Independent engineering project · ANSYS

Semiconductor package thermo-mechanical analysis.

A verified numerical workflow for comparing heat input, cooling and solder regional response.

01 / ANSYS

A verified basis for thermal–mechanical screening.

An independent learning and engineering project connecting a five-body package model, numerical checks and automated parameter studies.

5 W baseline · ambient / TREF 25 °C · h = 1000 W/(m²·K)

Heat input
5 W
Maximum temperature
38.79 °C
Maximum deformation
3.19 μm
Solder regional average stress
16.99 MPa
The final 5 W condition spans 36.93–38.79 °C.

02 / ANSYS

Heat transfer and expansion mismatch.

How does die heat reach the cooling boundary, and how does expansion mismatch turn into deformation and solder stress? Controlled power and cooling comparisons address these questions without changing the geometry, materials or constraints.

03 / ANSYS

One package. Two sequential physics models.

Si / SAC305 / Top Cu / AlN / Bottom Cu. Steady-State Thermal transfers temperature to Static Structural. Final L1 uses 23,108 nodes, 3,504 quadratic Hex20 elements, two elements through solder thickness and four Bonded MPC interfaces. Existing 3-2-1 constraints remove rigid-body motion.

SiLayer 1
SAC305Layer 2
Top CuLayer 3
AlNLayer 4
Bottom CuLayer 5
  1. 015 W Si heat input
  2. 02Steady-State Thermal
  3. 03Imported temperature
  4. 04Static Structural
  5. 05Deformation + solder stress
Inspect the final mesh
23,108 nodes and 3,504 quadratic Hex20 elements provide the common parameter-study mesh.

04 / ANSYS

Temperature becomes deformation and regional stress.

At 5 W / h = 1000, maximum deformation is 3.19 μm and solder regional average stress is 16.99 MPa. The average is taken from the existing NS_Solder Mechanical result; no new volume-weighted measure is implied. Native legends retain m and Pa units.

The 16.99 MPa regional average is the primary comparison metric. The local peak is mesh/contact sensitive.
Inspect total deformation
The imported thermal field produces a maximum displacement of 3.19 μm; use the legend for magnitude.

Native Time: 1 is the steady/static load-step label, not one second of transient response.

05 / ANSYS

Stable regional response. Sensitive local peak.

The L1→L2 mesh study found small changes in temperature, deformation and solder average, but a 10.67% peak change. Fixed-L1 contact tests changed the average by 1.85% from C0 to all-MPC C3. Mesh verification used the earlier contact formulation; a separate all-MPC L2 study is not claimed.

Contact formulation affects the local peak more strongly than the regional average.
Inspect the mesh convergence evidence
Solder average changes by 0.794099% from L1 to L2 in the earlier contact formulation.

06 / ANSYS

Validate once, then automate controlled comparisons.

Workbench Journal and Mechanical ExtAPI / IronPython handle parameter injection, thermal solve, temperature mapping, structural solve and extraction. Initial 5 W manual/automatic validation passed near the rounding precision of manual records; final-model regression was checked separately.

  1. 01Journal orchestration
  2. 02ExtAPI / IronPython
  3. 03Sequential solves
  4. 04CSV / JSON extraction
  5. 05Python comparison

07 / ANSYS

Higher power raises all three primary responses.

With h fixed at 1000 W/(m²·K), 3→7 W raises Tmax from 33.27 to 44.30 °C, maximum deformation from 1.91 to 4.47 μm and solder average from 10.19 to 23.79 MPa. All final cases share L1 and four Bonded MPC contacts.

Increasing heat input raises solder regional average stress under fixed cooling.

08 / ANSYS

Cooling remains effective, with diminishing temperature benefit.

At 5 W, h = 500→2000 W/(m²·K) lowers Tmax from 51.30 to 32.52 °C and solder average from 32.92 to 9.01 MPa. Rth = (Tmax − Tamb) / Power describes the modeled system, not a measured cooling device.

Stronger effective cooling reduces maximum temperature across the sampled cases.
Inspect effective system thermal resistance
This screening metric includes the modeled system and boundary; it is not an intrinsic material resistance.

09 / ANSYS

Compare regional stress, not an isolated peak.

Power and effective cooling consistently change the sampled regional response. Successive h doublings reduce Tmax by 12.51 and 6.27 °C. The Si/Solder interface-edge peak remains a numerical-sensitivity reference rather than an absolute strength criterion.

10 / ANSYS

A screening model, with explicit limits.

Steady-state and linear-elastic only. SAC305 viscoplasticity, creep, thermal cycling, Anand behavior, transient response and experimental calibration are not included. No actual fatigue life, failure probability or product qualification is established. The contact-stiffness warning disappeared; five other known warnings remain in each final case, with zero solver errors.