The one-way street: diodes and rectifiers

Last time we left Russell Ohl's cracked silicon rod, p-type on one side and n-type on the other . Where they met, a flashlight made about half a volt, and current flowed far more easily one way than the other . Every phone charger, every LED, every solar panel depends on that boundary . So what's actually happening where p meets n? Picture the two sides touching. The n side is crowded with free electrons, and the p side with holes . Nothing keeps them apart, so they spread out, the way perfume spreads across a room . Electrons wander into the p side, holes wander into the n side, and near the boundary they meet and cancel out, the recombination from episode one . But every electron that leaves the n side leaves its phosphorus behind, as a fixed positive ion . Every hole that leaves the p side leaves a fixed negative boron ion . Those ions can't move; they're locked into the crystal . So a thin layer forms at the boundary with almost no free carriers in it, just locked charges, positive on one side and negative on the other. It's called the depletion region . And those locked charges are why the spreading stops . Separated plus and minus charge makes an electric field, and that field pushes electrons back toward the n side and holes back toward the p side . Every carrier that crosses exposes more charge and makes the push stronger . It settles when the push exactly balances the spreading . In typical silicon the layer ends up less than a thousandth of a millimetre thick . In energy terms, that field is a hill . An electron on the n side has to climb it to reach the p side . The height of the hill, measured in volts, is called the built-in voltage . It keeps most electrons on their own side, and most holes on theirs . So how high is it? Heat gives electrons a spread of energies, and for every small step of height, about a fortieth of an electron volt, the climbers thin out to about a third . The hill grows until it thins the electrons by exactly the ratio between the two sides . With moderate doping on both sides, there are about a trillion times more electrons on the n side than on the p side, thanks to the minority-carrier effect from last time . Thinning a crowd a trillion-fold takes about twenty-eight of those steps, roughly 0.7 volts . So silicon's hill is a bit under a volt high, and its height is tied to the gap . Dope more heavily and the hill grows, from about 0.6 towards 1 volt, but the built-in hill can never get past the 1.1 electron volt gap . Germanium's gap is smaller, so its built-in hill is only about 0.3 volts . For the same reason, a smaller gap, germanium diodes switch on at around 0.3 volts, against silicon's 0.7 . Now Ohl's flashlight makes sense . Light frees electron-hole pairs, and any pair near the junction gets pulled apart by the hill, electrons one way, holes the other . Charge piles up on each side, and that's a voltage . Ohl's rod was a solar cell, and its half volt sits below the height of the hill, as it must . Now connect a battery, positive to the p side and negative to the n side . That pushes against the built-in field, so the depletion region narrows and the hill gets lower . Because the climbers thin out exponentially with height, a slightly lower hill lets through vastly more of them . In an ideal diode, every extra sixty millivolts or so multiplies the current by ten . This is called forward bias . So a diode doesn't snap on . Its current grows smoothly, but it starts so tiny that below about half a volt you'd barely notice it . Then it shoots up, and for silicon, the useful current arrives at about 0.6 to 0.7 volts . Set a multimeter to diode test and touch a silicon diode, and you'll see something like 0.67 volts . Notice that's below the hill's full height. You don't flatten the hill; you just lower it enough . Flip the battery round and the opposite happens. The applied field adds to the built-in one, the depletion region widens, and the hill gets higher . Hardly any electrons or holes have the energy to climb it, so the current almost stops . That's reverse bias, the closed direction of the one-way street . Almost . Remember the minority carriers from last time, the few electrons in p-type and the few holes in n-type ? For them, the hill runs downhill, so they slide straight across . There are so few that the current is tiny, and it barely depends on the voltage . A common small diode is specified to leak no more than twenty-five billionths of an amp . But heat makes more of those carriers, and at 150 degrees the same diode may leak up to two thousand times more . Push the reverse voltage hard enough, though, and the dam breaks . Carriers falling down the steep hill gain so much energy that they knock new electron-hole pairs loose, and those knock loose more, an avalanche . In very thin, heavily doped junctions, electrons can instead tunnel straight through the barrier . That's called Zener breakdown, after the physicist Clarence Zener . Breakdown itself isn't damage, only the heat is, and Zener diodes use it on purpose to hold a steady voltage . So what's a one-way street good for? Wall power in Australia is alternating current, two hundred and thirty volts, fifty cycles a second . It reverses direction a hundred times a second . Put a single diode in the way and it only conducts on the half-cycles that forward bias it . Out come humps of current, all in one direction, with gaps between them. That's a half-wave rectifier . Wasting half the wave is untidy, so mains power supplies usually use four diodes in a bridge . On each swing, one pair conducts and the other pair blocks, and either way the current goes the same direction through the load . Now every half-cycle becomes a hump, a hundred humps a second . The cost is two diode drops in the path, about 1.4 volts . Humps still aren't steady, so add a capacitor across the output . It fills up at each peak and drains into the load between peaks, like a bucket under a dripping tap . What's left is a small sawtooth called ripple, and a bigger capacitor or faster humps makes it smaller . A bridge feeding a big capacitor is the conventional front end of a plug-in power supply . None of this started with junctions . In 1874 a young physicist, Ferdinand Braun, touched a metal point to a crystal of galena and found that current flowed freely in only one direction . Decades later, that fine wire, the cat's whisker, became the detector in crystal radio sets . It rectified the radio wave so the much slower sound could reach the earphones . During World War Two, millions of point-contact diodes on silicon and germanium detected radar signals . In forward bias, electrons and holes cross over and recombine . In silicon, as episode one said, that energy mostly ends up as heat . Here's why. Silicon's gap is indirect: to fall across it, an electron also has to change its momentum . Light carries plenty of energy but almost no momentum, so a photon can't settle the bill on its own. It needs a crystal vibration to supply the kick at the same moment, which is rare, so the energy usually leaks away as vibration instead: heat . In materials like gallium arsenide, the gap is direct. No kick is needed, so the energy is more likely to leave as a photon . That's a light-emitting diode . Each photon carries about the gap's energy, so the gap sets the colour . A gap of about two electron volts gives red light . Visible light runs from about 1.8 electron volts at the red end to 3.1 at the violet . Gallium arsenide's 1.4 electron volt gap gives infrared . And because you have to lift each electron by that energy, an LED's turn-on voltage, in volts, roughly matches its photon energy in electron volts: around two volts for red, closer to three for blue . Red LEDs arrived around 1960, but blue remained a challenge for three decades . The answer was gallium nitride, but no one could grow good enough crystals, and making p-type gallium nitride was described as virtually impossible . Doping, last episode's trick, was a big part of the wall . Isamu Akasaki and Hiroshi Amano cracked it, and showed a bright blue diode in 1992 . That same year Shuji Nakamura found his own, simpler route to p-type gallium nitride, by heating it . That won them the 2014 Nobel Prize in Physics, for efficient blue LEDs . Shine blue on a phosphor and you get white light . So a diode is a one-way street, set by a hill of locked charges less than a micrometre wide . But an ordinary diode can't amplify. Nothing lets a small signal open or close it . The wartime crystal work pointed the way . In 1947, at Bell Labs, Walter Brattain used a slab of purified germanium to build something new . How did two wires on a crystal become a valve a tiny signal could control? That's next time.