# Why Diamond Transistors Are So Hard

## Метаданные

- **Канал:** Asianometry
- **YouTube:** https://www.youtube.com/watch?v=NLmd5vL0zmk

## Содержание

### [0:00](https://www.youtube.com/watch?v=NLmd5vL0zmk) Segment 1 (00:00 - 05:00)

In the movie “Diamonds are Forever”, James Covalent Bond is chasing some missing South African diamonds. He discovers that his nemesis Blofeld is cooking up an evil plot to use those diamonds for a big space laser that he intended to auction off. I find the plot impractical. Blofeld should have instead concentrated his criminal organization’s resources on developing diamond transistors! Had he succeeded, he might have reaped far greater profits than with some silly space laser. In this video, we look at a very hard gate. ## Silicon Sucks There are many good things about silicon that make it the most used semiconductor, but it is far from perfect. First is silicon's bandgap, or the amount of energy needed to excite the element's electrons from a resting to energized state. The wider the bandgap, the more energy needed to excite those electrons. Silicon's bandgap is just 1. 12 electron volts, which is relatively narrow. The upside of such a narrow bandgap is that we do not need such a high voltage to switch a silicon transistor. But it also means that heat can cause that transistor to inadvertently switch. With more heat, charge carriers can punch through the gate and generate a small leakage current. This current dissipates more heat, which worsens the leakage and makes even more heat. A negative feedback loop. This is why we need to be careful about how many silicon transistors within a given area are turned on during operations. Silicon also has a relatively low breakdown field. If exposed to a strong enough electric field, the silicon might lose its insulating properties - causing large currents to burst through. Think of water going through a pipe. If the water pressure gets too high, the pipe bursts. So silicon isn’t good to use in situations of high heat and strong electric fields. Like for example, the inverters in EVs that convert DC to AC power for the EV's electric motors to use. These have to withstand high voltages in the range of 400 volts or more. Or RF amplifiers used to amplify high frequency signals of anywhere from 1-6 gigahertz for 5G base stations. Yet higher frequencies for millimeterWave. ## Going Big Bandgap So for these, people have explored what are called wider bandgap materials. Called as such because their bandgaps are wider than silicon’s. There are a number of these known in the industry. Let me go through a few of them. Indium phosphide and gallium arsenide have somewhat higher bandgaps - 1. 35 and 1. 42 electron volts respectively. The bandgaps of silicon carbide and gallium nitride are even wider at 3. 26 and 3. 4 electron volts. But we need to consider more than just the raw bandgap. We must also consider economics, manufacturability and other things. For instance, the bandgaps of Indium phosphide and gallium arsenide are not that much higher than silicon's. Moreover, they are both toxic to humans and on the rarer side. Silicon carbide has relatively low carrier mobility. Meaning that charge carriers like electron and electron holes do not travel very fast through them. This is partly because silicon carbide’s oxide is silicon dioxide. The interactions between the silicon, oxygen, and carbon atoms create defects called traps that can impede the charge carriers' movement. Gallium nitride is nice and widely used in the LED and optical industries. However it gets unstable at high switching frequencies - leading to unpredictable behavior. And when exposed to the high heat and powerful electric fields, it falls apart faster than silicon. So while industries worth billions of dollars have emerged around these high-bandgap materials, research continues for other materials with possibly more to offer. In the late 1980s and early 1990s, people have explored what we call ultra-wide bandgap materials like aluminium nitride, gallium oxide, and of course diamond. ## Diamond Rocks. And yes, it is also a semiconductor. Diamond's bandgap is absolutely massive compared to silicon's: 5. 5 electron volts compared to 1. 12. And its breakdown field of 10 megavolts per centimeter can be up to 33 times higher than silicon's. And its charge carrier mobility is good, especially very pure, "electronic-grade" diamond. Mobilities have been measured at 3 times that of silicon for electrons and up to 9 times for electron holes. Though there are nuances to consider.

### [5:00](https://www.youtube.com/watch?v=NLmd5vL0zmk&t=300s) Segment 2 (05:00 - 10:00)

Perhaps most useful of all is diamond's unparalleled thermal conductivity. Thanks to strong covalent bonds which allow heat-carrying phonons to propagate without interference, Diamond's thermal conductivity is about 2,200 watts per meter-Kelvin. That is amongst the highest of all the known semiconductors. It blows away silicon's 150 watts per meter-Kelvin and exceeds copper's 400 silicon carbide's 370. So the advantages of this in a field effect transistor are pretty dead obvious. People are already experimenting with high electron mobility transistors made with diamond heat sinks beneath them. A diamond transistor has that unparalleled heat sink built right in. This means that we can dissipate far more power through diamond at the same temperature than with silicon and other materials. Enabling very high densities of active diamond transistors or less cooling. There are other things too. Diamond is also quite resistant to radiation, because of the tightness of the bonds. Such rad-hard qualities makes it suitable for extreme conditions like space travel. Perhaps the best thing is how diamond combines all of these properties - high thermal conductivity, high breakdown voltage, radiation-hardness, and high carrier mobility - in a single package. No other material is like this, and that makes it perfect for certain high-performance niches within the $55-60 billion power electronics market. Okay that is all the good stuff. The reasons to care. Now here comes the catch. We use diamonds in industry, taking advantage of its hardness or optical properties. But we do not have diamond transistors because those are very hard to make. Ha ha. Humans first fabricated a diamond-based semiconducting device in the late 1950s, when M. Bell used naturally P-typed diamond to create power diodes. But progress then stalled for several decades. ## Synthesis The first problem with diamond is making a lot of it, cheaply. We can produce 12-inch wide silicon wafers at scale by pulling crystals out of a molten silicon melt and cutting them. The Czochralski method, yes. Moreover, those wafers are single-crystal - meaning that we have a continuous, unbroken crystal lattice of silicon. Crystal defects - defined as dislocations of the atoms, voids in the structure, or disruptions in the planes - can interfere with the charge carriers and impede the device's performance. Czochralski is well known and works. But it cannot be applied to diamond because we cannot produce a vat of diamond "melt". Unless you put it under untenable high pressure conditions, molten diamond turns into graphite, which we cannot pull. So back in the 1950s, the only way to synthetically produce diamond was by replicating the high pressure and high temperature conditions of the earth's insides, called the HPHT method. HPHT involves putting carbon under a massive piston made from super-hard alloy - subjecting it to conditions of 5. 5 gigapascals - about as much pressure as my dad exerts during SATs - at 1,300 to 1,400 degrees Celsius. HPHT-produced diamond crystals have few lattice defects, which is good. But there are three problems. The first being that the diamonds are too small to do much with. And two, HPHT introduces unwanted impurities into the diamond from its surroundings - like boron, cobalt, and nitrogen. Ergo why so many HPHT diamonds are yellow, they have within them 10-100 parts per million nitrogen impurities. The third issue is that the HPHT method itself is expensive and high-maintenance. Thus, a new method has emerged: A form of chemical vapor deposition called Microwave Plasma CVD. This involves injecting a mix of precursor gases - methane about 97% diluted in hydrogen - into a chamber. Powerful microwaves then strip the gas atoms of their electrons, forming a plasma. The methane decomposes into carbon and hydrogen. The free carbon atoms then form chemical reactions on the surface of a seed diamond, building a diamond layer vertically on top of it. The fact that we use a seed diamond makes this MPCVD process a homoepitaxy - epitaxy on the same. We are essentially extending the seed diamond's existing crystal structure. Commercialized in the 1980s, MPCVD is the dominant method of producing gem-quality

### [10:00](https://www.youtube.com/watch?v=NLmd5vL0zmk&t=600s) Segment 3 (10:00 - 15:00)

diamond at scale. I once visited a MPCVD laboratory in Taichung - trying to produce electronic-grade synthetic diamond - and thought it was pretty cool technology. ## No Big Diamond Wafers However, more work needs to be done. The first problem is getting the MPCVD growth rate up without compromising on the quality of the grown crystal. The microwave plasmas are inherently non-uniform, and their flow over the growing diamond surface changes due to subtle things like even the presence of gaps or edges on the substrate-carrier. So areas of the wafer experience different conditions - leading to uneven quality. And carbon nucleates well on things, which makes diamond particularly apt to form as many small crystals separated by grain boundaries. We call this polycrystalline diamond, and it is littered with grain boundaries and other boundary defects that impede the travel of charge carriers and negate the material's inherent electronic advantages. A critical issue is the lack of large enough single-crystal seed diamonds. Most seeds are produced with HPHT, which grants us the single-crystal structure we want but inherently limits their size. And the complex dynamics of the gas precursors inside the chamber mean that single-crystal diamond growth via MPCVD is slow: 75 micrometers per hour, without nitrogen. About twice that with nitrogen, though nitrogen doping only goes so far because too much can lead to deficiencies like what are called nitrogen vacancies. I recall from my visit to the MPCVD lab them saying that a 2-carat gem-quality diamond might take about 2 weeks to produce. The target width to go commercial is about 4-inches with a killer defect rate of 1 per 10 square centimeters. But single-crystal wafers just 10 millimeters wide can cost up to 10,000 times of equivalent silicon. We are far from ready. And by the way. Interestingly enough, pure, electronic-grade diamond wafers are black colored. ## Trying to Get Big Improving the growth rates of the MPCVD method is tough. That is because it is a complex reaction, and isolating the impact of one factor requires a lot of runs. Though it can be improved by running multiple tests at the same time. Getting to 4 inches - or even just two - requires that we expand beyond the seeds' physical dimensions. Tried methods including using CVD to grow diamond laterally around the seed like a mold. This increases our size by 1-2 times in a single run. Good, though sadly still far short of our goal. Another idea has been to grow single-crystal diamond on the sides of the seeds. This is interesting, but generally does not work well because the seeds' sides are not that big. The method that I saw but did not recognize back in Taichung - is the mosaic method. We take multiple seeds and put them together in a single "mosaic", like a window pane. The diamond grows over the mosaic, fusing together to create a larger piece that we can lift off the mosaic. The quality is quite good since we are using single-crystal diamond seeds. A side-channel has been trying to make non-diamond seeds work - something called heteroepitaxy. If pure diamond does not work, then let's try something else entirely. The best candidate seems to be - surprisingly enough - Iridium. And not just a slab of it, very special iridium layers deposited on top of metal oxide and silicon substrates for cost and thermal reasons. Consensus is that this is best way to eventually get to 4-inch single-crystal wafers, but issues like defects persist. ## Doping Another major issue is doping. This is an essential part of the transistor production process. Without it, the material is electrically inert. In the case of silicon, we can produce electrically active, N or P-type silicon by injecting impurity atoms like phosphorus or boron into the material. Nowadays this is done with a method called ion implantation. We want to be able to do the same for diamond, and have it work just like a regular transistor, but it does not. If we want to imagine the transistor as transporting water from the source to the drain through the channel, we need to make sure that there is water in that source. With silicon, there is plenty of water - meaning charge carriers - available. The boron and phosphorous atoms inside the silicon are willing to give up their charge carriers when given just a small amount of thermal energy. At about 0. 045 electron volts for both boron and phosphorous, silicon's ionization energy threshold is low enough - or shallow enough, if we are to use industry lingo - that room temperature works.

### [15:00](https://www.youtube.com/watch?v=NLmd5vL0zmk&t=900s) Segment 4 (15:00 - 20:00)

But the threshold for doped diamond is multiples higher. Err. The correct lingo is to say, "deeper". Boron's is 0. 36 electron volts and phosphorous's is even deeper at 0. 57. Thus at room temperature, only a small percentage of the boron atoms give up their charge carriers. And so unless the diamond transistor is subjected to high temperatures, you get too few charge carriers for the transistor to run efficiently. That's great if we want to run our diamond transistor computer on the surface of Venus - and it legit might perform really well there - but we are talking about Earth. A few alternative dopants exist but the situation there is even worse. Practically speaking, it has got to be boron and phosphorous. We have to figure out a way around. ## Struggling with Boron Let us start with boron, which is used to create P-type diamond. First, they tried to overcome this low efficiency with brute force. If there is not enough boron atoms giving up their electron holes? Then let us just add more into the diamond lattice! Unfortunately, too much doping causes the diamond to act less like a semiconductor and more like a metal. They use the phrase "degenerate semiconductor", which is evocative. So they then tried something called delta-doping, which is where you create a very thin layer of heavily doped boron diamond sandwiched between two layers of normal diamond. But scientists have struggled to consistently do this, with methods like ion implantation known to cause serious damage to the diamond lattice. Note how efforts centered on boron and P-type diamond. In order to create a silicon CMOS-style arrangement, then we need the N-type diamond transistor too. Unfortunately the challenges there are even more daunting, since the dopant phosphorous activation energy is a magnitude deeper than boron. And the only other option is nitrogen - whose activation energy is a freakin' 1. 7 electron volts. ## Hydrogen-Termination The doping issues around diamond were so bad that it derailed technological progress in the field, with investment and effort going to silicon carbide and others. But in 1989, a chance experiment with CVD diamond discovered that if we terminated the dangling carbon bonds on the diamond surface, we can produce a conductive surface a trillion times higher than expected. What does this terminology mean? The carbon atoms on the diamond surface has its bonds left dangling out there. These dangling bonds are unstable and can grab passing charge carriers like electrons and interfere with their travel. Kind of like how anemones grab things in the ocean to eat them. To terminate these dangling bonds means to attach an atom to them to "satisfy" them. The result of this termination is a very thin layer of diamond that electron holes can travel through - giving us P-type conductive diamond while sidestepping the whole boron doping problem. It triggered a lot of research in the 1990s to confirm. Upon confirmation that it was indeed the hydrogen termination doing this, we were able to finally make switching devices. ## The MESFETs The first serious field effect transistor with diamond emerged in 1994. They demonstrated a MESFET - a metal semiconductor field effect transistor. Meaning a type of layered transistor with the source, drain, and gate all made from a metal. We start with the substrate, made from un-doped HPHT single-crystal diamond. A layer of diamond was grown on top of that with MPCVD. Hydrogen-termination via microwave-enabled hydrogen plasma is then used to make that layer P-type conductive, moving electron holes. Then two pieces of gold is deposited on top to serve as the source and drain. In between, we have an aluminium gate. You end up with a working MESFET, albeit rather quirkily and with slower charge carriers. Over the years, more MESFETs and variants have been made with modifications added for improvement. Like, hydrogen-termination being replaced with oxygen-termination, which is more stable in air. But major challenges remain. For instance, the transistors in modern power electronics are vertical - with the source and drain on opposite sides of the wafer - because it lets a power current flow through the bulk of the device for heat, efficiency, and voltage reasons. In the diamond MESFET, power flows only through the thin oxygen or hydrogen terminated layer at the surface of the device. Changing this means fixing the doping problem again. So yes, we are rather far from diamond transistors that can compete with silicon carbide or GaN power electronics. I should note that we do not have to achieve a full diamond transistor to get its benefits. I briefly mentioned earlier the Gate-after-diamond method, where a

### [20:00](https://www.youtube.com/watch?v=NLmd5vL0zmk&t=1200s) Segment 5 (20:00 - 21:00)

layer of diamond is laid down to serve as a heat sink for certain transistors that run hot, like High Electron Mobility Transistors, or HEMTs. Maybe that is the future. ## Conclusion The idea of a diamond transistor is fun. And there is real reason to say that diamond can produce the ultimate hard-core power electronics device. But as it so often turns out in semiconductors, superior material properties is just a small part of the equation. It needs to be manufacturable, scalable and economic. Hopefully with existing hardware. Even after overcoming diamond's existing size and doping shortcomings, it might take years more to bring something to market. Like with silicon carbide, where 4-inch wafers were ready by 1999 but no devices came out until CREE's in 2011. Nevertheless, the market’s growing needs and the material’s own inherent allure will continue to drive more research in the years to come.

---
*Источник: https://ekstraktznaniy.ru/video/26163*