skip to main content


Human scientific understanding is encoded in data relationships, and discovery of new relationships can be accelerated with judicious incorporation of automation and artificial intelligence (AI) in scientific research


Develop experiment and data science tools that facilitate the unification of different approaches to scientific discovery. For a given science goal, we analyze what type and size of data would enable us to link with complementary research methods (most commonly computational materials science), analyze the underlying physics of experiments that can provide that data, and design new instruments and/or modes of operation to deliver that data with the requisite balance of throughput and quality.

Target Technologies

The High Throughput Experimentation team is based in the Liquid Sunlight Alliance following its founding in the Joint Center for Artificial Photosynthesis (JCAP) to accelerate the discovery of materials for incorporation in a new generation of solar fuels generators. We also explore fuel cell electrocatalysts with the Toyota Research Institute (TRI-AMDD). Our other projects with DOE, DOD-AFOSR, and Google Accelerated Science are focused on identifying new, and hopefully better, ways to conduct materials science research.



The group was founded as part of the first solar fuels DOE energy innovation hub, the Joint Center for Artificial Photosynthesis (JCAP). The capabilities developed in that program are being deployed, improved, and expanded in the Liquid Sunlight Alliance (LiSA), a new energy innovation hub. This work involve the accelerated screening of (photo)catalyst materials to establish the basic science of solar fuel generation of liquid fuels. Our research includes a variety of projects starting with technique development and continuing with deployment for discovery of electrocatalysts, photoelectrocatalysts, and integrated materials photoelectrode. The choice of techniques and their deployment is guided by the above Strategy, resulting in team science that includes collaborations with the Agapie, Peters, Bajdich, Neaton, Persson, Yano, Toma, Cooper, and Drisdell groups (see Collaborators). For photoanodes, metal oxides comprise the most often studied class of materials in our LiSA research due to their oxidative stability, a practical consideration for solar fuels anode materials, and their enticing complexity, a strategic choice for accelerating the discovery of data relationships. For CO2 reduction to fuels, we study metal alloy catalysts and their modification with molecular coatings.


Our project with TRI features co-PI Carla Gomes (Cornell) whose prolific computer science career and innovative team empowers our development of new AI algorithms inspired by outstanding problems in materials research. We also work closely with TRI scientists Santosh Suram, Linda Hung, and the broader TRI-AMDD team to develop new tools for identifying materials with desired properties, leverage data resources, and benchmark machine learning algorithms.


We have also teamed with Carla Gomes and Jeff Neaton (Berkeley, Lawrence Berkeley National Laboratory) for the DOE project Energy Materials Chemistry Integrating Theory, Experiment and Data Science (EM-CITED), a multidisciplinary research effort focused on accelerating discovery of scientific knowledge via incorporation of data science and artificial intelligence in materials chemistry. EM-CITED has 3 focus areas: (i) Open Representations of Energy Materials, (ii) Energy Materials Synthesis Prediction, and (iii) Catalyst Evolution Prediction and Classification.

Carla Gomes is also the director of the Computational Sustainability Network where John Gregoire is a co-PI and works with the Gomes group to identify materials-inspired algorithms that can be deployed, and algorithms from other domains that can be deployed in materials science.


The Scientific Autonomous Reasoning Agent (SARA) project is led by PI Bruce van Dover (Cornell) and aims to incorporate theory guidance in a closed-loop experimental system, all guided by a network of AI "agents" that tackle different aspects of the materials research cycle. The multidisciplinary team includes co-PIs Mike Thompson (Cornell), Carla Gomes (Cornell), Bart Selman (Cornell), Alex Zunger (Colorado), Chris Wolverton (Northwestern), and John Gregoire.


An ongoing collaboration with Google Accelerated Science involves application of data science to design experimental protocols and automated interpretation and quality control, all in the absence of human bias.