A thin film cavity deposition system by ECR plasma inside an elliptical cavity is presented. The system uses the substrate copper cavity as the vacuum chamber. The ECR plasma will be created to produce direct niobium ion deposition. Central grid cylinder is biased to realize the energy controlled vacuum deposition. This report describes the design of several subcomponents including the vacuum chamber, RF feeding, biasing grid and magnet coils. Operation parameters are compared between the sample deposition system and this cavity deposition system, it is concluded that the plasma ignition in cavity deposition system is feasible.
Niobium thin film cavity deposition for a 500MHz copper cavity
G. Wu, H. L. Phillips, H. Wang, A-M. Valente
1. The vacuum chamber
The 500 MHz copper cavity is part of the vacuum system. A mechanic roughing pump and a CTI cryo-8 cryopump provides vacuum pumping. Based on the past experience on E-gun operation, a getter pump will be added to the system to pump the hydrogen residual gas.
Figure 1 is the illustration of the 500MHz cavity vacuum system. The copper cavity from Cornell University has a 13.25-in conflat flange build in at the end of the beam pipe. A box shaped vacuum chamber serves as the main housing of the 15-kW Rod-Fed e-gun. A customized vacuum Tee forms the main unit of the head assembly on top of the 500MHz copper cavity. All the feedthroughs will be incorporated in the head assembly shown in Figure 2. One getter pump will be attached through the head assembly. If necessary, a second getter pump will be connected through the side port, which will be very close to the E-gun main unit. The cavity, head assembly and one pancake coil (described later) will be put together inside clean room prior to the final assembly on top of the e-gun chamber.
2. RF system and the grid circular waveguide
The RF system is the same as the RF system used in sample deposition system with a 30-in long circular waveguide. RF travels in the circular grid in TE11 mode. The waveguide is also a grid floated against the base system.
Figure 3 is the brief diagram of the RF circuit.
Fig. 3 ECR-RF diagram
Same dimension RF window is adopted to the 10-inch conflat flange.
The 4.5-inch (ID) cylinder grid will be attached to the taper with a ceramic insulator. The thickness of the open gap is modeled to see whether the RF traveling will be affected.
Figure 4 shows the open gap located in the standing wave node where magnetic field is the maximum. The S11 due to the 0.125-in gap radiation is listed in table 1.
Figure 4. The open gap for radiation calculation.
Distance from end wall
Table 1. S-parameter for gaps at different standing wave position
For 0.0625-in gap, S11 for the 6.00-in gap position decreased to 0.96552, which translates to the 7% power radiation.
The guided wavelength of 2.45GHz for 4.5-in circular waveguide is 6.2-in. If we use a 33-in cylinder grid, the position of the gap is located at the standing wave E-maximum. Then the power radiation is minimized to less than 1%. If the plasma is on, then the RF becomes traveling wave. The magnetic field at the open gap is one half of the standing wave maximum; hence the radiation power is estimated at 1.75%.
Since we need to open the end of the cylinder grid to allow the niobium flux to travel in, the RF will radiate out once again. The S11 of a 3-in open hole for the circular waveguide is 0.46967, which translates 80% of RF power radiated. When two slots are added to the open holes like that in Figure 5 a, the S11 increases to 0.89852.
Since the current flows in a curved fashion on the end wall like that in the Figure 5 b, it is conceivable that adding curved slots like that in Figure 5 c would benefit the radiation shielding.
Fig. 5 The two slots added to the waveguide end (a), the surface current seen on the end wall of circular waveguide (b), the curved slots added to the waveguide end (c).
The S11 for the total four curved slots is 0.93413. If the end hole is reduced to 2.75-in diameter from 3.0-in, the S11 further reduces to 0.9543, which translates to 9% of power radiated. If there is plasma, the RF usually gets absorbed before it reaches the end openings.
Finally, the RF power can also be radiated through grid holes. The model in Figure 6 illustrates a hole pattern covers grid cylinder in a quarter wavelength. The hole is 0.25x0.25-in2 square. 40 of them evenly distribute azimuthally around the circular waveguide. Each ring is 0.34375-in apart along circular axis. The S11 due to 40x5 holes is 0.99924, which translates to 0.15% power radiation. For a 33-in long grid cylinder, all the grid holes have total 2.9% power radiation.
After all the radiation boundaries are considered, the total radiation power could be around 3.2% to 11.9% of input power. Table 2 lists the radiation contribution both for standing wave and traveling wave. One would note that for traveling wave situation, the plasma would absorb majority of the RF energy, leaving the radiation contribution even less. For a typical 250-watt input RF power, the RF leaked into the vacuum chamber will be 8.3 watt to 30-watt.
Table 2: Radiation contributions of the grid
Fig. 6 The radiation model for a grid cylinder.
3. Magnetic coils
Several configurations of permanent magnet/coil have been investigated. The all coil option as shown in Figure 7 gives the largest field uniformity over a large volume. The field in center 2 cm2 volume has magnetic field of gauss. Such a big volume with uniform field serves a trap to catch the neutral niobium vapor through electron cyclotron resonance.
Center coil is 66m, while top and bottom coils are 8.5m long. Three coils have total 83 m. The copper tubing has cross section of 1.3529 cm2. Total resistance for 83 m copper tubing is 0.00933 ohm.
For a 1779A current which provides the 875Gauss magnetic field, the total voltage drop is 16.6 V. Total tube joule heating is 30kW. For 26 m center coil alone, there will be 207C temperature increase. That means the center coil has to be split into three sections to allow the sufficient water-cooling. Figure 8 is the cavity deposition system with the coils added.
Fig. 7 The magnetic field by coils and yokes for ECR condition.
Fig. 8 The 500MHz copper cavity coating system
4. The bias structure.
The bias voltage will be applied to the insulated grid. Due to the potential for the insulator to be coated, a simple step such as that in Figure 9 in the connecting flange is needed to shield the niobium ions.
Fig. 9 The insulating step for the grid cylinder.
5. The plasma estimation
The plasma ignition is the complex process to simulate. Currently XOOPIC program can be used to simulate some of the plasma process in 2D, but not the ignition process nor in 3D. The PIC/MCC code development and XOOPIC code modification are both in progress. Meantime, some simple comparison between the cavity deposition system and the sample ECR system should provide some meaningful discussion. The critical condition for ECR ignition is the magnetic field configuration and the vapor flux density. Figure 10 shows the significant field uniformity in a 3x3-inch cross-section for the sample ECR system and the cavity ECR system. The field uniformity of the large area directly translates the ionization rate of the ECR neutrals.
Fig 10. The 3D visualization of the magnetic field in a 3-in-by-3-in cross section of the sample deposition ECR coil (a) and the cavity deposition coil (b).
The niobium neutral flux in the sample deposition system had an average 130A/s rate at 6-in distance. That was provided by 7.5kW e-beam gun. The distance from e-gun to the ECR volume will be around 15-in in cavity deposition system, which means the neutral flux rate will be reduced to roughly one sixth of the same e-beam gun. The e-beam gun for the cavity deposition system is expected to provide 14kW electron heating, which means two times the flux of that 7.5kW e-beam gun. Once the rod-fed mechanism is employed to the e-beam gun, the evaporation rate is expected to increase by three times according to the estimation by the e-beam gun vendor. Overall, the neutral flux rate should be quite close to that of the sample ECR system. The much better magnetic field configuration in cavity deposition system is expected to significantly reduce the neutral flux rate required by plasma ignition.
Future improvement to further reduce the ignition flux rate will be incorporating some coil/permanent magnet solution to provide magnetic mirror field to confine electrons to the ECR region both in Z-direction and the R direction.
6. The potential technical challenges
It is estimated the grid cylinder to be well below the 500C based on the thermal radiation. Whether the grid will experience overwhelming heating due to the e-gun thermal or electron radiation is unknown. Thus it remains a challenge that whether the grid expansion will seriously deform the grid, changes the field pattern or simply melts at the end.
The uniformity of the coating between beam pipe and the cavity equator is unknown, if uniformity were not achievable, what would be the sensible remedies. A simple plasma simulation code is under development to estimate the niobium ion flux at different locations.
The residue hydrogen will be difficult to remove, since the e-gun power may be quite close to the maximum 14kW. That will affects film growth or the final quality.
The simulation of the niobium film growth under molecular dynamics level is being pursued under the collaboration with Cornell University and NIST. Before that is proved accurate enough, there is not much theoretical understanding to guide the deposition process.