ is overproduced (Fig. 18c). The H₂O₂ triggers the dissociation of the gold nanoparticles, resulting in the decrease of the SERS signal of MBN at 2228 cm⁻¹, while the SERS signal of NAT at 1418 cm⁻¹ remains constant. This forms a ratiometric SERS signal of I₂₂₂₈/I₁₄₁₈ that reflects the concentration of H₂O₂ (Fig. 18d). The ratiometric PA images show that the inflamed tissues have a lower ratio than the normal tissues, indicating a higher level of H2O2 in the inflammation site (Fig. 18e). This study indicates that the future trajectory of detection technology lies in the amalgamation of diverse detection methods.

Plasmon-enhanced photoacoustic imaging can also guide the thermal effects for therapeutic purposes^12,186. For instance, plasmonic nanoparticles can be designed to accumulate specifically in diseased tissues^187,188. Once localized, these nanoparticles can be activated by laser irradiation to induce various therapeutic effects, such as hyperthermia, photothermal ablation, or drug release. The real-time imaging feedback from the photoacoustic system allows for the monitoring of the therapeutic process, ensuring accurate delivery and minimizing damage to surrounding healthy tissues. One example is the therapy of malignant glioblastoma, which is a highly aggressive and lethal brain tumor that is difficult to treat due to the existence of the blood–brain barrier (BBB) and the low penetration of near-infrared (NIR) light through the skull and scalp. Zhang and colleagues reported the treatment of orthotopic malignant glioblastoma with photoacoustic imaging guided NIR-II photodynamic therapy and chemotherapy by using drug-loaded ultra-small Cu_2−xSe nanoparticles (CS NPs) (Fig. 19a)¹⁸⁹. Furthermore, doxorubicin (DOX), a chemotherapeutic drug, were loaded on the CS NPs for enhanced therapeutic effect. Ultrasound was used focused to temporarily open the BBB and deliver the CS NPs into the tumor tissue. The authors demonstrated that the CS NPs could effectively degrade H₂O₂ and O₂ to produce ˙OH and ¹O₂ radicals, which could induce apoptosis of tumor cells (Fig. 19b). Besides, CS NPs could be efficiently transported into the orthotopic malignant glioblastoma with the assistance of focused ultrasound (Fig. 19c). It was observed that the tumor growth was significantly suppressed by the combined NIR-II photodynamic therapy and chemotherapy (Fig. 19d). This validates the advantages of using the NIR-II window for deeper penetration and higher permissible exposure of laser light.

Fig. 19: Photoacoustic imaging-guided plasmonic therapy.

a Photoacoustic imaging and imaging-guided NIR-II photodynamic therapy using ultra-small plasmonic Cu₂ − xSe nanoparticles¹⁸⁹. Illustration depicting the application of photoacoustic imaging in the treatment of orthotopic malignant glioblastoma in vivo. b Photoacoustic images of orthotopic malignant glioblastoma cells at various time intervals in mice subjected to either sonication or not, following intravenous injection of CS-D NPs. c The utilization of Ultra-small Cu₂ − xSe nanoparticles in generating reactive oxygen species for tumor therapy through electron transfer (for ‘OH generation)and energy transfer (for ¹O₂ generation) mechanisms. d H&E staining of brain slices from different treatment groups. Reproduced with permission¹⁸⁹. Copyright 2019 Royal Society of Chemistry. e Plasmonic non-wetting Au-on-AuNR homometallic nanostructures for photoacoustic imaging and imaging-guided therapy¹⁹⁰. The schematic diagram depicting non-wetting Au-on-AuNR homometallic nanostructures with the corresponding TEM image. f In vivo infrared thermal imaging of mice subjected to laser irradiation. g NIR-II photoacoustic images of mice administered with nanocoral-3c. h Excised tumors from mice following a complete treatment regimen, accompanied by photographic documentation of three groups (Control, Dark, Laser). i Quantification of signal intensity at the tumor site corresponding to f photoacoustic imaging. Reproduced with permission¹⁹⁰. Copyright 2020 John Wiley and Sons.

Beyond heterometallic nanostructures, homometallic plasmonic architectures have also demonstrated remarkable potential for synergistic photothermal and photoacoustic applications. By fine-tuning interfacial energy and growth kinetics, Au-on-Au nanohybrids with controllable morphologies were synthesized, ranging from spherical islands to branched nanocorals, to achieve continuous regulation of LSPR across the visible–NIR spectrum (Fig. 19e–i)¹⁹⁰. This structural modulation enabled strong broadband absorption and exceptional photothermal conversion, particularly within the NIR-II window. Both in vitro and in vivo evaluations confirmed that the nanocoral structures function effectively as dual-mode agents for photothermal therapy and photoacoustic imaging, offering precise energy deposition, deep-tissue penetration, and high therapeutic efficacy under NIR-II irradiation.

Integrative enhancements for steganography, encryption, and data storage

Plasmon-enabled steganography, encryption, and storage represent promising applications in the field^191,192,193. The core principle behind them is that plasmons are programmable. Plasmons can be manipulated by adjusting their structure, shape, size, and material composition to achieve specific optical responses. This enables plasmons to be used for embedding, concealing, or encoding information. By controlling the properties of plasmons, such as their frequency, amplitude, polarization state, or spatial distribution, it becomes possible to encode and encrypt information. Decoding requires understanding the encoding methods and how to interact with plasmons to recover the original information. Plasmonic encryption schemes can offer enhanced security compared to traditional encryption methods, as they rely on the unique properties of plasmonic materials that are difficult to replicate or decipher without specialized knowledge or equipment¹⁹¹. Additionally, plasmonic encryption can provide opportunities for multi-level security by incorporating multiple layers of encryption based on different plasmonic properties.

In a steganography application, variations in the resonant frequency of plasmonic nanostructures can encode binary data that is invisible to the naked eye but can be extracted using specialized detection techniques. For instance, Yang et al. developed a laser-scanning-based strategy to selectively remove polymers on top of plasmonic nanoparticles and guide the assembly of another type of nanoparticles onto the exposed surface to form dimer arrays with tunable resonant frequency (Fig. 20a)¹⁹⁴. The formation of dimer arrays is contingent upon the locally confined heating of photothermal effect of plasmonic particles. They demonstrated that the method can produce high-resolution patterns of plasmonic dimers from nanoparticles of different sizes and shapes, and that the patterned dimer arrays can display and erase plasmonic color by coating and removing polymer layers, enabling their use in information encryption (Fig. 20b). In addition to resonant frequency, the structural color and amplitude modulation can be combined to serve as an optical information concealment/retrieval strategy.

Fig. 20: Plasmonic metamaterials for optical steganography and encryption.

a Plasmonic thermal effect-enabled optical steganography²³³. Laser scanning guides the assembly of quasi-3D patterned arrays of plasmonic nanoparticle (NP) dimers. Localized photothermal heating selectively removes polymers on gold nanoparticles (GNPs). b The patterned quasi-3D NP dimers exhibit plasmonic coloration post-polymer removal, which can be restored by applying a polymer coating. Reproduced with permission²³³. Copyright 2021 John Wiley and Sons. c Plasmon-enhanced chirality for information encryption¹⁹⁷. Schematic representations of original (blue), vertical (red), and diagonal (green) grating patterns for LH structures. d Corresponding enlarged images. e Schematic illustrations depicting original grating patterns (left), accompanied by vertically formed nanopatterns (center), and diagonal nanopatterns (right) for LH-chiral plasmonic structures. f Schematic diagrams delineating applications in anti-counterfeiting and as chiral quantum dot (QD) codes. Reproduced with permission¹⁹⁷. Copyright 2023 Springer Nature Limited.

In the realm of encryption, the utilization of plasmonic properties for encoding and decoding information stands as a pivotal concern^195,196. Through the encryption procedure, the initial data undergoes transformation into a format delineated by distinct attributes of plasmons. Notably, chirality emerges as a distinctive feature of asymmetric plasmonic nanostructures, amenable to manipulation via mechanical stretching and metal deposition, as evidenced by Lee and colleagues (Fig. 20c)¹⁹⁷. They use the chiral diagonal nanopatterns as a substrate for coating quantum dots that emit circularly polarized light (CPL) under linearly polarized light (LPL) excitation (Fig. 20d, e). By controlling the handedness and intensity of the CPL emission, the authors can encode information in the form of chiral QR codes or anti-counterfeiting patterns (Fig. 20f). The information can only be decrypted by using a CPL filter and a specific wavelength of LPL. This plasmon encryption method is of high security: The information is hidden in the chiral plasmonic nanostructures that are not visible to the naked eye or conventional optical devices. The information can only be revealed by using a specific combination of CPL filter and LPL wavelength, which is difficult to guess or replicate.

Plasmons for data storage is an emerging field that explores the utilization of plasmonic properties for storing and retrieving information at the nanoscale^198,199. In plasmonic data storage, information is typically encoded as nanostructured surfaces, such as plasmonic gratings or arrays of nanoantennas. By selectively exciting plasmonic modes in these structures, data can be stored as variations in the intensity, wavelength, or polarization of light interacting with the plasmonic resonances. For instance, Song et al. developed a metasurface platform based on surface-relief aluminum nanoantennae that can reflect polarization-tunable plasmonic colors for data storage (Fig. 21a)²⁰⁰. Variations in the polarizer-analyzer angle combination produce four discernible colors (Fig. 21b). The sequence of four colors is utilized as a color code, which is then assigned a 3 bit information state, so that a piece of 3 bit information can be stored in a nanopixel (Fig. 21c, d). A parallel readout system consists of four white-light sources and four CCD cameras, each with a different analyzer angle (Fig. 21e, f). The system can simultaneously image multiple nanopixels in a frame and store their color information in a cache. They claim that the plasmon methods can achieve higher storage density and readout speed than conventional Blu-ray technology. Recently, rare earth (RE) dopants were proposed as optical centers for all-optical data storage and encryption (Fig. 21g, h)²⁰¹. The authors use a plasmon-driven rapid in situ crystal transformation strategy to obtain composite structures of Y2O3:Eu3+ embedded with Au nanoparticles, which can rapidly switch between two stable luminescent states (providing “0” and “1”) under resonant light excitation (Fig. 21g). The information of the light power is written into the composite structure during the plasmon-assisted crystal transformation, and can be read out by the luminescence switching behavior. The information can also be erased and rewritten by light irradiation with higher power, which causes the adaptive adjustment of the composite structure and the change of the light-to-heat conversion efficiency (Fig. 21h–j). This way, plasmon implements self-encrypted and reversible data storage based on the luminescence modulation of rare earth oxide crystals. In general, the key advantages of plasmonic data storage are its potential for ultra-high storage densities. The ability to manipulate light at the nanoscale allows for the packing of data at extremely high densities, surpassing the limits of conventional magnetic or optical storage technologies. Additionally, plasmonic data storage systems may offer faster read and write speeds, as well as improved durability and stability. Notably, there are also challenges and limitations associated with plasmonic data storage. These include issues related to scalability, compatibility with existing technologies, and the development of efficient readout mechanisms.

Fig. 21: Plasmonic metamaterials for data storage.

a Geometric and polarization characteristics of plasmon resonances²⁰⁰. Reproduced with permission²⁰⁰. Copyright 2021 John Wiley and Sons. Calculated phase difference in reflected light between x- and y-polarizations. The four upper insets show the elliptical polarization states of the reflected wave at four selected wavelengths, where the ellipse shape and orientation (rather than color) represent the polarization state. The lower inset depicts a schematic of a single unit cell housing an individual Al nanoantenna under the incident polarization state. b Simulated reflectance spectra for various polarizer–analyzer angle combinations, resulting in four discernible colors. c Plasmonic geometric and polarization variations for data storage demonstration. A frame (red box) of an plasmonic data-storage unit comprises 16 nanopixels, with each nanopixel (highlighted by the purple box) consisting of four unit cells (highlighted by the blue box). Each nanopixel functions as an indivisible storage unit, capable of storing 3 bits of information. d A distinct color code is generated for each nanoantenna orientation, which is then mapped to a 3-bit information state, facilitating the storage of 3 bits of information within a nanopixel. e Information retrieval mechanism. f Binary data is retrieved by comparing the color sequence of each nanopixel with the corresponding color codes in (c). g Schematic illustration of plasmon-assisted writing and reading of rare earth oxide crystal²⁰¹. Reproduced with permission²⁰¹. Copyright 2022 John Wiley and Sons. h Self-encrypted binary information storage using the rare earth oxide crystal. Optical stimuli-responsive Y₂O₃:Eu³⁺–Au NPs, exhibit rapid switching between enhanced and quenched states, representing “1” and “0”, respectively. Lower panel illustrates information retrieval outcomes under varying read powers. i Luminescence switching of Y₂O₃:Eu³⁺–Au NPs occurs via adjustment of excitation light intensity, toggling between enhanced and quenched states. j Input–output configuration and truth table elucidate the AND gate functionality, generating a photonic output solely when input light satisfies specific frequency and power criteria.

While these plasmonic architectures are primarily developed for high-density classical information storage and optical encoding, it is useful to briefly compare plasmonic-based data storage with quantum memory systems. Plasmonic-based data storage systems exploit deep-subwavelength optical confinement and classical light–matter interactions to achieve ultrahigh information density through spatial, spectral, polarization, or phase encoding. Such platforms typically operate at room temperature, are compatible with standard nanofabrication processes, and offer fast optical write/read speeds with good prospects for on-chip integration. However, their performance is constrained by intrinsic plasmonic losses, thermal effects under repeated cycling, and long-term material stability, and the stored information is limited to classical degrees of freedom without preserving quantum coherence. In contrast, quantum memory systems are designed to store and retrieve quantum states of light or matter, enabling the preservation of coherence, superposition, and entanglement, which are essential for quantum communication and networking. These advantages come at the cost of stringent requirements on coherence time, environmental isolation, and noise suppression, and often involve complex physical platforms and control schemes, sometimes requiring cryogenic operation.

Future perspectives and conclusions

Looking ahead, the future of integrative metamaterials is bright and multifaceted (Fig. 22). On the fundamental principles side, we can push into quantum metamaterials, examining how quantum effects (tunneling, nonlocal responses, quantum emitter–plasmon interactions) can be harnessed for even greater control and novel functionalities. Plasmonic devices operating in the quantum regime could facilitate single-photon transistors or enable room-temperature strong coupling with single molecules, opening doors in quantum information processing²⁰². Additionally, coupling metamaterials with spintronics (controlling spin-polarized currents via plasmons) and topological photonics (robust light transport immune to scattering) is on the horizon^203,204. Topologically protected plasmonic modes, for instance, might allow loss-mitigated propagation or new sensor paradigms leveraging topological sensitivity.

Fig. 22

Future development pathways of plasmonic metamaterials.

In terms of materials, we anticipate a broadening beyond noble metals. Emerging Dirac and Weyl semimetals²⁰⁵, graphene and its derivatives, conducting oxides (like ITO), and even hybrid organic–plasmonic materials (polymer composites with metal nanoparticles) are being explored to overcome limitations of gold and silver. Graphene plasmons, as discussed, bring tunability and extreme confinement; new materials like TiN, transparent conductive oxides, and 2D materials can support plasmons with lower losses or at more convenient spectral ranges (e.g., mid-IR). Hydrogels and soft plasmonic materials offer flexibility and environmental responsiveness, enabling plasmonic sensors that can stretch or biocompatible implants that integrate metamaterials into tissue²⁰⁶. Such material innovations will allow plasmonic systems that are adaptive, stretchable, or even self-healing.

On the technological front, plasmonic metamaterials will benefit from and contribute to advances in machine learning, edge computing, and in-sensor computing. Machine learning algorithms are already being used to inverse-design complex plasmonic structures (finding the pattern that yields a desired multi-wavelength, multi-polarization response)^{77,207,208,209,210,211}. This greatly accelerates innovation in multi-functional metasurfaces for simultaneously focusing and filtering light. Conversely, plasmonic circuits might implement neural network-like analog computing: plasmonic diffractive networks can perform matrix multiplications at the speed of light, relevant for edge computing tasks²¹². In-sensor computing could be achieved by plasmonic sensor arrays that perform optical pre-processing (like computing a principal component of a spectrum) before the signal is even digitized^{213,214,215,216,217,218,219,220}. The massive parallelism and sub-nanosecond latency of optical processes make metamaterials an attractive hardware platform for AI and signal processing at the sensor level.

In terms of emerging applications, the versatility of plasmonic multi-effects is poised to impact telecommunications (5G/6G)²²¹, virtual and augmented reality (VR/AR)^222,223,224, and point-of-care diagnostics²²⁵. For 5G/6G, plasmonic antennas and modulators can support the ultrawide bandwidths and tiny form factors needed in next-gen network devices; for instance, plasmonic THz modulators could be key in future wireless links²²⁶. In VR/AR, plasmonic metasurfaces could act as holographic displays or retina projections—creating vivid imagery in a lightweight glass by manipulating light at sub-wavelength scales (some startup companies are already developing metasurface-based AR displays)²²⁷. Plasmonic structural colors may replace pigment-based displays for more saturated, power-efficient screens. And in healthcare, point-of-care testing using plasmonic sensors could enable immediate detection of diseases (e.g., via a handheld SERS device analyzing a drop of blood for multiple biomarkers at once, with AI interpreting the results)²²⁵.

Collectively, plasmonic metamaterials are evolving into a cornerstone of nanoscience and technology, analogous to the role of the semiconductor transistor in the 20th century. It offers a path to continue the trajectory of device miniaturization and performance enhancement beyond the conventional limits by blending the best of multiple worlds, including optical bandwidth, electronic control, thermal tunability, and acoustic sensitivity. The challenges ahead, such as reducing losses, scalable manufacturing, and system-level integration, are significant but surmountable with interdisciplinary efforts bridging physics, engineering, chemistry, and biology. As these are addressed, we foresee plasmonic metamaterials moving from laboratories into everyday devices, such as invisibly embedded in high-speed chips, in smart wearables monitoring health, in secure authentication tokens, and in environmental sensor networks through the Internet of Things.

Building on the opportunities and the application examples, it is equally important to recognize the main obstacles that still limit truly integrative plasmonic systems. The points below are not intended to downplay the impressive progress in the field, but to highlight open questions that, in our view, deserve focused attention.

Multiple optical enhancement channels on a single platform

A first challenge is the design of nanostructures that can host several optical enhancement mechanisms within one compact footprint. Each technique favors different mode profiles, resonance wavelengths and metal–emitter separations, so a geometry optimized for one often compromises another. Hybrid and layered architectures, tunable metasurfaces and multi-modal resonators offer promising routes, but they demand very tight control of thickness, composition and surface quality to keep spectral overlap, field strength and selectivity simultaneously high.

Fusing optical, electrical, thermal, and acoustic effects

Going beyond all-optical schemes, cross-domain fusion of optical, electrical, thermal and acoustic responses introduces additional constraints. Each domain is governed by different figures of merit, such as quality factor and radiation efficiency for optics, mobility and contact resistance for electronics, mechanical impedance for acoustics, and conductivity and heat capacity for thermal transport. Plasmonic structures can, in principle, mediate energy flow between these channels, but examples where all four are efficiently coupled and controllable are still rare.

Molecular diagnostics and point-of-care implementations

For molecular diagnostics, integrative plasmonics must meet several demanding criteria at once: high sensitivity across multiple modalities, chemical selectivity in complex samples, and robust, quantitative performance. Hot-spot engineering is crucial for SERS and related methods, but substrates that maximize enhancement for one modality may not be ideal for SEIRA, SEF or photoacoustic readout. Designing platforms that provide sufficiently strong and well-distributed enhancement across different spectral windows and readout mechanisms is still challenging.

Plasmonic photonic integrated circuits

In plasmonic photonic integrated circuits, loss and integration remain the dominant bottlenecks. Plasmonic modes suffer from ohmic dissipation and scattering at bends, junctions and couplers; amplification using gain media or quantum emitters can offset these losses but adds complexity and power overhead. At the same time, emitters, modulators, switches and detectors often require different materials and dimensions, so heterogeneous integration on CMOS-compatible platforms is technically demanding.

In summarization, plasmonic metamaterials have evolved from purely optical structures into versatile platforms where light, charge, heat, and mechanical motion can be tightly coupled and engineered together, enabling advances in biosensing, imaging, information processing, and beyond. By concentrating electromagnetic fields and controlling their interaction with matter, they provide powerful routes to enhance spectroscopic sensitivity, steer carrier dynamics, tailor nanoscale temperature profiles, and interface with acoustic degrees of freedom. At the same time, realizing truly integrative devices that exploit several of these channels in a controlled and scalable way still demands progress in principles, materials, technologies, and applications. Looking ahead, continued efforts at the intersection of plasmonics, materials science, device physics, and data-driven optimization are expected to translate current proof-of-concept demonstrations into robust, multifunctional platforms that operate reliably in realistic environments and applications.