Minutes of the IR Engineering and Physics Meeting of 3 May 96 B1/Q1 Ion Pump Design Chuck Perkins has developed a design for a "Radial Ion Pump" between B1 and Q1, which uses the 15 kG field of the solenoid. The pump completely surrounds the beampipe, with a nominal operating pressure of 5 nTorr. The anodes run at 5.5 kV, which is the same potential as the arc DIP's, so identical power supplies can be used. The net pump speed on beamline is 315 L/s, including the conductance of the slots in the beampipe. The gross speed of the pump could be increased, but the larger-than-optimal cell radius of 0.18 inches provides better pumping at lower pressures, and is more tolerant of transverse magnetic fields which will be present, due to the stray fields from the P.M. B1 and Q1 magnets. The peak theoretical pumping speed requires a cell radius of 0.10 inches, and a cell length of 0.72 cm. Although the pumping speed per cell remains relatively constant for larger cell radii, the number of cells which fit in the area available decreases, so the total pump speed drops. This drop is 20% to 1500 L/s (saturated), when cell size is increased to 0.18 inches. However, the net speed on beamline is determined mostly by the conductance through the slots, so this 20% reduction of gross speed has a negligible effect on the net speed. Tolerance to Stray Fields The first cell of the ion pump is 3 cm away from the face of Q1, and the last is 16 cm away. At 3 cm, the transverse stray end fields from the Q1 quad and dipole fields is about 1.5 kG, based on simulations done by Gordon Bowden. This is 10% of the main solenoid field, directed in the plane normal to the solenoid and the cell axis. This effectively mis-aligns the field with respect to the cell axis, making the apparent diameter of the cell smaller, and the length longer. Both of these combine to reduce the pumping speed. Theoretically, at 12 degrees mis-alignment (21% transverse field), the pumping speed starts to drop off quickly. At 15 degrees, the speed is reduced by 32%. Empirical data shows that this trend begins at 10-15 degrees, but that the knee is not as step as theoretically predicted. Other test data shows that, for larger cell diameters, the pumping speed stays high to a larger angular mis-alignment, but then the knee is steeper. This suggests that the apparent length/diameter ratio of the cell should be in the range of 1.5-2. Above 2, the tested pumping speed drops off quickly. To ensure that the B1/Q1 Pump would not be sensitive to transverse fields which could effectively extinguish part of the pump, the cell radius was increased from 1.8 mm to 2.4 mm, making the new L/D ratio = 1.5. With this change, the knee of the pumping curve moved from 12 degrees to 20 degrees misalignment angle (36% transverse field). Conductance of Slots Since the gross speed of the pump is so high, the net speed on beamline is largely determined by the conductance of the slots in the beampipe. The baseline configuration assumes 12.5 cm long slots, 3 mm wide by 5 mm deep, backed by a 1 mm thick screen with 3 mm diameter holes located on 5 mm spacing. To minimize the possibility of dust dropping into the beamline from pump material above the chamber, the slots run only on the bottom 60% of the chamber. At the current radius of the beampipe, 40 slots fit. Using the more conservative of the theoretical conductance equations, and scaling down by 25% to correct for experience with the HER slots, gives a conductance of 414 L/s (10.4 L/s/slot). This low conductance throttles the high gross pumping speed down to 315 L/s on beamline. Changing either the number or width of slots has the greatest impact on pumping speed. If the slots completely circled the beampipe, 60 slots would fit, increasing the net speed by 70% to 418 L/s. However, dust sputtering into the beamline becomes an issue. Since these pumps can only operate when the detector solenoid is on, their expected maximum pressure should be fairly low. However, sputtering is typically a problem when the pump is run at higher pressures and currents. Thus, we may not have to contend with this problem. Alternatively, increasing the slot width from 3 mm to 4 mm increases the net pumping speed by 40%. Two issues affect the width of the slots. First, they produce an impedance to the image currents in the chamber wall. This was agreed to be negligible for two short pumps such as these. Second, the wider slots could allow more TM-mode radiation into the pump volume. This could heat the pumps or heat the thin screen behind the slots. Either could be very bad, since neither are thermally grounded very well. The H.O.M. characteristics of the two slot configurations will be studied before a decision is made, but the direction will be to try to increase conductance, if this does not pose heating problems. Pump Design The design of the pump is modeled after the Damping Ring DIP's. It has thick stainless steel anode plates, with 0.189 inch diameter holes drilled through them. The anodes are stood off the thin titanium cathode plates with a ceramic stud-and-cup arrangement to minimize leakage current, despite sputtering. The cathodes, at ground, are joined together by three threaded rods, and heat sunk through BeCu grounding fingers. The anodes are joined to the cathodes with the ceramics, then maintained at HV using a standard PEP-II HV feedthrough. Since space outside the pump is very tight, the feedthrough is interlaced with the BPM electrodes on the B1 Chamber, and the connector must be right-angled. This feedthrough and connector should be shielded to prevent RF coupling from the beam. The entire pump is tapered to fit in the tapered volume of the pump can and chamber. However, this makes assembly difficult, so the final pump configuration should be cylindrical. Near IR Pressure Profile Lou Bertolini ran new pressure profiles for the Near IR, using the 315 L/s B1/Q1 Ion Pump design presented by Chuck. Peak and average pressure values are shown below: Run Description B1/Q1 Speed Peak Press. Avg Press. "Old" Configuration 100 L/s 17 nTorr 6.5 nTorr "New" Configuration 315 L/s 10 nTorr 3.5 nTorr No B1/Q1 Pump at all 0.00 L/s 50 nTorr 20 nTorr On big qualitative difference is that the new configuration, with more pumping at B1, looks to be more conductance-limited, whereas the old configuration did not have sufficient pumping at B1, where gas is desorbed off the B1 masks (esp. on the LEB-upstream side). Lou estimated that the pumpdown time of the Near IR, from the "No Pump" configuration, when the solenoid is not on, should be 25 minutes. This is sufficiently short that it should not add to the recovery time of a detector access. Discussion on Backgrounds Since space is limited in the region between B1 and Q1, trade-offs must be made between maximizing pumping, at the expense of shielding, or putting in more shielding, and cutting down on pumping, and increasing the pressure through the I.P. Dave Kirkby felt that, based on previous simulations, the detector backgrounds were most sensitive to particles due to photons coming in from the upbeam HEB (-Z side). Thus, shielding may be more important on this side than pumping. Ideally, 5- 10 R.L. of shielding would be needed. Lou Bertolini will run another pressure profile with no pump on the -Z side, and the 315 L/s pump on the +Z side. Also, there being a consensus about the possible increases in slot conductance, Lou will also run a profile using 500 L/s net pumping on both sides. Dave Kirkby will then run background simulations for all three configurations: 1) 315 L/s pumps on both sides and 1-2 cm of shielding 2) 500 L/s pumps on both sides and 1-2 cm of shielding 2) 315 L/s and 1-2 L/s shielding on +Z side, 0 L/s and 4-5 cm shielding on -Z side. Martin Nordby and Chuck Perkins will also develop alternative slot configurations for modeling using MAFIA. These minutes, and agenda for future meetings, are available on the Web at: http://www.slac.stanford.edu/accel/pepii/near-ir/home.html