Dilute nitrides fire up red LEDs

The race to enter the general illumination market has fuelled a dramatic increase in blue and white LED performance. In sharp contrast, red, amber and yellow emitters used in applications such as large color screens, traffic lights and architectural lighting have shown slow improvement.

However, at Quanlight, a start-up based in San Diego, California, we have developed a radically different approach to chip manufacture that can overcome these obstacles and deliver a dramatic hike in LED performance. By switching from the conventional material, AlGaInP, to the novel dilute nitride, InGaPN, it is possible to produce higher-brightness LEDs that are also less sensitive to temperature changes. This will aid applications such as large color displays, which require stable red emission to produce high-quality images.

The roots of our key technology are found in the research carried out at the University of California, San Diego (UCSD). The UCSD team, headed by Charles Tu, had already made some progress with InGaPN before it spawned Quanlight. This material, which contains about 1% nitrogen, is seen as a promising candidate for red emitters because it combines with GaP to produce a heterostructure with a larger band-offset than AlInGaP-based LEDs. Calculations had shown that it can produce brighter LEDs, thanks to greater current-handling capabilities. What s more, InGaPN produced some encouraging initial results, even though the material s system was relatively immature, poorly understood and challenging to grow.

The researchers at UCSD managed to build a working prototype LED that delivered several predicted improvements in device performance, including a reduced shift in emission wavelength with temperature. However, although these facilities were sufficient for the initial study, contaminants in the material produced by the MBE tool limited LED brightness, so further development required a switch to a commercial facility employing an MOCVD platform.

Commercializing academic research
To make this transition we founded Quanlight. After assembling a development team and raising $4 million through two funding rounds from private investors, our company started to develop epiwafers in August 2006 through the foundry service Bandwidth Semiconductor. Excellent progress has been made to date and we are on track to manufacture and sell red LED epiwafers to chipmakers by the end of this year. We then plan to extend our range of epiwafer LEDs to cover orange and yellow wavelengths between 585 and 660 nm. We are also open to licensing our process, or forming a partnership with another company.

The three key advantages that InGaPN LEDs have over their AlGaInP cousins are lower manufacturing costs, greater color temperature stability and brighter emission at high current densities.

The lower production costs result from a simpler manufacturing process, which is carried out with essentially the same manufacturing tools that are used to make conventional red LEDs. Traditional AlInGaP-based emitters are grown on GaAs substrates. To boost output the epilayer is often transferred to a transparent GaP platform or a mirrored carrier. Our process eliminates this epilayer removal and bonding process, and it involves dilute nitride growth directly onto GaP. This cuts the number of process steps that are needed and also reduces the bill of materials.

There is a small lattice mismatch between the GaP substrate and the InGaPN material in our device. It means that the epilayers are pseudomorphically strained, but this allows for enough quantum wells to be incorporated within the LED for high-power output. This produces a structure with a quality akin to AlInGaP grown on GaAs, but with the caveat that there are no commercially available substrates grown by the vertical-gradient freeze (VGF) method – an approach that produces boules with very low defect densities.

Improving GaP quality
We are currently using 3 inch substrates grown by the liquid-encapsulated Czochralski technique, which should yield LEDs that are commercially competitive in terms of brightness and reliability. However, we are also working with PVA TePla to develop GaP boules grown by the VGF method. We expect this venture to be successful, because the VGF approach is well understood and it is already used to produce other forms of substrate. Although we cannot predict the precise benefits of a transfer to VGF material, we expect that it will improve the lifetime and output of our InGaPN LEDs. We have already started to conduct initial tests of LEDs grown on this platform and we expect to have preliminary results very shortly.

InGaNP has intrinsic properties that ensure the LED s peak wavelength shift with temperature is smaller than that of AlInGaP-based devices. This makes them more attractive for incorporation into color displays, as they require stable light sources. These improvements in the color stability of our red emitters were seen in the lab at UCSD using MBE-grown material, but they have also been replicated with LEDs produced by MOCVD. The results of this test, which are shown in figure 1, were obtained by externally heating LEDs from 25 to 125 °C and recording the peak emission wavelength at various temperatures. Our LED s peak emission wavelength varied by just 3 nm over this temperature range, which is one-fifth of the shift shown for an AlGaInP chip produced by a leading red LED manufacturer.

The third advantage of our InGaPN LEDs – brighter emission at higher temperatures – results from a superior band structure that improves carrier confinement in the active region. The InGaPN LEDs combine with GaP barriers to produce a band offset that is two to three times as great as AlGaInP quantum wells and AlGaInP barriers.

The stronger performance at elevated temperatures has been verified by a test that compared Quanlight s LED output with that of a conventional red emitter at external temperatures between 25 and 150 °C (figure 2).

At the top end of this range the Quanlight device emitted a power equal to 48% of its output at 25 °C, but the reference LED delivered only 25%. When our device development is completed, we expect our material to produce devices with the same brightness as AlInGaP chips at room temperature and a near two-fold increase at 150 °C. The improved performance at higher temperatures will make these InGaNP LEDs more attractive for use in red and yellow traffic lights, which have minimum flux standards in the US at 25 and 74 °C.

Improved carrier confinement in the active region also aids current handling, and values of up to 9 A/mm2 have been produced in developmental tests (figure 3). These measurements were performed on-wafer, rather than from individually diced chips, and it is reasonable to expect that the operational limits of production versions will be lower. Nevertheless, we can expect our InGaPN LEDs to deliver saturation current densities that are two to three times as great as those of their AlInGaP equivalents.

Higher current-handling capabilities will help LED package and application engineers. Switching from conventional chips to InGaPN devices enables the use of smaller components with greater drive currents, which combine to produce an equivalent amount of light, or the deployment of fewer large LEDs in a high-power array. Both approaches cut the LED footprint and reduce overall costs. Either fewer LEDs are deployed or an equivalent number are used that are cheaper due to their smaller size.

We are now testing the reliability of our red emitters. This will begin with 5000 h tests on development devices. We are also planning to compare the performance of LEDs grown on LEC and VGF substrates.

Preparing for the launch
We have already transferred our device growth to an MOCVD platform and are optimizing the epilayer design. Improvements are being seen in the light output from our devices. This benefit is not at the expense of color or thermal stability, which are related to the intrinsic properties of our dilute nitride.

Although many people in our community may regard dilute nitride as an awkward material that has not fulfilled its promise in the telecoms sector, we have good reason to believe that our devices will be a commercial success. The high indium concentration that is required in the epilayers of dilute nitride telecom lasers increases the strain in the material and degrades lifetime and reliability. However, red, orange and yellow InGaPN LEDs will not suffer from this because they contain far less indium.

We will also benefit from our extensive experience of producing dilute nitrides, which will give us a strong competitive edge over other companies that might start developing products using this material. Although our epiwafers are grown at Bandwidth Semiconductor, process knowledge and intellectual property resides with our technical team, which is on site for all development growth sessions. This team drives the material development.

When we launch our portfolio of powerful red, amber and yellow epiwafer LEDs covering 585 to 660 nm, we will be in a position to target a $500 million market with a rapidly growing high-brightness sector. The performance advantages of our products will then provide a great match for applications requiring high power or a stable color output. Back-lighting units for LCD televisions, light engines for projectors, outdoor displays and other red-green-blue color-mixing applications will benefit from the smaller temperature-induced wavelength shifts, which will translate into simpler control mechanisms. In addition, transportation, hazard, theatrical and architectural lighting will benefit substantially from our enhanced color stability and intensity output.

Applications such as traffic lights and automotive brake lights all use AlInGaP LEDs to reduce energy use and costs. For these types of high-power application, InGaPN LEDs will enable lamp designers to make further cost cuts by using smaller chips driven at higher currents, or fewer large LEDs in an array. As the Quanlight LED operates efficiently at higher temperatures, a more compact or heat-intensive enclosure may be used.

The only area where the advantages of InGaPN LEDs are less significant is low-power applications with less rigid output specifications, such as Christmas tree lights. This is the only market that we will not be targeting aggressively, as low-power AlInGaP chips can already be supplied cheaply.