Fabrication of carbon nanotube/copper and carbon nanofiber/copper composites for microelectronics

Carbon nanotubes (CNTs) and carbon nanofibers (CNFs) exhibit high ampacity, the key property needed for next-generation interconnects at miniaturized scales. Copper (Cu), the current state-of-the-art material used in interconnects, faces reliability issues at further miniaturized scales. This is due to high current density causing electromigration of Cu atoms. Therefore, CNTs and CNF were proposed to replace Cu in next-generation microelectronics. Using standalone CNT or CNF structures is very challenging and achieving the same properties as an individual CNT or CNF is difficult. The difficulty arises because the electrical properties of a grown forest depend on characteristics of the forest including self-alignment, density and mechanical stability. Thus, in this study, Cu is used to densify self-aligned CNTs and CNFs so that the mechanical stability and high density can be achieved. Parameters that impact the quality of the final Cu–CNT composite layer including the CNT qualities and fabrication techniques, CNT underlayer effect and different Cu deposition techniques are investigated. Different Cu deposition techniques on the as-grown CNTs are also experimented including electroplating, electroless plating and physical vapor deposition (PVD). The experiments show that although CNFs were successfully coated with Cu using electroless plating, CNTs are found to be fragile and are dissolved during the process of electroless plating. Furthermore, it was found that CNTs cannot act as seed layer for Cu. Other underlayer material such as Ti and TiN were found difficult to work with due to several reasons. Ti and TiN were not found a good material to grow vertically aligned CNTs using CVD. However, PECVD combined with TiN underlayer was successful in obtaining vertically aligned CNTs, although with a much slower growth rate compared to CNTs grown on Al₂O₃ underlayer. However, TiN was not successful in terms of electroplating.