Understanding Creep Deformation, metal thermal interface materials Posted by Bob Jarrett Monday, January 7th, 2008

Understanding Creep Deformation in Laser Diode/LED Assemblies

The assembly designs of laser and high-power light emitting diodes (LED) demand tight dimensional control under high thermal loads. The deflections and residual stresses caused by the differential thermal expansion of the materials in the stack up and the deflections are dauntingly complex. The last thing the engineer wants to hear is that the parts in the stack move over time at load. This time dependent deformation is known as creep.

Creep deformation is usually associated with high temperature metals (like those in jet engines) or semi-liquid materials like modeling clay or solder paste. Essentially all materials will creep—under the right (or should I say, wrong) set of conditions.

Attached is a simplified diagram of a stack-up showing the diode attached to the heat spreader/case which is attached to the heat sink. Heat is generated as the diode converts electricity to light. This waste heat must be removed by conduction to the heat sink while keeping the semi-conductor cool. Thermal interface materials conduct the heat diode-to-spreader (die attach or TIM1 thermal interface material, level 1) and spreader-to-sink (TIM2 thermal interface material level 2).

The die attach interface is often gold (Au-Sn, Au-Ge, Au-Si eutectics) or tin-silver-copper (SAC) alloy solder joint. These alloys form a “hard” solder joint with high tensile strength and low compliance. Soft solders can be used at the die attach level where the goal are compliance and low thermal stresses from the mismatch of the CTE for the die and the spreader. Silver-filled epoxy is a maturing alternative material which is in the mode of working out issues with the process and increasing the conductivity.

As would be expected, the soft solders will deform more easily under load and high temperatures than the hard solders. This accommodation/deformation results in relaxation of the thermal stresses. However, this movement may not be reversible. Under certain cyclic loading conditions, the soft solder may gradually take a permanent set.

In the case of the second level interface (or TIM2), there are a few more considerations. A range of solder alloys can be selected if both the case/heat spreader and the heat sink have a solderable surface such as copper or electroless nickel/immersion gold (ENIG). If an aluminum or nickel-plated component is used, soldering can only be achieved with an aggressive, high activity flux such as Indium Corporation’s RSA or Flux Number 3.

This flux breaks down the tenacious oxide layers to permit soldering, but may not be compatible with some of the electronic components. The recommended approach is to apply a thin layer of gold (<50 microinches) which is scavenged by the solder. The indium-containing solders readily form a gold-indium intermetallic bond. The intermetallic is inherently brittle so the gold coating should be kept to a minimal thickness, typically 10 microinches or less.

At the second level thermal interface material interface, the temperature and CTE mismatch effects are smaller than the first level thermal interface material. However, with a very soft material like pure indium, movement of the metal after soldering is still possible if the interface temperature approaches the melting point (156°C). Adding silver to indium increases the tensile strength with only a small trade-off in lowering the melting temperature (97%In-3%Ag is the eutectic at 143°C) and thermal conductivity. Adding 3%Ag to indium increases the tensile strength from 273 to 800 psi (1.9 and 5.5 MPa) and drops the conductivity from 86 to 73 W/m-K. Adding more 10% silver further increases the strength to 1650 psi (11 MPa) while the thermal conductivity decreases to 67 W/m-K. This alloy is off-eutectic so it has a melting range from the solidus at 143°C to the liquidus at 237°C.

Posted by Bob Jarrett at 15:21 PM (January 7th, 2008)

Tags: