As the semiconductor industry continually strives to increase the power density of single-chip packages, thermal management remains a critical challenge toward realizing both performance and reliability metrics. One of the primary bottlenecks inhibiting effective thermal management arises from the several interfaces that can exist between the chip and heat sink. Specifically, the thermal resistance of the thermal interface materials (TIMs) that are currently used to bridge these interfaces must be decreased. Carbon nanotubes (CNTs), with their extraordinarily high axial thermal conductivity, have generated tremendous interest as candidates for providing low-resistance TIMs. A potentially scalable and low-cost process was developed that utilizes spray coating to deposit thin polymer films onto the tips of CNTs for bonding at room temperature.
Polymers were dissolved into solution and spray-coated onto the tops of vertical CNT forests grown on Si substrates to evaluate the enhancement approach. The spray-coated CNTs were then bonded to Ag foil substrates by wetting the interface with solvent, and allowing it to dry under moderate pressure at room temperature. The total thermal resistance of the interface — including the contact resistance of the polymer bonded interface, the resistance of the CNT forest, and the contact resistance at the growth substrate interface — was measured using a photoacoustic method.
Two polymer systems were studied. The first, polystyrene (PS), was chosen since it is a low-cost, widely used aromatic polymer that is chemically stable at device operating temperatures; the second, poly-3-hexylthiophene (P3HT), was chosen because it has been shown to interact strongly with CNTs through π – π bonding and by preferentially wrapping around the nanotube axis. Additionally, due to its conjugated backbone, P3HT is chemically stable at higher temperatures as compared to PS.
The height of the CNT forest and the quantity of polymer sprayed were varied individually in an effort to understand their influence on the thermal resistance of the interface. To demonstrate how the spray coating process might be scaled for manufactured production, CNT forests were grown on both sides of Al foil to create a thermal interposer. The CNT-coated foil interposer eliminates the necessity to grow or transfer print CNTs directly onto the back of the chip or packaging. The CNT growth and spray-coating process can instead take place separately on the metal foil, before being incorporated into the electronic package.
It is difficult to quantify the fraction of polymer deposited onto the CNT forest for each 1-ml spray; therefore the quantity of polymer applied to each CNT forest is presented in terms of the number of 1-ml sprays. The number of sprays was altered among 1, 2, and 5 in order to examine the effects of the quantity of polymer on the resultant thermal resistance of the bonded CNTs. The spraycoating process restricts the deposition of polymer to the CNT tips and limits clumping due to capillary forces associated with the drying of the solvent.
Polymer spray-coating and bonding was demonstrated as an effective means for increasing the contact area and reducing the thermal resistance of CNT forest thermal interfaces. The thermal resistances did not change significantly after baking at 130 °C for 110 hours. The thermal resistances of dry and polymerbonded CNT interfaces were found by measurement to increase with CNT forest height because of the increased surface roughness of taller forests. The relatively low cost of polystyrene in addition to favorable bonding conditions, i.e. room temperature and low pressure (138 kPa), make the spray-coating and bonding process attractive for largescale implementation.
This work was done by John H. Taphouse, Thomas L. Bougher, Virendra Singh, Parisa Pour Shahid Saeed Abadi, Samuel Graham, and Baratunde A. Cola of Georgia Institute of Technology. DARPA-0013