Appl. Phys.

A30, 185-187 (1983)

Applied

so,,,,

Physics A Surfaces ""

9 Springer-Verlag 1983

Improvement of the Polishing Treatment for Niobium Surfaces of Superconducting Cavity Resonators
Y. Uzel, K. Schnitzke, and N. Krause Siemens AG, Research Laboratories, D-8520 Erlangen, Fed. Rep. Germany Received20 October 1982/Accepted 3 December 1982

Abstract. For superconducting cavities lrmde from niobium sheet, which are interesting in the field of accelerator applications, smooth and defect free inner surfaces are needed to achieve a high unloaded Q along with high accelerating field strength. These can be obtained using chemical polishing procedures. Whereas normally a mixture of HF, HNO3, and H3PO 4 is applied, we chose the known bath containing HF, HNO3, and H2SO 4 and worked above 50~ The surface quality was judged by visual inspection using a light microscope as well as by microwave measurements of one spherical resonator. The new method gave less grain boundary etching and, with high field levels, an enlarged unloaded

Q.
PACS: 85.25, 82.65

Superconducting cavity-resonators made of niobium and excited in the TM01o-mode are used especially for particle accelerators (e.g. [1,2]). Their main advantage compared to normal conducting types is the distinctly increased unloaded Q, resulting in lowered rf-power consumption. The very small power loss enables, for example, continuous accelerator operation. A prerequisite for high Q ' s - a n d moreover large accelerating f i e l d s - is a fairly smooth, clean and defect free niobium surface. In general, however, the inner surface of the resonator is severely damaged by machining. So a removal of this damage layer is needed and may be obtained using chemical or electrochemical polishing processes. Our aim is to find a procedure which leads to microwave properties as good as possible. One such process is known as electropolishing [3] and gives excellent results particulary with annealed material. Unfortunately the process works very slowly, dissolving the niobium at a rate of about 0.1 gm/ min. in order to avoid this time-consuming procedure alternative polishing processes have to be used. Regarding the frequency range of accelerator applications, which extends from about 300 MHz to some

3 GHz, resonator dimensions require manufacturing techniques based on the use of niobium sheets. The sheets are shaped by spinning or deep drawing and assembled by electron beam or argon arc welding. Therefore the strongly cold-worked condiuon of the material had to be taken into account.

Polishing Processes for Cold-Worked Niobium

In principle, removal of the damage layer is achieved by immersing the resonator into a mixture of both fluoric and nitric-acid (HF+HNO3) for a certain time. The thickness of the layer dissolved is determined by both the time of immersion (~) and the concentration of the acids. Using a mixture consisting of 1 part HF (40 %) and 1 part HNO 3 (65 %) and dissolving about 30 gin, the process unfortunately results in strong etching of the grain boundaries as well as in rough surfaces within the grains. An improved surface quality may be achieved with increased concentrations of the acids, e.g. HF (50%)+HNO 3 (100%), but this leads to very short immersing times in the range of 3...10s. To overcome this problem, addition of phosphoric acid (H3PO4) as a third component to the basic

186

Y. Uzel et al.

Fig. 1. Photograph of the surface of a niobium test sheet, polished by using H F (48%)+HNO3 (100%)+H3PO 4 (85%) (Mixture 1), volume ratio 1:1:1.0~t =37 "C z = 120 s, dissolved layer 35 p.m

mixture is .widely used. Other additions are also possible. We investigated particularly the use of sulphuric acid (H2SO4). A comparison was made between the H3PO 4 (Mixture I) and the H2SO 4 addition (Mixture II). The quality of the remaining surface after the polishing process was judged by a visual inspection of niobium test sheets using an optical microscope as well as by microwave measurements of the resonators. Polishing niobium with Mixture I was described by Kinter et al. [4]. The bath is composed of HF (48 %) + H N O 3 (100%)+H3PO~ (85 %) in a volume ratio of t : l : l . Compared to the basic mixture, addition of H3PO 4 reduces the rate of niobium removal and therefore makes the process more controllable. Mixture II was, in principle, described by Miller [-51 and was improved by our investigations concerning bath temperature and concentration of the acids. Best results we obtained using a polishing mixture of HF (40%), HNO 3 (65 %), and H2SO 4 (96...98 %), volume ratio 1:1:2. Bath temperature turned out to be a critical parameter. A starting temperature of about 80 ~ seems to be a good choice with regard to the polishing effect. Lower temperatures reduce the rate of removal but also the surface smoothness.
C o m p a r i s o n of Results by Surface Inspection

Fig. 2. Photograph of the surface of a niobium test sheet, polished by using H F (40%)+HNO3 (65%)+H2SO4 (96...98%) (Mixture ii), volume ratio 1:1:2. 0~t= 78 ~ ~=30 s, dissolved layer 35 gm
101(

J \ ~ -'~ , \~

,
o~,.D

I l

16

HF+HNO3+H3Pfl 4

"~ 109

Several test sheets of cold-worked niobium were polished using the two mixtures respectively. In each case about 30 gm of the original surface was removed, immersing the sheets into the bath for different times "c. Figure 1 gives an impression of the use of Mixture I (H3PO4). There is a good smoothing effect within the grains but grain boundary etching remains. The mixture was prepared following [4] and the process took about z = 120 s starting from a bath temperature of 0st=37 ~ Figure 2 shows the typical effect of adding HzSO 4 (Mixture II). Surface roughness within the grains is comparable to Fig. 1 but grain-boundary etching is drastically reduced. From this point of view therefore the latter method appears preferable. A certain disadvantage is the need for higher temperatures (0st =78~ and, relatively, short immersing times (~ ~ 30 s). To check the evidence of the optical estimation additional microwave measurements of one completely assembled resonator were carried out.
C o m p a r i s o n of Results by M i c r o w a v e M e a s u r e m e n t s

I
0

I
20

I
40

I
60

I
80

I
mT 100

magnetic flux-density/pc

Fig. 3. Measured unloaded Q vs. magnetic flux density B~c of a spherical Nb sheet resonator, polished using Mixture I or Mixture II in alternation

A spherical TMol 0 resonator designed for a resonance frequency of 9.5 GHz was manufactured using spin-

Polishing Treatment for Niobium Surfaces ning and electron-beam welding of 1 m m niobium sheet. Prior to the first measurement the resonator was polished in the H 3 P O 4 buffered mixture, dissolving about 30 gm of the original surface. This chemical polishing was followed by an oxipolishing treatment [63 and the resonator then attached to the microwave equipment. After cooling down to 1.5 K power was fed to the resonator and the unloaded Q as well as the corresponding microwave flux density B a~ were measured for increasing power levels until the critical flux density B~~ was reached and breakdown occurred. The resonator was warmed up, disconnected and, starting from the chemical polishing treatment, the whole procedure was repeated changing only to Mixture II (H2804). In this way a series of 6 measurements was taken, alternating the chemical polishing mixtures. In Fig. 3 the unloaded Q is shown versus the flux density B ac. The curves are numbered in the sequence of the single measurements. In all cases the well known decrease in Q with increasing magnetic flux density is observed. Nevertheless, the curves are arranged in two groups clearly distinguished by the gradient of the decrease. All measurements following the polishing with addition of H 3 P O 4 belong to the steeper gradient group, verifying the adverse influence of the stronger grain boundary etching (Fig. 1). This may be understood if one takes into account the field enhancement at peaks and sharp edges of the surfaces. The power loss related to these regions exceeds that of the surrounding smooth area giving rise to local surface temperature spots. These spots tend to grow because of their enlarged stirface resistance,

187 thereby causing an overproportional increase in the average resistance. An interesting detail not yet understood very well is connected with the oxipolishing procedure. If oxipolishing, e.g. anodizing and dissolving the oxide layer in a H F bath, is done only once using a voltage difference of 30 V ( ~ 30 n m oxide layer, Curves 5 and 6) the highest low-field Q's are reached as well as the ~steepest decrease in Q. On the other hand, oxipolishing twice (or more) with 100V (2 x l 0 0 n m oxide layer) yields a slower Q decrease.

Conclusion
We have shown that chemical polishing of niobium using a mixture of H F + H N O 3 + H z S O 4 leads to surfaces with better microwave properties than those obtained with a mixture of H F + H N O 3 + H a P O ~. This can also be seen by optical inspection.
Acknowled~emem. This work has been supported by the technologi-

cal program of the Federal Department of Research and Technology of the FRG.

References
1. Proc. Workshop on rf Supercondictivity, Karlsruhe, July 2-4, 1980 2. U. Klein,D. Proch, T. Grundey: IEEE Trans. NS-28, 3222-3224 (1981) 3. H. Diepers, O. Schmidt, H. Martens, F.S. Sun: Phys. Lett. 37A, 139-140 (1971) 4. M.L. Kinter, 1. Weissmann, W.W. Stein: J. AppI. Phys. 41, 828 (1970) 5. G.L. Miller: Tantalum and Niobium (Butterworths, London 1959) pp. 328-329 6. H. Martens, H. Diepers, B. Hillenbrand: Phys. Lett. 44A, 213214, (1973)