In-situ coating technology

1. In-situ coating technology A. Hershcovitch, M. Blaskiewicz, J.M. Brennan, W. Fischer, C-J Liaw, W. Meng, R. Todd Brookhaven National Laboratory, Upton, New York 11973, U.S.A A. Custer, M. Erickson, N. Jamshidi, R. Laping, H. J. Poole PVI, Oxnard, California 93031, USA J. M. Jimenez, A. Kuzucan, M. Taborelli, and C. Yin-Vallgren CERN, CH-1211 Genève 23, Switzerland N. Sochugov High Current Electronics Institute, Tomsk, 634055 Russia

3. Problems that must be solved in order to enhance RHIC luminosity Lower vacuum chamber resistivity and control electron cloud formation! The RHIC vacuum chamber is made of 7.1 cm ID stainless steel tubing with 1.6 m surface roughness in cold bore (2.1m in warm section). Stainless steel has high resistivity. Coat tube with copper!Electron clouds, which have been observed in many accelerators including RHIC can act to limit machine performance through dynamical instabilities and/or associated vacuum pressure increases. Formation of electron clouds is a result of electrons bouncing back and forth between surfaces, which cause emission of secondary electrons (electron multipacting effect). a-C coating!

4. Possible Solutions • Coating the stainless steel walls with about 10 micrometers of oxygen free high conductivity copper (OFHC) can prevent problems arising from high resistivity. • Mike Blaskiewicz made rigorous computations (in response to claim by Sergio Calatroni from CERN that due to anomalous skin effect a coating of 50 micrometer of copper is needed), which are appropriate for our case. Essence of these computation is as follows: he combined effects of low temperature and large magnetic fields will yield a net reduction in room temperature resistivity of RRR=50 in the copper coating. The mean free path of conduction electrons is 2 microns, which is equal to the skin depth at 20 MHz. It is therefore prudent to include the anomalous skin effect when calculating the effect of the coating. When this is done it is found that 10 micrometers of copper should be acceptable for even the most extreme future scenarios. Basically, those calculations that indicate that 10 micrometer of copper will suffice. Complete details can be found at the following BNL report: M. Blaskiewicz, “Copper Coating Specification for the RHIC Arcs”, C-A/AP/ note #413 Dec 2010 (unpublished). • Covering the OFHC with a very thin layer of amorphous graphite to reduce SEY (secondary electron yield) • But, Cu coatings thicker than 2 μm have columnar and other grain structures that might have lower SEY like gold black.

5. Coating Long Cold Stainless Steel Tube With OFHC & Other Materials Original project goal, by PVI of Oxnard, CA, in collaboration with BNL, was the development of a moving mole-like plasma deposition sputter device of Cu and a-C for application in 500 meter long sections of RHIC. But, the project evolved to encompass more R&D than anticipated due to some unforeseen challenges and opportunities.

6. DEPOSITION PROCESSES AND OPTIONS • Coating methods can be divided into two major categories: chemical vapor deposition (CVD) and physical vapor deposition (PVD). Due to the nature of the RHIC configuration, only PVD is viable for in-situ coating of the RHIC vacuum pipes. First, the temperature under which coating can be made cannot be high (400 C is required for some conventional CVD), since the RHIC vacuum tubes are in contact with superconducting magnets, which would be damaged. A second very severe constraint is the long distance between access points. Introduction of vapor from access points that are 500 meters apart into tubes with 7.1 centimeters ID would likely not propagate far and result in extremely non-uniform coating. • But these constraints also severely restrict PVD options. Obviously evaporation techniques (ovens, e-beams) cannot be used in 7.1 centimeters ID, 500-meter long tubes. Therefore, evaporation must be accomplished locally. One option is a plasma device on a mole that generates and deposits the vapor locally. • Presently, there are a variety of PVD methods used to deposit coatings on various substrates. By definition, physical vapor deposition entails purely physical processes of evaporating materials. The vapor then condenses on the desired substrate. There is a wide variety of vapor generation techniques ranging from high temperature evaporation to sputter bombardment by electron beams, ion beams and plasma. The latter involves a discharge like RF, glow, or an arc. The long distance between access points and the need to have a mole like deposition device precludes the use of RF plasmas.

7. MAGNETRON DEPOSITION STATE-OF-THE-ART • Of the plasma deposition devices like magnetrons, diodes, triodes, cathodic arcs, etc., magnetrons are the most commonly used plasma deposition devices. In magnetrons, magnetic fields are utilized to confine electrons that generate high density plasma (usually argon or xenon) near the surface of the material that is being sputtered. Major advantages of magnetron sputtering sources are that they are versatile, long-lived, high-rate, large-area, low-temperature vaporization sources that operate at relatively low gas pressure and offer reasonably high sputtering rates as compared to most other sputtering sources. Because of these superior characteristics magnetron sputtering is the most widely used PVD coating technique. Although arc discharges operate with higher intensity, they require the use of special filters to eliminate macroparticles that reduce the net deposition rate to those of magnetrons. • Typical coating rates by magnetrons (w/argon gas) are 5 Å/sec for a power of 50 W/inch2 on the magnetron cathode, though with intense cooling cathode power of 100 W/inch2 is achievable.

8. PLANNED DEPOSITION TECHNIQUE • Originally, the ultimate objective is to develop a plasma deposition device for in-situ coating of long, small diameter tubes with about 5 μm of Cu following by a coating of about 0.1 μm of a-C. The figure below is a scheme of a plasma deposition technique based on staged magnetrons. Plasma deposition sections consist of two, connected through an insulator, cylindrical magnetron devices. The first magnetron stage has oxygen free high conductivity copper cathode, while the second stage has a graphite cathode. Internal ring permanent magnets form the magnetic field. Magnetron assembly is to be mounted on a carriage (mole), which is to be pulled by a cable assembly driven by an external motor. To accommodate for any diameter variances, including bellow crossing, the carriage will have a spring-loadedguidewheelassembly.

9. Spool drive mechanism is shown in the below figures. A dragline, attached to end (opposite to the carriage) of the graphite cathode, is used to initially pull the magnetron assembly and cable bundle to the end, where coating begins. The dragline, (also motor driven), is a strong thin cable made of either high-tensile fishing line, or Teflon sleeved Inconel or equivalent. Should there be evidence that either the Teflon or the fishing line leave any residue, a pure metal line is to be used. During coating, the magnetron assembly and cable bundle are pulling the dragline (in a direction opposite to, which the dragline pulled on the magnetron assembly and cable bundle). If needed, a brushless DC servo-motor driving 4 rows of internal wheels moves the carriage, which has position feedback, assists carriage motion. Cable for pulling mole identified: ~1/4" diameter stranded SS with a Teflon sheath. This type of cable is typically used in aircraft for flexible linkage with the various airfoil surfaces (rudder, flaps...etc.). It is very strong (20K tensile) with low elongation.

10. At a Cu coating rate of 5 Å/sec, it would take 2.78 hours to deposit 5 μm of Cu, i.e., close to 3 hours to move one cathode length. With a 2 meter long cathode it would take 695 hours (or 29 days; a fraction of a typical RHIC shutdown period) to coat 500 m. And 2 m Cu cathode would not need reloading (2.1 meter long cylindrical magnetron cathodes exist in commercial systems). But magnetron weight and cooling capability would limit single deposition device length to about 50 cm. • Possible solution, if mole support wheels can be employed in areas, which were already copper coated, multiple magnetrons in a train like assembly, having a total exposed cathode length of 2+ meters, as shown below

11. Coating 500 meter section in-situ requires some RHIC bellows replacement with access bellows for copper cathode reloading

12. Other Options Being Considered • Presently replacement cathodes utilizing access bellows is not the leading option. • Taking few magnets out and coating a series of magnets at a time is being considered. • No decision final has been made

13. PREVIOUS EXPERIMENTAL RESULTS A mobile magnetron, shown below, with a 15 cm long cathode was designed, fabricated, and tested to coat 30 cm long samples of RHIC cold bore tubes with up to 6.1 μm with OFHC at an average coating rate of 30 Å/sec (factor 6 higher than absolutely needed). Copper deposition rates were measured with a 6 MHz crystal rate monitor.

14. Photo of copper coated RHIC stainless steel tube sample

15. Magnetron Operation & Coating Problems Encountered • Discharge ignition difficulties (operating on LHS of Paschen curve); pipe in big box • Edge effects in discharge intensity and coating rates • First coatings with DC power had poor adhesion. On occasion had success with AC at 40 kHz (square wave) coating. Inconsistent, adhesion did not always meet rigorous industrial standard (tape; nail). • Very poor copper utilization (magnetic field shape and magnet variation; 2 were 800+ Gauss others 500-600 Gauss). • Copper thickness 0.100” • At strongest magnets after operation. • Waist thickness 0.030” • Requires copper replacement • Magnetron integrity

16. Deposition Chamber (closed)

17. Deposition Chamber (open)

18. Photo of argon plasma between magnetron surface and tube (left); power, cooling, and instrumentation feed visible, plasma discharge and deposition are azimuthally uniform. But axially discharge is non-uniform causing very uneven erosion and poor copper cathode utilization. Not issue in long tube; identical magnets

19. Adhesion Pre-coating: Nickel (top industrial choice) is magnetic. Chrome (hard to sputter) Titanium (last choice). After some effort, got bi-metal magnetron to work. Copper to titanium adhesion excellent. But, Titanium to stainless steel poor.

20. Solution: Discharge Cleaning First step: apply a positive voltage (of about 1 kV) to the magnetron or a separate cleaning anode and to move the discharge down the tube with the pressure near 2 Torr. So far it worked well with existing magnetron (for long tube cleaning there concern of discharge cleaning debris affecting the copper cathode). Second step: is the conventional deposition step at a pressure of about 5 mTorr. First with Ti pre-coating; later with direct copper coating. No need for pre-coating!

21. RF Resistivity • First, without discharge cleaning, three 30 cm long RHIC SS tube samples were coated with 2.5 μm to 6.1 μm OFHC at 20 mTorr with AC power. Coating is axially non-uniform (thicker at edges; 6.1 μm had poor adhesion). Coating is matte in appearance (low SEY?). • Room temperature resistivity of one of the coated samples 2.5 μm or about 4.5 to 5 μm, was indistinguishable from copper at 180 MHz with coating that’s far from ideal.

22. A folded quarter wave resonator structure was built to measure the resistivity of coatings. The resistivity of samples (RED) was deduced from the Q value of the eight resonate modes between 180 and 2000 MHz. Microwave Studio separated the contributions to the Q from the samples and pure copper components ( Yellow) of the resonator. Calculated spectrum of the resonate modes of the test fixture

23. Test Fixture with two coated samples and pure copper reference sample

24. The input coupler was small and adjustable to provide weak coupling even at the top frequencies

25. Network Analyzer measurements agree with MicroWave Studio calculations Transmission coefficient from 10 MHz to 2200 MHz Q (Bandwidth) measurement at 390 MHz

26. Measured and computed Q values for stainless, coated, and copper samples

27. Coating 49cm long tubes (2 μm below); cut out the center 32cm for more RF testing

28. 49 cm tube, center cut 32cm; tack welded the ends that can be removed later.

29. Secondary Electron Yield (SEY) Three 29 mm diameter stainless steel discs and three 15x20 mm rectangular samples were Cu coated with thicknesses of 2 μm, 5 μm, and 10 μm. SEY of the rectangular samples were measured at room temperature; and SEY of the discs are to be measured at cryogenic temperatures. Measured samples were shiny in appearance (more of a crystalline structure?). Nevertheless, those exhibited lower SEY.

30. CERN SEY January Measurements Job by D.Letant-DelrieuxDate: 2012-01-17Number of samples: 3 Sample codification: 2606 2606-a: rectangle, 2 microns CERN, 11/29/11; 2606-b: rectangle, 5 microns CERN 2606-a: rectangle, 10 microns CERN, 12/20/11 Nature of samples: RHIC samples: copper coatings on stainless steel samples. Sample storage: in aluminium foils in plastic bags Aim of analysis: SEY measurements and checked the surface cleanliness by XPS. Results The secondary-electron yield (SEY) versus primary electron (PE) energy curves for the as-received samples are shown in figure 1. The higher SEY is δmax=1.78 and is observed for the 2 microns coating (sample 2606-a) at Emax=332eV. The maximum SEY value for the 5 microns (sample 2606-b) and 10 microns (sample 2606-c) coatings is δmax=1.65 and is observed at Emax=332eV.

31. C, O and Cu are detected on the 3 samples.Traces of Na are observed on sample 2606-a. The carbon contamination is ~30 at.% for samples 2606-a and 2606-b. Sample 2606-c is cleaner (C~18 at.%).

32. Attempting to Obtain Matte Coating Deposition modes: high or low pressure [20 to 40 mTorr or 5 mTorr or less (conventional DC industrial operation)], DC or AC power. Only high-pressure AC deposition results in matte, but not always. Based on visual observation. Awaiting SEY measurement of 2 μm, 5 μm, and 10 μm samples, in which only the top layers were deposited AC at 35 mTorr after DC operation at 5 mTorr.

33. Copper Coated Stainless Steel Samples

34. CERN SEY May Measurements Number of samples: 3 Sample codification: 2670 2670-a: rectangle, 2 microns CERN, 6/4/2012 2670-b: rectangle, 5 microns CERN, 9/4/2012 2670-c: rectangle, 10 microns CERN, 10/4/2012 Nature of samples: RHIC samples: copper coatings on stainless steel samples with different roughness. Sample storage: in aluminium foils in plastic bags Aim of analysis: SEY measurements and checked the surface cleanliness by XPS. Results The secondary-electron yield (SEY) versus primary electron (PE) energy curves for the as-received samples are shown in figure 1. The higher SEY is δmax=1.86 and is observed for the 10 microns coating (sample 2670-c) at Emax=382eV. The 5 microns sample (sample 2670-b) shows an δmax=1.84 . The lower SEY is observed for the 2 microns coating (sample 2670-a) and is δmax=1.79 (Emax=382eV). The resulting concentrations calculated from the XPS spectra are given in figure 2. C, O and Cu are detected on the 3 samples. The carbon contamination is ~20 at.% for the 3 samples. The binding energy of the Cu2p3/2 line indicates the Cu2O contribution and a further shoulder at higher binding energy from Cu+II, which can be ascribed either to Cu(OH)2 or CuO.

35. C & O observed on all samples; C contamination 20% in all. Highest SEY is δmax=1.86 for 10 microns coating at Emax=382eV. 5 microns sample had δmax=1.84. Lowest SEY is observed for the 2 microns coating δmax=1.79 (Emax=382eV).

36. Confused about SEY results; for 10µmdifference in SEY between samples is 0.21 We do not plan to measure all the samples at cryogenic temperatures, since the temperature does not affect geometry or roughness. You will soon get the data for the measurement of the sample which had the lowest SEY at room T. In fact the difference between the samples is only marginal. Keep in mind that he accuracy we estimate is +/-0.03 on the SEY, so basically the differences we see there at room T are close to detection limit. I do not know how your samples look like at microscopic level, but in order to modify the SEY you need high aspect ratio roughness. Just watching at the samples I would guess they do not show sufficient difference to justify a different SEY. So for me no surprise. Best regards Mauro

37. New Test Stand Comprising of full-size dipole vacuum tube with removable testing middle section, two types of bellows, differential pumping for magnetron insertion.

38. Agenda, Experiments in New Test Stand Axially uniformly 30 cm long SS tubes Cu coated 2 μm, 5 μm, & 10 μm thick (in dep. box). Design, fabricate, test magnetron with 50 cm long copper cathode (cooling, weight limits). Discharge cleaning in confined tube: debris, oxidation, or hydrocarbons (Ar-O2 or Ar-H2 glow discharge). Increasing cathode lifetime: thicker copper (x2), stronger uniform magnets, movable magnet package (pneumatic or utilize wheel rotation). Tesla coil/beta emitter (Ni-63) initiate/maintain discharge.

39. Magnetron with Moving Magnets

40. Status, Challenges and Opportunities • Coating method w/good adhesion developed. • Resistivity of coated RHIC tube samples close to copper; SEY lower than copper. • Cable for pulling mole identified. • Target to substrate distance 3 cm. Commercial coating equipment 10’s cm; 6.3 cm lowest experimental; unique omni-directional coating. • Encouraging results; but, there are still more questions to be answered.

41. Acknowledgement Work Supported by DOE under Contract No. DE-AC02-98CH1-886 Notice: This manuscript has been authored by Brookhaven Science Associates, LLC under Contract No. DE-AC02-98CH1-886 with the US Department of Energy. The U.S. Government retains, and the publisher, by accepting the article for publication, acknowledges, a world-wide license to publish or reproduce the published form of this manuscript, or others to do so, for the United States Government purposes.