Chen Zhu Research Group Eastern Institute of Technology, Ningbo
Wet-lab organic synthesis makes and measures. Dry-lab computation explains and predicts. AI and automation decide what comes next — AI for Science, closing the loop on itself.
The discovery loop
The laboratory runs as one closed circuit of making, understanding, and deciding. Automated stations generate real wet-lab data — yields, selectivities, spectra measured at the bench, not simulated. Computation turns that evidence into mechanisms and descriptors, and an active-learning engine proposes the next experiment — then learns again from what the bench sends back. Each pass through the loop trains the one that follows — a self-driving iteration that compounds, and a change in how research itself is run.
measureexplainchoose
Selected conditions return to the bench without a human hand-off; fresh data updates the model, and the loop turns again — each pass better informed than the last.
Three ways into the same question
We ask how organic synthesis can become more selective, sustainable, and predictive. The answer is not a stack of separate specialties but a working exchange between new reactivity, molecular explanation, and automated learning.
We build synthetic methods around the cleanest reagents chemistry has — photons and electrons. Transition-metal catalysis under photochemical, electrochemical, and photoelectrochemical drive assembles C – C and C – heteroatom bonds under mild conditions, with selectivity set by catalyst, ligand, and how the energy is delivered. Alternating-current photoelectrocatalysis extends that control into the time domain.
Computation is how the group understands its own chemistry. DFT and wavefunction analysis map catalytic cycles, locate transition states, and expose the electronic origins of reactivity and selectivity — explaining what experiment observes and predicting what it should try next. The same calculations distill molecules into descriptors that machine-learning models can reason over, welding theory to the automated loop.
The condition space of modern synthesis is far too large for one-variable-at-a-time habits. We are building an autonomous, self-driving laboratory — robotic stations running reactions and analyses in parallel, results streaming back as structured data, active-learning algorithms choosing the experiment that carries the most information. It is AI for Science made concrete: a dry – wet loop that changes not just what we discover, but how discovery itself is done.
Latest from the lab
2026 · 07
New work accepted in J. Am. Chem. Soc.
2026 · 05
Phase 1 of the automated dry – wet platform comes online.
2025 · 09
AC-PEC asymmetric coupling published in Nature Synthesis.
Selected publications
Three peer-reviewed results show the range of the method: controlling time, explaining mechanism, switching stereochemistry.
Nature Synthesis2025 · 4 · 1534
The first alternating-current-driven asymmetric cross-couplings: periodic polarity reversal suppresses electrode deposition of the metal catalyst, delivering chiral C–C and C–heteroatom bonds with up to 99% ee: the founding work of the group’s AC time-domain control.
Nature Chemistry2026 · 18 · 656
Programmable electrochemistry opens and multifunctionalizes strained rings; full-mechanism DFT modeling proposed and computationally validated the olefin slow-release pool mechanism.
Nature Catalysis2019 · 2 · 678
Nickel catalysis paired with single-electron and triplet-energy transfer builds stereodefined multisubstituted olefins; photocatalyst triplet energy switches the E or Z outcome.
Bring the next question into the loop.
We welcome researchers who want to connect rigorous organic chemistry with computation, AI, and laboratory automation — and collaborators whose methods can make that loop more capable.