Inferior light quality holds back the sales of most LEDbased replacements for halogen lamps in casinos, hotels, high-end retailers and cruise ships. But that’s not the case for Soraa’s lighting products, which produce full-spectrum emission with a violet LED pumping red, green and blue phosphors. Richard Stevenson reports.

A FEW YEARS AGO, the lighting industry debated whether there would be a revolution in solidstate lighting. Now, however, it’s not a question of it but when.

For chipmakers, this revolution can’t come to soon. They are operating in a market suffering from over capacity, and shipments to light bulb makers offer the opportunity for higher margins and increased sales.


Many LED makers are now working hard to forge strong relationships with lamp makers, so that as sales of solid-state lamps surge, so do their own chip revenues. But this is not the only road to success, and a few other firms are taking a different path – one that allows them to take far greater control of their own destiny. Companies that fall into this category include Cree, which has diversified from being an LED manufacturer to a maker of replacements for 40 W and 60 W incandescents, and Soraa, a start-up based in Goleta, CA, that has developed a novel chip that features in its solid-state replacement for 50-75 W halogen lamps.

Soraa is by no means the only maker of an LED-based replacement for the halogen lamp. But its product has a far higher colour quality than that produced many of its rivals – it has a colourrendering index (CRI) of 95, compared to a typical value of 80 – and that should ensure success in this market.




The Soraa SNAP System combines a unique 10 degree lamp with an innovative array of interchangeable beam and colour-shifting accessories

“We see our customers as those that have refused to consider LEDs in the past,” reveals Chief Operating Officer Douglas Devine. “They have refused to move away from the halogen, because saving a couple of bucks in the electricity bill isn’t worth damaging the ambience of the customer-facing areas that they have spent so much time and effort on.”

In May 2012, Soraa launched its full-colour MR16s that are wining deployment in businesses measuring revenue-per-square-foot, such as high-end retailers, casinos, cruise ships and hotels.
“These are areas where they employ lighting designers, who are more than sophisticated enough to appreciate the difference in the quality of light.”

To realise a full-colour spectrum, Soraa’s lamps pump red, blue and green phosphors with a violet-emitting chip. This is a markedly different approach from that used in most white-light sources, which employ a blue LED to excite a yellow phosphor, which is sometimes combined with a red variant.

“When you excite with a blue LED, it is not possible with a phosphor to tune the light down to violet, so typical competitor products are missing the violet all together,” explains Devine. This omission makes a massive difference to the appearance of anything containing whiteners, which are excited by violet light.

Customers investing in Sorra’s MR16 lamps are rewarded by getting light quality of a halogen source, but at a 75 percent gain in efficiency – despite the 10 percent reduction in efficiency for pumping with a violet, rather than blue. The high efficiency of the Soraa lamp stems from a novel chip design with very high extraction efficiency.

Triangular chips

Insights into the design of this LED are provided in an Applied Physics Letters paper, which details a triangular chip (note that the geometry reported in that journal does not represent a specific commercial device). The LEDs reported in that paper are grown by MOCVD on GaN substrates, and have a triangular shape to ensure in-plane light is extracted from the chip after one or two bounces within it.


50 mm diameter, free-standing SCoRA crystal

To optimise extraction, Soraa’s engineers model the proportion of light extracted from the LED at various chip heights. If it is just a few microns high, the device behaves like a thin-film chip with extraction resulting from just surface roughening. However, if the height of the chip increases to 50 μm or more, extraction is appreciably higher, thanks to light exiting the device via its sidewalls.

Triangular-shaped LEDs made at Soraa, which have a roughened top surface and sides with lengths of about 380 μm, produce a maximum external quantum efficiency of just over 73 percent. This high value is aided by the native substrate, which adds considerably to the device costs, but enables growth of an epistructure that is free from extended defects.

To prevent the LED costs within MR16s from becoming extortionate, devices are driven at incredibly high current densities, so very little chip real estate is required in each lamp. With a conventional LED – which is formed on sapphire, SiC or silicon – driving the device in this way leads to a significant reduction in efficiency, due to a controversial malady known as droop. This still affects Soraa’s LEDs, but the impact is far more modest, due to the combination of improved crystalline quality (due to native GaN substrate) and a light-generating active region that is far larger than a conventional device.

When the current is cranked up, the output produced by a single LED is sufficient for a halogen lamp (for the MR16 form factor, up to 75 W equivalence is achieved). Using a single source is beneficial, because it casts a single shadow. According to Soraa, rivals have multi-source lamps producing multiple shadows that fall far short of the 75 W equivalence mark. This divergence from the norm, in terms of LED count, plus a difference in the operating temperature regime, mean that Soraa’s lamp was never going to be an off-the-shelf product.
Instead, the lamp had to be designed specifically for a single GaN-on-GaN LED, making it better suited to a vertically integrated approach for the manufacture of MR16s.

But that is not to say every single manufacture steps takes place at Soraa. “We’re doing the design of the light chip and manufacturing the LED ourselves. Light chip manufacture and lamp [assembly] are done at sub-cons,” explains Devine. Sales of MR 16s then occur through web sites, and also via a network of distributors in Europe, Asia, Australia and New Zealand.

Making substrates

To try to reduce the cost of these lamps, Soraa is developing a novel ammonothermal process to make its own substrates. “We have a team working on bulk GaN development, because what we’d ideally like to buy is not available,” admits Devine. “We’ve developed a type of reactor, called a SCoRA (scalable, compact, rapid ammonothermal), that is making very good progress.”
The traditional ammonothermal method, which has been pioneered by Ammono of Warsaw, Poland, produces GaN of incredibly high quality, but growth rates can be as low as just a few microns per hour. In comparison, Soraa’s approach is capable of growth at greater than 40 μm/hr, and 10-30 μm/hr in all directions.

One of the key features of the Soraa reactor is its internal heating. This circumvents the material property limitations of a conventional ammonothermal chamber. The Soraa reactor is also claimed to be cheaper and easier to scale than traditional autoclaves, which are fabricated from nickel-based superalloys. The capsule that contains the raw materials – seed crystals, a polycrystalline GaN nutrient, a mineraliser and ammonia – is surrounded by a heater, followed by a ceramic shell providing structural support and thermal isolation, and finally an externallycooled outer shell providing mechanical confinement.

Thanks to low conductivity of the ceramic, the steel shell can remain below 200 °C at an operating temperature of 750 °C, allowing it to maintain high creep resistance.


Skin tones under typical 80 CRI blue-based LED, left, compared to 95 CRI 95 R9 Soraa Vivid LED, right

GaN crystals grown in this reactor are transparent but yellow, due to residual impurities, and have a dislocation density below 104 cm-2. These crystals have been sliced into wafers that have been used as the foundation for the growth of InGaN/GaN heterostructures. The intensity and full-width half maximum of the photoluminescence produced by this sample is virtually identical to that emitted by an identical structure formed on HVPE-grown GaN. However, electroluminescence emitted by the former structure is complicated by a yellow luminescence emanating from the substrate.

Although engineers are now working to try and improve the transparency and luminescence properties of their boules, it will be several years before Soraa is making its own substrates for device manufacture. “We’re not able to call the exact date,” explains Devine. “It will depend on the application. The specifications for lasers are not as stringent as those for LEDs. We have a laser business, and we could begin to use our own substrates earlier in our laser manufacturing, than for LEDs.” Solid-state lighting, however, will continue to provide the majority of the company’s revenue.

During the next 12 months Soraa will ramp its production of the MR 16 lamp family, while moving into expanded flood lamp configurations.

“As you go into 2015 and 2016, the possibilities broaden,” explains Devine. “For what the next area after the flood lamps will be, we need to do some more market research, as well as go through market acceptance of our flood lamp products.”
One option for broadening the portfolio is to launch a replacement for the 60 W incandescent.

This is not a trivial move, though, according to Devine, because it would mean a transition to a market where the price of light is more highly valued than its quality. “A market like that, which is a pure commodity, is a couple of years away from when it’s going to be the right time for us to approach it.”

Further reading
D. Ehrentraut et. al. Jpn. J. Appl. Phys. 52 08JA01 (2013)
M. Cich et. al. Appl. Phys. Lett. 101 223509 (2012)