In a leap forward for next-generation electronics, engineers at MIT have developed an innovative method to create and peel ultrathin layers of electronic material—akin to flexible, electronic “skins.” This breakthrough could pave the way for a new era of ultra-light, compact, and highly sensitive electronic devices, ranging from wearable sensors and flexible computing components to cutting-edge night vision systems.

As a proof of concept, the MIT team produced a 10-nanometer-thick membrane made from a heat-sensitive material known as pyroelectric film. This ultrathin film is capable of detecting minute changes in temperature and radiation across the far-infrared spectrum—an essential feature for high-performance thermal imaging.

“Reducing both the weight and cost, this film opens the door to lightweight, portable infrared sensors that could even be integrated into eyewear,” said Xinyuan Zhang, graduate student in MIT’s Department of Materials Science and Engineering and the study’s lead author.

Unlike conventional far-infrared sensors that rely on bulky, power-hungry cooling systems to function, MIT’s new film operates efficiently at room temperature. This allows for more compact designs that could transform current technologies, including night-vision goggles, which are often heavy and cumbersome.

The secret to this innovation lies in a surprising discovery: a certain heat-sensitive compound, PMN-PT, could be cleanly separated from its substrate without the need for an intermediate layer. Researchers found that lead atoms within the film acted like microscopic “nonstick” agents, allowing the membrane to lift away seamlessly and remain atomically smooth.

The team, in collaboration with researchers from the University of Wisconsin at Madison and other institutions, used this property to fabricate arrays of ultrathin heat-sensing pixels. These sensors exhibited sensitivity comparable to top-tier night-vision systems—without requiring cryogenic cooling—and showed potential for full-spectrum infrared detection.

“This technology could extend beyond defense and security,” said Zhang. “Its potential uses include autonomous driving in low-visibility conditions, real-time environmental monitoring, and even detecting overheating in semiconductor chips.”

The researchers are now working to integrate the films into practical devices, including lightweight, high-resolution night-vision glasses. They also believe their peeling technique could be applied to other types of ultrathin semiconductors, even those lacking lead, by engineering substrates to replicate the same peel-off effect.

The findings were published in Nature and include contributions from a broad team across MIT, the University of Wisconsin at Madison, Rensselaer Polytechnic Institute, and several other institutions.

At any given time, approximately 100,000 people in the U.S. are on the waitlist for a kidney transplant, but only about one-fifth receive a new kidney each year, while others die waiting. A new study co-authored by an MIT economist adds fresh insight into this life-or-death issue, offering the most precise estimates yet of how kidney transplants extend patient lives—and how the current system might be optimized to save even more.

Published in the latest issue of Econometrica, the paper—“Choices and Outcomes in Assignment Mechanisms: The Allocation of Deceased Donor Kidneys”—is the work of Nikhil Agarwal, professor of economics at MIT; Charles Hodgson of Yale University; and Paulo Somaini of Stanford University.

“There’s always this question about how to take the scarce number of organs being donated and place them efficiently, and place them well,” said Agarwal, in a statement. He emphasized that the goal of the study is to inform, not advocate, contributing rigorous data to help shape future kidney allocation policies.

The research finds that under the current U.S. kidney transplant system, recipients gain an average of 9.29 additional life-years—a metric known as LYFT (life-years from transplantation). If kidneys were distributed randomly, that figure would fall to 7.54 years. However, by restructuring the matching system, the study estimates that the LYFT could reach as high as 14.08 years.

To reach these conclusions, the researchers used comprehensive data from the Organ Procurement and Transplantation Network, covering patients from 2000 to 2010, and tracked survival outcomes through February 2020. The study is the first of its kind to take a quasiexperimental approach, accounting for complexities such as patients’ health status and the choices they make when accepting transplant offers.

“The [previous] methodology of estimating what are the life-years benefits was not incorporating this selection issue,” said Agarwal. The study found that patients are more likely to accept kidneys from younger donors, those without hypertension, those who died from head trauma (often a sign of otherwise healthy organs), and donors with perfect tissue-type matches.

One key finding is that healthier patients tend to benefit more from transplants than sicker ones—a fact that could pose a challenge to current policies, which often prioritize patients who have spent the most time on the waitlist, or those in more dire health.

“You might think people who are the sickest and who are most likely to die without an organ are going to benefit the most from it [in added life-years],” Agarwal noted. “But there might be some other comorbidity or factor that made them sick, and their body’s going to take a toll on the new organ, so the benefits might not be as large.”

This creates a policy dilemma, as the researchers write: “Our results indicate … a dilemma rooted in the tension between these two goals”—maximizing life-years versus prioritizing the sickest patients.

Ultimately, Agarwal stresses that the study’s aim is not to advocate for a specific allocation model, but to provide tools for better policymaking. “I don’t necessarily think it’s my comparative advantage to make the ethical decisions,” he said, “but we can at least think about and quantify what some of the tradeoffs are.”

As the conversation around kidney transplant allocation continues, the study provides essential evidence to guide efforts in balancing ethics, efficiency, and patient outcomes.

A new method is changing the way we think about molecule design, bringing us closer to the possibility of using large language models (LLMs) to streamline the creation of new medicines and materials. Imagine asking, in plain language, for a molecule with specific properties, and receiving a comprehensive plan on how to synthesize it. This futuristic vision is now within reach, thanks to a collaboration between researchers from MIT and the MIT-IBM Watson AI Lab.

A New era in molecular discovery

Traditionally, discovering the right molecules for medicines and materials has been a slow and resource-intensive process. It often involves the use of vast computational power and months of painstaking work to explore the nearly infinite pool of potential molecular candidates. However, this new method, blending LLMs with other machine-learning models known as graph-based models, offers a promising solution to speed up this process.

These researchers have developed an innovative system that augments an LLM with graph-based AI models, designed specifically to handle molecular structures. The approach allows users to input natural language queries specifying the desired molecular properties, and in return, the system provides not only a molecular design but also a step-by-step synthesis plan.

LLMs and graph models

LLMs like ChatGPT have revolutionized the way we interact with text, but they face challenges when it comes to molecular design. Molecules are graph structures—composed of atoms and bonds—which makes them fundamentally different from text. LLMs typically process text as a sequence of words, but molecules do not follow a linear structure. This discrepancy has made it difficult for LLMs to understand and predict molecular configurations in the same way they handle sentences.

To bridge this gap, MIT’s researchers created Llamole—a system that uses LLMs to interpret user queries and then switches between different graph-based AI modules to generate molecular structures, explain their rationale, and devise a synthesis strategy. The system combines the power of text, graphs, and synthesis steps into a unified workflow.

As a result, this multimodal approach drastically improves performance. Llamole was able to generate molecules that were far better at meeting user specifications and more likely to have a viable synthesis plan, increasing the success rate from 5 percent to 35 percent.

Llamole’s success lies in its unique ability to seamlessly combine language processing with graph-based molecular modeling. For example, if a user requests a molecule with specific traits—such as one that can penetrate the blood-brain barrier and inhibit HIV—the LLM interprets the plain-language request and switches to a graph module to generate the appropriate molecular structure.

This switch occurs through the use of a new type of trigger token, allowing the LLM to activate specific modules as needed. The process unfolds in stages: the LLM first predicts the molecular structure, then uses a graph neural network to encode the structure, and finally, a retrosynthetic module predicts the necessary steps to synthesize the molecule. The seamless flow between these stages ensures that the LLM maintains an understanding of what each module does, further enhancing its predictive accuracy.

“The beauty of this is that everything the LLM generates before activating a particular module gets fed into that module itself. The module is learning to operate in a way that is consistent with what came before,” says Michael Sun, an MIT graduate student and co-author of the study.

Simplicity meets precision

One of the most striking aspects of this new method is its ability to generate simpler, more cost-effective molecular structures. In tests, Llamole outperformed other LLM-based methods and achieved a notable 35 percent success rate in retrosynthetic planning, up from a mere 5 percent with traditional approaches. “On their own, LLMs struggle to figure out how to synthesize molecules because it requires a lot of multistep planning. Our method can generate better molecular structures that are also easier to synthesize,” says Gang Liu, the study’s lead author.

By designing molecules with simpler structures and more accessible building blocks, Llamole could significantly reduce the time and cost involved in developing new compounds.

The road ahead

Though Llamole’s current capabilities are impressive, there is still work to be done. The researchers built two custom datasets to train Llamole, but these datasets focus on only 10 molecular properties. Moving forward, they hope to expand Llamole’s capabilities to design molecules based on a broader range of properties and improve the system’s retrosynthetic planning success rate.

In the long run, the team envisions Llamole serving as a foundation for broader applications beyond molecular design. “Llamole demonstrates the feasibility of using large language models as an interface to complex data beyond textual description, and we anticipate them to be a foundation that interacts with other AI algorithms to solve any graph problems,” says Jie Chen, a senior researcher at MIT-IBM Watson AI Lab.

With further refinements, Llamole could revolutionize fields from pharmaceuticals to material science, offering a glimpse into the future of AI-driven innovation in molecular discovery.