IEEE Spectrum July, 2014 - 34

TraNSiSTOriziNg
THe vacuuM Tube

source

gate
drain
source

drain

Mo sFet

ciently for it to emit electrons. This is why vacuum tubes need
time to warm up and why they consume so much power. It's
also why they frequently burn out (often as a result of a minuscule leak in the tube's glass envelope). But vacuum-channel
transistors don't need a filament or hot cathode. If the device
is made small enough, the electric field across it is sufficient
to draw electrons from the source by a process known as field
emission. Eliminating the power-sapping heating element reduces the area each device takes up on a chip and makes this
new kind of transistor energy efficient.
Another weak point of tubes is that they must maintain a high
vacuum, typically a thousandth or so of atmospheric pressure,
to avoid collisions between electrons and gas molecules. Under
such low pressure, the electric field causes positive ions generated from the residual gas in a tube to accelerate and bombard
the cathode, creating sharp, nanometer-scale protrusions,
which degrade and, ultimately, destroy it.
These long-standing problems of vacuum electronics aren't
insurmountable. What if the distance between cathode and
anode were less than the average distance an electron travels
before hitting a gas molecule, a distance known as the mean
free path? Then you wouldn't have to worry about collisions
between electrons and gas molecules. For example, the mean
free path of electrons in air under normal atmospheric pressure is about 200 nanometers, which on the scale of today's
transistors is pretty large. Use helium instead of air and the
mean free path goes up to about 1 micrometer. That means an
electron traveling across, say, a 100-nm gap bathed in helium
would have only about a 10 percent probability of colliding
with the gas. Make the gap smaller still and the chance of collision diminishes further.
But even with a low probability of hitting, many electrons are
still going to collide with gas molecules. If the impact knocks a
34

jul 2014

north american

SPectrum.ieee.orG

gate

Vacu u M- ch a n n e l tr ans i sto r

bound electron from the gas molecule, it will become a positively charged ion, which means that the electric field will send
it flying toward the cathode. Under the bombardment of all
those positive ions, cathodes degrade. So you really want to
avoid this as much as possible.
Fortunately, if you keep the voltage low, the electrons
will never acquire enough energy to ionize helium. So if the
dimensions of the vacuum transistor are substantially smaller
than the mean free path of electrons (which is not hard to
arrange), and the working voltage is low enough (not difficult
either), the device can operate just fine at atmospheric pressure. That is, you don't, in fact, need to maintain any sort of
vacuum at all for what is nominally a miniaturized piece of
"vacuum" electronics!
But how do you turn this new kind of transistor on and
off? With a triode vacuum tube, you control the current flowing through it by varying the voltage applied to the grid-
a meshlike electrode situated between the cathode and the
anode. Positioning the grid close to the cathode enhances
the grid's electrostatic control, although that close positioning tends to increase the amount of current flowing into the
grid. Ideally, no current would ever flow into the grid, because
it wastes energy and can even cause the tube to malfunction.
But in practice there's always a little grid current.
To avoid such problems, we control current flow in our
vacuum-channel transistor just as it's done in ordinary
MOSFETs, using a gate electrode that has an insulating dielectric material (silicon dioxide) separating it from the current channel. The dielectric insulator transfers the electric field where
it's needed while preventing the flow of current into the gate.
So you see, the vacuum-channel transistor isn't at all complicated. Indeed, it operates much more simply than any of
the transistor varieties that came before it.
IllustratIon by

James Provost

Inset IMaGe: nasa aMes research center

A vacuum-channel transistor closely resembles
an ordinary metal-oxide semiconductor
field-effect transistor or MOSFET [left]. In a
MOSFET, voltage applied to the gate sets up
an electric field in the semiconductor material
below. This field in turn draws charge carriers
into the channel between the source and drain
regions, allowing current to flow. No current
flows into the gate, which is insulated from
the substrate below it by a thin oxide layer.
The vacuum-channel transistor the authors
developed [right] similarly uses a thin layer of
oxide to insulate the gate from the cathode and
anode, which are sharply pointed to intensify
the electric field at the tips.

http://SPectrum.ieee.orG

Table of Contents for the Digital Edition of IEEE Spectrum July, 2014