Researchers at Pennsylvania State University have developed a hybrid electronic memory that, by combining synthetic DNA and perovskite semiconductors, can perform data storage and processing in one location. According to the researchers, this technology requires about one-hundredth the energy compared to similar samples and could help develop a new generation of low-power hardware for artificial intelligence in the future.
DNA; an ultra-dense memory for data storage
DNA is not only the carrier of genetic information for living organisms, and because of its extremely high information density, it has long attracted the attention of researchers in the data storage field.
According to theoretical estimates, one gram of DNA can hold about 215 million gigabytes of data. Such capacity has made DNA an attractive option for storing the massive volumes of data that grow daily with the expansion of artificial intelligence and digital services.
However, using DNA in electronic devices faces a fundamental barrier: DNA is a biological material and its electrical properties differ greatly from the materials used in electronic circuits.
Researchers at Pennsylvania State University created a hybrid structure to solve this problem that combines two different materials: synthetic DNA and crystalline perovskite.
Combining synthetic DNA with perovskite
The synthetic DNA used in this research is made of designed chemical molecules. Unlike natural DNA, the length and structure of these short chains can be controlled more precisely at the molecular scale.
The second material is crystalline perovskite; a type of semiconductor that is also used in fields such as solar cells and lasers.
The main idea of the research is to combine DNA's ability to form dense, designable structures with the electronic properties of perovskite. The result is a hybrid material that can be used to build low-power electronic memories.
But for synthetic DNA to play a more effective role in conducting current, the researchers functionalized it with silver nanoparticles. This process, known as functionalization, helped increase the conductivity of the DNA and better organize its molecular units.
After that, the silver-modified DNA was combined with perovskite membranes and a structure formed that allowed electric current to pass within the device.
A memristor that remembers its previous state
The researchers used this hybrid structure to build a memristor; an electronic component that can retain its previous state and create a type of memory in a circuit. In this device, changing the direction of the electric current changes the memristor's state. The important point is that the device can retain this state even after the power source is cut off.
This feature can be highly significant in new computational architectures; because in such an architecture, data can be stored in the same place where it is processed and there is no longer a need to constantly move information between memory and processor.
Reducing this data movement can lead to lower energy consumption and increased efficiency of computing systems.
Stable performance at high temperature
Experiments on this device showed that electrons can reliably move in the hybrid structure with an applied voltage of less than 0.1 volts. The device also showed predictable responses when the current direction changed.
Thermal stability was another notable result of these experiments. According to the researchers, the device was able to maintain its performance at about 121 degrees Celsius and continued operating at room temperature for more than six weeks.
About 100 times lower energy consumption
The most important finding of this research relates to energy efficiency. The researchers say their hybrid memristor can perform similarly to comparable technologies with about one-hundredth the energy; in other words, its energy consumption is roughly 100 times lower.
The researchers also believe this structure can, in some cases, surpass traditional storage technologies like flash in terms of storage capacity.
If these features are preserved at larger scales, the new technology could become important for data centers and heavy computing systems; especially in conditions where the growth of artificial intelligence has led to a sharp increase in data volume and energy consumption.
Why is this technology important for artificial intelligence?
One of the main challenges of hardware used in artificial intelligence is moving massive amounts of data between memory and processor. Each data transfer requires energy, and as computational volume increases, the energy cost of this movement also rises.
For this reason, researchers in recent years have been seeking architectures that bring memory and processing closer together or even perform both processes in the same place.
Hybrid memristors can be considered one of the potential options for neuromorphic computing in this area. This field attempts to make computing system architectures more similar to the way neural networks in the brain operate.
Still some distance from entering computers
Despite promising results, the new technology is still in early development stages and cannot be considered a ready replacement for today’s memories.
The researchers intend to improve the device's performance, further study its stability and capabilities, and find other applications for using bioinspired materials in electronic devices.
The significance of this research is not limited merely to “using DNA in computers.” The main innovation is using synthetic DNA as part of a hybrid electronic material; a structure that can combine DNA’s dense storage properties with the electrical capabilities of semiconductors.
If subsequent research succeeds, such an approach could open a new path to building low-power memories and hardware suitable for next-generation artificial intelligence; a path in which biology, materials science, and electronic engineering come closer together than before.