Deterministic Nonvolatile Electrical Switching

• Deterministic nonvolatile electrical switching is a technology that achieves repeatable, stable state changes using targeted electrical stimuli, ensuring nonrandom, preset outcomes.
• It leverages diverse mechanisms—including superconducting vortex injection, phase-change transformations, and spin–torque dynamics—to enable reliable switching in memory and logic devices.
• This approach offers high endurance, low energy consumption, and robust state retention, paving the way for scalable, high-performance computing and neuromorphic systems.

Deterministic nonvolatile electrical switching refers to the ability to reversibly control the state of a physical system—such as resistance, critical current, magnetization, or phase—using electrical stimuli in a manner that is both precisely repeatable and stable once the driving signal is removed. This property underpins high-density, low-power, and high-speed memory or logic operations in diverse classes of emerging devices. Deterministic operation precludes stochastic or random fluctuations in the state after programming, while nonvolatility ensures robust retention in the absence of continued excitation.

1. Device Classes and Mechanistic Principles

Deterministic nonvolatile electrical switching spans several material, device, and physical domains. Key implemented classes include:

• Superconducting cross-bar junctions: Here, isolated Abrikosov vortex injection enables binary switching of the critical current $I_c$ between high and low states in a Josephson junction by driving a minimal current past a threshold under a small perpendicular magnetic field. The trapped vortex locally quenches the order parameter, sharply reducing $I_c$. The system's two vortex configurations (zero or one trapped vortex) form two nonvolatile, electrically toggleable states. The process is fully deterministic: the state can be set and reset indefinitely by the sign of applied current, with retention until an external erasure event, such as vortex annihilation, occurs. Switching is achieved at dramatically reduced current densities, and the underlying force dynamics and energy scaling are governed by Lorentz forces and surface barriers, as established by time-dependent Ginzburg–Landau simulations (Ma et al., 24 Jan 2026).
• Nonvolatile phase-change photonic and electronic devices (PCM-based): Materials such as Ge$_2$Sb$_2$Te$_5$, Sb$_2$S$_3$, and related chalcogenides exhibit reversible amorphous–crystalline phase transitions, triggered by Joule heating or optically driven heating above characteristic melting ($T_m$) or crystallization ($T_c$) temperatures. Device states are fixed by the local PCM structure, corresponding to pronounced contrasts in optical or electrical properties, and are absolutely retained absent further excitation (“zero-static-power” operation) (Gosciniak, 2021, Zheng et al., 2020, Sun et al., 2024). Deterministic multi-level states can be engineered by geometrical patterning of microheaters and tailoring the hotspot profile, allowing for analog modulation and robust cycling.
• Electronic phase transitions and correlated oxides: In Mott insulators, correlated electron systems exhibit electric-field-induced insulator-to-metal (IMT) or metal-to-insulator transitions, as in AM$_4$X$_8$ lacunar-spinels (Cario et al., 2013) and carbon-doped TMOs (CeRAM) (Araujo et al., 2022). Threshold electrical fields collapse the correlated gap, yielding sharp, deterministic resistive switching between ON and OFF states. The process is electronic, with minimal Joule heating, rapid (<2 ns) switching, and outstanding endurance—a sharp contrast to filamentary, stochastic oxide memories.
• Spintronic devices (magnetic, antiferromagnetic, and altermagnetic systems):
  • Spin–orbit torque–driven switching: Heavy metal/ferromagnet (HM/FM) stacks with perpendicular magnetic anisotropy are toggled deterministically by in-plane current pulses that inject spin currents, exerting damping-like and field-like torques. Symmetry breaking via interface engineering, built-in exchange bias, or use of low-symmetry substrates (e.g., WTe$_2$) enables field-free operation (Zhu, 2023, Kajale et al., 2023, Yang et al., 2019).
  • Antiferromagnetic and altermagnetic switching: Orthogonal damping-like torques in complex AFMs such as NiO achieve fully deterministic, 180$^\circ$ Néel vector reorientation, directly mapping electrical pulse polarity to AFM domain state, with high readout fidelity via spin Hall magnetoresistance (Qiao et al., 2024). In altermagnets (e.g., MnTe bilayers), parity symmetry breaking by atomic-scale layering or chemical environment makes state reversal uniquely determined by current direction, with switchability confirmed by band-structure and torque calculations (Chen et al., 2024).
• Ferroelectric and multiferroic-based nonvolatile switches: Nonvolatile polarization states in ferroelectric gate dielectrics modulate interfacial or channel properties in FeFETs, Mott-FeFETs, and magnetoelectric logic cells. Ferroelectric remanence enables data retention, and device-level innovations decouple programming from readout, allowing deterministic, low-voltage, and energy-efficient switching, as seen in VO$_2$-channel Mott-FeFETs (Vaidya et al., 2021) and hybrid ferroelectric/spin–orbit torque logic (Yang et al., 2019).
• Correlated and memristive nanoswitches: Model “perfect memristors” exhibit deterministic, nonvolatile switching between multiple resistance states by controlling a nanoparticle’s location in a double-well potential landscape, with transition triggered by local Joule heating, and state retention ensured by energy barriers (Savel'ev et al., 2013).

2. Quantitative Performance and Device Metrics

Deterministic nonvolatile electrical switching is evaluated using several critical figures of merit:

Device Class	Switching Speed	Endurance	Retention
Cross-bar SC junction	10 ns (vortex)	$>$100 cycles (no deg)	Inherent (supercurr)
PCM optical/electrical	5–200 ns	%%%%15$_8$16%%%% cycles	$>$10 years (85$^\circ$C)
CeRAM (C-doped TMO)	$<$2 ns	%%%%20$_8$21%%%% cycles	24 hr @$473$ K, $>$1h@$673$K
Mott RRAM	50 ns–10 $\mu$s	%%%%26$_8$27%%%% cycles	Months
AFM (NiO, MnTe, etc.)	(Sim: ns)	(Stab. $>$10 yr, barrier)	Room $T$
SOT-FM/Crossbar logic	$<$1 ns–ms	%%%%31$_8$32%%%% cycles	$>$10 yr

Additional system metrics include: high ON/OFF ratios ($10^3$–$10^4$), low intrinsic switching energies ($<1$ pJ to $<$10 fJ), low write-current densities (down to $5 \times 10^5$ A/cm$^2$ in SC junctions, $2.23 \times 10^6$ A/cm$^2$ in vdW SOT devices), and scalability to sub-50 nm nodes or beyond $10^8$ devices/cm$^2$ (Ma et al., 24 Jan 2026, Gosciniak, 2021, Cario et al., 2013, Chen et al., 2024, Kajale et al., 2023, Araujo et al., 2022, Qiao et al., 2024).

3. Physical Mechanisms Ensuring Determinism and Nonvolatility

Determinism in electrical switching arises from mechanisms that enforce a unique mapping between input pulse (amplitude, polarity, sequence) and device state, coupled with robust energy barriers for state retention:

• Superconducting Vortex Injection: Each current pulse of appropriate sign injects a single vortex; only one vortex may be trapped at a time for given $H_z$, yielding a uniquely defined $I_c$ due to geometrically-induced asymmetry. Vortex configurations remain stable at zero current (Ma et al., 24 Jan 2026).
• Phase Change (PCM): Each pulse exceeding $T_m$ or $T_c$ triggers a full volume-scale transformation between amorphous and crystalline states; atomic structures are immobile at room $T$, preventing drift (Gosciniak, 2021, Zheng et al., 2020, Sun et al., 2024).
• Correlated Electron Transitions: In Mott insulators and CeRAM, a field-driven, electronic band collapse or filling of a well-defined impurity band yields abrupt, repeatable insulator–metal switching; no ionic motion or stochastic filamentation, enabling immediate and reproducible switching, even at cryogenic or elevated temperatures (Cario et al., 2013, Araujo et al., 2022).
• Spin–Torque Switching: The torque symmetry and anisotropy ensure only one stable minimum per pulse protocol (e.g., 180$^\circ$ reversal in specific SOT or altermagnetic geometries); field-free switching is guaranteed by built-in or structurally engineered symmetry breaking (stacking, interfacial chemistry) (Qiao et al., 2024, Chen et al., 2024, Zhu, 2023).
• Ferroelectric Gating and Multiferroics: Remanent polarization defines a persistent surface or interface field, deterministically shifting thresholds (e.g., IMT in Mott-FeFETs) and modulating barrier heights even with $V_G=0$ (Vaidya et al., 2021). Fractional quantization of polarization upon sliding transitions in bilayer multiferroics yields unambiguous state assignment (Lu et al., 25 Dec 2025).

4. Theoretical Descriptions and Modeling

The modeling frameworks span Ginzburg–Landau vortex dynamics, time-dependent heat flow, electronic structure (Berry-phase, Kubo, and DFT treatments), and macrospin/dynamical equations for magnetic systems.

• For cross-bar Josephson junctions, time-dependent Ginzburg–Landau equations model vortex drift under current and field, reproducing observed thresholds and non-monotonic $I_c(H_z)$ (Ma et al., 24 Jan 2026).
• PCMs are modeled by coupled thermal and phase evolution equations, incorporating material-specific latent heat and temperature windows for nucleation–growth kinetics (Gosciniak, 2021, Zheng et al., 2020).
• Mott- and impurity-driven resistive switches invoke Mott-Hubbard models, order-parameter Landau theories as $F(m,E) = \alpha(E) m^2 + \beta m^4$, and correlated hopping transport, all governed by field-induced collapse of correlation gaps or filling of impurity bands (Cario et al., 2013, Araujo et al., 2022).
• Spintronic and altermagnetic devices use coupled LLG-type equations with symmetry-resolved and current-dependent spin–orbit torques, where deterministic switching emerges from landscape anisotropies and torque summations (Qiao et al., 2024, Chen et al., 2024).

5. Applications and Significance for Memory and Logic Integration

Deterministic nonvolatile electrical switching enables critical advances in memory and logic, including:

• Cryogenic RAM and superconducting logic: Cross-bar SC junctions yield dense, low-power, SQUID-free circuits suitable for scalable superconducting memory arrays ($10^8$ bits/cm$^2$), directly leveraging controlled vortex states (Ma et al., 24 Jan 2026).
• PICs and photonic computing: Phase-change nanophotonic modulators with deterministic control unlock multi-level, analog, and reconfigurable photonic circuits for in-memory and neuromorphic optoelectronic computing (Gosciniak, 2021, Sun et al., 2024).
• Correlated electron RAM (CeRAM): Field-driven, electronic (non-filamentary) switching in doped TMOs offers ultrafast, high-endurance, and low-variability arrays compatible with advanced node CMOS (Araujo et al., 2022).
• Antiferro- and altermagnetic MRAM: Memories based on deterministic, electrically programmed Néel vector or altermagnetic polarizations support sub-ns, zero-field operation with robust nonvolatility and fast access (Qiao et al., 2024, Chen et al., 2024).
• Spin–orbit torque logic and memory: FM, ferrimagnetic, and vdW ferromagnetic devices benefit from deterministic, field-free, bidirectional switching at low current densities, enabling Boolean and non-Boolean in-memory computation, nonvolatile logic, and compact, multi-level neuromorphic elements (Yang et al., 2019, Kajale et al., 2023, Zhu, 2023).

6. Outstanding Challenges and Future Directions

Despite remarkable progress, open issues remain:

• Achieving deterministic, field-free (macrospin) switching in scaled PMA devices without requiring engineered asymmetry or built-in fields.
• Reducing stochasticity and variability further in multi-level/analog PCM and electrothermal correlated oxide resistors to CMOS-logic reliability standards for non-binary applications.
• Engineering low-barrier, high-endurance ferroelectric and Mott-IMT heterostructures for fully integrated high-speed nonvolatile logic.
• Extending deterministic, symmetry-driven control to a broader class of antiferro- and altermagnets with robust, room-temperature operation.

Deterministic nonvolatile electrical switching thus represents a unifying physical and technological principle that underlies leading-edge development in superconducting, phase-change, correlated-electron, and spintronic memory and logic, and continues to drive research at the intersection of material physics, nanofabrication, and circuit architecture (Ma et al., 24 Jan 2026, Gosciniak, 2021, Cario et al., 2013, Lu et al., 25 Dec 2025, Chen et al., 2024, Qiao et al., 2024, Kajale et al., 2023, Araujo et al., 2022, Zheng et al., 2020, Sun et al., 2024, Yang et al., 2019, Vaidya et al., 2021, Zhu, 2023, Shuai et al., 2011, Savel'ev et al., 2013).