Why it was so difficult to make the blue LED | From Red and Green to Blue The Epic Journey of LED Innovation

LEDs do not get their color from their plastic covers and this makes sense, as seen in this clear LED glowing the same red hue. The light color actually originates from the electronics themselves with the casing being a distinguishing factor between various LEDs.

From Red and Green to Blue: The Epic Journey of LED Innovation

In the year of 1962, a general Electric engineer by the name of Nick Holonyak conceived the first visible LED, emitting a faint red glow. After some time, engineers from Monsanto developed a green LED. Nonetheless, for years only these two colors existed limiting the use of LEDs to applications like indicators, calculators, and watches. Imagine if only blue was feasible to produce, then we could blend red, green, and blue to produce white, as well as every other shade possible, ushering LEDs into all types of lighting across the globe. It was a dream worth pursuing, considering the potential of LEDs in various devices from light bulbs, phones, and computers to TVs and billboards. Unfortunately, blue seemed like an impossible endeavor at that time. Throughout the 1960s, leading electronics companies worldwide, from IBM to GE and Bell Labs, aggressively competed to engineer the elusive blue LED, realizing the immense value it could bring. Despite the collective efforts of countless researchers, no significant breakthrough was made. As time went on, hope for incorporating LEDs in lighting applications diminished with each passing year. 10 years post Holonyak's groundbreaking creation, the potential for LEDs in lighting seemed bleak, if not futile. As per an executive at Monsanto, LEDs were unlikely to replace standard kitchen lights. They were more suitable for appliances, car dashboards, and stereo systems providing a way to indicate whether the stereo was on. This stance might have remained unchanged until today; however, one bold engineer defied the status quo and made three groundbreaking advancements, culminating in the world's first blue LED.

How does an LED work?

When you have an isolated atom, each electron in that atom occupies a discreet energy level. You can think of these energy levels like individual seats from a hockey stadium, and all atoms of the same element, when they are far apart from each other, have identical available energy levels. But when bring multiple atoms together to form a solid, something interesting happens. The outermost electrons now feel the pole, not only of their own nucleus but of all the other nuclei as well. As a result, their energy levels shift. So instead of being identical, they become a series of closely spaced, but separate energy levels. An energy band. Electrons with the highest energy band are known as the valence band, and the next higher energy band is called the conduction band. You can think of it like the balcony level. In conductors, the valence band is only partially occupied. This means with a little bit of thermal energy, electrons can jump into nearby unfilled seats, and if an electric field is applied, they can jump from one unfilled seat to the next and conduct current through the material. In insulators, the valence band is full, and the energy difference between the valance band and the conduction bond is too large. So when an electric field is applied, no electrons can move. There are no vacant seats within the valence band and the band gap is too big for electrons to jump into the conduction band which brings us to semiconductors. Semiconductors are similar to insulators, except the band gap is much smaller. This means at room temperature, a few electrons will have sufficient energy to jump into the conduction band, and now they can easily access nearby empty seats and conduct current. Not only that, but the empty seats they left behind in the valence band can also move. Indeed, it is the adjacent electrons jumping into those empty seats. But if you look from afar, it appears as if the vacant seat or void is in motion, resembling a positive charge moving in the opposite direction to the electrons within the conduction band. By themselves, pure semiconductors are not that useful. To make them way more very functional, you have to add impurity atoms into the lattice. This is known as doping. For example, in silicon, you can add a small amount of phosphorus atoms. The phosphorus atom is similar to a silicon atom, that why it is easily fits into the lattice, but it brings with it one extra valence electron. This electron exists at a donor level just beneath the conduction band. So with little amount of thermal energy, all these electrons can jump into the conduction band and conduct current. Since most of the charges that can move in this type of semiconductor are negative electrons, this sort of semiconductor is called n-type, n for negative, but It is important to highlight that the semiconductor remains neutral. It's just that most of the moving(mobile) charge carriers are negative. They're electrons. So there is also another type of semiconductor where most of the mobile charge carriers are positive, and it's called p-type. To make p-type silicon, you need to add a small number of atoms of, boron. Boron fits into the lattice but brings with it one fewer valence electrons than silicon. So it creates an empty acceptor level just above the valence band. And with a bit of thermal energy, electrons can jump out of the valence band, leaving behind holes. It is these positive holes that are mostly responsible for carrying current in the p-type semiconductor. Again, the material overall is uncharged, it's just that most of the mobile charge carriers are positive holes. Where things get interesting is when you put a piece of p-type and n-type together. In the p-type semiconductor without even connecting to a circuit, some electrons will diffuse from n to p and fall into the holes. This makes the p-type a slight negatively charged, and the n-type a slight positively charged. So there is now an electric field inside an inert piece of material. Electrons keep diffusing until the electric field becomes so large, that it prevents them from crossing over And now we have established the depletion region, an area depleted of mobile charge carriers. There are no electrons in the conduction band and no holes in the valence band. Suppose you connect a battery the wrong way to this diode. In that case, it simply expands the depletion region until its electric field perfectly opposes that of the battery and no current flows. But if you flip the polarity of the battery, then the depletion region shrinks, the electric field decreases, and electrons can flow from n to p. When an electron falls from the conduction band into a hole in the valence band, that band gap energy can be emitted as a photon. The energy change of the electron is emitted as light, and this is how a light-emitting diode might work. The size of the band gap determines the color of the emitted photon. In pure silicon, the band gap is only 1.1 electron volts. So the photon released isn't visible range, it's infrared light. These LEDs are actually used in remote controls for your TV, and you can capture them on camera. Moving up the spectrum, you can see why the first visible light LEDs were red and then green, and why blue was so hard. A photon of blue light requires more energy, and therefore a larger band gap.

Overcoming Obstacles in Crystal Research

By the 1980s, after hundreds of millions of dollars had been spent hunting for suitable material, every electronics company had come up empty-handed. However, researchers had at least figured out the first critical requirement, high-quality crystal. No matter what material you used to make blue LED, it required a near-perfect crystal structure. Any defect present in the crystal lattice disrupts the flow of electrons. So instead of emitting their energy in the form of visible light, it is instead dissipated as heat. So the first step in Nakamura's(shuji Nakamura was a researcher at a small Japanese chemical company named Nichia) proposal to Ogawa was to disappear to Florida. He knew an old colleague there whose lab was beginning to use a new crystal-making technology called metal-organic Chemical Vapor Deposition, or MOCVD. An MOCVD reactor is essentially a giant oven for the production of clean crystals. It works by injecting vapor molecules of your crystal into a hot chamber, where they react with a base material called a substrate to form layers. The substrate lattice must match the crystal lattice being built on top of it to create a stable, smooth crystal. This is a precise art. The crystal layers often need to be as thin as just a couple of atoms. Nakamura joined the lab for a year to master MOCVD. But his time there was miserable. He wasn't allowed to use the working MOCVD, so he spent 10 of his 12 months assembling a new system, almost from scratch. Even worse, his lab mates shunned him because Nakamura didn't have a doctorate, or...academic papers to his name. His lab mates, all PhD researchers, dismissed him as a lowly technician. This experience fueled him. Nakamura wrote, "I feel resentful when people look down on me. I developed more fighting spirit. I would not allow myself to be beaten by such people." He returned to Japan in 1989 with two things in hand. One, an order for a brand new MOCVD reactor for Nichia, and two, a fervent desire to get his PhD. At that time in Japan, you could earn a PhD without having to go to university, simply by publishing five papers. Nakamura had always known his chances of inventing the blue LED were low. But now he had a backup plan. Even if he didn't succeed, he could at least get his PhD. But now the question was with MOCVD under his belt, which material should he research? By this time, scientists had narrowed the options down to two main candidates, zinc selenide, and gallium nitride. These were both semiconductors with band gaps, theoretically, in the blue light range. Zinc selenide was the far more promising choice. When grown in an MOCVD reactor, it had only a .3% lattice mismatch with its substrate, gallium arsenide. Therefore, zinc selenide crystal had about a thousand defects per square centimeter, within the upper limit for LED functioning. Its only issue was that while scientists had figured out multiple different ways to create n-type zinc selenide, no one knew how to create p-type. In contrast, gallium nitride had been forsaken by almost everybody for three reasons. First one, it was much harder to make a high-quality crystal. The best substrate for growing gallium nitride was sapphire, but its lattice mismatch was 16%. This resulted in higher defects, over 10 billion per square centimeter. As the research into blue LEDs progressed, it became evident that the main challenges revolved around gallium nitride. Distinct from silicon-based zinc selenide, the production of n-type gallium nitride was relatively straightforward. However, the elusive p-type gallium nitride posed a significant obstacle to commercial viability. For a blue LED to be considered commercially feasible, it would require a total light output power exceeding a thousand microwatts – a two-magnitude leap compared to existing prototypes.

The competition between zinc selenide and gallium nitride was fierce. With most researchers leaning towards zinc selenide due to its popularity, Shuji Nakamura made a pivotal decision to focus on gallium nitride. Despite the historical setback in 1972 – when RCA engineer Herbert Maruska's gallium nitride LED fell short in performance – Nakamura saw potential. The dim and inefficient LED may have deterred RCA from further exploration, but Nakamura was undeterred. At a pivotal physics conference in Japan, the preference for zinc selenide over gallium nitride was evident. With over 500 attendees in the former's talks contrasting with a mere five in the latter, the odds seemed stacked against gallium nitride. However, among those five attendees were the world-renowned experts: Dr. Isamu Akasaki and Dr. Hiroshi Amano from Nagoya University. Their breakthrough involving aluminum nitride buffer layers set a new course for gallium nitride research. While Nakamura struggled to grow gallium nitride crystals in his MOCVD reactor at Nichia, his persistence paid off. Long hours spent on refining the reactor, experimenting, and rebuilding parts eventually led to progress. Amidst this tireless pursuit, Nakamura adhered to a strict routine, working relentlessly without breaks – except for New Year's Day, a cherished holiday in Japan. The road to revolutionizing blue LEDs through gallium nitride was rife with challenges, but Nakamura's unwavering dedication laid the foundation for a breakthrough that would change the industry forever. After a year and a half of continuous work, he comes into the lab on a winter day in late 1990. In the morning, he went around the lab, grew a gallium nitride sample in the afternoon, and tested it, But this time, the electron mobility was four times higher than any gallium nitride grown directly on sapphire before. He called this the most exciting day of his life. His trick was to add an additional nozzle to the MOCVD reactor that gallium nitride reactant gases had been rising in the hot chamber, mixing in the air to form a powdery waste. But the second nozzle released a downward stream of inert gas pinning the first flow to the substrate forming a uniform crystal. For years, scientists avoided adding a second stream to MOCVD because they thought that it would only introduce more turbulence. Nakamura used a special nozzle to ensure that the streams maintained their laminar flow even when combined. He named his invention the two-flow reactor. Now, he was ready to take on Akazaki and Amano, but instead of copying their aluminum nitride buffer layer, his two-flow design allowed him to make gallium nitride so smooth and stable, that it could be used as a buffer layer on the sapphire substrate itself. Yielding an even cleaner crystal of gallium nitride on top, without the issues of aluminum. Nakamura now had the highest quality gallium nitride crystals ever made. But just as he started, things took a wrong turn. While happened to be in Florida, Nobuo Ogawa had stepped back from Nichia to become chairman. In his day, Nobuo had been a risk-taking scientist, designed the company's first products. This is why he supported Nakamura's lofty plans all this time. But in his place, his son-in-law, Eji Ogawa, became CEO of the company, and the younger Ogawa had a much stricter perspective. One Nichia client said, "He has a mind of steel, and he remembered everything." In 1990, an executive at Matsushita, an LED manufacturer, and Nichia's biggest customer, visited the company to give a talk on blue LEDs. In it, he claimed zinc selenide was the way forward, declaring "gallium nitride has no future." That very same day, Nakamura received a note from Eji to stop work on gallium nitride immediately. Eji had never supported the research and wanted to end what he saw as a colossal waste. But Nakamura crumpled up the note and threw it away, and he did so again, and again when a succession of similar notes and phone calls came from company management. Out of spite, he published his work on the two-flow reactor without Nichia's knowledge. It was his first paper. One down, four to go. With crystal formation settled, he turned to the second obstacle, creating p-type gallium nitride. Here Akazaki and Amano had again beat him to the punch. The gallium nitride sample doped with magnesium that they created initially did not perform the expected p-type behavior. However, after exposing it to an electron beam, it did behave as a p-type, the world's first p-type gallium nitride, after 20 years of trying. The catch was that no one knew why it worked. And the process of irradiating each crystal with electrons was too slow for mass production. At first, Nakamura copied Akazaki and Amano's approach, but he suspected the beam of electrons was overkill. Maybe all the crystal needed was energy. So he tried heating magnesium-doped gallium nitride to 400 degrees Celsius in a process known as annealing. The result is a completely p-type sample. This worked even better than the shallow electron beam, which only made the surfaces of the samples p-type, simply heating things up was a quick scalable process. His work also revealed why the p-type had been so difficult. To make gallium nitride with MOCVD, you supply the nitrogen from ammonia, ammonia also contains hydrogen. Where there should have been holes in the magnesium-doped gallium nitride, these hydrogen atoms were sneaking in and bonding with the magnesium, plugging all the holes. By adding energy into the system, the hydrogen was released from the material, thereby freeing up the holes once again. Nakamura had now obtained all the necessary components to develop a prototype blue LED, which he subsequently presented at a workshop held in St. Louis in 1992 and received a standing ovation, he was beginning to make a name for himself, but even though he had created the best prototype to date, it was more of a blue-violet color, and still extremely inefficient, with light output power of just 42 microwatts, well below the 1000 microwatt threshold for practical use. At Nichia, the new CEO's patience had run out. Eji sent written orders to Nakamura to stop tinkering and turn whatever he had into a product. His job was at risk, but in Nakamura's own words, "I kept ignoring his order. I had been successful because I didn't listen to company orders and trusted my own judgment." At this point, he only had the third hurdle left, getting his blue LED to a light output power of a thousand microwatts. A known trick to increase the performance of LEDs was to create a well, a thin layer of material at the p-n junction called an active layer that shrinks the band gap just a bit. This encourages more electrons to fall from the n-type conduction band into the holes in the p-type valence band. The best active layer for gallium nitride was already known to be indium gallium nitride, which would not only make the band gap easier to cross but also narrow it just the right amount to bring its blue-violet gap down to true blue. This time, Akasaki and Amano didn't scoop Nakamura. They were stuck trying to grow indium gallium nitride in the first place. Amano recalled, "It was generally said that gallium nitride and indium nitride would not mix, like water and oil." But Nakamura had an advantage, his ability to customize his MOCVD reactor. This allowed him to use brute force, adjusting the reactor to pump as much indium as he could onto the gallium nitride, in hopes that at least some would stick. To his surprise, the method worked, giving him a clean indium gallium nitride crystal. He quickly integrated this active layer into his LED, but the well worked a little too well and overflowed with electrons, leaking them back into the gallium nitride layers. Unphased, within a few months, Nakamura had fixed this too by creating the opposite of a well, a hill. He returned to his reactor one more time to make aluminum gallium nitride, a compound possibly with a larger band gap that could block electrons from escaping the well once inside. The structure of the blue LED had become far more complex than anyone could have imagined, but it was complete. 

By 1992, Shuji Nakamura had this. After 30 years of searching by many scientists, Nakamura had done it. He had created a glorious bright blue LED that could even be seen in daylight. It had a light output power of 1,500 microwatts and emitted a perfect blue at exactly 450 nanometers. It was over 100 times brighter than the previous pseudo-blue LEDs in the market. Nakamura wrote, "I felt like I had reached the top of Mount Fuji." Nichia called a press conference in Tokyo to announce the world's first true blue LED. The electronics industry was stunned. After few years later Nakamura published over 15 papers and he finally received his doctorate in engineering.

From red and green to white LED

The transition from blue to white was achieved in 1996 by covering the LED with a yellow phosphor. This chemical absorbs the blue light and re-radiates it in a broad spectrum across the visible range. Soon enough, Nichia was selling the world's first white LED. From house lights to streetlights. The result is a lighting revolution. In 2010, just 1% of residential lighting sales in the world were LED. In 2022, it was over half. Experts estimate that within the next 10 years, nearly all lighting sales will be LED. The energy savings will be enormous. Lighting accounts for 5% of all carbon emissions. Switching entirely to LED lights has the potential to reduce carbon dioxide emissions by approximately 1.4 billion tons, which is equivalent to removing nearly half of the world's cars from the roads. In 2014, Nakamura, Akasaki, and Amano were awarded the Nobel Prize in physics for creating the blue LED.

Summary

Blue LEDs, utilizing gallium nitride (GaN) with InGaN quantum wells to emit light, face several key challenges that affect their performance and efficiency:

1. Quality of GaN Crystals:  The crystals grown were plagued by a high dislocation defect density, leading to increased non-radiative recombination rates (SRH recombination) and leakage currents. Dislocation defects create deep-level traps, serving as non-radiative recombination sites for electrons and holes. Achieving optimal LED functionality hinges on maximizing radiative recombination, crucial for light emission.

2. P-type doping intricacies: An LED's diode facilitates the injection of charge carriers into quantum wells. Traditionally, identifying a suitable p-type dopant for GaN proved challenging. Nakamura's breakthrough involved utilizing magnesium as a p-type dopant under precise conditions (temperature and pressure). However, despite progress, magnesium doping remains suboptimal due to its high activation energy, with only a fraction (approximately 10%) of magnesium getting activated. Consequently, a high magnesium concentration of 1e19 yields a mere one billion holes generated.

3. Challenges of Low Hole Mobility and Resistivity in p-GaN: The excessively large size of Mg atoms coupled with uneven doping distribution in p-GaN results in a drastic decline in hole mobility and a rough surface texture. Consequently, the p-material is typically grown last as subsequent material layers exhibit inferior quality when grown atop p-GaN. The hole mobilities for p-GaN average around 10-20 cm^2/(Vs), exacerbating its resistive properties. Additionally, the inhomogeneous hole distribution in Quantum Wells (QWs) due to poor mobility leads to uneven light absorption, highlighting the critical nature of having bound carriers uniformly dispersed within the QW structure, a challenge yet to be resolved adequately.

4. Polarization Charge Impacts: III-N materials, noted for their polarity (e.g., GaN, AlN, and InN), exhibit asymmetry in their crystal structure along the polar c-plane. The formation of heterojunctions (such as GaN/InGaN) results in the creation of sheet charges at these interfaces, altering band diagrams and quantum wells, and thereby impeding hole injection. The existence of polarization charge hinders electron-hole wavefunction overlap, significantly impacting QW gain and manifesting the Quantum Confined Stark Effect, diminishing overall device performance and optical gain potential. Several mitigation strategies have been explored to counteract the detrimental effects of polarization charge and enhance LED performance.

5. Strain: All III-N materials have a large lattice size mismatch,, which means the AlN is much, much smaller than GaN, which is much, much smaller than InN along the a-plane.

   A typical LED has at least 10-20 layers grown one on top of another, by the MOCVD process.

   If you grow the layers one on top of another, the structure will crack due to strain, unless the layers aren't too much different in composition. For example, 10% AlGaN on GaN may be ok,, up to a certain height, but 25% AlGaN on GaN will crack. This restricts the design space.

   Despite all these drawbacks, the AlInGaN material system is so very promising, since it can go all the way from infrared (InN bandgap=0.7 eV) to mid-deep UV (AlN bandgap=6.2 eV).