Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://danp.sinp.msu.ru/Articles_GSIB/nimb_GCIB_Ar_Smoothing_Swenson.pdf Äàòà èçìåíåíèÿ: Fri Oct 7 20:32:30 2005 Äàòà èíäåêñèðîâàíèÿ: Mon Oct 1 22:39:16 2012 Êîäèðîâêà:

ARTICLE IN PRESS

Nuclear Instruments and Methods in Physics Research B xxx (2005) xxx­xxx www.elsevier.com/locate/nimb

Smoothing RF cavities with gas cluster ions to mitigate high voltage breakdown
D.R. Swenson
b c

a,*

, E. Degenkolb a, Z. Insepov b, L. Laurent c, G. Scheitrum

c

a Epion Corporation, 37 Manning Road, Billerica, MA 01821, USA Argonne National Laboratory, 9700 South Cass Ave., Argonne, IL 60439, USA Stanford Linear Accelerator Center, 2575 Sand Hill Road, Menlo Park, CA 94025, USA

Available online

Abstract Many studies have demonstrated that improving the surface smoothness and cleanliness of high voltage electrodes increases the voltage standoff capability, but none have specifically investigated the role of nano-scale and atomic level surface roughness. Using AFM imaging, we have studied the effect of gas cluster ion beams (GCIB) on oxygen-free Cu electrode material that is used in high gradient RF cavities. Using Ar clusters accelerated by 30 kV, with a dose of ° ° 6 · 1014 e cmþ2, we have effectively removed an asperity that was 3500 A wide and 350 A high. Subsequent processing ° ° with 5 kV acceleration reduced the surface roughness from an Ra value of 13.2 A to 4.8 A. This demonstrates the effectiveness of GCIB for reducing sub-micron roughness to atom level smoothness. ñ 2005 Elsevier B.V. All rights reserved.
PACS: 36.40.Wa; 39.90.+d; 36.40.Qv; 34.90.+q Keywords: Cluster­surface interactions; Kilpatrick limit; Cluster beam; RF breakdown

1. Introduction The electrical breakdown of a high voltage electrode in vacuum (Kilpatrick limit) is a fundamental constraint in the design of accelerators and of the other high voltage equipment. It often deterCorresponding author. Tel.: +1 978 215 6305; fax: +1 978 670 9119. E-mail address: dswenson@epion.com (D.R. Swenson).
*

mines economically important parameters such as the length of linear accelerators, the maximum energies of beams, the maximum RF power of waveguides and numerous other specifications. While it is widely recognized that surface smoothness and cleanliness are critical in achieving high breakdown voltages, and many technologies have been used to improve electrode finish, typical high voltage breakdown limits remain orders of magnitude less than the optimum predicted from

0168-583X/$ - see front matter ñ 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.nimb.2005.07.107


ARTICLE IN PRESS

2

D.R. Swenson et al. / Nucl. Instr. and Meth. in Phys. Res. B xxx (2005) xxx­xxx

Fowler­Nordheim field emission theory. It is widely speculated that nano-scale roughness may now be the limiting factor. Gas cluster ion beams (GCIB) is a new nano-smoothing technology that has several interesting features that make it applicable to high gradient technology. GCIB smoothing can be used on non-planar surfaces and is applied under high vacuum conditions. With GCIB the surface is bombarded by clusters of atoms typically 10,000 atoms in size with a charge of +3 and a velocity of 6 km/s [1]. At this velocity, the cluster impacts create shallow craters with ° diameters on the order of 100 A. With typical doses, every point on the surface is affected many times resulting in effective smoothing. GCIB processing has been successfully applied in several industries such as optical thin films, read/write heads for fixed disk memory systems, EUV mirrors and semiconductor fabrication. In these applications it has proven capable of producing atomic level smoothness. In this paper we investigate the smoothing capability of GCIB for OFE Cu electrode material that is typically found in high gradient RF accelerators. More specifically we investigate how large of a feature GCIB can practically remove from a surface.

experiment). A dipole magnet removes monomers and dimers from the beam. An electron flood provides neutralization of the space charge of the beam. The sample was mechanically scanned through the beam to provide uniform irradiation of the substrate. The Cu samples were prepared with typical machining, cleaning and polishing techniques used for high gradient X-band RF linac cavities but were handled in air when loading into the GCIB machine and when making the measurements. The sample was analyzed using atomic force microscopy (AFM) imaging. Fiducial marks were used to carefully align the sample and to find features on the surface. Using the optical microscopy capability of our AFM instrument, we were able to repeatedly locate and image a 10 · 10 lm area of ° the sample that contained a 3500 A diameter by ° 350 A high asperity. Using the coordinates determined from the 10 · 10 lm AFM images, we then were able to make detailed 2 · 2 lm images of the asperity. This technique makes it possible to see in detail the effects of GCIB processing on such a large asperity as well as the effect on the smoother region surrounding it. The GCIB processing was performed in sequential steps with the sample removed from the GCIB machine and imaged at each step.

2. Technique and results Fig. 1 shows schematically the GCIB beamline [2]. Clusters are formed as high pressure Ar expands supersonically into high vacuum inside a nozzle. The resulting jet of clusters is ionized by electron impacts and accelerated electrostatically (using either 30 kV or 5 kV potential for this 3. Results Figs. 2­5 are the AFM images of the asperity during the processing. The images are linear gray scale plots of the heights on the surface, with lighter shades denoting higher elevations. In Fig. 2, be-

Fig. 1. Schematic of the GCIB beamline.


ARTICLE IN PRESS

D.R. Swenson et al. / Nucl. Instr. and Meth. in Phys. Res. B xxx (2005) xxx­xxx

3

Fig. 2. Asperity on Cu electrode material before processing. ° The Z range = 471 A and the average roughness inside the ° black box is 10.0 A.

Fig. dose The 13.2

4. The same asperity after processing with an additional of 3 · 1014 ecmþ2 of Ar clusters accelerated with 30 kV. ° Z range = 272 A and the roughness inside the black box is ° A.

Fig. 3. The same asperity after a dose of 3 · 1014 ecmþ2 of Ar ° clusters accelerated with 30 kV. The Z range = 341 A and the ° average roughness inside the black box is 12.9 A.

° fore the processing, the 350 A high asperity can be seen as well as several smaller spike features. The Z range in the image (highest point­lowest point) and the Ra or average roughness (average deviation from the mean) are given in the caption. Fig. 3 shows that even after a very small dose of

Fig. 5. The same asperity after processing with an additional dose of 1 · 1015 e cmþ2 of Ar clusters accelerated with 5 kV. ° The Z range = 177 A and the roughness inside the black box is ° 4.9 A.

high energy GCIB, the smaller spikes are removed and the Z range is greatly reduced. The processing however roughens the smoother surrounding area (as measured inside the box). This roughening is


ARTICLE IN PRESS

4

D.R. Swenson et al. / Nucl. Instr. and Meth. in Phys. Res. B xxx (2005) xxx­xxx

Fig. 6. Summary of the results of GCIB processing of the asperity on Cu electrode material.

Fig. 7. Plot of relative etching obtained by pixel-by-pixel subtraction of Fig. 5 from Fig. 2.

caused by the large size of the high energy GCIB craters [3­8]. Fig. 4 shows the results of a second equal dose of high energy GCIB. The asperity is much smaller but the surrounding area is roughened more. Fig. 5 shows the results of processing with a larger dose of similar clusters but accelerated with only 5 kV. Here the GCIB craters are more shallow and the area in the box is smoothed ° ° from an Ra value of 13.2 A down to 4.9 A; which is much smoother than the initial roughness of ° 10.0 A. Fig. 6 summarizes the results of the processing and shows that a combination of high energy and low energy GCIB is effective at removing both the large asperity and smoothing the overall surface roughness.

two other asperities and further experimental and theoretical work is in progress. The equipment can easily provide beams of 200 lA and 100 lA using, respectively, 30 kV and 5 kV acceleration voltages, so it is very practical to process large areas with the doses we have studied. High voltage stress tests and Fowler­Nordheim field emission tests will be used to evaluate the effect of GCIB processing on the Kilpatrick limit.

Acknowledgements We acknowledge helpful discussions and assistance from J. Hautala, T. Tetreault, R. MacCrimmon, M. Tabat and the support of Epionós management A. Kirkpatrick, B. Libby and M.E. Mack

4. Conclusions These data show that GCIB can remove a wide range of surface roughness up to features on the ° order of 400 A. With larger doses even larger features might be smoothed. Fig. 7 is a pixel-by-pixel comparison of the initial and final images that shows that material is removed from high spots with some of the material redeposited in depressions. These data show that the etch rate is greatly accelerated for isolated asperities on a Cu surface. At this point it is not known whether this is a result of the mechanical weakness of such a structure, its thermal isolation, or whether the asperity was made of a different, more easily etched material. Similar results were obtained for

References
[1] D.R. Swenson, Nucl. Instr. and Meth. B 222 (2004) 61. [2] M.E. Mack, et al., in: Proceedings of the 14th International Conference on Ion Implantation Technology, Taos, New Mexico, 2002 (IEEE, 2003). [3] L.P. Allen et al., J. Appl. Phys. 92 (2002) 367. [4] I. Yamada, J. Matsuo, N. Toyoda, A. Kirkpatrick, Mater. Sci. Eng. R 34 (2001). [5] O. Rattunde et al., J. Appl. Phys. 90 (2001) 3225. [6] N. Toyoda, N. Hagiwara, J. Matsuo, I. Yamada, Nucl. Instr. and Meth. B 151 (2000) 980. [7] W.K. Chu et al., Appl. Phys. Lett. 72 (1998) 246. [8] Z. Insepov, I. Yamada, Mater. Sci. Eng. A 217 (1996) 89.