🔬 applied physics

A Physics-Based Circuit Model for Magnetic Tunnel Junctions

This paper presents a validated, physics-based equivalent circuit model for Magnetic Tunnel Junctions implemented in LTspice that accurately simulates key spintronic phenomena and enables efficient circuit-level analysis for applications ranging from neuromorphic computing to spintronic memory.

Original authors: Steven Louis, Hannah Bradley, Artem Litvinenko, Vasyl Tyberkevych

Published 2026-07-21

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Steven Louis, Hannah Bradley, Artem Litvinenko, Vasyl Tyberkevych

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine a world where your computer doesn't just store data as a series of zeros and ones on a hard drive, but as tiny, invisible magnets that can flip their direction instantly. This is the realm of spintronics, a branch of physics that treats the "spin" of electrons like a tiny compass needle. In these devices, information is stored by pointing these needles in different directions. The star player in this game is a component called a Magnetic Tunnel Junction (MTJ). Think of an MTJ as a microscopic sandwich: two layers of magnetic bread with a thin, non-magnetic filling in the middle. If the magnetic needles in the two bread slices point the same way, electricity flows easily through the sandwich. If they point in opposite directions, the flow gets blocked. This ability to switch between "easy flow" and "blocked flow" is what makes them perfect for memory chips that don't lose data when the power is turned off.

However, designing circuits with these magnetic sandwiches is notoriously tricky. Unlike a standard resistor or capacitor, an MTJ doesn't just sit there; its internal magnet is constantly dancing, wobbling, and spinning in response to electric currents and magnetic fields. To predict how a circuit with these wiggly magnets will behave, engineers usually have to solve incredibly complex math equations that describe the physics of spinning particles. It's like trying to design a rollercoaster by calculating the trajectory of every single air molecule in the wind. This paper introduces a clever new way to model these dancing magnets, turning their complex physics into a simple, easy-to-use circuit diagram that computer software can understand instantly.

The researchers behind this study, Steven Louis and his team, have built a new "translator" for magnetic tunnel junctions. Their goal was to create a model that acts like a bridge between the messy, complex world of spinning magnets and the clean, organized world of electronic circuit design. They wanted to see if they could replace the heavy-duty physics math with a standard circuit diagram that behaves exactly the same way, but runs much faster on a computer.

To do this, they imagined the MTJ not just as a resistor, but as a magical circuit box with two sides. On one side, there's the "charge circuit," which handles the electricity flowing in and out. On the other side, there's a "spin circuit," which is a special arrangement of capacitors, inductors, and resistors that mimics the behavior of the magnetic needle inside.

Here is the magic trick: In their model, the magnetic needle's wobble is represented by a capacitor and an inductor. Just as a capacitor stores energy in an electric field and an inductor stores it in a magnetic field, these circuit parts store the "energy" of the magnet's position. When the magnet wants to spin, it's like the capacitor is charging up and the inductor is pushing back. The "friction" that slows the magnet down (called damping) is modeled by a simple resistor. And the "kick" that the electric current gives the magnet to make it spin (called spin-transfer torque) is modeled by a tiny battery or current source inside the circuit.

The team tested this new model by pitting it against the "gold standard" of physics simulations. They ran the same scenarios on their new circuit model and on the traditional, heavy-duty math equations. The results were a perfect match. Whether they were testing the magnet's natural vibration (ferromagnetic resonance), watching it spin wildly to create a signal (spin-torque nano-oscillator), or forcing it to flip from one direction to another using a magnetic field or an electric current, the circuit model got it right.

For example, when they simulated a magnetic field switching the magnet's direction, the new model showed the magnet wobbling and settling down exactly the same way the complex math predicted. They even built a tiny memory cell, consisting of one transistor and one MTJ, and ran it through the circuit simulator. The model successfully predicted how the current would flip the magnet, just like the real physics would.

What makes this so exciting is that the new model is incredibly fast. While the traditional math takes a long time to crunch the numbers for every tiny movement of the magnet, this circuit model can be plugged into standard design software (like LTspice) and run almost instantly. This means engineers can now design complex computer chips with these magnetic components without waiting days for a simulation to finish. The researchers showed that this model works for different types of magnetic fields, different angles, and even for devices that generate microwave signals.

In short, the paper proves that you can describe the complex, dancing behavior of a magnetic memory bit using a simple, standard circuit diagram. It's like taking a chaotic, unpredictable dance routine and writing down a simple set of steps that anyone can follow to get the exact same result. This tool allows designers to mix these advanced magnetic components with regular electronics much more easily, paving the way for faster, smarter, and more energy-efficient computers in the future. The model is so accurate that the difference between the simulation and the real physics is less than a fraction of a percent, making it a reliable tool for building the next generation of technology.

Technical Summary: A Physics-Based Circuit Model for Magnetic Tunnel Junctions

Problem Statement
Magnetic Tunnel Junctions (MTJs) are critical components in modern spintronic applications, ranging from Magnetoresistive Random-Access Memory (MRAM) to microwave signal processing and neuromorphic computing. While accurate physical descriptions of MTJ dynamics exist via the Landau–Lifshitz–Gilbert–Slonczewski (LLGS) equation, integrating these devices into complex electronic system designs requires circuit-level simulation. Existing approaches often rely on numerically solving the LLGS equation within circuit simulators (e.g., LTspice) by mapping angular variables to circuit elements. However, these methods frequently lack a direct physical correspondence between circuit components and the underlying energy storage, dissipation, or torque mechanisms of the MTJ. Furthermore, some high-frequency impedance models fail to account for behavior under static inputs. There is a need for a computationally efficient, physically rigorous equivalent circuit model that accurately reproduces both magnetization dynamics and electrical behavior while facilitating seamless integration into standard circuit simulation environments.

Methodology
The authors propose a physics-based equivalent circuit model for MTJs operating within the macrospin approximation. The methodology involves deriving a circuit topology where electrical variables directly correspond to physical magnetic quantities:

Physical Mapping: The model maps the macrospin energy function $E(m)$ $E (m)$ to an equivalent nonlinear LC circuit.
- Conservative Interactions: Magnetic precession (driven by external fields, anisotropy, and demagnetization) is modeled by a nonlinear capacitor $C(m)$ and a nonlinear inductor $L(m)$ . These components are derived from the second derivatives of the energy function with respect to charge ( $Q$ ) and flux ( $\Phi$ ), which are defined as functions of the polar ( $\theta$ ) and azimuthal ( $\phi$ ) angles of the free layer magnetization.
- Dissipative Torques: Gilbert damping is represented by equivalent resistors ( $R_C$ and $R_L$ ).
- Spin-Transfer Torque (STT): The influence of electric current on magnetization is modeled using dependent current and voltage sources ( $I_s$ and $V_s$ ) that are functions of the MTJ voltage ( $V_{mtj}$ ).
Electrical Behavior: The MTJ resistance $R_{mtj}$ is modeled as a nonlinear intrinsic resistor dependent on the relative orientation of the fixed layer magnetization ( $p$ ) and the free layer magnetization ( $m$ ).
Implementation: The complete model, consisting of a "charge circuit" (handling electrical input/output) and a "spin circuit" (handling magnetic dynamics), was encapsulated as a standalone component in LTspice. The simulation workflow involves a self-consistent iteration where electrical inputs drive the spin circuit to update magnetization angles, which in turn update the resistance for the charge circuit.

Key Contributions

Physically Rigorous Topology: The paper introduces a circuit model where every element (capacitors, inductors, resistors, sources) has a direct physical origin corresponding to specific terms in the LLGS equation (precession, damping, STT).
Time-Dependent Capability: Unlike previous models limited to static inputs or specific high-frequency regimes, this model handles time-dependent magnetic fields and voltage-driven excitations.
Modularity and Extensibility: The framework is designed to be modular. The authors note that additional physical effects (e.g., field-like torque) or complex geometries (e.g., multiple fixed layers) can be incorporated by modifying the energy expression and updating the dependent sources, without redesigning the fundamental circuit topology.
Open-Source Availability: The model is implemented as a single LTspice component, with documentation and examples made available to facilitate adoption by the community.

Results
The validity of the model was established through direct comparison with numerical solutions of the LLGS equation across four distinct operating regimes:

Ferromagnetic Resonance (FMR): Simulations of free-layer relaxation showed excellent agreement with analytical FMR frequencies for fields applied along easy, intermediate, and hard axes. Relative errors were consistently below 0.02%, accurately capturing mode transitions.
Spin-Torque Nano-Oscillators (STNO): The model successfully reproduced large-amplitude, self-sustained oscillations in current-driven MTJs, including non-uniform precessional trajectories. The time evolution of the polar angle matched direct LLGS simulations exactly.
Field-Induced Switching: The model accurately simulated 180° magnetization switching driven by time-varying magnetic fields, including the post-switching "ringing" behavior.
Current-Induced Switching in 1T-1MTJ Cells: The model was integrated into a standard 1-transistor-1-MTJ memory cell. Simulations of write operations (current-induced switching) matched direct LLGS results, confirming the model's utility in hybrid circuit environments.

Significance and Claims
The authors claim that this work provides a "computationally efficient yet physically rigorous" representation of MTJs. The primary significance lies in bridging the gap between micromagnetic physics and circuit-level design. By offering a model that accurately captures macrospin dynamics without the computational overhead of solving differential equations directly within a circuit simulator, the work facilitates the integration of MTJs into complex electronic systems. The authors state that this capability is essential for the advancement of novel spintronic circuit architectures, specifically in the development of next-generation non-volatile memory (MRAM), microwave signal processing devices, and neuromorphic computing systems. The model is presented not as a final, exhaustive solution for all micromagnetic phenomena (such as spin-waves), but as a robust foundation for macrospin-based circuit analysis that can be extended to more complex scenarios.

Technical Summary: A Physics-Based Circuit Model for Magnetic Tunnel Junctions

More like this