Overview

Conducted computational physics research at the National Research and Innovation Agency (BRIN) as part of my undergraduate thesis and professional experience, forming the basis for an upcoming scientific manuscript. The work examined how nanowire dimensions and Dzyaloshinskii-Moriya Interaction (DMI) configurations influence domain wall dynamics in spintronic data storage. I performed 468 micromagnetic simulations using OOMMF and analyzed position, velocity, and energy data, producing structured reports and visualizations that informed research insights and supported the manuscript preparation.

Technical Workflow

This project was conducted using a structured micromagnetic simulation workflow, designed to ensure physical validity, numerical stability, and reproducible data analysis. The overall process integrates physical modeling, numerical simulation, and quantitative post-processing.

1. Physical Modeling

The system was modeled within the continuum micromagnetic framework, where the magnetization dynamics are governed by the Landau–Lifshitz–Gilbert (LLG) equation enhanced with Dzyaloshinskii–Moriya Interaction (DMI). Energy contributions include exchange, perpendicular magnetic anisotropy (PMA), demagnetization, and DMI.

2. Geometry and Material Configuration

Co₄₀Fe₄₀B₂₀ nanowires with perpendicular magnetic anisotropy were modeled with fixed length and varying widths and thicknesses. The simulation cell size was chosen to be smaller than the magnetic exchange length, ensuring numerical accuracy.

Figure 1. Schematic of the Co₄₀Fe₄₀B₂₀ nanowire geometry used in the micromagnetic simulations.

The nanowire has a fixed length of L = 10 µm, with varying width (W) and thickness (T) under perpendicular magnetic anisotropy (PMA). The color scale represents the out-of-plane magnetization component (Mz), where red and blue correspond to +Mz and −Mz states, respectively. A transverse domain wall initially separates oppositely magnetized domains. Nanosecond-scale current pulses (I_p = 1 ns) are injected along the x-direction to drive current-induced domain wall motion, while the Cartesian coordinate system used in the simulations is shown for reference.

3. Simulation Execution

All simulations were performed using OOMMF. The initial domain wall was defined as a Bloch wall and relaxed to its ground state. Due to DMI, the domain wall naturally transformed into a chiral Néel wall before current-driven dynamics were applied.

4. Current-Induced Dynamics

Nanosecond-scale current pulses were injected along the nanowire axis. Spin-transfer torque was incorporated into the LLG equation to drive domain wall motion. The domain wall behavior was analyzed during and after the current pulse to evaluate propagation stability.

Data Processing Pipeline

The data processing workflow was designed to be quantitative, reproducible, and minimally dependent on visual interpretation. All results were derived directly from numerical magnetization data.

1. Simulation Output

OOMMF generated time-resolved magnetization files (.omf) containing spatial distributions of the magnetization components (Mx, My, and Mz), along with auxiliary output files (.odt) primarily used for energy analysis.

Figure 3. Raw OOMMF output files displayed using Notepad++ (.omf).

Figure 4. Raw OOMMF output files displayed using OriginPro (.odt).

2. Data Conversion and Extraction

OOMMF magnetization files (.omf) were converted into tabular format (.odt) to enable precise identification of domain wall positions at specific times.

Although OOMMF directly generates .odt files during simulations, these outputs are primarily suitable for energy analysis and cannot be reliably used to determine domain wall position or velocity. During current-driven simulations, domain wall nucleation may occur, leading to the formation of additional domain walls that obscure the motion of the original one. To avoid this ambiguity, domain wall positions were extracted from spatial magnetization profiles at selected time steps (1.6 ns to 3.4 ns), enabling accurate time-resolved analysis.

3. Domain Wall Position Identification

The domain wall position was determined from the zero-crossing of the Mz component along the nanowire axis. This method provides higher robustness than direct snapshot inspection.

Figure 5. Domain wall position comparison under current injection for different nanowire geometries.

Panels (a–c) show DW positions for nanowire widths W = 50 nm, 100 nm, and 150 nm, respectively when T = 1 nm. Panels (c–e) show DW positions for nanowire widths W = 50 nm, 100 nm, and 150 nm, respectively when T = 2 nm. All simulations were performed at D = 0.6 mJ/m² and u = 1000 m/s.

4. Velocity and Energy Analysis

Domain wall velocity was obtained from linear fits of position–time data during the current-on regime. Energy evolution—including exchange, anisotropy, demagnetization, and DMI—was analyzed to identify stable and transient dynamic regimes.

Figure 6. Domain wall velocity comparison under current injection for different nanowire geometries

Panels (a–b) compare DW velocities for thicknesses T = 1 nm and 2 nm. Panels (c–e) show DW velocity for nanowire widths W = 50 nm, 100 nm, and 150 nm, respectively. All simulations were performed at all D variations and spin drift velocity u = 1000 m/s.

Figure 7. Energy Evolution Graphs Showing Exchange, Anisotropy, Demagnetization, DMI, and Total Energy During Current-Induced Domain Wall Motion with D=0 mJ/m² (T=1 nm & W=50 nm).

5. Visualization and Validation

ImageJ was used to validate domain wall structures (Bloch versus Néel), while Origin was employed to generate consistent, publication-ready plots for final analysis and reporting.

Figure 8. The Animation of Magnetization process during the ground-state simulation of a nanowire with D = 0.6 mJ/m². The magnetization configuration is obtained from micromagnetic simulation output and visualized using ImageJ.

6. Data Tabulation and Kinematic Analysis

Extracted domain wall positions were organized into tabulated datasets to calculate displacement and velocity as functions of time. Spreadsheet-based processing (Microsoft Excel) was used for preliminary data organization and linear fitting, while final plots and validation were performed using Origin to ensure consistency and publication-quality visualization.

Introduction

This concept was first proposed in the early 2000s by Dr. Stuart Parkin, a renowned physicist and researcher at the IBM Almaden Research Center. His fundamental paper on this technology was published in the journal Science in 2008. Parkin sought to create a "solid-state" device—one with no moving parts—that could rival the high storage density of magnetic disk platters.

This is my first research project with BRIN and Dr. Candra Kurniawan, M.Si., and I’m so excited about the topic. It is closely related to quantum and solid-state physics. This research gives me more time and opportunities to learn and improve my understanding.

HDDs and SSDs are non-volatile storage devices that retain data without power. HDDs use spinning platters and are slower than SSDs due to their mechanical components, which limit data access speed. In contrast, SSDs utilize semiconductor-based flash memory, offering significantly higher speed and reliability, albeit at a higher cost [1], [2]. However, SSDs have a limited lifespan due to the Program/Erase (P/E) cycle, although techniques such as wear leveling can extend their operational durability.

Current data storage technologies primarily rely on electrical charge, such as in silicon-based transistors. However, this approach faces fundamental limitations in terms of miniaturization, heat dissipation, and power efficiency as device dimensions approach the nanoscale. Spintronics provides an alternative paradigm by exploiting the electron’s spin degree of freedom instead of charge, enabling faster switching, lower energy consumption, and improved non-volatility [3], [4].

Racetrack Memory is an emerging spintronic storage technology in which digital information is represented by magnetic domains separated by domain walls that move along nanoscopic tracks [5]. A single racetrack can store multiple bits of data, enabling very high storage density. Domain walls are shifted using nanosecond-scale current pulses through mechanisms such as spin-transfer torque (STT) or spin–orbit torque (SOT), allowing high-speed data manipulation [6], [7]. In advanced systems such as synthetic antiferromagnetic racetracks, domain wall velocities can reach the order of ~1 km/s, demonstrating the potential for ultra-fast data transfer. Furthermore, optimized racetrack configurations have the potential to achieve sub-nanosecond operation speeds, making them denser and more efficient than conventional SRAM while maintaining non-volatility [8], [9].

Micromagnetic simulations play a crucial role in the development of racetrack memory, as they enable researchers to model and analyze domain wall dynamics without the need for costly physical prototypes. These simulations are typically based on the Landau–Lifshitz–Gilbert (LLG) equation, which describes the time evolution of magnetization under various physical interactions [13], [14]. Through simulation, researchers can optimize material properties, improve energy efficiency, enhance device durability, and refine writing and reading mechanisms. This approach significantly accelerates the development of racetrack memory as a faster, more efficient, and longer-lasting alternative to conventional storage technologies such as HDDs and SSDs [15], [16].

Read/Write Process in Racetrack Memory

Conventional HDDs store data in magnetic domains formed by groups of microscopic grains with aligned magnetization [1, 2]. Racetrack Memory uses thin nanowires where bits are represented by sequential magnetic domains separated by domain walls (DWs) [5, 17].

Data is written by injecting a current through an injector, using spin-transfer torque (STT) to switch atomic magnetization and create new bits [11, 18, 19]. Domains are then shifted along the wire via current-induced domain wall motion (CIDWM), enabling rapid, mechanical-free data movement [6, 7].

Reading is done with magnetoresistance-based sensors, such as magnetic tunnel junctions (MTJs), which detect resistance changes depending on domain alignment and translate them into binary values [5, 10]. This allows high-speed, efficient data storage compared to conventional HDDs [5, 9].

Domain Wall Motion

Engineering the motion of domain walls (DW) is a key factor in the development of Racetrack Memory, as it enables high-density data storage in three-dimensional structures, high-speed operation, and low energy consumption due to the absence of moving mechanical parts [5], [8]. Moreover, the dynamics of domain walls exhibit similarities to biological synaptic mechanisms, opening opportunities for applications in spintronic-based neuromorphic computing [12].

Meanwhile, micromagnetic simulations using software such as OOMMF are employed to model magnetic behavior at microscopic scales that are difficult to observe directly in laboratory experiments. These simulations are based on the numerical solution of the Landau–Lifshitz–Gilbert (LLG) equation, which describes the dynamics of magnetization in ferromagnetic materials [13], [14].

Through these simulations, researchers can visualize domain wall dynamics as well as complex structures such as skyrmions on nanosecond timescales. This approach greatly facilitates the optimization of material design and spintronic devices efficiently before proceeding to costly physical fabrication stages [15], [16].

Reference