Mesa Production 1958 Burn in Screening Early Failures
# Burn-In Screening: Catching Early Failures Before Shipment, Not After
## 1. Why the Weakest Devices Fail Fastest, and Why That Is Useful
This step runs every finished device, not just a sample, under electrical bias and elevated temperature for a short, fixed duration before packaging is finalized, deliberately trying to provoke exactly the kind of early failure Step 8's analysis showed this product is prone to — because a device carrying one of the marginal defects Step 8 characterized tends to fail quickly under stress, while a genuinely sound device tends to survive the same stress with no measurable change at all. The population of devices leaving Step 7's coating process is not uniform in reliability; it contains a mix of sound devices and a smaller fraction carrying latent defects — a thin coating spot, a borderline mesa sidewall — that have not yet failed but are disproportionately likely to fail soon. The failure rate of such a mixed population over time famously traces a bathtub shape, and burn-in exists specifically to consume the steep falling part of that curve before the device ever reaches a customer:
where $\lambda(t)$ is the instantaneous failure rate, $\eta$ a characteristic life, and $\beta$ the Weibull shape parameter. A shape parameter below one means the failure rate is highest at the very start of life and falls as weaker units are removed from the surviving population — which is exactly the signature of latent, pre-existing defects rather than wear-out, and it is why running every device through a short, controlled stress before shipment removes a disproportionate share of future field failures for a comparatively small cost in test time.
## 2. Real Diagram: Burn-In Removes Devices, It Does Not Repair Them
Burn-in is sometimes mistaken for a step that strengthens a device. It does not — it simply forces latent failures to happen now, on this line's own test bench, instead of later, in a customer's equipment. The devices that survive are not improved; they are simply the ones that did not carry the defect burn-in was designed to provoke.
## 3. Why Screening Is This Series' Cheapest Fix and Also Its Least Complete One
Step 7's coating and Step 8's failure-mode analysis both tried to address the exposed-junction problem at its physical source — reducing how much contamination reaches the sidewall, and understanding exactly how the ones that fail do so. This step addresses the problem at neither source; it only decides which already-manufactured devices get to leave the factory. That makes it the cheapest intervention in this series, requiring no change to the etch, the coating, or the materials, and it is also the most limited: burn-in catches a device with a defect severe enough to fail quickly under stress, but a marginal defect too mild to fail during a short burn-in window can still slip through and fail months later in the field, at a time this line cannot observe and a cause this line cannot trace back to itself.
Step 9 does not fix the exposed junction Step 5 created; it buys this line a last, economical chance to catch the worst of that defect's consequences before a customer, rather than this factory, discovers them.