A tiny gap makes the current very sensitive to distance
Binnig and Rohrer’s lecture follows local tunneling spectroscopy as it grew into scanning tunneling microscopy. With a bias applied, electrons tunnel between a conducting tip and a conducting sample brought very close together. The current depends so strongly on distance that tiny changes at the junction become measurable changes in current.
Turning this into a microscope took vibration isolation, a stable approach, sharp tips and scanning control, and Christoph Gerber and Edmund Weibel contributed to the work. An equation for barrier transmission is not enough to build a working microscope.
A simple barrier model explains the sensitivity
For a simple, constant vacuum barrier at low bias, the current falls off roughly exponentially with the gap. The decay rate depends on the electron mass and an effective barrier energy. A change in barrier height on the scale of one electronvolt, or a change in the state of the tip, can therefore alter the current even when the surface does not move.
The formula explains only how sensitive the current is to distance. It ignores the detailed densities of states of tip and sample, the way bias reshapes the barrier, and lateral structure. Fitting it pixel by pixel to turn current into height would assume a barrier that is known and the same everywhere.
Sources: [1]
The silicon scans showed the seven-by-seven reconstruction
The lecture shows the seven-by-seven reconstruction of silicon (111) twice: as a relief assembled from the original recorder traces and as a processed top view. Side by side, the two show how measured scan lines became an interpreted image of the surface.
The pattern narrowed down the competing models of the surface, but it did not directly give the positions of every atom beneath it. Tunneling depends on electronic states and on bias, so a bright spot need not sit over the geometric center of an atom.
Sources: [1]
Feedback speed and bias both shape the image
In constant-current mode, feedback raises and lowers the tip to hold the current at a set value. If the scan runs faster than the feedback can follow, slopes and edges are distorted, and a tip with several apexes can show the same feature more than once. In constant-height mode the tip stays at one height and the current varies instead; the risk of the tip hitting the surface is different.
Changing the bias changes which sample states take part and how the tip responds. A reversal of bright and dark can therefore be an electronic effect; the surface relief has not flipped. Reading an image quantitatively requires the junction conditions and a tunneling model.
Sources: [1]
The linked Lab models the barrier only
The existing tunneling Lab computes transmission through a barrier that you set. It has no full raster scan, no current feedback and no imaging of local densities of states, so it does not count as an STM. This article stands on its own as a historical reading; a dedicated STM instrument is planned as a separate project.
Sources: [1]