Publications

A Science-Based Understanding of Cermet Processing

Roach, R.A.; Kilgo, Alice C.; Susan, D.F.; Van Ornum, David J.

This report is a summary of the work completed in FY01 for science-based characterization of the processes used to fabricate 1) cermet vias in source feedthrus using slurry and paste-filling techniques and 2) cermet powder for dry pressing. Common defects found in cermet vias were characterized based on the ability of subsequent processing techniques (isopressing and firing) to remove the defects. Non-aqueous spray drying and mist granulation techniques were explored as alternative methods of creating CND50, the powder commonly used for dry pressed parts. Compaction and flow characteristics of these techniques were analyzed and compared to standard dry-ball-milled CND50. Due to processing changes, changes in microstructure can occur. A microstructure characterization technique was developed to numerically describe cermet microstructure. Machining and electrical properties of dry pressed parts were also analyzed and related to microstructure using this analytical technique.3 Executive SummaryThis report outlines accomplishments in the science-based understanding of cermet processing up to fiscal year 2002 for Sandia National Laboratories. The three main areas of work are centered on 1) increasing production yields of slurry-filled cermets, 2) evaluating the viability of high-solids-loading pastes for the same cermet components, and 3) optimizing cermet powder used in pressing processes (CND50). An additional development that was created as a result of the effort to fully understand the impacts of alternative processing techniques is the use of analytical methods to relate microstructure to physical properties. Recommendations are suggested at the end of this report. Summaries of these four efforts are as follows:1.Increase Production Yields of Slurry-Filled Cermet Vias Finalized slurry filling criteria were determined based on three designs of experiments where the following factors were analyzed: vacuum time, solids loading, pressure drop across the filter paper, slurry injection rate, via prewetting, slurry injection angle, filter paper prewetting, and slurry mixing time. Many of these factors did not have an influence on defect formation. In order of decreasing importance, critical factors for defect formation by slurry filling are vacuum time (20 sec. optimal), slurry solids loading (20.0 g of cermet with 13.00 g of DGBEA solvent (21.2 vol%)), filling with the pipette in a vertical position, and faster injection rates (%7E765 l/s) as preferable to slower. No further recommendations for improvement to this process can be suggested. All findings of the slurry filling process have been transferred to CeramTec, the supplier. Paste filling methods appear to show more promise of increasing production yields. The types of flaws commonly found in slurry-filled vias were identified and followed throughout the entire source feedthru process. In general, all sizes of cracks healed during isopressing and firing steps. Additionally, small to medium sized voids (less than 1/3 the via diameter) can be healed. Porosity will usually lead to via necking, which may cause the part to be out of specification. Large voids (greater 4 than 1/3 of the diameter) and partial fills are not healed or produce significant necking. 2.Viability of High-Solids-Loading-Cermet Paste for Filling Source Feedthru ViaThe paste-filling process is easy to implement and easier to use. The high solids loading (>40 vol %) reduces the incidence of drying defects, which are seen in slurry filled (%7E23 vol %) vias. Additionally, the way in which the vias are filled (the paste is pushed from entrance to exit, displacing air as the paste front progresses), reduces the chance of entrapped voids, which are common in the slurry filling process. From the fair number of samples already filled, the likelihood of this process being a viable and reliable process is very good. Issues of concern for the paste process, as with any new process, are any problems that may arise in subsequent manufacturing stages of the neutron tube that may be affected by subtle changes in microstructure. Both MC4277 and MC4300-type source feedthrus were paste-filled by hand. X-ray analysis showed a much lower existence of voids in the green parts as compared to slurry-filled parts. The paste shows improvements in shelf life (weeks) as compared to slurry (minutes). This method of introducing the cermet to the via also lends itself very well to an automated filling process where a machine can either drill vias or, with the aid of a vision system, find pre-drilled vias and fill them with paste. The pastes used in this work prove the concept of this automated filling process as MC4277 sources have been filled using such a prototype machine, however, better performing pastes can be developed which are less hazardous (aqueous systems). The paste process was also used to successfully fill MC4300 "dogleg" type sources.3.Optimize CND50 Two methods of creating granulated cermet powder for comparison with dry-ball milled CND50 were explored. The first method, non-aqueous spray drying, was performed at Niro Inc. used a 40/60 (wt %) ethanol/toluene solvent and three binder systems; polyvinyl butyral (B79), ethylcellulose (Ethocel), and hydroxypropylcellulose (Klucel). Due to the nature of small spray-dry systems, an excess amount of fines was present in the granulated powder, which may have contributed to the low angles of repose (68 to 78). This is a moderate increase in 5 flowability as standard dry-ball milled powder possesses an angle of repose of 79-89. Mist granulated powders were produced with a tert-butanol solvent and polyvinyl butyral binder system. The angles of repose were more promising (28). More investigation into the mist granulation method is required. Also, aqueous spray drying may be possible with cermet and should be explored. Compaction of all granulated powders is much closer to a proven pressing powder (Sandi94 - angle of repose 29) which should allow cermet to be pressed to near net shape where die filling is difficult for non-flowing powders.4.Microstructure Characterization An analytical technique was developed to numerically characterize microstructures in terms of molybdenum dispersion, homogeneity, and percolation indices. This technique was applied to dry-ball-milled samples of various ball-milling times (0.5 to 20 hours). Significant change in the microstructure could be seen with milling time. Increased milling time caused agglomeration of molybdenum particles, increasing the percolation index, whereas short milling times promoted higher dispersion indices. This phenomenon is contrary to conventional understanding of mixing. However, conventional ball milling does not usually incorporate granules with binder and separate particles. This discrepancy may explain the odd mixing behavior. It is important to note that the high percolation index possessed by long ball mill times showed lower electrical resistance than low-percolation-index microstructures. However, machinability of high percolation, low-dispersion-index microstructures were poor as compared to microstructures with high dispersion indices and moderate percolation indices. This trade-off between dispersion and percolation (at constant molybdenum levels) suggests that microstructures can be achieved that posses good mechanical and electrical properties. Coincidentally, microstructures that satisfy this condition are produced by the standard dry-ball-milled CND50 (4 hour ball mill time). The performance and sensitivity of the microstructure characterization technique should be evaluated, specifically for electrical conductivity. Processing techniques to decrease the percolation index (lowering molybdenum content, excess ball milling, 6 larger molybdenum particles, etc.) should be employed to determine the point where cermet is not conductive or falls below electrical conduction specifications.7