the scanning tunneling microscope
The scanning tunneling microscope (STM) does something that sounds impossible: it feels individual atoms on a surface with a needle, without ever touching them. You take a metal tip sharpened to a single atom at its very end and bring it to within about a nanometre of a surface — close, but with a clear gap. Then you apply a small voltage between tip and surface. Classically no current should flow across the empty gap, but quantum mechanics allows electrons to 'tunnel' across it, and a tiny tunneling current appears. That current is the signal. The instrument, invented by Gerd Binnig and Heinrich Rohrer in the early 1980s (Nobel Prize 1986), turned that faint quantum current into pictures of single atoms.
The magic is how sharply the tunneling current depends on the gap. It falls off exponentially with distance — shrinking by roughly a factor of ten for every extra tenth of a nanometre of gap. This extreme sensitivity is what gives vertical resolution finer than a single atom: move the tip a hair closer and the current leaps. In the usual mode, an electronic feedback loop moves the tip up and down as it scans, constantly adjusting height to hold the current fixed; the record of those height adjustments, mapped across the surface, becomes a contour map on which individual surface atoms appear as bumps. The tip can even be used to nudge single atoms into place — the famous demonstrations that spelled out words in atoms.
Here is the honest subtlety. The STM does not literally sense the nuclei; the tunneling current depends on the local density of electronic states available at the surface. So the 'bumps' are really where electrons are available to tunnel, which usually but not always coincides with atom positions — sometimes a bright spot is an electronic state rather than an atom, and the picture depends on the voltage you apply. And there is a hard requirement: the sample must conduct electricity (a metal or semiconductor), since an insulator cannot carry the tunneling current. Clean surfaces and often ultrahigh vacuum are needed too. Within those limits, the STM sees the electronic surface atom by atom — a genuinely atomic-resolution window, as long as you remember it is showing electrons, not nuclei.
On a clean silicon (111) surface the STM reveals the famous 7x7 reconstruction: an ordered array of bumps where the surface atoms have rearranged into a pattern with a repeat seven times the bulk spacing. The tunneling current, changing about tenfold for every 0.1 nm of height, lets the feedback loop trace that atomic corrugation faithfully.
The tunneling current's exponential distance dependence gives sub-atomic vertical resolution — but it maps electrons, not nuclei.
An STM image maps the local density of electronic states, not atomic nuclei directly, so a bright spot may be an electronic state rather than an atom, and the picture can change with the applied voltage. The sample must also be electrically conducting — insulators give no tunneling current.