(Invited) Nano-Controlled Etching of Polycrystalline Metals - a Key Enabler for Future Technologies

As Moore's law continues its relentless march, defying odds of its predicted demise well in to more than a decade, innovations are required to enable 10 nm and beyond technology nodes. The capability to etch polycrystalline metals with an unprecedented nanometer scale control offers unique challenges at the forefront of mechanistic and structural understanding of chemical reactivity and kinetics.

 Since the advent of Cu-based interconnects embedded in low-k dielectrics, the landscape of BEOL cleans has focused on development of process technologies that were aimed at enhancing metal compatibility in highly corrosive media. Fluorocarbon plasma-based etches and ashes needed to pattern interlayer vias leave behind a chemically-susceptible dielectric, surface modified metals, locally corrosive environments and copious amounts of etch polymer. Advanced formulations for BEOL via cleans incorporated metal corrosion inhibitors, buffering agents for precise pH control, additives for surface tension control and carefully selected solvent systems for precisely tuned viscosity control. Removal of post-plasma-etch residues with metal compatibility remained a finely tuned balancing act between clean wafers and undamaged wiring.

 Capping of the interconnect wires by selective metal growth as shown in Figure 1 has been extensively researched. While its benefits and limitations are beyond the scope of discussion here, two critical yield impactors, nodules of non-selective metal growth and protrusions of capping metal beyond the CD of the patterned line, are shown schematically in Figure 1. The scheme shown in Figure 2 relies on nano-controlled wet etching of the polycrystalline Cu in the line to create a recessed cavity which may be capped with a metal of choice followed by polishing to yield a capped metal flush with the dielectric. This scheme has the potential to realize benefits of metal capping without the yield-limiting defects shown in Figure 1.

  In this paper, we describe the development of a liquid phase chemical solution to etch a solid polycrystalline metal line composed of grains with potentially different sizes, orientations, phases, textures and inherently anisotropic grain boundaries with nanometer scale precision to yield a non-pitted and smooth metal surface with a RMS roughness of ~ 1.2 nm. Rates and mechanisms of metal dissolution reactions, in both acidic and basic media, are highly sensitive to dissolved oxygen concentration and the mass-transport of reactants and products. Development of etchant chemistry involved optimization of the chemistry components along with tool parameters to yield the process on a 300 mm wafer with < 3 nm center to edge etch depth uniformity (shown in Figure 3).

 Use of dopants to improve electromigration (EM) performance of Cu-interconnects has been a fairly standard practice in the semiconductor industry. Within the context of understanding dissolution of Cu lines with added EM-dopant, we observed somewhat expectedly, drastic differences in etch rates and profiles using the same chemical etch solution. The barrier metal, in direct contact with the Cu line, was also found to influence the mechanism and profile of metal etching. While Ta-based barriers yielded uniform etch rates in the middle of the Cu line relative to the edges, Ru-based barriers showed effects of galvanically-accelerated etching at contacted interfaces (Figure 4).

 Fundamental studies in the nature of metal etching at grain boundaries as a function of grain size, phase, orientation and texture are critically needed. The reaction mechanisms and equilibriums leading to oxidation, nature of transient oxides and their dissolution in the etchant as well as chelation-based stabilization of the dissolved metal to prevent re-deposition are not well characterized. While the application of nano-controlled etching to enable metal-capping for better EM performance and LWR control as described, is quite specific, it is neither unique nor exhaustive. Enhanced ability to selectively and controllably etch different metals may enable multiple innovations that will be needed at the 10 nm node and beyond to drive Moore's law.

 Fig 1. Metal capping by selective metal deposition: a) marginality in growth selectivity poses a shorting risk; b) cap growth typical produces an overhang which reduces dielectric CD and negatively impacts line width roughness (LWR) 

Fig 2. Nano-controlled etch of polycrystalline metal allows for deposition of a metal cap in the recessed cavity.

Fig 3. Nano-controlled etch of Cu lines with a Ta-based barrier

Fig 4. a) Nano-controlled Cu etch with a Ta-based barrier b) Nano-controlled Cu etch with a Ru-based barrier