Theoretical study of orbital torque: Dependence on ferromagnet species and nonmagnetic layer thickness

This study presents a systematic theoretical investigation of orbital torque in Ti/FM and Cu/FM bilayers, revealing that the torque's dependence on the ferromagnetic species varies with the nonmagnetic metal source and originates from the bulk nonmagnetic layer, thereby offering microscopic insights for designing light-metal-based orbitronic devices.

The Gist

The Big Idea: A New Way to Flip a Switch

Imagine you have a tiny magnetic switch (like the memory in your hard drive or a future super-fast computer). To turn this switch "on" or "off," you usually need to push it with a magnetic force. Traditionally, scientists have used heavy, expensive metals (like Platinum or Tungsten) to generate this push. These heavy metals act like a powerful magnetizer, but they are costly and heavy.

Recently, scientists discovered a "lighter" way to do this using Orbital Torque (OT). Instead of pushing with a magnetic spin, they use the "orbit" of electrons—imagine electrons not just spinning like tops, but also running in little circles (orbits) around the nucleus.

This paper asks a simple question: If we use light, cheap metals (like Titanium or Copper) to generate this "orbiting" push, does it work better with one type of magnetic switch (Cobalt) or another (Nickel)?

The Players in the Story

1. The Heavy Hitters (The Source): The researchers tested two light metals: Titanium (Ti) and Copper (Cu). Think of these as the "engines" that generate the orbital current.
2. The Targets (The Switches): They tested two magnetic metals: Cobalt (Co) and Nickel (Ni). These are the things getting pushed.
3. The Goal: To see which engine works best with which switch to create the strongest "push" (torque).

The Analogy: The Delivery Truck and the Warehouse

Imagine the Orbital Current is a delivery truck carrying packages (angular momentum) from a factory (the light metal) to a warehouse (the magnetic metal).

• The Spin Hall Effect (Old Way): The truck drives on a road that requires a special, heavy-duty license (Strong Spin-Orbit Coupling). Only heavy trucks (heavy metals) can do this.
• The Orbital Torque (New Way): The truck drives on a regular road. It doesn't need a special license to start the journey. However, once it arrives at the warehouse, the warehouse needs a special crane (Spin-Orbit Coupling) to unload the packages and turn them into a useful push.

What They Discovered

The researchers built a super-computer model to simulate these trucks and warehouses. Here is what they found, which was surprisingly tricky:

1. The "Titanium" Engine (Ti)

When they used Titanium as the engine:

• The Result: It worked best with Nickel.
• The Analogy: Think of Titanium as a high-speed race car. It delivers the packages so fast and efficiently that the Nickel warehouse (which has a slightly better crane) can unload them perfectly. The Cobalt warehouse was a bit slower at unloading, so the total push was weaker.
• Why? Nickel is slightly better at converting the "orbit" packages into a "spin" push.

2. The "Copper" Engine (Cu)

When they switched to Copper as the engine:

• The Result: The trend flipped! Now, Cobalt was the winner, and Nickel was weaker.
• The Analogy: Imagine Copper is a different kind of vehicle, maybe a bumpy off-road truck. The way it delivers the packages is different. It turns out the Cobalt warehouse is actually better at catching these specific bumpy deliveries. Nickel, which was great for the race car, wasn't as good at handling the off-road truck.
• The Lesson: There is no "one size fits all" rule. Just because Nickel is great with Titanium doesn't mean it will be great with Copper. It depends entirely on how the two materials "shake hands" at the interface.

3. The "Bulk" Secret (Thickness Matters)

The researchers also checked how thick the engine layer needed to be.

• The Finding: The "push" didn't just happen at the surface where the two metals touch. It came from the entire thickness of the light metal layer, sometimes extending over 10 nanometers (which is huge in the atomic world!).
• The Analogy: It's not just the driver at the front of the truck pushing the warehouse. It's as if the entire truck is vibrating and pushing. This means you can't just make the engine layer super thin; you need a certain amount of "bulk" material to get the full effect.

Why This Matters

1. No More Heavy Metals: This proves we can build efficient magnetic devices using cheap, light metals like Titanium and Copper, which are easier to manufacture and less toxic.
2. Design Rules: The biggest takeaway is that you can't just pick the "best" magnetic material (like Nickel) and assume it will work with any light metal. You have to match them carefully. A combination that works for one engine might fail with another.
3. Future Tech: This helps engineers design "Orbitronics"—a new generation of electronics that uses electron orbits instead of just spins. This could lead to faster, more energy-efficient computers and memory storage.

In a Nutshell

This paper is like a mechanic's guide for building a new type of engine. They found that Titanium loves Nickel, but Copper prefers Cobalt. They also learned that the engine needs to be thick enough to do its job. This discovery helps us stop guessing and start designing better, greener, and cheaper magnetic devices.

Technical Summary

1. Problem Statement

The manipulation of magnetization via Orbital Torque (OT) has emerged as a promising alternative to Spin-Orbit Torque (SOT) for energy-efficient devices, particularly because it can utilize light nonmagnetic metals (NMs) with weak spin-orbit coupling (SOC). While materials like Titanium (Ti) and Copper (Cu) are extensively studied as sources of orbital currents (via the Orbital Hall Effect, OHE, or Orbital Rashba-Edelstein effect), several critical gaps remain in the theoretical understanding of OT:

• Ferromagnet (FM) Dependence: Experimental results show conflicting trends regarding which 3d ferromagnet (Co vs. Ni) yields a larger torque. For instance, Ti/Ni often shows larger OT than Ti/Co, but Cu-based systems show different behaviors. It is unclear if this dependence is universal or specific to the NM source.
• Origin of Torque: There is ambiguity regarding whether the torque originates from the bulk NM, the NM/FM interface, or self-induced SOT within the FM layer.
• Limitations of Existing Models: Previous theoretical studies often relied on simplified pictures based on individual bulk properties (e.g., bulk orbital Hall conductivity) or were limited by the computational cost of fully ab initio methods for large systems, preventing the study of thickness-dependent effects up to tens of nanometers.

2. Methodology

The authors employed a systematic theoretical approach combining ab initio electronic structure calculations with realistic tight-binding models to overcome size limitations.

• Electronic Structure & Wannier Functions:

  • Bulk band structures for NMs (Ti, Cu) and FMs (Co, Ni) were calculated using the FLEUR code (Full-potential Linearized Augmented Plane-Wave) with the PBE functional.
  • Wannier90 was used to construct tight-binding models. Instead of maximally localized Wannier functions, the authors used 18 projected Wannier functions (s, p, and d orbitals) per spin to ensure accuracy for metallic systems.
  • To simulate realistic bilayers, the lattice constants of the FMs were adjusted to match the adjoining NM (fcc(111) orientation), and magnetization was aligned along the [111] direction.

• System Modeling:

  • Bilayer structures (NM/FM) were modeled as 2D hexagonal lattices.
  • The system size was scaled up to 50 atomic layers ($N_{NM} + N_{FM}$), allowing for the investigation of thickness effects up to $\approx 10$ nm, which is beyond the reach of standard ab initio transport calculations.

• Torque Calculation:

  • The study utilized linear-response theory within the Kubo formalism.
  • The damping-like torque (the primary component for switching) was calculated using the interband contribution ($t^{inter}_{yx}$), which is associated with Berry curvature mechanisms.
  • Crucially, the authors calculated the exchange torque ($\hat{T}_{XC}$) directly, which describes the transfer of angular momentum from conduction electrons to the local FM spin moment. This avoids ambiguities associated with defining non-conserved orbital current operators.
  • Disorder was modeled via a constant energy broadening parameter ($\Gamma$), representing quasiparticle lifetime.

3. Key Contributions

• Systematic Comparison of NM/FM Combinations: The first theoretical study to quantitatively compare OT in both Ti/FM and Cu/FM bilayers for Co and Ni FMs using identical, realistic models.
• Decoupling Bulk vs. Interface Effects: By varying the NM thickness ($N_{NM}$) up to 40 layers, the study isolates the bulk contribution of the NM from interfacial effects.
• Re-evaluation of FM Dependence: The work challenges the notion of a universal "best" FM for OT, demonstrating that the optimal FM depends on the specific NM orbital current source.
• Quantification of Crystal Field Torque: The study explicitly calculates the crystal field torque ($\hat{T}_{CF}$), showing that while a large portion of orbital angular momentum is quenched by the lattice, a significant exchange torque still remains.

4. Key Results

A. Ferromagnet Species Dependence (Co vs. Ni)

• Ti/FM Systems: The torque is significantly larger for Ni than for Co ($t_{yx} \approx -0.33 e a_0$ for Ti/Ni vs. $-0.18 e a_0$ for Ti/Co). This aligns with experimental observations and is attributed to Ni's stronger SOC, which facilitates more efficient orbital-to-spin conversion.
• Cu/FM Systems: The trend reverses. In pristine Cu/FM bilayers, Co produces a larger torque than Ni.
  • Implication: The FM dependence of OT is not universal. In Cu systems, factors other than SOC strength (such as orbital texture matching or interface transparency) dominate the torque efficiency.
  • Self-Induced SOT Check: The authors ruled out self-induced SOT as the cause of this reversal, as Ni has a much larger spin Hall conductivity than Co, yet yields a smaller torque in Cu/FM.

B. Nonmagnetic Layer Thickness Dependence

• Bulk Origin Confirmed: For both Ti and Cu systems, the torque magnitude increases with NM thickness and saturates only at large thicknesses ($>10$ nm). This confirms that the OT originates primarily from the bulk NM (via the OHE) rather than just the interface.
• Characteristic Lengths:
  • Ti/FM: The torque saturates slowly, consistent with experimental observations of long-range orbital transport (tens of nm).
  • Cu/FM: A distinct behavior is observed in Cu/Ni. In the clean limit (low disorder), the bulk Cu contribution to the torque has the opposite sign compared to Cu/Co. This suggests that the effective orbital Hall conductivity measured in torque experiments may differ from the theoretical bulk orbital Hall conductivity due to crystal field effects.

C. Mechanism and Disorder

• Crystal Field vs. Exchange Torque: The study found that the crystal field torque ($\hat{T}_{CF}$) is much larger than the exchange torque ($\hat{T}_{XC}$), indicating that most orbital angular momentum is quenched by the lattice. However, the remaining exchange torque is still sizable.
• Disorder Effects: Increasing disorder (higher $\Gamma$) generally reduces the torque magnitude. In Cu/Ni, the torque dependence on disorder is non-monotonic due to sharp energy dependencies near the Fermi level.

5. Significance and Conclusion

• Design Guidelines for Orbitronics: The results provide critical guidance for designing light-metal-based orbitronic devices. There is no "one-size-fits-all" ferromagnet; the choice of FM (Co vs. Ni) must be optimized based on the specific NM source (Ti vs. Cu).
• Beyond Bulk Properties: The study demonstrates that OT cannot be predicted solely by the bulk orbital Hall conductivity of the NM or the SOC of the FM. The material combination and interfacial/orbital hybridization play decisive roles.
• Experimental Validation: The theoretical prediction of a bulk-originated, long-range OT in Cu/FM systems (even without oxidation) supports recent experimental findings of orbital currents in Cu.
• Methodological Advance: By successfully modeling systems up to 10 nm using tight-binding models derived from ab initio data, the authors bridge the gap between microscopic electronic structure and macroscopic device performance, offering a robust framework for future theoretical studies in spintronics and orbitronics.