Minutes of the IR Engineering and Physics Meeting of 14 Jun 96 This meeting was devoted to an informal status report on the modification of the B1 and Q1 Magnet designs to incorporate the use of NdFeB permanent magnet material. The change is being investigated for its cost and technical merits, and this status report attempted to bring to light the technical advantages and disadvantages of this possible change. Q1 Quad Material Options Andy Ringwall provided a comparison of magnetic and mechanical properties for SmCo and NdFeB permanent magnet material. Sm2C017 is a pinning-type material, depending on precipitated phases forming at cell boundaries to prevent domain wall motion. To achieve high coercivity, the material is quenched to produce this precipitate. This quenching limits the size of the block. NdFeB is a nucleation-type material. Here, the grain boundaries stop domain wall motion, so quenching is not needed, and block sizes can be larger. Other points of comparison: NdFeB Advantages over SmCo Remanent field typically 20% larger than for comparable SmCo Lower cost raw material Less brittle, so larger blocks can be made, and grinding is less risky Larger number of material types and suppliers No liquid phase during sintering, so angle and strength tolerances can be tighter (for die-pressed material) NdFeB Disadvantages over SmCo Lower radiation resistance Lower curie temperature Temperature coefficient is 3 times that of SmCo Requires plating to prevent rusting Lower thermal conductivity More sensitive to de-magnetization from increased temperatures The Q1 magnet design was changed to better fit the BSC profiles of the beams through the Q1 chamber. The new design has a 57 mm radius through the first 80 cm of the magnet (Q1A), then it flares up to the 74 mm radius of the original design (Q1B). The outer radii were set by the strength of the NdFeB material. Scaling the material cost by the savings of material, the cost could be reduced from the prototype design by up to 77% (from $1,300k to $300k). The quad and dipole electro-magnet trims are moved to the outside of the magnet so the inner radius of the Q1A slice could be minimized. The entire package fits within the envelope of the prototype outside diameter, although the in-board end connection to B1 will be more difficult, since Q1A is so much smaller than the prototype. There is room both inside and outside the magnet for thermal shields to isolate the magnet. Depending on the heat load from the Q1 quad trim, a thermal shield may be needed outside the trim as well. B1 was also modified, by using NdFeB, removing the first small slice, and increasing the inner radius of all slices by 1 mm. This gives more clearance for thermal and alignment gap. Q1 Magnet Design Stan Ecklund presented the vital statistics for the quad and trim coils: Quad Blocks Dipole Blocks Q1A: R1 = 57 mm R1 = 92 mm R2 = 86 mm R2 = 117.4 mm 80 cm long Q1B: R1 = 74 mm R1 = 107.2 mm R2 = 101.2 R2 = 120.2 mm 30 cm long This assumed a remanent field of 12 kG, which is fairly standard for NdFeB. Stronger fields could be used, but this would limited the number of possible suppliers, and make it more difficult for them to produce the coercivity needed. The quad trim coils are outside the Q1A dipole blocks, with an inner radius of 140 mm. The trim gives G = 0.92 T/m (5% trim) at 400 A in a water cooled 0.255" square conductor. There are 45 turns per coil in four layers, for a total power consumption of 23 kW (4 coils). Due to the large radius, higher harmonics at the BSC are not an issue. They were in the noise of the numerical accuracy of the simulation (10E-7 for n = 14). BSC's and Synchrotron Radiation Mike Sullivan reported on the effects of the new B1 and Q1 on beam orbits and synchrotron radiation. B1 Magnet Despite the increase in inner radius and removal of slice #1 of B1, the magnet has gotten stronger, producing a 1.5 mrad increased bend to the LEB. This increases the beam separation through Q2-Q5, offsetting losses due to the changes in Q1. The separation at the 1st parasitic crossing is reduced by 0.9%. The first three slices of B1 are now all the same inner radius, large enough to fit over the welded eyelet on the Vertex Vacuum Chamber. This should eliminate the need for a clam-shell type collar. A final effect of the new B1 geometry is that it reduces the amount of offset needed for the Q1B quad field from 21 mm to 16 mm. This reduces the amount of dipole field (and dipole blocks) somewhat. Q1 Magnet Q1A: G = -129 kG/m K = -1.25 Q1B: G = -80 kG/m K = -0.774 This configuration assumes the quad trims are outside the Q1A part of the magnet. Another option which was studied was to add independently rotating quad rings on the out-board end of Q1. It doubles as a trim for Q1 and a replacement for SK1. However, lumping this trim at one spot decreased the beam separations at Q2-Q5 when trimming. Also, locating SK1 here reduced the dynamic aperture, and was not recommended by the Lattice group. Both the new B1 and Q1 designs increase the beam separation. This increase is 4.48 mm at Q2, 4.13 mm at Q4, and 3.10 mm at Q5. The increase is helpful in compensating for the here-to-fore unplanned-for loss of separation due to trimming. With the electro-magnet trims, and the new B1 configuration, beam separations changes for ±5% trim are: Q2: ±0.89 mm; Q4: ±1.35 mm; Q5: ±3.02 mm. The increase also buys back the ~1 mm separation lost when Q2 was changed to the "long-iron" design. Synchrotron Radiation Decreasing the I.D. of the Q1 Chamber brings it more into the path of some SR fans. To accommodate this, 2 extra mask tips are needed on the incoming LEB side in X (149 W and 0.1 W total power). These absorb the first part of the Q1 fan and shadow the B1/Q1 pump region. At the step in the chamber at 1.7 m, the chamber comes near(er) to the luminosity monitor SR stay-clear. On the incoming HEB side, 2 masks (in X) are needed to prevent scattering from incoming Q4 SR. Total power is 22 W and 54 W. Two vertical masks are also needed to prevent scattering into the I.P. region. Some outgoing Q1 SR strikes this chamber also. However, this cannot be masked, and some fraction of these photons cause scattering into the Vertex Vacuum Chamber. This solid angle is small, but backgrounds, due to this new source, could increase by as much as 10% (worst-case). All SR striking these new masks is from the body of the beam, not from tails, so the number of photons is not very sensitive to beam position or dispersion. Radiation Dose on B1 and Q1 Magnets David Kirkby is modifying his background and lost-particle simulations to count strikes on the B1 and Q1 magnets. His is currently refining the radial binning near the inner radii of the magnets to get a feel for expected peak dose at the inner wall of the magnet. David was out of town for the meeting, but should have results next week. Radiation Effects on NdFeB Hobey DeStaebler presented a survey of the available information on radiation sensitivity of NdFeB. This appears to be the biggest potential show-stopper in this design, and it focuses on 4 questions: 1. What is our radiation exposure? (David Kirkby is working on estimating this). 2. What is the effect of radiation on de-magnetizing B1 and Q1? 3. What are the optical effects of this demagnetizing? (Possible to simulate using Stan Ecklund's model). 4. Can we tolerate them? (Simulate by Lattice group and/or Magbends). Of these questions, the effects of radiation on the NdFeB is both the least understood and the hardest to simulate or test. David Kirkby's initial results suggest that radiation dose on B1 and Q1 may be at the Mrad- level per year. A possible mechanism for radiation damage was presented in a paper from Finland. Here, a grain is the smallest unit of material before sintering, while a domain is a region which is all magnetized in the same direction. Assume a domain is grain-sized, or about 1 micron across (1E11 atoms). It is immersed in an external magnetic field. Radiation heats a small region to above the Curie temperature. For a region which is 20% the length of the domain (1E7 atoms), 0.4 MeV would be needed. The resulting dose on this region is 1E4 MeV/g-cm^2. When cooling, the small region magnetizes in the direction of the external field, not that of the rest of the domain. Eventually, domain wall energetics causes the new little domain to grow, consuming the original domain (This has been shown to be the mechanism for thermal demagnetizing in an external field). Since NdFeB is a nucleation-type material, it follows from the above theory that it is more susceptible to radiation-induced de-magnetizing than SmCo. Articles have been found on radiation damage measurements from three labs: LLNL/Monterey Naval Post-Graduate School, TRIUMF, and LANL. Below is a summary of their results. For more details, see the hard-copy: Source Damage Rate Co60 (2 tests) 0% / 50 Mrads Charged Particles 100% / (10^14/cm^2) Reactor Neutrons 1% / (10^14/cm^2) Brems (82 MeV) 1-10% / Grad All measurement were made in the late 80's, but no newer articles have been found yet. The Brems should be most applicable to our application, suggesting that we may see damage at the 0.1% level. Two courses of action seem prudent. First, we need to get the energy spectra for recoil atoms from Coulomb scattering for our design. We should use this in the damage model to predict possible damage rate, and compared with the damage seen in the 85 MeV brem experiment by LLNL/NPGS. Second, we need to look at possible differences in radiation- susceptibility among suppliers, and investigate manufacturer tests. Hobey will also look into running our own test, possibly on the SPEAR injector, or other beamline which could better simulate our expected dose rate and type. Editor's Conclusion No obvious show-stoppers were uncovered in this investigation. However, there are clearly some increased technical risks associated with changing to NdFeB. The most significant ones discussed are (in order of presentation): 1. Increased dBr/dT, decreased thermal conductivity, and increased loss of magnetism at increased temperature all conspire to make the temperature stability of the magnet more critical than for SmCo. Coupled with the addition of masks in the Q1 Chamber and a 23 kW trim, this puts more pressure on the cooling water system design and implementation to stabilize temperatures of both B1 and Q1 2. Beam orbits and separation through the Near IR seems to be satisfactory with the new magnet configurations. However, these designs must be cleared with the Lattice group and run through their MAD simulation. Is beam orbit at out-board magnets a problem? Is the dynamic aperture acceptable? These should be secondary issues, but... 3. New masks in both Q1 Chambers will further add to the complexity of the vacuum system. Currently, the B1 and Q1 chambers are not able to be independently aligned, which may be a problem. Also, the stepped Q1 magnet, and 3/4-length trim add complications to the assembly method. None of these issues should be big problems, but may add to the complexity (read: cost) of components. 4. Expected radiation dose. Although this is being simulated, there are many unknowns associated with dumping, filling, mis-steering, etc. that could (?) irradiate these magnets more than expected. Could they produce a hard failure (as opposed to "just" shortening their lifetime)? 5. Radiation effect on NdFeB. There does not appear to be any directly applicable test already done to help us bound this problem. Of the other experiments, more remains to be understood about the results, and we may need to run our own test. This is clearly the single most significant risk with the NdFeB material. These minutes, and agenda for future meetings, are available on the Web at: http://www.slac.stanford.edu/accel/pepii/near-ir/home.html 7