Andrew A. Gewirth

Complexing agents in Cu chemical mechanical planarization slurries promote dissolution of the copper surface. We explored the effect of two chelating compounds, ammonium oxalate and oxalic acid, with and without hydrogen peroxide, on the dissolution and surface morphology of a Cu(111) surface at neutral pH. Without hydrogen peroxide, static etch rate (SER) measurements were highest for oxalic-acid-containing solutions, but with peroxide, ammonium-oxalate-containing solutions demonstrated the highest rates. The largest anodic and cathodic currents were seen in solutions containing ammonium oxalate and peroxide using cyclic voltammetry. In the presence of oxalic acid, atomic force microscopy revealed a thick layer on the surface, identified as Cu2OCu2O on the basis of X-ray photoelectron spectroscopy measurements, that changed little over time. In the presence of ammonium oxalate, the film was thinner and showed accelerated dissolution. It is proposed that the dissolution processes of oxalate and ammonium operate independently at the same time. In the presence of peroxide, more Cu2+Cu2+ is created which allows for the formation of more copper oxalate, and H2O2H2O2 also increases the concentration of copper oxide species at the surface, which drives the ammonia complexation reaction. These two events give rise to a large increase in SER.

The interactions of nitrate with Cu(100) and Cu(111) in acidic solution are studied by cyclic voltammetry (CV) and in situ electrochemical scanning tunneling microscopy (EC-STM). CV results show that reduction of nitrate on Cu(111) commences at 0.0 V vs. Ag/AgCl while the corresponding potential is −0.3 V on Cu(100). EC-STM images show that the terrace of both Cu(111) and Cu(100) are atomically flat at potentials more negative than −0.7 V. The Cu(100) surface exhibits flat terraces throughout the entire cathodic potential range. Close to OCP, step edges start to corrode. In contrast to Cu(100), the first layer of Cu(111) is converted to an atomically rough and defected surface—associated with nascent surface oxidation at potentials positive of −0.7 V. This surface oxidation is correlated with nitrate reduction.

We examined the effect of the structure of corrosion inhibitors benzotriazole (BTA) and 1,2,4-triazole (TAZ) on the Cu removal rate during chemical mechanical planarization. Removal rates were higher for solutions containing TAZ than solutions containing BTA. Corrosion inhibitor films were characterized using atomic force microscopy, cyclic voltammetry, impedance spectroscopy, surface-enhanced Raman spectroscopy, and mass spectrometry. Inhibitor films formed from TAZ were thicker, more permeable, and rougher than films formed from BTA. The addition of glycine to the corrosion inhibitor containing solutions showed an increase in corrosion for the TAZ-covered Cu surface, little change for the BTA-covered surface, and an increase in removal rate for both systems. This study correlates removal rate with the physical properties of these two different corrosion inhibitors.

The adsorption of saccharin on roughened nickel electrode has been studied with cyclic voltammetry (CV) and surface-enhanced Raman spectroscopy as well as complementary discrete Fourier transform calculations. CV measurements reveal a strong influence of both chloride concentration and pH of electrolyte on nickel surface and the inhibitory effect on nickel oxidation by the saccharin addition. Surface-enhanced Raman spectroscopy (SERS) measurements provide evidence for the strong interaction between the Ni surface and saccharin additives and dependence on the electrode potential. The overall behavior observed in SERS is explained based on the normal Raman spectra of saccharinate anion and its nickel-saccharinate complexes. Above the potential of zero charge (pzc) of the Ni surface, saccharinate-associated SERS peaks are dominant with O-bound nickel-saccharinate characteristics. Below the pzc, the SERS profile experiences a huge intensity enhancement and spectral change as well as substantial Raman frequency shifts. These shifts are interpreted in terms of a saccharinate-Ni complex, with both O and N bonding to the Ni electrode surface.

We examine the effect of different anions in solutions containing benzotriazole (BTA) on the Cu removal rate during chemical mechanical planarization (CMP). In solutions containing both Cl−Cl− and BTA, the Cu removal rate is nearly a factor of twenty lower than in solutions containing either Cl−Cl− or BTA alone. As-grown BTA films from solutions containing different anions are characterized using atomic force microscopy, ellipsometry, Raman spectroscopy, mass spectrometry, and open-circuit-potential measurements. Films grown from halide-containing solutions are found to be considerably thicker than those grown from other anions. The difference in Cu removal rate correlates well with the different as-grown film thicknesses.

In situ surface enhanced Raman scattering (SERS), infrared-visible sum frequency generation (SFG) spectroscopy, electrospray ionization mass spectrometry (ESI-MS), and detailed calculations are used to examine the interaction of bis(3-sulfopropyl)-disulfide (SPS) and mercaptopropylsulfonic acid (MPS) with Cu surfaces both in the absence and presence of chloride. MPS and SPS SERS spectra are similar to each other, and the lack of a specific S–S feature suggests that SPS interacts with Cu as a thiolate. SERS data show a significant shift in the S–O sulfonate region of the SPS or MPS spectra upon the addition of Cl−Cl− , a result which is interpreted as arising from CuCl coordination to the SO3SO3 moiety of these molecules. CuCl association with SPS is also evidenced in the ESI-MS data. The specific spectral shifts are corroborated by density functional theory calculations. In situ SFG measurements provide evidence that SPS and MPS interact with the Cu surface in solutions not containing Cl−Cl− at negative potentials. However, there is no evidence in either the SFG or the SERS for SPS or MPS-Cu coordination in the presence of Cl−Cl− at any potential. The data suggest that the thiolate moiety in the additives is not an active component for the acceleration behavior observed.