Abstract

Copper (Cu) has been used as the metal for interconnects for many years. Along with the continued scaling of semiconductor devices, especially beyond 10 nm nodes, alternative metals such as ruthenium (Ru) and molybdenum (Mo) have emerged as interesting candidates to replace Cu [1] at certain levels of the interconnect.Despite having higher bulk resistivities with respect to that of Cu, these alternative barrierless metals have lower resistivity at dimensions below ~12 nm. For instance, the resistivity of Mo was found to be lower than that of TaN/Cu/TaN stack of a total thickness below 8 nm [2]. In addition, besides W which is considered as a prime candidate as a first-generation buried power rail (BPR) metal [3,4], Mo has also been introduced as a potential candidate to achieve even lower BPR resistance.For both dual damascene and semi-damascene structures, typically, a wet cleaning step needs to be carried out prior to the next processing step, aiming at a complete removal of the residues generated during the dielectric patterning (dual damascene) or the metal patterning (semi-damascene). Therefore, the compatibility and etching behavior of the metals in chemical solutions must be well understood. In this work, the effect of various chemical solutions on the removal of post-etch residues and the etch characteristics of Mo is investigated.For the cleaning study, we first compared the etching rate on blanket Mo of two commodity chemicals (diluted HF of 0.05% and SC1) and formulated mixtures. The etching characteristics of the as-deposited and 420 °C-annealed Mo were similar for most chemical solutions. 0.05% HF and semi-aqueous alkaline mixtures were found to be highly compatible with Mo, while SC1 mixture significantly etched Mo. The high etching rate of Mo can be explained by the presence of H2O2 in the SC1 mixture, even at low concentration. The surface chemistry of the as-deposited Mo was compared with the surface after different treatments. Figure 1(a) shows the normalized XPS Mo 3d core-level spectra recorded on blanket Mo films after deposition, after Cl2-based plasma treatment and with subsequent wet cleaning. The narrow peak recorded at ~228.0 and 231.1 eV can be assigned to the Mo 3d5/2 and 3d3/2 components, respectively, of the metallic Mo. While the shoulder and broader peak recorded at higher binding energies, ~232.6 and 235.4 eV, are attributed to Mo oxides (3d5/2 and 3d3/2 components, respectively). It is clear that the surface became less oxidized after the plasma treatment. While the clean using 0.05% HF for 1 min did not show any change in terms of surface oxidation, the semi-aqueous mixture (FOTOPUR R-2403 formulation from BASF) clearly dissolved most of the Mo oxides present at the Mo surface. Since the XPS measurement was carried out ex-situ, this result could also be further explained by the dissolution of the surface Mo oxides combined with a better surface passivation provided by the formulated chemical vs. HF solution. In addition, these results also indicate that Mo surface is readily oxidized [5], and it needs to be protected after a dry etching or wet cleaning process. Figure 1(b) displays the atomic ratio of O 1s/Metallic Mo (measured from the Mo 3d level) for the surface described above, giving additional evidence of the low oxygen content of the Mo surface after being cleaned in the formulated chemical solution FOTOPUR R-2403.The assessment of post-etch residue removal was done on a semi-damascene structure (metal pitch = 32 nm), where Mo lines were formed by a non-optimized Cl2-based plasma etch. Figure 2 shows the TEM images obtained after the plasma etch, with the residues formed in between the lines (Figure 2(a)), and with subsequent cleaning using the semi-aqueous formulation FOTOPUR R-2403 from BASF (Figure 2(b)), showing efficient removal of the residues and good compatibility behavior with the top hard mask and the metal lines. The latter thus confirmed the etching rate results on blanket Mo films.Quantification of surface roughness and a study of the oxide growth on Mo surface will also be presented and discussed.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call