Magnetron sputtering cathode for low power density operation

In conventional magnetron sputtering discharges, E_z × B_r drift motion forms a doughnut-shaped plasma discharge region on the target surface. The doughnut-shaped plasma discharge contributes to plasma confinement on the target surface, and this more efficient plasma confinement enables operation at lower input power for plasma production, meaning that sputtering at lower power density is possible. Here, the power density (W/cm²) is generally defined as the input power divided by the target area. A sputtering process with low power density is useful to obtain a very smooth surface and preferential orientation of the grains.^1,2 From the viewpoint of magnetron sputtering process efficiency, the design of the magnetic field profile is highly important because the electric-field lines generally align in a vertical direction with respect to the target surface. A typical example of magnetic-field profile modification is an unbalanced magnetron sputtering, wherein the divergent magnetic field from the sputtering target to the substrate increases the ion and electron fluxes to the substrate and the self-bias voltage of the target.³ To improve the plasma confinement efficiency in the sputtering chamber, a multipolar magnetic plasma confinement setup has been added to unbalanced and conventional magnetron sputtering devices.^4,5 In addition, magnetic field control has been applied to electron cyclotron resonance and high-power pulsed sputtering.^6,7 Although many approaches have been followed to improve plasma confinement efficiency and low power density sputtering process results, a conventional planar magnetron cathode with a simple configuration is widely used in industry. Here, a novel planar magnetron sputtering cathode with magnetic mirror configuration is proposed, for low power density operation.

Figure 1(a) shows a schematic drawing of the Magnetic Mirror-type Magnetron Cathode (M3C). The magnetic field configuration of the M3C was designed, through simulation-based studies, to achieve an area of large |B_r| only above the target surface. The magnetic field lines around the target are shown in black. The magnetic mirrors positioned near the center and edges of the target surface are shown by red dashed circles. Figure 1(b) shows a top view of the two samarium-cobalt cylindrical permanent magnets embedded in the M3C. Figure 1(c) shows the calculated radial component of magnetic field strength, |B_r|, on the target surface. The |B_r| profile influences the plasma structure through the E_z × B_r drift motion. Actual |B_r| profile will be shifted outward by the conservation law of magnetic flux. The two permanent magnets, PM1 and PM2, are magnetized radially and axially, respectively. Both PM1 and PM2 are positioned behind the target, aligned radially with the center of the M3C. The angle of the magnetic-field line on the target center is vertically aligned as the inner diameter of PM1 decreases. The inner diameter of PM2 is the same as the target diameter. The top edge of PM2 is positioned at half the height of PM1 to maintain a vertical magnetic field near the target edges. The PM2 affects the magnetic mirror configuration at the target edge. The dimensions (thickness and height) and vertical position of PM2 were optimized on the basis of the various calculation results. The magnetic yoke, made of SUS430 (ferromagnetic metal), confines the E_z × B_r drift motion above the target surface. The magnetic mirror configuration positioned near the target edges is also controlled by the bracketing ends of the yoke. The magnetic mirror ratio is approximately 25, calculated on the basis of respective magnetic field strengths of ∼150 mT (|r| = 0, 25 mm and z = 0 mm), and ∼6 mT (|r| = 25 mm and z = 20 mm). The PM1 should be axially magnetized in the conventional magnetron sputtering cathode, so that the magnetic flux not converge effectively. A water-cooling component is installed inside the M3C.

The Ar gas pressure and flow rate were controlled over the ranges of 0.10-0.70 Pa and 7.0-47.0 sccm, respectively. A DC power supply (DCS0052B, ULVAC) with a power control range of 5-100 W was used to produce the plasmas. The aluminum target was 50 mm in diameter and 5.0 mm in thickness. Therefore, the target surface was roughly 20 cm². Eagle XG® (10 × 10 × 1.1 mm, Corning) was used as a substrate. The substrate center was positioned at |r| = 0 mm and z = 280 mm (i.e., target-substrate distance d = 280 mm).

The sputtered film thickness was measured by atomic force microscope (AFM: VN-8010, Keyence), and the sputtering rate was calculated by dividing the film thickness by the sputtering time of 60 min.

To compare the plasma discharge light emission patterns, the patterns were captured by an industrial camera (UI-3260CP-C-HQ Rev.2, iDS) with an interference filter (center wavelength, 488 nm; FWHM, 10 nm; Edmund Optics), because the spontaneous emission intensity is proportional to the electron density.^8,9 The erosion profiles of aluminum target were measured by stylus-type step profiler (SURFCOM 1800D, ACCRETECH TOKYO SEIMITSU).

Figure 2 shows a typical image of the plasma discharge using the M3C. The experimental conditions were as follows: Ar gas pressure p_Ar ∼ 0.10 Pa (7.0 sccm) and input DC power P_DC = 15 W. The light emission pattern of Fig. 2 is consistent with the |B_r| profile in Fig. 1(c). The film thickness was between 10 to 15 nm after 60 min of sputtering, so that the sputtering rate and the power density were estimated as ∼0.2 nm/min and ∼0.75 W/cm², respectively. It should be noted that the sputtering rate was 0.2 nm/min, despite the extremely long target-substrate distance of 280 mm, demonstrating that the M3C could achieve a practical sputtering rate. The power density and process gas conditions of various DC sputtering studies, including the present one, are listed in Table I. As can be seen, the power density of the M3C is sufficiently low compared with previous film production studies using DC discharges.^10–12 In general, pulsed discharge techniques are useful for low gas pressure operation of less than 0.5 Pa.^13–15 Plasma confinement by a magnetic mirror enables DC sputtering with low power density, in spite of low gas pressure of less than 0.5 Pa.

Figure 3 shows the dependence of the plasma discharge light emission pattern on p_Ar and P_DC. Low power density operation of 0.25 W/cm² was confirmed under experimental conditions of p_Ar = 0.26 Pa and P_DC = 5 W. Although the emission intensity increases with the increasing p_Ar, the doughnut-shaped plasma discharge region narrows. In practical terms, the short ignition discharge step at ≥ 0.7 Pa is required before a stable sputtering operation of < 0.7 Pa can be achieved. The gas pressure conditions of the ignition process generally requires a few pascals, even with the use of commercially available RF discharge cathodes. An unexpected plasma discharge between the target edge and the bracketing yoke occurs at p_Ar = 0.70 Pa and P_DC ≥ 30 W, shown in Fig. 3 as an outer blue ring. The electric field configuration should be optimized to obtain a stable plasma discharge at ≥ 0.70 Pa operation. In addition, a large-sized sputtering cathode with the same magnetic field configuration will be studied, to seek even lower power density operation.

Finally, we consider the plasma discharge region of M3C. The outer edge of the light emission region (the inner blue rings in Fig. 3) reaches to the target edges at p_Ar ∼ 0.10 Pa, suggesting that a higher magnetic mirror ratio near the target edge may effectively function to extend the plasma discharge region outward; and widening of the plasma discharge region induces higher target utilization efficiency. Figure 4 shows the erosion profiles of an aluminum target and the sputtering operation time dependence. The experimental conditions were p_Ar ∼ 0.10 Pa and P_DC = 100 W. The plasma confinement effect was effectively occurred at |r|∼ 5-25 mm as shown in Fig. 4. The erosion profiles have similar shape and these erosion-depths increase in proportion to the operation time. The plasma discharge region is determined by the balance between the confinement of E_z × B_r drift and the diffusion of plasma particles; so that a low pressure discharge with a long diffusion length leads to a large E_z × B_r region, resulting in significant target erosion area. The target utilization efficiency of M3C, using the ratio of sputtering volume to target volume, was roughly estimated at ∼50 % which is of practical value.

A novel magnetron sputtering cathode with magnetic mirror configuration has been developed, for low power density operation. Stable operation with a low power density of 0.25 W/cm² was successfully demonstrated with a DC plasma discharge of 5 W, and this power density is sufficiently low compared with previous film production studies. The sputtering rate was ∼0.2 nm/min with a DC power density of 0.75 W/cm², Ar gas pressure of 0.15 Pa, and target-substrate distance of 280 mm. The addition of a magnetic mirror configuration in the vicinity of the outer edges of the target resulted in a wide plasma discharge region reaching near to the edges of the target in low gas pressure operation.

The authors wish to acknowledge the assistance provided by Mr. Satoru Fukamachi and Mr. Osamu Matsuda of AIST.