Molybdenum disulphide two-dimensional semiconductor memory transistors are possible, and might be used to build logic-in-memory processors using a fabrication process that can be scaled, according to the Laboratory of Nanoscale Electronics and Structures (LANES) at Swiss lab EPFL, which has built a chip around floating-gate fets, similar to those used in flash memory.

“The unique electrical proprieties of MoS2 make it particularly sensitive to charges stored in floating gate fets, which is what enabled the LANES engineers to develop circuits that work as both memory storage units and programmable transistors. By using MoS2, they were able to incorporate numerous processing functions into a single circuit and then change them as desired,” according to EPFL.
In the research device, multiple distinct states were stored on the device for hours, and the scientists estimate 10 year data retention for two state operation would be possible.
EPFL-MoS2-logicInMemoryIn more detail (looking at the diagram) , the transistors are built on a silicon wafer (white) with a SiO2 insulating layer (yellow). The transistor has a metal bottom gate (also white), then a metal floating gate (red) separated by a 30nm HfO2 blocking oxide. The channel (green Mo and yellow S) is separated from the floating gate by a 7nm HfO2 tunnelling oxide. Drain and source are yellow and offer ohmic-like contact.

According to the Nature paper ‘Logic-in-memory based on an atomically thin semiconductor’, the fabrication techniques are scalable – no exfoliated (flaked and deposited) materials were used.
The team built simple logic gates (two-input NOR and three-input NOR) from two or three transistors, which can be combined to create many other Boolean functions. The went on to propose a two-input logic-in-memory unit cell capable of acting as a universal logic gate, performing any logic operation from a complete set of two-input logic operations, according to the paper, and be extended to more complex operations, such as the addition of two numbers using a half-adder.
Logic-in-memory processing moves away from traditional architectures like von Neumann where memory and logic are separate.
“This ability for circuits to perform two functions is similar to how the human brain works, where neurons are involved in both storing memories and conducting mental calculations,” said LANES head Andras Kis. “Our circuit design has several advantages. It can reduce the energy loss associated with transferring data between memory units and processors, cut the amount of time needed for computing operations and shrink the amount of space required. That opens the door to devices that are smaller, more powerful and more energy efficient.”