Fundamental study of nanoscale processes for the integration of carbon nanotubes in IC interconnects

The ongoing downscaling of the dimensions of the integrated circuit (IC) building blocks forces the semiconductor industry to search for new material combinations and innovative technological solutions in order to satisfy the requirements set for future generations. The fabrication of interconnects is no exception to this trend. The current technologies used to metalize interconnect vias and trenches are facing their limits as the dimensions of the vias become increasingly smaller. One alternative technology that could meet the requirements set by the ITRS, is a carbonnanotube based interconnect. Carbon nanotubes (CNTs) have unique electrical, thermal and mechanical properties that make them ideal candidates for future interconnect material. From theoretical calculations, it was shown that at a sufficiently high CNT shell density (> 1013 shells/cm2) the electrical resistance of the CNT bundle can compete with Cu. Also the theoretical current carrying capacity of CNTs is order of magnitudes higher compared to Cu, providing a better resistance towards electromigration. From a technological perspective, CNTs offer an additional advantage as the growth process is truly bottom-up. Therefore no void issues are expected in the fill of the via. In this thesis, the main challenges associated with the integration of CNTs in IC interconnects are addressed. The first challenge is to obtain a high CNT forest density (>1013 cm-2) on a conductive substrate. For large diameter vias, several densely packed CNT forests are needed to meet the technology requirements. To obtain high density CNT forests on conductive substrates, a high density of catalytic nanoparticles is needed. The first part of this thesis deals with how the density of catalytic nanoparticles can be increased. By inhibiting the growth of nanoparticles during electrochemical deposition, the nanoparticle density can be increased up to the point that film closure occurs when the film thickness is below 1 nm. CNT forests from these highly inhibited nanoparticles were densely packed and showed promise for obtaining high density CNT forests on conductive substrates. Another challenge for the CNT integration process is togrow high quality CNTs at a low growth temperature. The growth temperature is limited to CNTgrowth in this low temperature range. The balance between hydrogen and the hydrocarbon species played a dual role during CNT growth as no growth was observed in the absence of H2. In contrast, when the amount of hydrogen was too high, catalyst-mediated etching of the CNT forests occurred. Decreasing the growth temperature resulted in smaller CNT forest lengths. A cross-over in CNT length between Ni and Co catalyzed CNT forests was observed at 455 °C. As a result, at 400 °C CNT forests could only be grown using a Ni catalyst particle. Furthermore, the CNT growth mechanism was linked with its termination mechanism thereby limiting the ultimate CNT length that could be achieved at a given temperature. The last part of this thesis deals with the development of an anodized aluminum oxide (AAO) template that enables the study of CNT growth in confined dimensions. Especially for DRAM contacts where one multi-walled CNT per via is sufficient to meet the technology requirements, achieving CNT growth in sub-10 nm vias is crucial. The successful fabrication of a test vehicle with catalytic nanoparticles selectively at the bottom of the pores was demonstrated. Preliminary results have indicated that bottom-up CNT growth is possible in AAO structures with relaxed geometrical dimensions (40 nm pore diameter, AR = 7.5), but further work is still required to study CNT growth in sub-10 nm high aspect ratio pores. Aside from the development of the CNT structure, it was also found that by changing the composition of the Al film, the vertical pores in the AAO template become horizontally interconnected. The formation mechanism of the horizontal interconnects was studied in more detail and a model was proposed describing the events. These 3-dimensional AAO templates wereused to fabricate 3-dimensional Ni nanowire networks that are well suited for energy storage applications.status: publishe