Juan Perez-Mercader
Harvard University, MA·Chemistry
Today's artificial intelligence runs on silicon chips that use enormous amounts of energy. One alternative is to compute with chemistry itself: networks of chemical reactions whose changing concentrations act like signals, much as neurons do.
- Fields
Chemistry
Physics
Engineering
- Topics
chemical computing
chemical reaction networks
oscillating reactions
- Grants
- 1 known grant, through Jan 2029
What they work on
- Determine the smallest workable volume for computing chemical reaction networks, using picoliter and femtoliter reactor arrays to find the molecule count that keeps signal noise acceptable.
- Develop ways to couple oscillating chemical reaction networks, reactions whose concentrations rise and fall rhythmically, across separate microreactors by inhibiting or accelerating reaction paths.
- Build a physicochemical and condensed-matter physics description of the coupled reactors that links a chemical computer's inputs to its outputs.
- Develop methods for dynamically reconfiguring miniaturized chemical computing architectures, working with IBM Research partners who fabricate reactors by lithography and thermal scanning probe lithography.
More about the research
Many chemical computing demonstrations so far use flasks holding roughly 10^23 molecules per milliliter. It is not yet known how small such reaction networks can be made before random noise swamps the signal. It is also unclear how to feed in inputs, read out tiny amounts of chemicals quickly, and link reactors so they can do complex calculations.
Why it matters
Training and running AI on semiconductor hardware consumes resources at an enormous scale. Small chemical computing devices could offer a much lower-energy, smaller-footprint way to compute. If miniaturized reaction networks can be coupled reliably, they may eventually support neural networks trained on large data sets using a fraction of current energy use.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Closest labs
Raisul Islam
Purdue University, IN·Engineering
Fabricate and characterize the Diffusive Ferroelectric Field-Effect Transistor (DFeFET), combining engineered drain contact metals with amorphous oxide semiconductor channels in a ferroelectric-gated transistor to obtain controllable volatile hysteresis in drain current-voltage curves.
Shantanu Chakrabartty
Washington University in St Louis, MO·Engineering
Model single-electron transport across a Fowler-Nordheim quantum tunneling barrier as the equivalent of a neural spike in NeuroSA, a spiking architecture where neuron firing thresholds follow tunneling physics.
Eric Fullerton
University of California-San Diego, CA·Materials Science
Fabricate new quantum materials and characterize their properties with advanced techniques to understand and control their behavior for non-volatile memory.
The school
Harvard University: 42 labs on PI Pages, $1.5B a year on research, 712 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.