John Bardeen

John Bardeen

Physicist and electrical engineer, 1908–1991. Study focus: semiconductor interfaces, amplification, and the assumptions behind a microscopic theory.

Bardeen worked with Walter Brattain on the point-contact transistor and with Leon Cooper and J. Robert Schrieffer on the BCS theory of superconductivity. These achievements came from different collaborations linking theory with experiments. The technical lesson is to identify what a model explains, which observations challenge it, and where its assumptions stop applying.

Historical setting

Bardeen's education included electrical engineering at Wisconsin and doctoral research under Eugene Wigner at Princeton. At Bell Laboratories he studied semiconductor behavior in a group led by William Shockley. Bardeen and Brattain obtained transistor action in December 1947; the successful experiment on December 16 and demonstration to management on December 23 are distinct events. Their point-contact device was also distinct from Shockley's subsequent junction-transistor work. Museum history of the experiment, 1948 original paper record

Bardeen moved to the University of Illinois in 1951. With Cooper and Schrieffer he developed the theory published as Theory of Superconductivity in 1957. He shared the 1956 physics Nobel recognition for semiconductor research and the transistor, and the 1972 physics Nobel recognition for superconductivity. The prizes identify collaborative work; they are not a substitute for explaining it. Illinois biography, BCS paper

Why the surface changed the experiment

An electric field can alter carrier concentrations in a semiconductor. Early attempts to obtain the expected field effect failed to match simple predictions. Bardeen proposed that electronic states at the surface could capture charge and screen the applied field. Bardeen and Brattain investigated those states experimentally, leading to the point-contact device. This is a specific explanation of a mismatch between a model and an experiment, not the claim that nobody previously knew interfaces had physics. Surface-state problem

Here is an original, deliberately simplified charge-budget model. Suppose an electrode induces 100 arbitrary units of charge. If interface traps take 80 units, only 20 remain to change the mobile-carrier population in the intended channel. Reducing active trapping to 20 units leaves 80 for that population: four times the previous modulation for the same total induced charge.

The arithmetic illustrates why an interface can dominate a bulk-oriented design. It is not a physical transistor equation. Real devices require electrostatics, energy levels, capacitances, occupation probabilities, transport, and geometry. A useful experiment would vary surface preparation while holding other conditions fixed, then measure whether the predicted response changes. One should not explain every failed transistor by invoking surface states without such discrimination.

Amplification does not create energy

A transistor can use a small input variation to control energy drawn from a power supply. In an original small-signal example, an input current change of 20 microamperes accompanies an output current change of 1 milliampere. The incremental current gain is 0.001/0.000020=50.

That ratio does not establish a power gain of 50. If the input signal voltage is 0.1 volts and the output signal voltage is 0.5 volts, the corresponding simplified signal powers are 2 microwatts and 500 microwatts, a ratio of 250. These numbers assume compatible amplitude conventions and an appropriate linear operating regime. The additional output energy comes from the bias supply, not from nowhere.

An amplifier can also distort or saturate. If it can vary its output by at most 2 milliamperes, extrapolating gain 50 to a 100-microampere input would predict 5 milliamperes, outside that range. A measured local gain is not a globally valid transfer law. Computing with transistors relies on characterized operating regions, noise margins, timing, and power constraints.

Modern integrated MOS devices are not physically identical to the original point-contact transistor. They share the broader goal of controlling electrical signals, but geometry, materials, fabrication, and operating principles differ. A history of the transistor should not collapse all subsequent device engineering into one unchanged invention.

What BCS explains

The 1957 theory treats an effective attraction that can arise through interaction with lattice vibrations and competes with screened electrical repulsion. Its superconducting state involves correlated pairing and a gap in the excitation spectrum. The original paper connects the model with experimental features including the isotope effect, specific heat, and magnetic response. Original BCS abstract

The familiar picture of one electron deforming a lattice and another responding is an intuition for an effective interaction, not a literal account of two tiny balls tied together. Cooper's pairing problem concerns electrons near a filled Fermi sea, not any arbitrary pair of electrons in vacuum. A superconducting state is a collective quantum state.

It is also inaccurate to say that nothing in a superconductor can scatter or lose energy. Quasiparticles, defects, magnetic fields, finite temperatures, and excessive currents require careful treatment. A gap and phase coherence help explain important properties within a specified regime. They do not make every experimental circumstance dissipation-free, nor does the simplest phonon-mediated model settle every unconventional superconductor. APS account of BCS development

For a constructed energy-scale exercise, suppose a particular measurement requires creating two excitations, each costing at least 1.5 millielectronvolts. The combined minimum is 3 millielectronvolts. A 2-millielectronvolt input cannot create that pair under the stated idealized rule. A 4-millielectronvolt input clears the energy threshold, but allowed transitions, coupling, and conservation laws still determine what occurs. Distinguishing necessary conditions from sufficient conditions is part of reading physical explanations responsibly.

Exercises

1. In the charge-budget model, 120 units are induced and 90 trapped. A treatment reduces trapped charge to 45 without changing total induced charge. Compute the original and new mobile contributions and their ratio. Give one reason the model is insufficient to predict a real device's gain.
2. An amplifier has local current gain 40 and a maximum output variation of 1.2 milliamperes. Predict the output variation for a 10-microampere input, then explain the problem with applying the same law to 50 microamperes. Where does amplified signal energy come from?
3. A summary says “BCS proves that all paired electrons cannot scatter, so every superconductor has zero loss at any temperature.” Identify three separate problems with this inference, using the model's stated scope.

Compare your reasoning with the worked solutions

1. Initially 120-90=30 units contribute; afterward 120-45=75. The ratio is 2.5. This budget omits carrier mobility, electric-field distribution, contact behavior, and transport. More available mobile charge may improve conductivity modulation, but gain is a property of a complete biased device and circuit, not this subtraction alone.
2. For 10 microamperes, the local model predicts 400 microamperes, within the stated output range. For 50 microamperes it predicts 2 milliamperes, beyond the maximum, so saturation or another nonlinear response must occur. The bias supply provides the output energy. Gain describes control of that energy and does not violate energy conservation.
3. Pairing alone is not a blanket prohibition on all scattering; the many-body state, excitation spectrum, and perturbations matter. The model has a temperature-dependent superconducting regime rather than validity at any temperature. Finally, conventional phonon-mediated BCS theory is not an automatic complete description of every superconducting material or every loss mechanism. A claim about ideal DC resistance cannot be substituted for zero dissipation under all experimental conditions.