Dots Deliver Efficient, Phosphor-free White Lighting
LED light bulbs have many great attributes. Their efficiency trounces that of the incumbent source, the incandescent bulb, and unlike their energyefficient rival, the compact fluorescent, they are not ridden with mercury and reach full brightness in an instant. However, sales are poor, because retail prices are very high – on average an LED bulb sells for $28, according to the UK-based market analyst IMS Research.
To slash LED bulb prices, manufacturers must trim every major contribution to the overall cost. The packaged LED is the obvious place to start, since it accounts for almost 60 percent of the total bill of materials, according to a recent report from the US Department of Energy. If the efficiency of these chips can be increased, that will not only trim the price of the light bulb, thanks to a reduction in the number of LEDs needed to produce a given power; it will also reduce running costs, and in turn make the purchase more attractive.
White LED prices are very high, because the production process involves coating red, yellow or green phosphors on blue-emitting chips to produce white emission via colour mixing. This combination of chips and phosphors limits device efficiency, yield and reliability. But it is possible to tackle these issues by building novel, phosphor-free GaN-based dot-in-a-wire white LED. This technology promises to unlock the door to very-high-efficiency, more affordable light bulbs, according to our team at McGill University, Montreal, Canada.
If solid-state lighting products are to be competitive, they must deliver high-performance in the blue, green and red spectral range. Today, blue LEDs built from GaN-based quantum wells are a relatively mature, high performance technology, but their longer wavelength green, yellow and red cousins produce relatively low quantum efficiencies. What’s more, the device efficiency plummets at increasing current densities, a weakness that is commonly referred to as ‘efficiency droop’.
The low quantum efficiency of these green, yellow and red LEDs – and their severe efficiency droop – stems from material characteristics associated with III-nitride planar heterostructures, such as polarization fields and high densities of defects and dislocations. These traits are behind the unique carrier dynamics found in conventional III-nitride quantum-well LEDs, and have been claimed to be the root cause for electron leakage or overflow, Auger recombination, and poor hole transport in this type of device.
In contrast, our one-dimensional nanowire heterostructure LEDs do not suffer from many of these issues, which plague their planar counterparts. They can be built with drastically reduced dislocations and polarization fields, and they can enhance light extraction efficiency, thanks to far larger surface-to-volume ratios.
Conventional wisdom indicates that the way to realise green and red emission with nitride devices is to embed the InGaN quantum wells or ternary wires in GaN nanowire structures. But this approach is flawed: Only a small proportion of injected carriers transfer to the lateral surfaces of the wire, due to relatively poor carrier confinement in the nanoscale heterostructures; and the non-radiative carrier recombination on the wire surfaces significantly degrades the quantum efficiency of the device.
Our unique InGaN/GaN dot-in-a-wire heterostructures address this critical issue. In our case, InGaN quantum dots are incorporated in defect-free GaN nanowires that provide three-dimensional carrier confinement, a prerequisite for ultra-high-efficiency emission (see figure 1(a)). On top of this, our novel nanostructures offer unprecedented colour tunability. The size and the composition of the dots govern the emission wavelengths, and it is possible to create intrinsic whitelight sources from single GaN nanowires by varying the structural properties of the dots during a single epitaxial growth process.
Figure 1 (a) An illustration of InGaN/GaN dot-in-a-wire nanoscale heterostructures on a silicon (111) substrate.
We have employed a scanning electron microscope to acquire images of our InGaN/GaN dot-in-a-wire arrays that are grown directly on silicon (111) substrates by radio-frequency plasma-assisted MBE (see Figure 1 b). Catalyst-free nanowires, which are vertically aligned to the substrate and exhibit excellent size uniformity, form spontaneously under nitrogen-rich conditions.
Figure 1 (b) A 45o tilted scanning electron microscopy image of the InGaN/GaN dot-in-a-wire LED heterostructures grown on silicon (111).
A scanning transmission electron microscope can uncover more detailed images of our devices (see Figure 1 c). This tool reveals multiple InGaN quantum dots near the wire centre, due to strain-induced selforganization. The composition of these dots can berevealed with energy dispersive X-ray spectrometry, a technique that exposes variations in indium content from 10 percent to 50 percent (see Figure 1 d).
Figure 1 (c) A low-magnification, bright-field scanning transmission electron microscopy image clearly showing that InGaN quantum dots are well positioned in the center of a GaN nanowire. (d) An energy-dispersive, X-ray spectrometry spectrum image showing the quatitative variation of indium and gallium along the InGaN dots. The inset shows the line along which the electron energy loss spectrometry spectrum image is taken.
These structures produce strong photoluminescence emission across almost the entire visible range (see Figure 1 e).
(e) Room-temperature photoluminescence spectrum of white-emitting dot-in-a-wire LED heterostructures
To realize the full potential of our nanowire LEDs, we have significantly enhanced hole injection into the active region, while drastically reducing electron overflow out of it. Left unchecked, injected holes tend to reside in the small region close to the p-GaN, due to their heavy effective mass and low mobility. Meanwhile, the likelihood of electrons leaking out of or over the active region can be high, due to surface states and defects. The resulting hot carrier effect can supress the likelihood of radiative recombination and diminish LED efficiency under high injection conditions. In our opinion, these efficiency-limiting processes have not been well recognized and addressed in emerging nanowire devices – until now.
We combat these issues with a two-pronged approach to improve the recombination efficiency in this class of LED. We enhance hole transport by doping GaN barriers with magnesium to produce p-type modulation of the quantum dot active region. And by inserting a p-doped AlGaN electron-blocking layer between the quantum dot active region and p-GaN, we temper electron leakage and overflow from the light-generating zone.
To realize the full potential of our nanowire LEDs,we have significantly enhanced hole injection into the active region,while drastically reducing electron overflow out of it. Left unchecked, injected holes tend to reside in the small region close to the p-GaN, due to their heavy effective mass and low mobility
Figure 2: (a) Schematic illustration of InGaN/GaN dot-in-a-wire LEDs fabricated on an n-silicon (111) substrate. A single dot-in-a-wire LED structure with the incorporation of p-type modulation doping and an AlGaN electron-blocking layer (EBL) is shown in the inset. The flat energy band diagram is also illustrated. (b) Current-voltage characteristic of an InGaN/GaN dot-in-a-wire LED and the optical microscopy image of the device (inset). The series resistance is in the range of 20 - 50 Ω
Commercial promise The processes that we use to create highly uniform, densely packed InGaN/GaN dot-in-a-wire arrays on silicon substrates are well suited to the fabrication of large-area LEDs (see Figure 2 a). Fabrication involves ‘planarization’ of the nanowire arrays using a polyimide resist, before a p-type Ni/Au/indium-tin-oxide contact is deposited on the top of the nanowire surface and an n-metal Ti/Au contact is attached to the backside of the silicon substrate. This set of processes yields devices with excellent diode characteristics and negligible leakage current (see Figure 2 b).
By carefully selecting the height of the dots and their composition, we are able to fabricate InGaN/GaN dot-ina- wire LEDs on silicon substrates with strong green, yellow, orange and red emission (see Figure 3 a). And by combining dots of different colours in nanowires, we have formed high-performance, phosphor-free white LEDs on a silicon platform. These devices combine a strong white-light output with highly stable emission. This high level of performance over a wide range of operating conditions is seen in pulsed bias measurements of relative external quantum efficiency with injection current (see Figure 3b). There is no degradation in room-temperature device efficiency up to injection current densities of 2.2 kA cm-2.
Figure 3 (a)(Top) Optical images of the green, yellow, orange, and red-emitting dot-in-a-wire LEDs. (b) (left) Room-temperature relative external quantum efficiency (EQE) of the InGaN/GaN dot-in-a-wire LED device with the use of p-type modulation doping and an AlGaN electron-blocking layer. The simulated internal quantum efficiency (IQE) using the ABC model is shown for comparison. The optical image of the white LED is also shown in the inset. (c) (right) The Commission Internationale de l’Eclairage chromaticity diagram showing highly stable emission characteristics, with x and y in the ranges of ~ 0.33 – 0.35, and 0.36 – 0.38, respectively
We estimate that our internal quantum efficiency is about 60 percent, using an approach that is essentially based on the well-known ‘ABC’ model (the basis of this model is that carriers in an LED undergo one of three processes: Shockley-Reed Hall recombination, a nonradiative process that is proportional to the carrier density; radiative recombination, which is proportional to the square of the carrier density; or other higher order carrier loss processes, such as Auger recombination that depends on the cube of the carrier density). Values of the internal quantum efficiency extracted from our model agree with those obtained from optical pumping and electrical injection measurements.
Our dot-based devices set a new benchmark for internal quantum efficiency for any class of LED operating in the green, red, and entire visible spectral range. What’s more, the light emission characteristics are incredibly stable over a wide current range (from 333 A cm-2 to 1100 A cm-2) (see Figure 3 c)
Another great attribute of our novel LEDs is their absence of droop over a very wide operating range (see Figure 4). Simulations of the internal quantum efficiency, using what is essentially an ABC model, reveal that the third-order non-radiative carrier recombination coefficient is of the order of 10-34 cm6 s-1 – nearly four decades smaller than the commonly reported Auger coefficients in GaN-based quantum-well LEDs.
Figure 4 Relative external quantum efficiency (EQE) of a dot-in-a-wire white LED measured at different injection currents in the temperature range of 6 K to 440 K. The simulated internal quantum efficiency (IQE) (solid curve) using the ABC model is also shown for comparison.
This extremely small value should not raise any eyebrows, given the absence of efficiency droop. What’s more, it provides unambiguous evidence that Auger recombination plays a negligible role on the performance of InGaN/GaN dot-in-a-wire LEDs operating in the entire visible spectral range.
Our technology fundamentally addresses some of the major bottlenecks for the growth of phosphor-free solid-state lighting, such as low quantum efficiency and efficiency droop. Though still in its infancy, this remarkable dot-in-a-wire LED technology is already showing enormous potential for applications in future lighting and full-colour displays.
What’s more, it provides an extremely powerful, unprecedented approach for controlling the LED emission properties at the wafer level, which can significantly reduce manufacturing cost and improve device yield.
To advance the promise of these LEDs, we are now focusing on methods to transfer nanowire devices to transparent substrates, a step that will lead to high external quantum efficiency and effective thermal management. In addition, we are undertaking a detailed investigation of device reliability.