10.6 Surface plasmons

Surface plasmon fundamentals

Surface plasmons are collective oscillations of free electrons at the interface between a metal and a dielectric material. They form the basis of plasmonics and are central to applications in sensing, imaging, and nanophotonics. This section covers their physical origin, the conditions they require, and how their dispersion differs from free-space light.

Definition of surface plasmons

Surface plasmons are coherent electron oscillations that exist at the interface between a dielectric and a conductor, evanescently confined in the perpendicular direction. When these oscillations couple with electromagnetic waves, they form surface plasmon polaritons (SPPs).

SPPs are transverse magnetic (TM) waves that propagate along the metal-dielectric interface and decay exponentially into both media. The TM polarization requirement is fundamental: TE-polarized surface modes cannot satisfy the boundary conditions at a metal-dielectric interface, so only TM modes survive.

Conditions for existence

For surface plasmons to exist, the real part of the metal's dielectric function must be negative, and its magnitude must exceed the dielectric constant of the insulator:

$\text{Re}(\varepsilon_m) < 0 \quad \text{and} \quad |\text{Re}(\varepsilon_m)| > \varepsilon_d$

Several metals satisfy this in the visible and near-infrared range, most notably silver, gold, and aluminum. The dielectric side can be air, glass, or any insulating material with a positive dielectric constant. Silver tends to have the lowest losses (sharpest resonances), while gold is preferred in biological applications because of its chemical stability.

Dispersion relation

The SPP dispersion relation connects frequency $\omega$ to the in-plane wavevector $k_{\text{SPP}}$:

$k_{\text{SPP}} = \frac{\omega}{c} \sqrt{\frac{\varepsilon_m \, \varepsilon_d}{\varepsilon_m + \varepsilon_d}}$

where $\varepsilon_m$ and $\varepsilon_d$ are the dielectric functions of the metal and dielectric, respectively.

The key consequence: $k_{\text{SPP}}$ is always larger than the free-space wavevector $k_0 = \omega/c$ at the same frequency. This means the SPP dispersion curve lies to the right of the light line, so free-space photons cannot directly excite SPPs. You always need a coupling mechanism to supply the extra momentum.

At low frequencies, the SPP dispersion approaches the light line. At high frequencies, it asymptotically approaches the surface plasmon frequency $\omega_{sp} = \omega_p / \sqrt{1 + \varepsilon_d}$, where $\omega_p$ is the bulk plasma frequency.

Propagation length

The propagation length $L_{\text{SPP}}$ is the distance over which the SPP intensity decays to $1/e$ of its initial value. It is governed by:

• Absorption losses in the metal (ohmic dissipation)
• Scattering losses from surface roughness and grain boundaries

$L_{\text{SPP}}$ depends on frequency and material choice. At visible and near-infrared frequencies, typical values on silver and gold surfaces range from a few micrometers to several tens of micrometers. There is a fundamental trade-off here: tighter field confinement (desirable for nanoscale applications) comes at the cost of shorter propagation lengths, because the field penetrates more deeply into the lossy metal.

Excitation of surface plasmons

Because SPPs carry more momentum than free-space light at the same frequency, you need a technique to bridge the momentum gap. Three main approaches exist: prism coupling, grating coupling, and near-field excitation.

Prism coupling techniques

Prism coupling uses evanescent wave coupling through a high-index prism to provide the extra in-plane momentum. The two standard configurations are:

1. Kretschmann configuration: A thin metal film (typically 40-50 nm) is deposited directly on the prism base. Light undergoes total internal reflection at the prism-metal interface, generating an evanescent wave that tunnels through the thin film and excites SPPs on the far (metal-dielectric) side.
2. Otto configuration: The prism is placed near the metal surface with a small air gap (on the order of the wavelength). The evanescent wave from total internal reflection at the prism-air interface tunnels across the gap to excite SPPs on the metal surface.

The Kretschmann configuration is far more common in practice because it's easier to fabricate. The Otto configuration is useful when you don't want to deposit metal directly on the prism, or when studying bulk metal surfaces.

In both cases, SPP excitation appears as a sharp dip in the reflected intensity at a specific angle of incidence (the resonance angle), where energy is transferred from the incident beam into the surface plasmon.

Grating coupling methods

A periodic surface structure (diffraction grating) with period $\Lambda$ adds discrete momentum kicks of $\pm m \cdot 2\pi/\Lambda$ (where $m$ is the diffraction order) to the in-plane component of the incident light's wavevector. Coupling to SPPs occurs when:

$k_{\text{SPP}} = k_0 \sin\theta \pm m \frac{2\pi}{\Lambda}$

The grating period and angle of incidence together determine which SPP mode is excited and how efficiently. This approach is particularly useful for integrated plasmonic devices because the grating can be fabricated directly into the metal surface.

Near-field excitation

Near-field techniques use the evanescent field of a subwavelength source to couple directly to SPPs. Common sources include:

• Scanning near-field optical microscope (SNOM) tips
• Quantum dots or other nanoscale emitters
• Sharp defects or nanostructures on the surface

Because the evanescent field inherently contains high-$k$ components, no additional phase-matching is needed. This allows localized excitation of surface plasmons with spatial resolution well beyond the diffraction limit, which is valuable for probing the local properties of SPPs and for nanoscale optical device applications.

Properties of surface plasmons

Field distribution

The electromagnetic field of an SPP is tightly confined to the metal-dielectric interface. The field intensity peaks at the interface and decays exponentially into both media:

• Into the dielectric: the decay length (skin depth) is typically on the order of half the wavelength, so a few hundred nanometers at visible frequencies.
• Into the metal: the decay length is much shorter, roughly 20-30 nm for silver and gold at optical frequencies, set by the metal's skin depth.

This strong confinement produces intense field enhancement near the surface, which is the physical basis for techniques like surface-enhanced Raman spectroscopy (SERS) and surface-enhanced fluorescence (SEF).

Localization and confinement

Surface plasmons can be localized in subwavelength structures such as nanoparticles, nanoantennas, and nanogaps. These localized surface plasmons (LSPs) exhibit even stronger field confinement and enhancement than propagating SPPs. The resonance properties of LSPs can be tuned by controlling the size, shape, and composition of the nanostructure.

This tunability enables manipulation of light at the nanoscale, with applications in nanophotonics, optical data storage, and quantum information processing.

Sensitivity to surface conditions

SPP properties are extremely sensitive to the dielectric environment at the metal surface. Even small changes in the refractive index of the surrounding medium, or the adsorption of a molecular monolayer, can measurably shift the SPP dispersion relation and resonance conditions.

This sensitivity is what makes surface plasmon resonance (SPR) biosensing possible: the binding of analyte molecules to the metal surface changes the local refractive index, which shifts the resonance angle or wavelength in a detectable way.

Comparison to bulk plasmons

Applications of surface plasmons

Surface-enhanced spectroscopy

The intense near-field enhancement around plasmonic structures amplifies optical processes occurring near the metal surface.

• SERS (Surface-Enhanced Raman Spectroscopy): The Raman scattering cross-section scales roughly as $|E/E_0|^4$, where $E/E_0$ is the local field enhancement factor. Enhancement factors of $10^6$ to $10^{10}$ are routinely achieved, enabling detection down to the single-molecule level.
• Surface-Enhanced Fluorescence (SEF): The enhanced local field boosts both excitation and emission rates of fluorophores near the surface, improving sensitivity and reducing photobleaching. The metal-fluorophore distance must be carefully controlled, though: too close (< ~5 nm) and non-radiative quenching dominates.

Biosensing and chemical sensing

SPR biosensing is one of the most commercially successful applications of surface plasmons. The basic principle:

1. A thin gold film is functionalized with receptor molecules (antibodies, aptamers, etc.).
2. A sample containing the target analyte flows over the surface.
3. Binding of analyte to the receptors changes the local refractive index.
4. This shift is detected as a change in the SPR resonance angle or wavelength.

SPR sensors are label-free (no fluorescent tags needed) and provide real-time kinetic data on binding events. They are widely used in drug discovery, environmental monitoring, and clinical diagnostics. Typical refractive index sensitivities are on the order of $10^{-5}$ to $10^{-6}$ refractive index units (RIU).

Subwavelength optics

Surface plasmons can confine and guide light below the diffraction limit. Plasmonic nanostructures such as nanoantennas, nanolenses, and nanogratings concentrate light into subwavelength volumes, enabling:

• High-resolution imaging beyond the diffraction limit
• Nanoscale lithography
• Plasmonic metamaterials with engineered optical properties, including superlenses and negative refractive index materials

Plasmonic waveguides and circuits

SPPs can be guided along metal-dielectric interfaces in waveguide geometries, confining light to subwavelength cross-sections. This opens the door to integrating photonic and electronic functionality on the same chip. Plasmonic circuits can perform switching, modulation, and logic operations using external stimuli (electric fields, optical pulses, thermal effects).

The main challenge remains the propagation loss: plasmonic waveguides are inherently lossy, so there is an ongoing trade-off between confinement and propagation distance in circuit design.

Localized surface plasmons

Localized surface plasmons (LSPs) are non-propagating excitations of conduction electrons in metallic nanostructures such as nanoparticles, nanorods, and nanoshells. Unlike propagating SPPs, LSPs can be excited by direct illumination without any phase-matching technique.

Definition and properties

When light illuminates a metallic nanoparticle much smaller than the wavelength, the oscillating electric field drives the conduction electrons collectively back and forth. At the resonance frequency, this produces a strong, localized enhancement of the near field around the particle.

The resonance frequency depends on:

• Size of the nanostructure (larger particles red-shift)
• Shape (rods, triangles, and stars have different and often multiple resonances)
• Composition (silver vs. gold vs. alloys)
• Dielectric environment (higher surrounding refractive index red-shifts the resonance)

Resonance conditions

For a spherical nanoparticle much smaller than the wavelength (quasi-static limit), the polarizability is:

$\alpha = 4\pi a^3 \frac{\varepsilon_m - \varepsilon_d}{\varepsilon_m + 2\varepsilon_d}$

where $a$ is the particle radius. The resonance occurs when the denominator is minimized, giving the Fröhlich condition:

$\text{Re}(\varepsilon_m) = -2\varepsilon_d$

For non-spherical particles, the resonance conditions become more complex. The factor of 2 is replaced by a geometry-dependent depolarization factor, and multiple resonance modes can appear. Numerical methods such as the discrete dipole approximation (DDA) or finite-difference time-domain (FDTD) simulations are typically needed to compute the resonances of complex shapes.

Field enhancement

LSPs generate intense field enhancement near the nanostructure surface, with the field decaying rapidly over a distance comparable to the particle size. Enhancement factors can reach several orders of magnitude, and the strongest enhancements occur at:

• Sharp tips and edges
• Narrow gaps between adjacent nanostructures ("hot spots")
• Regions of high curvature

These hot spots are responsible for the enormous SERS enhancement factors observed in nanoparticle aggregates and engineered nanostructure arrays.

Applications in sensing and imaging

LSPR sensing works by monitoring shifts in the resonance wavelength (or changes in the extinction spectrum) of nanostructures as analyte molecules bind to their surface. Compared to propagating SPR sensors, LSPR sensors:

• Have smaller sensing volumes (more localized)
• Can detect binding events at the single-molecule level
• Are simpler to implement (no prism or grating needed; just measure the extinction spectrum)
• Can be integrated into multiplexed, chip-based formats

LSPs also enable high-resolution near-field imaging and tip-enhanced spectroscopy, exploiting the subwavelength field confinement around the nanostructure.

Advanced topics in surface plasmons

Nonlinear plasmonics

The strong field enhancement near plasmonic structures dramatically boosts nonlinear optical processes. Because nonlinear signals scale as high powers of the local field (e.g., second harmonic generation scales as $|E|^4$, third harmonic as $|E|^6$), even modest field enhancements translate into large nonlinear signal gains.

Plasmonic nanostructures can serve as nanoscale sources of nonlinear signals, enabling techniques like nonlinear microscopy with subdiffraction resolution. More exotic phenomena, including plasmonic solitons and self-phase modulation, have also been explored theoretically and experimentally.

Quantum plasmonics

Quantum plasmonics investigates what happens when surface plasmons interact with quantum emitters (quantum dots, single molecules, nitrogen-vacancy centers in diamond) in regimes where quantum effects matter.

Key phenomena include:

• Purcell enhancement: The high local density of optical states near a plasmonic structure can dramatically increase the spontaneous emission rate of a nearby emitter.
• Strong coupling: When the plasmon-emitter coupling rate exceeds both the plasmon damping rate and the emitter decay rate, the system enters the strong coupling regime, producing hybrid plasmon-exciton states (plexcitons).
• Single-photon sources and nanolasers: Plasmonic cavities with extremely small mode volumes can enable single-photon emission and lasing at the nanoscale (spasers).

Quantum plasmonics also probes fundamental questions about decoherence and entanglement in lossy, nanoscale systems.

Graphene plasmonics

Graphene, a single atomic layer of carbon, supports surface plasmons with properties quite different from those in metals:

• Extreme confinement: Graphene plasmon wavelengths can be 10-100 times shorter than the free-space wavelength at the same frequency.
• Tunability: The carrier density in graphene (and thus the plasmon properties) can be tuned in situ via electrostatic gating or chemical doping. This is a major advantage over metal plasmonics, where the optical properties are fixed by the material.
• Frequency range: Graphene plasmons are most prominent in the terahertz and mid-infrared, complementing metal plasmonics which operates mainly in the visible and near-IR.

Applications include tunable THz and mid-IR sensors, modulators, and detectors, as well as novel optoelectronic devices.

Chiral surface plasmons

Chiral surface plasmons arise in plasmonic structures that lack mirror symmetry (helices, twisted dimers, gammadion-shaped nanostructures). These modes interact differently with left- and right-circularly polarized light, producing enhanced circular dichroism (CD) signals.

The strong field enhancement associated with chiral plasmons can amplify the inherently weak chiral optical response of molecules by several orders of magnitude. This has practical implications for:

• Detection and discrimination of enantiomers in pharmaceutical applications
• Chiral sensing at very low concentrations
• Fundamental studies of light-matter interactions involving chirality

Numerical methods for surface plasmons

Designing and understanding plasmonic structures almost always requires numerical simulation, because analytical solutions exist only for the simplest geometries (infinite planar interfaces, spheres in the quasi-static limit).

Finite-difference time-domain (FDTD)

FDTD solves Maxwell's equations directly on a discrete spatial grid, updating the electric and magnetic fields at alternating half-time-steps (the Yee algorithm).

Strengths:

• Naturally handles broadband simulations (a single run gives the response over a wide frequency range)
• Can incorporate dispersive, nonlinear, and anisotropic materials
• Conceptually straightforward

Limitations:

• Requires a uniform or structured grid, so resolving very small features (e.g., 1 nm gaps) demands a very fine mesh and large memory
• Computationally expensive for large 3D structures
• Staircase approximation of curved surfaces can introduce errors unless subpixel averaging is used

Finite element method (FEM)

FEM divides the computational domain into an unstructured mesh of elements (typically triangles in 2D, tetrahedra in 3D) and approximates the fields using polynomial basis functions on each element.

Strengths:

• Handles complex, irregular geometries naturally through adaptive meshing
• Can refine the mesh locally near features of interest (e.g., sharp tips, narrow gaps)
• Well-suited for eigenmode analysis (finding resonance frequencies and mode profiles)

Limitations:

• Can be memory-intensive for large 3D problems
• Careful mesh generation is needed to ensure accuracy
• Typically solves at a single frequency per run (frequency-domain), so broadband characterization requires multiple runs

Green's function techniques

Methods based on integral formulations of Maxwell's equations, including the discrete dipole approximation (DDA) and the boundary element method (BEM), express the fields in terms of the system's Green's function.

• DDA represents the scatterer as a collection of polarizable point dipoles. The coupled-dipole equations are solved to find the dipole moments, from which the scattered fields are computed. DDA is flexible for arbitrary shapes but becomes expensive for large particles.
• BEM discretizes only the boundaries of the scatterer (not the volume), solving for surface currents and charges. This makes it very efficient for problems with homogeneous regions separated by well-defined interfaces.

Both methods are efficient when the background medium is homogeneous, and they provide physical insight through quantities like the induced surface charge distribution. They are less naturally suited to inhomogeneous or nonlinear media.

Comparison of numerical methods

The choice of method depends on the geometry, material complexity, and the type of information you need from the simulation.

Experimental techniques for surface plasmons

Near-field scanning optical microscopy (NSOM)

NSOM uses a subwavelength aperture or sharp tip scanned across the sample surface to map the evanescent field of surface plasmons with nanoscale spatial resolution (typically 20-100 nm, well below the diffraction limit).

The technique provides direct images of SPP propagation, interference, and confinement. It can also map the local density of optical states (LDOS), which governs how strongly a quantum emitter couples to the plasmonic field at each point. NSOM is widely used for characterizing both propagating SPPs on films and localized plasmons on nanostructures.

Electron energy loss spectroscopy (EELS)

EELS measures the energy lost by fast electrons (in a transmission electron microscope) as they pass through or near a plasmonic nanostructure. The energy loss spectrum reveals the frequencies of the plasmon modes excited by the electron beam.

Because the electron beam can be focused to sub-nanometer spots, EELS provides spatial maps of plasmonic modes with unmatched resolution. It can detect both bright modes (which couple to light) and dark modes (which don't radiate and are invisible to optical techniques), giving a more complete picture of the plasmonic mode spectrum than purely optical methods.