Access Official Technical PDF

Share Summary:LinkedInEmail

Autonomous materials development applies artificial intelligence (AI) and robotics to parts of the materials research process, with the aim of exploring experimental possibilities that would be difficult to assess manually. At the Samsung Advanced Institute of Technology (SAIT), this work sits within a broader research portfolio spanning semiconductors, devices, materials, health sensing and machine intelligence. For industrial automation leaders, the effort offers a useful view of how data-driven methods and laboratory automation may reshape research and development (R&D).

The goal is not simply to automate repetitive laboratory work. It is to connect experimental results with algorithms that can help select what to investigate next, then use synthesis tools to carry out experiments. That combination could expand the range of materials researchers can examine while allowing human specialists to focus more of their time on questions requiring creativity and judgment.

How autonomous materials development could work

SAIT’s autonomous materials development project is focused on an AI algorithm trained on experimental datasets. The intended next step is for the algorithm to drive robotic synthesis tools. In this model, experimental data informs algorithmic decisions, while automated synthesis makes it possible to test selected material candidates.

Materials research often involves considering many combinations of elements and configurations. Exploring that space presents a substantial analysis challenge for human researchers. AI could process large datasets and search experimental spaces that people may not have considered. The approach is promising, but it is presented as a development effort rather than a fully autonomous research capability already in routine use.

SAIT’s stated long-term ambition is for machines to undertake all synthesis, potentially within 10 to 20 years. If that goal is achieved, researchers could concentrate more on work that depends on superior creativity. The practical value will depend on how effectively the algorithm learns from experimental results and how well it connects with synthesis equipment; the available description does not specify a completed, end-to-end system or its performance.

From AI models to advanced materials

The project builds on a research program that has produced work in quantum dots, two-dimensional materials and other nanostructures. These examples show why material discovery matters beyond the laboratory: material properties can influence the design and performance of electronic components, displays and energy-storage technologies.

Quantum dots and semiconductor materials

Quantum dots illustrate one path from materials research to commercial products. SAIT researchers contributed to the development of cadmium-free quantum dot technology used in Samsung QLED televisions, which reached the market in 2015. The material’s sharply defined colour spectrum supports the displays’ colour characteristics. The broader research potential identified for quantum dots includes optoelectronic applications such as photovoltaics and sensors, as well as investigation of electroluminescence.

SAIT researchers are also investigating two-dimensional materials, including graphene, as possible alternatives to conventional silicon-based approaches in some applications. The research team reported extending graphene layers to an area on the scale of computer chips and demonstrating barristor arrays fabricated on a silicon wafer. Such work addresses the challenge of improving electronic components as their dimensions approach physical limits.

Another example is amorphous boron nitride, which researchers encountered while working with crystalline boron nitride. Rather than discarding the unexpected material, the team measured its properties and reported an ultra-low dielectric constant alongside robust electrical and mechanical properties. The finding demonstrates how experimental observation can redirect a research path—an important consideration when designing systems intended to learn from data and guide experiments.

Automation within a wider research ecosystem

Autonomous materials development is one strand of SAIT’s research, not a standalone solution to every R&D challenge. The institute is also exploring neuromorphic computing: electronic systems designed to reflect aspects of biological neural organization. This research seeks to capture brain-like network capabilities that differ from the strengths of conventional artificial neural network processors, particularly in learning and cognition.

To investigate biological neural networks, researchers developed an array of approximately 4,000 vertical nanoelectrodes controlled by a complementary metal-oxide-semiconductor (CMOS) chip. The system recorded activity inside approximately 1,700 rat neurons, supporting a map of about 300 excitatory and inhibitory synapses. The research team’s next stated objective is to use intracellular recording data to drive a non-volatile memory network that could encode synaptic connectivity.

Other research areas connect materials science with energy and health. SAIT has studied graphene structures for lithium battery applications and is pursuing all-solid-state batteries, which replace a liquid electrolyte and separator with a solid electrolyte. The institute has also developed wearable sensing approaches, including cuffless blood pressure measurement using pulse waveform analysis and non-invasive blood glucose sensing based on an optical skin spectrum. These efforts reflect a broader interest in technologies that combine materials, electronics, data analysis and practical applications.

What industrial R&D teams can take away

For engineering and operations organizations, the wider lesson is about system integration. A research automation program may need to connect data collection, analytical models, experimental equipment and material characterization. The SAIT example highlights several considerations:

  • Build on experimental data: AI-driven research depends on datasets that can inform algorithm development and guide investigations.
  • Connect decisions to physical processes: The proposed workflow links algorithmic analysis with robotic synthesis rather than treating AI as a separate planning tool.
  • Preserve human expertise: The stated ambition is to shift researchers toward work requiring creativity, not to remove the human role from R&D.
  • Allow for unexpected findings: The amorphous boron nitride example shows why experimental results should be assessed rather than dismissed when they diverge from expectations.

Conclusion

Autonomous materials development represents an ambitious use of AI and robotics: learning from experimental datasets, searching a broad range of material possibilities and directing synthesis tools. SAIT’s program remains a research effort with a long-term vision, while its wider work demonstrates how materials innovation can intersect with semiconductors, batteries, sensing and neuromorphic systems. For industrial R&D teams, the practical implication is to think of automation as an integrated research workflow—one that combines data, equipment and human expertise to explore problems at a scale that is difficult to achieve through manual analysis alone.

Access Official Technical PDF

About The Author