spacecraft, semiconductor reliability, single controlled baseline lot

Figure 1

In commercial integrated-circuit production flows, to achieve the lowest possible cost for a particular product, the yield is maximized and fab loading optimized to ensure a reliable and economical product supply chain. Planners typically distribute the fabrication of consumer or commercial-off-the-shelf (COTS/COTS+) product to multiple qualified wafer-facilities and assembly and test (A/T) sites. As Figure 1 shows, many larger manufacturers use 3 to 4 wafer fabs and A/T sites. If a natural disaster were to damage a fab or assembly site in one location, production is readily re-routed to the other unaffected fabs and A/T sites to take up the production, thereby avoiding a supply short-fall. At each manufacturing and A/T site, the production recipe, material set, and final assembly and test flows are changed/adjusted locally to optimize each for cost, yield, cycle-time, etc. The end-product from any site, by definition, must meet the data sheet specifications, BUT parameters and properties will be unique for each flow, introducing more variation in the specific quality and performance of product from each flow. Additionally, the same parts from the same manufacturing flows will vary over time (e.g. due to die revisions, mold compound change, etc.).

spacecraft, semiconductor reliability, single controlled baseline lot

Figure 2

When buying parts from a distributor, it is likely that the population of units that you’ll receive will be from different production runs, different fabs, and even different A/T sites. This implies that a typical COTS/COTS+ population of units is likely to have the largest variability of any other grade of semiconductor product. This wide variability makes the requirement for proper up-screening much more critical for ensuring that units of known quality and performance end up in a spacecraft system. It implies buying many times the number of units needed, testing all of the units at multiple temperatures, sacrificing some for SEE and TID studies, and doing some combination of burn-in, qualification, and/or HTOL on all units to fly in an effort to screen early failures and give high confidence of mission success. Included with this huge test over-head is the long cycle-time and the constant risk that units may not meet the reliability requirements after all the testing and/or that units may end up being overly stressed during the up-screening.

In comparison to the COTS/COTS+ flow, any semiconductor grade that produces populations of units from a single controlled baseline (SCB) lot flow (Figure 2) will produce units with much more constrained variation. This type of flow offers significantly improved quality and reliability. Product obtained from a SCB lot is from a single production lot (12-25 wafers processed at the same time on the same equipment within the fab) – all run in a single qualified wafer fab and assembled and tested at a single A/T site with one set of packaging materials.

spacecraft, semiconductor reliability, single controlled baseline lot

Figure 3

By running material consistently through the same fab and using the same single assembly/test facility, a much more consistent material is guaranteed. Demonstrating the variability from lot-to-lot and fab-to-fab, the Figure 3 shows parametric NMOSFET (x-axis) and PMOSFET (y-axis) threshold voltage data collected from two manufacturing fabs over several lots of materials. It is clear that getting units from multiple lots will increase parameter variation and getting units from multiple manufacturing sites will increase it further (red data points vs. single lot green data points). It should be noted that SCB lots are usually fabricated in QML (Qualified Manufacturer List) facilities, where the process and operating procedures are regulated and inspected by tight military specifications with all material tracked through the fab/assembly/test and after-packaging, so there is unit/lot traceability. In commercial flows such control and tracking is impossible due to cost pressures especially on super-high volume commodity parts selling for a few cents per component.

In summary, any product grade based on an SCB Lot flow will provide units with significantly reduced manufacturing variation as compared with product manufactured on a COTS/COTS+ flow. Thus, these types of grades provide a population of units with significantly higher quality and reliability and a more stable radiation performance (allowing “what you test is what you fly”). With proper up-screening, these higher grades offer many of the key advantages of using full space-grade parts but with a reduced price to make them more attractive in the larger volumes being sought after by the small-satellite market sectors.