Online proceedings - EDA Publishing Association

24-26 September 2008, Rome, ItalyIn-situ measurement of various thin Bond-Line-Thickness Thermal Interface Materials withCorrelation to Structural FeaturesB. Wunderle 1 , J. Kleff 1 , R. Mrossko 2 , M. Abo Ras 1 , D. May 1 , R. Schacht 1 , H. Oppermann 1 , J. Keller 2 , B. Michel 11 Fraunhofer Institute Reliability and Microintegration, Gustav-Meyer-Allee 25, 13355 Berlin, GermanyEmail: bernhard.wunderle@izm.fraunhofer.de2 Berliner Nanotest und Design GmbH, D-12489 Berlin, GermanyThermal characterisation of thermal interfaces becomeseven tougher a challenge at low bond line thicknesses andhigher thermal conductivities of the interface materials asmore accurate measurement techniques are required. As inparallel the quest for high conductivity adhesives andgreases is ongoing, a correlation between thermal bulk orinterface properties and structure is in high demand. Wehave developed test-stands for various classes of thermalinterface materials. These permit characterisation formaterials with thin bond line thickness and high thermalconductivity still using steady state techniques. Themethods are benchmarked for greases, adhesives andsintered silver. For the latter, the technology developmentis described. Then, structural features such as particledensity and porosity are examined. It will be the aim tocompare and correlate them to thermal resistance. Part ofthe work has been accomplished within the running EUProject “Nanopack”.I. INTRODUCTIONThermal interface resistance represents one of the majorbottlenecks in advanced thermal packaging solutions. Notsurprisingly, many research institutions and companies havebeen busily trying to design, manufacture and commercialisethermal interface materials (TIMs) which are to feature bothhigh thermal conductivity and supremely low thermalinterface resistance at the mating boundaries. This is true formost classes of TIMs such as e.g. mono-metal die attach,solders, particle-filled polymer adhesives, thermal pads orgreases. An urgently required systematic furtherimprovement of thermal interfaces, however, presupposesunderstanding of the governing effects of heat transfer, theidentification of key determinants in materials design andbonding processes as well as the capability to accurateexperimental characterisation. The latter has to compriseboth structural characterisation and measurement of thethermal performance concerning conductivity and interfaceresistance as function of interface properties and processconditions. In this vein a structure-property correlation withrespect to thermal properties is required to establish designguidelines for manufacturers. This becomes even moreinteresting as thin bond line thicknesses (BLTs) below 20microns are realised, having at the same time to assure a lowmechanical stress bond for reliability reasons and permitprocessability by featuring low viscosity. Here it becomesevident, that advanced thermal management and choice/development of TIMs has to go hand-in-hand withtechnology development issues as well as design-forreliability.This comprehensive approach is the content and goal ofthe “Nanopack” project, with special focus on the use ofnano-structural features to decrease thermal resistance innew thermal technology development as to materials,surfaces and processes. Appropriate experimental analysesand simulations are methods to achieve this aim and gaininsight into thermal heat transfer on a micro- and nano-scale.Under investigation are the following material systems,where material/structural and process parameters are varied.- Thermal greases, e.g. filled silicones,- Thermal adhesives, e.g. Ag and NCT-filled epoxyresins,- Mono-metal systems, e.g. sintered (nano-) Ag layers.Later, these materials will be employed in conjunctionwith different surface finishes/technologies to investigatetheir influence to achieve a lower interface resistance. Still,at the beginning of such a vast study there is first need forsome reliable thermal characterisation method.Thermal characterisation methods for TIMs narrate a longstory of confusion, as results from different characterisationmethods often disagree formidably. Even worse, thermalconductivity values will be different when applied to the realdevice, likely to cause over-, or more often, fatally underdesignedthermal heat paths. The reason for thismisjudgement is often that TIM characterisation is doneunder much too favourable conditions (e.g. polishedsurfaces, excessive pressure conditions) or disregardingtechnological influences (e.g. cure regime for adhesives,©EDA Publishing/THERMINIC 2008 112ISBN: 978-2-35500-008-9