Announcement

Collapse
No announcement yet.

LEDs and hot electrons

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • LEDs and hot electrons

    http://www.theregister.co.uk/2013/04...fficient_leds/

    One of the problems with using LED-based lamps to replace incandescent or fluorescent lamps is that they're expensive: not only do they need more electronics than the alternatives, LED efficiency is capped by a fall in light output at higher current.

    To answer the so-what question: getting rid of the “droop effect” allows LEDs to be made more efficient – and that would make them cheaper. But to eliminate the droop, you first need to find out why it happens.

    Enter a group of scientists from the US Department of Energy-sponsored Centre for Energy-Efficient Materials at the University of California, Santa Barbara. Working with colleagues from France's École Polytechnique, they have watched what's going on at an atomic level and concluded that the “droop” is down to what's known as an “Auger recombination” process.

    The process has been known for some time, and is well-understood in semiconductors, but its status as a candidate for LED droop has been controversial since it was first proposed in 2011. The Auger effect describes how an electron falling from a high energy level to an inner-shell state can release energy as a photon, or transfer its energy to another electron that leaves the atom.

    As Wikipedia explains, in Auger recombination, an electron and a hole recombine and energy is transferred to an electron in the conduction band. The result is that some of the electrons energy arriving in the LED's drive current is given off as heat instead of light.

    Observing the Auger recombination taking place should pave the way for more efficient LED designs, the researchers explained to Phys.org.

    One of the researchers, professor James Speck said the findings “will enable us to design LEDs that minimise the non-radiative recombination and produce higher light output.”

    The researchers observed the phenomenon using electron emission spectroscopy to see what was taking place in a gallium nitride (GaN) LED. The surface of the LED was prepared so as to allow the researchers to directly measure the energy spectrum of electrons emitted by the device.

    “The signature of Auger electrons is observed through high energy peaks which appear in the electron energy distribution curves (EDCs) at high injected current densities. The Auger electron current is found to correlate with the simultaneously observed droop in emission efficiency”, they write in their paper, available at Arxiv.
    The paper is also to be published in Physical Review Letters. ®

    Bootnote: In the original version of this story I wrote "electrons" where I meant to write "energy". Thanks for pointing out the error. ®

    This is interesting, because hot electrons are (or were anyway) one of the issues facing cutting edge semiconductor design 10 years ago. Essentially at higher energy levels, the electrons don't behave as they are supposed to, they generate heat - a lot of it - and this damages the materials they are supposed to travel through.

    In semiconductors - the desired behavior is movement while in LEDs the desired behavior is the emission of a photon.

    The above article talks about loss of lighting power, but damage to the underlying semiconductor device would be an issue as well.

    Fixing this problem is non-trivial. On the semi side - the emphasis was never on fixing it, but rather identifying areas where it would be a potential problem and changing the current flows such that the damage would not affect device longevity.

  • #2
    Re: LEDs and hot electrons

    Originally posted by c1ue View Post
    http://www.theregister.co.uk/2013/04...fficient_leds/

    This is interesting, because hot electrons are (or were anyway) one of the issues facing cutting edge semiconductor design 10 years ago. Essentially at higher energy levels, the electrons don't behave as they are supposed to, they generate heat - a lot of it - and this damages the materials they are supposed to travel through.

    In semiconductors - the desired behavior is movement while in LEDs the desired behavior is the emission of a photon.
    Actually, a "hot electron" is not necessarily something that generates heat at all. And in this system they are usually behaving exactly like they're "supposed to." They don't (directly) damage the material, unless you're talking about electromigration, which depends on the number of carriers, and not their momentum, and thus still only connects to hot electrons very tangentially.

    And LEDs ARE made of semiconductors.

    Just ask Britney Spears.

    Originally posted by c1ue View Post
    The above article talks about loss of lighting power, but damage to the underlying semiconductor device would be an issue as well.

    Fixing this problem is non-trivial. On the semi side - the emphasis was never on fixing it, but rather identifying areas where it would be a potential problem and changing the current flows such that the damage would not affect device longevity.
    A hot electron is just one that has high momentum (is traveling rapidly in the semiconductor). In indirect band gap semiconductors like silicon, this can be a big deal, because the dispersion relation (E-k diagram) is such that a higher momentum state may have a lower energy than the 0-k state. As such, interband scattering is limited only by kinetics, and not energetics.

    However, in the direct band-gap semiconductors used for optoelectronic devices, this MAY result in interband scattering only when kinetic energy reaches the indirect gap energy, which is higher than the band gap. Only then can thermalization be affected. But devices are simply designed not to operate in this regime in any event. That isn't even particularly hard for LEDs, in which the switching speeds are inconsequential.

    Gallium Nitride has a wide and direct bandgap. That's why you can make LEDs for lighting out of it. Hot electron losses are not the big problem.

    The work is interesting because of what it reveals about the nature and consequences of Auger recombination, which IS a big problem, not only for LEDs, but more so for lasers (it makes it harder to sustain the population inversion, and causes heating).
    Last edited by astonas; 04-25-13, 12:01 AM. Reason: Spelling of Britney Spears' name.

    Comment


    • #3
      Re: LEDs and hot electrons

      Originally posted by astonas View Post
      Actually, a "hot electron" is not necessarily something that generates heat at all. And in this system they are usually behaving exactly like they're "supposed to." They don't (directly) damage the material, unless you're talking about electromigration, which depends on the number of carriers, and not their momentum, and thus still only connects to hot electrons very tangentially.

      And LEDs ARE made of semiconductors.

      Just ask Brittany Spears.



      A hot electron is just one that has high momentum (is traveling rapidly in the semiconductor). In indirect band gap semiconductors like silicon, this can be a big deal, because the dispersion relation (E-k diagram) is such that a higher momentum state may have a lower energy than the 0-k state. As such, interband scattering is limited only by kinetics, and not energetics.

      However, in the direct band-gap semiconductors used for optoelectronic devices, this MAY result in interband scattering only when kinetic energy reaches the indirect gap energy, which is higher than the band gap. Only then can thermalization be affected. But devices are simply designed not to operate in this regime in any event. That isn't even particularly hard for LEDs, in which the switching speeds are inconsequential.

      Gallium Nitride has a wide and direct bandgap. That's why you can make LEDs for lighting out of it. Hot electron losses are not the big problem.

      The work is interesting because of what it reveals about the nature and consequences of Auger recombination, which IS a big problem, not only for LEDs, but more so for lasers (it makes it harder to sustain the population inversion, and causes heating).
      LEDs are to lightbulbs as transistors are to vacuum tubes.

      Remarkable how long it is taking to get useful light out of silicon.

      Comment


      • #4
        Re: LEDs and hot electrons

        Originally posted by EJ View Post
        LEDs are to lightbulbs as transistors are to vacuum tubes.
        This is a fantastic analogy! It applies on several different levels.

        Originally posted by EJ View Post
        Remarkable how long it is taking to get useful light out of silicon.
        Well, yes and no. And then yes again. ;)

        Since it is an indirect bandgap semiconductor, one really wouldn't expect to get light out of silicon. As a narrower bandgap material, even if you could, it would be too long wavelength (low energy photons) for lighting.

        But on the other hand, if one wanted to get light out for, say, optical interconnects for direct, high-speed communication between integrated circuits in a computer, that could conceivably be achieved with the use of photoelectrochemically etched nanoporous silicon.

        The problem is that this couldn't switch fast enough for IC-level communications, and would generate too much heat (both problems are intractable, as they stem from the same properties that enable light emission).

        I think I've solved this problem, but am still early in the process of verifying it experimentally, and haven't yet disclosed it. To maintain patentability, I probably shouldn't go into too many more details here. It is one of four potential applications (in surprisingly varied fields) for the technology I am developing in my startup.
        Last edited by astonas; 04-25-13, 12:02 AM.

        Comment


        • #5
          LED's vs Temperature

          Is this Auger problem separate from the temperature problem? LED's lose efficiency when the junction temperature rises above 40C. I thought this was just higher temperature---> more phonons ---> more inelastic collisions between phonons and electrons.

          Since increasing the current/light output raises the power dissipation and temperature, you would expect efficiency to decrease at high current densities.

          Comment


          • #6
            Re: LEDs and hot electrons

            Originally posted by astonas View Post
            This is a fantastic analogy! It applies on several different levels.



            Well, yes and no. And then yes again. ;)

            Since it is an indirect bandgap semiconductor, one really wouldn't expect to get light out of silicon. As a narrower bandgap material, even if you could, it would be too long wavelength (low energy photons) for lighting.

            But on the other hand, if one wanted to get light out for, say, optical interconnects for direct, high-speed communication between integrated circuits in a computer, that could conceivably be achieved with the use of photoelectrochemically etched nanoporous silicon.

            The problem is that this couldn't switch fast enough for IC-level communications, and would generate too much heat (both problems are intractable, as they stem from the same properties that enable light emission).

            I think I've solved this problem, but am still early in the process of verifying it experimentally, and haven't yet disclosed it. To maintain patentability, I probably shouldn't go into too many more details here. It is one of four potential applications (in surprisingly varied fields) for the technology I am developing in my startup.
            I learned just enough to be dangerous from the former CEO of Luminus Devices and from a good friend, MIT grad, and Managing General Partner of a VC firm here in town who has the material science and chip expertise on the team.

            One major disadvantage of a chip over a filament is that light emits 360 degrees from a filament whereas a chip emits light at a narrow angle in one direction making even dispersion problematic. The fact that light emits more broadly from the flat surface of the chip versus from a sharp point of light of a filament makes it hard to confine the light to a narrow beam to project it long distances. Cooling is inherently problematic and most of the innovations that have produced high intensity LEDs are in heat dissipation. The key advantages of LEDs is that 80% to 90% of the input energy goes into generating light output versus 80% to 90% heat output in the case of incandescent bulbs, and LEDs can last for decades versus months.

            Best of luck with your invention.

            Comment


            • #7
              Re: LEDs and hot electrons

              Originally posted by EJ View Post
              . . .

              One major disadvantage of a chip over a filament is that light emits 360 degrees from a filament whereas a chip emits light at a narrow angle in one direction making even dispersion problematic. The fact that light emits more broadly from the flat surface of the chip versus from a sharp point of light of a filament makes it hard to confine the light to a narrow beam to project it long distances.
              .

              This is a big issue for bicycle headlights. The best of them use both a reflector and a lens to focus the beam.

              A parabolic reflector with the filament at the focus produces parallel rays---good for headlights.

              It is more difficult for LED's, but the combination of lens and reflector can make a tightly focused beam.

              http://www.ebay.com/itm/CREE-XML-XM-...item460e90c41e

              These do a good job.

              Comment


              • #8
                Re: LEDs and hot electrons

                Originally posted by astonas
                Actually, a "hot electron" is not necessarily something that generates heat at all. And in this system they are usually behaving exactly like they're "supposed to." They don't (directly) damage the material, unless you're talking about electromigration, which depends on the number of carriers, and not their momentum, and thus still only connects to hot electrons very tangentially.

                And LEDs ARE made of semiconductors.
                You might note that I never said LEDs weren't made of semiconductors.

                What I said was that an LED was seeking to generate light by charging up an electron, and having said electron slip to a lower energy level thus generating a photon.

                This is different than standard semiconductors where electrons are primarily used as electrical current.

                Sure, LEDs do also use electrons to generate electrical current, but this is secondary to what they are intended to do: generate light.

                Originally posted by astonas
                A hot electron is just one that has high momentum (is traveling rapidly in the semiconductor).
                Completely wrong. The reason they're called 'hot' is because the recombination of holes and electrons generates heat. The heat absolutely does generate damage, and furthermore this damage tends to be inside the carrier material.

                For those not familiar - a 'hole' is a gap where an electron used to be. In semiconductors, the 'hole' can be used in a similar fashion to the missing electron. Think of a hole as a bubble going through water, while an electron might be a bowling ball rolling down the lane. When a bowling ball rolls into a bubble, the energy either/both were carrying tends to convert to heat.

                Electromigration in contrast damages material by moving ions of said material to a different location. This is a completely different mechanism - it is caused by fast moving electrons transferring momentum to the ions of the conductor, causing said ions to move and corresponds to what you note above.

                You have a lot of knowledge in certain areas, but don't assume you have greater knowledge of semiconductors. This is what I did for a long time.

                Originally posted by Polish Silver
                Is this Auger problem separate from the temperature problem? LED's lose efficiency when the junction temperature rises above 40C. I thought this was just higher temperature---> more phonons ---> more inelastic collisions between phonons and electrons.
                I doubt it. The process of charging up the electrons is a more standard semiconductor device function - and temperature affects those performance curves. In particular, the hotter a semiconductor device gets, the more leakage occurs in the devices, which in turn degrades the expected behavior.
                Last edited by c1ue; 04-25-13, 03:03 PM.

                Comment


                • #9
                  My experience with Hot electrons

                  This article agrees with my experience and understanding of hot carriers in semiconductors.

                  1) In mosfets, hot carriers cause increased body current, which is undesired.

                  2) In mosfets, they are believed to gradually change the threshold voltage, by creating defects at the oxide interface. This is undesired.

                  3) Hot carriers have a velocity comparable to (or greater than) the saturation velocity. Most devices are designed to work with the carriers well below the saturation velocity.

                  The hot carrier current increases exponentially with the drain-source voltage, above a threshold which depends on the channel length.
                  It is the greatest when the gate voltage is about 1/2 way between the drain and source voltage. It is small when Vgs~0V or when Vgd ~ 0V.
                  Last edited by Polish_Silver; 04-26-13, 08:33 AM.

                  Comment


                  • #10
                    Re: My experience with Hot electrons

                    Originally posted by Polish Silver
                    This article agrees with my experience and understanding of hot carriers in semiconductors.
                    The article talks about hot carrier injection. This isn't the hot electron effect.

                    The wiki even notes:

                    That is, high temperatures caused by the effect are unrelated to the phrase "hot electron effect".

                    Comment


                    • #11
                      something completely different!

                      Originally posted by c1ue View Post
                      The article talks about hot carrier injection. This isn't the hot electron effect.

                      The wiki even notes:
                      Hot carrier injection! Something completely different! Or at least, somewhat different.

                      Comment


                      • #12
                        Re: My experience with Hot electrons

                        Originally posted by c1ue View Post
                        The article talks about hot carrier injection. This isn't the hot electron effect.

                        The wiki even notes:
                        Clue, you haven't seen all the conversations in the select section (and hence, some of the background and skill that Polish_Silver and others display there) so I'll just give a bit of friendly advice.

                        This is the time to admit that you don't understand the science, apologize for the debate "tactic" of trying to bluff your way through by referring to your semiconductor experience, and move on.

                        Please. It's nicer when the debate stays civil.

                        Comment


                        • #13
                          Re: LEDs and hot electrons

                          Originally posted by EJ View Post
                          Best of luck with your invention.
                          I'd like to second that astonas and offer additional good wishes on your patent applications.

                          It's nice to see that there's still a segment of the population trying to grind it out in the Producer economy, through the utilization of new technology and good old innovation - mixed with sweat equity.

                          One of our companies is awaiting 4 international patents. Been there before, but it can sometimes feel glacial and frustrating. I did a presentation last night on the saviour and servitude elements of future technologies to a business crowd. The presentation was only 1 hour long, but I stayed for 2 and a half more hours to extinguish all the questions. I was a little shocked at the level of interest and high quality of the questions. The "kid's table" as they're known (students) in the club - peppered me to death with interest, knowledge, and even questions regarding starting-up high-tech businesses. I left feeling a little bit more optimistic about the future.

                          Best wishes on your venture.

                          Comment


                          • #14
                            Re: LEDs and hot electrons

                            Originally posted by EJ View Post
                            I learned just enough to be dangerous from the former CEO of Luminus Devices and from a good friend, MIT grad, and Managing General Partner of a VC firm here in town who has the material science and chip expertise on the team.

                            One major disadvantage of a chip over a filament is that light emits 360 degrees from a filament whereas a chip emits light at a narrow angle in one direction making even dispersion problematic. The fact that light emits more broadly from the flat surface of the chip versus from a sharp point of light of a filament makes it hard to confine the light to a narrow beam to project it long distances. Cooling is inherently problematic and most of the innovations that have produced high intensity LEDs are in heat dissipation. The key advantages of LEDs is that 80% to 90% of the input energy goes into generating light output versus 80% to 90% heat output in the case of incandescent bulbs, and LEDs can last for decades versus months.

                            Best of luck with your invention.
                            Thank you very much! There's no doubt that small businesses need all the luck they can get these days!

                            The problem of light extraction from raw GaN to air being limited (by total internal reflection) to a ~22 degree cone is indeed a big one.

                            But that too, could be addressed much better than it is in the industry today. My graduate work, for example, resulted in a surface-shaping mechanism that permitted far higher light extraction than even today's commercial LEDs achieve. The raw die increased in forward extraction by a factor of 5. (I patented the method, and it was licensed by a major player who didn't understand how to do it properly, and wound up killing it.)

                            So there's still LOTs of good new ideas worth pursuing in that regard, and it confuses me that the industry still insists on iterating the older ones, and killing R&D efforts on the newer ones to save costs. I can't see their project-level balance sheets and income statements, of course, but there are a still a few tricks that could be tried today that might even reduce LED unit cost, if they were only taken more seriously by the players involved. The engineering is very sound, and for certain approaches the financials should be as well.

                            If my current efforts don't pan out quickly enough, I suppose that is an area of research I could return to. It wouldn't take much effort to extend my graduate work in a way that resulted in a new and separate patent (providing a sustainable advantage), since I've been mulling the problem over in the back of my mind since then. I've even got some initial experiments designed.

                            As far as cooling is concerned, you are right that it is a major issue. But it is not the more fundamental issue. A better answer is to generate less heat in the first place. An ideal LED should generate light with far less heat than today's LEDs, just as they do in the GaAs, InP, and other direct bandgap semiconductor systems. The problem in the GaN system is that heating occurs mostly because of all the non-radiative recombination paths that exist in the material, due to both point and line defects (threading dislocations). To address THAT requires innovation in epitaxial growth technology, which is being pursued very heavily indeed, as well as substrate technology, which is being pursued by some startups stemming from the place I did my graduate work. There's some very real promise there, but cost remains an issue at this point.

                            By the way, a company started by one of my graduate advisors, (they're taking the-reduce-dislocations-via-better-materials approach) just released its first LED product!

                            Comment


                            • #15
                              Re: LEDs and hot electrons

                              Originally posted by Fiat Currency View Post
                              I'd like to second that astonas and offer additional good wishes on your patent applications.

                              It's nice to see that there's still a segment of the population trying to grind it out in the Producer economy, through the utilization of new technology and good old innovation - mixed with sweat equity.

                              One of our companies is awaiting 4 international patents. Been there before, but it can sometimes feel glacial and frustrating. I did a presentation last night on the saviour and servitude elements of future technologies to a business crowd. The presentation was only 1 hour long, but I stayed for 2 and a half more hours to extinguish all the questions. I was a little shocked at the level of interest and high quality of the questions. The "kid's table" as they're known (students) in the club - peppered me to death with interest, knowledge, and even questions regarding starting-up high-tech businesses. I left feeling a little bit more optimistic about the future.

                              Best wishes on your venture.
                              Thanks, Fiat, for your support and advice! It is always good to know that there are some who find a path through to make it all work.

                              I am intrigued by your presentation on the savior and servitude elements of future technologies. This is not a phrase I'm familiar with, and while I think I can extrapolate what you mean by it, I would be interested in learning more. Do you have a link or file that you wouldn't mind sharing, or would you perhaps be willing to summarize?

                              As far as the "kid's table" goes, I'd say you're absolutely right. There's certainly some great talent around, if you know which rocks to look under. And entrepreneurship is a culture that is making its way into graduate schools everywhere in a big way. One of the things that makes it possible to do my own work on a shoestring budget is a close collaboration with a research university, and so I get to interact with the "kids" quite a bit. There's obviously a range, but the good ones give a great deal of hope.

                              Comment

                              Working...
                              X