Walter Brattain

Walter Brattain

Experimental physicist and transistor co-inventor · 1902–1987 · United States

A device can rectify current without amplifying a signal. A transistor can amplify without generating energy. These distinctions are central to understanding what Bardeen and Brattain achieved and why measuring input, output, and power supply separately matters.

Historical bearings

Walter Houser Brattain was born on 10 February 1902 in Amoy, now Xiamen, and died on 13 October 1987. He studied at Whitman, Oregon, and Minnesota. His doctoral work under John Tate concerned electron-impact excitation and scattering in mercury vapor, not electron emission from hot surfaces. He worked at the National Bureau of Standards before joining Bell Laboratories in 1929. Bardeen's National Academy memoir

The wartime story also needs correction. Bardeen records Brattain's work on airborne magnetometers for submarine detection, and says their group learned about wartime semiconductor developments afterward. It is inaccurate to turn the wider radar-detector programme into Brattain's personal wartime project. His experience with surfaces and contacts was important without this invented assignment. Colleague's account

In December 1947, Bardeen and Brattain achieved amplification using closely spaced gold contacts on germanium. Their device was demonstrated to Bell Labs officials on 23 December and publicly announced in June 1948. Shockley led the broader group and later developed a different junction-transistor approach. The 1956 Nobel Prize was shared by Bardeen, Brattain, and Shockley. Museum history, NAS memoir

Surfaces are part of the device

A semiconductor surface does not necessarily behave like an ideal continuation of its interior. Electronic states associated with surfaces can hold charge and alter the electric field and carrier distribution nearby. This helped explain why early field-effect experiments failed to behave as a simple bulk model predicted. Bardeen's theoretical analysis and Brattain's experiments developed together; a strict division into a person who only thought and another who only built misrepresents scientific work. Historical account

A useful engineering lesson is to inspect assumptions at interfaces. Surface preparation, contact geometry, impurities, and bias can change the measured behavior. An observation that disagrees with a simplified model may indicate a missing boundary mechanism rather than a failure of the whole physical theory.

The point-contact device should not be drawn as a modern textbook bipolar junction transistor without explanation. Bardeen and Brattain's 1949 account discusses holes traveling either through a surface layer or through the germanium body. Its geometry and collector behavior have details beyond a generic three-layer cartoon. Primary research abstract

Worked mechanism: amplification needs a power supply

This is an original small-signal circuit model, not a reconstruction of the first transistor's measured specifications. Suppose a biased amplifier has incremental input resistance 100 ohms and current gain magnitude 0.8 into a 10,000-ohm resistive load. Assume a small sinusoidal input, linear behavior around the bias point, and negligible additional loading.

For an RMS input voltage of 1 millivolt:

input current = 0.001 V / 100 ohms = 10 microamperes
output current = 0.8 × 10 microamperes = 8 microamperes
output voltage = 8 microamperes × 10,000 ohms = 0.08 V.

The voltage gain magnitude is 80, even though the current gain is below one. Input signal power is V²/R = 10^-6/100 = 10^-8 W, or 10 nanowatts. Load signal power is I²R = (8×10^-6)²×10,000 = 6.4×10^-7 W, or 640 nanowatts. Power gain is 64.

Where does the extra 630 nanowatts come from? The DC supply provides the energy that the input signal controls. The signal does not create energy. A complete efficiency calculation would also need the bias power, device dissipation, and other losses; power gain alone supplies none of those quantities.

The calculation demonstrates why current gain alone cannot establish whether an amplifier has power gain. The impedances and operating point matter. Nor does the calculation guarantee gain over every frequency or amplitude. Real devices have finite bandwidth and nonlinear limits; large signals may clip or alter the bias.

A carrier-transport model, with boundaries

As a separate original approximation, suppose an excess minority-carrier concentration decays with distance as

excess(x) = excess(0) exp(-x/L).

Here L is a diffusion length under a simplified steady-state model with uniform coefficients. At distance L the fraction remaining is about 0.368; at 2L it is about 0.135. If a collector is moved farther from an injecting contact, fewer injected carriers may reach it before recombination, other factors being equal.

This illustrates why geometry and carrier lifetime matter. It is not a complete model of the 1947 device: electric fields, surface transport, collector effects, and contact preparation can invalidate the simple one-dimensional approximation. A formula is useful only together with its conditions.

Modern MOSFET channel control uses an insulated gate and field effect. It is not simply the same minority-carrier injection process scaled down. Later transistor families, materials processing, and integrated-circuit design required further inventions. Toshiba device explanation

The point-contact transistor also had actual applications and production. Calling it an immediately discarded curiosity erases that history. Conversely, transistor development did not instantly eliminate vacuum tubes or make every later circuit cool, reliable, and low-power under all operating conditions. Museum account of Type A production

Exercises

1. Repeat the amplifier calculation with the same input resistance and current gain but a 1,000-ohm load. Find voltage gain and power gain. Derive a general expression for power gain in terms of current gain and input/output resistances under this model.
2. In the exponential transport model, let L=100 micrometers. Compare the surviving fractions at separations 50 and 150 micrometers. By what factor does the fraction decrease? State two effects omitted by the model.
3. Evaluate this claim: “A 100-fold voltage increase proves that the transistor generates energy, and every MOSFET works by the same injected-hole mechanism.” Explain both errors and specify the measurements needed to establish signal power gain experimentally.

Compare your reasoning with the worked solutions

1. Output current remains 8 microamperes under the model's assumed constant gain, so output voltage is 8 millivolts and voltage gain is 8. Output power is (8×10^-6)²×1,000=64 nanowatts, giving power gain 6.4. Generally Pout/Pin=(Iout/Iin)²(Rout/Rin)=|Ai|²Rout/Rin. This assumes the resistances and gain apply at the chosen bias and frequency; changing a real load may change the device's gain.
2. The fractions are exp(-0.5)≈0.6065 and exp(-1.5)≈0.2231. Their ratio is exp(-1)≈0.3679, so the farther point retains about 36.8% as much excess concentration, a decrease by a factor of about 2.718. Electric-field drift and surface recombination are two omitted effects; contact geometry and nonuniform material parameters are others. Concentration at a point is not automatically identical to measured collector current.
3. Voltage gain alone omits current and impedance. Even a passive transformer can increase voltage while decreasing current. An active transistor circuit draws energy from its supply. Measure input and output voltage and current with their phase relationship, frequency, load, and bias specified; for purely resistive sinusoidal ports RMS products give average powers. MOSFET gate-controlled channel conduction is physically different from the simplified injected-hole model, so shared switching or amplification functions do not imply identical carrier mechanisms.

Reading the original work

The 1948 paper record identifies Bardeen and Brattain's initial publication. The 1949 abstract adds important physical qualifications. The worked circuits here use explicitly chosen teaching parameters; they are not historical measurements or claims about a particular commercial device.