Optimization of Chemical Etching Process in Niobium Cavities

Superconducting niobium cavities are important components of linear accelerators. Buffered chemical polishing (bcp) on the inner surface of the cavity is a standard procedure to improve its performance. The quality of bcp, however, has not been optimized well in terms of the uniformity of surface smoothness. A finite element computational fluid dynamics (cfd) model was developed to simulate the chemical etching process inside the cavity. The analysis confirmed the observation of other researchers that the sections closer to the axis of the cavity received more etching than other regions. A baffle was used by lanl personnel to direct the flow of the etching fluid toward the walls of the cavity. A new baffle design was tined using optimization techniques. The redesigned baffle significantly improves the performance of the etching process. To verify these results an experimental setup for flow visualization was created. The setup consists of a high speed, high resolution ccd camera. The camera is positioned by a computer-controlled traversing mechanism. A dye injecting arrangement is used for tracking the fluid path. Experimental results are in general agreement with computational findings. INTRODUCTION The nuclear industry provides a significant percentage of the world, as well as of the United States, with electricity. Nuclear power plants produce thousands of tons of spent fuel. Some of this spent fuel can be radioactive for thousands of years. The US DOE is currently exploring the possibility of creating a permanent storage site at Yucca Mountain, Nevada, for spent nuclear fuel. The US Congress has recently authorized exploring an alternative way to deal with spent nuclear fuel: Accelerator Transmutation of Waste (ATW). In this approach, a particle accelerator produces protons that react with a heavy metal target to produce neutrons that hit spent fuel and shorten the life of radioactivity through nuclear reactions. A major component of the system is a linear accelerator (linac) that can accelerate over 100 mA of protons to several GeV [1]. Los Alamos National Laboratory (LANL) is an active participant in developing a high-current superconducting rf (SCRF) linear accelerator. It has three major components: niobium cavities, power couplers, and cryo modules. This paper principally deals with niobium cavities. Niobium cavities have several advantages including significantly small power dissipation compared to copper cavities due to the superconductivity. These cavities are usually made of multiple elliptical cells, Figure 1. They are formed from sheet metal using various techniques such as deep drawing or spinning. The cells then are welded together using electron-beams. Multi-cell units are usually tuned by stretching or squeezing them. 1 Copyright © 2004 by ASME Figure 1. A LANL APT Five-Cell Niobium Cavity Assembled with Helium Vessels The performance of a niobium cavity deteriorates if its surface has stains of chemical products on the surface. Insufficient degreasing, liquid retention in pits, areas with varying metal structure introduced during the manufacturing, or mechanical surface damage also affect the performance. Foreign particles sticking to the cavity surface also cause field emission that degrades the cavity performance. To ensure the success of the niobium cavities, they are chemically polished and then subjected to high pressure rinsing, [2]. Palmieri, [3], stated that chemical etching is the industry standard. The following is a brief overview of some of the research activities in this area. Kneisel [4] compared chemical etching to electro polishing. He presented etching rates for different etching fluids. Ono [5] studied the effects of many factors including chemical polishing on the performance of superconducting cavities. He concluded that chemical etching might not improve the performance after a certain depth (≅100 μm). Kneisel and Palmieri, [6], however, stated that good performance of seamless cavities requires removal of a relatively large amount of material removal. Singer et al. [7] used a combination of electro polishing and chemical etching to improve the surface quality. Aune et al. [8] presented a comprehensive study on developing nine-cell cavity. They applied chemical etching to the surface of the cavity during different stages of manufacturing. The researchers noticed that the surface of the cavity experiences more etching near the iris than near the equator of the cavity. To improve the uniformity of the etching process, researchers at Los Alamos National Laboratory (LANL) proposed inserting a baffle inside the cavity to help direct the etching fluid toward the equator walls of the cavity, Figure 2. This baffle design will be labeled LANL Baffle in the remainder of the paper. The objectives of this paper are, (i) Evaluate the effectiveness of LANL design using a computational fluid dynamics (CFD). (ii) Propose an alternative baffle design if needed (iii) Optimize the baffle design to improve the uniformity of the flow speed on the cavity surface. (iv) Visualize and verify the CFD results using an experimental setup composed of a transparent prototype of the cavity, fabricated optimized baffle, CCD camera and a traversing mechanism for positioning the camera. Figure 2. Current Etching Configuration of Niobium Cavities (LANL Design). BACKGROUND The composition of the etching fluid is presented in Table I. The etching fluid has the following characteristics: Density=1532 kg/m Dynamic Viscosity=0.0221 Ns/m Average inlet velocity is 0.0475 m/s. The etching fluid is actively chilled in a reservoir after it exits the cavity. Temperature is maintained below 15oC. The flow is moving against gravity. The fluid leaves the cavity through holes in the baffle. Table I. Chemical Composition of the Etching Fluid Part (by Volume) Acid Reagent Grade 1 Nitric Acid (HNO3) (69-71%) 1 Hydrofluoric Acid (HF) (48%) 2 Phosphoric Acid (H3PO4) (85%) MODELING The problem is modeled using finite element analysis. The model is axi-symmetric. The inlet condition is described by etching fluid flowing through the cavity. The flow is laminar and has a fully developed laminar velocity profile. Since the etching fluid is a Newtonian fluid and its density is constant at isothermal conditions the Navier-Stokes and the continuity equation characterize the axi-symmetric flow. ( ) v b v p & r v ρ ρ μ = + ∇ + ∇ − 2 (1) 0 = ⋅ ∇ u (2) 2 Copyright © 2004 by ASME where η denotes the dynamic viscosity (kg ms). u Velocity vector (ms), ρ Density of the fluid (kgm), p Pressure (Pa). The applied boundary conditions are: i. 0 ) , .( = v u n at symmetry where, n and t are the normal and tangential vectors respectively; ii. p=0 at exit iii. Velocity at inlet is )) 2 ( , 0 ( 2 max s s v u − = where, vmax is the maximum velocity in axial direction. s is a parameter that varies from 0 to 1 from boundary to the centerline of the cavity. The velocity varies parabolically between the walls and centerline. Simulation results of fluid flow using LANL design are shown in Figure 3. The results showed that the baffle succeeded in directing the flow toward the cavity surface. The flow was however restricted to the neighborhood of the narrow (iris) regions of the cavity with very limited circulation in the wide (equator) regions, which confirms the observations of Aune et al. [8]. The current design also experiences back flow behind the second through fifth cells. There is also a significant increase in velocity at the outlet. Figure 3. Velocity Field for the Current Baffle Design (inset: zoomed–in view of the flow at the exit) PERFORMANCE INDEX Internal boundaries are created close to the inner walls of the cavity, Figure 4.a. As Figure 4.b. shows, each cell is divided into six sections as can be seen in the zoomed-in view in the same figure: • Bottom iris • Bottom straight • Bottom equator • Top equator • Top straight • Top iris Inlet and outlet sections are represented using one boundary each. The velocity is integrated along each section. Average velocity for the LANL design along these internal boundaries is equal to 0.0482 m/s while the standard deviation is 0.2422. A performance index is defined using two quantities. The first quantity describes the average velocity along the internal boundaries of the cavity surface while the second one defines its standard deviation. The objective of the optimization is to maximize the first and minimize the second variable as follows: