Altera and Taiwan Semiconductor Manufacturing Company, Ltd. (TSMC) have developed a methodology called FabMatch® to ensure process technology is duplicated across all fabs, thereby resulting in identical devices being produced at multiple wafer fabs. The FabMatch approach compares the process, reliability and device performance data on products from two separate facilities to make certain the process technology is duplicated, and that the products have identical electrical and product performance characteristics. The FabMatch program was initiated in 1998, and adapted by TSMC in 2000. Since then, the program has been used with multiple fabs and processes without issue. FabMatch virtually eliminates capacity constraints and enhances manufacturing flexibility.
The process technology from both fabs is matched by replicating both input and output parameters. In-line measurements of the process parameters, such as the gate oxide thickness, interlevel-dielectric (ILD) thickness, poly etch critical dimension, salicide thickness, and all metal layer etch critical dimensions, are taken and compared to make sure they all are within specification limits. Structural cross-sections of the same device built at both fabs are compared to make sure they are physically identical. The electrical parameters are measured and compared. These include: NMOS/PMOS VT and Isat; N+/P+ contact and sheet resistances; N+/P+ poly contact and sheet resistances; metal 1-9 and poly resistances; and via 1-8 resistances.
Process reliability tests are conducted on the devices from both fabs to confirm that the devices meet reliability requirements. The reliability tests include: high temperature operating life (HTOL) test, (125°C junction, 1.8V, 1000 hours); high temperature storage (HTS) test, (150°C, 1000 hours); temperature cycling (TC) test, (Condition B, 1000 cycles); highly accelerated stress test (HAST), ( 130°C, 85 percent relative humidity, 1.7 ATM pressure, 1.5V bias, 96 hours); biased moisture resistance test, (85°C, 85 percent relative humidity, 1000 hours); and an accelerated unbiased moisture resistance test, (130°C, 85 percent relative humidity, 168 hours).
Product performance measurements are taken and compared to demonstrate that devices from both fabs meet the product performance specifications. The parameters measured and compared can include timing path measurements, phase-locked loop (PLL) measurements, and low-voltage differential signaling (LVDS) measurements.
The FabMatch process compares the process technology, reliability, and product performance data to ensure that all critical parameters measured are well within the product acceptance specifications. Based upon such data, a process transfer from one fab to another can be fully qualified by Altera.