Evaluation of low-loss alumina material for high-power RF windows

Conventional RF vacuum windows are made of metalized ceramics, hermetically brazed to a pillbox cavity. High-power windows, operating in ultra-high frequency (UHF) band, require the fabrication of ceramic disks with diameters on the order of 9”. Furthermore, a Titanium Nitride (TiN) multipactor suppression coating must be applied to the ceramic surfaces. The large size and complex internal geometry of these windows create challenges in validating the coating in the fully fabricated assembly. This study evaluates a novel low-loss alumina AO479U, provided by Kyocera, and a reactive sputtering process suitable to deposit a 10-20 nm thick TiN coating on a large diameter window. The paper will report the changes in the TiN coating through chemical cleaning and vacuum braze processes using stylus profilometry, optical microscopy, Scanning Electron Microscopy (SEM), and Rutherford Backscattering Spectroscopy (RBS).


Introduction
Kyocera AO479U is a 99.6% purity alumina with properties developed for low dielectric loss applications, such as high-power radio frequency (RF) windows.Specifically, the loss factor, is about an order of magnitude lower than loss factors of typical ceramics (>1−2•10 -3 ) [1,2].For accelerator facilities that require multi-Megawatt power transmission, a high loss factor can result in significant heat dissipation in the ceramic [3].Induced stress from thermal gradients can lead to cracking and major cascading impacts.Prior work surveyed numerous materials and identified AO479U as the leading candidate [4].
Another common issue for RF windows is charge build up leading to multipactor discharge.To mitigate this, a thin film of Titanium Nitride (TiN) can be applied to the vacuum-side of the ceramic faces [5].TiN supresses multipactoring by reducing the Secondary Electron Yield (SEY) coefficient from approximately 7 to 1.5 [6].For pillbox window geometries in rectangular waveguide or coaxial RF transmission lines, the ceramic face is at least partially obscured when viewed externally.Therefore, line of sight physical vapor deposition (PVD) processes must deposit the coating prior to final assembly.
Fabrication of RF window assemblies is usually achieved by brazing processes in hydrogen or vacuum environments using silver-copper or gold-copper alloys at temperatures > 700° C. Hydrogen brazing improves liquid metal wettability by removing surface oxides but is more costly and presents safety hazards.Vacuum brazing requires more attention to part cleanliness and joint design, but is a more widely available toll service offered by specialized vendors.The motivation of this work was to assess the impact of the fabrication process on a TiN coated, AO479U window.

Methods
Two types of coupons were selected.Washer shaped Kyocera AO479U alumina coupons of outer diameter 38.75 mm (1.53'') and inner diameter 16.95 mm (0.67'') were chosen to represent a miniaturized version of a full scale, coaxial window.These coupons are subsequently referred to as ceramic coupons.Commercially available, 10 mm × 10 mm × 1 mm (0.39''× 0.39''× 0.039'') optical grade sapphire substrates (SPI P/N 501SS10X10-AB) were selected for the compositional similarity, flatness, polished surface finish, and availability.These coupons are subsequently referred to as sapphire coupons.TiN coating thickness measurement is a challenge.At 10-20 nm, the surface roughness of the ceramic coupons, Ra ≈ 250 nm, prevents accurate thickness measurements with mechanical profilometry or ellipsometry techniques.The sapphire coupons with Ra ≤ 0.2 nm do not present these difficulties.
In collaboration with Acree Technologies LLC, Concord California, TiN was deposited via reactive RF magnetron sputtering.In this process, a pure titanium target is bombarded by high energy plasma ions confined by a magnetic field.Ejected target atoms deposit on the substrate positioned below the target.To form TiN, a nitrogen and argon mixture is used as the process gas.The process was designed for compatibility with a 225 mm (9'') ceramic disk.

Experimental Plan
Two identical coupon coating runs were conducted.Coupons were numbered, tracked, and distributed symmetrically to assess the uniformity of the coating across a theoretical 225 mm (9'') window diameter.Coupon layout in the deposition chamber is shown in Fig. 1.Sapphire coupons were masked with Kapton tape to form a step measurable with a stylus profilometer.Sputtering was performed with an approximately -20 V substrate bias, 2 mTorr process partial pressure, and a 0.3 nm/sec deposition rate.Deposition was performed using a 102 mm (4'') diameter target and the substrate platen was rotated during deposition.The distance between target and substrates was approximately 127 mm (5'').
To document the as-coated condition (Fig. 2), sapphire coupons were characterized using stylus profilometry, optical microscopy and Rutherford Backscattering Spectroscopy (RBS).Ceramic coupons were analyzed with optical microscopy and Scanning Electron Microscopy (SEM).
Select coupons were subjected to thermal cycle in a graphite internal vacuum furnace.A vacuum pressure of 9•10 -6 torr or lower was maintained throughout the run.Peak temperatures reached approximately 1000° C for 3 minutes, simulating a 50Au/50Cu alloy brazing profile.After braze process exposure, coupons were characterized with the same techniques used for the as-coated condition

Stylus Profilometry
Thickness measurements of sapphire coupons were per-formed using a Sloan Dektak 3030 ST profilometer.The nine sapphire coupons from Run 1 yielded a 14.7 nm average thickness and a 1.3 nm standard deviation.Run 2 yielded a 15.3 nm average thickness with a standard deviation of 1.9 nm.All values were within the 10-20 nm target range.Profilometer measurements for four sapphire coupons before and after braze cycle exposure showed no significant change in TiN thickness (Table 1).Optical microscopy was performed using a Keyence VHX-5000 digital microscope.Imaging was conducted with mixed coaxial and ring lighting, no polarization filter, and 20-200× magnification.Slight differences in color and reflectivity were noticeable with the naked eye for both coupon types, indicating a change in the coating independent from the type of substrate.Under magnification, the change in coloring was much more pronounced.Post-braze imaging revealed several small black particulates on the surface of both types of coupons, presumably graphite from the substrate and/or furnace (Fig. 3).No cleaning was performed after the braze cycle and before the imaging.Imaging of both coupons was very similar.No parameters needed adjustment and both surfaces appeared to be completely covered, from peaks to valleys.The post-braze condition coupon appeared slightly darker, suggesting a change in coating morphology leading to better surface conductivity.Two foreign object debris (FOD) were identified on the surface of the post-braze coupon (Fig. 5).Both were estimated to be on the order of 30 micron. .Spot EDS analysis of each FOD showed compositional variation, but both were primarily silicon based (Fig. 6).Trace amounts of other elements including potassium, calcium and sodium could indicate a glass as the contamination source.Since coupons were cleaned prior to imaging, these debris were likely on the coupon prior to high temperature exposure, during which they became adhered.This identifies the brazing furnace as a contamination risk for particulates other than graphite.Though FOD raises concerns of electrical breakdown on RF window surfaces during high power operation, there is no established criteria for contamination classification or acceptance.

RBS Analysis
Select sapphire were analyzed with RBS by Infinita Labs, Newark California.RBS is commonly used to analyze thin films by impinging the sample surface with high energy ions.Substrate stopping power is de-pendent on both atomic number, Z, and depth of ion beam penetration into the substrate thickness.Therefore, elastically backscattered ion counts and energies can be used to determine thin film thickness and composition.The data was collected using a 1.9 MeV 4 He + beam, 5 nA beam current, 4° incident angle, and 165° scattering angle.Two different sets of sapphire coupons were used for the as-coated and post-braze analysis.

Conclusion
The impact of vacuum brazing on a TiN multipactor suppression coating for use on RF windows was evaluated.The coating was deposited on Kyocera AO479U alumina and sapphire coupons and analysed with profilometry, optical microscopy, SEM, and RBS before and after exposure to a mock braze cycle.Coating thickness uniformity and repeatability was demonstrated for windows up to 225 mm (9'') in diameter.Despite a change in visual appearance, no significant changes in coating thickness, composition or morphology was noted.How-ever, the brazing process introduced various microscopic FOD that presents contamination concerns.

Figure 1 .
Figure 1.Coupon layout in deposition chamber.Dimensions in millimeters.

4. 3 .Figure 4 .
Figure 4. 1000× magnification of ceramic coupon in as-coated condition (a) and after braze cycle exposure (b).Imaging of both coupons was very similar.No parameters needed adjustment and both surfaces appeared to be completely covered, from peaks to valleys.The post-braze condition coupon appeared slightly darker, suggesting a change in coating morphology leading to better surface conductivity.Two foreign object debris (FOD) were identified on the surface of the post-braze coupon (Fig.5).Both were estimated to be on the order of 30 micron.

Figure 5 .
Figure 5. 500× SEM image of FOD on post-braze cycle exposure ceramic coupon.. Spot EDS analysis of each FOD showed compositional variation, but both were primarily silicon based (Fig.6).Trace amounts of other elements including potassium, calcium and sodium could indicate a glass as the contamination source.Since coupons were cleaned prior to imaging, these debris were likely on the coupon prior to high temperature exposure, during which they became

Table 1 .
Profilometer Thickness Data for Sapphire Coupons before and after Braze Cycle Exposure

Table 2 .
RBS Data for Sapphire Coupons RBS analysis revealed no significant changes in thickness and composition (Tab.2).Thickness values were consistently ~40% lower than what was measured with the stylus profilometer.Since RBS estimates thickness assuming a close packed crystal structure, this difference suggests the sputtered deposit is not fully dense.Both sets of measurements indicate inclusion of 2-6 at.% Cr or Si in the coating.The purity decreased modestly, but consistently for post-braze coupons. 6