🔬 materials science

Optimizing spin-based terahertz emission from magnetic heterostructures

This paper presents a systematic theoretical analysis of spin-based terahertz emission from magnetic heterostructures using a superdiffusive spin-transport model to establish optimization guidelines regarding layer thickness, interface properties, and excitation protocols, ultimately revealing trade-offs between bandwidth and efficiency while proposing a double-pulse trilayer protocol for enhanced broadband emission.

Original authors: Francesco Foggetti, Francesco Cosco, Peter M. Oppeneer, Henri Jaffrés, Niloufar Nilforoushan, Juliette Mangeney, Sukhdeep Dhillon

Published 2026-07-13

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Francesco Foggetti, Francesco Cosco, Peter M. Oppeneer, Henri Jaffrés, Niloufar Nilforoushan, Juliette Mangeney, Sukhdeep Dhillon

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 have a tiny, super-fast flashlight that doesn't just shine light, but shoots out invisible "THz" waves—radiation that sits right between the radio waves in your phone and the infrared heat you feel from a fire. Scientists have figured out how to make these waves using a special sandwich of metal layers, but they've been struggling to tune the sandwich to make the waves as wide and powerful as possible.

In this study, the authors built a detailed computer simulation (a digital twin of the real world) to figure out exactly how to bake the perfect metal sandwich. They didn't just guess; they used a specific model called "superdiffusive spin transport" to track how tiny, hot electrons zoom around inside the metal after being hit by a laser.

The Perfect Sandwich Recipe
Think of the metal sandwich as a two-layer cake: a ferromagnetic layer (the "magnetic" part, like Cobalt) and a non-magnetic layer (the "heavy metal" part, like Platinum). When a super-fast laser pulse hits the cake, it wakes up "hot" electrons in the magnetic layer. These electrons are like a crowd of excited kids running out of a classroom. Because they are "spin-polarized," they run in a specific direction, creating a current.

When these running kids hit the second layer (the Platinum), they get converted into an electrical charge that shoots out a THz wave. The authors found that the thickness of the layers matters a lot:

The Magnetic Layer: Should be very thin, around 2 to 3 nm.
The Heavy Metal Layer: Should be about 5 to 6 nm thick.

If the heavy metal layer is too thick, the "kids" (electrons) get lost or run around too long, making the signal blurry and slow. If it's too thin, they bounce back into the first layer too quickly. The 5–6 nm sweet spot lets them run just long enough to create a sharp, wide burst of THz waves.

The Laser Pulse: Speed vs. Shape
The authors also tested how the laser itself changes the show. They found that using a shorter laser pulse (like a quick snap of a whip) makes the THz waves have a wider bandwidth (more variety in the frequencies). However, there's a cool twist: changing how fast the laser snaps does not change the main "note" or peak frequency of the wave. The peak frequency is determined by the shape and size of the metal sandwich itself, not by how fast you hit it.

The Walls of the Room
Imagine the metal sandwich is in a room. If the walls are mirrors (perfectly reflective), the electrons bounce back and forth, staying in the system longer. This makes the signal stronger (more power) but narrower in frequency. If the walls are black holes (spin sinks) that swallow the electrons, the signal is very short and wide (great bandwidth), but you lose a lot of energy because the electrons disappear. The authors suggest that if you want the widest possible signal, you should let the electrons escape (use "spin sinks"), but you have to accept that the signal won't be as loud.

The Double-Click Trick
Finally, the authors proposed a clever new idea: what if you hit the sandwich with two laser pulses instead of one? They simulated a three-layer sandwich (Magnetic/Heavy Metal/Magnetic) and hit both sides with lasers that are slightly out of sync.

Think of it like two people pushing a swing. If they push at the exact wrong time, they cancel each other out. But if they time it just right, they can squeeze the swing's motion into a super-short, super-fast burst. In their simulation, using two pulses with a specific delay created a signal that was even wider in bandwidth than the best two-layer sandwich. However, there's a catch: squeezing the signal this way makes it quieter. The energy efficiency drops, meaning you get a wider signal but a weaker one.

What They Didn't Find
The authors were careful to point out what doesn't work or what isn't the main driver. They noted that simply making the laser pulse more intense (brighter) doesn't change the shape or bandwidth of the wave; it just makes the whole thing louder. They also clarified that while other theories exist (like thermalized transport), their model, which tracks individual hot electrons moving ballistically and then diffusively, matched real-world experiments on a Co(2 nm)/Pt(4 nm) sandwich perfectly. This gives them confidence that their "electron running" model is the right way to understand these devices.

In short, to get the best THz emission, you need a thin sandwich, a quick laser snap, and maybe a double-tap with two lasers if you're willing to trade some volume for a wider frequency range. It's all about tuning the geometry and the timing to get the electrons to dance just right.

Problem Statement
While spin-based terahertz (THz) emitters, utilizing femtosecond laser excitation of ferromagnetic/nonmagnetic (FM/NM) heterostructures, offer a promising alternative to traditional semiconductor-based THz sources, the mechanisms for optimizing the amplitude and bandwidth of the emitted THz pulses remain poorly understood. Specifically, there is a lack of systematic guidelines regarding how layer thicknesses, interface properties, material choices, and pump laser parameters influence the THz emission profile. Furthermore, the proportionality between the THz electric field and the generating source (e.g., charge current vs. its time derivative) remains a subject of debate in the literature.

Methodology
The authors employ a theoretical framework based on the semiclassical, superdiffusive spin-transport model to simulate ultrafast demagnetization and spin-current generation in FM/NM heterostructures. Key methodological features include:

Superdiffusive Transport: The model tracks hot, spin-polarized electrons (majority and minority) with energy-dependent velocities and lifetimes. It accounts for ballistic transport in the first ~100 fs, transitioning to diffusive transport up to ~1 ps.
Energy-Dependent Spin Hall Effect (ISHE): A critical refinement in this work is the inclusion of the energy dependence of the spin Hall conductivity in the heavy metal (Pt). The model recognizes that the spin Hall angle is large for hot electrons near the Fermi energy but drops significantly for energies above 0.5 eV, requiring electrons to scatter and lose energy before contributing effectively to spin-to-charge conversion.
Interface and Boundary Conditions: The simulations incorporate spin-dependent transmission and reflection coefficients at the FM/NM interface and at the outer boundaries (substrate/capping layers).
THz Emission Calculation: The time-varying charge current, derived from the spin current via the inverse spin Hall effect, is used to calculate the emitted THz electric field. The authors adopt the proportionality $E(\nu) \propto \int dz J_c(z, \nu)$ , accounting for the effective time integration introduced by the parabolic mirror in typical THz detection setups.
Validation: The model is first validated against experimental data for a Co(2 nm)/Pt(4 nm) bilayer, showing good agreement in both peak frequency and bandwidth.

Key Results
The study systematically varies structural and excitation parameters to identify optimization guidelines for THz emission:

Laser Pulse Width: Reducing the pump pulse duration (FWHM) increases the THz bandwidth, as shorter pulses excite faster electron dynamics. However, the peak frequency ( $\nu_{max}$ ) is found to be largely independent of the pulse width.
Layer Thicknesses:
- Nonmagnetic (NM) Layer: Optimal bandwidth is achieved with thin NM layers (5–6 nm for Pt). Thicker layers lead to longer-lived currents and lower bandwidths. If the NM layer is too thin (<3 nm), hot electrons may re-enter the FM layer, reducing emission efficiency.
- Ferromagnetic (FM) Layer: Thin FM layers (2–3 nm) are optimal for bandwidth.
- Peak Frequency: The peak frequency shifts to lower values as the NM layer thickness increases, correlating with the longer temporal spread of the current in thicker layers.
Interface Properties:
- FM/NM Interface: Maximum bandwidth is achieved when the interface acts as a "filter" that allows high transmission for electrons moving from FM to NM (low reflection, $R_{Right}$ ) but high reflection for electrons attempting to return from NM to FM (high reflection, $R_{Left}$ ). This traps electrons in the NM layer, enhancing the spin-to-charge conversion duration.
- Trade-off: While optimizing interface reflection for bandwidth is possible, it comes at the cost of reduced total integrated power (energy efficiency).
Boundary Conditions: Lower reflectivity at the outer boundaries (acting as spin sinks) results in larger bandwidths and higher peak frequencies because the spin current is evicted from the system quickly, creating a shorter temporal pulse. However, this configuration significantly reduces the total emitted power compared to perfectly reflecting boundaries, where energy is conserved within the system.
Double-Pulse Trilayer Emitter: The authors propose a novel FM/NM/FM trilayer excited by two time-delayed laser pulses. By tuning the delay ( $\Delta t$ ) to approximately the pulse FWHM, the two counter-propagating spin currents interfere destructively, effectively "squeezing" the signal in time. This results in a significantly broader bandwidth than standard bilayers. However, this comes with a trade-off: the total integrated power is approximately one order of magnitude lower than that of an optimized bilayer.

Significance and Claims
The paper claims to provide a rigorous theoretical basis for optimizing spintronic THz generation. By validating the superdiffusive model against experimental data, the authors establish a reliable tool for predicting THz emission profiles. The primary contribution is the establishment of specific design guidelines:

Thin NM layers (5–6 nm) and FM layers (2–3 nm) are optimal for bandwidth.
Interface engineering should favor high transmission into the NM layer and high reflection back from it.
External boundaries should ideally act as spin sinks to maximize bandwidth, though this reduces signal amplitude.
A double-pulse excitation protocol on a trilayer system offers a pathway to broadband emission, albeit with reduced energy efficiency.

The authors conclude that their results contribute to the ongoing discussion regarding the physical origin of spintronic THz emission, supporting the superdiffusive spin-transport model as a suitable description for these ultrafast processes. They emphasize that their findings offer actionable guidelines for experimentalists aiming to maximize THz electric fields and bandwidths in spintronic devices.

More like this