Particles in fields
The Hall effect
Edwin Hall found in 1879 that a magnet pushes the current in a strip of gold leaf towards one edge. The moving carriers feel a sideways force $q\mathbf v \times \mathbf B$, crowd towards one edge and leave the other short of them. Their charge makes a field across the strip, which grows until it cancels the magnetic push. The voltage across the strip then tells you how many carriers there are and the sign of their charge.
Before you look
Predict first: The carriers are electrons, the current runs to the right and B points out of the board. Which edge goes negative?
What you are looking at
A strip of a model semiconductor 5 µm wide and 0.4 µm thick, carrying 5 µA to the right in 1 T out of the board. The dots are the carriers, the white lines a few carriers’ recent paths. The coloured bands and the + and − signs along the edges show the edge charge; the amber arrows are the Hall field in the middle of the strip.
The voltmeter reads the bottom edge minus the top; the trace below follows it against IB/nqt. Time is in picoseconds: a carrier collides every 2.8 ps on average.
The physics
In the steady state the Hall field cancels the magnetic force on the average carrier, , where is the carriers’ drift speed. A strip of width and thickness , with carriers of charge per cubic metre, carries the current , so
So the Hall coefficient gives the carrier density, and its sign the carriers’ sign.
Going deeper
Vector calculus and Maxwell’s equations
In the Drude model the mean velocity obeys . With no current across the strip, the steady state has along it and across, whatever the collision time: depends on alone.
Until the edge charge builds, the current is turned through the Hall angle, : 27° here at 1 T.
What to measure
V_H = IB/nqt, with the carriers’ sign
- The checks average the voltmeter over 60 collision times and compare it with IB/nqt (within 3%), for electrons and for holes, whose voltage has the opposite sign. Doubling B or the current doubles the voltage (within 4%); with no field it stays at zero.
What the model leaves out
- One kind of carrier with the free electron’s mass and a single collision time. Real semiconductors have effective masses, scattering that depends on energy, and sometimes electrons and holes together, all of which change .
- Contacts: the strip is periodic along its length.
- Heating. The carriers drift at a third of their thermal speed, and fast ones are pushed further than slow ones, so one edge runs warmer (the Ettingshausen effect). That shifts the reading by under 1% here.
Things to try
- Switch to holes: the same current gives the opposite voltage.
- Set B to zero: the carriers go straight on and the voltmeter reads only noise.
- Halve the current: the voltage halves, and so does the edge charge.
References
- E. H. Hall (1879), On a new action of the magnet on electric currents, American Journal of Mathematics 2, 287.
- N. W. Ashcroft and N. D. Mermin (1976), Solid State Physics, chapter 1, the Drude theory of metals.
More experiments, and how the lab works, are in the lab’s overview.
