Plasmon Effect in Silicon Detectors

• The plasmon effect in silicon detectors is the collective excitation of valence electrons, generating sharp resonant modes that enhance optical response and detection sensitivity.
• Device-level implementations employ both bulk and surface plasmon modes, with resonances around 15–20 eV that improve signal-to-noise ratios in dark matter and sub-bandgap photodetection.
• Engineered structures like MSM waveguides and nanoparticle integrations optimize plasmonic field confinement, yielding up to 100-fold event rate enhancement and 20–25% gains in quantum efficiency.

The plasmon effect in silicon detectors refers to the collective excitation of valence electrons in the silicon lattice, manifesting as sharp resonant modes (plasmons) in the energy-loss function. These modes result from the longitudinal oscillation of the electron gas and play a central role both in the optical response of silicon and in its sensitivity to various weakly coupled particles, including photons, dark matter, and axion-like particles. Plasmonic phenomena are exploited to enhance signal strengths in direct detection physics, sub-bandgap photodetection, and photonic applications. Critical aspects include the theoretical description of bulk plasmons, conditions for resonant excitation, device-level plasmonic field confinement, and the translation of plasmon resonances into measurable electronic signals.

1. Bulk Plasmon Modes in Silicon: Dielectric Response and Energy Loss

The collective modes in silicon are described by the longitudinal dielectric function, most accurately provided by the Lindhard formula in the random-phase approximation (RPA):

$$\epsilon(\omega, q) = 1 + \frac{4\pi e^2}{q^2}\chi_0(\omega, q)$$

where $\chi_0$ is the density–density response of noninteracting electrons. In the long-wavelength limit ($q\to0$), the bulk plasma oscillation frequency becomes:

$\omega_{\rm pl}^2 = \frac{4\pi n_e e^2}{m_e^*}$

with $n_e$ the valence-electron density ($n_e \simeq 6\times10^{23}~\text{cm}^{-3}$ for silicon), and $m_e^*\approx 0.26\,m_e$ is the effective mass. At finite $q$, the plasmon dispersion is:

$$\omega_{\rm pl}(q) \simeq \omega_{\rm pl}\left[1+\frac{3}{10}\left(\frac{q v_F}{\omega_{\rm pl}}\right)^2\right]$$

where $v_F$ is the Fermi velocity. The sharpness and position of the plasmon resonance are further refined by Drude or Mermin extensions and density-functional (DFT+RPA) band-structure corrections, shifting the physical peak to $15$–$17$ eV for momentum transfers $q<3$ keV (Liang et al., 2024, Essig et al., 2024). The energy-loss function (ELF):

$${\rm Im}\left[-\epsilon^{-1} (q, \omega)\right] = \frac{\rm Im\,\epsilon(q,\omega)}{|\epsilon(q,\omega)|^2}$$

exhibits a pronounced peak at these energies, signaling efficient energy coupling between probe particles and collective modes.

2. Plasmon Excitation by External Probes: Kinematic Thresholds and Resonance

Resonant excitation occurs when a probe particle (photon, electron, dark matter particle, etc.) can kinematically transfer energy and momentum to match the plasmon dispersion. For particle scattering (e.g., dark matter, millicharged particles, ALPs), energy–momentum conservation leads to a minimal velocity threshold for plasmon excitation. For dark matter with mass $m_\chi$ and velocity $v_\chi$, the minimal velocity is:

$$v_\text{min}(q) = \frac{\omega_{\rm pl}(q) + q^2/(2m_\chi)}{q}$$

For relativistic probes, the excitation condition generalizes to

$$\omega = \sqrt{p_\chi^2 + m_\chi^2} - \sqrt{(p_\chi - Q)^2 + m_\chi^2}$$

with $\omega$ matched to $\omega_{\rm pl}$ on resonance (Liang et al., 2024, Essig et al., 2024). The ELF sharply amplifies the cross section near $\omega \simeq \omega_{\rm pl}(q)$, producing a resonant enhancement in the detector’s energy spectrum for events depositing $\sim$16–20 eV.

3. Device-Level Plasmonic Effects: Surface Plasmons and Internal Photoemission

In engineered silicon devices, plasmonic enhancement exploits surface plasmon-polariton (SPP) and localized surface plasmon resonance (LSPR) modes at metal-silicon interfaces or metallic nanoparticles atop silicon. A prototypical structure is the metal/semiconductor/metal (MSM) waveguide, supporting SPPs tightly confined at subwavelength scales:

$$k_{\rm sp} = \frac{\omega}{c}\sqrt{\frac{\epsilon_m\,\epsilon_d}{\epsilon_m + \epsilon_d}}$$

Here, $\epsilon_m$ (complex permittivity of metal) and $\epsilon_d$ (of silicon) determine the propagation constant and evanescent decay. SPPs enable high local field intensities at the interface, leading to efficient hot-carrier generation, which for photon energies below the bandgap ($h\nu<E_g$) is detected via the internal photoemission (IPE) process. IPE efficiency is governed by:

$$\eta_i = \frac{1}{2}\left(1 - \sqrt{\frac{\Phi_B}{h\nu}}\right)^2$$

where $\Phi_B$ is the Schottky barrier at the metal–semiconductor junction (Goykhman et al., 2014, Pashaki et al., 2018, Pashaki et al., 2019). Field enhancement, quantum efficiency, and responsivity are direct functions of device geometry (gap width, metal choice, plasmonic coupling efficiency) (Pashaki et al., 2018, Pashaki et al., 2019).

4. Impact on Detector Sensitivity and Application to New Physics Searches

The plasmon effect enables silicon detectors to reach and surpass sensitivity thresholds for low-mass and weakly coupled particles. For relativistic dark-matter or axion-like particle scattering, the ELF-induced plasmon resonance shifts the expected recoil energy spectrum toward higher electron-hole (e–h) pair multiplicities (peaking at 4–6 e–h pairs):

• For sub-MeV dark matter, the inclusion of plasmon modes produces a pronounced bump near 15–20 eV, offering $\sim$100-fold rate enhancement over heavy-mediator cases, as shown in SENSEI data from SNOLAB and in reactor ALP searches (CONNIE, Atucha-II) (Liang et al., 2024, Essig et al., 2024, Gong et al., 12 Jan 2026).
• Backgrounds (dark counts, radiogenic events) peak at lower Q ($\lesssim2$ e–h pairs), so plasmon-induced energy thresholds improve signal-to-background discrimination.
• In dark matter nuclear recoils (via inelastic plasmon bremsstrahlung), event rates for the 16 eV plasmon (with $\sim$4–5 e–h pairs) may be four to five orders of magnitude higher than for ordinary photon bremsstrahlung (Kozaczuk et al., 2020).

These features are unmatched by optical (PAI) or free-electron models, which miss the plasmon resonance, instead predicting event rate maxima at lower (3–5 eV) energies (Essig et al., 2024).

5. Plasmonic Enhancement in Photovoltaic and Photodetector Applications

Surface and guided-mode plasmons also underpin light-harvesting strategies in silicon solar cells and detectors. Metallic nanoparticles atop silicon (e.g., Ag nanospheres over SiO$_2$/Si) generate strong near-field (LSPR) and far-field (scattering) enhancements. Optimization of particle size (D) and dielectric spacer thickness (t) produces IQE gains of $\sim$20–25% over bare silicon at specific $D$–$t$ combinations, supporting broadband light trapping and increased photocurrent (Rui et al., 2011). Antireflection functionality and spectral shaping can be engineered by tuning these parameters. In guided-mode plasmonic Schottky detectors, field-confinement and interface roughness, film thickness, and band engineering (e.g., SiGe core) further modulate internal quantum efficiency and bandwidth (Goykhman et al., 2014, Pashaki et al., 2019).

Device/Method	Resonant Energy (eV)	Enhancement Mechanism
Bulk plasmon excitation	15–20	Collective electron mode
SPP-based MSM detectors	0.3–0.8 (via $\Phi_B$)	Hot-carrier generation
NP/Si surface (LSPR)	Tunable (400–1100)	Near/far-field coupling
ALP–plasmon resonance	15–50	ELF resonance, reactor ALPs

6. Experimental Design: Thresholds, Readout, and Performance Metrics

State-of-the-art plasmon-sensitive silicon detectors, such as low-noise skipper-CCDs, achieve single-e–h pair sensitivity, $\sim$1 eV resolution, and sub-electron background rates. Design targets include:

• Energy threshold $\lesssim$ 15 eV (4 e–h pairs) to match the plasmon resonance (Liang et al., 2024, Essig et al., 2024).
• Momentum acceptance $q\lesssim3$ keV, set by pixel size and band-structure corrections.
• Multi-arm or differential layouts (e.g., balanced MSM) to minimize dark current and noise (Pashaki et al., 2018).
• Responsivity enhancements (up to $0.89~\text{A/W}$ at 1550 nm for SiGe-core MSM devices) and GHz-scale electrical bandwidth (Pashaki et al., 2019).
• Surface engineering for field-confinement optimization and antireflection.

Such optimizations enable enhanced reach for new physics searches and high-performance photonic and photovoltaic applications.

7. Prospects for Future Development and Fundamental Limitations

Future improvements in plasmon-based silicon detectors include:

• Wider spectral coverage via distributed NP sizes/spacers or multilayer dielectrics (Rui et al., 2011).
• Tailored core composition (e.g., high Ge fraction) for mode confinement, balanced against dark current (Pashaki et al., 2019).
• Use of metals or alloys with optimized plasma frequency, and interface roughening for k-parallel momentum relaxation (Goykhman et al., 2014).
• Integration of resonant or slow-light photonic components for extended interaction lengths and higher hotspots (Goykhman et al., 2014, Pashaki et al., 2018).
• Scaling to kg$\cdot$yr exposures for dark sector searches (e.g., Oscura’s projected reach to $g_{a\gamma\gamma}\sim10^{-7}~\text{GeV}^{-1}$) (Gong et al., 12 Jan 2026).

A plausible implication is the eventual convergence of plasmonic enhancement strategies across photon-sensing and rare-event detection, realizing detectors with both extreme spectral shape control and ultra-low noise/threshold characteristics.

---

References:

(Rui et al., 2011, Goykhman et al., 2014, Pashaki et al., 2018, Pashaki et al., 2019, Kozaczuk et al., 2020, Liang et al., 2024, Essig et al., 2024, Gong et al., 12 Jan 2026)