Boundary-Induced Absorbing Transitions

• Boundary-induced absorbing transitions are nonequilibrium phase transitions where spatial boundaries localize reactions, creating distinct active and inactive states.
• Models like the Burov–Kessler framework use boundary-exclusive reactions and bulk diffusion to uncover critical thresholds, novel scaling, and aging phenomena.
• These transitions exhibit non-universal scaling with continuously drifting critical exponents, emphasizing the impact of microscopic boundary conditions on system dynamics.

Boundary-induced absorbing transitions refer to nonequilibrium phase transitions in which the presence, properties, or structure of spatial boundaries fundamentally alter the nature, criticality, or universal behavior of transitions between active and absorbing (inactive) states. These phenomena arise when boundary-driven processes such as reactions, particle activation, or dynamical constraints result in critical thresholds and scaling exponents distinct from those dictated solely by bulk dynamics. Boundary-induced transitions challenge the paradigm that absorbing-state critical behavior is universal and controlled only by bulk features, revealing regimes with non-universal exponents, novel scaling, and aging phenomena.

1. Mechanisms and Model Classes

The canonical setting for boundary-induced absorbing transitions involves systems where activation or extinction mechanisms are spatially localized to a boundary, interface, or domain wall, while the bulk dynamics are governed by diffusion or other transport processes. The paradigmatic model introduced by Burov and Kessler consists of particles of type $A$ residing and reproducing only at the boundary site (site $0$ of a semi-infinite lattice), with offspring $O$ diffusing in the bulk ($x>0$) until eventual return to the boundary (Burov et al., 2010).

Relevant mechanisms include:

• Boundary-exclusive reactions: Creation and annihilation processes are permitted only at the boundary, leading to long-time feedback via bulk return.
• Domain-wall activation: In systems with multiple absorbing states, activity can be reinitiated by boundary interfaces between distinct inactive domains, even when bulk is inactive (Lee et al., 2010).
• Inhomogeneous driving or caging: In particulate systems, spatial variations in driving or geometric constraints can generate boundary layers with distinct critical behavior (Bhowmik et al., 2024, Ness et al., 2020).

2. Microscopic Rules and Exact Solutions

A core feature of boundary-induced models is the separation between boundary activation and bulk transport. In the Burov–Kessler model, the microscopic rules are:

• At most $N$ active particles $A$ may occupy the boundary. $N=1$ yields the Barato–Hinrichsen (BH) class; $N\rightarrow\infty$ removes exclusion.
• Each $A$ at site $0$ either dies ($A\rightarrow\varnothing$ with probability $1-p$) or splits ($A\rightarrow A+O$ with probability $p$), after which the $A$ is removed.
• Each $O$ executes normal or anomalous diffusion in $x>0$, returning after a time drawn from a heavy-tailed return-time distribution $\psi(t)\sim t^{-(1+\beta)}$ ($0<\beta\leq 1$).

For $N\rightarrow\infty$, the problem maps to an age-dependent Galton–Watson branching process, admitting a full generating-function solution and explicit equations for extinction probabilities, critical points, and amplitude prefactors (Burov et al., 2010).

3. Critical Points, Exponents, and Non-Universality

Boundary-induced transitions depart sharply from classic universality:

• Critical point: For the Burov–Kessler model, the critical reproduction probability is $p_c=1/2$ for any boundary occupancy constraint.
• Critical exponents: In the $N\to\infty$ limit (mean-field), the survival exponent is

$\delta=\beta/2$

with $\beta=1/2$ for normal diffusion, giving $\delta=1/4$.

• Occupation restriction and $\kappa$ parameter: Introducing finite $N$ or a blocking time $\Delta$ after reproduction interpolates between the mean-field ($N\to\infty$, $\delta=1/4$) and exclusion ($N=1$, $\delta\approx 0.16$) universality classes, with $\delta(N)$ exhibiting a continuous drift as $N$ increases and no evidence of asymptotic universality (Burov et al., 2010).
• Parameter dependence: Varying bulk diffusion constant $D$ or boundary lifetime $\Delta$ tunes a dimensionless parameter $\kappa\propto D\Delta$, allowing critical exponents to be continuously modified.

In generalized contact processes with two absorbing states, the parameter region where the bulk is inactive (subcritical), but boundary creation at domain walls ($\sigma>0$) induces activity, produces a “soft” boundary-induced transition—a finite $\sigma$ causes a nonzero density of active sites, even for subcritical bulk healing rates (Lee et al., 2010). The phase diagram then features distinct lines: a bulk (DP-class) transition, a parity-conserving (PC/DP2) boundary-induced line, and an endpoint with unique scaling.

Model/Regime	Mechanism	Survival Exponent $\delta$
BH ($N=1$)	Single occupancy, exclusion	$\approx 0.16$
Interpolating ($N$)	Partial exclusion	$0.16 \lesssim \delta(N) < 0.25$
Mean-field ($N\to\infty$)	No exclusion ($\kappa\to0$)	$1/4$
PC-line (generalized CP)	Domain-wall activation	$0.289(5)$ (PC/DP2)

4. Boundary-Induced Aging and Dynamical Scaling

Boundary-localized reaction-diffusion models can exhibit aging at criticality, a property not generic in bulk-driven absorbing transitions. For $0<\beta<1/2$, two-time survival probabilities exhibit aging scaling:

$P_S(t, s) \sim (s/t)^{1-\beta},\quad t \gg s$

For $\beta>1/2$, scaling becomes stationary, with logarithmic corrections at $\beta=1/2$. This emergence of non-stationary critical scaling (aging) is a hallmark of the boundary-induced class for subdiffusive or fractal bulk return times (Burov et al., 2010).

5. Structural Sensitivity and Phase Diagram Topologies

Boundary-induced transitions are acutely sensitive to details of system geometry, control parameters, and local rules:

• Jamming in particulate matter: In driven granular systems, absorbing states arise via geometric isolation or caging. Variations in the caging rule (defining the "dead-zone" shell width $\chi$) directly control the topology of the phase diagram—either generating two successive continuous (Manna-class) transitions or a direct weakly first-order absorbing-to-absorbing jump (Ness et al., 2020).
• Control of exponents: The critical exponents at phase boundaries can vary with the strength of caging interactions or the driving parameters, with measured values $\beta^I=0.63\pm0.02$, $\beta^{II}=0.67\pm0.03$ for the two Manna-class boundaries. The relaxation time diverges with exponents $\nu_\parallel^I=1.24\pm0.04$, $\nu_\parallel^{II}=1.21\pm0.03$.
• Spatial inhomogeneity: In spatially inhomogeneous (smooth or discontinuous) driving, the location of absorbing-diffusive transition boundaries and exponents can shift nontrivially. For smoothly varying drive, the critical point coincides with homogeneous theory. For step-like discontinuities, the phase boundary shifts and the order parameter exponent $\beta$ is continuously reduced as the cell size $L_c$ or the fraction of pinned particles increases (Bhowmik et al., 2024).

6. Continuum Theories and Mean-Field Descriptions

Boundary-induced transitions are often captured using coarse-grained mean-field or continuum PDE frameworks. For example, a Manna-like two-component field theory for alive and dead particle densities,

$$\frac{\partial \rho_A}{\partial t} = D\nabla^2 \rho_A + \alpha \rho_A (\phi-\rho_A) - \beta \rho_A (1-\phi) - \nu \rho_A \phi^2$$

with $\alpha\sim\Delta$ (driving), $\beta$ (isolation), and $\nu$ (caging) explains the U-shaped phase boundary and rationalizes how caging or driving inhomogeneity switches the system between direct and two-step boundary-induced absorbing transitions (Ness et al., 2020). In random-organization models with spatially varying drive, one-dimensional continuum equations with spatially dependent coefficients capture the phase boundary shift and exponent suppression observed in simulations (Bhowmik et al., 2024).

7. Universality Class Interpolation, Non-Universal Scaling, and Implications

Boundary-induced absorbing transitions violate the standard universality conjecture for non-equilibrium phase transitions. Exponents drift continuously with boundary occupation number $N$, boundary-blocking time $\Delta$, bulk diffusion $D$, or other microscopic details (Burov et al., 2010, Lee et al., 2010, Bhowmik et al., 2024, Ness et al., 2020). In particular, the survival exponent $\delta$ does not settle to a universal value even as $N\to\infty$, and varying geometrical or dynamical detail interpolates smoothly between known bulk universality classes (DP, Manna) and new boundary-controlled regimes. Domain-wall activation in generalized contact processes produces a unique “soft” boundary-induced region where the system is activated not by critical fluctuations but by finite density, non-critical domain-wall processes.

A plausible implication is that many experimental systems where activity is nucleated or reinitiated at boundaries, flaws, or interfaces may exhibit critical behavior outside conventional bulk universality classes, especially when interface activation, geometric constraints, or inhomogeneous driving are relevant. The non-universality and sensitivity of these transitions call for careful attention to microscopic detail and boundary conditions in both modeling and experimental analysis of absorbing-state phenomena.