Quantum well intermixing: The quest for orange and yellow lasers
The LED light bulb is destined to consign incandescents and fluorescents to the history books. That's because this lighting source has much to recommend it "“ it is efficient, long lasting, and hits full brightness in an instant "“ and its sales are accelerating as prices plummet.
But the LED might not have the final say in the rise of solid-state lighting. In time, it could be superseded by white-light sources formed by mixing the output of diode lasers emitting at different wavelengths. Attractions of this laser-based technology are the promise of even higher efficacies, a superior colour quality and light-based communication rates that are an order of magnitude higher than those that could be produced by LEDs.
Before lasers can become the dominant source for solid-state lighting, however, there is a need to address their biggest weakness, which is their poor spectral coverage. Efficient blue and red sources are readily available, but losses rise in the green, and there are no commercial, directly emitting chips in the orange and yellow. This compromises the quality of white-light-based laser sources produced by colour mixing.
Our team at the King Abdullah University of Science and Technology (KAUST) in Saudi Arabia is working to overcome this issue by developing orange and yellow lasers. We are making good progress, having already had significant success with a novel form of quantum well intermixing "“ it involves applying strain to the active region via the growth of a SiO2 film on top of the structure. Highlights of our work include fabricating a phosphide-based laser that operates down to 608 nm at room temperature, and the world's first yellow superluminescent LED. The latter produces a total output of around 5 mW at 583 nm.
Figure 1: Neither the GaN/InGaN laser diodes (LDs), nor those made from InGaP/InAlGaP, can deliver emission in the yellow and orange.
One of the first questions that any researcher must face when attempting to build orange or yellow lasers is this: what material they should use to make their device? Commercial lasers based on the nitrides, and sporting InGaN/GaN quantum wells, can span the violet to green (405 nm to 530 nm), while cousins based on the phosphides emit in the red, covering 632 nm to 690 nm. In between these two spectral ranges there is the green-yellow-orange wavelengths of 530 nm to 630 nm, where no laser diodes are for sale.
If white-emitting light sources were made from what is commercially available, the colour-rendering index (CRI) would be limited to a value of 80 or less, which is not acceptable. To meet customer requirements, the CRI must exceed 90. This target can be meet, but requires lasers to be used in conjunction with phosphors emitting in the green, yellow, and orange/red. However, this addresses colour quality at the expense of system efficiency, which falls due to the difference in energy between the photons that are pumping the phosphor and those that are being emitted by it "“ and also energy losses from some phosphors that are absorbing the light that other ones are emitting.
At first glance, it would appear that the best way to realise a high CRI with a phosphor-free system would be to extend the spectral range covered by either InGaP/InAlGaP or InGaN/GaN systems. With production of these classes of chips firmly established, manufacture could quickly follow development.
However, it is very challenging to have success with these material systems. Phosphide-based systems require an increase in the aluminium content to propel emission to the green-yellow-orange, but moving to shorter wavelengths reduces confinement of electrons and holes in the quantum wells, prohibiting the growth of efficient active regions. Efficacy takes a further knock from the tremendous hike in the density of the non-radiative centres.
With the nitrides, different issues are at play. The indium content has to increase to reach longer wavelengths, but this introduces significant strain in the quantum wells. Indium segregation can also occur, making it even harder to produce high quality green, yellow and orange sources.
Prior to our efforts, the only phosphide-based laser diodes emitting in the orange-to-green spectral range were chips held at cryogenic temperatures and subjected to high external pressures. Such devices clearly are impractical for commercial applications.
An alternative approach, which fails to tackle the issue head on, is to turn to frequency doubling of a diode-pumped solid-state laser or an infrared laser diode. One downside of this technology is that the non-linear crystals used for second-harmonic generation are not that efficient. What's more, they require an externally distributed Bragg reflector and a good heat sink, which increase complexity.
Figure 2: A transmission electron microscopy image of the InGaP/InAlGaP laser structure with an InGaP single quantum well.
Another option is to use a gas-based laser. But these sources have sizes ranging from that of a shoebox to filling an entire room. So it is of no surprise that there is huge demand for a tiny chip that could replace these complex, expensive and power consuming lasers that emit within the colour gap.
Attempts to fabricate lasers in the colour gap date back four decades. During that time, much effort has been devoted to improving the epitaxy and design of phosphide-based material systems. The first milestone came in 1971, when a group at RCA Laboratories reported orange lasing with low-efficiency and high threshold current densities at 80 K. Two decades latter came the next significant breakthrough, when, in 1992, a team from Toshiba realised room temperature orange emission at 625 nm, with a device featuring six pairs of multi-quantum barriers. However, this emitter produced a very low output power. In 2003, another advance was reported "“ the fabrication of yellow-orange emitters, based on a strained InGaP quantum well, that were grown on a transparent, compositionally graded AlInGaP buffer. Although the devices did not lase, they set a new benchmark for LED power for this wavelength range, producing 0.18 μW per facet.
Mixing the wellsAn additional option for controlling the emission wavelength of a laser is quantum well-intermixing (QWI). This established technique is already used in the production of lasers, with pioneers including the maker of red and infrared sources, Intense of North Brunswick, NJ. Our contribution to this technology is its application to orange and yellow lasers.
Figure 3: Photoluminescence peak emission of InGaP/InAlGaP laser samples capped with SiO2 and annealed at 950ËšC for 1,2 and 3 cycles. Also shown is the process at a reduced annealing temperature of 900ËšC. Emission in orange and yellow was accomplished via different combinations of annealing time, duration and the number of cycles of annealing.
Figure 4: Intermixed laser structures with front and back contacts can span the red, orange, yellow and green.
Figure 5: Optimized photo-luminescence sample after annealing at 950ËšC for 9 and 5 cycles of 30s duration. These samples are used to fabricated orange laser emitting at 608 nm and yellow super-luminescent diode at 583 nm. Also shown is the photo-luminescence of the as-grown sample.
The purpose of QWI is to selectively tune the band edge of a semiconductor heterostructure. The process for accomplishing this begins by forming a disordered region that is near to, but separated from, the quantum well active region. Annealing follows, driving the diffusion of vacancies/defects from the disordered region into the quantum well region, where they enhance inter-diffusion at the junctions between the well and barrier. Repeating this process of introducing disorder and subsequent annealing of the heterostructure shifts the bandgap to ever shorter wavelengths.
Only a handful of groups have applied the QWI process to devices that are made with the InGaP/InAlGaP material system and have wavelengths as short as 640 nm. These attempts have been unsuccessful, as there has been a minimal blueshift in the emission wavelength.
One example of this is work by the team from Xerox:. They annealed as-grown and SiO2 capped samples of InGaP/InGaAlP at 900ËšC for four hours. The emission blue-shifted, but only by 10 nm.
It is worth noting that longer wavelengths have led to slightly more success, with a sample coated in a layer deposited by sputtering, rather than PECVD. This sample underwent a shift from 670 nm to 640 nm. However, this group from the University of Glasgow reported photoluminescence for devices cooled by liquid nitrogen, suggesting that the QWI process degraded the active layer. The researchers postulated that the intermixing created point defects at the sample surface, with damage in the p-cladding impairing device performance. The point defects are thought to originate from plasma-induced damage, which occurs when plasma species are accelerated across the space-charge region.
Our team has been much more successful, using a refined QWI process that breaks the record for the tuning of the bandgap in the InGaP/InAlGaP material system. Emission can now span from 640 nm in the red all the way to 565 nm, which is just inside the start of the green spectrum.
Figure 6: Power-current characteristics as a function of current density of (a) an as-grown red laser and (b) an intermixed orange laser. Inset: Red and orange lasing spot emitting at a current injection of 1.2Jth.
Figure 7: Room temperature lasing spectra obtained at a current injection of 1.2Jth. Inset: orange lasing spot emitting at 608 nm.
Figure 8: Power-current characteristics as a function of current density from an InGaP/InAlGaP intermixed sample. Inset: amplified spontaneous emission spectra at 583 nm.
At the heart of our process is the PECVD of a dielectric layer. Its thickness is comparable to, or thicker than, the upper-cladding layer of the quantum well. The role of this thick dielectric layer, which is formed from SiO2, is to induce significant strain within the quantum well region, without damaging contacts or the p-cladding. Note that if the dielectric is too thin, the QWI process fails to deliver a significant shift in emission wavelength, while impairing surface morphology and electrical and optical characteristics.
Using a structure with a SiO2 layer of appropriate thickness, we undertook annealing in a rapid thermal processor that promotes quantum-well intermixing. We used multiple cycles at a relatively low temperature to reduce chances of the point defects diffusing to form cluster defects "“ these large imperfections are highly undesirable, as they cannot be removed by annealing, and they degrade crystal quality. A lower temperature also helps to reduce redistribution and diffusion of zinc dopants. If this occurred, it could destroy the diode's p-n junction.
During our low-temperature annealing process, atoms in the thin InAlGaP quantum barrier and InGaP quantum well interdiffuse, to relax the strain induced by the thick dielectric layer. This creates grown-in strain relaxation and an atomic composition change, which work together to produce a shift in the bandgap.
We have fabricated a range of emitters with our novel QWI technique. This includes simple, broad-area devices with a wavelength as short as 608 nm. They lase at room temperature, produce a total output power of 46 mW, and break new ground for banishing the colour gap from the long-wavelength side. This success has enabled us to produce the first report of lasing action from a post-growth interdiffused process.
We have also demonstrated the first yellow superluminescence at a wavelength of 583 nm. This device has a total two-facet output of 5 mW, which is the highest optical power ever reported at this wavelength in this material system.
Our results are very encouraging, given that the cladding does not contain a complicated multi-quantum barrier to suppress carrier overflow. If we were to introduce this, it should enable a cut in threshold, plus a hike in output power.
We hope that our successes − demonstrating an orange laser and yellow superluminescent diode "“ will pave the way to the fabrication of the first yellow semiconductor laser. Uniting this with cousins emitting in the green, red and blue could spawn a new revolution in solid-state lighting. There is the promise of a source that combines incredibly high efficacies with a CRI that is close to 100, and a light-based replacement for WiFi that delivers breath-taking
data rates.
Multi-colour, monolithic sources
To produce monolithically integrated multiple photonic devices on a single chip, there needs to be a process for shifting the bandgap energy of a selected area to a value that is different from that of another area. With the quantum well intermixing process developed at King Abdullah University of Science and Technology (KAUST), the strain-induced disordering of the quantum well and barrier depends on the thickness of the dielectric. It is also possible to introduce different dielectrics, such as silicon nitride and silicon oxide, to induce different degrees of strain, and ultimately produce a different emission wavelength in a different area of the substrate. By adopting this type of approach, the team at KAUST have produced a multiple bandgap semiconductor light-emitting device.
Opportunities for green, orange and yellow lasers
COMMERCIALLY available true-green, orange and yellow laser diodes and LEDs would be attractive alternatives to the gas and diode-pumped solid-state lasers that currently serve the spectral range spanning 550 nm to 620 nm.
Potential applications include:
- Solid state lighting A combination of blue and yellow − or blue, green and red − can produce efficient white light for lighting applications.
- Visible light communications and Li-Fi Modulating the light beam from diode lasers, rather than LEDs, has extreme advantages in the field of visible light communication. Current Li-Fi technology uses LED technology, which is compromised by its relatively slow modulation rates. It could be replaced by visible lasers with a GHz modulation bandwidth.
- Displays The availability of high efficiency red, green and blue lasers would enable high-definition projectors, HD TV and micro-projector technologies.
- Optical communication using plastic fibre Plastic fibre has a peak transmission wavelength at 580 nm, making yellow lasers ideal sources.
- Skin care Yellow lasers can treat wrinkles, sun damage, pigmentation and other skin problems, with a dramatic improvement after just a few treatments. Success results from absorption of yellow light by melanin cells under pigmentation.
- Retinal diabetes surgery This form of surgery currently uses a DPSS laser emitting at around 577 nm. Working with this wavelength allows the curing of tissues at the back of the eye. This can be accomplished without damaging tissue that the laser beam must pass through.
- Flow cytometry This technique in the field of bioscience uses photoluminescence to quickly uncover the flow of species or certain type of cells. Solid-state lasers emitting at 595 nm are used for this application, but they could be replaced by laser diodes formed by quantum-well intermixing that emit at ideal wavelengths.
- Horticulture LED and laser lighting can be fined tuned to deliver the optimum spectral output for different phases of the plant growth cycle. This approach could also have nutritional benefits.
- Pharmaceutical Industry Lighting based on LEDs, superluminescent diodes and laser diodes could be used in the construction of genetically engineered plants for the production of pharmaceutical products (referred as Pharming). This form of lighting is ideal, because transgenic plants need particular illumination wavelengths at particular phases of their life cycle.
