GBA Logo horizontal Facebook LinkedIn Email Pinterest Twitter X Instagram YouTube Icon Navigation Search Icon Main Search Icon Video Play Icon Plus Icon Minus Icon Picture icon Hamburger Icon Close Icon Sorted
Green Building News

A Reprieve for the Incandescent Bulb?

Researchers have tweaked conventional incandescents to make the bulbs far more efficient

Standard incandescent light bulbs are famous energy hogs, but a change in design can make them even more efficient than fluorescents and LEDs. University researchers have successfully developed a "proof-of-concept" bulb using the new technology.
Image Credit: Jeff Kubina / Creative Commons license / Flickr

UPDATED January 14, 2016

The incandescent light bulb, maligned as an epic waster of energy, could be revived with a change in design that boosts their efficiency, university researchers said.

In a news release, the Massachusetts Institute of Technology said that researchers at MIT and Purdue University collaborated on a new design that could prove to be more efficient than either fluorescent lamps or LEDs.

The new bulb uses a metal filament, just like a conventional light bulb. But instead of allowing radiant energy to escape, as it does in a conventional bulb, the radiant energy is reflected back to the filament where it’s re-absorbed and re-emitted as visible light. MIT described these “secondary structures” surrounding the filament as a type of photonic crystal composed of abundant materials and made with conventional material-deposition technology.

In the U.S., conventional incandescent bulbs are being phased out under energy-efficiency regulations because as much as 95% of the energy they consume is wasted, mostly in the form of heat. Fluorescents and LEDs are rapidly taking their place, even as some consumers grouse about their cost or the quality of light they produce.

MIT said that one characteristic of a lighting source is luminous efficiency, a metric which takes the reaction of the human eye to light into account. The luminous efficiency of a conventional incandescent bulb is between 2% and 3%, with fluorescents measuring between 7% and 15% and most commercial LEDs between 5% and 20%. These new bulbs have efficiencies as high as 40%.

Results of the research were reported in Nature Nanotechology. (The report is behind a pay wall.)

Still in the lab

Don’t run out to the store quite yet. The first “proof-of-concept” bulbs produced by researchers don’t approach the 40% efficiency they believe the redesigned bulbs are capable of. Yet the measured 6.6% luminous efficiency is still three times as high as a conventional incandescent bulb, and as good as some fluorescents and LEDs.

The team is “not seriously contemplating” commercialization at this point. The point of the research, the team said, was to “understand the science of how (and to what extent) thermal emission from high temperature sources can be tailored through this concept of light recycling, which might have many potential applications for energy conversion.”

But the research is very promising. The team succeeded in designing a crystal that works for a variety of light wavelengths and angles. The stack of layers can be assembled so that the desired wavelengths pass through and out of the bulb, while the infrared wavelengths are reflected.

One of the researchers said the technology might be applied to a number of other products.

“LEDs are great things, and people should be buying them,” said Marin SoljaÄić of MIT, one of the researchers. “But understanding these basic properties” about the way light, heat, and matter interact and how the light’s energy can be more efficiently harnessed “is very important to a wide variety of things.”

Dr. Ognjen Ilic, one of researchers, added some information from the report comparing the experimental light bulb to a standard incandescent, a compact fluorescent lamps (CFL), and a LEDs lamp.

A typical incandescent produces about 15 lumens of light per watt (lum/W), compared to the 40-100 lum/W for a compact fluorescent (CFL) 40-140 for a replacement LED (state-of-the-art LED chips have higher ratings). The experimental bulb produced an estimated 45 lum/W.

In terms of the percentage of electricity that is converted into light, an incandescent is about 5% efficient, compared to 15% to 20% for CFL and 20% to 25% for an LED replacement lamp. The proof-of-concept bulb is about 16% efficient and in principle could go much higher.


  1. JC72 | | #1

    True story.
    We've been living in our townhome for 15 yrs (bought new) and all of the bulbs (incandescent) in the canister lights, bathrooms, hallways (2) and two bedrooms are as old as the house.


  2. GBA Editor
    Martin Holladay | | #2

    Response to Chris M
    And even though you are a GBA reader, you were never tempted, during those 15 years, to try a CFL or an LED?

  3. JC72 | | #3

    Response to Martin
    Well we have used CFL's for the bulbs which were replaced (couple of table lamps, garage light) and our kitchen light has always been fluorescents which have been replaced numerous times.

    With just the two of us, the reality is that we just never leave lights on and don't spend any time in some areas of the house. It's a 3 bdrm dwelling.

    I've been tempted to replace the row of 8 clear incandescent bulbs above a mirror in the master bathroom simply because of all the heat they throw off but only 2 have been replaced in the 15 yrs since we moved in and the little lady has said no.

    CFL's have their place for areas where occupants need light for an extended period of time.

  4. cussnu2 | | #4

    Those wonderful CFLs that
    Those wonderful CFLs that cost exponentially more and last 1 10th of the time they are purported to last and as a wonderful environmental benefit they include trace amounts of mercury. Anybody with half a brain would know that bulbs get broken regularly but hey lets spray our homes with mercury....and BTW that heat incandescents throw off is not wasted during the heating season. All of this and the light is still not as good as the original. Thanks for the wonderful progress.

  5. Expert Member
    Dana Dorsett | | #5

    Laughable statistics, not supported by facts (response to #4)
    Seriously- I haven't read that level of disinformation about CFLs in almost a decade.

    Exponentially higher cost? Show your math! (An exponent of what, pray tell? Show the formula.)

    Lifecycles 10% of what's advertised? Let's see the field monitored data.

    Bottom of the line CFLs have some "infant mortality" issues, but in general those that last past the first month go the distance unless power quality in the neighborhood is exceptionally poor, or the thing is being cycled on/off a LOT.

    A mercury exposure problem?

    There's enough mercury in an old-school wall thermostat to build between 1000-5000 CFL twistys. If you're unfortunate enough to break one, the exposure to the humans cleaning it up is roughly the equivalent of eating 5 tuna sandwiches, unless they're silly enough to be cleaning up the mess with their tongue.

    The amount of mercury emitted into the environment for the additional power generation necessary to produce an equivalent amount of light with an incandescent is orders of magnitude higher in areas with significant coal-fired generation than burying all the CFLs in landfills at end of life. (And in most states CFLs and other fluorescents are disposed of more benignly than that, separated from the regular trash stream for processing as mercury contaminated waste.)

    Heating with resistance electricity (from incandescent lights or other) is usually several times more expensive than any alternatives with the possible exception of propane in the highest propane cost markets. Though not "wasted" , it's not exactly doing the budget any good. Heating with cheaper (and usually greener) BTUs from other sources would be preferable. The air conditioning load from incandescent lighting adds another ~25% to the operating cost of the bulb during the cooling season. Put into perspective, just the air conditioning power use of an incandenscent would power it's CFL-equivalent in summer.

    CFLs are on the way out, being displaced in the marketplace by LEDs, but the cost (environmental or financial) has never been anywhere near what the popular press critics have claimed. The "You can have my incandescent bulbs when you pry them from my cold dead (if burned) hand" crowd is getting a bit long in the tooth, but they're still around, I s'pose. I personally know a guy who stockpiled a barn full of old school incandescent bulbs a couple of years ago just to make sure he was covered for life- they're definitely out there.

  6. STEPHEN SHEEHY | | #6

    I chuckle when liberal environmentalists get blamed for the phase-out of incandescent bulbs. It was signed into law by President Bush, after passing the House and Senate with substantial support from both parties. It was largely written by the light bulb manufacturers.
    Sort of like those Obama phones that have been around since the Reagan administration.

  7. DavidAlex | | #7

    resurgent incandescent
    Like so many technologies that are announced via MIT press release, I don't believe this one will ever see the inside of a big box store. Or any other commercial enterprise, for that matter. It's a lab toy and always will be.

  8. charlie_sullivan | | #8

    35 years late
    This is the same technology that was developed by Duro-Test in 1977-1980--their "heat mirror" lamp, which got 30 lm/W in the production version. The difference is that the new result uses more layers to make a higher performance reflector, and uses a flat reflector instead of spherical, presumably to make the deposition of the layers easier.

    For more on the heat mirror lamp:

    The Duro-Test lamp was constrained by practical considerations like lamp life, and the degradation of the reflector by evaporated tungsten deposited on it. This new work locates the reflector close to the lamp where it would degrade faster. The paper mentions that that could be a problem but doesn't make any estimate of the lamp life. In essence, a close reading of the paper shows that they acknowledge that it's just a theoretical examination of the limits of this 38-year-old strategy, not a breakthrough or new idea, and they don't claim to have evaluated what would be feasible considering constraints such as lamp life.

    Meanwhile GE, Osram, and perhaps others have been making lamps using this approach, with simpler lower-cost reflectors. GE calls it "halogen IR" or HIR; Osram calls it IRC (IR Coating). GE has been doing this since about 2000. The results are in the 20-30 lm/W range. Presumably they understand the cost vs. efficacy trade-off involved in the design of the filter and the choice of the number of layers in it, and have found that for higher efficacy it's more cost effective to go with other technologies--increasingly with LEDs.

    The PI on this research paper has a history of using MIT press releases to generate hype that vastly exceeds the novelty of his work. The Nature Photonics paper, and the page put up by the post-doc Ognjen Ilic, both acknowledge that this is neither a new idea or a practical light source. It's unfortunate that MIT and the PI choose to hype it way beyond its real significance. This is particularly sad because we are just at the point where cheap LEDs are readily available in higher efficacy than CFLs, and are being rapidly adopted by consumers. This hype could slow that transition. Hopefully it will soon be forgotten.

  9. user-349933 | | #9

    I like my LEDS but a certain percentage seem to fail.

  10. AlanB4 | | #10

    This would theoretically
    This would theoretically produce 100lm/w, todays 60W eq LEDs produce 60lm/W. In 5 years LED bulbs will probably exceed 100 lm/W (Cree has a 303 lm/W prototype at 90CRI iirc) and this will still be in the prototype stages if they are lucky. In 10 years its likely we will have over 200lm/W LEDs and if lucky these incandescent may be marketable by then.
    We actually have 200 lumen/watt LEDs today, but only at 1W and $5-15/chip (a bulb needs chips (usually multiple), electronics, heatsink, diffuser, etc). Cree XP-L, XHP50 and XHP70 were released within about the past year and are used in modern flashlights and available from many sources)
    This is a good idea, but about 10-15 years too late. Also HIR halogen is an established technology that never caught on, same concept, and are still available for some cars with single filament bulbs (they don't work for dual filament high/low beam headlights).

  11. Tim C | | #11

    CREE 303 lm/W
    CREE did not include a CRI in their 303 lm/W announcement. A 90 CRI at that efficiency would be quite incredible; at that CRI the maximum achievable efficiency is around 360 lm/W.

  12. AlanB4 | | #12

    @ Tim
    Your correct, i don';t

    @ Tim
    Your correct, i don';t know where i got 90CRI form

    How did you calculate a 360lm/w theoretical maximum at 90CRI?

  13. Tim C | | #13

    Maximum Efficiency
    No calculation of my own, I just eyeballed it on a graph :)

    It's not actually a theoretical maximum - the theoretical maximum at 90 CRI is about 408 lm/W, using four narrow band emitters (e.g. LEDs rather than phosphor) with red, yellow, green, and blue wavelengths. But I excluded that from consideration because getting even 200 lm/W with RYGB would require significant breakthroughs in yellow and green LED efficiency. It's also worth noting that CRI isn't well designed for narrow band emitters; this approach "games" CRI to a non-trivial extent and would have visibly worse performance than current typical bulbs with similar CRI Ra and R9 ratings.

    The 360lm/W I provided was for a truncated black body spectrum (where CRI is lowered by truncating into the visible spectrum on the extreme blue & red ends). Blue LEDs with green+red phosphor conversion don't exactly follow that spectrum, but the best achievable spectral efficiency ends up being quite similar anyway.

  14. AlanB4 | | #14

    Interesting, can you provide
    Interesting, can you provide some sources for more information or link to these graphs, i am curious what the theoretical efficiency is at 80CRI, given most LED chips have spectral graphs in their datasheets (well Cree does anyways). Basically i am curious what percent efficiency the 80CRI bulbs we can buy today are, given most produce 800 lumens on 10ish watts.

  15. Tim C | | #15

    Sources is a good source on the spectral efficiency of truncated black bodies. is a slideshow, so it can be opaque and confusing at points, but it has an excellent overview of where we're at and what the limits of different approaches are. (Note that it switches between spectral efficiency and wall plug efficiency in different portions).

    If you relax from 90 CRI to 80 CRI, you can't go much higher than 408 lm/W (maybe 415-420) with RGYB. However, more significantly, it does let you hit 400-410 lm/W with just RGB emitters (but unfortunately with garbage rendering of reds, not good for skin tones). With blue LED + phosphor conversion (with or without an additional red LED), the best spectral efficiency we can do is more like 370 lm/W, so that would be a reasonably fair benchmark for retail bulbs.

  16. charlie_sullivan | | #16

    Sylvania ColorCalculator
    Alan, the efficiency from wall plug to optical radiation is about 27% for an 80 lm/W, 80 CRI bulb.

    If you like exploring these ideas, Osram/Sylvania makes software for people using their LED chips that calculates CRI and efficacy for different spectra. It's pretty amazing and it's free.

    I just fired it up and looked at the standard 2700K, CRI = 82 spectrum. That spectrum gives 300 lumens per watt of radiation. So a bulb producing that spectrum would give 80 lumens per watt of electrical power if the efficiency of conversion from wall plug to optical radiation is 27%. That's probably made up of something like 80% efficiency in the driver (power supply) and 34% efficiency for the LED itself.

Log in or create an account to post a comment.



Recent Questions and Replies

  • |
  • |
  • |
  • |