How Does MRAM Use Electron Spin to Store Non-Volatile Bits?

Spintronic devices such as Magnetoresistive RAM store classical binary data by exploiting the intrinsic spin of electrons rather than relying on electric charge stored in capacitors. By aligning the magnetic orientation of ferromagnetic layers inside a magnetic tunnel junction, these devices represent ones and zeros as distinct electrical resistance states. Because magnetic orientation persists without continuous electrical power, spintronic memory achieves true non-volatility alongside high speed and endurance.

From Electric Charge to Electron Spin

Conventional memory technologies, such as Dynamic RAM (DRAM) and Flash, encode binary information by trapping or releasing electric charge. DRAM stores charge in tiny capacitors that leak over time, requiring constant refreshing every few milliseconds. Flash memory traps charge on a floating gate, which provides non-volatility but suffers from slower write speeds and eventual wear-out due to high-voltage tunneling.

Spintronics, or spin-transport electronics, takes a different approach by utilizing the quantum mechanical property of electron spin. An electron possesses an intrinsic angular momentum that gives it a tiny magnetic moment, commonly conceptualized as either "spin-up" or "spin-down." In normal conductors, these spin states are evenly balanced. However, in ferromagnetic materials like iron, cobalt, or nickel, an imbalance exists between spin-up and spin-down electrons, creating a net spin polarization.

The Magnetic Tunnel Junction (MTJ)

The fundamental building block of Magnetoresistive RAM (MRAM) is the Magnetic Tunnel Junction (MTJ). An MTJ consists of three primary layers:

  1. The Pinned (Reference) Layer: A ferromagnetic layer whose magnetic orientation is fixed in a single direction, often stabilized by an adjacent antiferromagnetic layer.
  2. The Tunnel Barrier: An ultra-thin insulating layer, typically made of magnesium oxide (\(\text{MgO}\)), just a few nanometers thick.
  3. The Free Layer: A second ferromagnetic layer whose magnetic orientation can be switched between two stable states—either parallel or antiparallel to the pinned layer.

Reading Bits via Tunnel Magnetoresistance

Data is read from an MTJ by passing a small sensing current through the stack and measuring its electrical resistance. Electrons quantum-mechanically tunnel through the thin \(\text{MgO}\) barrier, and the probability of tunneling depends on the relative alignment of the two ferromagnetic layers—a phenomenon known as Tunnel Magnetoresistance (TMR).

  • Parallel State (Low Resistance, Logic "0"): When the free layer's magnetization is aligned in the same direction as the pinned layer, majority-spin electrons easily find available states to tunnel into, resulting in low electrical resistance (\(R_P\)).
  • Antiparallel State (High Resistance, Logic "1"): When the free layer's magnetization is opposite to the pinned layer, fewer matching spin states are available on the other side of the barrier, impeding tunneling and producing high electrical resistance (\(R_{AP}\)).

The contrast between these two states is quantified by the TMR ratio:

\[TMR = \frac{R_{AP} - R_P}{R_P}\]

A higher TMR ratio makes it easier and faster for sense amplifiers to distinguish between a stored 0 and a stored 1.

Writing Bits with Spin-Transfer Torque

Early MRAM designs used external magnetic fields generated by nearby current-carrying wires to flip the free layer, which limited how densely cells could be packed. Modern MRAM overcomes this using Spin-Transfer Torque (STT-MRAM).

In STT-MRAM, a larger write current is passed directly through the MTJ. As electrons travel through the pinned layer, they become spin-polarized. When these spin-polarized electrons cross the tunnel barrier and enter the free layer, they transfer their angular momentum to the local magnetic moments. If the current exceeds a critical threshold, this torque is strong enough to flip the free layer's magnetization into the desired parallel or antiparallel orientation.

Advantages in Classical Computing

Because the magnetic state of the free layer is determined by atomic-scale magnetic interactions rather than trapped electrons, it does not leak away when power is removed. This gives MRAM inherent non-volatility, instant-on capability, and immunity to radiation-induced charge upsets. Furthermore, because writing involves reorienting magnetic moments rather than moving atoms or forcing charge through thick oxides, STT-MRAM offers near-SRAM speeds and virtually unlimited write endurance, bridging the gap between fast working memory and permanent storage.