Dislocation-Driven Interconnects

• Dislocation-driven interconnects are crystalline architectures in which line defects and interfacial disconnections serve as active pathways for transmitting stress, charge, and spin.
• They integrate mechanisms like shear-coupled grain boundary migration, partial-dislocation channels in bilayer graphene, and quantum templating in diamond NV centers.
• Design principles hinge on mode selectivity and field control to transition between jammed, fluctuating, and ordered regimes for reliable mechanical and electronic transport.

Dislocation-driven interconnects are architectures in which crystalline line defects—or the closely related interfacial disconnections that combine dislocation and step character—function as carriers or scaffolds for connectivity. In current research usage, the idea spans several distinct but related settings: mechanically coupled interface migration and grain rotation in microstructures, current-carrying partial-dislocation channels in bilayer graphene, quasi-one-dimensional arrays of spin defects patterned near dislocations in diamond, and twist-programmed dislocation networks that imprint polar and electronic superlattices in oxides (Han et al., 2021, Weckbecker et al., 2018, Zhang et al., 16 Jul 2025, Sandholt et al., 28 Mar 2026). Across these settings, the unifying premise is that a line defect is not merely a passive imperfection but an organized pathway for transmitting shear, stress, charge, spin, polarization, or structural transformation.

1. Definitions and scope

In crystalline microstructure theory, a conventional lattice dislocation is a line defect characterized by a Burgers vector $\mathbf{b}$, whereas a disconnection is a line defect confined to an interface and characterized by both Burgers vector $\mathbf{b}$ and step height $h$. A disconnection mode is therefore written as $(\mathbf{b},h)$, and its glide simultaneously translates one grain tangentially relative to the other and advances or retreats the interface locally (Qiu et al., 2023). In that specific sense, disconnections are interfacial dislocation-driven interconnects: they move along an internal boundary and transmit shear, stress, and lattice rotation while transporting the boundary itself.

The phrase also appears in electronic and quantum settings with a different emphasis. In bilayer graphene, partial dislocations define current-carrying channels and node states (Weckbecker et al., 2018). In diamond, dislocations are proposed as quasi-one-dimensional templates for positioning NV centers into lines of spin qubits that may act as quantum interconnects (Zhang et al., 16 Jul 2025). In twisted SrTiO$_3$, a twist-controlled screw-dislocation network programs vortex-antivortex polar textures and an electronic superlattice (Sandholt et al., 28 Mar 2026).

Context	Active line defect	Interconnect function
Grain and phase boundaries	Disconnection $(\mathbf{b},h)$	Shear-coupled migration, stress transmission, grain rotation
Bilayer graphene	Partial dislocation network	Current-carrying line states and node states
Diamond spin platforms	Dislocation core plus nearby NV centers	Quasi-1D spin-qubit arrays
Twisted oxides	Screw-dislocation network	Polar/electronic patterning with line- and node-localized states

A common misconception is that the term denotes a single standardized device class. The literature instead uses it as an umbrella for line-defect-mediated connectivity across mechanical, electronic, polar, and quantum regimes.

2. Interfacial disconnections as mechanical interconnects

The continuum theory developed by Salvalaglio, Han, and Srolovitz represents an arbitrarily curved interface as a superposition of disconnections associated with crystallographically selected reference interfaces $R^{(k)}$ (Han et al., 2021). For a two-dimensional boundary parameterized by arc length $s$, the local tangent is constrained by the step densities through relations of the form

$$\sum_m h_m^{(1)} \rho_m^{(1)} = -l_2, \qquad \sum_m h_m^{(2)} \rho_m^{(2)} = l_1,$$

so the interface geometry is encoded directly in the line densities $\rho_m^{(k)}(s)$ of the available disconnection modes (Qiu et al., 2023). The state variables are therefore not just the interface profile $\mathbf{b}$0, but also the evolving mode populations.

The governing equations couple anisotropic capillarity, bulk free-energy difference, and elastic interactions of Burgers vectors. In the 2023 grain-rotation formulation, the interface profile evolves under a mobility tensor $\mathbf{b}$1 and driving terms containing the grain-boundary stiffness $\mathbf{b}$2, a resolved shear stress $\mathbf{b}$3, and a bulk free-energy-density jump $\mathbf{b}$4 (Qiu et al., 2023). Because step and Burgers vector are bound together within each mode, normal migration and tangential shear are kinematically inseparable. This is the core reason disconnections act as mechanical interconnects rather than merely as carriers of curvature-driven motion.

A phenomenological interpretation is provided by the modified Cahn–Taylor model for a rectangular embedded grain. For each reference interface $\mathbf{b}$5,

$\mathbf{b}$6

with $\mathbf{b}$7 an effective shear-coupling factor produced by multiple disconnection modes. The resulting rotation rate is

$\mathbf{b}$8

This embeds energetic anisotropy, kinetic anisotropy, and multi-mode disconnection statistics in a single rotation law (Qiu et al., 2023).

The physical consequence is nonlocal grain reorientation without intragranular plasticity. During capillarity-driven shrinkage, simulations show that $\mathbf{b}$9—the Burgers vector density component normal to the local grain boundary—builds up on slow-moving facets, which act as reservoirs of perpendicular dislocation content; under $h$0-driven growth, $h$1 reverses sign and localizes on different facets (Qiu et al., 2023). The elastic interaction kernel decays only logarithmically with distance in two dimensions, so disconnections on one segment transmit stress to distant segments of the same closed boundary. That nonlocality gives the interface the character of a mechanically wired network.

The same framework was benchmarked against molecular dynamics simulations of Trautt and Mishin for four [100] tilt boundaries in Cu—$h$2, $h$3, $h$4, and $h$5—at $h$6 K. The continuum model reproduced the sign of misorientation evolution in all cases, and the ordering of shrinkage rates agreed as $h$7; deviations in shrinkage rates were within $h$8 (Qiu et al., 2023). This establishes that disconnection-mediated interconnect behavior is quantitatively relevant for high-angle boundaries, not only for idealized low-angle constructions.

3. Partial-dislocation channels in bilayer graphene

In bilayer graphene, dislocation-driven interconnects arise from a stacking-dependent Dirac-Weyl Hamiltonian with off-diagonal interlayer coupling $h$9, where the local relative displacement field $(\mathbf{b},h)$0 is determined by the partial-dislocation network (Weckbecker et al., 2018). A single partial is modeled as a smooth transition between AB and AC stackings over a width of about $(\mathbf{b},h)$1 nm, and realistic networks include wandering, annihilating, and intersecting partial lines.

The relevant Burgers vectors are the three partial translations of magnitude $(\mathbf{b},h)$2. The paper finds that charge accumulation states at partials are sensitive to Fermi energy and partial Burgers vector, but not to the screw versus edge character of the partial (Weckbecker et al., 2018). That insensitivity is significant: channel functionality is controlled primarily by topology and local stacking rather than by the detailed local orientation of the line.

The energy dependence is sharp. In the experimental network, with domain size around $(\mathbf{b},h)$3 nm over an area of roughly $(\mathbf{b},h)$4, charge accumulates predominantly on type 3 partials below the Dirac point, while above the Dirac point it accumulates on type 2 partials; very near the Dirac point, localization switches from partials to intersections of partials (Weckbecker et al., 2018). In the model hexagonal network, strong localization occurs on type 3 partials at $(\mathbf{b},h)$5 meV, on nodes at $(\mathbf{b},h)$6, and on type 2 partials at $(\mathbf{b},h)$7 meV.

These states are not merely localized; they are current carrying. The current density executes a spiral motion along the dislocation line, with a strong interlayer component, and near the Dirac point the node states show complex current flow around the intersections (Weckbecker et al., 2018). The interlayer current changes sign along a partial, so forward motion in one layer is periodically transferred to the other. This produces a helical or spiral transport channel embedded within the bilayer rather than a purely planar wire.

The result is a self-assembled network of line channels and junctions whose activity can be selected by Fermi level. A plausible implication is that electrostatic gating can reweight line-based versus node-based conduction without changing the underlying defect structure.

4. Quantum and polar implementations

In diamond, the proposal for dislocation-driven quantum interconnects is to use dislocations as quasi-one-dimensional templates for spin-defect arrays, with NV centers placed close to or at the dislocation core (Zhang et al., 16 Jul 2025). The microscopic motivation is that dislocations possess long-range elastic fields that attract vacancies and impurities, and previous work found NV formation energies near dislocation cores $(\mathbf{b},h)$8–$(\mathbf{b},h)$9 eV lower than in bulk. The 2025 study examined $_3$0 NV configurations near a $_3$1 glide partial and $_3$2 near a $_3$3 partial. Of the former, $_3$4 ($_3$5) favored the triplet over the singlet and $_3$6 ($_3$7) had the triplet at least $_3$8 eV lower; of the latter, $_3$9 ($(\mathbf{b},h)$0) favored a triplet ground state and $(\mathbf{b},h)$1 ($(\mathbf{b},h)$2) had triplet at least $(\mathbf{b},h)$3 eV lower (Zhang et al., 16 Jul 2025).

Charge stability and optical addressability remain viable near the line defect. The representative NVs have lower formation energy than bulk across the entire Fermi-level range within the gap, can be stabilized in the NV$(\mathbf{b},h)$4 charge state over a wide Fermi-level range, and the dislocations themselves do not introduce deep mid-gap levels that would pin the Fermi level (Zhang et al., 16 Jul 2025). The zero-phonon line for the $(\mathbf{b},h)$5 transition varies from $(\mathbf{b},h)$6 to $(\mathbf{b},h)$7 eV, and Debye–Waller factors range from $(\mathbf{b},h)$8 to $(\mathbf{b},h)$9, compared with a bulk NV ZPL of about $R^{(k)}$0 eV and DWF of about $R^{(k)}$1 (Zhang et al., 16 Jul 2025).

The most distinctive quantum-control result concerns coherence. Near dislocations, the axial zero-field splitting $R^{(k)}$2 ranges from $R^{(k)}$3 to $R^{(k)}$4 MHz and the transverse splitting $R^{(k)}$5 reaches as high as $R^{(k)}$6 MHz, whereas ideal bulk NV centers have $R^{(k)}$7 (Zhang et al., 16 Jul 2025). The induced clock transitions at low field produce $R^{(k)}$8 values at $R^{(k)}$9 that can be up to an order of magnitude longer than bulk, and for a representative $s$0-core NV, CPMG with $s$1 pulses yields $s$2 ms in the long-$s$3 limit (Zhang et al., 16 Jul 2025). At the same time, only about $s$4 ($s$5) of representative configurations exhibit intersystem-crossing behavior deemed desirable for a bulk-like optical cycle. The feasibility case is therefore selective rather than universal.

A different implementation appears in twisted freestanding SrTiO$s$6. For twist angles $s$7, interfacial reconstruction produces an ordered square network of screw dislocations with spacing $s$8; at $s$9, both experiment and modeling give a periodicity around $$\sum_m h_m^{(1)} \rho_m^{(1)} = -l_2, \qquad \sum_m h_m^{(2)} \rho_m^{(2)} = l_1,$$0 nm (Sandholt et al., 28 Mar 2026). Four-dimensional STEM reveals long-range ordered vortex-antivortex arrays with nearly continuous polarization rotation, and the associated strain gradients are of order $$\sum_m h_m^{(1)} \rho_m^{(1)} = -l_2, \qquad \sum_m h_m^{(2)} \rho_m^{(2)} = l_1,$$1. Phase-field modeling yields polarization magnitude $$\sum_m h_m^{(1)} \rho_m^{(1)} = -l_2, \qquad \sum_m h_m^{(2)} \rho_m^{(2)} = l_1,$$2 along dislocation lines, decaying within about $$\sum_m h_m^{(1)} \rho_m^{(1)} = -l_2, \qquad \sum_m h_m^{(2)} \rho_m^{(2)} = l_1,$$3 nm from the interface (Sandholt et al., 28 Mar 2026).

Electronic structure calculations further show states within about $$\sum_m h_m^{(1)} \rho_m^{(1)} = -l_2, \qquad \sum_m h_m^{(2)} \rho_m^{(2)} = l_1,$$4 eV above the valence-band maximum localized along screw dislocation lines and nodes (Sandholt et al., 28 Mar 2026). In this setting, the interconnect is not a moving defect but a twist-programmed defect lattice that acts as a fixed mesh of polar and electronic pathways. This suggests a route from dislocation engineering to mesoscale oxide circuitry based on line-localized band-edge states and vortex nodes.

5. Dynamic phases, stochastic transport, and network formalisms

When dislocations themselves are the mobile carriers, their collective dynamics become a transport problem. A two-dimensional discrete dislocation dynamics model of $$\sum_m h_m^{(1)} \rho_m^{(1)} = -l_2, \qquad \sum_m h_m^{(2)} \rho_m^{(2)} = l_1,$$5 straight edge dislocations with overdamped glide,

$$\sum_m h_m^{(1)} \rho_m^{(1)} = -l_2, \qquad \sum_m h_m^{(2)} \rho_m^{(2)} = l_1,$$6

showed three dynamical regimes under increasing drive: jammed, fluctuating, and dynamically ordered (Zhou et al., 2012). In normalized units, the yielding threshold occurs at $$\sum_m h_m^{(1)} \rho_m^{(1)} = -l_2, \qquad \sum_m h_m^{(2)} \rho_m^{(2)} = l_1,$$7, the fluctuating regime extends to about $$\sum_m h_m^{(1)} \rho_m^{(1)} = -l_2, \qquad \sum_m h_m^{(2)} \rho_m^{(2)} = l_1,$$8, and high drive produces moving unipolar walls with nearly linear transport. The fluctuating phase exhibits broadband noise with

$$\sum_m h_m^{(1)} \rho_m^{(1)} = -l_2, \qquad \sum_m h_m^{(2)} \rho_m^{(2)} = l_1,$$9

whereas the ordered phase exhibits narrow-band noise and a sharply peaked velocity distribution (Zhou et al., 2012). For interconnect purposes, the ordered regime is the low-noise analogue of coherent channel flow, while the fluctuating regime is intrinsically noisy and intermittent.

A complementary atomistic-to-mesoscale surrogate casts dislocation glide as a continuous-time random walk on a graph (Moraes et al., 2020). In bcc Fe at $\rho_m^{(k)}(s)$0 K, the MD mobility obeys

$\rho_m^{(k)}(s)$1

with $\rho_m^{(k)}(s)$2 (Moraes et al., 2020). The glide direction is coarse-grained into a ring graph whose nodes are bins along the periodic glide coordinate; forward and backward jumps occur with rates $\rho_m^{(k)}(s)$3 and $\rho_m^{(k)}(s)$4, total rate $\rho_m^{(k)}(s)$5, and KMC timestep

$\rho_m^{(k)}(s)$6

The surrogate reproduced the MD mobility with relative error about $\rho_m^{(k)}(s)$7, while a single MD run required about $\rho_m^{(k)}(s)$8 hours on $\rho_m^{(k)}(s)$9 Intel Xeon CPUs versus about $\mathbf{b}$00 seconds for a single surrogate realization; $\mathbf{b}$01 surrogate runs took about $\mathbf{b}$02 minutes (Moraes et al., 2020). This is a network-theoretic formulation of dislocation-mediated transport, suitable for uncertainty propagation and, plausibly, for interconnect design studies.

Network topology also evolves statistically under load. In fcc Cu DDD simulations during strain hardening, link lengths on active slip systems follow a double-exponential distribution, while inactive systems follow a single exponential distribution (Akhondzadeh et al., 3 Sep 2025). The long tail is attributed to stress-induced bowing of long links on active systems and disappears when the applied stress is removed. The same paper shows that a one-dimensional Poisson splitting process augmented by a super-linear growth law for links above a critical length reproduces the double-exponential form (Akhondzadeh et al., 3 Sep 2025). This identifies a sparse backbone of long, fast links as the transport-dominant subset of the network.

At the most general continuum level, large-strain crystal plasticity can be written directly in terms of dislocation flow. The kinematics use space-time slip trajectories and a crystal scaffold $\mathbf{b}$03, with plastic distortion $\mathbf{b}$04, and the plastic rate is

$\mathbf{b}$05

where $\mathbf{b}$06 is the geometric slip rate associated with dislocations of Burgers vector $\mathbf{b}$07 (Hudson et al., 2021). The corresponding configurational stress is

$\mathbf{b}$08

which reduces in the linearized limit to the classical Peach–Koehler force (Hudson et al., 2021). This formulation places dislocation-driven interconnects within a fully nonlinear field theory of transport by line defects.

6. Field control, design principles, and limitations

The most direct demonstration of active control is the electric-field-driven motion of partial dislocations in single-crystalline ZnS (Li et al., 2022). In that system, $\mathbf{b}$09 partial dislocations move back and forth solely under an applied electric field, while $\mathbf{b}$10 partial dislocations remain motionless. A positively charged $\mathbf{b}$11 Zn-core partial moved away from a positively biased tip and reversed direction under negative bias, with total displacement about $\mathbf{b}$12 nm and maximum velocity about $\mathbf{b}$13 nm/s (Li et al., 2022). Negatively charged $\mathbf{b}$14 S-core partials likewise reversed direction with field polarity, with displacements up to $\mathbf{b}$15 nm and velocities up to $\mathbf{b}$16 nm/s (Li et al., 2022).

The threshold fields were of order $\mathbf{b}$17–$\mathbf{b}$18. In one geometry, motion began around $\mathbf{b}$19 V, corresponding to a local field of about $\mathbf{b}$20 and an estimated critical field of about $\mathbf{b}$21; in another, motion began around $\mathbf{b}$22–$\mathbf{b}$23 V, corresponding to local fields of about $\mathbf{b}$24–$\mathbf{b}$25 (Li et al., 2022). DFT and climbing-NEB calculations identified the underlying mechanism: charged, nonstoichiometric cores experience field-dependent barrier lowering. For a negatively charged $\mathbf{b}$26 S core, the glide barrier drops from about $\mathbf{b}$27 eV/nm at zero field to about $\mathbf{b}$28 eV/nm at $\mathbf{b}$29 in the favorable direction (Li et al., 2022).

Taken together, these studies suggest several general design rules. First, functionality is highly mode selective: not all dislocations or disconnections are active under a given stimulus, as illustrated by the $\mathbf{b}$30 contrast in ZnS and by the fact that only a subset of NV configurations near diamond dislocations exhibit a desirable optical cycle (Li et al., 2022, Zhang et al., 16 Jul 2025). Second, the useful operating regime is not arbitrary: driven dislocation assemblies have jammed, fluctuating, and ordered phases, and only the ordered phase resembles a stable low-noise transport lane (Zhou et al., 2012). Third, crystallography remains decisive. In interface mechanics, the allowed $\mathbf{b}$31 modes determine shear-coupling factors and the sign of grain rotation; in bilayer graphene, the Burgers vector type selects which partials host charge accumulation; in twisted oxides, the twist angle fixes dislocation spacing and therefore the periodicity of the polar and electronic network (Qiu et al., 2023, Weckbecker et al., 2018, Sandholt et al., 28 Mar 2026).

Several limitations recur across the literature. The disconnection-based interface theories are predominantly two-dimensional and use phenomenological mobilities (Han et al., 2021, Qiu et al., 2023). The diamond quantum proposal does not simulate explicit state-transfer protocols along the dislocation-populated line (Zhang et al., 16 Jul 2025). The twisted-perovskite work identifies near-edge states along the dislocation grid but does not yet report direct transport along those lines (Sandholt et al., 28 Mar 2026). The ZnS experiments establish reversible motion under electric field, but do not address large-cycle reliability or device-level integration (Li et al., 2022). These limitations do not negate the central result; they define the present boundary between mechanistic understanding and engineered circuitry.

Dislocation-driven interconnects therefore comprise a family of line-defect-based transport architectures rather than a single mechanism. In microstructures they are embodied by disconnections that wire together migration, shear, stress redistribution, and grain rotation. In low-dimensional and oxide electronics they appear as dislocation-bound current channels, node states, and polar/electronic superlattices. In quantum materials they provide a thermodynamic and structural scaffold for one-dimensional qubit arrays. The shared research problem is the controlled use of line defects as functional pathways inside crystals.