Mesa Production 1958 Scale Furnace Schedule Boat
# Scale the Furnace Schedule From One Wafer to a Production Boat: One Setpoint, Eight Positions
## 1. Why a Furnace That Works for One Wafer Can Still Fail a Boat of Eight
This step takes the diffusion schedule that worked correctly for a single demonstration wafer in 1956 and 1957 and asks it to produce the same result simultaneously at every position along a quartz boat holding several wafers at once — and the furnace does not treat every position identically, because gas flow, radiant heating, and dopant-source depletion all vary slightly from the boat's mouth to its back. A furnace tube held at a single nominal temperature still carries a real spatial temperature gradient along its length, and because diffusion depth depends on temperature through an exponential diffusivity term, even a small gradient produces a measurable spread in junction depth from one end of the boat to the other:
where $z$ is position along the boat, $T(z)$ the local temperature at that position, $D_0$ a material constant, and $E_a$ the diffusion activation energy. Because $D$ depends exponentially on $T$, a temperature difference of only a few degrees between the front and back of the boat can produce a junction-depth spread large enough to separate a wafer that meets specification from one that does not — a spread that simply could not exist when this project's earlier series diffused one wafer at a time.
## 2. Real Diagram: The Same Recipe, Eight Different Outcomes
A single furnace recipe, run once, no longer produces a single result. It produces eight simultaneous results, one per wafer position, and the spread between the best and worst of those eight positions is a new quantity this project's earlier series never had to measure, because they never ran more than one wafer at a time.
## 3. Why Neither 1956 nor 1957 Needed a Position-Dependent Specification
The 1956 drift-transistor process and the 1957 photolithography process both specified furnace schedules as a single temperature and a single time, because each of those series' articles implicitly described a furnace holding one wafer. This step is the first in this project's history where a furnace schedule must be qualified not as a single setpoint but as a *profile* across physical position, because position itself has become a process variable the moment more than one wafer shares a tube. The fix is not a different furnace or a different dopant; it is a schedule written to a tolerance that holds across every position at once, which is a genuinely new kind of engineering discipline this project's earlier, single-wafer series never had reason to develop.
Step 2 does not change what any wafer in this boat is diffused with; it decides whether the same furnace cycle can be trusted to treat all eight of them the same way, a question that simply did not exist before volume made the boat bigger than one wafer.