Temperature-driven enhancement and sign reversal of field-like torque in Py/FePS$_3$ bilayers

This study demonstrates that interfacing Py with the van der Waals antiferromagnetic insulator FePS$_3$ significantly enhances and reverses the sign of the field-like spin-orbit torque through interfacial effects driven by antiferromagnetic ordering, while leaving the damping-like torque largely unaffected.

The Gist

Imagine you are trying to push a heavy swing (a magnet) to make it move. Usually, you need to push it in a very specific way to get it going efficiently. In the world of tiny electronics (spintronics), scientists use electricity to "push" these magnetic swings to store data in computers. This push is called Spin-Orbit Torque (SOT).

This paper is about a clever experiment where scientists tried to make that push stronger and more controllable by adding a special "magic layer" to the swing.

Here is the story of what they found, explained simply:

1. The Setup: The Swing and the Magic Blanket

• The Swing (Py): The scientists used a thin layer of a magnetic metal called Permalloy (Py). This is the part that actually holds the data (the magnet).
• The Magic Blanket (FePS₃): They placed a very thin, flaky layer of a material called FePS₃ on top of the swing.
  • What is FePS₃? Think of it as a "quiet neighbor." It is an antiferromagnetic insulator.
  • Insulator: It doesn't let electricity flow through it (like a rubber glove).
  • Antiferromagnetic: Inside, its tiny magnetic parts are arranged in a perfect, rigid dance where they cancel each other out. It has no net magnetism of its own, but it has a very strong internal order.

2. The Experiment: Pushing with a Twist

The scientists ran an electric current through the metal swing (Py). Because of physics rules, this current creates a "twist" or a "push" on the magnet. They wanted to see if adding the "Magic Blanket" (FePS₃) changed how hard or in what direction the magnet was pushed.

They measured two types of pushes:

• The "Damping" Push: This is like a brake or a stabilizer. It helps the swing settle down.
• The "Field-Like" Push: This is like a nudge that tries to tilt the swing sideways.

3. The Big Surprise: The Temperature Switch

The scientists did something very smart: they cooled the device down from room temperature to freezing cold, watching how the "push" changed.

What happened to the "Damping" push?
Nothing. It stayed the same. The blanket didn't change the brakes.

What happened to the "Field-Like" push?
This is where the magic happened!

1. It got much stronger: As the device got colder, the push became about 5 times stronger than it was in the metal alone.
2. It flipped direction: This is the wildest part. As they cooled it down, the direction of the push suddenly reversed. It was pushing one way at room temperature, and then, like a switch flipping, it started pushing the exact opposite way when it got cold.

4. Why Did This Happen? (The "Ghost" Effect)

You might ask: "If the blanket (FePS₃) doesn't let electricity through, how did it change the push?"

Think of it like this:
Imagine you are shouting at a friend through a thick wall. You can't walk through the wall, but your voice (the magnetic influence) vibrates the wall, and the wall vibrates back at you.

• No Electricity Flow: The scientists proved that almost no electricity actually went through the FePS₃ blanket. It stayed an insulator.
• The Interface is Key: The change happened right at the boundary (the interface) where the metal and the blanket touch.
• The "Order" Matters: The FePS₃ blanket has a special internal order (antiferromagnetism) that only becomes "active" and rigid when it gets cold. As it cools, this internal order "shakes hands" with the metal layer, changing the rules of how the magnetic push works.

The Takeaway

This paper shows that you don't need to run electricity through a material to change how a magnet behaves. You can just touch it with a special "quiet" material (an antiferromagnetic insulator).

• The Analogy: It's like putting a specific type of snow on a car tire. The snow doesn't power the car, but it changes how the tire grips the road. In this case, the "snow" (FePS₃) changes how the "engine" (electricity in Py) pushes the "wheel" (magnetization).

Why does this matter?
This gives engineers a new tool. Instead of just making things smaller, they can now tune how magnetic memory works just by changing the temperature or choosing the right "blanket" material. This could lead to computers that are faster, use less energy, and are much harder to crash.

In a nutshell: By stacking a special magnetic "insulator" on top of a metal, the scientists discovered they could turn a weak magnetic push into a super-strong one that even flips its direction when cold, all without any electricity flowing through the insulator. It's a new way to control magnets using "ghostly" magnetic connections.

Technical Summary

1. Problem Statement

The electrical manipulation of magnetization via current-induced spin–orbit torques (SOTs) is a critical mechanism for developing non-volatile, energy-efficient spintronic memory devices. While bulk spin–orbit effects (like the Spin Hall Effect) and interfacial effects (like the Rashba–Edelstein effect) are well-studied in metallic systems, the role of antiferromagnetic (AFM) insulators in modulating SOTs remains an active area of research.

• The Gap: There is a need to understand how layered van der Waals (vdW) AFM insulators, specifically FePS3, influence the symmetry (field-like vs. damping-like) and efficiency of torques when interfaced with ferromagnets (FM).
• The Challenge: Distinguishing whether torque modulation arises from bulk charge transport through the AFM layer or from interfacial magnetic coupling, especially given that AFM insulators do not conduct charge current.

2. Methodology

The researchers fabricated and characterized Py/FePS3 bilayer devices (Py = Permalloy/NiFe; FePS3 = an insulating layered AFM).

• Device Fabrication:
  • Substrate: Si/SiO2 (300 nm).
  • Ferromagnet: A 10 nm thick Py layer was deposited via thermal evaporation onto Hall-bar structures (2 × 30 µm² channel).
  • Antiferromagnet: FePS3 flakes were mechanically exfoliated from bulk crystals and deterministically transferred onto the Py Hall bars using a micromanipulator at 100°C to ensure clean, atomically sharp interfaces.
• Measurement Technique:
  • Harmonic Hall Measurements: A low-frequency sinusoidal current ($I = I_0 \sin(\omega t)$) was applied.
  • Signal Detection: Both first-harmonic ($1\omega$) and second-harmonic ($2\omega$) transverse Hall voltages were measured using a lock-in amplifier.
  • Angular Dependence: An in-plane magnetic field was rotated 360° to analyze the angular dependence of the Hall resistance.
  • Separation of Torques: The second-harmonic signal was fitted to a standard harmonic model to separate contributions from:
    • Field-like torque ($B_{FL}$): Associated with $\cos(2\phi)\cos(\phi)$ symmetry.
    • Damping-like torque ($B_{DL}$): Associated with $\cos(\phi)$ symmetry.
    • Thermoelectric artifacts: Longitudinal measurements were used to subtract effects like the Anomalous Nernst Effect (ANE).
• Temperature Control: Measurements were conducted in a closed-cycle cryostat from room temperature (290 K) down to 5 K to probe the evolution of torques relative to the magnetic ordering of FePS3.

3. Key Contributions

• Isolation of Interfacial Effects: The study successfully demonstrates that significant torque modulation can occur in an insulating AFM/FM heterostructure without bulk charge current flow through the AFM layer.
• Decoupling Torque Symmetries: The work reveals a unique decoupling where the field-like torque is strongly enhanced and temperature-dependent, while the damping-like torque remains negligible.
• Sign Reversal Phenomenon: The authors report a clear sign reversal of the field-like torque efficiency upon cooling, a phenomenon directly correlated with the antiferromagnetic ordering of FePS3.

4. Key Results

A. Transport and Static Magnetic Properties

• Insulating Nature Confirmed: First-harmonic Hall measurements (Planar Hall Effect and Anomalous Hall Effect) for Py/FePS3 bilayers were nearly identical to Py single-layer devices. This confirms that charge current flows almost exclusively through the metallic Py layer, and the FePS3 layer acts as an insulator with negligible shunting.
• Magnetic Ordering: SQUID magnetometry identified the Néel temperature ($T_N$) of FePS3 at approximately 118 K.

B. Torque Efficiency and Temperature Dependence

• Field-Like Torque ($B_{FL}$):
  • Enhancement: The Py/FePS3 bilayer exhibited a significantly higher field-like torque efficiency compared to the Py reference device across all temperatures.
  • Sign Reversal: As temperature decreased, the field-like torque efficiency underwent a sign reversal (from negative to positive) around 250 K.
  • Non-monotonic Behavior: Below $T_N$ (118 K), the torque magnitude showed a complex, non-monotonic evolution, increasing significantly as the temperature approached 5 K.
• Damping-Like Torque ($B_{DL}$):
  • The damping-like torque contribution was found to be negligible within experimental uncertainty for both the bilayer and the reference device across the entire temperature range.

C. Mechanism Analysis

• The distinct temperature dependence and sign reversal observed only in the bilayer (and not in the Py reference) rule out intrinsic Py effects or bulk transport mechanisms.
• The correlation between the torque evolution and the AFM ordering of FePS3 suggests the mechanism is driven by interfacial spin–orbit coupling, orbital hybridization, and exchange interactions at the vdW interface.

5. Significance and Implications

• Active Role of AFM Insulators: The study challenges the view of AFM insulators as passive spacers. Instead, it proves they actively participate in spin conversion and torque generation through interfacial magnetic correlations.
• Torque Engineering: The ability to tune the symmetry (enhancing field-like while suppressing damping-like) and sign of the torque via temperature and magnetic ordering offers a new degree of freedom for designing spintronic devices.
• Energy Efficiency: Since the effect relies on interfacial coupling rather than bulk charge injection into the AFM, it suggests a pathway for low-power spintronic devices where the AFM layer modulates the FM layer's dynamics without dissipative charge currents.
• Fundamental Physics: The results provide experimental evidence supporting theoretical predictions that antiferromagnetic insulators can generate predominantly field-like torques via interfacial exchange coupling, distinct from the bulk spin-Hall mechanisms found in heavy metals.

In conclusion, this work establishes FePS3/Py as a robust platform for studying and engineering interfacial spin–orbit torques, highlighting the potential of van der Waals antiferromagnetic insulators to control magnetic switching symmetry and efficiency in next-generation memory technologies.