Advance brings low-cost, bright LED lighting closer to reality

Advance brings low-cost, bright LED lighting closer to reality
Timothy D. Sands, at left, director of Purdue's Birck Nanotechnology Center in Discovery Park, and graduate student Mark Oliver, operate a "reactor" in work aimed at perfecting solid-state lighting, a technology that could cut electricity consumption by 10 percent if widely adopted. Inside the reactor, a material called gallium nitride is deposited on silicon at temperatures of about 1,000 degrees Celsius, or 1,800 degrees Fahrenheit. Purdue researchers have overcome a major obstacle in reducing the cost of the lighting technology, called light-emitting diodes. (Purdue News Service photo/David Umberger)

Researchers at Purdue University have overcome a major obstacle in reducing the cost of "solid state lighting," a technology that could cut electricity consumption by 10 percent if widely adopted.

The technology, called light-emitting diodes, or LEDs, is about four times more efficient than conventional incandescent lights and more environmentally friendly than compact fluorescent bulbs. The LEDs also are expected to be far longer lasting than conventional lighting, lasting perhaps as long as 15 years before burning out.

"The LED technology has the potential of replacing all incandescent and compact fluorescent bulbs, which would have dramatic energy and environmental ramifications," said Timothy D. Sands, the Basil S. Turner Professor of Materials Engineering and Electrical and Computer Engineering.

The LED lights are about as efficient as compact fluorescent lights, which contain harmful mercury.

But LED lights now on the market are prohibitively expensive, in part because they are created on a substrate, or first layer, of sapphire. The Purdue researchers have solved this problem by developing a technique to create LEDs on low-cost, metal-coated silicon wafers, said Mark H. Oliver, a graduate student in materials engineering who is working with Sands.

Findings are detailed in a research paper appearing this month in the journal Applied Physics Letters.

LEDs designed to emit white light are central to solid-state lighting, semiconducting devices made of layers of materials that emit light when electricity is applied. Conventional lighting generates light with hot metal filaments or glowing gasses inside glass tubes.

The LEDs have historically been limited primarily to applications such as indicator lamps in electronics and toys, but recent advances have made them as bright as incandescent bulbs.

The light-emitting ingredient in LEDs is a material called gallium nitride, which is used in the sapphire-based blue and green LEDs, including those in traffic signals. The material also is used in lasers in high-definition DVD players.

The sapphire-based technology, however, is currently too expensive for widespread domestic-lighting use, costing at least 20 times more than conventional incandescent and compact fluorescent light bulbs.

One reason for the high cost is that the sapphire-based LEDs require a separate mirrorlike collector to reflect light that ordinarily would be lost.

In the new silicon-based LED research, the Purdue engineers "metallized" the silicon substrate with a built-in reflective layer of zirconium nitride.

"When the LED emits light, some of it goes down and some goes up, and we want the light that goes down to bounce back up so we don't lose it," said Sands, the Mary Jo and Robert L. Kirk Director of the Birck Nanotechnology Center in Purdue's Discovery Park.

Ordinarily, zirconium nitride is unstable in the presence of silicon, meaning it undergoes a chemical reaction that changes its properties.

The Purdue researchers solved this problem by placing an insulating layer of aluminum nitride between the silicon substrate and the zirconium nitride.

"One of the main achievements in this work was placing a barrier on the silicon substrate to keep the zirconium nitride from reacting," Sands said.

Until the advance, engineers had been unable to produce an efficient LED created directly on a silicon substrate with a metallic reflective layer.

The Purdue team used a technique common in the electronics industry called reactive sputter deposition. Using the method, the researchers bombarded the metals zirconium and aluminum with positively charged ions of argon gas in a vacuum chamber. The argon ions caused metal atoms to be ejected, and a reaction with nitrogen in the chamber resulted in the deposition of aluminum nitride and zirconium nitride onto the silicon surface. The gallium nitride was then deposited by another common technique known as organometallic vapor phase epitaxy, performed in a chamber, called a reactor, at temperatures of about 1,000 degrees Celsius, or 1,800 degrees Fahrenheit.

As the zirconium nitride, aluminum nitride and gallium nitride are deposited on the silicon, they arrange themselves in a crystalline structure matching that of silicon.

"We call this epitaxial growth, or the ordered arrangement of atoms on top of the substrate," Sands said. "The atoms travel to the substrate, and they move around on the silicon until they find the right spot."

This crystalline formation is critical to enabling the LEDs to perform properly.

"It all starts with silicon, which is a single crystal, and you end up with gallium nitride that's oriented with respect to the silicon through these intermediate layers of zirconium nitride and aluminum nitride," Sands said. "If you just deposited gallium nitride on a glass slide, for example, you wouldn't get the ordered crystalline structure and the LED would not operate efficiently."

Using silicon will enable industry to "scale up" the process, or manufacture many devices on large wafers of silicon, which is not possible using sapphire. Producing many devices on a single wafer reduces the cost, Sands said.

Another advantage of silicon is that it dissipates heat better than sapphire, reducing damage caused by heating, which is likely to improve reliability and increase the lifetime of LED lighting, Oliver said.

The widespread adoption of solid-state lighting could have a dramatic impact on energy consumption and carbon emissions associated with electricity generation since about one-third of all electrical power consumed in the United States is from lighting.

"If you replaced existing lighting with solid-state lighting, following some reasonable estimates for the penetration of that technology based on economics and other factors, it could reduce the amount of energy we consume for lighting by about one-third," Sands said. "That represents a 10 percent reduction of electricity consumption and a comparable reduction of related carbon emissions."

Incandescent bulbs are about 10 percent efficient, meaning they convert 10 percent of electricity into light and 90 percent into heat.

"Its actually a better heater than a light emitter," Sands said.

By comparison, efficiencies ranging from 47 percent to 64 percent have been seen in some white LEDs, but the LED lights now on the market cost about $100.

"When the cost of a white LED lamp comes down to about $5, LEDs will be in widespread use for general illumination," Sands said. "LEDs are still improving in efficiency, so they will surpass fluorescents. Everything looks favorable for LEDs, except for that initial cost, a problem that is likely to be solved soon."

He expects affordable LED lights to be on the market within two years.

Two remaining hurdles are to learn how to reduce defects in the devices and prevent the gallium nitride layer from cracking as the silicon wafer cools down after manufacturing.

"The silicon wafer expands and contracts less than the gallium nitride," Sands said. "When you cool it down, the silicon does not contract as fast as the gallium nitride, and the gallium nitride tends to crack."

Sands said he expects both challenges to be met by industry.

"These are engineering issues, not major show stoppers," he said. "The major obstacle was coming up with a substrate based on silicon that also has a reflective surface underneath the epitaxial gallium nitride layer, and we have now solved this problem."

Source: Purdue University

Explore further

Silica paradox: Scientists discover seemingly 'impossible' material

Citation: Advance brings low-cost, bright LED lighting closer to reality (2008, July 17) retrieved 25 May 2019 from
This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no part may be reproduced without the written permission. The content is provided for information purposes only.

Feedback to editors

User comments

Jul 17, 2008
Excellent article. I am particularly re-assured by the statement about the real world engineering issues. It sounds like they have actually given good thought about the manufacturing side.
I recall that they can sometimes use a buffer layer to over come differences in the thermal expansion coefficient that acts like a paint primer.
Now just sell/donate the idea to HP and we can get these devices in the shops. That way we can stop using haz mat mercury CFLs.

Jul 17, 2008
what happened to those leds we read about a few months back that were about the size of a tic tac and produced as many lumens as a 600 watt street lamp in a spectrum about that of the sun? I need some for my grow room.

Jul 17, 2008
I dont think those were LED but more like the lamps being used in the Panasonic LIFI system.

Jul 17, 2008
DGBEACH, you seem to be the resident LED and electrical expert. Thoughts?

Jul 18, 2008
"Two remaining hurdles are to learn how to reduce defects in the devices and prevent the gallium nitride layer from cracking as the silicon wafer cools down after manufacturing. "

I'm no scientist or anything but wouldn't gradually reducing the temperature reduce the chance of cracking?

Jul 18, 2008
The best LED fixtures on the market are actually about twice as efficient as compact fluorescent fixtures now, with 65 lumens per watt system efficacy. See

Jul 18, 2008
This is good news... The combination of LED and LCD screens could be cheaper...

Jul 18, 2008
Love to hear new research on pushing LEDs to the forefront of lighting technology. One of the biggest problems in LEDs is Numerical Aperture due to total internal reflection. Much of the produced light is reflected due to the angle at which it is emitted. Using Snell's Law, the emission angle can be calculated rather easily. For a semiconductor with refractive index around 3.5 and air around 1, the angle is around 19 degrees.

What this all really means is that the created light struggles to get out of the chip. Many diodes have internal quantum efficiencies (how many photons produced for input power essentially) near 100% but that light is scattered in the device. The best LEDs out there use techniques like shaping the emitting region and putting it on a reflector. One of those techniques is the TIP or Truncated Inverted Pyramid. It changes the angle at which light rays are incident upon the air/semiconductor interface and improves the external efficiency. A rear reflector will help the external efficiency by having some more light reflected but you're still losing some to scattering and heat. It will be neat to see how they eventually pattern and package them because that's how the real efficiency is extracted.

Being able to grow the LED on a silicon wafer is a wonderful accomplishment because the infrastructure for high volume silicon processing is already available. Silicon is really cheap and has good thermal conductivity. One of the techniques certain groups use when growing strained epitaxial layers is a patterned etch. Think about the expansion cracks on a sidewalk. A 300mm wafer has to be diced after growth so one could theoretically etch or pattern a crosshatch that would allow the differing expansion rates to occur and improve the dicing. A little more difficult on (111) silicon than (100) due to the propensity to cleave in triangles instead of squares but that's all possible to overcome.

Last comment; on the high efficiency LIFI source. That source is a blackbody radiator whereby an RF source excites a gas mixture forming a plasma (if I recall correctly). They used their 'puck' to control and confine the plasma and to conduct heat away. Probably more valuable for street lamps and the aforementioned grow room than home lighting. Apparently they are already going into high end TVs.

Jul 18, 2008
Two remaining hurdles are to learn how to reduce defects in the devices and prevent the gallium nitride layer from cracking as the silicon wafer cools down after manufacturing.

Due to the myriads of defective devices obtained during the manufacturing process, many must be made just to produce one, thus the higher price.
Those made successfully must afterwards be sorted into different lots according to their actual emitted wavelengths (white has many hues it would seem), another expensive process. Another example of a device which requires this type of sorting is crystal resonators, which must be sorted according to their thermal stability characteristics, making the lower temp ones more expensive.
And for everybody's information dimming of an LED is accomplished by turning it on and off very quickly, with the amount of on time deciding how bright it "appears" (Pulse Width Modulation). This requires circuitry which is more complex than that needed to dim an incandescent...even more cost.

So as you see, trying to reduce the costs of LEDs would involve many different levels of the supply chain, but it inevitably will happen, and I'll be right there fighting to help make it happen because it is something which I believe will help us in the long run.

Jul 18, 2008
Use of LEDs will reduce heat significantly, causing a reduction in air conditioning costs. However, in the winter climates, an increase in heating costs will result.

Jul 19, 2008
joefarah, I recently saw a study about that. In colder climates the use of CFLs actually *increases* carbon emissions during the winter because of increased heating through natural gas. I suspect that the efficiencies on LEDs are high enough to make up for the difference, unlike the craptacluar CFLs:P.

Jul 20, 2008
Heat energy created from resistance has an efficiency ratio of under 1. While heat pump air conditioning can acheive a ratio of about 3. This is because the heat pump transfers heat from outside the heating area (outside) into the heating area. Therefore the loss of heat generated from the incandecent light bulb would be generated by the heat pump more efficiently so overall there would be a reduction in energy used.

Jul 20, 2008
I think you may have missed the point aussie... how many places use heat pump heaters?:P I've never even seen one. I've seen 3 types of heaters in common use: 1)natural gas, 2)electric, 3)wood/coal burning. This isn't "how efficient could a home be theoretically", because if we're talking about that, then we can just skip CFLs and move straight to LEDs anyway:P... or maybe we can pretend that each home can skip over all that primitive, inefficient solar/wind/geothermal/fusion garbage and instead having its own self-contained Quantum Chamber to generate electricity! Woot! No, what I was talking about was, "How efficient a home or business is in actual observed conditions in real life".

Jul 21, 2008
Well about 600,000 heat pump air conditioners sell each year in Australia, a country which has a population of 22,000,000.

Jul 21, 2008
I just want to know which company is going to bring the first $5 LED bulb to market, so I can buy lots of their stock NOW!

Jul 21, 2008
Pretty much my entire neighborhood uses heatpumps.

I don't know where they're getting this 'need for saphire' idea. Most efficient LEDs use good old GaAs/N/P substrates, which are still super expensive; Not to mention any process that needs an MBE growth is going to be expensive.

What ever happened to the light emitting nanowires. Those guys were dirt cheap to make and efficient, but it seems they never figured out implementation?

Jul 24, 2008
Heat pumps are NOT commonly used as PRIMARY HEATING units in the world as a whole. When they are, you can talk about the efficiency of a heat pump verses something else. But since most of our heat comes from electric resistance, wood, coal, or natural gas, those are what must be compared to right now. Geesh. I don't know why that is such a difficult concept to grasp.

What you're saying is exactly the same as saying something like "electric cars don't produce distributed pollution, therefore smog controls on cars aren't necessary, only smog controls on powerplants are necessary". That statement would be true, if EVs were what the majority of people in the world were CURRENTLY driving.

Jul 29, 2008
I am thinking that mass producing LEDS onto silicon wafers - will result in a new type of dimming technique where dimming is produced by turning off many of the light emitting components on a sliding scale. This will make the LED light emitters closer in use to the incandescent bulbs.

Coupled with increased number of emitters per component this should generate a very smooth overall effect. (Even if it is digital)

Heat pumps are very efficient and I have seen a company selling them for household water heaters as well as air conditioners. The Household heat pump water heater used less electricity than the solar water heater (which does not heat without sunlight). What I don't understand is why such efficiency has been ignored and Gov's still promote the solar hot water and disregard the heat pump hot water.

The heat pump water heater cost less than off-peak hot water heaters as well so seems a good idea to me to use anywhere where a houshold is connected to electricity at all.

For those of you still burning timber for heat - (I hope it is out of necessity instead of choice)

Jul 30, 2008
For those of you still burning timber for heat - (I hope it is out of necessity instead of choice)

In order to heat my house (4 kids, Canadian winters,you do the math) it costs over $1800 using just electricity...or $800 on 10 cords of hardwood (usually maple). However I don't drive my car to work (140kms daily), but bus it in, so I don't add much more to the atmosphere than someone who heats electric but drives in.

Oct 08, 2008

In order to heat my house (4 kids, Canadian winters,you do the math) it costs over $1800 using just electricity...or $800 on 10 cords of hardwood (usually maple). However I don't drive my car to work (140kms daily), but bus it in, so I don't add much more to the atmosphere than someone who heats electric but drives in.

The problem with wood heat is the combustion products are much worse than just about any other fuel. Paticulates from burning wood are considered by some to be a major cause of the polar sea ice melting.

Please sign in to add a comment. Registration is free, and takes less than a minute. Read more