AVS 59th Annual International Symposium and Exhibition
    Advanced Surface Engineering Tuesday Sessions
       Session SE+PS-TuM

Paper SE+PS-TuM3
The Magnetic Field Configuration’s Effect on Plasma Parameters in High-Power Pulsed Magnetron Sputtering

Tuesday, October 30, 2012, 8:40 am, Room 22

Session: Pulsed Plasmas in Surface Engineering
Presenter: H. Yu, University of Illinois at Urbana Champaign
Authors: H. Yu, University of Illinois at Urbana Champaign
L. Meng, University of Illinois at Urbana Champaign
P. Raman, University of Illinois at Urbana Champaign
T.S. Cho, University of Illinois at Urbana Champaign
D.N. Ruzic, University of Illinois at Urbana Champaign
Correspondent: Click to Email
Magnetic field design is crucial in DC magnetron sputtering operation, but has been largely overlooked in high power pulsed magnetron sputtering (HPPMS). In a HPPMS discharge, plasma disperses after each short pulse, unlike in the DC operation which requires a good magnetic confinement to maintain the plasma. It is thus interesting to study the effect of magnetic field configuration on HPPMS discharge for further optimization. A special magnet pack was fabricated with the magnet positions fully adjustable. Different designs were made, with the corresponding magnetic field calculated by Comsol. For example, the magnetic field (B) strength in the racetrack was varied, as 200, 500, and 800 Gauss. Different racetrack shapes were used, such as ring-like and bean-like. Designs were also made to have incomplete racetrack, and weak racetrack in which piles of magnets were placed evenly across the whole pack and polarities of adjacent piles were opposite. These magnetic configurations were then tested in a large magnetron system with a 36 cm target using a conventional HPPMS plasma generator. The I-V discharge characteristics were measured. A time-resolved triple Langmuir probe was employed to study the temporal evolution of electron temperature (Te) and density (ne) across the substrate. Ionization fractions of sputtered metal were also measured using quartz crystal microbalance combined with electrostatic filters. The results showed that higher B field strength and longer racetrack produced higher pulse current. The 500 Gauss configuration however had higher ne in the pulse than 800 Gauss did, likely because it allowed easier plasma diffusion. In all the designs with a racetrack, the distribution of ne on the substrate was non-uniform that ne was typically about 10 times higher right below the racetrack than at the center. The weak racetrack configuration was found to work and showed certain superiority over the normal DC magnet design. The plasma was generated from almost the entire target surface indicating an improvement in target utilization without the need for rotation. The discharge current was comparable with the racetrack designs. Furthermore, the plasma density distribution over the substrate was very uniform.