The Blue LED Nobel Prize: Historical context, current scientific understanding, human benefit

The paths that connect scientific understanding with tools and technology are rarely linear. Sometimes scientific understanding leads and enables, sometimes tools and technologies lead and enable. But by feeding into each other, they create virtuous spirals of forward and backward innovation 1. In perhaps no area of human knowledge has this been more evident than in semiconductors 2: the interactive progress in semiconductor science and technology has transformed and continues to transform both our scientific understanding of the universe and the technologies with which we live our daily lives. A testament to this are the seven Physics Nobel Prizes listed in Figure 1 that have been awarded in the broad area of semiconductors. Some of the Prizes were for technology breakthroughs – as with research surrounding Si planar processing and the integrated circuit. Some of the Prizes were for science breakthroughs – as with research surrounding the integer and fractional quantum Hall effects. And some were for nearly simultaneous science and technology breakthroughs – as with the transistor. The most recent 2014 Physics Nobel Prize, for the blue light-emitting diode (LED), is clearly for a technology breakthrough. Indeed, it is an extreme example in which the then-prevailing scientific understanding appeared to discourage continued effort. All the more amazing, then, was the perservence of Isamu Akasaki, Hiroshi Amano and Shuji Nakamura (“AAN” for short), who in the late 1980's and early 1990's braved seeming impossibilities and skeptics with an undaunted spirit. In this Opinion piece, we give an informal tour through: (1) the historical conext of semiconductor science and technology breakthroughs that preceded the most recent blue LED breakthrough; (2) our current scientific understanding of the blue LED breakthrough; and (3) the massive human benefit the blue LED breakthrough is unleashing. Up until the AAN breakthroughs, the four successive semiconductor technology generations could be thought of as: Ge, Si bipolar, Si MOS, and the “conventional” III-V's. Each generation followed a similar pattern of manufacturing process development from substrates to chips, and built and improved on hard-won knowledge from the previous generation. And each generation became part of a long chain of semiconductor innovations that continue to transform the world. The first semiconductor technology was based on Ge, and its foundation 4 was laid by a Bell Laboratories team led by William Shockley, whose goal was a field-effect amplifying device to replace the triode tube. Laying that foundation was not without its challenges. It was only after repeated failures to achieve power gain in Ge that Walter Brattain and John Bardeen finally turned to the use of a cat-whisker contact on a bar of n-type Ge, and the week before Christmas in 1947 discovered power gain. Shockley, familiar with work by Richard Haynes and Gerald Pearson also at Bell Laboratories, was ultimately able to explain the observed gain not as due to a field effect but instead due to minority carrier injection, a realization that later led to the much more manufacturable bipolar junction transistor 5. The combined discovery and explanation by Bardeen, Brattain and Shockley of the transistor is the most foundational and important breakthrough in the history of semiconductor science and technology, and was awarded the Physics Nobel Prize in 1956. If measured by economic or social benefit to mankind, then it might also be the most important breakthrough of the 20th Century. It initiated a now 6+-decade chain of ever-more-powerful semiconductor technologies, including those that eventually circled back to include field-effect amplifying devices. A significant handicap of Ge is its limited operating temperature range because of its narrow 0.69 eV band gap. Si, with its wider band gap of 1.1 eV, expands the operating temperature range, and thus became an intense focus of research. Initially, slight variations on the bag of fabrication tricks developed for Ge – crystal growth, grown/alloyed/diffused junction formation, alloyed contacts and cleaved or sawed chip cutting – were used for Si. These early Si devices were good enough to replace tubes in consumer applications, 62 but the more demanding computer, telephone and military applications could not tolerate the effects of contamination introduced at the pn junction edges. This led, in 1958-9, to the development by Jean Hoerni at Fairchild of the native oxide (SiO2) both for passivation and as a mask for dopant diffusion. Combined with photolithography, the resulting “planar process” became the most important combination of steps in Si processing. This in turn led almost immediately to the practical integrated circuit, by Robert Noyce (also at Fairchild), that enabled the interconnection of multiple circuit elements which Jack Kilby at Texas Instruments had earlier demonstrated in Ge. Ultimately, Kilby (though neither Hoerni, who died in 1997, nor Noyce, who died in 1990) was awarded half of the Physics Nobel Prize in 2000 for the invention of the integrated circuit (IC). But the implementation by Noyce using Hoerni's planar transistor process3 was a necessary precondition to take integration beyond the range of a handful of circuit elements interconnected by flying wires, and to initiate what is now known as the Moore's Law evolution of IC density and performance. Continued mastery of the planar process, especially clean, controlled oxidation, eventually led to the metal-oxide-semiconductor field-effect transistor (MOSFET) – the solid-state version of the triode tube that Brattain, Bardeen and Shockley were trying to replicate. In retrospect, MOS transistors would have been practically impossible to implement in the Ge technology available in 1947. Even in Si, it took Fairchild and Intel nearly a decade to clean up their Si crystal growth and oxidation processes to the point where MOS transistors could match the stability of bipolar transistors. Even after reaching acceptable performance and yields in both p-MOS and n-MOS transistors, it took another half decade to make complementary MOS (CMOS), the work-horse technology whose low power consumption enabled scaling to ever higher densities and very large scale integration (VLSI). Enroute, a variant of the MOSFET, the charge-coupled device (CCD), was developed, with the Physics Nobel Prize in 2009 awarded to Willard Boyle and George Smith of Bell Laboratories for its invention and use as a solid-state image sensor. Ultimately, it is CMOS and VLSI technology that have enabled the explosion of computing power that has transformed the world to one in which smartphones and the Internet feed each other information 24/7. And, to come full circle, that computing power in turn enables science breakthroughs of even the most fundamental sort – including the recent data-analysis-rich experimental confirmation of the Higgs Boson, whose prediction was awarded the Physics Nobel Prize in 2013. Though the most important, Si is by no means the only important semiconductor material. III-V semiconductors, materials which are compounds of elements from column III and V of the periodic table, have found utility as well. The easier of these materials to work with, the III-P's, III-As's and III-Sb's, are mostly so-called direct-bandgap semiconductors. Thus, they emit and absorb light much more efficiently than Si does, hence are useful for optoelectronics; and they have much higher electron and hole mobilities than Si has, hence are useful for low-noise and high-speed electronics. These III-V semiconductors could build to some extent on the fabrication processes developed for Si. Most importantly, they have more degrees of compositional freedom than Si. The binary III-V compound GaAs is nearly completely miscible with the binary III-V compound AlAs and so the ternary alloy AlxGa1-xAs is possible, as are a zoo of other ternary, quaternary and even pentanary III-V alloys. With these additional degrees of compositional freedom come the ability to fabricate devices with tailored spatial variations not only in doping but in bandgap. Thus, the development of precision composition-tailoring epitaxy techniques – liquid-phase epitaxy (LPE), vapor-phase epitaxy (VPE), molecular beam epitaxy (MBE) and metal-organic vapor-phase epitaxy (MOVPE) – was critical. And heterostructures – whose development by Herb Kroemer (at Varian) and Zhores Alferov (Ioffe Physico-Technical Institute) was awarded half of the Physics Nobel Prize in 2000 – were also critical. Indeed, precision epitaxy and “bandgap engineering” via heterostructures are used in virtually all modern III-V semiconductor devices. Among them are the critical forerunners to the blue LED: the red and yellow LEDs pioneered by Nick Holonyak (University of Illinois) and George Craford (Monsanto and Hewlett Packard) that are useful in their own right for signaling and displays. Also among them are the single-mode laser diodes that enable the broadband dense-wavelength-division-multiplexed optical fiber networks that are the backbone of the Internet, and the low-noise high-speed transistors used for “last mile” wireless communications to and from smartphones. And, again to come full circle, precision epitaxy and bandgap engineering technology enabled the scientific discovery of quantum fluids with fractionally charged excitations by Robert Laughlin (Lawrence Livermore National Laboratory), Horst Stormer (Bell Laboratories) and Dan Tsui (Bell Laboratories), awarded the Physics Nobel Prize in 1998. The one III-V semiconductor materials system that had remained elusive through the years of intense development of Ge, Si and the “conventional” III-V technologies were the “unconventional” III-N semiconductor materials. Early work at RCA had established a plausibility proof in the 1965-70 period that some blue light could be coerced out of GaN. But the sheet resistance of the p-layer was atrociously high (Vf = 20-30 V) and the conversion efficiency was unacceptably low. The major challenges: no substrate in sight that was reasonably lattice-matched to GaN so as to keep dislocation density acceptably low (seemingly a must for any minority carrier device); and no dopant process for low-sheet-resistance p-layers. Indeed, by the 1980's the challenges seemed insurmountable, and most researchers had abandoned the field. All the more amazing and unexpected then were the three pivotal AAN breakthroughs in the late 1980's and early 1990's: (a) the “magic” AlN and GaN buffer layers on sapphire with reduced dislocation densities; (b) an unexpected method to activate p-type Mg doping of GaN; and (c) the unexpected resistance of InGaN quantum well luminescence to defects. Indeed, one can see, from the citation plot in Figure 2, how these breakthroughs unleashed huge subsequent worldwide efforts in III-N semiconductor research. Even AAN themselves, though, fantastic as their intuition was, did not fully understand the mechanisms underlying their breakthroughs. Those mechanisms took 1-2 more decades to unravel; and here we discuss briefly our best current scientific understanding of those mechanisms. The first breakthrough, the discovery of the low-temperature (LT) AlN buffer layer on sapphire, 7 is undoubtedly the critical foundational step in the history of III-N materials. In the absence of bulk GaN substrates, sapphire is one of the few available substrates that is compatible with the severe growth conditions associated with GaN MOVPE. But, given their mismatch both in crystallography and lattice constant, growth of GaN on sapphire is an extreme exercise in heteroepitaxy by any measure. By the early 1980s, the task was taken by most to be at best quixotic – Akasaki might very well have been the only researcher persevering in this journey, joined by a handful of associates and assistants in his group (including Amano who was then a graduate student). Even for these two talented experimentalists, countless growth experiments failed, but eventually led them to reformulate the problem as the difficulty of nucleation of GaN on highly lattice-mismatched sapphire. In the inventors’ own accounts, Akasaki described the need for “a softer or flexible thin layer” while Amano considered the insertion of Al-containing layers, based on the empirical finding that “AlGaN layers tend to have a better morphology.” Ultimately, they circumvented the nucleation barrier through the use of an AlN buffer layer grown at an (at first accidental) much-reduced temperature. Then, a few years later, Nakamura discovered that a GaN buffer layer grown at a reduced temperature in a homebuilt “cross-flow” or “two-flow” MOVPE reactor gave yet improved results 8. Throughout the 1990s there were debates and much confusion at major conferences the of buffer layer from the two as well as the of reactor of the GaN at the were to of buffer In we now that two were and are at in what has come to be the and used growth The first is the of GaN on sapphire both with a high of with the underlying sapphire, and in a narrow and range of at an reduced growth temperature a of in the – a not from a thin layer of by on a in Then, because GaN has a than AlN does, GaN is and its are more to processes as and epitaxy These processes enable the crystal and density of the to be a range by growth The is the of the GaN a subsequent growth The GaN that the in the without much with the underlying sapphire substrate and is a process now of the to the and with one another the and subsequent The full understanding of these growth is and nearly all the important and in crystal and growth including the and to a also a yet significant from work on the GaN his of in temperature as by In his growth the of the was through and by the GaN layer on sapphire. The an the developed, and to the world the of the growth evolution of GaN on sapphire. of and are and in GaN epitaxy research. The breakthrough, the of p-type is no than the In the late even with a GaN layer from the process, p-type doping was Akasaki and Amano p-type but turned out highly or to the a was that resistance to in semiconductors was fundamental and due to so-called by native point – some semiconductors, have not been p-type even to this A first of Amano observed in a electron that no of p-type gave a blue luminescence electron beam This discovery became an in the years to the of any even confirmation was Akasaki and Amano the and demonstrated p-type with a now A few years later, Nakamura the by with a more process, while also passivation as the for the observed high The GaN light-emitting diode was in In we now that is to the p-type doping breakthrough, but a The first is MOVPE of there is an of from and carrier the is able to in GaN growth and is able to with and to these to Akasaki and by or the of native which would p-type is a that the of and Mg into GaN without the The is by the Mg immediately growth there is no p-type The must be through or through the must take in a and it must be at a high enough temperature not only to the but to enable to and then a significant barrier to and the of passivation was well in for and GaAs in the and these or not have they did the of and the of as a for The of this is to by the it in the early The first two breakthroughs, as from were by the science and technology with only This is in both Amano and Nobel in which they at the GaN at The The at the from was that as in semiconductors, and and them from to emit light GaN on sapphire, even with the developed AlN and GaN buffer layers on sapphire, were more than GaAs by at of The for blue light was materials with low density and high a It was by no means light devices could be from GaN. The in that Nakamura to results of and higher efficiency InGaN of his of of with the LEDs is no that Nakamura was for LEDs from to his discovery was GaN technology was to all materials were for the science and technology remained at countless no the in but did to their In the subsequent few the GaN two a to up with and and a scientific to the mechanisms underlying the in GaN. the most from electron from LEDs and the of luminescence efficiency of InGaN to the of those defects. It took at years of and to a of this with mechanisms The current is that alloy is are much than into GaN and In growth conditions discovered by the InGaN alloy to into and in to This of is to the of as and InGaN has a bandgap than and an of and that their and them from by the Thus, one of most important was a which InGaN is at an which the to and the InGaN alloy a process became the to a The blue LED breakthrough led first and to solid-state which is in the of how we light our world. we discuss the current and of the blue LED breakthrough was a of breakthroughs in the foundational III-N semiconductor materials so we also discuss other technologies that this materials system might enable. The most of the blue LED is that it is on the of the light that the human can Thus, the and LEDs that were in the from these blue LEDs can be both into yellow and red via and also with those Thus, blue LEDs enable light that can be used for – what we now solid-state in the two decades the AAN breakthroughs has been of for a of III-N blue LEDs are similar to red and LEDs in the they could the chip and developed for those III-V light has so from associated with a of applications, as signaling and to Most importantly, because is a semiconductor technology, it has the for very high efficiency and an evolution in as and this point in efficiency for for research devices is much higher than those of historical technologies as and And because is so critical to and so much of the and of the the of is A recent by the of based on for and in in the of of to of and of consumer and to the of power would be these by – a huge benefit to to any technology that these those for in the of most other in the efficiency of other technologies, though, these are not but has progress in efficiency and but is from only are in efficiency and but that make more are as With to as is higher than and higher than But at there is yet for LED efficiency as electron density – an better and would enable major and efficiency III-N LED efficiency also as from the blue to the – a which better and would enable light without of with the human is a – the of the red light by current to which the human is With to technology can much more than efficiency LEDs are point of they can in be controlled in LEDs can be and combined with or other their can be controlled LEDs have very operating their can be in and LEDs are they can be efficiently with or power of the long of can be to and more integration of into on the example of to match up for to match up to and the of light on the and to match human up for and information All this is to a of science and a of the of research has even an unexpected the human a that and the with for to and come full back to has the to the consumption of light and via that are on only where and And to beyond as did previous generations of semiconductor technologies, almost create that we even Some have and we that and its be so and as to be by another technology as the blue LED led to in turn leads to other out in light are the most both and so they could also the for and for the Internet of III-N materials emit light a range, they enable current and applications as: for and optical fiber for and LEDs and for and optical because materials have very they are useful for power devices high and high and which can efficiently at high temperature. These are for power the of the and for power and the and that on that of a large and of and in III-N semiconductor and and of of the applications they we Akasaki, Amano and also our to the researchers on whose work AAN built and from – including those awarded previous Physics Nobel Prizes in semiconductor science and And we those whose and on the of all this enable yet more breakthroughs, and human most importantly, we both and at the of this Opinion piece, the that scientific understanding with tools and technologies is rarely linear. scientific understanding enables and tools and But scientific understanding is and technology researchers must as did its for their breakthroughs. this scientific understanding up but ultimately as it is limited by the tools and technologies it in the of understanding and a system which and are to one Also as at the of this Opinion piece, the of technology breakthroughs is evident in in recent more and more Physics Nobel Prizes have been awarded for technology breakthroughs, and of these by the most have been for semiconductors. live in the information and the critical of the information are semiconductor with and we with processing and we long and with laser diode and and we through displays. we use to the world via chips, the most recent and important of which are for technologies benefit to mankind, they are the very of daily are for from and This work was by the of of and and by by the of of National Laboratories is a and by a of for the of National