(Intro)

There's something that's arguably the foundation of the modern world but that most people have probably never seen before.  Behold the humble transistor.  It may look like Dr. Evil's latest gadget for world domination, but a transistor is just an electrical switch that can be turned on or off by another circuit.  That's exactly what you need for the digital logic that computers run on.

In the 1950s and 60s, transistors took us from giant computers made of fiddly vacuum tubes to slightly less giant computers made of silicon and ever since, they've kept getting smaller and smaller and smaller.  Today, the amazing shrinking transistor continues to make our devices smaller, more powerful, and more efficient, but transistors are running out of room to shrink, at least, the ones made of silicon are.

So the question becomes, once we can't improve silicon anymore, what's next?  Some experts are looking to two-dimensional materials which consist of sheets one atom thick.  For transistors, probably the most promising material is a silvery, greasy feeling substance called molybdenum disulfide, or MoS2.  The 2D form of MoS2 is a sheet of molybdenum atoms sandwiched between two layers of sulfur and it could help with many of the problems we're facing with silicon, leading to faster, cheaper, and lower powered electronics, but there are some serious manufacturing challenges to solve first.

Most transistors used in computers have four basic components.  As an electrical switch, a transistor needs places for the electricity to come from and to flow to.  Those electrical contacts are called the source and drain.  Electricity flows from the source to the drain through a region called the channel, which is usually a short ridge of silicon.  It's like the electron at the source is a marble at the top of a track.  It wants to roll down toward the drain.  The channel is the track, the path the ball would take as it rolls down.  Just above the channel is the final component, the gate, which switches the transistor on or off.  When the gate is off, its electric field creates a barrier in the channel.  The field pushes electrons away, leaving a sort of dead zone that electrons don't have the energy to pass through, so the channel acts like a track with a towel tossed over it.  

That marble is going nowhere, but when the gate is turned on, its electric field allows electrons to flow across the channel.  It's like lifting the towel off to let the ball through.  When it comes to computer chips, the length of the channel is super important.  The shorter it is, the smaller you can make the whole transistor, which gets you tons of benefits.  For one thing, electrons have less distance to travel, so information flows around faster.  Tiny transistors also let you build more sophisticated circuits. 

Modern chips need very complex circuitry for all their speed-up tricks, like guessing in advance what decisions a program might make next.  That means squeezing more transistors onto each chip.  Another advantage is that smaller circuits don't push back as hard against changes in current, so turning them on and off draws less power, so tinier transistors let you improve speed, complexity, and efficiency, but like all good things, at some point, transistor shrinkage must come to an end.

The obvious limit is the size of an atom. You can't possibly make a channel shorter than that, but with silicon, problems known as short channel effects kick in well before you get down to that size.  One problem is what's called drain-induced barrier lowering.  When the channel is too short, the drain's electric field wipes out the barrier that the gate is supposed to create.  It's as though the electron is a marble not sitting on a track but resting on a squishy block of foam a few inches away from a heavy brick.  If you put the brick far from the marble, then the foam in between will keep the marble from rolling down.  It will only fall toward the brick if you compress the foam, just like a transistor's gate only lets electricity reach the drain when it's open.  But if you push the brick too close, the depression around it reaches all the way to the marble.  That marble's rolling whether you touch the foam or not.

The same thing happens with the drain's electric field.  Another issue is surface scattering, where the electrons in the channel get distracted by the charge on the gate and bounce off tiny imperfections in the channel's surface, and then of course there's quantum mechanics, the branch of physics that deals with the super tiny.

In quantum mechanics, things don't always work the way you would expect based on the everyday scale we're used to.  Quantum mechanics says an electron's location is sort of a smear.  If the channel is short, it's easy for them to tunnel where they just show up on the other side even when the transistor is supposed to be off.  One solution is to make the channel not just super short but super thin, like by using MOS2 . Since the molybdenum is only an atom thick, the gate's electrical field can control the channel better, like how it's easier to squeeze off a centimeter diameter hose than a 15 cm one, and because MOS2 is a perfect sheet, there's no surface roughness or even interior atoms for electrons to bump into.  They just glide along.

MOS2 also has some bonus features unrelated to thickness.  For instance, getting electrons moving in it at all takes more of a kick, so electrons tunnel across the channel or leak via barrier lowering so rarely it's hard to even measure . Of course, it's not all rainbows and unicorns.  Short channel effects do still show up in MOS2.  They just aren't as strong.   Also, short channels aren't the only size limitation.  Chips with tiny transistors have to pack a lot of power-hungry circuitry into a small space, making them hard to cool.

An even bigger problem is the copper connections that wire transistors together.  If we miniaturize those further, forcing electrons through them starts to take too much power.  Then there are the challenges of manufacturing the MOS2 itself.  Finding ways to grow and manipulate MOS2 with the same cost and versatility as silicon would take enormous investment, but people are working on solutions to all of these problems and if electronics are going to keep improving the way we've come to expect, we may need MOS2 or its two-dimensional relatives.

So despite the challenges in the world of semi-conductors, it's an exciting time to be flat.

Thanks for watching this episode of SciShow and thanks especially to our community on Patreon.  It's because of you that we're able to do deep dives into technical but super important topics like this one.  

If you want to help us keep making episodes like this, just check out Patreon.com/SciShow.  

(Endscreen)