Electron Beam Lithography Acceptance Test Plan: Evidence Before Handover

1. Freeze acceptance scope before installation

Define what the facility expects to write: critical dimension, pitch, pattern area, layer count, overlay, substrate, resist, workload, vacuum and metrology. Separate platform acceptance from process-development targets. A supplier may need to optimize a new resist, but the handover decision should still have a clear minimum evidence package.

Identify the exact ZEL configuration, source, voltage, current, write-field list, stage, generator, software and optional functions. The same model name can cover different settings, so the plan should carry the configuration record into every coupon and raw file.

2. Design coupons that expose each risk

Use isolated and dense lines, equal line/space, a large-area pattern, field-crossing structures, alignment marks, charging test areas and a representative device fragment. Include a layout that stresses the expected write field and stage travel. The coupon should be small enough to repeat but rich enough to reveal field, process and metrology errors.

Record layout revision, file conversion, orientation and layer names. If a discharge layer or conductive substrate is required, include it in the coupon and in the process record. A coupon that omits charging or field stitching cannot accept those functions.

• Resolution and density
• Field stitching and overlay
• Charging and substrate
• Representative area and workload

3. Lock beam and resist conditions

State acceleration voltage, beam current, aperture, spot or image-resolution condition, dose, step or dwell, resist, thickness, bake, developer and substrate. Request SEM images with scale and process conditions. A quoted linewidth without those conditions is not a reproducible acceptance result.

Use the facility's intended resist or label an alternative clearly. Higher voltage, current and dose trade resolution, damage and write time; the plan should test the operating point the lab will actually use rather than the supplier's best demonstration setting.

4. Verify field calibration, stitching and overlay

Expose marks across a single field and across field boundaries. Measure field distortion, stitch residual and overlay with the agreed metrology. Repeat in multiple locations and, for multilayer work, after the second exposure. Record calibration file, stage coordinate, mark design and software revision.

The ZEL304G documentation lists a maximum field and field-calibration, stitching and overlay functions; acceptance should show the result under the buyer's pattern class and substrate. A best-case field image does not close a large-area or multilayer requirement.

5. Test charging, vacuum and thermal state

Use the actual substrate and discharge-layer plan. DTU Nanolab guidance emphasizes conductive grounding or a discharge layer for insulating substrates; include that layer, removal step and process impact in the acceptance record. Monitor beam stability, pattern placement and defects when the coupon charges or when the process changes.

Record pump-down, base pressure, outgassing or conditioning, stage temperature and warm-up. Vacuum recovery and thermal state affect both reproducibility and throughput. A facility should know when the tool is ready for a production-representative job, not only when the chamber door closes.

6. Measure write time on a reference workload

Define a reference job with pattern area, density, dose, beam current, field, correction, stage moves and number of layers. Time file preparation, loading, alignment, exposure, unload and inspection separately. A sparse coupon can hide the cost of a dense device or proximity-effect correction.

Compare the result with the facility's expected workload and schedule. Record whether the supplier used a production-representative generator and correction setting. Throughput is an acceptance item when tool time is a capacity constraint, even if resolution is the primary technical goal.

7. Run repeatability and fault recovery

Repeat the coupon after reload, warm-up, a second operator and a controlled restart. Test alarms, beam blanking, stage stop, file resume and safe recovery. Keep failed results; they reveal whether the process or tool is robust. Define how a failed measurement is repeated and who owns the corrective action.

Handover should include calibration backup, recipe, job files, maintenance schedule, training record and service contacts. Any software or hardware change after acceptance should trigger a defined re-test rather than an informal sign-off.

8. Sign only a traceable handover

The final record should include scope, configuration, coupons, process conditions, raw files, measurement tables, uncertainty, pass/fail, open items, exceptions, training and requalification triggers. Distinguish guaranteed limits from typical or process-dependent results and keep the source documents with the tool file.

SENFU can review an EBL acceptance plan against the documented ZEL304G functions and the facility's device stack. A complete handover makes the first production job a controlled continuation of the demonstration rather than a new experiment.

9. Define data ownership and requalification

The acceptance record should say where job files, recipes, calibration backups, SEM images, metrology tables, software versions and maintenance logs are stored and who may change them. Keep a configuration snapshot with the accepted source, generator, stage, write-field settings and optional functions. If the facility cannot reproduce a result from the archived data, the result is difficult to defend during a service call or a process transfer.

Set requalification triggers before handover: source or column service, stage or chuck change, software update, calibration file replacement, vacuum intervention, new resist or substrate class, or a major recipe change. A focused re-test can be proportional to the change, but it should cover the affected risk and compare with the original control coupon. This gives procurement a lifecycle plan and lets the nanofabrication team distinguish tool drift from process development.

Name the acceptance owner for each risk and make the open-item list visible. Resolution, stitching, charging, vacuum, throughput and training may be owned by different teams, but the handover should have one decision record that links their evidence. If a result is conditional on a future resist, stage setting or software feature, label it as development work and schedule the follow-up test. That clarity prevents a commissioning demonstration from being misread as a universal performance guarantee.

Keep the acceptance coupon and its measurement recipe available for a later comparison. A repeat run after maintenance should use the same field-crossing marks, charging structure, workload and metrology sites unless the change itself requires a new design. The comparison can then show tool drift, process drift or a measurement change without reopening the entire purchase debate.

• Configuration and file ownership
• Calibration and software snapshot
• Service and process change triggers
• Targeted re-test against the control coupon