Ni₄W Memory Breakthrough Enables Magnet-Free Switching

The latest technological advancements, ranging from big data to artificial intelligence (AI) to the Internet of Things (IoT), collect and process tons of data. For that, they need high power efficiency, low-latency data transfer, and high-speed processing. 

Here, advances in high-performance computing (HPC) are crucial in enhancing data processing capabilities, for which they take advantage of parallel processing, powerful hardware, and sophisticated software.

However, memory access tends to be the bottleneck, as such creating a strong need for memory technology that’s compatible with these demands.

Memory technology enables the access, storage, and changing of data. The information here is represented by collections of bits, with each bit being either zero or one (alternatively, true or false).

Ideally, memory reads and writes in negligible time, consumes little power, occupies insignificant space, and retains its stored value indefinitely. But of course, in practice, no memory technology fulfills these ideal conditions. Different technologies have their own strengths and weaknesses, with there being no one best memory technology.

Memory technology is primarily divided into two categories:

• Volatile
• Non-volatile

This is based on the cell design. Cells are the basic units of memory, actually an ‘array’ of memory ‘cells’, where each cell holds one bit of data, and the characteristics of a single cell reflect those of the overall array.

A volatile memory is one that works as long as it is powered and loses the stored information when power is turned off. Hence, this type of memory can be used to store data temporarily.

A non-volatile memory, in contrast, retains its stored value even when power is removed. For this particular type of memory, sophisticated semiconductor technology is applied, as it is more challenging to manufacture and difficult to write to electronically.

With the increasing availability of more sophisticated memory technology on the market, the distinction between these two memory categories is becoming increasingly blurry.

Breakthroughs in Memory Technology

Given the importance of memory technology for the operation and performance of various electronic devices and systems, as it allows computers and other devices to store and retrieve information needed for use, researchers have continually explored new ways to make it more efficient.

Over the years, several breakthroughs have revolutionized tech. With the goal of overcoming the limitations of current RAM and storage solutions, ongoing research is driving faster, more energy-efficient computing and enabling new applications in areas such as AI and neuromorphic computing.

PCM and Low-Power Innovations

Some of the key advancements in this area include new PCM (Phase Change Memory) materials for creating a single memory type that combines the speed of RAM with the non-volatility of flash storage.

In the PCM realm, late last year, scientists discovered¹ a new technique to lower the energy requirements of PCM by up to 1 billion times.

“One of the reasons why phase-change memory devices haven’t reached widespread use is due to the energy required,” said author Ritesh Agarwal, a professor of materials science and engineering at Penn Engineering, which means the potential of the findings of this new technique is “tremendous” for designing low-power memory devices.

This particular discovery relies on the unique properties of indium selenide (In2Se3), a semiconductor material exhibiting both piezoelectric (materials that physically deform when exposed to an electric charge) and ferroelectric (materials that can generate an internal electric field without requiring an external charge) characteristics. 

When indium selenide was exposed to a continuous current, the researchers observed that sections of it amorphized, disrupting the crystalline structure and opening “up a new field on the structural transformations that can happen in a material when all these properties come together.”

Multiferroics & Efficient Data Storage

Multiferroic materials that exhibit both ferroelectric and ferromagnetic properties for non-destructive data storage are also being explored by researchers. 

One such material is cobalt-substituted BiFeO3 (BiFe0.9Co0.1O3, BFCO), which exhibits strong magnetoelectric coupling, allowing an energy-efficient way of writing data. Last year, researchers from the Tokyo Institute of Technology developed² BFCO nanodots with single ferroelectric and ferromagnetic domains.

This year, the researchers made progress³, building on the research to demonstrate real-world switching functionality in oriented thin films. The dynamic control demonstrates actual electric‐field‐driven magnetization switching in a more device-compatible format.

Ferroelectric Solutions & New Memory Designs

Chiplet technology is another approach where multiple smaller chips, or chiplets, are mounted on a substrate that connects them, enabling higher memory bandwidth and density. Meanwhile, advancements in NAND flash and DRAM technologies continue towards smaller process nodes, with a focus on increasing bandwidth and power efficiency.

While NAND flash memory is one of the most prevalent technologies for mass data storage due to its ability to store more data in the same area by stacking cells in a 3D structure, it depends on charge traps to store data, which means higher operating voltages and slower speeds.

A promising solution to this is hafnia (Hafnium oxide)-based ferroelectric memory, but the challenge with them is limited memory for data storage.

A team from POSTECH addressed this issue⁴ by doping the ferroelectric materials with aluminum, which created high-performance ferroelectric thin films. Additionally, they employed an innovative metal-ferroelectric-metal-ferroelectric-semiconductor (MFMFS) structure, rather than the typical MFS structure.

This allowed them to successfully control voltage in every layer by fine-tuning factors like the thickness and area ratio of the layers. As a result, the team achieved a memory window surpassing 10 volts (V), as opposed to a mere 2V in conventional devices.

Spin-Orbit Torque and Magnetic Memory Evolution

Even quantum computing is seeing a lot of traction as an emerging technology paving the way for more powerful, efficient, and versatile computing devices of the future.

Then there’s energy-efficient Spin-Orbit Torque Magnetic Random Access Memory (SOT-MRAM), where electrical currents are used to switch magnetic states and achieve high speed and low power consumption.

Earlier this year, a team of researchers from the JGU Institute of Physics shared their innovation⁵ based on SOT-MRAM, which shows potential to reduce energy consumption by over 50% and boost efficiency by 30%. It also reduces the input current needed for magnetic switching to store the data by 20% and achieves a thermal stability that ensures data storage longevity.

Photonic and Magneto-Optical Memory

Controlling optical memory chips by light and magnets is yet another way to improve processing speed and efficiency.

In one development, scientists designed a programmable photonic latch⁶ built on a silicon photonic platform. Each memory unit in the system is driven by its own light source, allowing multiple units to function independently. This prevents signal degradation that optical power loss can cause, making the architecture more scalable for larger systems.

Farshid Ashtiani of Nokia Bell Labs explained the potential:

“Large language models like ChatGPT rely on massive amounts of simple mathematical operations, such as multiplication and addition, performed iteratively to learn and generate answers.”

And while full-scale optical computers are still years away, this optical memory represents a significant step in that direction.

Meanwhile, another team showed a new magneto-optical memory technology⁷ using cerium-substituted yttrium iron garnet (Ce:YIG). This material exhibits tunable optical behavior when exposed to magnetic fields. By embedding microscopic magnets, the researchers could store and manipulate data through changes in light propagation.

This way, they introduced a novel class of magneto-optical memories that have switching speeds 100 times faster than advanced photonic integrated technology and consume about one-tenth the power. Magneto-optical memories can also be rewritten more than 2.3 billion times.

Ni₄W: Field-Free Magnetization Achieved

Researchers from the University of Minnesota Twin Cities have now reported a new achievement in memory tech. 

Published in the peer-reviewed scientific journal Advanced Materials, the study detailed the development⁸, which involved the use of Ni₄W, an alloy of nickel and tungsten. This metal flips magnetism without requiring magnets, and as such, shows potential to power next-generation electronics.

With the team showcasing a way to produce spin currents to control magnetization in devices, the study opens the door to cheaper, faster, and more efficient computer memory and logic devices.

Switching the Magnetism of Metal Without Magnets

With the demand for emerging memory technology growing, researchers are actively exploring different alternatives to existing memory solutions that can increase the functionality of everyday tech while consuming less energy.

So, University of Minnesota researchers turned to a new material to make computer memory faster and more energy-efficient.

The material is a nickel-tungsten alloy, a class of material known for its high density, strength, and resistance to wear and corrosion. In these alloys, the specific composition of the metals influences their properties. 

In this study, the researchers used Ni₄W, a material that shows powerful magnetic control properties.

To choose Ni₄W, the team first searched the material database for potential candidates with stable phases within the I4/m space group, then used density functional theory (DFT) calculations, which identified Ni4W as the most promising candidate due to showing large theoretical SOT efficiency and being the ground state for the Ni-W binary intermetallic system.

The team verified the existence of unconventional spin Hall conductivity (USHC) for Ni4W (100) as well as Ni4W (211), but chose to focus their experimental efforts on the latter due to its better SOT efficiency, which exceeded the former. 

“Theoretical calculations confirm that Ni4W (211) is about the most optimal crystal orientation for USHC,” noted the study, adding that its hexagonal-like lattice structure makes it easier to grow experimentally.

The material can make computer memory faster as well as significantly reduce energy use in electronic devices. The researchers have secured a patent on the technology.

“Ni₄W reduces power usage for writing data, potentially cutting energy use in electronics significantly,” said senior paper author Jian-Ping Wang, who’s a Distinguished McKnight Professor and Robert F. Hartmann Chair in the Department of Electrical and Computer Engineering (ECE) at the U of M.

Unlike conventional materials, the low-symmetry Ni₄W allows for ‘field-free’ switching. What it means is that the material can switch its magnetic states without needing magnets. It is by generating spin currents in multiple directions that enable Ni₄W to flip magnetic states’ field-free’ without requiring external magnetic fields. 

In their work, the team provides new insight into the material while showcasing a more effective approach to control magnetization in small electronic devices using this combination of nickel and tungsten.

As per the study, the researchers found that Ni₄W generates strong spin-orbit torque (SOT), a way to manipulate magnetism in next-gen memory technologies.

SOT is an emerging technology that allows for an efficient manipulation of spintronic devices, which utilize the intrinsic spin of electrons as well as their charge, to store and manipulate information.

This mechanism emerges from the effects of spin-orbit coupling (SOC), like the anomalous Hall effect (AHE), spin Hall effect (SHE), and Rashba effect, and shows superior performance in terms of efficiency and speed. 

While SOT offers an efficient way to manipulate the magnetization of ferromagnetic materials (which exhibit permanent magnetizations and possess a permanent magnetic moment in the absence of an external field) in memory devices, conventional SOT materials like heavy metals and topological insulators are limited by their high crystal symmetry.

As a result, researchers either use materials with low symmetry or break the high symmetry using an external magnetic field to produce unconventional spin currents, enabling field-free deterministic switching of perpendicular magnetization. 

Despite the progress, the SOT efficiency of these materials continues to remain low, limiting their practical application. This, however, is not the case with the new material, which shows a large SOT efficiency of 0.3 at room temperature.

“We observed high SOT efficiency with multi-direction in Ni₄W both on its own and when layered with tungsten, pointing to its strong potential for use in low-power, high-speed spintronic devices.”

– Paper’s co-first author Yifei Yang, who’s a fifth-year Ph.D. student in Wang’s group

A large SOT efficiency of 0.73 was also observed in W/Ni4W (5 nm), but that could be from extrinsic effects.

Notably, the new material is made from common metals and, as such, can be manufactured using standard industrial processes. This ease of manufacturing makes it a low-cost process, in turn, making Ni₄W attractive to industry partners. This also means that the technology can be implemented into everyday products like phones and smart watches easily and in the near future.

“We are very excited to see that our calculations confirmed the choice of the material and the SOT experimental observation.”

– Paper’s co-first author Seungjun Lee, a postdoctoral fellow in ECE

So, the study has found Ni4W to be a promising unconventional SOT material for energy-efficient spintronic devices. Being cheap to produce, it can find its widespread application in devices like phones as well as data centers, making the future of electronics both smarter and more sustainable.

In the next steps, the team will grow these materials into a device, smaller than their previous work.

Investing in Memory Tech

Micron Technology (MU ), a leading player in DRAM, NAND, and high-bandwidth memory solutions, is heavily investing in next-generation memory, such as HBM, for AI workloads. In the future, we can expect the company to integrate novel solutions, such as spintronic or SOT-based memory, when they become commercially viable.

Micron Technology (MU )

With a market cap of $126.7 billion, MU shares are currently trading at $112.78, up 34.54% so far this year. It has an EPS (TTM) of 5.52 and a P/E (TTM) of 20.53. The dividend yield that shareholders can earn is 0.41%.

Regarding the company’s financial position, it reported $9.30 billion in revenue for the third quarter of fiscal 2025, which ended May 29, 2025. This represents a 15.5% increase from the prior quarter and a 36.5% increase from the same period last year.

GAAP net income for the period was $1.89 billion, or $1.68 per diluted share, and non-GAAP net income was $2.18 billion, or $1.91 per diluted share. Its operating cash flow also increased to $4.61 billion.

Micron ended the quarter with $12.22 bln in cash, marketable investments, and restricted cash.

The record revenue, CEO Sanjay Mehrotra noted, was driven by all-time-high DRAM revenue, including almost 50% sequential growth in HBM revenue. Revenue from data centres also hit a record in the quarter, while consumer-oriented end markets recorded strong sequential growth.

“We are on track to deliver record revenue with solid profitability and free cash flow in fiscal 2025, while we make disciplined investments to build on our technology leadership and manufacturing excellence to satisfy growing AI-driven memory demand.” 

– CEO Sanjay Mehrotra

Amidst all this, the company announced that its HBM3E 36GB 12-high offering will be integrated into AMD’s next-gen GPUs (Instinct™ MI350 Series), critical for training large AI models and handling complex HPC workloads like data processing and computational modeling.

Micron also announced a $200 billion U.S. expansion plan that includes domestic memory manufacturing and R&D, which is expected to create 90,000 direct and indirect jobs. At the same time, it finalized a direct funding of $275 million in the CHIPS Act.

Latest Micron Technology (MU) Stock News and Developments

Final Thoughts on the Future of Memory Tech

Memory technology continues to evolve and reshape the foundation of modern computing. From phase-change innovations to spintronic breakthroughs, all these advancements promise faster, more energy-efficient, and scalable solutions for AI, big data, and next-gen consumer electronics.

The latest discovery of Ni₄W alloy, with its field-free magnetization switching, could prove to be a game-changer, bridging the gap between cost-effectiveness and high-performance memory solutions and potentially making way for the widespread adoption of spin-orbit torque memory in mainstream electronics in the coming years.

Click here for a list of top non-silicon computing companies.

References:

1. Modi, G.; Parate, S. K.; Kwon, C.; Han, S. H.; Kim, Y.; Wang, X.; Lee, S.; Wu, L.; Kwon, J.; Kim, K.; Zhang, Y.; Milliron, D. J.; Duerloo, K.-A. N.; Kim, M. J.; Jeong, Y.; Park, J. Electrically Driven Long-Range Solid-State Amorphization in Ferroic In₂Se₃. Nature, 635, 847–853 (2024). Published online November 6, 2024. https://doi.org/10.1038/s41586-024-08156-8
2. Ozawa, K.; Nagase, Y.; Katsumata, M.; Shigematsu, K.; Azuma, M. Electric-Field Control of the Magneto-Optical Effect in a Transparent Perovskite Oxide. ACS Applied Materials & Interfaces, 16 (16), 20930–20936 (2024). Published online April 24, 2024. https://doi.org/10.1021/acsami.4c01232
3. Itoh, T.; Shigematsu, K.; Das, H.; Meisenheimer, P.; Maeda, K.; Lee, K.; Manna, M.; Reddy, S. P.; Susarla, S.; Stevenson, P.; Ramesh, R.; Azuma, M. Electric‑Field‑Driven Reversal of Ferromagnetism in (110)‑Oriented, Single‑Phase, Multiferroic Co‑Substituted BiFeO₃ Thin Films. Advanced Materials, published online April 28, 2025, e2419580. https://doi.org/10.1002/adma.202419580
4. Kim, I.–J.; Lee, J.–S.; … Lee, J.–S. Unlocking Large Memory Windows and 16-Level Data‑Per‑Cell Memory Operations in Hafnia‑Based Ferroelectric Transistors. Science Advances, published online June 7, 2024, 10 (23): eadn1345. https://doi.org/10.1126/sciadv.adn1345
5. Gupta, R.; Bouard, C.; Kammerbauer, F.; Ledesma‑Martín, J. O.; Bose, A.; Kononenko, I.; Martin, S.; Usé, P.; Jakob, G.; Drouard, M.; Kläui, M. Harnessing Orbital Hall Effect in Spin‑Orbit Torque MRAM. Nature Communications, 16, 130 (2025). Received 18 September 2024; Accepted 12 December 2024; Published 2 January 2025. https://doi.org/10.1038/s41467-024-55437-x
6. Goto, T.; Onbaşli, M. C.; Ross, C. A. Magneto-Optical Properties of Cerium‑Substituted Yttrium Iron Garnet Films with Reduced Thermal Budget for Monolithic Photonic Integrated Circuits. Optics Express, 20 (27), 28507–28517 (2012). Received 24 October 2012; Revised 20 November 2012; Accepted 21 November 2012; Published online December 10, 2012. https://doi.org/10.1364/OE.20.028507
7. Pintus, P.; Dumont, M.; Shah, V.; Murai, T.; Shoji, Y.; Huang, D.; Moody, G.; Bowers, J. E.; Youngblood, N. Integrated Non‑Reciprocal Magneto‑Optics with Ultra‑High Endurance for Photonic In‑Memory Computing. Nature Photonics, 19, 54–62 (2025). Received 18 January 2024; Accepted 14 September 2024; Published 23 October 2024. https://doi.org/10.1038/s41566-024-01549-1
8. Yang, Y.; Lee, S.; Chen, Y. C.; Jia, Q.; Dixit, B.; Sousa, D.; Odlyzko, M.; Garcia‑Barriocanal, J.; Yu, G.; Haugstad, G.; Fan, Y.; Huang, Y. H.; Lyu, D.; Cresswell, Z.; Liang, S.; Benally, O. J.; Low, T.; Wang, J. P. Large Spin‑Orbit Torque with Multi‑Directional Spin Components in Ni₄W. Advanced Materials, published online May 15, 2025, e2416763. https://doi.org/10.1002/adma.202416763