No abstract

The oxidation and removal of copper(0) foil and copper(I) oxide (Cu2O) powder were studied in hexanes or condensed CO2 (at a constant density of 0.8 g/cm3) using excess tert-butyl peracetate (t-BuPA), as the oxidant, and excess 1,1,1-trifluoro-2,4-pentanedione (tfacH), as the chelating agent from 19 to 68 °C. Concentration and temperature effects were used to develop kinetic models. Under our conditions, independent of solvent choice, the reaction of Cu2O was first order with respect to t-BuPA and tfacH while for Cu(0) foil the reaction was independent of t-BuPA concentration and second order with respect to tfacH. The calculated effects of stir rate and reaction temperature on the mass transfer coefficient derived from a correlation for flow past a sphere were used to confirm that the reaction was kinetically controlled along with the fact that the reaction rates and orders were independent of stirring rate. Copper removal rates were calculated from the kinetic expressions (at 40 °C using 0.78 M tfacH and t-BuPA):  copper from the foil was removed at 2.7 nm/min in hexanes and 140 nm/min in condensed CO2 while the Cu2O powder, based on initial surface area, was consumed at 120 nm/min in hexanes and 7.0 nm/min in condensed CO2. The Arrhenius expressions were determined by the temperature-dependent kinetics of the product, Cu(tfac)2, formation. While hexanes and condensed CO2 have similar solvent properties in many respects, the apparent activation energies, E a, for Cu(0) foil were 88 kJ/mol in hexanes and 49 kJ/mol in CO2. The apparent E a for Cu2O were 30 kJ/mol in hexanes and 53 kJ/mol in CO2. The differences in the activation energies are attributed to a complex combination of solvent properties and the energetics of the reacting species.

Chemical-mechanical planarization (CMP) is a process of oxidizing and chelating the copper overburden present in an interconnect device while mechanically polishing the surface of the wafer. Because the use of condensed CO 2 as the solvent for CMP would be environmentally and technically advantageous, several substituted bis(acetylacetonate)ethylenediimine (R 4 BAE, where R ) CH 3 or CF 3 ) and lithium or sodium dialkyldithiocarbamate (M + (R 2 DTC -), where M + ) Li + or Na + and R ) ethyl, n-propyl, n-butyl, or 1,1,1trifluoroethyl) ligands were used with t-butylperacetate (t-BuPA, as oxidant) for the oxidative dissolution of copper(0) in supercritical (sc) CO 2 at 40 °C and 170-210 bar or in hexanes at 40 °C and atmospheric pressure. The reaction products from the copper etching were determined to be Cu(R 4 BAE) or Cu(R 2 DTC) 2 , respectively. The R 2 DTCligands had higher etch rates than the R 4 BAE ligands with comparable substituents, and the lithium dialkyldithiocarbamate salts gave higher copper etching rates than their sodium counterparts. The highest average etch rates were observed for Li((CF 3 CH 2 ) 2 DTC): 16.7 nm/min in sc CO 2 and 11.2 nm/min in hexanes. While hexanes have similar physical properties when compared to sc CO 2 , the rates of copper(0) removal with the R 2 DTCligands were 17-49% higher in sc CO 2 than in hexanes at comparable temperatures and solvent densities. Scanning electron microscopy (SEM) images of the postreaction copper surfaces using the various ligands showed significant variations in surface roughness. X-ray photoelectron spectroscopy (XPS) measurements indicated that the lower R 4 BAE etch rates may be due to surface passivation by the R 4 BAE ligands and/or the Cu(R 4 BAE) complexes.