Researchers at the University of New Hampshire have harnessed artificial intelligence to accelerate the discovery of new functional magnetic materials, creating a searchable database of 67,573 magnetic materials, including 25 previously unrecognized compounds that remain magnetic even at high temperatures.

“By accelerating the discovery of sustainable magnetic materials, we can reduce dependence on rare earth elements, lower the cost of electric vehicles and renewable-energy systems, and strengthen the U.S. manufacturing base,” said Suman Itani, lead author and a doctoral student in physics.

The newly created database, named the Northeast Materials Database, helps to more easily explore all the magnetic materials which play a major role in the technology that powers our world: smartphones, medical devices, power generators, electric vehicles and more. But these magnets rely on expensive, imported, and increasingly difficult to obtain rare earth elements, and no new permanent magnet has been discovered from the many magnetic compounds we know exist.

How AI is transforming materials research

The research, published in the journal Nature Communications, outlines how the UNH team built an artificial intelligence system that can read scientific papers and extract those key experimental details.

This data fed computer models that identified whether a material is magnetic, and how high a temperature it can withstand before losing its magnetism and organized it into a single, searchable database.

Scientists know that many undiscovered magnetic compounds exist, but testing every possible combination of elements—potentially millions—in the lab is prohibitively time-consuming and expensive.

“We are tackling one of the most difficult challenges in materials science—discovering sustainable alternatives to permanent magnets—and we are optimistic that our experimental database and growing AI technologies will make this goal achievable,” said Jiadong Zang, physics professor and co-author.

Researchers, which also include co-author Yibo Zhang, a postdoctoral researcher in both physics and chemistry, say that moving forward the modern large language model behind this project could have widespread use beyond this database, particularly in higher education. For instance, converting images to a modern rich text format could also be used to modernize library holdings.

Despite the never-ending drama over whether to ban the app, President Donald Trump’s volatile tariff regime, and executive shake-ups, TikTok’s ecommerce division is still seeing massive growth.

TikTok doesn’t disclose aggregate sales figures, but the price and sales volume of individual products are visible on the app. Based on that data, the analytics firm EchoTik estimates TikTok Shop sold $19 billion worth of products globally from July through September of this year. The United States, its largest market, accounted for $4 billion to $4.5 billion in sales, an increase of about 125 percent compared to the second quarter of 2025.

To put those numbers into perspective, consider that TikTok Shop is now on the same scale as eBay, which saw $20.1 billion in total sales in the last quarter. TikTok Shop only launched in the US in September 2023, while eBay has been around for over 30 years. That speed of growth is remarkable.

“We have mostly focused on TikTok from this point of view of the TikTok ban, and we have almost forgotten that TikTok Shop’s share in US ecommerce just continues to grow slowly,” says Juozas Kaziukėnas, an independent ecommerce analyst.

What You See Is What You Get

TikTok Shop broke into the extremely saturated ecommerce landscape in the US by excelling at an age-old platitude: show, don’t tell. Creators post short videos on TikTok trying on clothes or using home gadgets and include links to buy the products on the same platform. It creates a much more straightforward experience for consumers, who can see real people testing products instead of needing to wade through written reviews on traditional ecommerce sites.

Ivy Yang, the founder of Wavelet Strategy, a strategic public relations consultancy, says she recently bought a dust-mite-removing vacuum on Amazon shortly before she scrolled past a TikTok Shop video featuring a similar product. She quickly realized the TikTok Shop version had more features, so she ordered it, tried it out, and returned the one from Amazon. To her, that’s the appeal of shopping on TikTok. “I need to see how it works in action,” Yang explains.

In theory, that’s what makes livestream shopping even more popular, at least in China, because now influencers can tout products on camera in real time, and there’s little editing involved that might mask any potential product defects. In recent years, livestream shopping has completely reshaped how people buy things in China and has become one of ByteDance’s main revenue pillars. But despite how hard TikTok has tried, it simply hasn’t been able to replicate that success in the US. Kaziukėnas says that TikTok Shop’s performance likely still falls short of ByteDance’s expectations, especially when it comes to livestream shopping.

The transition from a carbon-based fuel economy to that centered on hydrogen has gained interest worldwide given the focus on sustainability. As researchers in corrosion, it became obvious for us to look at the underlying interaction of hydrogen with materials as it forms the backbone of the hydrogen infrastructure, especially with respect to hydrogen transportation. For example, pipelines carrying hydrogen blended with natural gas offer an economic means of transporting hydrogen over long distances.

Of critical interest for such applications is the hydrogen diffusion characteristics in such steels as it gives fundamental knowledge of the threshold amount of hydrogen that can cause failure.

Reliably measuring the diffusion coefficient of hydrogen in steels is of great value to researchers working in the area of hydrogen-material interactions.

When we set out to measure the diffusion characteristics of hydrogen in steels, we thought it could simply be followed from the ASTM (American Society for Testing of Materials) standard already available. We thought that we would indeed measure a typical hydrogen permeation transient using a classical Devanathan-Stachurski double permeation cell.

In this approach, upon hydrogen charging on one side of the sample, the first atomic hydrogen is detected on the other side after a breakthrough time, followed by a “rise” in the hydrogen flux and finally attaining a steady state from which the diffusion coefficient could be evaluated.

Although it looked straightforward, we faced challenges in implementing this in our lab. The first question we struggled with was obtaining the so-called steady state hydrogen permeation flux. For a typical electrochemical permeation measurement, we had to charge the sample with hydrogen at a certain current density.

The only question was by how much? From what we saw in literature, we tried to use severe charging conditions in alkaline electrolyte to begin with and we could not achieve this steady state. The flux reached a maximum and started to decrease thereafter, showing atypical behavior.

Trying to repeat the measurements were in vain, but what we noticed and what indeed puzzled us was some visible color change on the hydrogen charging side of the steel surface just after the measurement.

So, we immediately investigated the surface using scanning electron microscopy (SEM) to indeed observe cracked layers and randomly distributed particles all over the sample. These particles showed a peak corresponding to oxygen when analyzed with energy dispersive X-ray spectroscopy (EDS), prompting us to think they were iron oxides and encouraging us to use complementary characterization techniques to further identify them.

We used Raman spectroscopy to identify mixed iron oxides comprising of magnetite (Fe₃O₄), hematite (Fe₂O₃), and lepidocrocite (γ-FeOOH). Further, we calculated, using X-ray photoelectron spectroscopy (XPS), depth profiling, the thickness of the oxide to be around 50 nm.

We could also confirm this using Focused Ion Beam (FIB) milling and SEM cross-section imaging. But, formation of iron oxides during hydrogen charging was really surprising because the electrochemical conditions we used don’t generally support iron corrosion.

So, we proposed a hypothesis that during hydrogen charging, the formation of hydrogen bubbles occurs, and they attach to the surface of the steel. Due to this, the polarization potential applied to the steel is actually not realized on the surface as there is continuous and excessive hydrogen bubble formation.

As a result, an Ohmic drop across the bubbles occurs which, along with a higher pH value due to hydrogen evolution, could result in iron corrosion, according to the Pourbaix diagram.

This results in iron oxide formation, which we also confirmed by measuring the thickness using XPS and observation of particles on the surface using SEM for an independent electrochemical hydrogen charging experiment.

The results of this study were published in Corrosion Science.

But one might wonder how does the formation of iron oxide explain the atypical behavior of the hydrogen permeation flux. We suggested that these hydrogen bubbles, after growing up to a critical size, detach from the surface and therefore expose the underlying iron oxide.

The oxides then immediately undergo reduction owing to the electrochemical potential applied, and further result in the formation of fresh catalytic iron that enhances the hydrogen activity and promotes higher hydrogen flux.

On the other hand, the formation of iron oxide could also block hydrogen permeation, which could explain the decrease after reaching the maximum in the hydrogen permeation flux.

Having found out that severe charging leads to iron corrosion and surface effects during hydrogen permeation, we employed electrochemical impedance spectroscopy to further prove that the iron oxide grows during hydrogen charging.

By measuring a corresponding higher charge transfer resistance for the oxide, we indeed showed that it influences the hydrogen permeation behavior. We also made use of the electron backscattered diffraction (EBSD) technique to show that such severe charging leads to generation of new dislocations that introduce artifacts into the measurement of the hydrogen diffusion constant.

All this meant that we had to devise a strategy to avoid severe charging, so we came up with the idea of “soft” charging where we used much lower hydrogen charging current densities for performing the hydrogen permeation measurement.

Guess what, the idea worked!

We could measure a steady-state in the hydrogen permeation flux which did not decrease with time. We could clearly correlate this observation to the significant decrease in the amount of iron oxides visible on the surface using SEM and the almost negligible number of dislocations introduced using EBSD.

Thus, we suggest the use of “soft” hydrogen charging to measure reliably the diffusion constant of hydrogen in steels.

In essence, we report a surprising observation of iron corrosion during hydrogen charging in an electrochemical permeation measurement and suggest ways to circumvent this for reliably measuring the diffusion constant of hydrogen in steels. We believe this could be of great use to researchers working in the area of hydrogen-material interactions, the electrochemistry and corrosion community.

This story is part of Science X Dialog, where researchers can report findings from their published research articles. Visit this page for information about Science X Dialog and how to participate.

Vijayshankar Dandapani is an Associate Professor in the Metallurgical Engineering and Materials Science Department, Indian Institute of Technology (IIT), Bombay where he heads the Electrochemistry at Interface Lab. He works in the area of hydrogen, electrochemistry and corrosion.