The most advanced chips are printed with light of 13.5 nm. No material is transparent to it, so the machine that prints them steers the light with mirrors only. Each mirror returns about three quarters of the light and keeps the rest as heat.
This project follows one such mirror all the way: why a single surface reflects almost nothing, how 40 pairs of thin layers bring that to 74%, where the remaining light ends up, and how hot it makes the glass. Each idea comes with the reason it is there, an experiment to move with your hands, a full explanation and the formulas with their origin.
The problem in one picture
Between the source and the wafer the light meets a chain of mirrors. With the ideal mirror of these lessons, which reflects 74.2%, ten reflections leave 5% of the light. Everything else has become heat in the mirrors, most of it in the first ones.
Two questions follow, and they organise the whole project. How can a mirror reflect as much as 74% of a light that every material absorbs? And what does the quarter it keeps do to it?
What the mirror is
A thick disc of glass carries, on its surface, 40 pairs of layers: molybdenum, which reflects, and silicon, which keeps the molybdenum layers at the right distance. Each layer is a few nanometres thick, a few tens of atoms. The light is reflected a little at every boundary, and the thicknesses are chosen so that all those small reflections add up.
How each module is made
- An index of ideas. Every module opens on the list of its ideas, each with a line on what it explains and which experiment it holds. Any of them can be opened directly.
- Why here. Each idea begins by saying why it comes at that point: what it builds on and what will use it later.
- An experiment. Sliders and graphs to move, with a few presets marked Try that set a telling case and say what to look at.
- The explanation and the formula card. What the experiment shows and why, then each formula with where it comes from, what it assumes, what its terms mean and a check you can do yourself.
- No tests. Nothing has to be answered or completed. The last idea of each module ends with a short list of questions for retracing the reasoning on your own.
All numbers are computed in the page for an ideal mirror: 40 pairs with perfectly sharp, smooth boundaries, at 13.5 nm and 6° from perpendicular. Real mirrors reflect somewhat less. Optical constants from the CXRO database (Henke, Gullikson, Davis; Lawrence Berkeley National Laboratory).