A reconfigurable brain-inspired memory device shows promise to simplify semiconductor circuit design, enhance computational power and speed, and advance edge computing.
In an era of digitalisation, many believe that we are losing our human touch. Ironically, the technologies that seem to make us less human are becoming increasingly human themselves. Today, many commonly used electronic devices are equipped with semiconductor logic circuits that are meant to mimic the decision-making process of the human brain. However, even state-of-the-art circuits have yet to reach, much less, surpass the sophisticated decision-making enabled by the complex interconnections of our brain’s neurons. The functions of man-made circuits are limited to basic ones and are based on switches that can only perform predefined logic functions.
To advance the performance of logic circuits, physicists from the National University of Singapore collaborated with an international team of researchers to develop a molecular memristor that is energy-efficient, capable of enhanced computational power and speed, and has exceptional memory reconfigurability. A memristor is an electronic memory device that is based on memory and resistor functions whereby its resistance can be programmed (resistor function) and subsequently remains stored (memory function). Unlike standard hard-wired circuits, the molecular device can be reconfigured using voltage to embed different computational tasks.
Drawing inspiration from neuronal interconnections, the researchers reimagined the design strategy of a logic circuit and fundamental electronic circuit elements. Research team member Dr. Sreebrata Goswami conceptualised and designed a molecular system belonging to the chemical family of phenyl azo pyridines that have a central metal atom bound to organic molecules called ligands. He explained that “these molecules are like electron sponges that can offer as many as six electron transfers resulting in five different molecular states. The interconnectivity between these states is the key behind the device’s reconfigurability.”
With this new system, first author Dr. Sreetosh Goswami created a tiny electrical circuit that consists of a 40-nanometre layer of molecular film and sandwiched the circuit between a top layer of gold and a bottom layer of gold-infused nanodisc and indium tin oxide. Upon applying a negative voltage to the device, he was surprised to find an unprecedented current-voltage profile. Further investigations revealed that these organic molecular devices were unlike conventional metal-oxide memristors that are usually switched on and off at one fixed voltage. Their device could switch between on-off states at several discrete sequential voltages.
To understand the reason behind this difference, the team used Raman spectroscopy to analyse the spectral signatures in the vibrational motion of the organic molecule. According to Dr. Sreebata Goswami, the multiple transitions occurred because “sweeping the negative voltage triggered the ligands on the molecule to undergo a series of reduction, or electron-gaining which caused the molecule to transition between off and on states.” They were able to analyse the molecules’ behaviour with the help of a decision tree algorithm with “if-then-else” statements.
After getting a good grasp of the underlying principles of the molecular system, the team further built upon Dr. Goswami’s findings by using the molecular memory devices to run programs for various real-world computational tasks. In their proof-of-concept, the researchers demonstrated that the technology could perform complex computations using only one step and could be reprogrammed to perform a different task immediately after. With only one molecular memory device, they were also able to perform the same computational functions that usually require thousands of transistors, thus making the novel technology a more powerful and energy-efficient memory option.
“Similar to the flexibility and adaptability of connections in the human brain, our memory device can be reconfigured on the fly for different computational tasks by simply changing applied voltages. Furthermore, like how nerve cells can store memories, the same device can also retain information for future retrieval and processing,” explained Dr. Sreetosh Goswami.
Besides its reconfigurability, Associate Professor Ariando also believes that “this new discovery can contribute to developments in edge computing as a sophisticated in-memory computing approach to overcome the von Neumann bottleneck, a delay in computational processing seen in many digital technologies due to the physical separation of memory storage from a device’s processor.”
Having demonstrated the powers of their new device, the team is excited about the potential applications of their molecular memory system. Currently, they are in the midst of building new electronic devices that can incorporate the system and is working with collaborators to conduct simulation and benchmarking related to existing technologies, like cell phones and sensors. [APBN]
Source: Goswami et al. (2021). Decision trees within a molecular memristor. Nature, 597, 51-56.