Shining a Light on Solid-state Green LEDs: Could C-GaN Be the Answer?

March 07, 2022 by Dale Wilson

High-efficiency, solid-state green LEDs remain an elusive goal for semiconductor manufacturers. Research funded by ARPA-E targets these LEDs using standard complementary metal-oxide-semiconductor (CMOS) processing tools.

October 9, 2022, will mark the 60th anniversary of the invention of the first visible-light, solid-state LED by Nick Holonyak, Jr. 

On top of this, we are almost 30 years removed from the invention of the blue light LED. Yet, even though LEDs can generate light at both ends of the visible spectrum directly, the green middle of the spectrum remains effectively out of reach for solid-state, direct-emission LEDs. 


Visible light spectrum

Visible light spectrum. Image used courtesy of ThoughtCo


Dr. Can Bayram from the University of Ilinois at Urbana-Champaign is hoping to finally break through the barriers facing green LEDs under a recent ARPA-E grant. Aiming to understand this feat better, AAC had the opportunity to talk with Dr. Bayram and dig into this initiative, fittingly taking place at the Holonyak Lab.


Creating Green LEDs Is Really Hard

Currently, green light is typically generated by adding phosphors to blue LEDs to shift the frequency output to the longer green wavelengths. 

This indirect emission process can be inefficient and prone to degradation over time, impacting the ability to create the white light used in standard lighting applications and making it difficult to create colors using the RGB additive color-mixing process accurately. 

We asked Dr. Bayram what has made creating direct-emission green LEDs so difficult.  

“I would say the material engineering and the limits on how we can get an alloy with the right bandgap.” 

Researchers have tried adding more aluminum to red LEDs and more indium to blue LEDs; however, excess doping creates many problems, including lattice mismatch and polarization that creates an internal electric field. 

Most notably, these LEDs are very inefficient. 


Substrate Engineering to Create Cubic GaN

When approaching this research, Bayram’s team begins with silicon substrate wafers and uses only standard CMOS processing tools. The crystalline nature of silicon allows them to etch small, triangular trenches into the silicon.

“This is a new form of material synthesis that relies on precise substrate engineering,” Bayram explained. 

They then grow hexagonal gallium nitride or GaN (h-GaN) onto the exposed open faces of those trenches. However, h-GaN can be inefficient for light generation. 

What they really need is cubic GaN (c-GaN), where the cubic nitrides are known to be better than hexagonal, but they do not form naturally with epitaxial processes.

As these h-GaN surfaces grow together, the magic happens—the h-GaN growth automatically transitions to c-GaN. 

“What we are doing is creating artificial boundaries in the silicon patterning, forcing the hexagonal GaN into cubic GaN,” Bayram noted, “so that we have 100% cubic materials rather than a mixed phase of cubic and hexagonal.”


Novel Green LEDs for the Ultimate Solid-State Lighting. Credit: Richard Liu. Image used courtesy of Richard Liu and the University of Illinois


Bayram provides a helpful visual for this process by explaining that if you hold a cube in your hand and rotate it to the correct angle, you see a hexagon. 

“So long as we have a perfect angle, you are going to transition these two hexagons into a cube,” Bayram explained. “We do that in crystallography.”

You can try it yourself. Rotate a cube so that one corner is directly facing you. The perimeter, as you see it, is a hexagon.


High Yield, Low Cost: Volume Manufacturing Green LEDs

Much of the work in this field has focused on using smaller, more expensive substrates like 100-150 mm sapphire wafers. Using standard silicon substrates will allow Bayram’s team to manufacture the green LEDs in high volume and at low cost on 300 or even 450 mm wafers. 

Under this program, they are partnering with fabs that will process their technology on 300 mm wafers. When asked if their process would require adding any specialized equipment to the processing lines, Bayram laughed and said, “they don’t allow us to change anything in the CMOS fabs.”


Green LEDs for Energy Savings and Health Benefits

Estimates vary, but lighting accounts for about 15-20% of the global electricity consumption and approximately 5% of global greenhouse gas emissions. 

Bayram believes that using efficient, direct-emission, solid-state green LEDs in lighting could reduce the lighting-related energy needs and greenhouse gas emissions by approximately 25%. 


Northwestern Europe lights at night as seen from the ISS

The lighting of northwestern Europe as seen from the International Space Station. Image used courtesy of NASA


Not only is this a huge waste of power, but it is also a health problem. 

The LED lighting we use today is shifted towards the blue spectrum and typically gets worse as the green phosphor-coated LEDs age. This blue light impacts sleep patterns, wildlife, and possibly even the health of our eyes.


Measuring Success for Green Solid-State LEDs

The program, entitled “Green Light Emitting Diodes for the Ultimate Solid-State Lighting,” will run for three years. 

When asked about their goals, Bayram listed several, including:

  • > 30% efficiency at 100 A/cm2 current density, up from the 13% available today
  • < 30% efficiency droop at higher injection currents, compared to the typical values of 70% available now
  • 555 +/- 15 nm wavelength
  • < 5 nm spectral drift

Overall, Bayram stated that “even if you reach about half of these targets” for efficiencies and injection currents, these green LEDs would “still be a huge improvement over the state-of-the-art.” Additionally, they could be useful in virtual and augmented reality headsets, high bandwidth optical interconnects in data centers, and displays. 

However, if Bayram and his team are successful, perhaps one day we will use these solid-state, direct emission green LEDs to light our homes and offices. 

1 Comment
  • B
    Boggart March 11, 2022

    “Currently, green light is typically generated by adding phosphors to blue LEDs to shift the frequency output to the longer green wavelengths.”

    Have I just woken up in an alternate universe or something? Direct emission green LEDs are readily available in both the 565nm and 525nm bandwidth regions, they don’t use phosphors, they are extremely common, in fact. I sell thousands of them a year.

    525nm range green LEDs are some of the brightest direct emission LEDs you can get, they have the highest output ratings of any LEDs on a per watt basis, they vastly outshine blue, orange, yellow, red etc. Yes, 565nm LEDs are more difficult to make bright, but they have improved massively in the last few years, I sell a Chinese 3W 525nm power LED that puts out 120 lumens at 700mA, pretty good for that wavelength.

    Of course, phosphor based LEDs are brighter, as you would expect, but the article is very vague about which wavelengths it’s actually talking about, and just keeps stating “green LEDs” instead of stating specific wavelengths.

    And I don’t know what they are going on about as far as general lighting is concerned, LED lighting mostly uses white LEDs, which are based on blue LEDs (mostly, some use UV) with a broad spectrum phosphor coating to produce the selected colour temperature and CRI. Almost no white LED lighting uses separate red, green and blue LEDs to produce white light as the output is too peaky and so CRI is crap.

    The article needs a rewrite, and the writer probably needs to get out of their lab and see what the Chinese have been doing for years.

    Like. Reply