Industry 1

Semiconductors & advanced materials

Process-integration and materials R&D leads at a fab or a materials supplier: the photolithography, precursor and packaging-materials groups.

Written for

Process-integration and materials R&D leads at a fab or a materials supplier: the photolithography, precursor and packaging-materials groups.

The decision

Which formulation or process variant goes to the next wafer lot.

How it is done today

Synthesis plus wafer runs, with DFT and vendor data as a guide.

When it goes wrong A wrong lot costs weeks of tool time and a missed node milestone.

Applications

ApplicationYou giveYou getStatus
EUV photoresist candidate ranking

Proof: Four of four formulations ranked correctly, every gap within 0.3% of the exact reference, on IBM Heron

Candidate resist structuresA ranked list with the energy gap each ranking rests onLive on quantum hardware
Precursor and dopant chemistry (ALD/CVD precursors, dopant activation)Candidate moleculesA ranking with binding and activation energiesReady · computed, hardware next
Low-k and high-k dielectric and interconnect materialsCandidate compositionsA ranking with the smallest difference it resolvesReady · computed, hardware next
Defect and yield analysis across a tool chainThe process dependency mapWhich steps fail together and where excursions concentrateMapped
Fab scheduling and lot dispatch

Proof: Production scheduling 6 h → 1 min, in daily use at a chemicals plant

Lots, tools, recipes and due datesA re-plan in minutesDeployed through the platform network
Lithography process-window explorationA process landscapeThe settings that hold the specMapped

What changes for you

  • Candidates are ranked before synthesis, with the energy difference each ranking rests on stated beside it.
  • The chemistry model trains in minutes instead of 173 GPU-hours, so a screen runs inside one planning cycle.
  • Rankings are read as same-run comparisons, so they hold at sub-percent energy differences, below what wafer data resolves.
  • A full-size run uses the whole processor: 26 working units carrying 1,448 entangling operations on a 156-qubit chip.

First engagement

You send Two to twelve candidate structures, or the active space from the package you already run.

You get back The ranking, the gaps, and whether your DFT already settles it. Two weeks.

Proof · EUV photoresist candidate screening

Hardware result · award, 2nd place

4/4

candidate formulations ranked correctly on a quantum processor; energy differences within 0.3% of the exact reference

Measured on quantum hardwareClassical computationJudged resultPlatform IBM Heron, 156 qubits

Decision

Which photoresist formulation goes to wafer testing

What is modelled

An industrial tin-oxo photoresist cluster and its counter-anion variants, as a 40-qubit quantum chemistry model.

Operational outcome

  • Ranked candidate formulations before synthesis
  • Fewer wafer runs on candidates that will not win
  • Model training in minutes instead of 173 GPU-hours
Problem size
40-qubit chemistry model · ranking run on 156 qubits · 120 qubits in a single job
Scope
A ranking of candidates; it does not predict fab performance.

Results, each with its baseline

  • 4 of 4 photoresist formulations ranked in the correct order on a quantum processor, energy differences within 0.3%. Baseline: the exact reference energy for every candidate.Measured on quantum hardwareWhere IBM Heron, 156 qubits
  • 40-qubit quantum chemistry on the industrial tin-oxo resist cluster, within chemical accuracy of the reference at every step. The published method runs at about 20 qubits.Classical computationMeasured on quantum hardwareWhere GPU computation with hardware sampling
  • Model training: 173 GPU-hours → minutes on one GPU, for the same challenge method.Classical computationWhere one GPU
  • Award winner, 2nd place worldwide, GIC 2026 Advanced Materials track (Mitsubishi Chemical Group × AIST). A judged result; it does not imply endorsement of any product.Judged resultWhere GIC 2026 Advanced Materials track
  • Full-processor runs on a 156-qubit superconducting processor, and 126 qubits in a single job, at a fixed circuit depth of three entangling layers at every width.Measured on quantum hardwareWhere 156-qubit and 126-qubit superconducting runs
  • 144-atom quantum dynamics on a neutral-atom processor: measured scaling exponent 0.382 ± 0.105 against a predicted 0.386; the classical simulator we ran did not converge at bond dimensions up to 128.Measured on quantum hardwareWhere 144-atom neutral-atom processor
Bare cluster3.49 eVSulfonate3.34 eVCarboxylate3.01 eVHydroxide2.21 eV
Four resist formulations ranked before synthesis by predicted electronic gap (lower activates more easily under EUV light). The ranking held on quantum hardware (IBM Heron, 156 qubits); every difference within 0.3% of the exact reference.
60%70%80%90%100%74.0% · 12q90.4% · 40q
Accuracy that improves with scale: 74.0% of the target at 12 qubits, 90.4% at 40 qubits, with circuits kept to three entangling layers. GPU computation plus hardware sampling.

Benchmark

Semiconductors & advanced materialsEUV photoresist screening

Published reference

Published device runs on the same photoresist chemistry: 14 and 22 qubits.

Sources: arXiv:2607.23988 (2026)

Merlin

4/4 candidates ranked correctly on hardware · 40-qubit chemistry model · training 173 GPU-hours → minutes

Measured on quantum hardwareClassical computation

Baseline

Exact reference energies; the published method's training cost

Read the story

Your industry, next

Bring us the decision your industry has to make next.

A resist, a grid plan, a controller, a cipher: you get the answer, the method it was checked against, and its status. The free check runs in your browser.

No keyNo cardAn answer in about a second