🔬 applied physics

Modeling of a magnetic field sensor based on spin Hall magnetoresistance

This paper presents a comprehensive multiphysics model, validated by experiments on Pt/CoFeB and Ta/CoFeB bilayers, to optimize the design and performance of Spin Hall Magnetoresistance-based magnetic field sensors in Wheatstone bridge configurations by accounting for complex interplays between SMR, AMR, and SOT effects.

Original authors: Syeda Farwa Bukhari, Alessandro Magni, Witold Skowroński, Elena Losero, Vittorio Basso, Carlo Appino, Piotr Wiśniowski, Juergen Langer, Berthold Ocker, Dario Daghero, Michaela Kuepferling

Published 2026-02-23

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Syeda Farwa Bukhari, Alessandro Magni, Witold Skowroński, Elena Losero, Vittorio Basso, Carlo Appino, Piotr Wiśniowski, Juergen Langer, Berthold Ocker, Dario Daghero, Michaela Kuepferling

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 you are trying to listen to a whisper in a very noisy room. That is essentially what engineers face when trying to build the next generation of magnetic field sensors. These tiny devices are the "ears" of modern technology, used in everything from the hard drives in your computer to the navigation systems in your car and even in medical devices that listen to the magnetic whispers of your heart.

For a long time, the industry standard has been a type of sensor called TMR (Tunneling Magnetoresistance). Think of these like high-end, noise-canceling headphones. They work great, but they are expensive to manufacture, pick up a lot of static noise (called "1/f noise"), and often have a "drift" where they slowly lose their calibration over time.

This paper introduces a new, promising alternative: a sensor based on Spin Hall Magnetoresistance (SMR). The authors didn't just build one; they built a sophisticated computer model (a digital twin) to predict exactly how it would behave, and then they built real-life prototypes to prove the model was right.

Here is a breakdown of how it works, using simple analogies:

1. The Core Concept: The "Spin" Traffic Jam

To understand SMR, imagine electricity not just as a flow of water, but as a crowd of people walking down a hallway.

Traditional sensors rely on the crowd's direction changing based on a magnet.
SMR sensors use a special trick called the Spin Hall Effect. Imagine the hallway has a special floor (a heavy metal like Platinum or Tantalum). When the crowd (electrons) walks down it, the floor spins them. Some spin left, some spin right.
These spinning electrons hit a wall (a magnetic layer) next to the hallway. Depending on how the wall is oriented, it either absorbs the spinners or bounces them back. This "bouncing" changes how hard it is for the crowd to move, which changes the electrical resistance.

2. The Problem: The "Barber Pole" vs. The "Magic Push"

Old sensors (AMR) needed a physical structure called a "barber pole" (a striped pattern) to force the magnetic field to sit at a perfect 45-degree angle. This was like trying to keep a spinning top balanced on a tilted table—it was complex and prone to wobbling (hysteresis).

The new SMR sensor uses Spin-Orbit Torque (SOT). Think of this as a "magic push." The electrical current itself generates a force that gently nudges the magnetic layer into the perfect 45-degree angle automatically.

The Benefit: No need for complex physical stripes. The current does the work. This makes the device simpler, thinner, and more stable.

3. The Model: The "Digital Weather Forecast"

The authors created a massive computer simulation to predict how this sensor would behave.

The "Stoner-Wohlfarth" Model: Imagine a field of tiny compasses. In a perfect world, they all turn together like a single giant compass. The model assumes this first.
The "Truncated Astroid" Twist: In the real world, compasses don't always turn perfectly together; sometimes they get stuck in groups (magnetic domains) and jump suddenly. The authors added a special rule to their model (the "truncated astroid") to account for these messy, real-world jumps. It's like upgrading a weather forecast from "It will be sunny" to "It will be sunny, but there might be a sudden thunderstorm in the northeast."

4. The Experiment: The Wheatstone Bridge

To test their theory, they built a Wheatstone Bridge.

The Analogy: Imagine a four-lane race track where cars (electricity) run in a loop. If the track is perfectly symmetrical, the finish line is dead even (zero voltage). But if a magnetic field changes the speed of the cars in two lanes differently, the balance tips, and you get a signal.
They made sensors using Platinum (Pt) and Tantalum (Ta) paired with a magnetic alloy (FeCoB).
The Results:
- Platinum sensors were electrically efficient (low power) but had a bit more magnetic "stiffness."
- Tantalum sensors were magnetically "soft" (very flexible and linear), meaning they responded very smoothly to magnetic changes, but they needed a bit more power to run.

5. Why This Matters

The paper concludes that this new approach is a game-changer for a few reasons:

Simplicity: The sensor is just a thin film sandwich (two layers), making it incredibly easy to manufacture compared to the complex layers of older sensors.
Transparency: Because the layers are so thin (nanometers), the sensor is almost transparent. This means it could be used in devices where you need to see through the sensor, like in advanced optical medical devices or smart contact lenses.
Precision: The model allows engineers to tweak the "recipe" (layer thickness, material choice) to find the perfect balance between low power consumption and high sensitivity.

In a nutshell:
The authors built a digital blueprint and a physical prototype for a new type of magnetic sensor. By using a "magic push" from electricity to align the sensor, they created a device that is simpler to make, potentially cheaper, and capable of being transparent, all while maintaining high sensitivity. It's a step toward making the "ears" of our technology smaller, smarter, and more efficient.

1. Problem Statement

Traditional magnetic field sensors, particularly those based on Tunneling Magnetoresistance (TMR), face significant limitations hindering their widespread adoption in next-generation applications. These limitations include:

Complex Manufacturing: TMR devices require intricate fabrication of multiple magnetic layers.
Noise and Offsets: They suffer from high $1/f$ noise and significant magnetic offsets.
Linearization Issues: Conventional Anisotropic Magnetoresistance (AMR) sensors often require complex "barber pole" structures or flipping coils to achieve linear output and reduce hysteresis.

The paper addresses the need for a robust, low-power, and highly sensitive magnetic field sensor that overcomes these drawbacks. It focuses on Spin Hall Magnetoresistance (SMR) in heavy metal/ferromagnet (HM/FM) bilayers, which offers a simpler structure and the potential for intrinsic linearization via Spin-Orbit Torque (SOT). However, accurately modeling the interplay between SMR, AMR, and SOT in the presence of magnetic domains remains a challenge.

2. Methodology

The authors developed a comprehensive multiphysics modeling framework and validated it against experimental data from fabricated Wheatstone bridge sensors.

A. Theoretical Modeling

The model integrates three distinct physical phenomena:

Current Distribution (Fuchs-Sondheimer): Analyzes the current density distribution in the HM/FM bilayer, treating the layers as parallel resistive paths with distinct magnetoresistive effects (AMR in the FM layer and SMR in the HM layer).
Magnetization Dynamics (Landau-Lifshitz-Gilbert - LLG): Incorporates SOT effects (both field-like and damping-like torques) generated by the spin Hall effect. This allows for the calculation of equilibrium magnetization angles under external fields and current-induced torques.
Hysteresis and Domain Effects (Modified Stoner-Wohlfarth):
- Standard Stoner-Wohlfarth models assume uniform magnetization (single-domain).
- To account for real-world devices with micrometer-sized features, the authors employed a "truncated astroid" approach. This treats the device as an ensemble of non-interacting single-domain particles with a statistical distribution of local anisotropy axes.
- It distinguishes between coherent rotation (reversible) and domain wall (DW) motion (irreversible Barkhausen jumps), parameterized by a factor $T$ (where $T=1$ is pure rotation, $T<1$ includes DW motion).

B. Device Configuration

Structure: Wheatstone bridge configuration using HM (Pt or Ta) / FM (Fe $_{60}$ Co $_{20}$ B $_{20}$ ) bilayers.
Geometry: Two configurations were tested:
- p45: Bridge branches at 45° to the induced magnetic anisotropy (optimized for linear response).
- p0: Bridge branches parallel to the anisotropy.
Fabrication: Samples were sputter-deposited on Si wafers, patterned via optical lithography and ion-beam etching, and annealed in a 1 T magnetic field at 300°C to induce uniaxial anisotropy.

C. Experimental Validation

Materials: Pt/FeCoB and Ta/FeCoB bilayers.
Characterization:
- Longitudinal Magnetoresistance: Measured on Hall bars to determine AMR and SMR contributions.
- Magneto-Optic Kerr Effect (MOKE): Used to image domain structures and measure hysteresis loops ( $M(H)$ ) to extract anisotropy constants and domain wall densities.
- Wheatstone Bridge Output: Measured output voltage vs. applied magnetic field to validate the sensor model.

3. Key Contributions

Unified Multiphysics Model: The paper presents a novel model that simultaneously accounts for SMR, AMR, SOT, and non-uniform magnetization (domain walls). This bridges the gap between microscopic magnetic behavior and macroscopic sensor output.
Truncated Astroid Framework: The introduction of the "truncated astroid" within the Stoner-Wohlfarth model allows for the statistical description of domain wall motion during magnetization reversal, significantly improving the accuracy of hysteresis predictions compared to uniform rotation models.
Design Guidelines for Power and Sensitivity: The authors derived analytical expressions linking material properties (resistivity, SOT efficiency) and geometry to power consumption and sensitivity. They identified that maximizing the Spin Hall torque efficiency and minimizing the resistivity ratio ( $\rho_{HM} \ll \rho_{FM}$ ) are critical for optimization.
Experimental Correlation: The model was rigorously validated against Pt and Ta-based devices, successfully reproducing experimental hysteresis loops and Wheatstone bridge transfer curves, including the effects of slight misalignment angles.

4. Key Results

Model Accuracy: The modified Stoner-Wohlfarth model (with truncated astroids) showed excellent agreement with experimental MOKE hysteresis loops and Wheatstone bridge output voltages for both Pt and Ta samples.
Material Comparison:
- Pt-based Sensors: Exhibited higher resistivity contrast ( $R_{FM} \gg R_{HM}$ ), leading to a larger magnetoresistance signal. However, they showed complex domain structures.
- Ta-based Sensors: Exhibited "softer" magnetic properties with lower coercivity and a higher tendency for single-domain behavior (lower domain wall density). This resulted in a wider linear, hysteresis-free range ( $\pm 200$ $\mu$ T).
Performance Metrics:
- Sensitivity: $\approx 0.1$ $\Omega$ /mT (Pt) and $0.7$ $\Omega$ /mT (Ta).
- Field Error: $\approx 8$ $\mu$ T (Pt) and $4.2$ $\mu$ T (Ta).
- Power Consumption: Estimated minimum power for linearization was $\approx 290$ mW for Pt and $2.17$ W for Ta. The higher power for Ta is attributed to its higher resistivity (comparable to the FM layer), which reduces the current efficiency for generating SOT.
Linearization: The study confirmed that SOT can effectively replace the "barber pole" structure, forcing the magnetization to a 45° angle relative to the current, thereby linearizing the sensor output without complex patterning.

5. Significance

This work provides a critical roadmap for the development of next-generation spintronic magnetic sensors.

Simplification: It demonstrates that SMR-based sensors can achieve high performance with a simple bilayer structure, eliminating the need for complex multi-layer stacks or mechanical flipping coils.
Optimization Tool: The developed model serves as a predictive tool for engineers to find the "sweet spot" in the multi-dimensional parameter space (material choice, layer thickness, geometry) to balance sensitivity, linearity, and power consumption.
Future Applications: The simple, semi-transparent nature of the bilayer structure makes these sensors ideal for niche applications requiring optical access (e.g., biomedical monitoring or integrated optoelectronic systems).
Path Forward: The authors suggest that future improvements could come from exploiting orbital torques to enhance the field-like torque, potentially overcoming current limitations in power efficiency and sensitivity.

In conclusion, the paper successfully validates a sophisticated multiphysics model for SMR sensors, proving that with proper material selection (favoring low-resistivity HM for power efficiency or soft magnetic properties for linearity) and design optimization, SMR sensors can rival or exceed the performance of traditional TMR and AMR sensors while offering superior structural simplicity.