Thermal performance and reliability of thermal interface materials: A review

Interface thermal resistance minimization has been identified as a critical issue for the thermal management of electronic systems by the NEMI 2004 technology roadmap [1]. In current high-performance air-cooled microelectronic applications, the component-to-heat sink interfacial contact thermal resistance can be comparable to that of the actual heat sink [2]. Consequently, improved thermal interface materials are now the focus of much on-going research [2]-[4]. Typically, TIM applications are categorized as either die-to-heat spreader (TIM 1), die-to-heat sink (TIM 1.5), heat spreader-to-heat sink (TIM 2), or die-to-die carrier (die attach). An overview of thermal interface technologies is given in [4]-[6]. The majority of TIMs are polymeric-based materials, such as adhesives, greases, phase change materials (PCMs) and thermal pads. The thermal performance of such systems is constrained by their bulk thermal conductivity and interfacial thermal contact resistance. The bulk thermal conductivities of newly developed silver-filled or carbon fiber-loaded epoxy resin die attach adhesives, which are intended for power ICs, are claimed to exceed those of eutectic solders. However, to fully exploit the potential of such materials, the die attach assembly process (adhesive deposition, curing) requires careful optimization to ensure good structural integrity of the bulk adhesive (maximum surface area coverage, minimum voiding) and optimum bond line thickness [7]. Apart from progress in manufacturability, thermal interface minimization also requires new modeling techniques for micro/nano resistance across hard/soft interfaces, and an understanding of fatigue-induced failures in bonded interfaces. Solder alloys have a thermal conductivity of up to twenty times that of conventional TIMs [8], and are therefore an attractive solution, in terms of allowing higher component power dissipation while maintaining operating temperature within safe limits. Despite this advantage, solder-based TIM1 and TIM2 materials have been rarely employed to date, partly due to reliability concerns associated with the coefficient of thermal expansion (CTE) mismatch between the silicon die (3 to 5 ppm/°C) and metallic heat spreader (17 to 23 ppm/°C for copper and aluminum). Furthermore, conventional reflow soldering processes employed to directly bond silicon to copper tend to induce residual stresses within the package during the cooling phase, and require the use of flux.A novel selective soldering technology has recently been proposed for TIM1 and TIM2, which melts the solder layers locally by exothermic reactions at room temperature [9]. As heating is confined to the interface region, the residual stresses that would be induced in the package in a conventional reflow process are eliminated. However, there is a need to assess the reliability of such novel solder materials in electronic TIM applications. The characterization methods currently employed by vendors to characterize TIM thermal performance do not permit nominal measurement data to translate to application environments [3]. This arises from the characterization environment being poorly representative of the mechanical and thermal application. In addition, characterization methods typically differ between vendors, preventing direct comparison of TIM performance. For thermal resistance measurement, standard ASTM D 5470 specifies a contact pressure of 3 MPa, which considerably exceeds the contact pressures typically employed in real applications, which are of the order of 0.1 MPa [2]. Consequently, vendor thermal resistance measurements essentially represent the bulk thermal resistance. When used in an application environment, nominal vendor data would therefore lead to an overestimation of actual TIM thermal performance. Such inconsistencies make the selection of an appropriate TIM, and its integration into an application environment, difficult for end-users. Improvement in current thermal characterization methods are clearly required. This paper presents a survey of the thermal performance of a range of TIMs (adhesives, greases, phase change materials (PCMs) and thermal pads) from different vendors. Previously published studies on TIM thermal performance characterization and reliability assessment for electronics packaging are reviewed. The potential limitations of ASTM D 5470 standard, and TIM industry thermal impedance characterization practices utilizing this standard, are discussed, as well as alternative characterization approaches. From this analysis, guidelines are derived on TIM selection from thermal performance and reliability perspectives.