Hurdles and Hopes Examined
LEDs have become entrenched as a modern lighting technology. But LEDs can also emit ultraviolet light. In this article, Faiz looks at the technical challenges and possible applications for deep UV (DUV) LEDs.
Light-emitting diodes (LEDs) have taken over most lighting applications, since the first illumination-capable devices appeared in the mid-1990s. LED-powered luminaires have largely displaced both incandescent and fluorescent lamps, and in doing so have become one of the most common objects around us. However, LEDs are not limited to emitting light that we can see with our eyes. Visible-radiation-emitting LEDs were, historically, preceded by infrared-emitting devices that appeared back in the 1960s. Beginning in the late 1980s, LEDs capable of emitting near-UV radiation also came out, and now LEDs cover a broad range of wavelengths from the infrared to the UV.
You can usually find an LED to satisfy almost any application that requires a compact, solid-state source of radiation in the regions of the electromagnetic spectrum between X-rays and microwaves. Because of their widespread use, LEDs are now produced in extremely large numbers, and have become inexpensive commodity items. But the march of technology never stops, and in the world of LEDs, this amounts to increasing their brightness, enhancing the wall-plug efficiency and extending their emission wavelengths. The latter change has proved especially difficult to achieve, if we want to move deeper into the infrared or the UV. Here we take a look at the challenges in developing deep UV (DUV) LEDs, and some of the application possibilities that they promise to realize.
THE SITUATION WITH UV LEDs
LEDs that emit UV radiation have been available commercially for many years. They tend to compete with UV-emitting discharge tubes, and, where possible, replace the latter. On the one hand, LEDs offer distinct advantages of small size, low weight, low voltage operation, long life and ease of radiation extraction. Discharge-tube-based UV emitters, on the other hand, are bulky, fragile, require high operating voltages (greater than 100V) and contain toxic mercury.
Outwardly, there seems to be no competition, since LEDs have so many desirable attributes when compared with traditional mercury vapor UV discharge tubes. In reality, however, the delicate UV tubes offer stiff competition in many applications, simply because UV LEDs are often just not bright enough. This becomes increasingly obvious as we go deeper into the UV region—toward shorter and shorter wavelengths. Diode-based UV emitters then rapidly lose efficiency, and eventually get so dim that they cannot be used in any realistic applications. This is the reason that mercury discharge tubes are still going strong in many applications, where there is need for significant amounts of UV radiation at reasonable cost.
With LEDs that emit visible light maturing to become commonplace and inexpensive, attention has been turning toward developing similar devices for use in the UV region. Figure 1 shows the various parts of the UV region of the electromagnetic spectrum.
Near-UV LEDs are widely available and are roughly comparable in performance to blue- and violet-emitting devices (Figure 2). Mid-UV LEDs are also fairly common, and are used in most applications where UV LEDs have made inroads, such as handheld disinfection devices, air purifiers, stereolithography printers and so on. While not as bright or efficient as LEDs at longer wavelengths, these devices are now good enough to be used in cost-sensitive consumer electronics products. However, when we reach the deep UV region, the situation changes dramatically, because LED efficiencies fall precipitously.
— ADVERTISMENT—
—Advertise Here—
Different regions of the UV spectrum. Terminology used by the biomedical community is shown at the top, and that used by physicist and engineers at the bottom.
Ultraviolet (365nm) CBM-120-UV-Gen 4 LED Device (Image courtesy of Luminus)
Although commercial DUV LEDs are available, they are not capable of giving out decent amounts of UV radiation. This region is overflowing with application possibilities, but high-power, reasonably priced DUV LEDs are non-existent. Everything from water treatment plants and large building-scale air purifiers to industrial resin-curing systems and a variety of scientific instruments can make use of high-power DUV LEDs, if only such devices were available. These, and other applications, represent a potential market worth several hundreds of millions of dollars a year. This fact has not been lost on the companies that manufacture UV LEDs and, consequently, an enormous amount of effort is being directed toward increasing the efficiency of all UV LEDs, and especially for developing high-brightness DUV LEDs.
COMPLICATING FACTORS
So, what ails short-wavelength UV LEDs? As it turns out, both physics and materials science place hurdles in developing DUV LEDs. The color of light emitted by any LED depends on the material used for making the device. Different semiconductors have different energy band gaps, ΔE, which determine the wavelength (color) of radiation emission in accordance with the Planck relation:
Here, h is Planck’s constant (6.636×10-34 Joule-seconds), v is the frequency, λ is the wavelength of light emitted, and c is the speed of light (3×108m/s).
With that in mind, indium gallium arsenide phosphide (InGaAsP) is typically used for making red and orange LEDs, gallium arsenide phosphide (GaAsP) for yellow LEDs and aluminum gallium phosphide (AlGaP) for green LEDs. All blue and violet LEDs, and also higher-power green LEDs, are now made from indium gallium nitride (InGaN). Going beyond to the UV region, we need wider band-gap materials to make LEDs. The so-called gallium nitride (GaN) family semiconductor AlGaN is suitable for this, and currently is used for making almost all the world’s UV LEDs. Typically, gallium can be replaced with a mixture of gallium and aluminum, such as 10% gallium and 90% aluminum. The other half of the nitride semiconductor is always nitrogen. This is suitable for making near- and mid-UV LEDs, which emit UV radiation of wavelengths from 400nm to below 300nm.
To get down to even shorter UV wavelengths, we need material with even larger band gap. Theoretically, this is quite simple to do. Just increase the fraction of aluminum in AlGaN, say 20% aluminum with 80% gallium. Increasing the proportion of aluminum causes the band gap to get wider, so the LEDs emit shorter-wavelength radiation.
In practice, however, this turns out to be more complicated. One runs into two principal difficulties while trying to follow this approach. Adding aluminum beyond a certain fraction starts causing aluminum segregation, where aluminum-rich and aluminum-deficient regions get formed. Such material shows wild band gap fluctuations, making it unsuitable for making devices. The other difficulty with wide-band-gap materials is that it is difficult to dope them p-type—to obtain sufficiently hole-rich material. This is essential for making pn-junction devices, such as LEDs. But generally, the wider the band gap of a semiconductor, the harder it is to obtain strongly p-type material. These two materials-related issues (aluminum segregation and p-type doping difficulty) make it difficult to construct good, high-efficiency, UV-emitting LEDs, especially at shorter UV wavelengths.
Below 350nm, UV LEDs are conspicuously less efficient, and by the time the wavelengths go below 300nm, these devices put out only miniscule amounts of radiation. Short-wavelength UV LEDs lose efficiency because of increased in-device resistances contributed by the semiconductor and the device contacts. This causes most of the input electrical energy to be converted into wasted heat. This results in reduced UV output, and also makes it a challenge to remove the heat quickly, before device degradation sets in. Clearly, better materials and device designs are called for to avoid such problems.
OTHER MATERIALS
The material segregation problem can be solved, to a large extent, by exploiting wide-band-gap semiconductors other than AlGaN. There are several contenders for this. Zinc oxide (ZnO)—an ingredient in many sunblock creams—is probably the most studied alternative. ZnO LEDs are capable of emitting near-UV radiation around 380nm [1]. By adding some magnesium, one obtains magnesium zinc oxide (MgZnO), which emits shorter and shorter UV wavelengths down to around 200nm as the amount of magnesium is increased. These materials are capable of emitting UV radiation with efficiencies that significantly exceed that of traditional AlGaN. Seemingly, they would make perfect UV LED materials, due to their low cost, wide availability and easy processability.
Unfortunately, however, it turns out that these so-called ZnO-family semiconductors are extremely hard to dope p-type, and unless a viable method for their p-type doping is developed, there is no prospect for the commercial production of ZnO and MgZnO LEDs. Laboratory devices have been constructed from these semiconductors (Figure 3) and have shown good performance, but with insufficient lifetimes.
— ADVERTISMENT—
—Advertise Here—
Visible and near-UV radiation emission from an experimental ZnO LED
Other, even less mature, prospective materials for DUV LEDs include boron nitride, gallium oxide and even diamond. All these materials are hard to produce in a structural form suitable for making LEDs. In addition, there is no known method of producing hole-rich (p-type) material with them. GaN had the same problem with p-type doping until well into the 1980s when a team of Japanese scientists developed a technique for robust p-type doping of GaN and InGaN. This resulted in the commercial availability of blue LEDs, and earned the technique’s developers the Nobel Prize in Physics for 2014. Hopefully, one day we’ll have an analogous technique for p-type doping of one or more other wide-band-gap semiconductors. That will truly usher in the age of DUV LEDs.
A DIFFERENT STRATEGY
There is at least one other possible way to make short-wavelength emitters from wide-band-gap semiconductors—one that does not require p-type material, thus avoiding the complications of p-type doping. This requires making a capacitor-like structure (Figure 4), in which the active light-emitting material is sandwiched between two insulated conductors—one a shiny metal, and the other a transparent conductor, such as indium tin oxide.
Device structure of an impact-ionization UV emitter
By applying a high AC voltage, ranging from tens of volts up to a few hundred volts, it is possible to accelerate free electrons in the active material. When the accelerated electrons collide with the material’s atoms, they knock additional electrons off the atoms. Each time this happens, we get a knocked-off electron and an atom with one electron missing. This is, essentially, an electron and hole pair. When these entities merge (recombine) with each other, radiation gets emitted. With wide-band-gap materials, this radiation is in the UV region.
The process described above is called “impact ionization,” for obvious reasons. A device employing impact ionization can emit radiation without having n- and p-type regions in it. Accordingly, this kind of a device is not a pn-junction diode—it is not an LED. In this way, one can obtain a radiation-emitting device without having to create n- and p-type regions in a semiconductor and forming a pn-junction diode.
An impact ionization device is not just conceptual. For example, many night lights generate their glow from precisely such a construction. The active material in this case is usually zinc sulfide (ZnS) doped with manganese (Mn), which emits light in the visible region. Somewhat similar devices using wide-band-gap materials, such as MgZnO, also have been demonstrated to emit UV radiation around 285nm. The impact ionization scheme is a promising way to build UV and even DUV emitters from wide-band-gap semiconductors, which, as explained earlier, are nearly impossible to dope p-type reliably.
APPLICATIONS ABOUND
Solid-state UV emitters are suitable for many applications, ranging from air and water purification to materials processing and characterization. Traditionally, such applications have used mercury vapor lamps as sources of UV radiation, but solid-state lamps are much more desirable because of their small size and weight, longevity, low cost and absence of hazardous mercury.
As noted at the beginning of this article, the rather low efficiency of existing semiconductor-based UV and DUV sources have precluded them from many applications. With the successful realization of higher-efficiency devices, many existing applications will migrate to the use of semiconductor UV sources. This has already happened for longer UV wavelengths in the range of 400nm to 350nm, where mercury lamps increasingly have been replaced with UV LEDs. For shorter wavelengths, such as 254nm, where mercury lamps have strong emission, LEDs have not made any significant inroads, except in applications such as analytical instruments, where even weak emitters can serve the purpose.
There is a vast market waiting to be tapped, once high-efficiency DUV semiconductor devices with efficiencies of 20% or more become available. These applications include desktop printing, 3D printing and stereolithography, semiconductor lithography, resin curing, surface disinfection and production of biochemicals. Although UV LEDs are already being used to a limited extent in some of these applications, their market penetration is small because of their very low efficiency (currently, around 4%). Better sources will enable a much greater market capture and will also open up new application areas.
SELECTIVE DISINFECTION
An interesting possibility is the use of DUV sources for selectively destroying dangerous pathogens, such as the SARS-CoV-2 (COVID-19) virus, without harming humans. Generally, short wavelength UV radiation is harmful to all living organisms, because it causes irreversible changes in DNA strands. Genetic mutations brought about by such changes are almost always harmful, and can cause lasting damage. DUV radiation can also cause deleterious changes in proteins. For this reason, staring at sources of DUV radiation during their operation is strongly discouraged, because it can induce cataracts and even retinal damage.
Given this scenario, it is interesting to note that some studies indicate that UV radiation in the 200nm to 222nm range can kill viruses and other pathogens, but are not harmful to human cells [2]. This is largely because the wavelength of such radiation is smaller than the size of ordinary mammalian cells, so there is barely any optical interaction between them. Also, radiation in this range is absorbed by the outermost layer of skin, and does not penetrate deeper into the tissues. With even shorter wavelengths, however, such as radiation below 200nm, there is sufficient energy in each photon to ionize the medium it passes through. This can cause electrochemical damage to tissues, somewhat similar to that caused by radiation from radioactive materials.
The existence of a “safe window” for humans opens up the interesting possibility of making virus killers with sources that emit only within this limited wavelength region. At present, this possibility is being seriously investigated as a potential defense against viral epidemics.
CONCLUSION
Solid-state UV emitters present plenty of application possibilities, but are also challenging to make. Work that is being done now will, hopefully, result in better devices in the coming years, and this will open up a huge market-in-waiting. With visible-light LEDs maturing, UV and especially DUV LEDs are now the next challenge for materials scientists and device engineers. Vigorous activity going on in this area suggests there is hope that high-performance devices will become available in a few years. 
References:
[1] Zinc oxide light-emitting diodes: a review
Optical Engineering (OE) published peer-reviewed papers reporting on research, development, and applications of optics, photonics, and imaging science and engineering.
https://www.spiedigitallibrary.org/journals/optical-engineering/volume-58/issue-01/010901/Zinc-oxide-light-emitting-diodes-a-review/10.1117/1.OE.58.1.010901.full?SSO=1
[2] Effect of DUV on pathogens and human cells.
GMS Hygiene and Infection Control. 2021; 16: Doc07.
https://www.egms.de/static/pdf/journals/dgkh/2021-16/dgkh000378.pdf
— ADVERTISMENT—
—Advertise Here—
PUBLISHED IN CIRCUIT CELLAR MAGAZINE • OCTOBER 2021 #375 – Get a PDF of the issue 
About the author
Faiz Rahman, PhD, is a materials and device engineer, working on novel materials and device structures for both electronics and photonics. He develops new LEDs and solid-state lighting concepts at Electrospell and is also a visiting professor in Electrical Engineering at Ohio University. Faiz is a senior member of the Optical Society of America and of IEEE. His current research interests include the development of nano LEDs, plasmonic structures, and advanced nanofabrication technologies.
