Comparing ionized physical vapor deposition and high power magnetron copper seed deposition

Abstract

A computational modeling comparison is made between ionized physical vapor deposition (IPVD) and high power magnetron (HPM) deposition of copper. For the comparison the point of view of the feature scale is stressed where the two reactors are distinguishable by the magnitude and ratio of specie (Cu,Cua,Cu+,Ar+) flux, the angular distribution of the specie, and the energy of the ions incident on the feature surface. The HPM is characterized for the conditions studied by a metal flux content made up almost entirely of copper athermals, an Ar+ ion flux about four times the Cu metal flux, decreasing Cu+ fraction and increasing Cu athermal flux to surface with increasing target power, and both no sputter and sputter regimes at the wafer possible. The IPVD reactor is characterized for the conditions studied by a Cu metal flux with a large neutral fraction but significant ions and athermals, an Ar+ ion flux on the order of the Cu metal flux, and only a sputter regime at the wafer possible. An increase in target power increases the deposition rate and decreases the Cu+ fraction in both systems. In IPVD the bottom coverage increases and the side wall coverage decreases due to a decrease in the sputter rate and an increase in the Cu neutral and athermal fraction. In HPM bottom coverage is reduced with increasing target power due to the lower Cu+ fraction. An increase in wafer power decreases the deposition rate in both systems by increasing the sputter rate. A lower ion current to the wafer for IPVD versus HPM gives the ions a higher energy at the wafer for the same power. In HPM lower energy ions are sufficient for the same sputtering rate versus IPVD due to the higher ion/neutral fraction. With no wafer bias HPM has thicker bottom versus IPVD since no sputtering of the feature bottom is occurring and the more focused athermals (versus neutrals) are less shadowed to the feature bottom. The IPVD side wall deposits have more thickness variation than HPM due to the large Cu neutral component in IPVD. An increase in wafer power increases side wall coverage and decreases feature bottom coverage in both systems as metal deposited at feature bottom is redeposited to the sidewalls. For increased coil power in IPVD the Cu+ fraction increases and the Cu neutral fraction decreases. Both bottom and side wall coverage increase as more Cu enters the feature as focused ions.