Gallium-Nitride Power-Device Reliability Director — Fast Charging

The mandate

A fast-charging platform adopted enhancement-mode GaN devices to raise switching frequency and shrink converter size, but field returns now show intermittent efficiency loss and gate damage that standard qualification did not predict. Device suppliers cite compliant parts, while platform teams changed drive conditions, layout and thermal interfaces across revisions. The defined problem is to establish which combined stresses govern lifetime and whether current screening and derating protect the deployed mission.

The deliverables are a field-to-physics failure map, configuration genealogy, mission-profile stress model, dynamic on-resistance characterisation, gate-margin evidence, accelerated-test plan, screening decision and platform reliability case. Outputs must separate recoverable trapping from destructive degradation, bind device lot to converter conditions and explain extrapolation limits. The project will recommend design or operating changes without pretending one acceleration factor governs every mechanism.

Four milestones structure five months: reconciled return population and competing failure hypotheses by 23 October 2026; characterised electrical, thermal and switching stress corners by 27 November; witnessed acceleration and mitigation evidence by 15 January 2027; and accepted lifetime case, field action and client-team transfer by 12 February. Samples must include returned units and production-representative builds from declared lots.

Acceptance requires failure analysis to reproduce the dominant signatures, converter engineering to demonstrate mitigation across agreed corners, quality to approve population treatment and the client reliability lead to rerun the lifetime model independently. The sponsor returns a single variance list within six working days of each milestone. Supplier qualification reports or curve fitting without mechanism stability cannot satisfy the final acceptance criteria.

The client provides field telemetry, returned chargers, device and board genealogy, gate-drive waveforms, layout revisions, thermal data, laboratory access, supplier contacts and authorised stress limits. It funds agreed sample preparation and declares any safety-critical operating boundary. The director cannot release products, change live charger firmware, determine customer compensation or assert supplier liability; client personnel execute all production and field changes.

Why this is external work

The disputed failures sit between semiconductor behaviour and converter implementation, giving each supplier a plausible local explanation. Internal teams also designed the changing revisions and need an independent mechanism test before committing to a fleet action. A finite reliability programme can end with accepted lifetime evidence, bounded corrective action and a client team able to reproduce the reasoning.

What you will own

• Reconcile returned-unit symptoms with device lot, board revision, gate drive, switching history, thermal path and charger duty cycle.
• Distinguish trapping-related dynamic resistance, gate overstress, current collapse, package damage and system-control artefacts through physical evidence.
• Translate charging sessions, ambient conditions, grid events and maintenance behaviour into defensible laboratory stress sequences.
• Design acceleration studies with mechanism checks, censored-data treatment, sample sufficiency and explicit lifetime-extrapolation limits.
• Challenge supplier qualification against the platform’s actual voltage slew, parasitics, dead time, thermal cycling and transient exposure.
• Decide whether screening, derating, layout change, driver modification or population containment best addresses each evidenced mechanism.
• Deliver the accepted reliability case, field disposition logic, monitoring triggers and independently executable client model.

Candidate qualifications

• Directed GaN device or power-converter reliability investigations from field signature through mechanism and corrective action.
• Can evidence a dynamic or gate-related failure that standard component qualification did not represent adequately.
• Understands trapping, dynamic resistance, gate behaviour, switching parasitics, thermal coupling, acceleration and statistical lifetime inference.
• Has challenged both device and platform teams without assigning causation before configuration and physical evidence converged.
• Designed production-representative stress and screening strategies that balanced escapes, false rejects, cost and device damage.
• Communicated extrapolation uncertainty to executives making fleet, supplier and customer-treatment decisions.

Non-negotiables

• Will attend all Tokyo, Nagoya and Taipei laboratory and supplier sessions in the milestone plan.
• Has personally interpreted destructive physical analysis alongside dynamic electrical characterisation of GaN hardware.
• Accepts milestone rejection if returned-unit genealogy or accelerated mechanism stability cannot support the claimed lifetime conclusion.
• Will disclose power-device, foundry, charging-platform and failure-analysis laboratory relationships before engagement.

1. 49 words maximum. Which GaN failure appeared compliant at component qualification yet failed under a real converter mission?
2. 49 words maximum. How would you prove accelerated stress preserved the field mechanism rather than creating another one?
3. 49 words maximum. What evidence separates supplier device weakness from damaging platform implementation?