Nowadays, a planar DC magnetron sputtering system has been one of the most popular thin-film fabrication tools and has been widely employed in the microelectronics industry. As the semiconductor industry sets in 300mm wafer process era and the flat panel display industry enters mass production system where the size of the glass substrate is greater than one square meter, the dimensions of magnetron sputtering system have continued to grow. However, the scale-up must sustain the major performance indices such as target utilization, target life-time, deposition rate, and deposition uniformity, which requires the reduction of development cost and time through systematic computer-aidedengineering(CAE). In this study, we present a computational study on plasma generation and film deposition in a planar DC magnetron sputtering system. Design optimization of a planar DC magnetron sputtering system needs a precise multi-scale simulation considering a target erosion by magnetron plasma, a macro film deposition by collisional transport, and a micro deposition topography by collisionless transport.
The physical interpretation of magnetron sputtering process mainly consists of three parts; a target sputtering erosion by magnetron plasma, a macro film deposition onto substrate, and a micro film deposition onto micro deposition geometries such as trench or via-hole. But their modeling time and dimension scales are very different from each other so that most conventional simulations of magnetron sputtering system have been limited to the individual simulation of each physical step until now. In order to integrate the whole above physical steps, it is necessary to develop a systematic multi-scale simulation scheme including the definition of data flow between each physical step and the efficient particle-based algorithm per each physical step. In this study, we present a systematic CAE methodology, integrating all of the physical steps, such as plasma generation and macro/micro film deposition processes in magnetron sputtering system.
In PIC-MCC target sputtering erosion simulation step, at frist, we present a discussion about how to determine the optimum detection range for the quasi-steady state of magnetron plasma in two-dimensional PIC-MCC plasma simulation. Within the optimum detection range determined for each experimental set, we have measured the ion current density distribution on the sputter target simulated by two-dimensional PIC-MCC magnetron plasma simulation using XOOPIC. The simulated ion density distribution has passed two calibration steps for the energy dependence of sputtering yield and the gas rarefaction effect due to sputtering wind. Finally the simulated erosion profiles are compared with the experimental erosion results.
In macro film deposition simulation step, Thompson kinetic energy distribution and cosine angular flux distribution are used for the initial sputtered atoms ejecting from the cathode target. A variable hard sphere model is used to calculate the collision cross-section of sputtered atoms. To consider the multiple collisions of fast sputtered atoms within a global time step, we applied a pseudo-time step for fast sputtered atoms. To calculate the post-collision velocity of scattered atoms, isotropic scattering model is used. Bilinear weighting algorithm is used to count the deposition rate of sputtered atoms onto substrate, and the simulated deposition profiles are compared with the experimental profiles for various process pressures and target-substrate distances.
In micro film deposition simulation step, to consider the effect of the asymmetric angular flux distribution on the micro deposition topography, we present a new acceptance-and-rejection method, which can generate an asymmetric angular flux distribution, based on Monte Carlo method, and periodic boundary condition is applied for the vertical domain boundaries. To represent evolving film surface in three-dimensional deposition geometries, an equi-volume rate model is used, and the simulated micro deposition profiles are compared with the experimental profiles along radial distance. The variation of the degree of asymmetry and the step coverage at the side/bottom wall along radial distance are simulated for various process pressures and target-substrate distances.
In addition, through the design optimization based on our multi-scale simulation, we have developed a new high performance rectangular type magnetron sputtering systems of high target utilization, which is called rocking magnet gun. Compared to the conventional rotary magnet gun, the target erosion rate and utilization are enhanced in the rocking magnet gun by restricting the rotary angle of moving magnet within an elaborate rocking angle.
In conclusion, we present a multi-scale simulation scheme integrating the plasma generation and macro/micro film deposition processes of the magnetron sputtering system, which has been limited to the individual simulation of each physical step in the past. Furthermore, we present a systematic CAE methodology of magnetron sputtering system through the multi-scale simulation of plasma generation and film deposition in a planar DC magnetron sputtering system.