betaFeedback

WorksPhysics and Astronomy

Boundary homogenization for target search problems

Abstract

In this review, we describe several approximations in the theory of Laplacian transport near complex or heterogeneously reactive boundaries. This phenomenon, governed by the Laplace operator, is ubiquitous in fields as diverse as chemical physics, hydrodynamics, electrochemistry, heat transfer, wave propagation, self-organization, biophysics, and target search. We overview the mathematical basis and various applications of the effective medium approximation and the related boundary homogenization when a complex heterogeneous boundary is replaced by an effective much simpler boundary. We also discuss the constant-flux approximation, the Fick-Jacobs equation, and other mathematical tools for studying the statistics of first-passage times to a target. Numerous examples and illustrations are provided to highlight the advantages and limitations of these approaches.

Introduction

The diffusion of Brownian particles near complex boundaries is ubiquitous for many natural phenomena and engineering applications. Common examples are heterogeneous catalysis, diffusion-limited aggregation, tracer dispersion over complex canopies in atmosphere and ocean, cell communication via chemical signals, gas exchange in the human lungs and placentas, and many others [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18]. The mathematics of this process remains unchanged even if the diffusive quantity is continuous (e.g., temperature profile, velocity field, wave amplitude, magnetic field, vorticity) and thus not directly related to any physical particle or Lagrangian marker. In this regard, diffusion is connected to the broad family of phenomena, which are collectively referred to as Laplacian transport. The unified analytical framework of the underlying Laplace equation, which also includes the diffusion equation at the steady state, is thus applicable in fields as diverse as chemical physics, acoustics, electrostatics, fluid dynamics, electrochemistry, wetting, etc. [19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30].

The analytical results for diffusive transport near inhomogeneous boundaries are readily available only for simple shapes (plane, sphere, cylinder, etc.). More complex boundary profiles require the application of advanced analytical methods (e.g., conformal mapping, eigenfunction decomposition, perturbation theory, scaling arguments, etc.) and often lead to extensive numerical simulations via Monte Carlo, finite element, or Lagrangian methods. The aim of this review is to show how the conventional framework of the effective medium approximation and boundary homogenization can be applied to target search problems to enable analytical progress even for boundaries of rather complex profiles. We show that this approach gives access not only to some “averaged” properties of the system (e.g., the total flux) but also to the first-passage time (FPT) statistics. The latter plays the central role in target search problems [31, 32, 33, 34] as the proxy of efficiency of search strategies [35, 36, 37].

Generally speaking, the complexity of the boundary in the context of diffusive transport can be attributed to two independent factors: (i) the geometrical complexity of the boundary profile (spiky, fractal, patchy, etc), and (ii) the distribution of the absorbing properties on that boundary. For instance, some parts of the boundary can be absorbing and the other be reflecting. This pattern can be rather complex even if the geometrical profile of the boundary is simple (e.g., flat boundary). The present review deals with both kinds of boundary complexity. Instead of solving the target search problem for a given complex boundary, boundary homogenization aims at replacing such a boundary by an equivalent simpler one, for which the target search problem admits an exact simple solution. The review is thus focused on the fundamental question of constructing such an equivalent boundary, and provides theoretical answers and practical ready-to-use recipes.

We begin with an illustrative example. Consider particles diffusing in a layer 0<y<H0 < y < H between two parallel planes separated by distance HH. At steady state, the particle concentration CC within the layer obeys the Laplace equation

ΔC=0,(1)\Delta C = 0, \tag*{(1)}

where Δ=∂2/∂x2+∂2/∂y2+∂2/∂z2\Delta= \partial^2/\partial x^2 + \partial^2/\partial y^2 + \partial^2/\partial z^2 is the Laplace operator. Assume that the bottom plane y=0y = 0 is absorbing (i.e., any particle hitting this plane disappears so that C=0C = 0 at y=0y = 0), and the concentration of the particles at the top plane y=Hy = H is kept constant: C=C0C = C_0 (i.e., it is a permanent source of particles). One can also think of CC as a temperature profile inside a layer, whose two boundaries are kept at constant temperatures 00 and C0C_0. The solution of this trivial problem yields the linear profile of particle concentration

C=C0Hy=j0Dy,(2)C = \frac{C_0}{H}y = \frac{j_0}{D}y, \tag*{(2)}

where j0=DC0/Hj_0 = DC_0/H is the particle flux density, and DD is the particle diffusivity.

Next we assume that the bottom boundary is subject to a periodic deformation with the period W≪HW \ll H (Figure 1a). Near the deformed boundary, the particle concentration is strongly non-uniform, but for y≫Wy \gg W the transverse components of the particle flux rapidly disappear and the concentration profile takes the linear form (Figure 1b):

Schematic view of Laplacian transport and numerical solutions

Figure 1. (a) A schematic view of Laplacian transport from the top (flat) source at y=Hy = H (in red) towards an absorbing boundary with sinusoidal profile (in blue), with the period (or wavelength) WW and amplitude h0h_0. Dashed vertical lines delimit the periodic cell that is repeated along xx coordinate. The system is translationally invariant along zz coordinate, which is perpendicular to the figure. (b) A numerical solution of the Laplace equation inside this domain, with reflecting boundary conditions instead of periodic ones on two vertical segments. This numerical solution (obtained by a finite-element method via PDEtool in Matlab) determines the diffusive flux JJ onto the absorbing boundary, which in turn sets via Eq. (4) the offset λ\lambda (e.g., we get λ≈0.26\lambda\approx0.26 for H=5H = 5, W=2W = 2, and h0=0.5h_0 = 0.5). (c) The solution C=C0(y−λ)/(H−λ)C = C_0 (y - \lambda)/(H - \lambda) in a simpler case with a flat absorbing boundary at y0=λy_0 = \lambda, which yields the same diffusive flux.

C≈C0H−λ(y−λ),(3)C \approx\frac{C_0}{H-\lambda}(y-\lambda), \tag*{(3)}

where the offset parameter λ\lambda has the dimension of length and can be positive or negative (note that another convention with −λ-\lambda instead of λ\lambda in Eq. (3) is often used). This parameter aggregates all information about the boundary profile and inhomogeneity of its absorbing properties that is relevant at far field y≫Wy \gg W. For instance, in terms of λ\lambda, the diffusive flux through the system is given by a simple expression

JJ0=11−λ/H,(4)\frac{J}{J_0} = \frac{1}{1-\lambda/H}, \tag*{(4)}

where J0=j0A0=DC0A0/HJ_0 = j_0\mathcal{A}_0 = DC_0\mathcal{A}_0/H is the diffusive flux to a flat absorbing boundary, and A0\mathcal{A}_0 is the cross-sectional area of the periodic cell. At the same time, this relation allows one to determine the offset λ\lambda from the diffusive flux JJ, which can be found numerically by solving the Laplace equation or via Monte Carlo simulations:

λ=H(1−J0/J).(5)\lambda= H(1-J_0/J). \tag*{(5)}

Similarly, the far-field solution (3) remains valid for a flat bottom boundary with absorbing and reflecting parts (see examples below). If the bottom boundary is fully reflecting, the concentration CC is constant, and the diffusive flux JJ is zero. This situation corresponds to the limit λ=−∞\lambda=-\infty. Adding more and more absorbing spots onto the bottom boundary increases the diffusive flux and thus the offset λ\lambda, which thus ranges from −∞-\infty (flat reflecting boundary) to 0 (flat absorbing boundary). Moreover, if the bottom boundary is both deformed and heterogeneous, the far-field solution may correspond to positive λ\lambda.

The linear form (3) provides a simple interpretation for the offset λ\lambda: the condition y=λy=\lambda determines the position of the equivalent absorbing boundary that provides the same diffusive flux through the system. The goal of boundary homogenization is to relate λ\lambda to the geometrical properties of the original complex boundary. As this problem is relevant for various disciplines, the aggregated parameter λ\lambda bears different names: slip length in the context of superhydrophobic coating [25], blockage coefficient in fluid dynamics [25, 26], grid parameter in electrostatics [27], displacement length in heat transfer [29, 30]. The mathematical equivalence between steady-state diffusion and electrostatics, both described by the Laplace equation, is particularly helpful for bringing intuitive interpretations [19, 38]. In fact, if C0C_0 is understood as the applied electric potential (voltage) between two metal electrodes located at the top and at the bottom, then JJ becomes the electric current, while Z=C0/JZ=C_0/J is the electric resistance (or impedance) of this system. For instance, if the bottom electrode is flat, one retrieves the classical formula Z0=C0/J0=H/(DA0)Z_0=C_0/J_0=H/(D\mathcal{A}_0) for the resistance of a metal wire of length HH and cross-sectional area A0\mathcal{A}_0, while 1/D1/D plays the role of electric resistivity. As a consequence, Eq. (4) provides another interpretation of the offset λ\lambda as the relative difference between the electric resistances ZZ and Z0Z_0:

Z0−ZZ0=λH.(6)\frac{Z_0-Z}{Z_0}=\frac{\lambda}{H}. \tag*{(6)}

Moreover, common rules for computing the overall resistance of an electric circuit, composed of sequential and/or parallel connections between its elements, can shed light onto various expressions for the offset λ\lambda that will be discussed below.

In chemical physics, it is more conventional to relate the offset λ\lambda to the trapping parameter κ\kappa that appears in the Robin (or radiation) boundary condition on a partially absorbing (reactive) flat boundary located at some height y0y_0:

−D∂C∂n=κC,(7)-D\frac{\partial C}{\partial n}=\kappa C, \tag*{(7)}

where ∂/∂n\partial/\partial n is the normal derivative oriented outwards the domain [39]. The trapping parameter κ\kappa, which has units of meter per second, is also known as surface reactivity, forward reaction constant, permeability, relaxivity, or the inverse of surface resistivity, depending on the application field. It can be checked by direct substitution of Eq. (3) into Eq. (7) that these approaches are equivalent and

κ=Dy0−λ.(8)\kappa=\frac{D}{y_0-\lambda}. \tag*{(8)}

While the offset λ\lambda is uniquely determined by the shape of the bottom boundary and the inhomogeneity of its absorbing parts, the value of the trapping parameter κ\kappa depends on the choice of the location y0y_0 of the effective flat boundary. This ambiguity actually offers some flexibility for boundary homogenization. From the mathematical point of view, the trapping parameter κ\kappa should be restricted to nonnegative values, with two relevant limits, κ→0\kappa\to0 and κ→∞\kappa\to\infty, corresponding to the perfectly reflecting and perfectly absorbing boundary, respectively. When λ<0\lambda< 0, one can naturally choose y0=0y_0 = 0. However, if λ>0\lambda> 0, it is mandatory to choose y0≥λy_0 \ge\lambda (note that the choice y0=λy_0 = \lambda corresponds to an absorbing boundary with κ=∞\kappa= \infty, as discussed above).

For simplicity of notations and more compact formulas, we will employ both parameters λ\lambda and κ\kappa interchangeably, keeping in mind that they are related via Eq. (8). One of the aims of this review is to transfer the analytical results from the different areas of Laplacian transport to the context of particle diffusion and target search problems. Evaluation of κ\kappa for a given boundary profile is a difficult task that requires solving the Laplace equation in a complex shape domain. Below we present some analytical results for different types of complex boundaries. In § 0.0.2, we consider boundary profiles and inhomogeneities that are periodic in xx coordinate and translationally invariant along zz coordinate, e.g., periodic stripes or riblets. The invariance along zz coordinate reduces the original 3D problem to a planar one. In turn, § 0.0.5 focuses on absorbing spots or deformations, which are periodic in two directions along the surface. § 0.0.11 presents a constant-flux approximation, which may offer more accurate results, even though the derived expressions are usually more sophisticated. In § 0.0.12, we discuss the Fick-Jacobs equation to deal with Laplacian transport in channels of variable cross-section and its application for computing the mean first-passage time (MFPT). § 0.0.15 concludes this overview by summarizing the main steps and discussing other aspects of Laplacian transport.

Periodic patterns in two dimensions

In this section, we focus on various arrangements of infinitely long stripes or deformations so that the system is invariant along one direction (we call it zz), and the original three-dimensional problem is thus reduced to a two-dimensional problem in the plane xyxy. Several profiles of stripes and deformations are illustrated in Figure 2. For such profiles, there is a wealth of analytical results for λ\lambda and κ\kappa due to the conformal invariance of the Laplace equation in two dimensions that enables application of conformal mapping. The general framework for evaluation λ\lambda (and hence κ\kappa) is given in Refs. [25, 27, 30, 40, 41, 42, 43] and references therein.

Various configurations of absorbing stripes or deformations

Figure 2. Various configurations of absorbing stripes or deformations that are infinite along zz direction, which is perpendicular to the figure. (a) A flat boundary with a periodic arrangement of absorbing (blue solid) and reflecting (green dashed) stripes, of period WW and absorbing fraction σ\sigma. (b) A flat periodic boundary with a pattern of two equal absorbing stripes (blue solid) and two unequal reflecting stripes (green dashed), with the gap σgW\sigma_g W between two absorbing stripes. (c, d) A periodic boundary with flat absorbing stripes (blue solid) and non-flat reflecting parts (green dashed) modeled by circular arcs, with the angle θ\theta at the joint point between two stripes: (c) upward orientation for θ>0\theta> 0; (d) downward orientation for θ<0\theta< 0. (e) An absorbing comb-like boundary. (f) An absorbing boundary with rectangular riblets of width ww and height h0h_0. (g) An absorbing boundary of saw-tooth riblets of width WW and height h0h_0. (h) An absorbing boundary of riblets with trapezoidal grooves. (i) An absorbing comb-like boundary with reflecting base (green dashed).

Flat boundaries

For the flat boundary with a periodic alternating pattern of absorbing and reflecting stripes (Figure 2a), one has [44] (see also [45])1

λ=Wln⁡[sin⁡(πσ/2)]π<0,κ=DWπln⁡[1/sin⁡(πσ/2)],(9)\lambda= W \frac{\ln[\sin(\pi\sigma/2)]}{\pi} < 0,\qquad\kappa= \frac{D}{W}\frac{\pi}{\ln[1/\sin(\pi\sigma/2)]}, \tag*{(9)}

where WW is the period of the pattern, σ\sigma is the fraction of absorbing stripes, and we set y0=0y_0 = 0 to introduce κ\kappa. As σ→0\sigma\to0, the trapping parameter κ\kappa vanishes logarithmically, κ∝1/ln⁡(1/σ)→0\kappa\propto1/\ln(1/\sigma) \to0, so the heterogeneously absorbing boundary becomes perfectly reflecting. In the opposite limit σ→1\sigma\to1, κ\kappa diverges as a power-law, κ∝1/(1−σ)2→∞\kappa\propto1/(1-\sigma)^2 \to\infty, and the boundary becomes perfectly absorbing.

It is worth noting that the same result is applicable for pressure-driven incompressible Stokes flow in a cylindrical pipe with alternating no-slip and no-stress stripes [46]: when the stripes are parallel to the cylinder axis (Figure 3a), Eq. (9) determines the trapping parameter κ\kappa of the homogenized cylinder; in turn, if the stripes are perpendicular to the cylinder axis (Figure 3b), Eq. (9) should be divided by factor 2, see details in [46]. In the context of target search problems, the trapping parameter κ\kappa can be used to estimate the MFPT to the absorbing stripes on the cylinder from the explicitly known solution for a homogeneous partially absorbing cylinder:

Cylindrical domains

Figure 3. Cylindrical domains. (a) An infinite reflecting cylinder with absorbing stripes (in blue) oriented parallel to the cylinder axis. (b) An infinite reflecting cylinder with absorbing stripes (in blue) oriented perpendicular to the cylinder axis. (c) A long reflecting cylinder with an absorbing disk (in blue) centered on the cylinder axis. (d) A capped cylinder of height HH and radius RR, with the source located at the cylinder wall and an absorbing disk (in blue), located at the center of the bottom reflecting wall. (e) Two capped coaxial reflecting cylinders of height HH and radii ρ\rho and RR, with an absorbing stripes (in blue) located on the inner cylinder. (f, g) Cross-section of an infinite absorbing cylinder with triangular riblets (the cylinder axis is perpendicular to the figure). (h) Cross-section of an infinite reflecting cylinder with absorbing spikes. (i) An appropriate conformal map ϕ\phi can transform an arbitrary simply-connected cross-section of an infinite cylinder into an equivalent circular cross-section with a conformal radius rcr_c.

T(r)=R2−r24D+R2κ,(10)T(r)=\frac{R^{2}-r^{2}}{4D}+\frac{R}{2\kappa}, \tag*{(10)}

where rr is the radial coordinate of the starting point. In the case of heterogeneously absorbing cylinder, this is an approximate solution, which is valid far from the boundary, R−r≫WR-r\gg W, where WW is the period. Moreover, if the starting point is uniformed distributed in the cylinder, the average of T(r)T(r) over the volume yields

T‾=1πR2∫02πdθ∫0Rdr r T(r)=R28D+R2κ.(11)\overline{T}=\frac{1}{\pi R^{2}}\int_{0}^{2\pi}d\theta\int_{0}^{R}dr\,r\,T(r)=\frac{R^{2}}{8D}+\frac{R}{2\kappa}. \tag*{(11)}

Exceptionally, the exact expression for the MFPT in the cylinder with absorbing parallel stripes is known [47, 48, 49], but it is rather complicated, especially for many stripes. In turn, Eqs. (10), (11), which we use to illustrate an application of boundary homogenization, provide a rapid and simple way to estimating the MFPT.

For the configurations of stripes with two and more absorbing stripes per period, κ\kappa cannot be described exclusively by the surface fraction σ\sigma since the relative position of the absorbing stripes is also important due to their competition for particles (the phenomenon known as diffusional screening or diffusive interaction). The general formula for κ\kappa is rather cumbersome [50], but for two equal absorbing stripes per period (not equally spaced), the result has a simple explicit form:

λ=Wln⁡(F)2π<0,F=sin⁡2[π(σ+σg)/2]−sin⁡2[(πσg)/2],(12)\lambda=W\frac{\ln(F)}{2\pi}<0,\qquad F=\sin^{2}[\pi(\sigma+\sigma_g)/2]-\sin^{2}[(\pi\sigma_g)/2], \tag*{(12)}

where σ\sigma is the surface fraction of the absorbing stripes, and σg\sigma_g is the surface fraction of the gap between them (Figure 2b). For σg=0\sigma_g=0 we return to the previous formula (9), while in the general case σg≠0\sigma_g\ne0, the parameter σg\sigma_g describes the diffusive interaction between the absorbing stripes. We note that the functional form of this interaction is nontrivial: even for π(σ+σg)/2≪1\pi(\sigma+\sigma_g)/2\ll1, one has λ≈Wln⁡(π22σ(σg+12σ))/π\lambda\approx W\ln(\frac{\pi^{2}}{2}\sigma(\sigma_g+\frac{1}{2}\sigma))/\pi, i.e. λ\lambda (and thus κ\kappa) is determined by the product σ(σg+12σ)\sigma(\sigma_g+\frac{1}{2}\sigma), which is a signature of strong coupling (two tiny stripes strongly interact even when they are separated by a large distance). Equations (9) and (12) have been extensively validated numerically with Brownian dynamics simulations, including homogenization of cylindrical surface with different orientations of stripes (longitudinal, transverse and spiral) [51, 52].

Non-flat boundaries

Next we consider some examples of non-flat boundaries. We begin with the profile, in which the reflecting parts of the boundary are non-flat. This boundary has a different value of the parameter κ\kappa because the curved reflecting parts affect the particle ability to reach the absorbing parts of the boundary. To appreciate this effect, we can use the results of Ref. [53], in which the reflecting parts were modeled as circular arcs (see Figure 2c,Figure 2d). In this case, the formula for κ\kappa includes a correction factor

κ(θ)κ(0)=3(π−θ)23π2−4πθ+2θ2(13)\frac{\kappa(\theta)}{\kappa(0)} = \frac{3(\pi-\theta)^2}{3\pi^2 - 4\pi\theta+ 2\theta^2} \tag*{(13)}

where θ\theta is the angle between reflecting and absorbing parts at their joint, and κ(0)\kappa(0) is given by Eq. (9), in which σ≪1\sigma\ll1 (dilute limit) was assumed. At θ=0\theta= 0, one retrieves the flat boundary, while negative angles correspond to the opposite (downwards) orientation of the reflecting arcs. When θ\theta varies from −π/2-\pi/2 (half-circle oriented downwards) to π/2\pi/2 (half-circle oriented upwards), the correction factor changes from 1.2273 to 0.5, i.e., the boundary becomes less absorbing due obstruction of the absorbing intervals by the reflecting arcs of the boundary. Curiously, the correction factor is larger than 1 for θ<0\theta< 0, i.e., the inclusion of reflecting arcs oriented downwards enhances the overall reactivity. Indeed, the particle can spend more time in these regions and thus have higher chances to arrive to the absorbing intervals, before returning to the source. Other examples and their correction factors can be found in [54].

An important analytical result for the offset λ\lambda was obtained for a sinusoidal absorbing boundary of a small amplitude h0h_0 (Figure 1a). In this case, the formula for λ\lambda can be derived by means of perturbation theory [40]:

λ≈πh02W>0,h0≪W.(14)\lambda\approx\frac{\pi h_0^2}{W} > 0,\quad h_0 \ll W. \tag*{(14)}

This formula states that the sinusoidal deformation shifts the equivalent boundary upwards, though this shift is small since λ/h0≈πh0/W≪1\lambda/h_0 \approx\pi h_0/W \ll1. The direction of this shift is also understandable: the bottom of the troughs of the deformed boundary is less accessible to the particles, so that most of the particle absorption occurs near the crests (i.g., above the middle line located at y=0y = 0), which play the role of “active zones” of absorption (see [17, 55] for this concept). As we can see below, for a small-amplitude deformation of the boundary, the scaling λ∝(h02/W)\lambda\propto(h_0^2/W) is rather general: it can be interpreted as the product of the amplitude h0h_0 and a typical slope ∝h0/W\propto h_0/W [40]. We stress that this result is perturbative; for instance, substitution of h0=0.5h_0 = 0.5 and W=2W = 2 from the example in Figure 1 into Eq. (14) yields λ≈0.39\lambda\approx0.39, whereas a numerical solution of the Laplace equation gave λ≈0.26\lambda\approx0.26. Vandembroucq and Roux proposed an improved expression

λ≈2h0πarctan⁡(π2h02W),(15)\lambda\approx\frac{2h_0}{\pi}\arctan\left(\frac{\pi^2 h_0}{2W}\right), \tag*{(15)}

which is reduced to Eq. (14) when π2h0/(2W)\pi^2 h_0/(2W) is small. For our example, this expression gives λ≈0.28\lambda\approx0.28, which is much closer to the above numerical value. Note that if the amplitude h0h_0 was reduced to 0.1, Eq. (14) would yield 0.016, which is already close to the numerical solution 0.013. Expectedly, the perturbative expression (14) becomes more accurate as h0/Wh_0/W gets smaller.

For the comb-like absorbing boundary (Figure 2e), there is the exact formula [41]

λ=Wln⁡[cosh⁡(πh0/W)]π>0,(16)\lambda= W\frac{\ln\left[\cosh\left(\pi h_0/W\right)\right]}{\pi} > 0, \tag*{(16)}

where h0h_0 is the height of the riblets. This offset is always positive, i.e., the effective absorbing boundary stands above the base of the comb. This is expected because the spikes efficiently absorb particles and thus increases the diffusive flux JJ, as compared to the flux J0J_0 onto the absorbing base without spikes. For πh0/W≪1\pi h_0/W \ll1, Eq. (16) leads to a similar scaling as in Eq. (14):

λ≈πh022W,h0≪W.(17)\lambda\approx\frac{\pi h_0^2}{2W},\quad h_0 \ll W. \tag*{(17)}

The difference by factor 2 between these expressions can be attributed to distinct shapes of the sinusoidal profile and the comb structure (note also that h0h_0 was the amplitude of the sine profile, not the height). In contrast, for high riblets h0/W≫1h_0/W \gg1, there is a saturation limit

λ≈h0−Wln⁡2π,h0≫W.(18)\lambda\approx h_0 - \frac{W\ln2}{\pi}, \qquad h_0 \gg W. \tag*{(18)}

This result implies that most of the particles are absorbed within the distance ln⁡2πW≈0.22W\frac{\ln2}{\pi}W \approx0.22 W from the tips of the comb-like structure making the rest of the structure irrelevant in this context. This result is the simplest way of illustrating the concept of “active zones” discussed above. As λ\lambda remains below h0h_0, it is natural to impose the effective partially absorbing boundary at y0=h0y_0=h_0 (at the tip of the combs), in which case Eq. (8) defines the trapping parameter as

κ=Dh0−Wln⁡[cosh⁡(πh0/W)]/π.(19)\kappa= \frac{D}{h_0-W\ln[\cosh(\pi h_0/W)]/\pi}. \tag*{(19)}

In two limits h0/W≪1h_0/W \ll1 and h0/W≫1h_0/W \gg1, one gets κ/D≈1/h0+π/(2W)\kappa/D \approx1/h_0+\pi/(2W) and κ/D≈(π/ln⁡2)/W\kappa/D \approx(\pi/\ln2)/W, respectively.

The direct generalization of the comb-like boundary is a periodic structure with the riblets of a finite thickness. The closed-form solution is only available for high rectangular spikes of height h0h_0 and width ww (Figure 2f). If h0/W>0.6h_0/W>0.6, one gets [56]

λ≈h0−W2πG(w/W),(20)\lambda\approx h_0-\frac{W}{2\pi}G(w/W), \tag*{(20)}

where G(ζ)G(\zeta) is the dimensionless function:

G(ζ)=2(1−ζ)ln⁡(2−ζ)+ζln⁡(2ζ−ζ2),ζ=w/W≤1.(21)G(\zeta)=2(1-\zeta)\ln(2-\zeta)+\zeta\ln(2\zeta-\zeta^2), \qquad\zeta=w/W\leq1. \tag*{(21)}

In the limit w→0w\to0, one has G(0)=2ln⁡2G(0)=2\ln2, so that Eq. (18) for the comb-like boundary is retrieved. In the opposite limit ζ=w/W→1\zeta=w/W\to1, one has G(ζ)≈(1−ζ)2G(\zeta)\approx(1-\zeta)^2, i.e., the second term in Eq. (20) vanishes, and one gets λ=h0\lambda=h_0 as expected for a flat absorbing boundary at the height h0h_0. As previously, it is natural to impose the effective partially absorbing boundary at y0=h0y_0=h_0 (at the tips of the riblets), in which case the trapping parameter is

κ≈DW2πG(w/W).(22)\kappa\approx\frac{D}{W}\frac{2\pi}{G(w/W)}. \tag*{(22)}

The last family of the 2D profiles that we considered in this section is the saw-tooth boundary (Figure 2g). The profile is characterized by its height h0h_0 and the base width WW, or by the saw-tooth angle 2α=2arctan⁡(W2h0)2\alpha=2\arctan\left(\frac{W}{2h_0}\right). In this case, one gets [57]

λ=h0−W2πS(α),(23)\lambda=h_0-\frac{W}{2\pi}S(\alpha), \tag*{(23)}

where S(α)S(\alpha) is a dimensionless function of angle α\alpha:

S(α)=γ+2ln⁡2+πtan⁡α−πα+ψ(1+απ),(24)S(\alpha)=\gamma+2\ln2+\frac{\pi}{\tan\alpha}-\frac{\pi}{\alpha}+\psi\left(1+\frac{\alpha}{\pi}\right), \tag*{(24)}

where ψ(⋅)\psi(\cdot) is the digamma function, and γ≈0.5772\gamma\approx0.5772 is the Euler constant. One can check that S(α)S(\alpha) is a monotonously decreasing function of α\alpha. For α=0\alpha=0 (comb-like boundary), one has ψ(1)=−γ\psi(1)=-\gamma and thus S(0)=2ln⁡2S(0)=2\ln2, in agreement with Eq. (18). For α=π/2\alpha=\pi/2 (flat surface), one gets ψ(3/2)=−γ−2ln⁡2+2\psi(3/2)=-\gamma-2\ln2+2, so that S(π/2)=0S(\pi/2)=0, as expected. Another explicit result corresponds to α=π/6\alpha=\pi/6 (equilateral triangle), for which S(π/6)=π3/2−3ln⁡(3)/2S(\pi/6)=\pi\sqrt{3}/2-3\ln(3)/2 and thus λ/W=3/4+3ln⁡(3)/(4π)\lambda/W=\sqrt{3}/4+3\ln(3)/(4\pi) [30]. As previously, setting y0=h0y_0=h_0, one can also introduce the trapping parameter as

κ=DW2πS(α).(25)\kappa=\frac{D}{W}\frac{2\pi}{S(\alpha)}. \tag*{(25)}

More complex profiles appear when the saw-teeth are separated by a piece of flat boundary, with a period WW (the so-called trapezoidal grooves, Figure 2h). The generalized formulas were proposed in Ref. [57], e.g., for high riblets, one has

λ≈h0−[(W−w)ln⁡2π+w2πS(α)],h0/W>1,(26)\lambda\approx h_0-\left[(W-w)\frac{\ln2}{\pi}+\frac{w}{2\pi}S(\alpha)\right],\qquad h_0/W>1, \tag*{(26)}

where ww is the width of the foot of the riblet, and α=atan⁡(12w/h0)\alpha=\operatorname{atan}(\frac{1}{2}w/h_0). Moreover, one can account for rounded wedges at the tips [57]. More numerical and analytical estimates of λ\lambda for different types of boundaries (piece-wise, pre-fractals, convex, with scalloped and club-like riblets, etc.) can be found in Ref. [57]. We also mention an interesting result established numerically in Ref. [30] regarding the strong non-monotonic dependence of the parameter λ\lambda on geometrical parameters of MM-shape profiles.

To conclude this section, we present an example of non-flat boundaries with mixed boundary conditions (absorbing and reflecting). For the comb-like boundary with absorbing spikes and the reflecting base (Figure 2i), the result is [42]

λ=Wln⁡[sinh⁡(πh0/W)]π.(27)\lambda=W\frac{\ln[\sinh(\pi h_0/W)]}{\pi}. \tag*{(27)}

For πh0/W≫1\pi h_0/W\gg1, the saturation limit, λ≈h0−Wln⁡(2)/π\lambda\approx h_0-W\ln(2)/\pi, remains the same as for the absorbing base (since particles do not reach the bottom of the comb structure, the boundary condition at the base becomes irrelevant). As the comb height h0h_0 decreases, the offset parameter λ\lambda also decreases and then becomes negative. In the limit h0/W≪1h_0/W\ll1, the offset parameter diverges to −∞-\infty logarithmically, and the boundary slowly becomes perfectly reflecting. In this limit, Eqs. (9) and (27) become identical provided we set h0/W=σ/2h_0/W=\sigma/2. The factor 2 can be explained by symmetry reasoning: a horizontal absorbing interval laying on the reflecting base creates the diffusive flux twice smaller than the flux created by the same vertical interval standing on that base (due to two accessible sides of this interval). These results can be extended to incorporate other shapes of small absorbers by finding an equivalent absorbing interval with the same log-capacity and using it in Eq. (9). An effective radiative boundary can be set at y0=h0y_0 = h_0 with the trapping parameter

κ=Dh0−Wln⁡[sinh⁡(πh0/W)]/π(28)\kappa= \frac{D}{h_0 - W \ln[\sinh(\pi h_0/W)]/\pi} \tag*{(28)}

which vanishes very slowly in the limit h0→0h_0 \to0.

As earlier in the case of cylinders with absorbing stripes, the above expressions for the trapping parameter κ\kappa, which were derived for a flat boundary, are applicable to curved surfaces. For instance, if the cylinder of radius RR contains grooves or riblets that are parallel to the cylinder axis (Figure 3f,Figure 3g,Figure 3h), one can get κ\kappa by choosing an appropriate expression according to the groove shape, and then substitute it to Eqs. (10), (11) for estimating the MFPT. Even though the height h0h_0 of these grooves and the period WW are formally expected to be relatively small with respect to RR (which plays here the role of HH for the flat boundary), such an estimate remains quite accurate for a broad range of parameters. Moreover, there is no smoothness constraint on such a cylindrical surface with grooves or riblets (e.g., one can even consider a comb-like structure). This feature reveals the crucial role of conformal mapping, on which many of the above expressions rely, and thus distinguishes them from other approximations obtained, e.g., via perturbation theory. More generally, for a cylinder of any simply-connected cross-section Ω\Omega (Figure 3i), one can construct a conformal map from Ω\Omega onto a disk and thus determine its conformal radius rcr_c, from which the MFPT can be estimated by replacing RR by rcr_c in Eq. (11) and setting κ=∞\kappa= \infty. It is important to stress the two-step character of this approximation: one first estimates rcr_c and then evaluates the MFPT for an equivalent disk. In contrast, even though a direct application of conformal mapping to the first-passage problem is possible (see [49] for details), the Poisson equation governing the MFPT is not conformally invariant that brings substantial complications to the exact solution. Despite the simplicity and potential utility of boundary homogenization, its applications are not yet fully explored in the context of target search problems, while systematic investigations on their accuracy and validity range are still missing.

Periodic patterns in three dimensions

When the domain is not translationally invariant along one axis, the original 3D problem is not reducible to a planar one that limits the use of conformal mapping. As a consequence, the analysis of patterns that are periodic in two directions, is much more difficult, with a limited number of available analytical results.

Flat boundaries

Similar to the 2D case, we begin with the trapping parameter of a smooth boundary covered by absorbing and reflecting spots. The first results by Berg and Purcell [60, 61] and by Zwanzig [62] dealt with the particle trapping by small absorbing spots covering a reflecting sphere. The opposite case (a small reflecting spot on the absorbing sphere) was analyzed in [63]. This is still an area of active research [64, 65].

The appeal to electrostatic analogy provides a valuable insight and clear interpretation of the main result on this topic [6, 61]. According to this analogy, the trapping capacity KK of an object and its electric capacity (or capacitance) CC are related as

K=CDK = CD

[6, 61] (here we employ the convention that the capacity of a sphere of radius RR is 4πR4\pi R). There are many analytical methods for estimating CC from the shape of the object [66, 67, 68, 69] that enables fruitful translation of many results from electrostatic to diffusion kinetics and target search problems.

We begin with an illustrative example: a sparse periodic array of absorbing disks of radius aa on a reflecting surface (Figure 4a). If diffusional screening between disks could be ignored, an estimation of the trapping parameter for such a boundary would be straightforward [70]

Periodic cells on a hexagonal lattice with mixed absorbing-reflecting properties

Figure 4. Periodic cells on a hexagonal lattice with mixed absorbing-reflecting properties (dark area is absorbing and light area is reflecting): (a) absorbing disk on the reflecting base; (b) reflecting disk on the absorbing base; (c) a large reflecting disk on the absorbing base, which is equivalent to a thin polygonal absorbing ring on the reflecting base (d).

κ=KA0=CDA0=4aDA0,(29)\kappa= \frac{K}{\mathcal{A}_0} = \frac{CD}{\mathcal{A}_0} = \frac{4aD}{\mathcal{A}_0}, \tag*{(29)}

where A0\mathcal{A}_0 is the area of the periodic cell, and C=4aC = 4a is half of the capacity of a disk in the space R3\mathbb{R}^3 (the factor 1/21/2 accounts for the half of the flux to one side of the disk surface). Even though diffusional screening between disks is always present, the formula (29) is still valid if CC is understood as the capacity of a disk in the lattice, which depends on the arrangement of the disks. In the formal limit when the surface fraction of absorbing disks σ\sigma is equal to 1 (full coverage), one retrieves the infinite capacity of the perfectly conducting plane: C=∞C = \infty. The calculation of the “renormalized” CC (and hence κ\kappa) that accounts for the diffusive interaction between absorbers in the lattice is the main aim of the effective medium approximation in this context. Different analytical methods provide different levels of rigor and accuracy.

The aforementioned electrostatic analogy allows us to reveal some important features of boundary homogenization in three dimensions. The trapping parameter κ\kappa as function of the surface fraction σ\sigma of the absorbing part can be strongly dependent on its shape. To illustrate this point, we analyze two complementary configurations: a “spot configuration” (Figure 4a) formed by a hexagonal lattice of absorbing disks (C=0C=0) on the reflecting base (dC/dy=0dC/dy=0), and a “mesh configuration” (Figure 4b), which is an inverse of the first one: the same lattice but of the reflecting spots (dC/dy=0dC/dy=0) on the absorbing base (C=0C=0). In the limit of small σ\sigma, the first configuration reduces to a sparse periodic arrangement of small disks, so according to Eq. (29), one has κ∝a∝σ\kappa\propto a \propto\sqrt{\sigma}, while the second configuration transforms to a lattice of flat rings (Figure 4c,Figure 4d) and κ\kappa decays much slower: κ∝1/ln⁡(1/σ)\kappa\propto1/\ln(1/\sigma) (this follows from the formula for the capacity of a flat ring [71], see also Eq. (9) for the stripe configuration). This comparison illustrates that the knowledge of σ\sigma alone is insufficient for estimating the trapping parameter without additional information on geometrical settings.

It is instructive to pursue the analysis of these two configurations and to deduce explicit formulas for κ\kappa and then by comparison to appreciate their differences. Our choice of the hexagonal lattice and of the circular shape of absorbing/reflecting spots helps to avoid additional complications related to anisotropy or other asymmetries. To simplify analytical treatment, the periodic cell of this system is often approximated by a circular tube of the same cross-section containing a single disk [72].

Absorbing disk on reflecting base

For the first configuration (Figure 4a), it is convenient to write the trapping parameter κ\kappa in the following form [73, 74, 75]

κ=DA0F1(σ),(30)\kappa= \frac{D}{\sqrt{\mathcal{A}_0}}F_1(\sigma), \tag*{(30)}

where σ\sigma is the surface fraction of absorbing area, and A0\mathcal{A}_0 is the area of the periodic cell of the lattice. In turn, the dimensionless function F1(σ)F_1(\sigma) can be written as

F1(σ)=4πσf(σ)F_1(\sigma)=\frac{4}{\sqrt{\pi}}\sqrt{\sigma}f(\sigma)

to ensure f(0)=1f(0)=1 according to Eq. (29) in the limit σ→0\sigma\to0. The asymptotic behavior of the function f(σ)f(\sigma) as σ→0\sigma\to0 and σ→1\sigma\to1 can be deduced from the known analytical results, including the capacitance approximation.

In 1941, Fock studied the steady flow of electric current down a long circular tube filled with a conducting medium, with a thin circular hole in its center (Figure 3c). In particular, Fock obtained a representation of the function f(σ)f(\sigma) (and hence F1(σ)F_1(\sigma)) as a power-law series of σ\sqrt{\sigma}, evaluated the coefficients of this series in terms of the integrals of Bessel functions, and gave numerical values of the first 12 coefficients [76]. Fock also proposed a simple truncated expression

f(σ)≈11−1.41σ1/2+0.34σ3/2+0.07σ5/2.(31)f(\sigma)\approx\frac{1}{1-1.41\sigma^{1/2}+0.34\sigma^{3/2}+0.07\sigma^{5/2}}. \tag*{(31)}

This expression provides a very accurate behavior of f(σ)f(\sigma) as σ→0\sigma\to0. Moreover, it ensures the expected divergence of f(σ)f(\sigma) as σ→1\sigma\to1 but fails to yield the correct power-law exponent. This issue was resolved by Leppington and Levine [71], who derived the asymptotic behavior near σ→1\sigma\to1:

f(σ)≈12(1−σ)2≈2(1−σ)2,σ→1,(32)f(\sigma) \approx\frac{1}{2(1-\sqrt{\sigma})^2} \approx\frac{2}{(1-\sigma)^2}, \quad\sigma\to1, \tag*{(32)}

where the second relation was obtained by writing 1−σ=(1−σ)(1+σ)≈2(1−σ)1-\sigma=(1-\sqrt{\sigma})(1+\sqrt{\sigma})\approx2(1-\sqrt{\sigma}) as σ→1\sigma\to1. Motivated by the relative smallness of the coefficient of the last term in the denominator of Eq. (31) and combining two asymptotic results together, one arrives at the following approximation

F1(σ)≈4πσ 1+Aσ−(A−1)σ3/2(1−σ)2⏟≈f(σ),(33)F_1(\sigma) \approx\frac{4}{\sqrt{\pi}}\sqrt{\sigma}\,\frac{1+A\sqrt{\sigma}-(A-1)\sigma^{3/2}}{\underbrace{(1-\sigma)^2}_{\approx f(\sigma)}}, \tag*{(33)}

which is applicable for any 0<σ<10<\sigma<1. The coefficient A=1.41A=1.41 is fixed by matching the coefficient in front of σ\sqrt{\sigma} in the Taylor expansion of the Fock’s expression (31) as σ→0\sigma\to0, whereas the coefficient A−1A-1 in front of σ3/2\sigma^{3/2} ensures the correct asymptotic behavior (32) as σ→1\sigma\to1. We stress that this approximation has no fitting parameter. Note that Bernoff et al. derived a similar result by other means [77]. A very similar approximation was earlier proposed by Berezhkovskii et al. by fitting the numerical solution of the Laplace equation [73, 74, 75]

F1(σ)≈4πσ 1+Aσ−Bσ2(1−σ)2⏟≈f(σ).(34)F_1(\sigma) \approx\frac{4}{\sqrt{\pi}}\sqrt{\sigma}\,\frac{1+A\sqrt{\sigma}-B\sigma^2}{\underbrace{(1-\sigma)^2}_{\approx f(\sigma)}}. \tag*{(34)}

The parameters AA and BB were estimated from fitting and were shown to depend on the considered lattice, e.g., A≈1.37A\approx1.37 for the hexagonal lattice [73, 74, 75]. Note that the second parameter BB should be fixed as A−1A-1 to ensure the correct asymptotic behavior (32) as σ→1\sigma\to1.

One can compare Eq. (30) with other results for κ\kappa found in the literature [77]. The solution of the celebrated Berg-Purcell problem [60] for a reflecting sphere covered by a large number N≫1N\gg1 of small absorbing disks of radius aa reads in our notations as

κBP=DaNπR2=4Dπaσ,(35)\kappa_{\mathrm{BP}}=\frac{DaN}{\pi R^2}=\frac{4D}{\pi a}\sigma, \tag*{(35)}

where σ=N(a/2R)2\sigma=N(a/2R)^2. As the Berg-Purcell solution ignores diffusional screening between absorbing spots, it is valid only for σ≪1\sigma\ll1 (the dilute limit). Later Zwanzig heuristically modified the Berg-Purcell result (35) to explicitly account for the reduction of the surface area due to absorbing disks [62]:

κZW=κBP1−σ.(36)\kappa_{\mathrm{ZW}}=\frac{\kappa_{\mathrm{BP}}}{1-\sigma}. \tag*{(36)}

While the diffusive transport towards a spherical boundary is quite different from that on the plane, one can still compare the predictions given by Eqs. (36) and (30). In the latter case, the surface area of the “periodic cell” can be written as A0=4πR2/N\mathcal{A}_0 = 4\pi R^2/N so that Eq. (30) becomes

κ≈DaπσF1(σ)=4Dσπaf(σ).(37)\kappa\approx\frac{D}{a\sqrt{\pi}}\sqrt{\sigma}F_1(\sigma)=\frac{4D\sigma}{\pi a}f(\sigma). \tag*{(37)}

As outlined by Bernoff et al. [77], Zwanzig’s correction (36) yields a slower divergence, (1−σ)−1(1-\sigma)^{-1} instead of (1−σ)−2(1-\sigma)^{-2}, as σ→1\sigma\to1. In turn, Eq. (37), which was originally derived by homogenizing the flat boundary, is applicable for the whole range of σ\sigma even for a sphere.

Reflecting disk on absorbing base

For the second configuration (Figure 4b), we also write

κ=DA0F2(σ).(38)\kappa= \frac{D}{\sqrt{\mathcal{A}_0}}F_2(\sigma). \tag*{(38)}

An important insight comes from the identity κ=D/B\kappa= D/\mathcal{B} [26, 28, 78], where B\mathcal{B} is the blockage coefficient of a potential flow in a tube with a blocking disk. The value of the blockage coefficient was calculated in [26, 79], yielding the following asymptotic behaviors

F2(σ)≈π3/2ln⁡(2/σ),σ≪1,(39)F_2(\sigma) \approx\frac{\pi^{3/2}}{\ln(2/\sigma)},\qquad\sigma\ll1, \tag*{(39)}
F2(σ)≈3π3/24(1−σ)3/2,σ→1.(40)F_2(\sigma) \approx\frac{3\pi^{3/2}}{4(1-\sigma)^{3/2}},\qquad\sigma\to1. \tag*{(40)}

Combining two asymptotic results, one arrives at the following interpolation

F2(σ)≈3π3/24(1−σ)3/2ln⁡(P+Qσ−4/3),(41)F_2(\sigma) \approx\frac{3\pi^{3/2}}{4(1-\sigma)^{3/2}\ln(P+Q\sigma^{-4/3})}, \tag*{(41)}

with two constants PP and QQ that follow by matching Eq. (41) with the asymptotic relations (39a, 39b):

P=exp⁡(1)−24/3≈0.20,Q=24/3≈2.51.(42)P=\exp(1)-2^{4/3}\approx0.20,\qquad Q=2^{4/3}\approx2.51. \tag*{(42)}

While the values of constants AA, BB, PP, QQ are affected by the lattice type (see [73, 74, 75, 77] for details), the functional form of F1F_1 and F2F_2 remains unchanged. As illustrated on Figure 4c,Figure 4d, the limit σ→0\sigma\to0 is particularly interesting because such a configuration forms a sort of spider web of thin flat absorbing rings on the hexagonal lattice. Moreover, using the capacitance argument, one can replace thin flat stripes by thin absorbing cylinders whose arrangement resembles a graphene layer. While a direct computation of the trapping parameter for such a structure is challenging, Eq. (41) can provide a first approximation. The expression (41) for the function F2(σ)F_2(\sigma) can be compared with the results of [78, 80] deduced from fitting the numerical solution:

F2(σ)=3π3/2g(ζ)4ζ3[1−ln⁡(1−ζ)](43)F_2(\sigma)=\frac{3\pi^{3/2}g(\zeta)}{4\zeta^3[1-\ln(1-\zeta)]} \tag*{(43)}

where ζ=1−σ\zeta=\sqrt{1-\sigma}, g(ζ)=1+0.6ζ+2ζ2−1.5ζ3−0.8ζ100g(\zeta)=1+0.6\zeta+2\zeta^2-1.5\zeta^3-0.8\zeta^{100}, and σ\sigma is the surface fraction of absorbing parts. If Eq. (43) may be considered as a benchmark, such a comparison reveals a moderate quality of the interpolation formula (41). Moreover, the asymptotic relation (39) is accurate only at very small σ\sigma, whereas next-order corrections are needed for moderately small σ\sigma. One can also compare these results to the capacitance approximation κ=DC/A0\kappa=DC/\mathcal{A}_0 with the capacity CC of a flat circular ring, given in [71]. Further investigations could clarify these points and bring more accurate approximations.

Hindering effect of a porous membrane

So far we were focused on the effect of mixed absorbing/reflecting patterns on the plane at y=0y=0. In a similar way, boundary homogenization can be employed to estimate the hindering effect of pores inside a semi-permeable membrane. Let us consider a flat absorbing boundary at y=0y=0, which is separated from the source at height HH by a semi-permeable membrane of thickness ww, located at some height h0h_0. The presence of the membrane reduces the diffusive flux from the basic value J0=DC0/HJ_0=DC_0/H without membrane, to [81, 82]

J=C0H−wD+wDσ+2κm(44)J=\frac{C_0}{\frac{H-w}{D}+\frac{w}{D\sigma}+\frac{2}{\kappa_m}} \tag*{(44)}

where σ\sigma is the surface fraction of pores in the membrane, and κm\kappa_m is given by Eq. (30) or Eq. (33), depending on the arrangement of pores. As discussed in § 0.0.1, the ratio C0/JC_0/J can be interpreted as the overall diffusive resistance (or impedance) of the system, allowing for a simple physical interpretation: the bulk resistance (H−w)/D(H-w)/D of the region without membrane is sequentially added to the resistance of the membrane. The latter has two contributions: the resistance w/(Dσ)w/(D\sigma) of multiple pores (connected in parallel) and the additional contributions 2/κm2/\kappa_m accounting for heterogeneous concentration of arrived particles onto the membrane. Expectedly, the overall resistance and thus the flux do not depend on the height h0h_0 due to the sequential addition of resistances. As a consequence, Eqs. (5), (8) define

λ=w(1−1/σ)−2Dκm<0,κ=1wD(1/σ−1)+2κm(45)\lambda=w(1-1/\sigma)-\frac{2D}{\kappa_m}<0,\qquad \kappa=\frac{1}{\frac{w}{D}(1/\sigma-1)+\frac{2}{\kappa_m}} \tag*{(45)}

where the effective radiative boundary was set at y0=0y_0 = 0. Expectedly, the trapping parameter κ\kappa decreases when the width ww of the membrane increases or the pores vanish (σ→0\sigma\to0). Note that the factor 2 in the contribution 2/κm2/\kappa_m comes from two sides of the membrane; if the membrane lies on the absorbing boundary, 2/κm2/\kappa_m should be replaced by 1/κm1/\kappa_m.

Non-flat boundaries

There is only a limited number of analytical results for the trapping parameter κ\kappa for the boundaries with complex 3D profiles (e.g., boundaries with protrusions). Some of these results will be presented below.

Similar to the 2D sinusoidal profile (Figure 1a), one can analyze a wavy absorbing surfaces in three dimensions (i.e., a doubly-periodic sinusoidal deformation with a square or hexagonal pattern). The problem can be treated by perturbation theory that yields [29]

λ≈qh02W,(46)\lambda\approx\frac{q h_0^2}{W}, \tag*{(46)}

where q=2πq = 2\pi for the square pattern and q=4π3q = \frac{4\pi}{\sqrt{3}} for the hexagonal pattern (note a weak dependence on the lattice geometry). These results were validated by numerical simulations [29]. For small values of h0/Wh_0/W (up to 0.1), the agreement between the asymptotic and numerical results is good, confirming the parabolic scaling near the origin in Eqs. (14), (46) for both 2D and 3D inhomogeneities.

A lattice of absorbing/reflecting cylinders (pillars) is the natural extension of the model of rectangular riblets considered in § 0.0.2. For a square lattice of absorbing pillars on the absorbing base (Figure 5a), the trapping parameter was computed and analyzed in [83]. Since the exact expression for κ\kappa is rather cumbersome, we just mention on two limits.

Examples of irregular boundaries

Figure 5. Examples of irregular boundaries: (a) a square lattice of absorbing pillars on a flat boundary; (b,c) a 3D coating of periodic riblets without or with gaps.

(i) For thin and tall pillars, one gets

λ≈h0−Wln⁡(1/σ)−3/4,h0≫W, σ≪1,(47)\lambda\approx h_0 - W\sqrt{\ln(1/\sigma)-3/4}, \qquad h_0 \gg W,\ \sigma\ll1, \tag*{(47)}

where WW is the lattice period and σ\sigma is the surface fraction occupied by the tips of the pillars. Setting an effective radiation boundary at y0=h0y_0 = h_0, one gets

κ≈DW1ln⁡(1/σ)−3/4(48)\kappa\approx\frac{D}{W}\frac{1}{\sqrt{\ln(1/\sigma)-3/4}} \tag*{(48)}

Similar to the 2D case, this formula is independent of h0h_0 since the particles are mostly absorbed near the top of the pillars.

(ii) For thin and short pillars, one has

λ≈h0−W2ln⁡(1/σ)−3/4h0,κ≈DWh0/Wln⁡(1/σ)−3/4,h0≪W, σ≪1.(49)\lambda\approx h_0-W^2\frac{\ln(1/\sigma)-3/4}{h_0},\qquad\kappa\approx\frac{D}{W}\frac{h_0/W}{\ln(1/\sigma)-3/4},\qquad h_0\ll W,\ \sigma\ll1. \tag*{(49)}

In this regime, the value of λ\lambda is affected by the height h0h_0: as the pillar gets shorter, λ\lambda decreases, as expected. The limit of h0=0h_0=0 is given by Eq. (33). In both regimes of high and short pillars, one has λ→−∞\lambda\to-\infty and κ→0\kappa\to0 in the limit σ→0\sigma\to0 because infinitely thin pillars (needles) are inaccessible to diffusion in three dimensions. This is in sharp contrast to the planar case, for which rectangular riblets of width ww were reduced to a comb-like boundary as σ=w/W→0\sigma=w/W\to0, and the latter was still accessible to diffusing particles; in particular, the offset λ\lambda and the trapping parameter κ\kappa remain finite, see Eq. (27).

An approximation for the offset λ\lambda can be derived analytically for a coating made of a periodic system of absorbing riblets with periodic gaps in each riblet (Figure 5b,Figure 5c). This seems to be one of the simplest realizations of a 3D non-differentiable profile. The result is [84]

λ≈W−2λsπln⁡(cosh⁡[πh0W−2λs]),(50)\lambda\approx\frac{W-2\lambda_s}{\pi}\ln\left(\cosh\left[\frac{\pi h_0}{W-2\lambda_s}\right]\right), \tag*{(50)}

where WW is the distance between riblets,

λs=Lπln⁡[sin⁡(πσs2)]<0,(51)\lambda_s=\frac{L}{\pi}\ln\left[\sin\left(\frac{\pi\sigma_s}{2}\right)\right]<0, \tag*{(51)}

L=s+gL=s+g is the period of the solid-gap structure of an individual riblet, ss is the width of the solid part of the riblet (per period), g=L−sg=L-s is the width of the gap, and σs=s/L\sigma_s=s/L. For the case g=0g=0 (no gaps), one has σs=1\sigma_s=1, λs=0\lambda_s=0, and we return to the 2D solution given by Eq. (16).

For a lattice of small absorbers of arbitrary shape on the reflecting plane, an estimation of κ\kappa can be deduced from the appeal to electrostatic analogy (capacitance arguments), see Eq. (29). In fact, a given absorber can be replaced by an equivalent disk, which has the same capacity as the absorber; after that, Eq. (30) can be applied. Unfortunately, this approach being physically appealing has a limited range of validity, namely, a very diluted limit (see [84] for details). Another approach is to use relation K=1/BK=1/\mathcal{B}, where B\mathcal{B} is the blockage coefficient for a potential flow in a tube with a blocking soft body of the shape of the protrusion. For different shapes of bodies, expressions for B\mathcal{B} are given in [26].

We also mention the perturbative computation of the trapping parameter κ\kappa for nonspherical axisymmetric boundaries with partial reactivity κ0\kappa_0 [85]. In sharp contrast to most earlier studies dedicated to deformations of the plane, these boundaries appear as small deformations of a sphere of radius RR, which can be written in spherical coordinates as r(θ)=R+δf(θ)r(\theta)=R+\delta f(\theta), where δ≪R\delta\ll R, and f(θ)f(\theta) is a dimensionless profile depending on the polar angle θ\theta. If a constant concentration C0C_0 is imposed at infinity, the diffusive flux reads as

J=4πDRC01+D/(κ0R)(1+B01+2D/(κ0R)1+D/(κ0R)(δ/R)+O((δ/R)2)).(52)J=\frac{4\pi D R C_0}{1+D/(\kappa_0R)}\left(1+B_0\frac{1+2D/(\kappa_0R)}{1+D/(\kappa_0R)}(\delta/R)+O\left((\delta/R)^2\right)\right). \tag*{(52)}

where B0=12∫0πf(θ)sin⁡(θ) dθB_0=\frac{1}{2}\int_0^\pi f(\theta)\sin(\theta)\,d\theta. When δ=0\delta=0, one retrieves the diffusive flux onto a partially absorbing sphere found by Collins and Kimball [39], which is further reduced to the Smoluchowski flux J0=4πDC0RJ_0=4\pi D C_0R for a perfectly absorbing sphere. Comparison of the diffusive fluxes for δ=0\delta=0 and δ≠0\delta\ne0 allows one to introduce the trapping parameter κ\kappa of an effective spherical boundary as:

1κ=1κ0−(1κ0+RD)B01+2D/(κ0R)1+D/(κ0R)(δ/R)+O((δ/R)2).(53)\frac{1}{\kappa}=\frac{1}{\kappa_0}-\left(\frac{1}{\kappa_0}+\frac{R}{D}\right)B_0\frac{1+2D/(\kappa_0R)}{1+D/(\kappa_0R)}(\delta/R)+O\left((\delta/R)^2\right). \tag*{(53)}

This definition is meaningful only when the right-hand side is positive. One can also introduce the offset λ\lambda, even though the original definition via Eq. (8) has to be adapted to this problem.

To conclude this section, we mention two examples of a coating made from mixed absorbing/reflecting parts (reflecting protrusions on the absorbing base). For a hexagonal lattice of cylindrical pits of height h0h_0 inside a reflecting matrix, one has [58, 59]

Dκ=h0σ+A0F1(σ).(54)\frac{D}{\kappa}=\frac{h_0}{\sigma}+\frac{\sqrt{\mathcal{A}_0}}{F_1(\sigma)}. \tag*{(54)}

where σ\sigma is the absorbing fraction on the plane, A0\mathcal{A}_0 is the area of the periodic cell, and F1(σ)F_1(\sigma) is given by Eq. (33). In turn, for a hexagonal lattice of reflecting cylinders (pillars) on the absorbing plane, the result reads

Dκ=h0σ+A0F2(σ).(55)\frac{D}{\kappa}=\frac{h_0}{\sigma}+\frac{\sqrt{\mathcal{A}_0}}{F_2(\sigma)}. \tag*{(55)}

where F2(σ)F_2(\sigma) is given by Eq. (41). The structure of Eqs. (54), (55) is very similar, with two contributions. The first term represents the diffusion resistance of the pit, while the second term accounts for access resistance (a delay due to the time needed for a particle to find an opening in the interface). The relative contributions of these terms depend on the geometrical properties of the boundary profile, and the functional form in the limits h0→0h_0\to0 and σ→0\sigma\to0 is very different.

Constant-flux approximation

In two previous sections, we mainly focused on absorbing spots or stripes distributed periodically on the plane. The same arguments and often similar results can be derived for curved boundaries such as a sphere (e.g., the Berg-Purcell’s problem [60]) or a cylinder (e.g., Stokes flow in a cylindrical pipe with stripes [46]), as discussed earlier. Moreover, the periodicity assumption can also be relaxed by considering random, uniformly distributed arrangements of spots. In fact, specific local variations of the concentration near the absorbing spots are averaged out in the far field, ensuring a similar form for the offset λ\lambda, up to minor changes (in the same way as different lattices yielded different constants in functions F1(σ)F_1(\sigma) and F2(σ)F_2(\sigma) in § 0.0.5). At the same time, heterogeneous distribution of absorbing spots or their polydispersity (i.e., distribution of their sizes or shapes) can significantly alter the offset λ\lambda, as we illustrated in § 0.0.5 by considering two complementary configurations of absorbing patterns (Figure 4a,Figure 4b). In such situations, other approximations may be needed for estimating the diffusive flux and the trapping parameter.

To give an example, we mention the so-called narrow escape problem when a particle diffusing inside an Euclidean domain Ω⊂Rd\Omega\subset\mathbb{R}^{d} with reflecting boundary ∂Ω\partial\Omega searches for a small hole Γ\Gamma located on that boundary [86, 87]. Many efforts were dedicated to estimating the MFPT TT to this hole [47, 49, 63, 65, 77, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97]. To get accurate estimates in general domains, one needs to employ elaborate mathematical tools such as conformal mapping (see [49, 87] and references therein) or matched asymptotic analysis (see [98] for details). At the same time, boundary homogenization can still be useful even in this setting of a single small absorbing spot. For instance, one of the first results in this field is attributed to Lord Rayleigh who estimated T≈∣Ω∣/(4aD)T \approx|\Omega|/(4aD) for a small circular hole of radius aa on the reflecting sphere of volume ∣Ω∣|\Omega|. Curiously, this leading-order result can be alternatively deduced by replacing the small absorbing spot on the reflecting sphere by an effective homogeneous partially absorbing sphere with the reactivity κBP\kappa_{\mathrm{BP}} given by the Berg-Purcell’s formula (35) with N=1N=1. In this homogeneous setting, the MFPT can be easily found as

T‾=R215D+R3κBP(56)\overline{T}=\frac{R^2}{15D}+\frac{R}{3\kappa_{\mathrm{BP}}} \tag*{(56)}

after averaging over the starting point. Substituting κBP\kappa_{\mathrm{BP}} from Eq. (35), one notices that the second term in Eq. (56) is equal to ∣Ω∣/(4aD)|\Omega|/(4aD) and provides the dominant contribution if a≪Ra \ll R. In other words, despite highly non-uniform distribution of absorbing spots, boundary homogenization still yields the correct result in the leading order. In turn, more advanced techniques are needed for getting next-order corrections or analyzing the dependence on the starting point.

In this section, we briefly describe another approximation that allows to deal with non-uniform distributions of absorbing spots. The so-called constant-flux approximation (also known as self-consistent approximation) was proposed by Shoup et al. to calculate the Smoluchowski-type constant for reactions with a small center situated on the otherwise impenetrable surface of a spherical domain [99] (see also earlier works by Keller and Stein [72]. Let us formulate it in a more general setting of diffusion in an unbounded domain Ω\Omega, whose reflecting boundary ∂Ω\partial\Omega contains partially absorbing spots Γ\Gamma characterized by reactivity κ0\kappa_0. As previously, the steady-state concentration CC obeys the Laplace equation, ΔC=0\Delta C = 0, with a constant concentration C0C_0 at infinity, and mixed boundary conditions:

−D∂nC=κ0Con Γ,(57)-D\partial_n C = \kappa_0 C \quad\text{on } \Gamma, \tag*{(57)}
−D∂nC=0on ∂Ω\Γ.(58)-D\partial_n C = 0 \quad\text{on } \partial\Omega\backslash\Gamma. \tag*{(58)}

The reactivity κ0\kappa_0 can either represent finite reaction probability on the spots, or result from the previous step of boundary homogenization on Γ\Gamma. Note that the limit κ0=∞\kappa_0 = \infty corresponds to the absorbing spots discussed in § 0.0.5. An exact solution of this boundary value problem is difficult even for simple domains such as a disk or a sphere. In particular, even for planar domains, conformal mapping is not much useful for finite κ0\kappa_0 because such a transformation will make the reactivity dependent on the boundary point. The basic idea of the constant-flux approximation consists in replacing the right-hand side of Robin boundary condition (57) by a constant flux density j0j_0. In other words, one aims at approximating the solution CC of the original problem by the solution C^\hat{C} of the modified problem, ΔC^=0\Delta\hat{C} = 0, with inhomogeneous Neumann boundary condition:

−D∂nC^=j0IΓ(x)on ∂Ω,(59)-D\partial_n\hat{C} = j_0 I_{\Gamma}(\mathbf{x}) \quad\text{on } \partial\Omega, \tag*{(59)}

where IΓ(x)I_{\Gamma}(\mathbf{x}) is the indicator function of the spots: IΓ(x)=1I_{\Gamma}(\mathbf{x}) = 1 for x∈Γ\mathbf{x} \in\Gamma, and 0 otherwise. The unknown parameter j0j_0 can be fixed by imposing the self-consistent condition that the Robin boundary condition (57) is satisfied on average on Γ\Gamma:

κ0∫ΓC^(x) dx=∫Γ(−D∂nC^) dx=j0∣Γ∣,(60)\kappa_0 \int_{\Gamma} \hat{C}(\mathbf{x}) \, \mathrm{d}x = \int_{\Gamma} (-D\partial_n\hat{C}) \, \mathrm{d}x = j_0 |\Gamma|, \tag*{(60)}

where ∣Γ∣|\Gamma| is the surface area of the spots Γ\Gamma. Since C^(x)\hat{C}(\mathbf{x}) implicitly depends on j0j_0, this equation can be used to determine j0j_0 (in the limit κ0=∞\kappa_0 = \infty, this equation is reduced to ∫ΓC^(x) dx=0\int_{\Gamma} \hat{C}(\mathbf{x}) \, \mathrm{d}x = 0, which still determines j0j_0). Even though finding the solution C^\hat{C} of the modified problem is in general not elementary, it is much simpler than solving the original problem with mixed boundary conditions (56).

Since the diffusive flux density of the exact solution, −D∂nC-D\partial_n C, is not constant, this approximation fails in a vicinity of the absorbing spot, in the same way as the far-field approximation (3) failed near the boundary. However, the approximate solution C^\hat{C} turns out to be accurate far from the absorbing spots. In particular, the approximation J=j0∣Γ∣J = j_0|\Gamma| for the diffusive flux is very accurate in many settings. In this light, the constant-flux approximation can be seen as an alternative way of boundary homogenization discussed in sections 2 and 3. The major advantage of the constant-flux approximation is that the diffusive flux density is not uniformly distributed over the boundary ∂Ω\partial\Omega, as it was for homogenized boundaries. The accuracy of two methods was compared for absorbing stripes on a cylinder [52].

The original constant-flux approximation by Shoup et al. was later extended and applied to other problems such chemical kinetics of active colloidal particles [100], MFPT in cylindrical and spherical domains [101, 102], and the whole distribution of first-passage times [103, 104, 105]. For instance, for diffusion between two capped coaxial cylinders of height HH and radii ρ\rho and RR (Figure 3e), the MFPT to an absorbing stripe located on the inner cylinder was found in [102]. When the starting point is uniformly distributed, the constant-flux approximation yields

T=H(R2−ρ2)πDρf,f=∑n=1∞Gnn(sin⁡(πσn)πσn)2,(61)T = \frac{H(R^2-\rho^2)}{\pi D\rho}f,\qquad f=\sum_{n=1}^{\infty}\frac{G_n}{n}\left(\frac{\sin(\pi\sigma n)}{\pi\sigma n}\right)^2, \tag*{(61)}

with

Gn=−I1(πnR/H)K0(πnρ/H)+K1(πnR/H)I0(πnρ/H)I1(πnR/H)K1(πnρ/H)−K1(πnR/H)I1(πnρ/H),(62)G_n=-\frac{I_1(\pi nR/H)K_0(\pi n\rho/H)+K_1(\pi nR/H)I_0(\pi n\rho/H)}{I_1(\pi nR/H)K_1(\pi n\rho/H)-K_1(\pi nR/H)I_1(\pi n\rho/H)}, \tag*{(62)}

where Iν(z)I_\nu(z) and Kν(z)K_\nu(z) are the modified Bessel functions of the first and second kind. In turn, using Eq. (9) for the trapping parameter κ\kappa (with the correction factor 1/21/2 due the perpendicular orientation of stripes, see [46]), one gets a much simpler approximation f≈−ln⁡(sin⁡(πσ/2))f \approx-\ln(\sin(\pi\sigma/2)). While the simplicity of this expression is advantageous for getting rapid estimates, the constant-flux approximation in Eq. (61) is more accurate [52].

Another example of the fruitful use of the constant-flux approximation is the computation of the steady-state reaction rate of diffusion-controlled reactions in sheets [106]. In some applications, the absorbing spot is located on one side of a flat layer (a sheet) between two parallel reflecting boundaries and can accessed from the lateral sides (Figure 3d). For instance, one can think of a capped cylinder {(x,y,z)∈R3:x2+y2<R2, 0<z<H}\{(x,y,z)\in\mathbb{R}^3:x^2+y^2<R^2,\ 0<z<H\} of height HH and radius RR, with a constant concentration imposed on the cylinder wall and an absorbing disk of radius ρ\rho lying on the bottom. A semi-analytical solution of the Laplace equation, its constant-flux approximation and asymptotic behavior were derived in [106]. In particular, the constant-flux approximation yielded the following asymptotic behavior

JJ0≈π2(163π+ρHln⁡R2H)−1(ρ≪H≪R),(63)\frac{J}{J_0}\approx\frac{\pi}{2}\left(\frac{16}{3\pi}+\frac{\rho}{H}\ln\frac{R}{2H}\right)^{-1}\qquad(\rho\ll H\ll R), \tag*{(63)}

where J0=4ρDC0J_0=4\rho DC_0 is the diffusive flux on the half-disk in the upper half-space (with 4ρ4\rho being the half of the capacity of the disk). This ratio determines the offset λ\lambda and the trapping parameter κ\kappa. Other asymptotic limits were discussed in [106].

Diffusion inside channels

Finding buried binding sites by diffusing particles can be seen as another variant of target search problems [58, 59, 107]. In this problem, the binding chemical sites (usually modeled as a finite domain with absorbing boundary) are at the bottom of the deep pits made inside a reflecting porous matrix or at the bottom of a deep reflecting canopy (Figure 6a). The wall of the pit creates the geometrical constraint (an entropy barrier) preventing particles to access the absorbing bottom. This setting is an extension of the example of a semi-circular canopy shown in Figure 2c. In this section, we briefly discuss a powerful approximation via the Fick-Jacobs equation that allows one to reduce the original 2D or 3D problems to an effective one-dimensional problem that can be solved explicitly [108, 109, 110, 111, 112, 113, 114]. We start with the planar setting and then briefly mention the 3D case, which is very similar.

Arrays of long pits with buried binding sites at the bottom

Figure 6. Arrays of long pits with buried binding sites at the bottom (in blue). (a) Sinusoidal profile. (b) Rectangular profile of width ww and height h0h_0, on the period WW, with a flat absorbing site at the bottom. (c) Smoothly varying profile w(y)w(y) of a pit of height h0h_0, with a flat absorbing site at the bottom. Light gray indicates inaccessible solid matrix. The particles arrive from a flat source at y=Hy = H located far above the entrances of the pits. Vertical dotted lines delimit one periodic cell.

Offset and trapping parameter of a pit

Let us consider a periodic array of 2D deep pits with reflecting walls and the absorbing bottom (Figure 6b,Figure 6c). As previously, one can focus on a single periodic cell of width WW. The Fick-Jacobs equation inside the pit has the form [108, 109, 110, 111, 112]

ddy[D(y)w(y)ddyc(y)w(y)]=0,0<y<h0,(64)\frac{d}{dy}\left[D(y)w(y)\frac{d}{dy}\frac{c(y)}{w(y)}\right]=0,\quad0<y<h_0, \tag*{(64)}

where w(y)w(y) is the width of the pit at height yy, c(y)c(y) is the effective one-dimensional concentration, and D(y)D(y) is the effective diffusivity determined by w(y)w(y). Among numerous approximations for D(y)D(y), the model of Kalinay and Percus,

D(y)=Datan⁡(12w′(y))12w′(y),(65)D(y)=D\frac{\operatorname{atan}\left(\frac{1}{2}w'(y)\right)}{\frac{1}{2}w'(y)}, \tag*{(65)}

is often employed [108, 109, 110]. For a smooth profile (w′(y)≪1w'(y)\ll1), we assume D(y)=D=constD(y)=D=const in a first approximation [114]. The absorbing bottom of the pit implies the boundary condition c=0c=0, whereas we set c=c0=constc=c_0=const at some y=H≫Wy=H\gg W at the far field. The offset λ\lambda, which is determined by the far-field behavior (3) at large yy, is then evaluated from the explicit solution of Eq. (64) as

λ=h0−W∫0h0dyw(y)(66)\lambda= h_0 - W \int_{0}^{h_0} \frac{\mathrm{d}y}{w(y)} \tag*{(66)}

Setting an effective radiative boundary at y0=h0y_0 = h_0, one gets the trapping parameter

κ=DW(∫0h0dyw(y))−1.(67)\kappa= \frac{D}{W}\left(\int_{0}^{h_0} \frac{\mathrm{d}y}{w(y)}\right)^{-1}. \tag*{(67)}

If there are discontinuities in the profile w(y)w(y) (e.g., near the pit entrance, y=h0y = h_0, as illustrated in Figure 6b,Figure 6c), then the Fick-Jacobs equation is not applicable near these points. In this case, one first employs boundary homogenization to estimate the trapping parameter κ\kappa in the 2D periodic cell, and then imposes a transmission boundary condition for the one-dimensional concentration c(y)c(y) (see [108, 109, 110, 111, 112] and references therein). This boundary condition results in the additional term in the formula (65) for κ\kappa, see [115]. The physical nature of this term (the so-called access resistance) is a delay in particle diffusion due to the time taken by the particle to find the pit entrance.

The above analysis can be carried on in three dimensions for a lattice of long pits by replacing the pit width w(y)w(y) by its cross-sectional area A(y)\mathcal{A}(y) that leads to a general formula

λ=h0−A0∫0h0dyA(y),(68)\lambda= h_0 - \mathcal{A}_0 \int_{0}^{h_0} \frac{\mathrm{d}y}{\mathcal{A}(y)}, \tag*{(68)}

where A0\mathcal{A}_0 is the area of the periodic cell. In this way, the trapping parameter κ\kappa can be analytically derived for a variety of pit profiles (cone, spheroid, spindle, see Refs. [78, 80, 81] for details). We note that the change of the pit profile can modify the functional form of κ(σ)\kappa(\sigma) [26]. Similar to Eq. (54), the discontinuity in the pit profile near the opening leads to an additional contribution to κ\kappa [78, 80, 81].

Mean first-passage time

In a similar way, the Fick-Jacobs equation can be applied to estimate the MFPT to a small target inside an 2D elongated domain with reflecting boundaries [116]. The profile w(y)w(y) of the domain is assumed to be smooth, slowly changing, but otherwise general, whereas the target is small but of an arbitrary shape. To treat the original 2D problem as one-dimensional (Figure 7), we first replace the absorbing target by an equivalent absorbing interval (i.e., the interval of the same log-capacity). In turn, this absorbing interval is then replaced by a partially absorbing horizontal segment of length w(yT)w(y_T) and trapping parameter κ\kappa. The latter can be estimated from Eq. (12) by thinking of the elongated domain as a stripe of width w(yT)w(y_T) with reflecting boundaries. We stress that the value of the trapping parameter κ\kappa depends on the position yTy_T of the target that enables formulating the MFPT within the effective medium approximation. After that, the Poisson equation for MFPT in the elongated domain reduces to 1D equation that resembles the Fick-Jacobs equation:

Three-step transformation of the original MFPT problem inside an elongated planar domain

Figure 7. Three-step transformation of the original MFPT problem inside an elongated planar domain (left) into an effectively one-dimensional problem with semi-permeable semi-absorbing boundary condition (right). A small target of arbitrary shape (here, a red rhombus) is replaced by an equivalent interval (in red) of length ss of the same log-capacity, which is then replaced by a semi-permeable semi-absorbing boundary (in gray) with reactivity κ\kappa.

Dw(y)ddy[w(y)dTdy]=−1,(69)\frac{D}{w(y)}\frac{d}{dy}\left[w(y)\frac{d\mathcal{T}}{dy}\right] = -1, \tag*{(69)}

subject to the effective semi-permeable boundary condition at the target location yTy_T:

T(yT−0)=T(yT+0),(70)\mathcal{T}(y_T-0)=\mathcal{T}(y_T+0), \tag*{(70)}
D[dTdy(yT+0)−dTdy(yT−0)]=κT(yT).(71)D\left[\frac{d\mathcal{T}}{dy}(y_T+0)-\frac{d\mathcal{T}}{dy}(y_T-0)\right]=\kappa\mathcal{T}(y_T). \tag*{(71)}

This problem was then solved analytically for a variety of profiles w(y)w(y) and favorably validated by numerical simulations [116].

This framework has been extended to find the MFPT in 3D elongated domains with axial symmetry [117]. Here, Eq. (69) remains the same, with w(y)w(y) being replaced by a cross-sectional area A(y)A(y) of the domain. The target is modeled by an equivalent disk and the trapping parameter κ\kappa is given by Eq. (30) via boundary homogenization. We stress that the ideas of boundary homogenization can provide some analytical tools not only for calculating the MFPT, but also for evaluating the FPT statistics, in particular, in the long-time limit [118].

Conclusion

Laplacian transport near complex boundaries is relevant in many disciplines, including chemical physics, electrostatics, hydrodynamics, and heat transfer. Despite the long history of intensive research, this field remains very active, with numerous recent advances and applications. In this review, we overviewed several approximations that yielded a simplified yet accurate description of this phenomenon. In particular, we discussed the effective medium theory, boundary homogenization, constant-flux approximation and perturbative analysis that allow one to express the diffusive flux. and other integral properties of the system in terms of its geometric and absorption parameters. For instance, we described the procedure of boundary homogenization when the complex boundary is replaced by an equivalent flat boundary with Robin boundary condition. The fraction of absorbing parts and the amplitude of boundary profile variations do not need to be small, while the boundary can be rough and spiky. The proposed framework may be useful for a rapid modeling and engineering prototyping before proceeding with extensive computer simulations.

At the same time, many other aspects of Laplacian transport remain unexplored in this review: (i) the narrow escape problem and the related matched asymptotic analysis; (ii) diffusive search in porous media formed by absorbing/reflecting spherical beads and efficient semi-analytical methods [119, 120, 121]; (iii) Laplacian transport towards fractal boundaries [17, 18, 20, 21, 122, 123, 124, 125] and its electrochemical measurements [126, 127, 128, 129, 130, 131, 132]; (iv) the impact of partial reactivity of complex boundaries, its approximate treatments (including land surveyor approximation [19, 38, 133]), and the spectral theory based on the Dirichlet-to-Neumann operator [22, 134, 135, 136, 137]; (v) time-dependent (non-stationary) diffusion in complex media, the distribution of first-passage times, and efficient search strategies [14, 32, 35, 36, 138]. We believe that further applications of boundary homogenization and other approximate tools can help uncovering the crucial role of complex geometry and heterogeneity in Laplacian transport and target search problems.

Acknowledgments

A.T.S thanks Alexander M. Berezhkovskii and Paul A. Martin for useful discussions.

References

References

  1. [1]D. ben-Avraham and S. Havlin, Diffusion and Reactions in Fractals and Disordered Systems (Cambridge University Press: Cambridge, UK, 2010).
  2. [2]Z. Schuss, Brownian Dynamics at Boundaries and Interfaces in Physics, Chemistry and Biology (Springer: New York, USA, 2013).DOI
  3. [3]A. Bejan, Shape and Structure, from Engineering to Nature (Cambridge University Press: Cambridge, UK, 2000).
  4. [4]S. Rice, Diffusion-Limited Reactions (Elsevier: Amsterdam, the Netherlands, 1985).
  5. [5]B. D. Hughes, Random Walks and Random Environments (Clarendon Press: Oxford, UK, 1995).
  6. [6]P. L. Krapivsky, S. Redner, and E. Ben-Naim, A Kinetic View of Statistical Physics (Cambridge University Press: New York, USA, 2010).
  7. [7]P. R. Nair and M. A. Alam, Dimensionally frustrated diffusion towards fractal adsorber, Phys. Rev. Lett. 99, 256101 (2007).
  8. [8]H. Nepf, M. Ghisalberti, B. White, and E. Murphy, Retention time and dispersion associated with submerged aquatic canopies, Water Resource Reseach 43, W04422 (2007).
  9. [9]S. L. Edburg, G. Allwine, B. Lamb, D. Stock, H. Thistle, H. Peterson, and B. Strom, A Simple Model to Predict Scalar Dispersion within a Successively Thinned Loblolly Pine Canopy, J. App. Meteor. Climatology 49, 1913–1926 (2010).DOI
  10. [10]G. Z. Ramon and E. M. V. Hoek, Transport through composite membranes, part 2: Impacts of roughness on permeability and fouling. J. Membr. Sci. 425-426, 141–148 (2013).DOI
  11. [11]M. Felici, M. Filoche, and B. Sapoval, Renormalized Random Walk Study of Oxygen Absorption in the Human Lung, Phys. Rev. Lett. 92, 068101 (2004).
  12. [12]D. S. Grebenkov, M. Filoche, B. Sapoval, and M. Felici, Diffusion-Reaction in Branched Structures: Theory and Application to the Lung Acinus, Phys. Rev. Lett. 94, 050602 (2005).
  13. [13]A. S. Serov, C. Salafia, D. S. Grebenkov, and M. Filoche, The Role of Morphology in Mathematical Models of Placental Gas Exchange, J. Appl. Physiol. 120, 17-28 (2016).DOI
  14. [14]K. Lindenberg, R. Metzler, and G. Oshanin (Eds.) Chemical Kinetics: Beyond the Textbook (World Scientific Press: New Jersey, USA, 2019).
  15. [15]D. A. Lauffenburger and J. J. Linderman, Receptors: Models for Binding, Trafficking, and Signaling (Oxford University Press: New York, USA, 1993).
  16. [16]M. Reva, D. A. DiGregorio, and D. S. Grebenkov, A first-passage approach to diffusion-influenced reversible binding: insights into nanoscale signaling at the presynapse, Scient. Rep. 11, 5377 (2021).DOI
  17. [17]M. Filoche, D. S. Grebenkov, J. S. Andrade Jr., and B. Sapoval, Passivation of irregular surfaces accessed by diffusion, Proc. Nat. Acad. Sci. USA 105, 7636 (2008).DOI
  18. [18]T. A. Witten Jr. and L. M. Sander, Diffusion-Limited Aggregation, a Kinetic Critical Phenomenon, Phys. Rev. Lett. 47, 1400 (1981).
  19. [19]B. Sapoval, General Formulation of Laplacian Transfer Across Irregular Surfaces, Phys. Rev. Lett. 73, 3314 (1994).
  20. [20]M. Filoche and B. Sapoval, Transfer Across Random versus Deterministic Fractal Interfaces, Phys. Rev. Lett. 84, 5776 (2000).
  21. [21]P. Levitz, D. S. Grebenkov, M. Zinsmeister, K. M. Kolwankar, and B. Sapoval, Brownian flights over a fractal nest and first passage statistics on irregular surfaces, Phys. Rev. Lett. 96, 180601 (2006).
  22. [22]D. S. Grebenkov, M. Filoche, and B. Sapoval, Mathematical Basis for a General Theory of Laplacian Transport towards Irregular Interfaces, Phys. Rev. E 73, 021103 (2006).
  23. [23]M. Z. Bazant, Exact solutions and physical analogies for unidirectional flows, Phys. Rev. Fluids 1, 024001 (2016).DOI
  24. [24]J. P. Rothstein, Slip on superhydrophobic surfaces, Ann. Rev. Fluid Mech. 42, 89 (2010).
  25. [25]D. G. Crowdy, Frictional slip lengths for unidirectional superhydrophobic grooved surfaces, Phys. Fluids 23, 072001 (2011).DOI
  26. [26]P. A. Martin and A. T. Skvortsov, On blockage coefficients: flow past a body in a pipe, Proc. R. Soc. A 478, 20210677 (2022).DOI
  27. [27]D. P. Hewett and I. J. Hewitt, Homogenized boundary conditions and resonance effects in Faraday cages, Proc. R. Soc. A 472, 20160062 (2016).
  28. [28]P. A. Martin and A. T. Skvortsov, Steady state diffusion in tubular structures: Assessment of one-dimensional models, Eur. J. Appl. Math. 34, 262-279 (2023).DOI
  29. [29]M. M. Fyrillas and C. Pozrikidis, Conductive heat transport across rough surfaces and interfaces between two conforming media, Int. J. Heat and Mass Transfer 44, 1789-1801 (2001).DOI
  30. [30]M. G. Blyth and C. Pozrikidis, Heat conduction across irregular and fractal-like surfaces, Int. J. Heat and Mass Transfer 46, 1329-1339 (2003).DOI
  31. [31]S. Redner, A Guide to First-Passage Processes (Cambridge University Press, 2001).
  32. [32]R. Metzler, G. Oshanin, and S. Redner (Eds) First-Passage Phenomena and Their Applications (World Scientific Press: Singapore, 2014).
  33. [33]L. D. Stone, J. O. Royset, and A. R. Washburn, Optimal Search for Moving Targets (Springer International Publishing, 2016).
  34. [34]J. Masoliver, Random Processes: First-passage and Escape (World Scientific, 2018).
  35. [35]O. Bénichou, C. Loverdo, M. Moreau, and R. Voituriez, Intermittent search strategies, Rev. Mod. Phys. 83, 81-130 (2011).
  36. [36]O. Bénichou and R. Voituriez, From first-passage times of random walks in confinement to geometry-controlled kinetics, Phys. Rep. 539, 225-284 (2014).DOI
  37. [37]P. C. Bressloff and J. Newby, Stochastic models of intracellular transport, Rev. Mod. Phys. 85, 135-196 (2013).
  38. [38]B. Sapoval, Transport Across Irregular Interfaces: Fractal Electrodes, Membranes and Catalysts, in “Fractals and Disordered Systems”, A. Bunde, S. Havlin (Eds) (Springer-Verlag, Berlin, 1996), pp. 233-261.DOI
  39. [39]F. C. Collins and G. E. Kimball, Diffusion-controlled reaction rates, J. Colloid Sci. 4, 425-437 (1949).
  40. [40]D. Vandembroucq and S. Roux, Conformal mapping on rough boundaries I: Applications to harmonic problems, Phys. Rev. E 55, 6171-6185 (1997).DOI
  41. [41]A. T. Skvortsov and A. Walker, Trapping of diffusive particles by rough absorbing surfaces: Boundary smoothing approach, Phys. Rev. E 90, 023202 (2014).
  42. [42]A. T. Skvortsov, A. M. Berezhkovskii, and L. Dagdug, Steady-state flux of diffusing particles to a rough boundary formed by absorbing spikes periodically protruding from a reflecting base, J. Chem. Phys. 150, 194109 (2019).DOI
  43. [43]P. A. Martin, On acoustic and electric Faraday cages, Proc. R. Soc. A 470, 20140344 (2014).
  44. [44]J. R. Philip, Flows satisfying mixed no-slip and no-shear conditions, Z. Angew. Math. Phys. 23, 353-372 (1972).DOI
  45. [45]B. Ya. Moizhes, Averaged electrostatic boundary conditions for metallic meshes, Zh. Tech. Fiz. 25, 167-176 (1955) (in Russian).
  46. [46]E. Lauga and H. A. Stone, Effective slip in pressure-driven Stokes flow, J. Fluid Mech. 489, 55-77 (2003).DOI
  47. [47]A. Singer, Z. Schuss, and D. Holcman, Narrow escape, part II: the circular disk, J. Stat. Phys. 122, 465 (2006).DOI
  48. [48]J. S. Marshall, Analytical Solutions for an Escape Problem in a Disc with an Arbitrary Distribution of Exit Holes Along Its Boundary, J. Stat. Phys. 165, 920-952 (2016).DOI
  49. [49]D. S. Grebenkov, Universal formula for the mean first passage time in planar domains, Phys. Rev. Lett. 117, 260201 (2016).DOI
  50. [50]A. T. Skvortsov, Mean first passage time for a particle diffusing on a disk with two absorbing traps at the boundary, Phys. Rev. E 102, 012123 (2020).DOI
  51. [51]L. Dagdug, A. M. Berezhkovskii, and A. T. Skvortsov, Trapping of diffusing particles by striped cylindrical surfaces. Boundary homogenization approach, J. Chem. Phys. 142, 234902 (2015).DOI
  52. [52]D. S. Grebenkov, R. Metzler, G. Oshanin, L. Dagdug, A. M. Berezhkovskii, and A. T. Skvortsov, Trapping of diffusing particles by periodic absorbing rings on a cylindrical tube, J. Chem. Phys. 150, 206101 (2019).DOI
  53. [53]D. G. Crowdy, Slip length for longitudinal shear flow over a dilute periodic mattress of protruding bubbles, Phys. Fluids 22, 121703 (2010).DOI
  54. [54]D. G. Crowdy, Analytical formulae for longitudinal slip lengths over unidirectional superhydrophobic surfaces with curved menisci, J. Fluid Mech. 791, R7 (2016).DOI
  55. [55]B. Sapoval, M. Filoche, K. Karamanos, and R. Brizzi, Can one hear the shape of an electrode. I. Numerical study of the active zone in Laplacian transfer, Eur. Phys. J B 9, 739-753 (1999).DOI
  56. [56]S. Richardson, A model for the boundary condition of a porous material. Part 2, J. Fluid Mech. 49, 327-336 (1971).DOI
  57. [57]D. W. Bechert and M. Bartenwerfer, The viscous flow on surfaces with longitudinal ribs, Fluid Mech. 206, 105-129 (1989).DOI
  58. [58]L. Dagdug, A. M. Berezhkovskii, S. M. Bezrukov, and G. H. Weiss, Diffusion-controlled reactions with a binding site hidden in a channel, J. Chem. Phys. 118, 2367-2373 (2003).DOI
  59. [59]A. M. Berezhkovskii, A. Szabo, and H.-X. Zhou, Diffusion-influenced ligand binding to buried sites in macromolecules and transmembrane channels, J. Chem. Phys. 135, 075103 (2011).DOI
  60. [60]H. C. Berg and E. M. Purcell, Physics of chemoreception, Biophys J. 20, 193-219 (1977).
  61. [61]H. C. Berg, Random Walks in Biology (Princeton University Press, 1993).
  62. [62]R. Zwanzig, Diffusion controlled ligand binding to spheres partially covered by receptors: An effective medium treatment, Proc. Nat. Acad. Sci. U.S.A. 87, 5856-5857 (1990).
  63. [63]L. Dagdug, M.-V. Vazquez, A. M. Berezhkovskii, and V. Yu. Zitserman, Boundary homogenization for a sphere with an absorbing cap of arbitrary size, J. Chem. Phys. 145, 214101 (2016).DOI
  64. [64]G. Handy and S. D. Lawley, Revising Berg-Purcell for finite receptor kinetics, Biophys. J. 120, 2237-2248 (2021).DOI
  65. [65]A. E. Lindsay, A. J. Bernoff, and M. J. Ward, First passage statistics for the capture of a Brownian particle by a structured spherical target with multiple surface traps, SIAM Multiscale Model. Simul. 15, 74-109 (2017).DOI
  66. [66]W. R. Smyth, Static and Dynamic Electricity, 3rd Ed. (McGraw-Hill, 1958).
  67. [67]Y. L. Chow and M. M. Yovanovich, The shape factor of capacitance of a conductor, J. Appl. Phys. 53, 8470-8475 (1982).
  68. [68]O. K. Dudko, A. M. Berezhkovskii, and G. H. Weiss, Rate constant for diffusion-influenced ligand binding to receptors of arbitrary shape on a cell surface, J. Chem. Phys. 121, 1562-1565 (2004).
  69. [69]A. M. Berezhkovskii and A. V. Barzykin, Simple formulas for the trapping rate by nonspherical absorber and capacitance of nonspherical conductor, J. Chem. Phys. 126, 106102 (2007).
  70. [70]L. Hill, Effect of rotation on the diffusion-controlled rate of ligand-protein association, Proc. Natl. Acad. Sci. U.S.A. 72, 4918 (1975).
  71. [71]F. G. Leppington and H. Levine, Some axially symmetric potential problems, Proc. Edinburgh Math. Soc. 18, 55 (1972).
  72. [72]K. H. Keller and T. R. Stein, A two-dimensional analysis of porous membrane transport, Math. Biosci. 1, 421-437 (1967).DOI
  73. [73]A. M. Berezhkovskii, Yu. A. Makhnovskii, M. I. Monine, V. Yu. Zitserman, and S. Y. Shvartsman, Boundary homogenization for trapping by patchy surfaces, J. Chem. Phys. 121, 11390 (2004).
  74. [74]A. M. Berezhkovskii, M. I. Monine, C. B. Muratov, and S. Y. Shvartsman, Homogenization of boundary conditions for surfaces with regular arrays of traps, J. Chem. Phys. 124, 036103 (2006).
  75. [75]C. B. Muratov S. Y. and Shvartsman, Boundary Homogenization for Periodic Arrays of Absorbers, Multiscale Modeling and Simulation 7, 44-61 (2008).
  76. [76]V. A. Fock, A theoretical investigation of the acoustical conductivity of a circular aperture in a wall put across a tube, Dokl. Acad. Nauk SSSR 31, 875 (1941).
  77. [77]A. J. Bernoff, A. E. Lindsay, and D. D. Schmidt, Boundary Homogenization and Capture Time Distributions of Semipermeable Membranes with Periodic Patterns of Reactive Sites, SIAM Multiscale Model. Simul. 16, 1411-1447 (2018).DOI
  78. [78]A. T. Skvortsov, L. Dagdug, A. M. Berezhkovskii, and S. M. Bezrukov, Block-age coefficient of cylindrical blocker and diffusion resistance of membrane channels, Phys. Fluids 35, 011702 (2023).DOI
  79. [79]P. A. Martin and A. T. Skvortsov, Scattering by a sphere in a tube, and related problems, J. Acoust. Soc. America 148, 191 (2020).DOI
  80. [80]L. Dagdug, A. T. Skvortsov, A. M. Berezhkovskii, and S. M. Bezrukov, Blocker effect on diffusion resistance of a membrane channel. Dependence on the blocker geometry, J. Phys. Chem. B 126, 6016-6025 (2022).
  81. [81]A. T. Skvortsov, L. Dagdug, E. F. Hilder, A. M. Berezhkovskii, and S. M. Bezrukov, Permeability and diffusion resistance of porous membranes: Analytical theory and its numerical test, J. Chem. Phys. 158, 054114 (2023).DOI
  82. [82]A. T. Skvortsov, L. Dagdug, A. M. Berezhkovskii, I. R. MacGillivray, S. M. Bezrukov, Evaluating diffusion resistance of a constriction in a membrane channel by the method of boundary homogenization, Phys. Rev. E 103, 012408 (2021).DOI
  83. [83]D. S. Grebenkov and A. T. Skvortsov, Diffusion toward a nanoforest of absorbing pillars, J. Chem. Phys. 157, 244102 (2022).DOI
  84. [84]A. T. Skvortsov, D. S. Grebenkov, L. Chan, and A. Ooi, Slip length for a viscous flow over spiky surfaces, Europhys. Lett. 143, 63001 (2023).arxiv.org/abs/2302.07518
  85. [85]F. Piazza and D. S. Grebenkov, Diffusion-controlled reaction rate on non-spherical partially absorbing axisymmetric surfaces, Phys. Chem. Chem. Phys. 21, 25896-25906 (2019).DOI
  86. [86]D. Holcman and Z. Schuss, Control of flux by narrow passages and hidden targets in cellular biology, Phys. Progr. Rep. 76, 074601 (2013).DOI
  87. [87]D. Holcman and Z. Schuss, The narrow escape problem, SIAM Rev. 56, 213-257 (2014).
  88. [88]I. V. Grigoriev, Y. A. Makhnovskii, A. M. Berezhkovskii, and V. Y. Zitserman, Kinetics of escape through a small hole, J. Chem. Phys. 116, 9574 (2002).
  89. [89]A. Singer, Z. Schuss, D. Holcman, and R. S. Eisenberg, Narrow escape, part I, J. Stat. Phys. 122, 437-463 (2006).
  90. [90]A. Singer, Z. Schuss, and D. Holcman, Narrow escape, part III: non-smooth domains and Riemann surfaces, J. Stat. Phys. 122, 491 (2006).
  91. [91]Z. Schuss, A. Singer, and D. Holcman, The narrow escape problem for diffusion in cellular microdomains, Proc. Nat. Acad. Sci. USA 104, 16098-16103 (2007).
  92. [92]O. Bénichou and R. Voituriez, Narrow-escape time problem: time needed for a particle to exit a confining domain through a small window, Phys. Rev. Lett. 100, 168105 (2008).
  93. [93]S. Pillay, M. J. Ward, A. Peirce, and T. Kolokolnikov, An asymptotic analysis of the mean first passage time for narrow escape problems: part I: two-dimensional domains, Multiscale Model. Simul. 8, 803-835 (2010).DOI
  94. [94]A. F. Cheviakov, M. J. Ward, and R. Straube, An asymptotic analysis of the mean first passage time for narrow escape problems: part II: the sphere, Multiscale Model. Simul. 8, 836-870 (2010).
  95. [95]A. F. Cheviakov, A. S. Reimer, and M. J. Ward, Mathematical modeling and numerical computation of narrow escape problems, Phys. Rev. E 85, 021131 (2012).
  96. [96]S. A. Isaacson, A. J. Mauro, and J. Newby, Uniform asymptotic approximation of diffusion to a small target: Generalized reaction models, Phys. Rev. E 94, 042414 (2016).
  97. [97]A. J. Bernoff and A. E. Lindsay, Numerical approximation of diffusive capture rates by planar and spherical surfaces with absorbing pores, SIAM J. Appl. Math. 78, 266-290 (2018).
  98. [98]M. J. Ward and J. B. Keller, Strong Localized Perturbations of Eigenvalue Problems, SIAM J. Appl. Math. 53, 770-798 (1993).DOI
  99. [99]D. Shoup, G. Lipari, and A. Szabo, Diffusion-controlled bimolecular reaction rates. The effect of rotational diffusion and orientation constraints, Biophys. J. 36, 697 (1981).
  100. [100]G. Oshanin, M. N. Popescu, and S. Dietrich, Active colloids in the context of chemical kinetics, J. Phys. A: Math. Theor. 50, 134001 (2017).
  101. [101]D. S. Grebenkov and G. Oshanin, Diffusive escape through a narrow opening: new insights into a classic problem, Phys. Chem. Chem. Phys. 19, 2723-2739 (2017).
  102. [102]D. S. Grebenkov, R. Metzler, and G. Oshanin, Effects of the target aspect ratio and intrinsic reactivity onto diffusive search in bounded domains, New J. Phys. 19, 103025 (2017).
  103. [103]D. S. Grebenkov, R. Metzler, and G. Oshanin, Towards a full quantitative description of single-molecule reaction kinetics in biological cells, Phys. Chem. Chem. Phys. 20, 16393-16401 (2018).
  104. [104]D. S. Grebenkov, R. Metzler, and G. Oshanin, Full distribution of first exit times in the narrow escape problem, New J. Phys. 21, 122001 (2019).
  105. [105]D. S. Grebenkov, R. Metzler, and G. Oshanin, Distribution of first-reaction times with target regions on boundaries of shell-like domains, New J. Phys. 23, 123049 (2021).
  106. [106]D. S. Grebenkov and D. Krapf, Steady-state reaction rate of diffusion-controlled reactions in sheets, J. Chem. Phys 149, 064117 (2018).
  107. [107]R. Samson and J. M. Deutch, Diffusion-controlled reaction rate to a buried active site, J. Chem. Phys. 68, 285-290 (1978).
  108. [108]P. Kalinay and J. K. Percus, Extended Fick-Jacobs equation: Variational approach, Phys. Rev. E 72, 061203 (2005).
  109. [109]P. Kalinay and J. K. Percus, Corrections to the Fick-Jacobs equation, Phys. Rev. E 74, 041203 (2006).
  110. [110]P. Kalinay and J. K. Percus, Approximations of the generalized Fick-Jacobs equation, Phys. Rev. E 78, 021103 (2008).
  111. [111]M. Mangeat, T. Guérin, and D. S. Dean, Dispersion in two dimensional channels – the Fick-Jacobs approximation revisited, J. Stat. Mech. 123205 (2017).
  112. [112]M. Mangeat, T. Guérin, and D. S. Dean, Dispersion in two-dimensional periodic channels with discontinuous profiles, J. Chem. Phys. 149, 124105 (2018).
  113. [113]A. M. Berezhkovskii, M. A. Pustovoit, and S. M. Bezrukov, Diffusion in a tube of varying cross section: Numerical study of reduction to effective one-dimensional description, J. Chem. Phys. 126, 134706 (2007).
  114. [114]A. M. Berezhkovskii, L. Dagdug, and S. M. Bezrukov, Range of applicability of modified Fick-Jacobs equation in two dimensions, J. Chem. Phys. 143, 164102 (2015).
  115. [115]L. Dagdug, A. M. Berezhkovskii, V. Yu. Zitserman, and S. M. Bezrukov, Trapping of particles diffusing in two dimensions by a hidden binding site, Phys. Rev. E 103, 012135 (2021).DOI
  116. [116]D. S. Grebenkov and A. T. Skvortsov, Mean first-passage time to a small absorbing target in an elongated planar domain, New J. Phys. 22, 113024 (2020).
  117. [117]D. S. Grebenkov and A. T. Skvortsov, Mean first-passage time to a small absorbing target in three-dimensional elongated domains, Phys. Rev. E 105, 054107 (2022).DOI
  118. [118]D. S. Grebenkov and A. T. Skvortsov, Survival in a nanoforest of absorbing pillars, J. Phys. A: Math. Theor. 56, 165002 (2023).
  119. [119]M. Galanti, D. Fanelli, S. D. Traytak, and F. Piazza, Theory of diffusion-influenced reactions in complex geometries, Phys. Chem. Chem. Phys. 18, 15950-15954 (2016).
  120. [120]D. S. Grebenkov and S. D. Traytak, Semi-analytical computation of Laplacian Green functions in three-dimensional domains with disconnected spherical boundaries, J. Comput. Phys. 379, 91-117 (2019).DOI
  121. [121]D. S. Grebenkov, Diffusion toward non-overlapping partially reactive spherical traps: fresh insights onto classic problems, J. Chem. Phys. 152, 244108 (2020).
  122. [122]B. B. Mandelbrot and C. J. G. Evertsz, The potential distribution around growing fractal clusters, Nature 348, 143 (1990).DOI
  123. [123]D. S. Grebenkov, What Makes a Boundary Less Accessible, Phys. Rev. Lett. 95, 200602 (2005).
  124. [124]J. S. Andrade Jr., A. D. Araújo, M. Filoche, and B. Sapoval, Screening Effects in Flow through Rough Channels, Phys. Rev. Lett. 98, 194101 (2007).
  125. [125]A. Rozanova-Pierrat, D. S. Grebenkov, and B. Sapoval, Faster Diffusion across an Irregular Boundary, Phys. Rev. Lett. 108, 240602 (2012).
  126. [126]R. de Levie, The influence of surface roughness of solid electrodes on electrochemical measurements, Electrochimica Acta 10, 113 (1965).DOI
  127. [127]S. H. Liu, Fractal Model for the ac Response of a Rough Interface, Phys. Rev. Lett. 55, 529 (1985).
  128. [128]T. C. Halsey, Frequency dependence of the double layer impedance at a rough surface, Phys. Rev. A 35, 3512 (1987).
  129. [129]R. de Levie, Fractals and rough electrodes, J. Electroanal. Chem. 281, 1 (1990).DOI
  130. [130]T. C. Halsey and M. Leibig, The double layer impedance at a rough surface. Theoretical results, Ann. Phys. 219, 109 (1992).
  131. [131]T. Pajkossy, Impedance of rough capacitive electrodes, J. Electroanal. Chem. 364, 111 (1994).
  132. [132]E. Chassaing and B. Sapoval, Electrochemical Impedance of Blocking Quasi-Fractal 3D Electrodes, J. Electrochem. Soc. 141, 2711 (1994).
  133. [133]B. Sapoval, J. S. Andrade Jr., and M. Filoche, Catalytic effectiveness of irregular interfaces and rough pores: the “land surveyor approximation”, Chem. Engnr. Sci. 56, 5011-5023 (2001).
  134. [134]D. S. Grebenkov, Scaling Properties of the Spread Harmonic Measures, Frac-tals 14, 231-243 (2006).
  135. [135]D. S. Grebenkov, Analytical representations of the spread harmonic measure, Phys. Rev. E 91, 052108 (2015).
  136. [136]D. S. Grebenkov, Spectral theory of imperfect diffusion-controlled reactions on heterogeneous catalytic surfaces, J. Chem. Phys. 151, 104108 (2019).
  137. [137]D. S. Grebenkov, Paradigm Shift in Diffusion-Mediated Surface Phenomena, Phys. Rev. Lett. 125, 078102 (2020).
  138. [138]D. S. Grebenkov, R. Metzler, and G. Oshanin, Strong defocusing of molecular reaction times results from an interplay of geometry and reaction control, Commun. Chem. 1, 96 (2018).

Paper details

vlnc:W_6dtwqdvw
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 10

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents