[Paper Review] Ultra fast bit addressing in a magnetic memory matrix with crossed wire write line geometry

H. W. Schumacher|arXiv (Cornell University)|Feb 10, 2005

Magnetic properties of thin films4 citations

TL;DR

This paper proposes an ultra-fast bit addressing scheme for magnetic random access memory (MRAM) using a crossed wire geometry, where sub-nanosecond magnetic field pulses induce controlled precession of the free layer's magnetization. By tuning pulse parameters to achieve half or full precessional turns, the method suppresses magnetization ringing and enables write clock rates exceeding 1 GHz, significantly enhancing MRAM speed and scalability.

ABSTRACT

An ultra fast bit addressing scheme for magnetic random access memories (MRAM) in a crossed wire geometry is proposed. In the addressing scheme a word of cells is programmed simultaneously by sub nanosecond field pulses making use of the magnetization precession of the free layer. Single spin simulations of the free layer dynamics show that the pulse parameters for programming an arbitrary word of the array can be chosen such that the magnetization of the cells to be written performs either a half or a full precessional turn during application of the programming pulse depending on the initial and final magnetization orientation of the addressed cells. Such bit addressing scheme leads to a suppression of the magnetization ringing in all cells of the memory array thereby allowing ultra high MRAM write clock rates above 1 GHz.

Motivation & Objective

• To address the need for faster, scalable magnetic memory architectures with high write speeds.
• To overcome limitations in conventional MRAM addressing, such as slow write rates and magnetization ringing.
• To develop a bit addressing scheme that enables simultaneous programming of multiple memory cells with sub-nanosecond pulses.
• To suppress magnetization ringing across the entire array through precise control of precessional dynamics.
• To achieve write clock rates above 1 GHz in a practical, scalable magnetic memory matrix.

Proposed method

• The method employs sub-nanosecond magnetic field pulses applied via crossed write lines to induce controlled precession in the free layer of magnetic tunnel junctions.
• Pulse parameters (amplitude and duration) are tuned based on initial and final magnetization states to achieve either half or full precessional turns.
• The approach leverages the dynamics of magnetization precession to ensure deterministic switching without overshoot or ringing.
• Single spin simulations are used to model free layer dynamics and validate pulse design across various initial and target magnetization orientations.
• The scheme enables simultaneous programming of an entire word line by ensuring all addressed cells reach their target state within one precessional cycle.
• The geometry allows for high spatial resolution and scalability, with minimal crosstalk due to precise field control.

Experimental results

Research questions

• RQ1 Can sub-nanosecond field pulses be used to achieve deterministic switching in magnetic memory cells with minimal ringing?
• RQ2 How can precessional dynamics be controlled to allow for both half and full precessional turns depending on the target magnetization state?
• RQ3 What pulse parameters are required to simultaneously program an entire word of cells without inducing magnetization ringing?
• RQ4 Can the proposed scheme achieve write clock rates exceeding 1 GHz in a scalable magnetic memory matrix?
• RQ5 How does the crossed wire geometry enable high-speed, low-crosstalk addressing in MRAM architectures?

Key findings

• The proposed addressing scheme enables write clock rates exceeding 1 GHz by suppressing magnetization ringing through controlled precession.
• Sub-nanosecond field pulses can reliably switch the magnetization of multiple cells simultaneously by achieving either half or full precessional turns.
• Pulse parameters are optimized based on initial and final magnetization states to ensure deterministic switching without overshoot.
• Single spin simulations confirm that the method effectively eliminates ringing across all cells in the array.
• The scheme is scalable and compatible with conventional crossed wire MRAM architectures, enabling high-speed operation.
• The method achieves full switching fidelity with minimal energy dissipation by leveraging natural precessional dynamics.