Chinese Researchers Report 1,000 Times Faster Growth of Atomically Thin Chip Material

A Faster Way to Grow a Missing Chip Ingredient

Chinese researchers have reported a way to grow a semiconductor material about 1,000 times faster than conventional techniques, tackling a production bottleneck for electronics built from layers just a few atoms thick. Their modified process produces tungsten silicon nitride at approximately 20 micrometers per minute, with experimental sheets reaching about 36 by 18 millimeters.

The team, led by Zhu Mengjian alongside Ren Wencai and Xu Chuan, brings together scientists from the National University of Defence Technology and the Chinese Academy of Sciences. Ren and Xu are from the academy's Institute of Metal Research. Their findings appeared in the journal National Science Review on March 26.

The advance concerns the speed of growing a semiconductor material, a crucial manufacturing step before it can become part of a working chip. Tungsten silicon nitride is particularly interesting because it naturally provides a type of electrical behavior that has been difficult to obtain reliably in two dimensional semiconductors.

The researchers changed the surface on which the material grows, using a liquid gold and tungsten bilayer instead of a conventional solid substrate. A substrate is the supporting surface beneath a growing film. Their approach combines a smooth growth surface with a nearby supply of tungsten, one of the ingredients in the semiconductor.

As of April 16, 2026, the result offers a possible route toward faster production of materials for future chips. Turning the experimental films into commercial devices will still require larger sheets, fewer defects and compatibility with the many processing steps inside a semiconductor factory.

What the Production Figures Show

The researchers describe an increase of roughly three orders of magnitude, or about 1,000 times, in growth speed. The reported rate is approximately 20 micrometers per minute. A micrometer is one thousandth of a millimeter, so even this faster process operates on dimensions far smaller than most everyday objects.

The earlier production benchmark was five hours to grow just 0.001 millimeters. Expressed in inches, the comparison was roughly 0.00004 inches over five hours against about 0.0008 inches per minute with the new method. Those measurements describe the distance over which material grows.

The experimental films measured approximately 36 by 18 millimeters. Producing continuous sheets at that size moves the work toward manufacturing on wafers, the flat platforms on which chip factories build electronic devices. Commercial fabrication commonly uses wafers measuring 300 millimeters across, leaving a substantial difference between the current demonstration and a production platform.

Growth speed matters because a process that takes hours to produce a tiny sample can become impractical when manufacturers need much larger quantities. Faster growth could help address that problem, while the electrical quality and consistency of the resulting material remain equally important.

Why Silicon Faces Increasing Pressure

Silicon has supported more than five decades of progress associated with Moore's Law, the observation that the number of transistors on a microchip doubles approximately every two years while costs fall. Transistors are tiny switches that control electrical signals. Packing more of them onto chips has helped drive advances from smartphones to supercomputers and supported a semiconductor industry worth a trillion dollars.

Transistors now operate at dimensions where the behavior of individual particles becomes increasingly difficult to manage. Below approximately 5 nanometers, quantum effects become a growing concern for silicon devices. A nanometer is one billionth of a meter.

Quantum tunneling allows electrons to pass through insulating barriers that are intended to block them. This leakage wastes electricity and makes it harder to maintain a clean difference between a switch being on and off. Placing more components close together also makes removing heat increasingly difficult.

Artificial intelligence adds pressure to keep improving computing performance. Large language models and other AI systems require rapidly increasing processing capacity. Industry forecasts cited in the account put potential AI computing demand in 2030 at 1,000 times its 2023 level. The growing demand, alongside the physical constraints on silicon, is driving research into materials that could extend or replace conventional manufacturing approaches.

How Extremely Thin Materials Could Help

Two dimensional semiconductors are exceptionally thin materials, sometimes only a single atom thick, whose electrical behavior differs from that of three dimensional crystals. Their natural thinness offers a different starting point for making the channel, the part of a transistor through which electrical current flows.

Conventional silicon channels must be formed from a solid body of material. Two dimensional materials already have an extremely thin structure, reducing the need to remove material to achieve that channel thickness. Their appeal comes partly from beginning with dimensions that silicon manufacturers must work hard to create.

Quantum confinement can also change their electrical properties in useful ways. When electrons are restricted to movement in two dimensions, their minimum energy increases. This can widen the bandgap, the energy separation that helps determine whether a semiconductor allows current to flow.

A suitable bandgap helps a transistor switch between on and off states while limiting unwanted leakage. These properties make two dimensional materials candidates for technologies below the 5 nanometer node, with the potential for lower power use and less leakage than silicon at comparable dimensions.

Molybdenum disulphide and molybdenum diselenide have already shown promising characteristics. Both belong to a family called transition metal dichalcogenides and possess natural bandgaps. Producing these materials in laboratories has been relatively straightforward, but building practical circuits requires a complementary electrical partner.

The Need for Two Types of Semiconductor

Modern digital circuits generally combine two semiconductor types. Materials described as n-type carry current through electrons. Materials described as p-type carry current through holes, which are missing electrons in the crystal structure and behave as positive charge carriers.

These partners underpin complementary metal oxide semiconductor technology, known as CMOS. The architecture is the foundation of contemporary digital logic. Its basic NAND, NOR and NOT gates depend on the two types working together to perform the operations from which computation is built.

Researchers have had relatively accessible n-type materials for two dimensional electronics, including the molybdenum compounds. High performance p-type materials that also remain stable have been harder to produce. Zhu Mengjian described that shortage as a critical bottleneck and a fiercely competitive area of semiconductor research.

A strong n-type material alone cannot supply the complementary pair needed for these CMOS logic gates. The challenge is to find a p-type partner with suitable electrical performance that can also survive the handling, heating and chemical processing involved in making a chip.

Tungsten silicon nitride offers a possible answer because it naturally exhibits p-type behavior. It does not require external doping to obtain that behavior. Doping changes a semiconductor's electrical properties by introducing other substances, but the process can also introduce instability.

What Tungsten Silicon Nitride Offers

The experimental tungsten silicon nitride films showed high hole mobility and strong current density. Hole mobility describes how readily the positive charge carriers move through a material. Current density describes how much electrical current passes through a given area.

The films also showed strong mechanical properties and thermal conductivity. Mechanical strength matters because an extremely thin layer must remain intact through fabrication. Thermal conductivity matters because heat needs to move away from operating devices.

These characteristics suggest the material could withstand the multiple lithographic processing steps used to manufacture chips without cracking or separating from its supporting surface. Lithography defines patterns used to build the tiny features of a circuit. A useful semiconductor must retain its quality through that patterning and the processing around it.

The method also provides control over doping properties, allowing engineers to tune the electrical characteristics of the films. That adjustability is relevant to integrating two dimensional materials into existing CMOS architectures, where different components must operate together predictably.

Natural p-type behavior, adjustable electrical properties and physical durability address different parts of the same manufacturing problem. Researchers need a material that carries charge effectively, can be matched with other components and remains usable after fabrication.

Why Liquid Gold and Tungsten Change Growth

Chemical vapor deposition, or CVD, grows films by passing precursor gases over a heated substrate. Precursors supply the chemical ingredients from which the material forms. In conventional approaches, atoms gradually arrange into a crystal, making the growth of two dimensional semiconductor films painstakingly slow.

The Chinese team replaced the standard solid growth surface with a liquid gold and tungsten bilayer. The liquid gold surface removes the grain boundaries found in conventional solid surfaces and provides a uniformly smooth template for crystal formation.

Precursor atoms can move across this surface more rapidly than across a conventional solid substrate. That movement helps the atoms reach places where they can join the growing material. The smoother surface changes a central part of the growth process without relying solely on a faster supply of gases.

The underlying tungsten layer serves a second purpose: it supplies tungsten directly for the tungsten silicon nitride film. Providing an ingredient from beneath reduces dependence on delivery through the gas phase, which can limit growth rates.

Together, these changes enabled the reported acceleration and the production of monolayer films approximately 36 by 18 millimeters in size. A monolayer is a single structural layer of the material. The resulting films combined the electrical performance, durability and heat handling properties needed for further work on advanced chips.

Compatibility with existing semiconductor infrastructure could simplify future integration. The current demonstration still requires substantial engineering before it can become a process used across full production wafers.

A Related Advance Uses a Different Material

Professor Peng Hailin at Peking University has been working on a separate two dimensional transistor project using bismuth oxyselenide. His work also addresses architectures intended to move beyond the limits of conventional silicon designs.

Peng explained the importance of balancing the two semiconductor types needed for complementary circuits. His team's research reported speeds 40% faster than Intel's latest 3nm chips while using 10% less power.

The Peking University group described its approach as changing lanes rather than taking shortcuts. The comparison reflected its pursuit of a different material and device approach, rather than relying entirely on continued reductions in silicon dimensions.

The two projects tackle different parts of the effort to develop future electronics. The tungsten silicon nitride research focuses on faster material growth and the shortage of suitable p-type films. The bismuth oxyselenide project provides a separate example of transistor research aimed at performance and energy savings.

Research Advances Amid Strategic Competition

China's semiconductor research is advancing during intense competition over manufacturing capacity, computing technology and access to equipment. Data from the CAS Content Collection show a sharp rise in Chinese journal publications and patents related to microelectronics, including many intellectual property filings with a commercial focus.

The United States has committed $52.7 billion through the CHIPS and Science Act to strengthen domestic semiconductor manufacturing and research. Europe is also investing heavily through its own CHIPS initiatives as governments seek more secure and diversified supply chains.

Export controls introduced in October 2022 restricted China's access to advanced semiconductor equipment and computing chips. The restrictions aim to limit Chinese progress in AI and defense technologies and have also encouraged domestic research directed toward technological independence.

Alternative materials offer China a possible way to develop manufacturing platforms beyond conventional equipment routes. The liquid metal growth method is part of that wider research effort, with potential value for processing methods as well as for the particular semiconductor demonstrated.

The main dates and milestones described in the account are:

• October 2022: Export controls restricted China's access to advanced semiconductor equipment and computing chips.
• 2023: The baseline year used in forecasts of a possible 1,000 times increase in AI computing demand by 2030.
• March 26, 2026: The tungsten silicon nitride growth research appeared in National Science Review.
• April 16, 2026: The reported advance remained at the stage of experimental films and further manufacturing development.
• 2030: The forecast horizon for both the cited AI demand increase and a projected United States semiconductor workforce shortfall.

A Growing Market Still Demands Factory Readiness

The semiconductor equipment market was valued at approximately $111.84 billion in 2025 and is projected to reach $268.06 billion by 2034. The forecast reflects demand for advanced manufacturing, including technologies below 5 nanometers that require extreme ultraviolet lithography and new materials.

Asia-Pacific holds nearly 69% of the market, while Europe and North America are investing to reduce dependence on concentrated supply chains. Equipment growth creates a commercial setting for research into new manufacturing methods, although adopting a new material requires much more than a successful growth experiment.

Moving from laboratory demonstrations to commercial production typically takes years and involves materials scientists, process engineers and device physicists. Defects must become rare enough for billions of transistors to work reliably on a single chip.

New films must also be compatible with lithography, etching and metallization. Etching removes selected material, while metallization creates the metal connections between components. Each step must preserve the properties that made the semiconductor useful in the first place.

The liquid metal substrate approach could influence manufacturing methods even if another material eventually reaches mass production. Its central contribution is a way to change how atoms move across a growth surface and how an essential ingredient reaches the forming film.

What Must Be Solved Before Commercial Production

Scaling from films measuring 36 by 18 millimeters to production wafers measuring 300 millimeters across requires consistent quality over a much larger surface. Uniformity becomes increasingly difficult as the growth area expands. Enlarging a CVD chamber alone would leave the need to maintain precision at the atomic level across that area.

Heat management remains another concern. Tungsten silicon nitride has favorable thermal properties compared with some alternatives, but extremely thin materials have limited cross sections through which heat can flow. Densely packed transistors require effective cooling to preserve performance and avoid damage.

Mechanical strength also needs to survive repeated manufacturing steps. Etching, deposition and cleaning can place stress on delicate films. A material that performs well as a laboratory sample must retain that performance after the full fabrication sequence.

No timetable for commercial production was given. The next stage requires progress on larger growth areas, defect density, heat removal and integration with existing factory processes.

People, Water and Energy Also Shape Expansion

Semiconductor factories require skilled workers as well as suitable materials and equipment. The Semiconductor Industry Association projected that the United States alone could face a shortage of approximately 67,000 technicians, engineers and scientists by 2030.

Similar shortages affect manufacturing hubs worldwide. A lack of trained workers can slow the introduction of new technology even after researchers establish that it works, because factories need people who can operate, maintain and improve complex production processes.

Water and energy requirements also affect expansion. A modern fabrication plant producing wafers measuring 12 inches across can consume approximately 10 million gallons of water a day. Taiwan's severe drought in 2021 showed how water shortages can threaten semiconductor manufacturing.

Manufacturers are adopting recycling technologies and other sustainability measures, but chip production remains demanding in both water and energy. Building facilities for new materials brings those resource needs alongside the technical work of making reliable devices.

The Chinese team's result addresses a specific and valuable part of that larger challenge: growing a promising p-type semiconductor much faster. Commercial success will depend on carrying the material's speed, quality and electrical performance through the rest of the manufacturing process.

The Bottom Line

• Researchers from the National University of Defence Technology and the Chinese Academy of Sciences reported roughly 1,000 times faster growth of tungsten silicon nitride.
• The modified chemical vapor deposition process reached approximately 20 micrometers per minute and produced films measuring about 36 by 18 millimeters.
• A liquid gold and tungsten bilayer provides a smooth growth surface and a direct supply of tungsten.
• Tungsten silicon nitride naturally exhibits p-type behavior, addressing a shortage of stable complementary materials for two dimensional CMOS circuits.
• The research appeared in National Science Review on March 26, 2026, amid competition over advanced chips, equipment access and domestic manufacturing.
• Commercial production requires larger and more uniform films, low defect levels, reliable heat management and compatibility with existing fabrication steps.
• Workforce shortages and substantial water and energy needs remain practical constraints on semiconductor industry expansion.