Abstract
The tribochemical model for copper CMP that has been described previously was tested for acidic slurries containing both glycine and BTA using electrochemical tests employing either a pattern-defined polyurethane pad fabricated in our laboratory or a commercially-available IC1000 pad. As predicted, these tests confirmed that the electrochemical oxidation rate of copper is relatively insensitive to applied pressure. Somewhat unexpectedly, the oxidation rate was also insensitive to sliding velocity. However, the measured oxidation rates were markedly lower than the material removal rates that are typically encountered in copper CMP. This was confirmed in tests using a blanket copper wafer, in which the material removal rates were about two orders of magnitude higher than the electrochemical oxidation rates. Hence we conclude that the principal material removal mechanism is mechanical removal of unoxidized copper. What remains to be resolved is why the material removal rate should be sensitive to the chemical conditions of the slurry.