Mesa Production 1958 Kilby Monolithic Integrated Circuit
# Context: Kilby's Monolithic Integrated Circuit, Same Year at TI
## 1. Why the Same Year That Perfected One Transistor Also Made the Single Transistor Obsolete as a Unit of Progress
In September 1958, while this line was qualifying the 2N696 and 2N697 for IBM one discrete device at a time, Jack Kilby at Texas Instruments built a working phase-shift oscillator with every component — transistor, resistors, and capacitors — fabricated on a single piece of germanium and connected by fine wires, rather than assembled from separately packaged parts soldered onto a circuit board. The problem Kilby's device addressed had nothing to do with any individual component's quality, and everything to do with how many connections a circuit needs once it contains more than a handful of parts. Every soldered or wire-bonded joint in a circuit carries some small probability of failure, and because a circuit needs every one of its joints to work, system reliability falls off sharply as the joint count grows:
where $p$ is the failure probability of a single joint, $N_{\text{joints}}$ the total number of connections in the circuit, $n$ the number of discrete components, and $k$ the average number of leads each one needs connected. Even a small $p$ becomes a serious problem once $n$ grows into the hundreds, which is exactly the regime electronics was entering by 1958 — a condition the industry was already calling the tyranny of numbers. Kilby's answer was not to make $p$ smaller; it was to make $N_{\text{joints}}$ smaller, by eliminating the external connections between components entirely and leaving only the much smaller number needed to connect the finished circuit to the outside world.
## 2. Real Diagram: Two Reliability Crises, Solved Two Different Ways, in the Same Year
This series has spent Steps 5 through 10 fighting a reliability problem that lives at the edge of a single exposed junction. Kilby's problem lives in the wiring between many separately packaged components. Both are genuine 1958 reliability crises; neither one's solution touches the other's cause at all.
## 3. Why This Context Does Not Change Anything About This Series' Own Problem
It would be a mistake to read Kilby's monolithic circuit as a solution, even in principle, to the exposed-junction problem this series has tracked since Step 5. Integration reduces the number of external connections a circuit needs; it says nothing about whether any individual transistor's own junction edge is protected from its environment, and in fact a monolithic circuit built the way Kilby demonstrated in 1958 would have exactly the same kind of exposed-junction vulnerability this series has been fighting, multiplied by however many transistors sit on the same piece of germanium. The real fix for *that* problem — not discrete-versus-integrated, but exposed-versus-protected at the junction itself — is still a year away, waiting on a different insight entirely: growing the oxide over the junction and never removing it, rather than cutting it away at the mesa etch the way this series' Step 5 still does.
Step 13 does not belong to this production line's own story; it is this year's other half, a reminder that 1958 was solving two separate reliability crises at once, and that this series' crisis — the one still open as this account reaches its final step — would need a fix from the other side of the industry before it was ever truly closed.