Rate-Grown Transistor
A rate-grown transistor is a grown-junction transistor in which the p-n junctions are produced by changing how fast the crystal is pulled from the melt, not by dropping new impurities into it. The method was developed at the General Electric research laboratory in 1951, with the work of Robert N. Hall and W. C. Dunlap often cited, and it let one melt containing both a donor and an acceptor yield n-type or p-type crystal depending on pull speed alone.
The physical basis is segregation, the tendency of impurities to be taken up by a growing crystal in a different proportion from their concentration in the melt. Each dopant has its own distribution coefficient, the ratio of how much enters the solid compared with the liquid, and for most dopants in germanium this is much less than one, so the melt is gradually enriched in the impurity left behind. Crucially, the effective value depends on growth conditions, including how fast the crystal grows, and different impurities respond by different amounts. A melt deliberately doped with both a donor such as antimony and an acceptor such as gallium therefore gives a crystal whose type depends on which impurity is captured more at the moment. Speeding up or slowing down the pull shifts the balance, so the crystal can flip between n-type and p-type as the rate is varied, with each switch forming a junction.
In practice the process is a pulling schedule rather than a series of additions. A melt containing both impurities is prepared, a seed is dipped in, and the crystal is pulled at one rate to grow the first n-type region. The rate is then changed for a short interval to grow a thin p-type layer, and then returned to the original value to continue with n-type. The thickness of the base layer is controlled by how long the changed rate lasts, which can be manipulated more finely than the time needed to add a pellet and wait for it to dissolve. The result is the same n-p-n sandwich as in the simple grown-junction device, then sliced, cut into bars and fitted with leads in the same way. The attraction was a thinner and more controllable base than simple melt doping allowed, though how much better depended on the crystal and the operator, so no single figure should be taken as typical.
The method's strengths and weaknesses follow from relying on a growth-rate sensitivity. Because the layers come from a speed change rather than from a mechanical addition to the melt, the transitions are quicker and the base can be thinner, and the process fits naturally onto a crystal puller that already has a controllable drive. The weaknesses are that the ratio of dopant capture is also sensitive to temperature, stirring and melt composition, so the process needs very steady conditions, and that the crystal still has to be sliced into bars with the layer found afterwards, so the yield per crystal remains low and the cutting waste high. The thin region is also only as straight and flat as the freezing interface at the time it grew, which is never perfectly planar.
It is easy to confuse with several neighbouring methods. Rate growing is a form of grown junction, not an alloy, a diffusion or a surface-barrier process; in all of those the junction is made on an already grown wafer, while here it is built in as the crystal forms. It is sometimes loosely attributed to RCA or lumped with the RCA and GE alloy transistors of the same period, but the pull-rate technique is the GE laboratory's, and RCA's early commercial junction devices were mainly alloy types. A further pitfall is to think the pull rate by itself creates the dopants; it only changes how much of each of the two already present impurities is incorporated. Later variants, such as meltback and grown-diffused methods, combined growing with a further heat treatment to push impurities across the base and make it still thinner.
Its legacy is as one of several ingenious attempts to extract a transistor from the crystal-growing step before wafer processing took over. Grown and rate-grown junctions were costly and low-volume, and they were overtaken by the alloy types for cheap audio parts and by diffusion and the planar process for everything else. Yet the physics behind them, that incorporation of an impurity depends on the rate and conditions of growth, remains central in modern crystal growth, epitaxy and the control of dopant profiles in silicon ingots, where unwanted striations in resistivity are a well-known consequence of the same effect.
Read the rate-grown transistor as the grown junction refined by letting the speed of the pull, instead of a pellet in the melt, decide which impurity wins: a clever use of segregation physics that gave thinner bases, and a reminder that the junction can be written into a crystal as it freezes.