Computers Math

Compared to robots, human bodies are flexible, capable of fine movements, and can convert energy efficiently into movement. Drawing inspiration from human gait, researchers from Japan crafted a two-legged biohybrid robot by combining muscle tissues and artificial materials. Publishing on January 26 in the journal Matter, this method allows the robot to walk and pivot.
“Research on biohybrid robots, which are a fusion of biology and mechanics, is recently attracting attention as a new field of robotics featuring biological function,” says corresponding author Shoji Takeuchi of the University of Tokyo, Japan. “Using muscle as actuators allows us to build a compact robot and achieve efficient, silent movements with a soft touch.”
The research team’s two-legged robot, an innovative bipedal design, builds on the legacy of biohybrid robots that take advantage of muscles. Muscle tissues have driven biohybrid robots to crawl and swim straight forward and make turns — but not sharp ones. Yet, being able to pivot and make sharp turns is an essential feature for robots to avoid obstacles.
To build a nimbler robot with fine and delicate movements, the researchers designed a biohybrid robot that mimics human gait and operates in water. The robot has a foam buoy top and weighted legs to help it stand straight underwater. The skeleton of the robot is mainly made from silicone rubber that can bend and flex to conform to muscle movements. The researchers then attached strips of lab-grown skeletal muscle tissues to the silicone rubber and each leg.
When the researchers zapped the muscle tissue with electricity, the muscle contracted, lifting the leg up. The heel of the leg then landed forward when the electricity dissipated. By alternating the electric stimulation between the left and right leg every 5 seconds, the biohybrid robot successfully “walked” at the speed of 5.4 mm/min (0.002 mph). To turn, researchers repeatedly zapped the right leg every 5 seconds while the left leg served as an anchor. The robot made a 90-degree left turn in 62 seconds. The findings showed that the muscle-driven bipedal robot can walk, stop, and make fine-tuned turning motions.
“Currently, we are manually moving a pair of electrodes to apply an electric field individually to the legs, which takes time,” says Takeuchi. “In the future, by integrating the electrodes into the robot, we expect to increase the speed more efficiently.”
The team also plans to give joints and thicker muscle tissues to the bipedal robot to enable more sophisticated and powerful movements. But before upgrading the robot with more biological components, Takeuchi says the team will have to integrate a nutrient supply system to sustain the living tissues and device structures that allow the robot to operate in the air.
“A cheer broke out during our regular lab meeting when we saw the robot successfully walk on the video,” says Takeuchi. “Though they might seem like small steps, they are, in fact, giant leaps forward for the biohybrid robots.”
This work was supported by JST-Mirai Program, JST Fusion Oriented Research for disruptive Science and Technology, and the Japan Society for the Promotion of Science. More

Our understanding of the world relies greatly on our knowledge of its constituent materials and their interactions. Recent advances in materials science technologies have ratcheted up our ability to identify chemical substances and expanded possible applications.
One such technology is infrared spectroscopy, used for molecular identification in various fields, such as in medicine, environmental monitoring, and industrial production. However, even the best existing tool — the Fourier transform infrared spectrometer or FTIR — utilizes a heating element as its light source. Resulting detector noise in the infrared region limits the devices’ sensitivity, while physical properties hinder miniaturization.
Now, a research team led by Kyoto University has addressed this problem by incorporating a quantum light source. Their innovative ultra-broadband, quantum-entangled source generates a relatively wider range of infrared photons with wavelengths between 2 μm and 5 μm.
“This achievement sets the stage for dramatically downsizing the system and upgrading infrared spectrometer sensitivity,” says Shigeki Takeuchi of the Department of Electronic Science and Engineering.
Another elephant in the room with FTIRs is the burden of transporting mammoth-sized, power-hungry equipment to various locations for testing materials on-site. Takeuchi eyes a future where his team’s compact, high-performance, battery-operated scanners will lead to easy-to-use applications in various fields such as environmental monitoring, medicine, and security.
“We can obtain spectra for various target samples, including hard solids, plastics, and organic solutions. Shimadzu Corporation — our partner that developed the quantum light device — has concurred that the broadband measurement spectra were very convincing for distinguishing substances for a wide range of samples,” adds Takeuchi.
Although quantum entangled light is not new, bandwidth has thus far been limited to a narrow range of 1 μm or less in the infrared region. This new technique, meanwhile, uses the unique properties of quantum mechanics — such as superposition and entanglement — to overcome the limitations of conventional techniques.
The team’s independently developed chirped quasi-phase-matching device generates quantum-entangled light by harnessing chirping — gradually changing an element’s polarization reversal period — to generate quantum photon pairs over a wide bandwidth.
“Improving the sensitivity of quantum infrared spectroscopy and developing quantum imaging in the infrared region are part of our quest to develop real-world quantum technologies,” remarks Takeuchi. More

People who were more skeptical of human-caused climate change or the Black Lives Matter movement who took part in conversation with a popular AI chatbot were disappointed with the experience but left the conversation more supportive of the scientific consensus on climate change or BLM. This is according to researchers studying how these chatbots handle interactions from people with different cultural backgrounds.
Savvy humans can adjust to their conversation partners’ political leanings and cultural expectations to make sure they’re understood, but more and more often, humans find themselves in conversation with computer programs, called large language models, meant to mimic the way people communicate.
Researchers at the University of Wisconsin-Madison studying AI wanted to understand how one complex large language model, GPT-3, would perform across a culturally diverse group of users in complex discussions. The model is a precursor to one that powers the high-profile ChatGPT. The researchers recruited more than 3,000 people in late 2021 and early 2022 to have real-time conversations with GPT-3 about climate change and BLM.
“The fundamental goal of an interaction like this between two people (or agents) is to increase understanding of each other’s perspective,” says Kaiping Chen, a professor of life sciences communication who studies how people discuss science and deliberate on related political issues — often through digital technology. “A good large language model would probably make users feel the same kind of understanding.”
Chen and Yixuan “Sharon” Li, a UW-Madison professor of computer science who studies the safety and reliability of AI systems, along with their students Anqi Shao and Jirayu Burapacheep (now a graduate student at Stanford University), published their results this month in the journal Scientific Reports.
Study participants were instructed to strike up a conversation with GPT-3 through a chat setup Burapacheep designed. The participants were told to chat with GPT-3 about climate change or BLM, but were otherwise left to approach the experience as they wished. The average conversation went back and forth about eight turns.
Most of the participants came away from their chat with similar levels of user satisfaction.

“We asked them a bunch of questions — Do you like it? Would you recommend it? — about the user experience,” Chen says. “Across gender, race, ethnicity, there’s not much difference in their evaluations. Where we saw big differences was across opinions on contentious issues and different levels of education.”
The roughly 25% of participants who reported the lowest levels of agreement with scientific consensus on climate change or least agreement with BLM were, compared to the other 75% of chatters, far more dissatisfied with their GPT-3 interactions. They gave the bot scores half a point or more lower on a 5-point scale.
Despite the lower scores, the chat shifted their thinking on the hot topics. The hundreds of people who were least supportive of the facts of climate change and its human-driven causes moved a combined 6% closer to the supportive end of the scale.
“They showed in their post-chat surveys that they have larger positive attitude changes after their conversation with GPT-3,” says Chen. “I won’t say they began to entirely acknowledge human-caused climate change or suddenly they support Black Lives Matter, but when we repeated our survey questions about those topics after their very short conversations, there was a significant change: more positive attitudes toward the majority opinions on climate change or BLM.”
GPT-3 offered different response styles between the two topics, including more justification for human-caused climate change.
“That was interesting. People who expressed some disagreement with climate change, GPT-3 was likely to tell them they were wrong and offer evidence to support that,” Chen says. “GPT-3’s response to people who said they didn’t quite support BLM was more like, ‘I do not think it would be a good idea to talk about this. As much as I do like to help you, this is a matter we truly disagree on.'”
That’s not a bad thing, Chen says. Equity and understanding comes in different shapes to bridge different gaps. Ultimately, that’s her hope for the chatbot research. Next steps include explorations of finer-grained differences between chatbot users, but high-functioning dialogue between divided people is Chen’s goal.
“We don’t always want to make the users happy. We wanted them to learn something, even though it might not change their attitudes,” Chen says. “What we can learn from a chatbot interaction about the importance of understanding perspectives, values, cultures, this is important to understanding how we can open dialogue between people — the kind of dialogues that are important to society.” More

Chemists of the University of Amsterdam (UvA) have developed an autonomous chemical synthesis robot with an integrated AI-driven machine learning unit. Dubbed ‘RoboChem’, the benchtop device can outperform a human chemist in terms of speed and accuracy while also displaying a high level of ingenuity. As the first of its kind, it could significantly accelerate chemical discovery of molecules for pharmaceutical and many other applications. RoboChem’s first results were published on 25 January in the journal Science.
RoboChem was developed by the group of Prof. Timothy Noël at the UvA’s Van ‘t Hoff Institute for Molecular Sciences. Their paper shows that RoboChem is a precise and reliable chemist that can perform a variety of reactions while producing minimal amounts of waste. Working autonomously around the clock, the system delivers results quickly and tirelessly. Noël: ‘In a week, we can optimise the synthesis of about ten to twenty molecules. This would take a PhD student several months.’ The robot not only yields the best reaction conditions, but also provides the settings for scale-up. ‘This means we can produce quantities that are directly relevant for suppliers to the pharmaceutical industry, for example.’
RoboChem’s ‘brain’
The expertise of the Noël group is in flow chemistry, a novel way of performing chemistry where a system of small, flexible tubes replaces beakers, flasks and other traditional chemistry tools. In RoboChem, a robotic needle carefully collects starting materials and mixes these together in small volumes of just over half a millilitre. These then flow through the tubing system towards the reactor. There, the light from powerful LEDs triggers the molecular conversion by activating a photocatalyst included in the reaction mixture. The flow then continues towards an automated NMR spectrometer that identifies the transformed molecules. These data are fed back in real-time to the computer that controls RoboChem. ‘This is the brain behind RoboChem,’ says Noël. ‘It processes the information using artificial intelligence. We use a machine learning algorithm that autonomously determines which reactions to perform. It always aims for the optimal outcome and constantly refines its understanding of the chemistry.’
Impressive ingenuity
The group put a lot of effort into substantiating RoboChem’s results. All of the molecules now included in the Science paper were isolated and checked manually. Noël says the system has impressed him with its ingenuity: ‘I have been working on photocatalysis for more than a decade now. Still, RoboChem has shown results that I would not have been able to predict. For instance, it has identified reactions that require only very little light. At times I had to scratch my head to fathom what it had done. You then wonder: would we have done it the same way? In retrospect, you see RoboChem’s logic. But I doubt if we would have obtained the same results ourselves. Or not as quickly, at least.’
The researchers also used RoboChem to replicate previous research published in four randomly selected papers. They then determined whether Robochem produced the same — or better — results. ‘In about 80% of the cases, the system produced better yields. For the other 20%, the results were similar,’ Noël says. ‘This leaves me with no doubt that an AI-assisted approach will be beneficial to chemical discovery in the broadest possible sense.’
Breakthroughs in chemistry using AI
According to Noël, the relevance of RoboChem and other ‘computerised’ chemistry also lies in the generation of high-quality data, which will benefit the future use of AI. ‘In traditional chemical discovery only a few molecules are thoroughly researched. Results are then extrapolated to seemingly similar molecules. RoboChem produces a complete and comprehensive dataset where all relevant parameters are obtained for each individual molecule. That provides much more insight.’
Another feature is that the system also records ‘negative’ data. In current scientific practice, most published data only reflects successful experiments. ‘A failed experiment also provides relevant data,’ says Noël. ‘But this can only be found in the researchers’ handwritten lab notes. These are not published and thus unavailable for AI-powered chemistry. RoboChem will change that, too. I have no doubt that if you want to make breakthroughs in chemistry with AI, you will need these kinds of robots.’ More

The branch of mathematics known as topology has become a cornerstone of modern physics thanks to the remarkable — and above all reliable — properties it can impart to a material or system. Unfortunately, identifying topological systems, or even designing new ones, is generally a tedious process that requires exactly matching the physical system to a mathematical model. Researchers at the University of Amsterdam and the École Normale Supérieure of Lyon have demonstrated a model-free method for identifying topology, enabling the discovery of new topological materials using a purely experimental approach.
Topology encompasses the properties of a system that cannot be changed by any ‘smooth deformation’. As you might be able to tell from this rather formal and abstract description, topology began its life as a branch of mathematics. However, over the last few decades physicists have demonstrated that the mathematics underlying topology can have very real consequences. Topological effects can be found in a wide range of physical systems, from individual electrons to large-scale ocean currents.
As a concrete example: in the field of quantum matter, topology rose to fame thanks to so-called topological insulators. These materials do not conduct electricity through their bulk, but electrons move freely along their surfaces or edges. This surface conduction will persist, unhindered by material imperfections, as long as you do not do something drastic like changing the entire atomic structure of the material. Moreover, currents on the surfaces or edges of a topological insulator have a set direction (depending on the electron spin), again enforced by the topological nature of the electronic structure.
Such topological features can have very useful applications, and topology has become one of the frontiers of materials science. Aside from identifying topological materials in nature, parallel research efforts focus on designing synthetic topological materials from the bottom up. Topological edge states of mechanical structures known as ‘metamaterials’ present unmatched opportunities for achieving reliable responses in wave guiding, sensing, computation, and filtering.
Impractical mathematical models
Research in this area is slowed down by the lack of experimental ways to investigate the topological nature of a system. The necessity of matching a mathematical model to a physical system limits research to materials for which we already have a theoretical description, and forms a bottleneck for identifying and designing topological materials. To tackle this issue, Xiaofei Guo and Corentin Coulais of the Machine Materials Laboratory at the University of Amsterdam teamed up with Marcelo Guzmán, David Carpentier and Denis Bartolo of ENS Lyon.
“Until now, most experiments were intended to prove theories or showcase theoretical predictions in journals,” says Guo. “We found a way to measure topologically protected soft or fragile spots in unknown mechanical metamaterials without the need for modelling. Our approach allows for practical exploration and characterisation of material properties without delving into complex theoretical frameworks.”
Poking and prodding

The researchers demonstrated their method with mechanical metamaterials consisting of a network of rotors (rigid rods which can rotate) connected by elastic springs. Topology in these systems can make some regions of such a metamaterial particularly floppy or stiff. Bartolo: “We realised that selectively probing a material locally could give us all the necessary information to unveil soft or fragile spots in the structure, even in regions far removed from our probes. Using this, we developed a highly practical protocol applicable to a diverse range of materials and metamaterials.”
Prodding individual rotors in the metamaterial and tracking the resulting displacements and elongations in the system, the researchers identified different ‘mechanical molecules’: groups of rotors and springs which move as a single unit. In analogy to electrostatic systems, they then determined an effective ‘polarisation’ of each molecule, calculated from the molecules’ movements. This polarisation will suddenly flip direction in the presence of a topological feature, making inherent topology easy to identify.
The researchers applied their method to various mechanical metamaterials, some of which were known from previous studies to be topological, while others were new structures without an associated mathematical model. The results demonstrate that the experimentally determined polarisation is very effective in pointing out topological features.
This model-free approach is not just limited to mechanical systems; the same method could be applied to photonic or acoustic structures. It will make topology accessible to a broader range of physicists and engineers, and will make it easier to construct functional materials that go beyond laboratory demonstrations. More

Certain materials have desirable properties that are hidden, and just as you would use a flashlight to see in the dark, scientists can use light to uncover these properties.
Researchers at the University of California San Diego have used an advanced optical technique to learn more about a quantum material called Ta2NiSe5 (TNS). Their work appears in Nature Materials.
Materials can be perturbed through different external stimuli, often with changes in temperature or pressure; however, because light is the fastest thing in the universe, materials will respond very quickly to optical stimuli, revealing properties that would otherwise remain hidden.
“In essence, we shine a laser on a material and it’s like stop-action photography where we can incrementally follow a certain property of that material,” said Professor of Physics Richard Averitt, who led the research and is one of the paper’s authors. “By looking at how constituent particles move around in that system, we can tease out these properties that are really tricky to find otherwise.”
The experiment was conducted by lead author Sheikh Rubaiat Ul Haque, who graduated from UC San Diego in 2023 and is now a postdoctoral scholar at Stanford University. He, along with Yuan Zhang, another graduate student in Averitt’s lab, improved upon a technique called terahertz time-domain spectroscopy. This technique allows scientists to measure a material’s properties over a range of frequencies, and Haque’s improvements allowed them access to a broader range of frequencies.
The work was based on a theory created by another of the paper’s authors, Eugene Demler, a professor at ETH Zürich. Demler and his graduate student Marios Michael developed the idea that when certain quantum materials are excited by light, they may turn into a medium that amplifies terahertz frequency light. This led Haque and colleagues to look closely into the optical properties of TNS.
When an electron is excited to a higher level by a photon, it leaves behind a hole. If the electron and hole are bound, an exciton is created. Excitons may also form a condensate — a state that occurs when particles come together and behave as a single entity.

Haque’s technique, backed by Demler’s theory and using density functional calculations by Angel Rubio’s group at Max Planck Institute for the Structure and Dynamics of Matter, the team was able to observe anomalous terahertz light amplification, which uncovered some of the hidden properties of the TNS exciton condensate.
Condensates are a well-defined quantum state and using this spectroscopic technique could allow some of their quantum properties to be imprinted onto light. This may have implications in the emerging field of entangled light sources (where multiple light sources have interconnected properties) utilizing quantum materials.
“I think it’s a wide-open area,” stated Haque. “Demler’s theory can be applied to a suite of other materials with nonlinear optical properties. With this technique, we can discover new light-induced phenomena that haven’t been explored before.”
Funding provided by the DARPA DRINQS Program (D18AC00014), the Swiss National Science Foundation (200021_212899), Army Research Office (W911NF-21-1-0184), the European Research Council (ERC-2015-AdG694097), the Cluster of Excellence ‘Advanced Imaging of Matter’ (AIM), Grupos Consolidados (IT1249-19), Deutsche Forschungsgemeinschaft (170620586), and the Flatiron Institute. More

Computer scientists have invented a highly effective, yet incredibly simple, algorithm to decide which items to toss from a web cache to make room for new ones. Known as SIEVE, the new open-source algorithm holds the potential to transform the management of web traffic on a large scale.
SIEVE is a joint project of computer scientists at Emory University, Carnegie Mellon University and the Pelikan Foundation. The team’s paper on SIEVE will be presented at the 21st USENIX Symposium on Networked Systems Design and Implementation (NSDI) in Santa Clara, California, in April.
A preprint of the paper is already making waves.
“SIEVE is bigger and greater than just us,” says Yazhuo Zhang, an Emory PhD student and co-first author of the paper. “It is already performing well but we are getting a lot of good suggestions to make it even better. That’s the beauty of the open-source world.”
Zhang shares first authorship of the paper with Juncheng (Jason) Yang, who received his master’s degree in computer science at Emory and is now a PhD candidate at Carnegie Mellon.
“SIEVE is an easy improvement of a tried-and-true cache-eviction algorithm that’s been in use for decades — which is literally like centuries in the world of computing,” says Ymir Vigfusson, associate professor in Emory’s Department of Computer Science.
Vigfusson is co-senior author of the paper, along with Rashmi Vinayak, an associate professor in Carnegie Mellon’s computer science department. Yao Yue, a computer engineer at the Pelikan Foundation, is also a co-author.

In addition to its speed and effectiveness, a key factor sparking interest in SIEVE is its simplicity, lending it scalability.
“Simplicity is the ultimate sophistication,” Vigfusson says. “The simpler the pieces are within a system designed to serve billions of people within a fraction of a second, the easier it is to efficiently implement and maintain that system.”
Keeping ‘hot objects’ handy
Many people understand the value of regularly reorganizing their clothing closet. Items that are never used can be tossed and those that are rarely used can be moved to the attic or some other remote location. That leaves the items most commonly worn within easy reach so they can be found quickly, without rummaging around.
A cache is like a well-organized closet for computer data. The cache is filled with copies of the most popular objects requested by users, or “hot objects” in IT terminology. The cache maintains this small collection of hot objects separately from a computer network’s main database, which is like a vast warehouse filled with all the information that could be served by the system.
Caching hot objects allows a networked system to run more efficiently, rapidly responding to requests from users. A web application can effectively handle more traffic by popping into a handy closet to grab most of the objects users want rather than traveling down to the warehouse and searching through a massive database for each request.

“Caching is everywhere,” Zhang says. “It’s important to every company, big or small, that is using web applications. Every website needs a cache system.”
And yet, caching is relatively understudied in the computer science field.
How caching works
While caching can be thought of as a well-organized closet for a computer, it is difficult to know what should go into that closet when millions of people, with constantly changing needs, are using it.
The fast memory of the cache is expensive to run yet critical to a good experience for web users. The goal is to keep the most useful, future information within the cache. Other objects must be continuously winnowed out, or “evicted” in tech terminology, to make room for the changing array of hot objects.
Cache-eviction algorithms determine what objects to toss and when to do so.
FIFO, or “first-in, first-out,” is a classic eviction algorithm developed in the 1960s. Imagine objects lined up on a conveyor belt. Newly requested objects enter on the left and the oldest objects get evicted when they reach the end of the line on the right.
In the LRU, or “least recently used,” algorithm the objects also move along the line towards eviction at the end. However, if an object is requested again while it moves down the conveyor belt, it gets moved back to the head of the line.
Hundreds of variations of eviction algorithms exist but they have tended to take on greater complexity to gain efficiency. That generally means they are opaque to reason about and require high maintenance, especially when dealing with massive workloads.
“If an algorithm is very complicated, it tends to have more bugs, and all of those bugs need to be fixed,” Zhang explains.
A simple idea
Like LRU and some other algorithms, SIEVE makes a simple tweak on the basic FIFO scheme.
SIEVE initially labels a requested object as a “zero.” If the object is requested again as it moves down the belt, its status changes to “one.” When an object labeled “one” makes it to the end of the line it is automatically reset to “zero” and evicted.
A pointer, or “moving hand,” also scans the objects as they travel down the line. The pointer starts at the end of the line and then jumps to the head, moving in a continuous circle. Anytime the pointer hits an object labeled “zero,” the object is evicted.
“It’s important to evict unpopular objects as quickly as possible, and SIEVE is very fast at this task,” Zhang says.
In addition to this quick demotion of objects, SIEVE manages to maintain popular objects in the cache with minimal computational effort, known as “lazy promotion” in computer terminology. The researchers believe that SIEVE is the simplest cache-eviction algorithm to effectively achieve both quick demotion and lazy promotion.
A lower miss ratio
The purpose of caching is to achieve a low miss ratio — the fraction of requested objects that must be fetched from “the warehouse.”
To evaluate SIEVE, the researchers conducted experiments on open-source web-cache traces from Meta, Wikimedia, X and four other large datasets. The results showed that SIEVE achieves a lower miss ratio than nine state-of-the-art algorithms on more than 45% of the traces. The next best algorithm has a lower miss ratio on only 15%.
The ease and simplicity of SIEVE raises the question of why no one came up with the method before. The SIEVE team’s focus on how patterns of web traffic have changed in recent years may have made the difference, Zhang theorizes.
“For example,” she says, “new items now become ‘hot’ quickly but also disappear quickly. People continuously lose interest in things because new things keep coming up.”
Web-cache workloads tend to follow what are known as generalized Zipfian distributions, where a small subset of objects account for a large proportion of requests. SIEVE may have hit a Zipfian sweet spot for current workloads.
“It is clearly a transformative moment for our understanding of web-cache eviction,” Vigfusson says. “It changes a construct that’s been used blindly for so long.”
Even a tiny improvement in a web-caching system, he adds, can save millions of dollars at a major data center.
Zhang and Yang are on track to receive their PhDs in May.
“They are doing incredible work,” Vigfusson says. “It’s safe to say that both of them are now among the world experts on web-cache eviction.” More

Researchers with the Department of Energy’s SLAC National Accelerator Laboratory, Stanford University and the DOE’s Lawrence Berkeley National Laboratory (LBNL) grew a twisted multilayer crystal structure for the first time and measured the structure’s key properties. The twisted structure could help researchers develop next-generation materials for solar cells, quantum computers, lasers and other devices.
“This structure is something that we have not seen before — it was a huge surprise to me,” said Yi Cui, a professor at Stanford and SLAC and paper co-author. “A new quantum electronic property could appear within this three-layer twisted structure in future experiments.”
Adding layers, with a twist
The crystals the team designed extended the concept of epitaxy, a phenomenon that occurs when one type of crystal material grows on top of another material in an ordered way — kind of like growing a neat lawn on top of soil, but at the atomic level. Understanding epitaxial growth has been critical to the development of many industries for more than 50 years, particularly the semiconductor industry. Indeed, epitaxy is part of many of the electronic devices that we use today, from cell phones to computers to solar panels, allowing electricity to flow, and not flow, through them.
To date, epitaxy research has focused on growing one layer of material onto another, and the two materials have the same crystal orientation at the interface. This approach has been successful for decades in many applications, such as transistors, light-emitting diodes, lasers and quantum devices. But to find new materials that perform even better for more demanding needs, like quantum computing, researchers are searching for other epitaxial designs — ones that might be more complex, yet better performing, hence the “twisted epitaxy” concept demonstrated in this study.
In their experiment, detailed this month in Science, researchers added a layer of gold between two sheets of a traditional semiconducting material, molybdenum disulfide (MoS2). Because the top and bottom sheets were oriented differently, the gold atoms could not align with both simultaneously, which allowed the Au structure to twist, said Yi Cui, Professor Cui’s graduate student in materials science and engineering at Stanford and co-author of the paper.
“With only a bottom MoS2 layer, the gold is happy to align with it, so no twist happens,” said Cui, the graduate student. “But with two twisted MoS2 sheets, the gold isn’t sure to align with the top or bottom layer. We managed to help the gold solve its confusion and discovered a relationship between the orientation of Au and the twist angle of bilayer MoS2.”
Zapping gold nanodiscs

To study the gold layer in detail, the researcher team from the Stanford Institute for Materials and Energy Sciences (SIMES) and LBNL heated a sample of the whole structure to 500 degrees Celsius. Then they sent a stream of electrons through the sample using a technique called transmission electron microscopy (TEM), which revealed the morphology, orientation and strain of the gold nanodiscs after annealing at the different temperatures. Measuring these properties of the gold nanodiscs was a necessary first step toward understanding how the new structure could be designed for real world applications in the future.
“Without this study, we would not know if twisting an epitaxial layer of metal on top of a semiconductor was even possible,” said Cui, the graduate student. “Measuring the complete three-layer structure with electron microscopy confirmed that it was not only possible, but also that the new structure could be controlled in exciting ways.”
Next, researchers want to further study the optical properties of the gold nanodiscs using TEM and learn if their design alters physical properties like band structure of Au. They also want to extend this concept to try to build three-layer structures with other semiconductor materials and other metals.
“We’re beginning to explore whether only this combination of materials allows this or if it happens more broadly,” said Bob Sinclair, the Charles M. Pigott Professor in Stanford’s school of Materials Science and Engineering and paper co-author. “This discovery is opening a whole new series of experiments that we can try. We could be on our way to finding brand new material properties that we could exploit.” More