Jordan Ross

The word Creep is not something an engineer wants to hear when designing a metal TIM into their Laser Diode Stack up. But is creep really an issue, and do we really understand creep when talking about indium? The answer is relative. Attached you will find a stack up diagram I made of a typical Diode stack. This could be a laser application or an LED application. The Die Attach layer is usually a high temp solder or a silver filled epoxy. Can indium be used at this level? The answer is yes, if and only if the temperature of the junction is far away from the melting point of pure indium, which is 156C. Soft solders are just that, they are soft, and in die attach some customers have used indium at this level, but it is not nearly as common as other higher temp solders like AuSn, Sac alloys or SnAg. Silver filled epoxies are really getting better and better but there are some issues with their conductivity and their process. When using a solder as a die attach the die itself can float or move during reflow. In this case some kind of mechanism can be used to hold it in place during reflow to ensure that its alignment is perfect. So to answer the question, solder can be used as die attach in this application, but the alloy chosen will really determine how effective it is and how reliable it is.

In the case of TIM2 (thermal interface material level 2), there are a few more considerations here. Let us first assume that we are going to reflow at this level. Copper at the heat-spreader level will not be a problem, but Nickel at the spreader/sink level will be a problem. Aluminum will also be a problem here. The problem is that these materials are hard to solder to, but it can be done. A high activity flux such as Indiums RSA or Flux number 3 can be used to break the oxide layer that will be present. However a layer of gold on the surface will help assure soldering will be effective. Indium recommends not to exceed 50 micro inches of gold, and recommends that the thinner the better, usually 10 micro inches will do it. Indium the element will actually dissolve the gold or other wise known that the gold will diffuse into the indium. During soldering an Indium/Gold inter-metallic will form. This is a brittle layer and if too much gold is used can induce reliability issues and cracking of the joint. So back to our original question; why would you use indium here and can you use indium here? This is the most common area where creep of indium can be an issue. Creep can be acceptable however. Indium will not creep to the degree that it will come out like pump out or like play-doe. The degree of creep is related to the pressure that is put on it, ie: CTE movement or direct pressure from clamping, as well as the temperature that the interface sees. If the junction temp is less that 20 degrees of the melting point and some movement is allowable, such as in an LED application, this is acceptable. However in a laser application, Indium Corporation usually advises that we do not go with pure indium at this level and rather choose an alloy such as Indium Silver or Tin Silver. The more Silver you add to the alloy the harder the material will become. Consider this. Pure Indium has a conductivity of 86W/mk, mp=156C Eutectic and a tinsel strength of 273 psi. Add 3% silver and the MP goes to 143C Eutectic, the conductivity goes down to 73 but the tinsel goes up to 800 psi. Better yet, add 10% silver and the MP is now plastic from 143 to 237, conductivity goes down to 67 but the tensil goes up to 1650 psi. Then consider SnAg which has a melt point of 221C a conductivity of 33W/mk and a tensile of 5800 psi. What is the best for your application? The answer lies in what is acceptable to you, if CTE is an issue go with less silver, if temp is an issue look at no indium, if conductivity is an issue look at high indium.

When considering a compressible metal at the TIM2 level, the issue now at had is how much pressure you have, the temp of the junction, and the planarity of the surface. Obviously with a compressible interface you no longer need any gold, and there will be no issues with indium in direct contact with nickel. However, copper and indium can form an inter-metallic over time; however the oxide layer on the copper usually keeps this from happening. In fact in our thermal lab we only saw this happening when we actually baked the modules for over 1000 hours at 125C+. Even then the phenomena was nominal. If you do not believe you will rework the interface 4-5 years after its construction I would say that this will not be an issue, actually it will improve the thermal performance and reliability. Rework within 1-2 years will not be a problem. Many of our customers already use indium at this level as a compressible interface but few are aware that they can actually improve the performance if they convert to a Heat-Spring ™. The Heat-Spring is a patented process that allows us to decrease the contact resistance of the metal if the pressure is at least 50 psi. This allows the stack up to use a thinner bond line thickness and improve the thermal performance of the Metal Thermal Interface. So what about creep at this interface? Once again altering the alloy can eliminate the chance of this happening, but converting from a standard indium flat foil to a Heat-Spring will further decrease these chances because your bond line is usually significantly less if you use a Heat-Spring. (On average about .003").

In Summary, the things to consider when using a Metal Thermal Interface or Die Attach Solder are as follows:
• What is the working temp of the interface?
• Is this temperature too close to the melting point of the Metal Thermal Interface?
• Can the device handle the reflow temperature of a higher temp solder such as Gold Tin or Tin Silver
• Is the thermal performance of the interface an issue so if you change the conductivity of the metal thermal interface by decreasing the indium content, you detriment the conductivity of the entire stack up? For example going from 86w/mK to 67w/mK.
• Is creep really an issue? If your device can accept a small degree of movement there is no issue to use indium. If even a slight issue will cause a problem, suggest an Indium Alloy and not pure indium.

In the end, Indium Corporation is here to help you. Please see our web site for additional information including our e-list of alloys to help you choose the best Metal Thermal Interface Material.

Amanda and I recently spoke at the last IMAPS workshop in San Jose last week. The event was a great success. I did however take away one thing especially from a young man that presented at the IMAPS about over clocking. The crowd kind of chuckled at this young engineer about the fact that these guys can burn out a processor in a month. Instantly I wondered if a better thermal interface material would help make a difference. In this particular case this young man presented data on a number of thermal interface materials but did not focus at all on 100% metal or fully metallic thermal interface materials. The range of materials was mostly thermal greases that were tested on an AMD lidded processor. I have since given my card to this young gentleman and I hope that he responds to me about testing some of our metallic thermal interface materials.

It was disappointing to me that some of his experiments seemed too radical to some of the participants, and I can understand why. But I must say that the work that some of these guys are doing is a great way to test interface materials and thermal set-ups in general. My opinion was, this guy has his own thermal test vehicle in his dorm. I think there is something to learn from this group.

I was searching the web and saw a blog entry on Liquid Metal TIM. The web address is http://reviews.pimprig.com/cooling/coollaboratory_liquid_metal.php.

Indium Corporation has been the leader in Metal Thermal Interface Materials for many years, and the demand for materials tends to have its own life pattern. Many years ago there was a large demand for metals that were liquid at room temp such as Indalloy 51. This demand was mostly due to the search for a replacement for Mercury.

This blog entry points out how wierd handling liquid metal can be. And actually the performance of the material turned out to be the second best of all the tested TIMs. This brought up another point. The Bond Line Thickness of the Metallic Thermal Interface Material can actually be too thin for the application. This is what I believe may have happened in this experiment. The Metallic Thermal Interface Material or Liquid Metal was probably about .001 to .002" thick at the interface level. If there were planarity issues on the lid and the heat sink you could have a potential gap of up to .004" to .006". These voids can cause a major increase in the thermal resistance of the interface, especially when using a metal thermal interface material that is much thinner than the BLT.

By: Pat Ryan

Do you know who the father of heat transfer is? You probably know techniques for heat dissipation in electronic devices can include heatsinks and fans, but did you know in 1822, a French mathematical physicist developed the equation we use today to analyze conductive heat transfer in electronic devices. The equation, Fourier's law of conductivity named after J. Fourier, is central to the study of heat transfer in electronic devices and the application of thermal interface materials to improve reliability and prevent premature failures.

Could Fourier have ever predicted that semiconductor materials, such as silicon and germanium would not only be excellent electrical conductors but thermal conductors as well? Could he have known that an exception to his rule of solid materials conduct heat better that liquid would be the metals indium and bismuth. Today these materials are in the forefront of thermal designs for electronic packages including power QFN's, diodes and power devices in PCB assemblies.

Not many people realize that Indium is actually 4 times softer than lead. In this picture you can see that with just a standard vise I could compress this penny into a .004" thick piece of Indium.

Welcome to the new Indium Blog. We have Indium Thermal Team Members that will constantly be adding to this blog. They will include Bob Jarrett, Amanda Whittemore, Adrian Low, Jim Hisert and many others.

Please send emails to jross@indium.com for comments.