|
Porous low-k thin films
|
Metal DiffusionCu is known to have high diffusivity into most of the promising low-k materials. The presence of Cu in low-k materials will result in serious device degradation and failure. PALS is found to be very sensitive to the diffusion of metal (Al or Cu) interconnects into the porous network of a low-k film with interconnected porosity. |
|
In the process of studying an open-porosity silica film, we attempted to cap the film by depositing 80 nm of thermally evaporated aluminum without the use of any sample cooling. Instead of the 140 ns component being shortened to a 98 ns component indicative of Ps diffusing within the interconnected pore structure of the film, the long-lived component totally disappeared and a short, 3 ns, component appeared in its place. We confirmed that the sample warmed up during the deposition, resulting in aluminum diffusion into the pore structure. A metal coating on the pores would explain the appearance of the 3 ns component since Ps annihilation would be enhanced by the high density of free electrons. We verified that the silica film did indeed have a large open pore structure by room temperature sputter-deposition of an Al cap and fitting a single 98 ns lifetime component. These results are shown in Figure 1.
Figure 1. PALS spectra of the sputtered Al- and evaporated Al-capped porous films
|
|
To test the metal interdiffusion hypothesis we performed a systematically controlled study involving the Al sputter-capped silica film with an inherent lifetime component of 98 ns. We heat-treated the film at progressively higher temperatures and collected data at room temperature in between each heating cycle. We monitored the intensity of the 98 ns component and searched for the appearance of a short 3 ns lifetime component. The results are shown in Figure 2. At approximately 450 ºC we see the apparent onset of Al interdiffusion. This temperature is consistent with that of bulk annealing of Al and is probably significantly higher than that experienced by the sample during the thermal evaporation process. We tentatively conclude that the sample temperature during vapor deposition is much more critical than subsequent processing temperatures in preventing metal interdiffusion.
Figure 2. PALS fitted intensities of the 3 ns and 98 ns components in the sputtered Al-capped porous film
|
|
Metal diffusion into porous low-k dielectrics has also been observed in Cu-capped porous films, as shown in Figure 3. A Cu-capped porous silica film presents two Ps lifetime components of 35ns and 3.5 ns instead of the single 94 ns component observed in the oxide-capped silica film. This reduction of the Ps lifetime is almost certainly due to Cu coating on the inner pore surfaces where Ps annihilation would be enhanced by the high density of free electrons in Cu. There appears to be significant Cu diffusion into the silica pore surfaces, even though the sample was held at room temperature in the sputter deposition process. (In this regard Cu diffusion is worse than Al diffusion.) These deductions are not surprising, as the need for diffusion barriers with Cu metalization on porous silica is an accepted fact. |
|
Figure 3. PALS spectra of uncapped, silicon oxide capped and Cu capped porous films References:
|
