The molecular component mimics the flexibility of the brain and could revolutionize treatments

electronics

Technology Innovation Website Writing – 09/03/2021

The single molecule stores multiple data, performs calculations and learns in the process with minimal energy consumption.
[Imagem: NUS]

artificial clamp

All computers and other electronic devices rely on logic circuits built with transistors connected together to perform predefined logic functions.

Now imagine a circuit in which each core component can be configured in real time, and can act as memory and computation, eliminating the separation of processor and memory, and still learning in a similar way to the synapses in the human brain.

Since there is no need for imagination, since an international team of researchers has just developed it, a new electronic component that serves as a memory and a processor can be reconfigured, learned and even created by a single molecule, reducing its energy consumption to a minimum.

“Like the flexibility and adaptability of connections in the human brain, our memory device can be rapidly reconfigured in time of flight for different computational tasks simply by changing the voltages. [eltricas] applied. Furthermore, just as neurons can store memories, the device itself can also hold information for future retrieval and processing.”

This is a giant step towards Membership memories created by the team last year, which in itself was surprising.

Leap in the field of molecular electronics.
[Imagem: Sreetosh Goswami et al. – 10.1038/s41586-021-03748-0]

molecular memory

Technically, the new component is a molecular memory – also known as synthetic synapses, the diary can perform calculations and “remember” the data that was previously stored in it.

The big difference is that it consists of a single molecule, consisting of only 77 atoms, belonging to the chemical family of phenyl azopyridine, which has a central metal atom bonded to organic molecules called bonds. Complex logic can be fully encoded in the properties of each molecule, creating a new basic electronic circuit component that is not limited to 0 and 1 second storage, like transistors.

“These molecules are like electron sponges, which can present up to six electron transfers, resulting in five different molecular states. The interconnection between these states is key behind reconfiguring the device,” Goswami explained.

And unlike traditional metal-oxide memristors, which are only turned on and off at a constant voltage, these new organic molecular components can switch on and off states at many separate series voltages.

The molecule uses a natural asymmetry in its metal-organic bonds to switch explicitly between different states, allowing it to perform ultra-fast decision-making functions. Goswami explains: “The negative voltage variation induces the molecule’s bonds, causing them to go through a series of reductions, or electron gains, causing the molecule to transition between on and off states.”

Molecular masculine behavior is directly described in the form of decision trees, as in computer programs.
[Imagem: Sreetosh Goswami et al. – 10.1038/s41586-021-03748-0]

neural computing engineering

The team then set out to explain how organic molecular notes work in the world of electronics and computing.

What they actually demonstrated is a new type of neural computing architecture – inspired by the brain – created by optimizing the electrical properties of the resulting crystals from their molecules.

The molecules were dispersed in a microcircuit consisting of a 40-nm-thick molecular membrane layer sandwiched between a top gold layer and a bottom layer composed of a gold-indium tin oxide (ITO) nanodisc doped. When a negative voltage is applied to the circuit, it exhibits an unprecedented voltage-current profile, with multiple transitions caused by the vibrational motion of the organic molecule.

The researchers describe this behavior of molecules using the decision tree algorithm in the “if-then-else” terms typically used in computer program coding — as opposed to the traditional approach of using fundamental physics-based equations.

With this, the team used molecular memory components to run programs for various computational tasks in the real world. As a proof of concept, the team showed that their technology can perform complex calculations in one step and can be reprogrammed to perform another task in the near future.

A single molecular memory circuit can perform the same computational functions as thousands of transistors, making the technology a more robust and energy-efficient memory option.

The team is already working on building more complex electronic circuits and working with collaborators from other universities to conduct simulations and evaluations of how their new architecture will perform against existing technologies.

Other news about: