Early Mos Yield and Reliability Problems at Volume

# Early MOS Yield and Reliability Problems at Volume

## 1. The Gap Between a Working Die and a Reliable Shipment

Shipping a single working PMOS die in a lab is a different problem from shipping thousands of them a month to a calculator maker who will solder them into a consumer product and expect years of field life. General Microelectronics and Electronic Arrays both struggled through the mid-1960s with gate-oxide contamination, threshold-voltage drift from mobile ion contamination, and the general difficulty of controlling a thin oxide layer's properties across an entire production run rather than a single hand-built wafer.

Mobile sodium-ion contamination in particular caused parts to shift their switching threshold months after shipment, a failure mode invisible at final test but devastating to a calculator maker's warranty return rate, and it took several years of process cleanup — better furnace tube materials, tighter handling discipline, phosphosilicate glass passivation layers — before PMOS yield and field reliability reached a level volume customers would tolerate.

## 2. Why This Mattered More for MOS Than for Bipolar

Bipolar logic families already had a decade of manufacturing discipline behind them by the mid-1960s; MOS did not, and every new PMOS fabrication line had to rediscover the same oxide-cleanliness lessons more or less from scratch. The companies that survived this period were the ones willing to treat yield and contamination control as a core engineering problem rather than an afterthought to the circuit design, a lesson that shaped how every MOS entrant after them, including a new company founded in 1968, approached process work from day one.