Semiconductor Data Matrix Marking: Code Quality and Traceability Acceptance
Separate successful decoding, DPM symbol quality, payload correctness and MES identity binding, then validate the complete traceability path under production conditions.
Explore semiconductor laser solutions Read the guideOn this page
Semiconductor Data Matrix marking acceptance needs four separate results: successful decoding, acceptable symbol quality, correct encoded data and correct binding to the physical device. Then test readability after the agreed downstream processes and challenge fault recovery. The buyer, quality team and equipment supplier should agree grade limits, dimensions, damage criteria and sampling before trials; no single grade is a universal semiconductor acceptance threshold.
This guide covers acceptance. For material and source selection, use the semiconductor laser marking selection guide. Review the UV laser marking and traceability system and the semiconductor manufacturing route for equipment and process context.
Separate the four acceptance results
Decoding is not symbol verification
A production reader determines whether its optics and algorithm can decode a symbol under current conditions. Verification evaluates symbol quality using an agreed method and produces a report. A capable reader may decode a marginal mark, so a successful scan screenshot is not a quality report. Cognex explains the difference between production decoding and the standard reference decode algorithm.
Symbol quality is not data correctness
A clear code can contain the wrong lot, obsolete recipe or duplicate serial number. Compare the decoded payload against the issued data, including leading zeros, case, field lengths and separators—not only a reformatted screen display. If GS1 DataMatrix is specified, check its FNC1 and Application Identifier rules. Ordinary Data Matrix is not automatically GS1 DataMatrix. The GS1 guideline describes that special start character and data structure.
Correct data is not correct identity binding
Check that the serial number belongs to the actual device, carrier pocket, work order and process state. Challenge the system with a wrong part, duplicate code, stale instruction and network retry. Acceptance should require rejection or controlled disposition under approved rules, rather than treating any database entry as a successful traceability transaction.
Specify the DPM verification method
ISO/IEC 15415:2024 defines two-dimensional symbol quality measurement and grading. Its public scope notes that using ISO/IEC 29158 alongside it can improve correlation with scanning performance for direct part marks. ISO/IEC 29158:2025 modifies lighting, measurement and grade reporting for DPM; it does not supply a universal purchasing grade. Specify the method, edition and application requirements. See ISO/IEC 15415:2024 and ISO/IEC 29158:2025.
Do not copy label-printing test conditions onto a laser-marked surface or change lighting merely to obtain a higher score. Record verifier identification, software, measurement configuration and relevant calibration status. Retain original images and complete reports. Test production readers separately at the actual working distance, orientation and cycle time.
Complete a buyer defined acceptance matrix
The following is a project framework, not a table of standard limits. Before trials, fill in numerical limits, sample quantities, failure actions and the approving owner.
| Check | Measure and retain | Buyer defined acceptance |
|---|---|---|
| Symbol quality | Overall grade, parameter results and configuration under the agreed DPM method | Minimum grade, method edition, conditions and sampling |
| Position and geometry | Module size, overall size, position and clear surrounding area | Drawing tolerances, available area and application rules |
| Production reading | First attempt outcomes, failures and retries within the required cycle | First read rate, timeout, retry policy and test quantity |
| Payload | Issued data, decoded fields and approved reference payload | Field rules, allowed characters, uniqueness and release blocks |
| MES binding | Device, pocket, work order, recipe revision and acknowledgement | Binding rules, duplicate prevention and transaction completion |
| Material and function | Risk based microscopy, appearance and electrical checks | Mark depth, defect extent and allowable functional change |
| Downstream retention | Read and verify the same device before and after specified processes | Processes, conditions, quality floor and traceability completeness |
For GS1 DataMatrix, the GS1 guideline requires a clear zone at least one module wide around the symbol. Apply the relevant application specification and device drawing when setting dimensions. Do not import grade or size tables intended for medical instruments into a semiconductor contract. GS1 DataMatrix quiet zone guidance.
Validate real devices in the complete cycle
- Freeze sample identity. Record material lot, package, surface treatment, carrier and work order; cover buyer approved incoming variation.
- Establish a process window. Compare settings against quality, damage and cycle time, retaining acceptable and boundary failure samples.
- Time the complete cycle. Include handling, alignment, marking, reading, judgement, upload acknowledgement and reject routing.
- Retest after agreed downstream operations. These might include actual cleaning, curing, reflow or handling steps; retain conditions and the same device identifiers.
- Challenge recovery. Test network loss, repeated triggers, restart, changeover and payload mismatch; check for missing numbers, duplicate release and incorrect record binding.
FAT should demonstrate the agreed configuration and samples. SAT should repeat relevant checks with site readers, carriers, networks and production interfaces. One optimum sample does not replace continuous-run evidence. Use the FAT and SAT acceptance matrix for responsibilities and records, and the automation integration guide to define interface boundaries.
Frequently asked questions
Must every semiconductor device meet the same code grade
No. Agree the grade and measurement conditions against the customer drawing, application rules and risk requirements. Confirm downstream reading performance separately.
Can a production camera quality score replace a verifier
First establish its method, optical conditions, calibration requirements and reporting capability. A proprietary score is not automatically the contractual standard verification result.
Does every code need offline verification
The quality plan decides. Separate per-device reading, per-device payload checks and sampled symbol verification. If inline verification of every device is required, include its cycle-time, hardware and storage scope.
Can production continue offline and upload later
Only under a buyer approved offline policy. Define local identity allocation, durable queues, duplicate-safe retries and recovery reconciliation. Keep unconfirmed devices in an explicit controlled state.
Prepare the validation inputs for a quotation
Supply drawings, materials and surfaces, code artwork, field rules, downstream reader conditions, MES transaction requirements, cycle-time targets and the acceptance matrix. Check the semiconductor equipment RFQ checklist, then send the requirements through the project inquiry form so laser processing, vision, verification and traceability interfaces share a defined delivery scope.
References
- ISO/IEC 29158:2025 official scope: DPM measurement and grading methodology.
- ISO/IEC 15415:2024 official scope: two-dimensional symbol quality assessment.
- GS1 DataMatrix guideline: GS1 data structure and symbol requirements.
- Cognex decode and verification explanation: production decoding versus reference decoding.