Nonlocal Optical Response and Surface Susceptibilities: A Systematic Derivation via Spatial Moment Expansion

This paper presents a systematic theory that derives effective surface susceptibilities for arbitrary curved interfaces by applying a spatial moment expansion to nonlocal optical response kernels, thereby generalizing Feibelman's $d$-parameters to include curvature corrections and establishing generalized Maxwell boundary conditions.

The Gist

The Big Picture: Why Light Gets "Confused" at the Edge

Imagine you are walking through a crowded room. If you are in the middle of the crowd, you can move freely; the people around you act like a uniform fluid. This is how light usually behaves inside a material like glass or metal. Physicists call this the "bulk" behavior.

But what happens when you reach the edge of the room, near the wall? Suddenly, the rules change. You can't walk through the wall, and the people near the wall behave differently than those in the middle. In the world of light and nanotechnology, the "wall" is the surface of a tiny object (like a nanoparticle), and the "people" are the electrons.

For a long time, scientists treated the surface of these materials as a sharp, perfect line where the rules just suddenly stop. But in reality, the transition isn't a sharp line; it's a fuzzy, blurry zone a few atoms thick. This paper is about figuring out exactly how that fuzzy zone changes the way light bounces off or passes through a surface.

The Problem: The "Blur" is Too Hard to Calculate

The paper starts by saying that describing this fuzzy zone is incredibly difficult.

• The Atomic Scale: At the very bottom, electrons are quantum particles. They don't just react to the light hitting them directly; they react to the light hitting their neighbors. It's like a crowd where everyone is holding hands; if you push one person, the whole line wiggles.
• The Optical Scale: We usually look at light with wavelengths much larger than atoms (like micrometers).
• The Gap: Trying to calculate the behavior of every single atom for every single photon is impossible for complex shapes like spheres or cylinders.

Scientists have used "Feibelman d-parameters" (a fancy name for two numbers) to describe flat surfaces. But this paper asks: What if the surface is curved? What if it's a sphere, a cylinder, or a weirdly shaped egg? The old rules didn't work well there.

The Solution: The "Moment Expansion" (The Crowd Analogy)

The author, F. Zolla, proposes a new way to solve this. Instead of trying to track every atom, they use a mathematical trick called a "Spatial Moment Expansion."

Think of it like this:
Imagine you are trying to describe the "weight" of a crowd of people standing in a hallway.

1. The Old Way: You try to weigh every single person individually. (Too hard!)
2. The New Way: You realize that for most of the hallway, the crowd is uniform. You only need to worry about the people right near the door.
3. The Trick: You take a "snapshot" of the crowd near the door and calculate a few key numbers (moments) that summarize their behavior.
   • Moment 1: How heavy is the crowd right at the door? (This gives us the Surface Susceptibility, $\chi_s$).
   • Moment 2: How does the weight change as you move away from the door? (This gives us the Curvature Corrections).

By doing this, the author shows that all the messy, complex quantum physics of the surface can be squashed down into just a few simple numbers.

The Key Discovery: Curvature Matters

The most exciting part of the paper is what happens when the surface is curved.

• Flat Surface (The Wall): The light sees a simple "fuzzy layer." The math is straightforward.
• Curved Surface (The Ball or Cylinder): The paper proves that the shape of the object changes the light's behavior.
  • If you have a tiny sphere, the "fuzzy layer" is squeezed differently than on a flat wall.
  • The author derives new formulas that include Curvature (how round the object is) and Gaussian Curvature (how the shape bends in two directions, like a saddle).

The Analogy: Imagine painting a wall.

• On a flat wall, you need a standard amount of paint.
• On a ball, the paint has to stretch over a curve. The paper calculates exactly how much extra "paint" (or how the light reacts) is needed because of that curve.

The Results: A New Set of Rules

The paper provides a "Universal Toolkit" for scientists.

1. One Number is Enough (Mostly): For many common materials (like metals), the light reacts the same way whether it hits the surface head-on or from the side. This simplifies things massively.
2. The "Curvature Correction": The paper gives a specific formula to fix the math for curved objects. It says: "Take the flat result, and add a small correction based on how round the object is."
3. Real-World Check: The author tested this on spheres, cylinders, and ellipsoids (egg shapes) using different types of "fuzzy layers" (mathematical models). In every case, the math worked perfectly and matched what we expect for flat surfaces when the object gets huge.

Why This Matters (According to the Paper)

The paper doesn't talk about curing diseases or building new phones. It focuses on precision physics.

• Nanoparticles: When objects get very small (like 10 nanometers), the "fuzzy layer" takes up a huge chunk of the object. The old rules (which ignore the fuzziness) start to fail. This paper gives the correct rules for these tiny objects.
• Mie Resonances: This is a fancy term for the specific colors that tiny balls of metal reflect. The paper shows that because of the surface fuzziness and the curve, these colors shift slightly. If you want to design a sensor that detects a virus using light, you need to know exactly where that color shift happens. This paper tells you how to calculate it.

Summary in One Sentence

This paper provides a systematic, mathematical recipe to turn the messy, complex physics of how light interacts with the fuzzy edge of a material into a few simple numbers that work for flat walls, round balls, and everything in between.

Technical Summary

Technical Summary: Nonlocal Optical Response and Surface Susceptibilities

Problem Statement
The interaction between light and matter involves multiple spatial scales: the atomic scale (Ångströms), the optical wavelength (micrometers), and the macroscopic object size. While macroscopic optics relies on local constitutive relations where polarization depends only on the local electric field, this approximation fails when relevant length scales approach the microscopic characteristic lengths of materials (e.g., in metals and semiconductors). In these regimes, spatial non-locality becomes significant: the polarization at a point depends on the electric field in a spatial neighborhood.

Despite extensive work on nonlocal effects in bulk media (e.g., excitonic polaritons, plasmonic nanostructures), a systematic, constructive link between the bulk nonlocal response kernel and the effective surface response for arbitrarily curved interfaces has remained elusive. Existing approaches often rely on phenomenological models (like the nonlocal hydrodynamic model) or specific microscopic calculations without providing a general derivation of how bulk kernels translate into boundary conditions for curved geometries. Specifically, the influence of interface curvature on nonlocal effects (Mie corrections) has not been rigorously derived from the bulk kernel alone.

Methodology
The paper presents a systematic derivation of linear surface susceptibilities from a general tensorial nonlocal response kernel. The methodology relies on two primary mathematical tools:

1. Spatial Moment Expansion: The nonlocal constitutive relation is expanded by performing a Taylor series of the electric field around an observation point. This decomposes the polarization into a bulk contribution (infinite homogeneous medium) and a boundary contribution. The coefficients of this expansion are spatial moments of the kernel.
2. Distributional Thin-Layer Limit: The boundary contribution, which is concentrated in a layer of thickness $\ell$ (the nonlocality range) near the interface, is treated as a distribution. By taking the limit where the layer thickness $\ell \to 0$ while keeping the integrated moments finite, the boundary terms are converted into singular distributions (Dirac deltas and their derivatives) supported on the interface $\partial\Omega$.

The derivation assumes a bulk-isotropic and centrosymmetric kernel (invariant under parity $R \to -R$), a condition satisfied by standard plasmonic metals and dielectrics. The geometry is treated generally using normal coordinates, incorporating the mean curvature ($H$) and Gaussian curvature ($K$) of the interface.

Key Contributions and Results

• Constructive Generalization of Feibelman Parameters: The authors derive a single scalar surface susceptibility, $\chi_s$, which serves as a constructive generalization of the Feibelman $d$-parameters. Unlike phenomenological parameters, $\chi_s$ is explicitly calculated from the zeroth-order radial moments of the bulk kernel. For centrosymmetric isotropic kernels, the tangential and normal surface susceptibilities are shown to be equal ($\chi_{s\parallel} = \chi_{s\perp} = \chi_s$).
• Hierarchy of Surface Terms: The distributional expansion reveals a hierarchy of surface effects:
  • Leading Order ($\ell^0$): A surface susceptibility term proportional to $\delta_{\partial\Omega}$.
  • First Order ($\ell^1$): Terms involving the normal derivative of the delta function and mean curvature vanish identically for isotropic centrosymmetric kernels due to parity arguments.
  • Second Order ($\ell^2$): Curvature corrections proportional to the mean curvature $H$ and Gaussian curvature $K$ appear. These are governed by the second-order moments of the kernel.
• Generalized Maxwell Boundary Conditions: The paper establishes explicit boundary conditions relating field jumps to the derived surface susceptibilities and curvature terms. These conditions modify the classical Fresnel relations, introducing terms dependent on surface gradients and curvature.
• Analytical Verification: The formalism is applied to four canonical geometries (planar, spherical, cylindrical, and ellipsoidal) and four kernel models (Gaussian, Yukawa, tensorial Lorentz, and local). In all cases, the calculations are fully explicit, recovering the local (Fresnel) limit as a special case.
• Numerical Examples:
  • Planar Interfaces: The modified Fresnel coefficients show that while reflectance intensity is affected only at second order in $\ell/\lambda$, the reflection phase is shifted at first order, making it detectable via ellipsometry.
  • Curved Interfaces (Mie Scattering): For cylindrical and spherical geometries, the nonlocal corrections shift Mie resonance frequencies and modify scattering efficiencies. The corrections scale as $\ell/R_0$ (where $R_0$ is the radius), becoming significant for nanoparticles where $R_0 \lesssim 10\ell$. The paper demonstrates that the tangential susceptibility ($\chi_{s\parallel}$) primarily affects resonance positions, while the normal susceptibility ($\chi_{s\perp}$) acts as a broadband renormalization of the exterior permittivity.

Significance and Claims
The paper claims to fill a critical gap in theoretical optics by providing a rigorous, parameter-free link between microscopic bulk nonlocality and macroscopic surface boundary conditions. Its primary significance lies in:

1. Universality: The derived hierarchy of surface terms depends only on the moments of the bulk kernel and the geometric invariants of the surface, making the framework applicable to any regular surface geometry and isotropic kernel.
2. Experimental Accessibility: The effective parameters ($\chi_s$, $\nu_\parallel$, $\nu_\perp$) are derived quantities that can, in principle, be extracted experimentally (e.g., via ellipsometry or dark-field spectroscopy) without requiring knowledge of the underlying microscopic model.
3. Curvature Effects: The work explicitly quantifies how interface curvature modifies nonlocal responses, providing a theoretical basis for understanding size-dependent optical properties in nanoparticles (nonlocal Mie corrections).
4. Reactive Nature: The analysis confirms that for real, even kernels, spatial non-locality acts as a purely reactive phenomenon (shifting phases) rather than a dissipative one, with energy conservation strictly maintained. Dissipation only arises if the kernel is frequency-dependent (absorbing) or non-centrosymmetric.

The authors note that this framework is restricted to linear, centrosymmetric media and that extensions to chiral media and nonlinear responses are treated in companion papers. The formalism is presented as a systematic alternative to ad-hoc "dead layer" models, offering a first-principles derivation of surface susceptibilities directly from bulk properties.