Why must EM limits account for temperature?

Short Answer

Electromigration speeds up sharply with temperature, so the current a wire can carry for its lifetime drops as it gets hotter. An EM check run at a nominal temperature passes wires that fail in the hot parts of the die. Run thermal analysis first, then set the EM temperature to the hot-region value, or check hot regions separately.

Technical Explanation

• Black's equation links mean time to failure to current density and temperature through an exponential in activation energy over temperature, so a modest rise in temperature cuts lifetime a lot at the same current.
• RedHawk sets EM temperature separately from extraction temperature. Priority runs from TEMPERATURES_EM (RH) per layer in the GSR, to TEMPERATURE_EM (RH) globally, to T_EM (RH) in the tech file, then the extraction temperatures, and finally TNOM_EM (RH). The log records what was used.
• EM_TEMP_RATING (RH) in the tech file gives the ratio of maximum current density at another temperature to the limit at TNOM_EM (RH), which is how a foundry limit is derated.
• RMS limits are tied to self-heating: RMS polynomial EM uses a temperature rise term set by DELTA_T_RMS_EM (RH).
• The EM temperature keywords TEMPERATURES_EM (RH) and TEMPERATURE_EM (RH) are per layer or global, not per region. On a die with a 20 C spread you either use the hot-spot temperature everywhere, which is safe but pessimistic, or run the hot region again at its own temperature. The exception is a tile-based THERMAL_PROFILE (RH) from Sentinel thermal modeling, which maps tile temperatures onto wires and vias and derates only the average EM limits through EM_TEMP_RATING (RH).
• Kelvin inside ICC2 gives the temperature map: analyze_thermal (ICC2) with thermal.tech_file (ICC2), which is mandatory, and report_thermal_qor -threshold (ICC2), threshold in Celsius, lists grids or instances above a temperature.
• Temperature also raises metal resistance, so the same hot region sees more IR drop. Check thermal, IR and EM together rather than in isolation.

What To Check

• Temperature used by the EM run, from the RedHawk log.
• Hottest regions from thermal analysis and their temperature.
• EM ratio of straps in hot regions at the hot temperature.
• RMS results for high-activity nets, which depend on self-heating.

Healthy, Suspicious & Hard-stop Results

• Healthy (illustrative): EM run at the 110 C hot-spot temperature with worst ratio 88%, confirmed by the log.
• Suspicious (illustrative): EM run at a nominal 85 C while thermal analysis shows 110 C in the compute cluster.
• Hard stop: Straps over 100% at the hot-region temperature, or no record of the EM temperature used.

What The Interviewer Is Testing

• Understanding that allowed current drops with temperature.
• Knowledge of how RedHawk sets EM temperature.
• Linking thermal analysis to EM and IR.

Follow-up Question & Model Response

"Why not always run EM at the hottest temperature on the die?"

Candidate Model Response: It is safe, and many teams do that for a first pass. On a die with a wide spread, though, it over-constrains the cool regions and can drive wider straps and more tracks where they are not needed. A second pass per hot region, or layer temperatures that match the thermal map, keeps the margin where the heat is without paying for it everywhere.