Atomic /Electro-Chemical Switches - Pro

• TiOx switching is controlled by oxygen vacancies• What can we do with these devices: Memory?• How reproducible are these switches?• How fast do they switch?• ……• Adjunct technology in context with CMOL or other

hybrid approaches

• Net: This was a research report but not a summary on the opportunity or capability of these devices!

• Metric missing

Atomic /Electro-Chemical Switches - Con• Mechanism: Ionic transport plus chemical reaction (the

slower process determines the switching speed) -> formation of atomic bridge

• Application: Memory, logic not quite clear– Switching time slow ~ 100ns (reported in array 4-30us, 1.1V;

50ns, 0.6V)– Cycling low 1E11– Data situation confusing to allow projection (not quite clear what is

intrinsic material property and what is process related)• Challenge: reliability, switching speed, endurance• Ion mobility small number limits migration time ~ 1ns

• New logic gates required for this device: fabrics need to be integrated with I/Os, density requires default tolerant architectures, what happens with defect driven lkg in the fabrics?

Atomic /Electro-Chemical Switches - QA

Q: If diffusion is grain boundary what about scalability

A: not quite clear if it isQ: Switching voltage pretty high about a voltA: Yes they are high. Not quite clear if this can be

reduced and maintain a high on/off ratioQ: Will power in these arrays dominated by array

and not by overheadA: Depends on array architecture