U.S. Department of Energy - Energy Efficiency and Renewable Energy

Building Technologies Office – Information Resources

Text Alternative Version: Life-Cycle Assessment of Energy and Environmental Impacts of LED Lighting Products

Below is the text-alternative version of the "Life-Cycle Assessment of Energy and Environmental Impacts of LED Lighting Products" webcast, held March 28, 2013.

Terry Shoemaker: Good morning, ladies and gentlemen. This is Terry Shoemaker with the Pacific Northwest National Laboratory. Welcome to today's webcast: Life Cycle Assessment of LED Lighting Products brought to you by the U.S. Department of Energy Solid State Lighting Program. We will now begin the presentation. Brad, please begin.

Brad Hollomon: All right. Thanks. Give me a second until we get the screen up and everything. But I am going to begin this presentation; it's really going to be in two parts. We have been engaged in the life cycle assessment of energy, environmental impacts of LED lighting products, and two parts of this project have actually taken place and the reports have been published and so on, and so I'm going to cover those sort of by way of context and background for today's piece de resistance, which is going to be Jason Tuenge's presentation of the third part, which is the end-of-life analysis of solid state lighting products. And so I will be as brief as I can, but I hope to provide at least the kind of overview to set the context for, as I say, Jason's presentation.

And basically we started I think with a chart from the Department of Energy that arose out of some controversy in the literature about what the impacts or directed impact – direct and indirect of solid state lighting technology might be, and I think that in the interest of what I best characterize as due diligence, the Department of Energy was interested in making sure that, first of all, what it was promoting is that efficient lighting technology had no hidden dangers and that there was no – nothing that was likely to be harmful, and if there were problems to make sure that they would be addressed through research and development technology programs as the technology of the industry progressed. And then finally, to be able to inform particularly state and local solid waste authorities, since in many instances they're the ones that have to weigh potential cost of benefits from an environmental point of view at the end of the product's life.

But anyway, what we started to do was review the existing literature on energy and environmental assessments of lighting options, and that was performed by Navigant Consulting with a team of people, which included some folks in Carnegie Mellon University, which had been investigating this topic for some time, so it wasn't really starting from scratch.

The next thing we wanted to do after reviewing the literature and looking particularly at energy impact was to expand the evaluation to go back into the manufacturing process and to look at not only the impacts in terms of energy, but the environmental impacts of producing all of the materials that went into these products and the materials that went into those materials and so on, ad infinitum, and also to look at a wider variety of different impacts, not just energy, but to look at things like potential for the chemical oxidants, for acidification, for global warming, and for a number of things that one would want to consider in terms of the overall sustainability of the product.

And then finally we wanted to focus in on what happens ultimately at the end of the life of these products, when sometime in the future somebody is going to dispose of them. And so basically what we've got here is energy and environmental impacts, and we wanted to start particularly in terms of the life cycle with the manufacturing, getting these things from one place to another, in both shipping, packaging, and so on. And their use, which obviously mostly impacts on energy consumption, and also, as I said, the other dimensions of environmental impact, which include what goes ultimately into the landfill, but also what goes into the air as you're making raw materials.

The first part focused on energy, as I said. And then part two is somewhat more comprehensive, and part three we'll be hearing from Jason about the at the end of life.

The first portion of this, the literature review and energy evaluation was completed in February 2012. There's a website there where you can find it. In fact, you can find all of the reports relating to this project at that website. You have to look down a little bit on the screen, the technical reports, but you'll see them all there, including some summaries. So I would recommend that.

And the – basically what this report did was – I'm sorry if that's too early – was to, drawing on a number of different existing studies, work out the energy consumption based on a sort of standard functional unit of 20 million lumen hours, which is effectively the output of a standard incandescent light, so that the energy consumption is basically on an apples-to-apples, megajoule-to-megajoule basis.

And this is what they came up with. It shows that for the energy consumption in megajoules for 20 million lumen hours is, obviously it's considerably higher for incandescents, including both standard and halogens, and lowered by a factor of several for compact fluorescent lights and LEDs. And then we've also got a project of what LEDs might represent, assuming progress in technology, particularly efficiency of these devices, which does appear to be improving. The half point – 2015 – was a projection. I believe you can now buy LEDs which are more efficient than the ones that they sold using the data in 2011.

As you can see, the material manufacturing, the package manufacturing and so on and transport are all quite small; they're almost invisible as components of the energy consumption. But in terms of the literature that was available at that time, one of the larger uncertainties was the energy that went into manufacturing. Although it's a small number to begin with, it was possible that some people felt that the manufacturing might amount to the 25% . And so the, you know, as I say, the LED lamps are similar and the incandescent lights are a good bit more energy-intensive.

And the takeaway from this is that the most important contributor to the energy consumption is the consumption and use of the power that goes through the filament or the tube that really dwarfs everything else. And, you know, as I said, there was a bit of uncertainty about the consumption in manufacturing and so that really became part of the focus of part two, which, covered not just the manufacturing, but also the supply of the inputs that go into the process.

So this is the next, the next part was the – the portion completed, again, was June of 2012. You can find it on the website. And it shows a much more detailed assessment of the manufacturing process that LED products use in that process. One of the things that we were able to do to take advantage of some work that Michael Shawn, the author, had done in the U.K. previously, and so he was up to speed on the manufacturing and also had fairly good connections within the industry, but not necessarily prepared to share everything. But he was able to get some basic information and the report was reviewed by several industrial representatives who were able to at least say that there was nothing that as far as they were willing to say was clearly wrong with his characterization of manufacturing.

And so he also did include a comparison with the projected characteristics of every lamp that might be available in 2017, so he basically bracketed the technology with the central workings of manufacturing and production the way it would be done today or perhaps the year before the report came out, and where it might be in 2017 if the DOE programs and others are successful.

The procedure that we followed is an emerging standard one. This life cycle analysis is something which is a kind of an approach which perhaps is a little bit more advanced in Europe than it is in this country and one of the reasons we were able to tap the resource that Michael represented is having done some work in the U.K., there is an ISO – there is an ISO standard associated with it, which is marked on the graph here. But the point is this is not something that we invented from scratch; it's something which we could do following the protocol that leads to results that hopefully would be consistent with what we want other people to come in the field.

And part of this, of course, is deconstructing the manufacturing and the production process, and that requires an understanding of impacts of all of these steps along the way, not just the manufacturing of the lamp itself, but also the aluminum that goes into the heat sink and all of that kind of thing. And this is – because this is a fairly evolved kind of approach in Europe, we drew upon a database, the Ecoinvent database, which is the Swiss Center for Life Cycle Inventories. And it represents an accumulation of a wide variety of different studies, as you would imagine there would have to be, and a set of parameters and so on, some of which have been reached by scientific consensus. And so we felt that it was appropriate to draw on that since it was perhaps more highly developed there than in other places.

So as we show on this slide, these are the assumptions about the LEDs. And basically what we're doing is we're reflecting what's there is 2011 technology. These things start with sapphire wafers and this is, again, that's an evolving kind of technology. The sapphire wafers for the semiconductor devices have a nice advantage of having a very high delicate constant, and at the same time they have high thermal connectivity, which makes them useful for keeping things from overheating. And there's a technology that is improving – let me get this out of the way, I guess – is improving. As the wafers become larger they become more efficient and it becomes more efficient to use them to develop – to use them to make the LED devices themselves.

And here are the rest of the assumptions about the lamps that we're comparing. And this is standard incandescent; fairly standard CFL; a current technology LED lamp, which may as I say, be a little bit obsolete at this point; and then there is an advanced one, which anticipates some improvements in technology. So in a way what we're doing is kind of bracketing the impacts of the LED lamps, including, you know, what we can be pretty sure is current technology, warts and all, but what you can possibly expect in the future.

And again, we're looking at primary energy consumption, which reflects pretty much the same – a little bit refined, but in somewhat the same kind of conclusion that you would draw from the part one study. And again, we've got the CFLs and the LED – current LEDs being roughly comparable and then the advanced LED being considerably better and the incandescent light is way off in the distance – a distant fourth.

Again, energy-in-use is a dominant consideration, and the CFL is perhaps slightly more harmful than the integrated ballasted LED on procuring it, because of hazardous waste to the landfill. And so the easiest way to represent this is with a rather interesting spider web here, which basically looks at all of the different environmental impacts and with the impacts normalized to the impacts of the incandescent light. So the outer perimeter of this spider web is basically the normalized impact of the incandescent light in terms of specific for electricity, ecosystem damage, land use, all of the things which go into this life cycle analysis procedure that has been, as I said, developed considerably in Europe, and then it compares the other lamps, which are represented by the lines in the center.

Now one of the things, since it's not clear that the incandescent is perhaps a little bit of a straw man at this point, because incandescents are, you know, you kind of know about them and they're in some ways being phased out anyway. So let's take out a magnifying glass and look at the same results that would just be CFLs and the few technologies for LEDs included, and you can see, relatively speaking, where they stand; the LED is a little bit better than the CFL in all respects except for hazardous waste landfill, which we believe is primarily due to the heat sink. The LED requires a fairly hefty heat sink. As technology improves, the heat sink may get smaller because of less waste heat to get rid of. And the other thing is that maybe we're considering ways of trying to recycle these devices because – in part because it would be possible to recycle the aluminum heat sink and avoid those impacts, and it may be easy to justify economically, since the aluminum is fairly valuable.

So that – we're postponing the questions I guess at the end of the session, but I'd be certainly happy – someone in the background is compiling them and is also perfectly welcome to submit them. And so what I would invite now is that you put away your magnifying glass and pick up your mass spectrometer and I'd like to turn things over to Jason Tuenge, who is the – who will be presenting the end-of-life analysis. And I want to thank you very much for listening.

Jason Tuenge: Okay. Thanks, Brad. Can you hear me? I think you're already muted. Okay. Please make sure you mute. Thank you, everyone, again for joining us for this third part of this study. We decided to take a closer look at the end-of-life considerations for these products as you go to dispose of them.

We decided to focus on elements that are used to determine whether a waste is going to be classified as being hazardous or not for this study, and the study was just now published, just posted to our website at the link given here. So you can take a look at that maybe later today or say over the weekend, if you get into that sort of thing.

Let me know if my slide does not advance here. You should be on to the next slide now.

For this study we looked at, we focused on regulations and test procedures used by the EPA at the federal level. Also looking at the State of California, which has more stringent criteria in place and some test methods of their own. All of these procedures involve grinding up products, so when you drop the product into acid for testing you aren't just directly dropping it into the acid; you're taking the thing apart, grinding it up, milling it really finely, as shown in the photos below, to maximize the surface area of all the bits within the product to make them environmentally available. This is a conservative, worst-case form of testing, but it's primarily done in this manner to make the testing very consistent and repeatable. So there is definitely a very good reason for doing this.

We followed precedent set by the State of California in a 2004 report where they set up some specific procedures for electronic devices, as well as special handling of lamps containing mercury, such as CFLs. We didn't actually do the testing ourselves, at DOE or at PNNL, but contracted to independent laboratories that are qualified to do this type of work.

Looking at the particular test methods that were used, there are three primary test methods. The first two shown here are used in California, whereas the last one is used both in California and at the federal level. So method 3050 and WET are used in California. 3050 is basically a complete digest; for pretty much anything you drop in the acid, it's going to get dissolved completely, so long as it's ground up beforehand. Anything aside from basically sand or glass will get digested when you do this. Whereas the WET is intended to be more like what happens in a landfill, where you have an acid pass over something and it will cause some things to be digested and carried away, but not everything; that's what actually happens to the landfill. They use both evaluations when testing products and they look at 17 different elements and compare the concentrations of these elements relative to thresholds or limits, the first one being the total threshold limit concentration, TTLC, and that's for 3050 versus the soluble threshold limit concentration, the STLC that you get from WET testing.

At the federal level they don't look at all 17 of these elements, they only look at 8 of them. And at the federal level they just look at what is soluble, what would actually happen in a landfill. And this test does differ, though, from the WET, it's a different procedure. And here we abbreviate the criteria as FRL for Federal Regulatory Level. One more thing to notice really quickly is the different units. The total digest is milligrams per kilogram, whereas the other two are milligrams per liter.

Looking at the actual regulations, the restrictions, we've got the federal criteria in a darker yellow. So arsenic, barium, cadmium, chromium, lead, mercury, selenium – I'm probably pronouncing that wrong – silver, those are all restricted at the federal level as well as in California, whereas in California they also look at antimony, beryllium, cobalt, copper, etc. The criteria you can see differ widely. And then when it looks the same between the STLC and the FRL for where the criteria overlap. For instance, looking at arsenic, just because the value numerically is the same doesn't mean it's the same restriction, because the test methods differ.

So this is kind of a clumsy illustration just to give you a sense of what's going on with these different test methods. The one on the left is the TTLC in California. That's the total digest shown in red, to basically indicate that you're going to be getting pretty much everything digested. You're going to get, you know, if there's any copper in the sample, all that copper is going to get extracted and you're going to have it all in solution, and so it's kind of a hot solution you might say.

The STLC does not get everything out necessarily; it typically only gets a portion of say a copper out of the material. And then if you run the TCLB you actually get typically a lower concentration than with the WET because the volume is different; the volume is basically twice what you'd have with the WET. I'm simplifying here, but this basically gives you a sense of what's going on. The California criteria are going to be more stringent in the federal as a rule, and you can't really compare these values directly across, but you can use the method 3050 results to estimate the maximum possible concentration that you could have from WET or TCLP testing.

This flowchart gives you an idea of – or an overview of how we conducted the testing. Basically all samples were subjected to method 3050 after they were photographed, weighed, broken down into components, and then milled separately as components. So method 3050, we've got data for all the lamps that we tested. As I mentioned before, you can estimate the maximum possible concentrations through the other tests using the method 3050 data, the test results. And so you can save yourself some trouble, some testing, if you can tell from the method 3050 results that there's basically no say silver in that sample. If it doesn't show up in method 3050 it's not going to show up in the other tests either. And so we basically step through these tests in that manner, saving time and money on testing if we knew that there was nothing there to be found.

Some of the products, if they did not fail the TTLC, that first criteria in this in California, if it didn't fail that threshold, it didn't exceed it, then they still have a pretty high level of that material, we would continue testing, going on to the WET test to look at the STLC. And then we'd do the same thing there. If it exceeded the STLC we'd stop testing at that point and say "There's too much in this thing." Otherwise if levels are fairly high but not too high, then we would move on to the third test at the federal level and see what it looked like there.

We evaluated – we obtained 22 lamps. These represented 11 different models, so two samples of a given model on average. But you can see this is varied a little bit; we actually tested four samples of one of the incandescent lamps, and that's the INC-1, and just one of the CFLs, the CFL-3. And this is just to allow us to do some comparisons across laboratories and within a laboratory using different techniques. So it gives a little bit of flexibility by doing this.

These lamp types were either omnidirectional, emitting light in all directions more or less uniformly, like you'd usually see from a typical light bulb, or directional, more like the bulbs you'd be sticking into a recessed can or a recessed downlight in your ceiling. So they fell into one of those two categories; we wanted to see both.

These are the basic characteristics of the omnidirectional lamps, just the basic bulbs, showing that they were selected to be interchangeable. You get comparable light output from any of these. Obviously if you're going to be installing a CFL or an LED you expect some energy savings. If the wattage is lower they're more efficacious. They naturally have different lifetimes, you know, so the CFLs and LEDs also feature longer life in addition to lower wattage, but the light output is the same, or at least comparable.

Also showing which lamps were disassembled and which weren't. we did experiment with a couple of these to see what the effect, if any, was of not disassembling and separating these things into different components, but just grinding the whole lamp all into one pile, rather than piles separated by components. In the case of directional lamps, they were all disassembled before being milled into piles of dust. But again, these are all similar in terms of output, but you might notice they're actually lower in output than the omnidirectional; we compare them separately or treat them separately.

We acquired all our samples – we started testing over a year ago, or about a year ago, I guess at this point, but actually acquired samples over a year ago. And then with just a couple of exceptions; in one case we ran out of material for one of the products as we were going through and doing testing, so we got another sample so that we could get another component from and continue testing on that component. Another one, basically it was actually an LED product, again, but in this case as far as we could tell a piece had just disappeared, almost as though somebody had hit it with a hammer and that piece just went somewhere. And basically the weight didn't look right, photos didn't identify the missing piece, so we just got a replacement sample. And it was acquired actually relatively recently, in October.

So here is an overview of the results looking at TTLC. Again, TTLC or the method 3050 was the test that was applied to all products; it was the first step in the process. The other ones we don't necessarily have data on because we might have stopped testing after running 3050. So looking at this chart, we get a good sense of how these products stack up against each other. Incandescent, covering both just old incandescent technology and the newer halogen is shown in red, whereas CFLs are in blue, LED is in green, and you can see that it's fairly spiky because each one of these bars, these skinny columns, represents a single sample, a single lamp.

So you can see basically a number of these elements that are evaluated, this is all 17, and the ones that are evaluated at the federal level or are considered at the federal level are indicated with an asterisk. When you go through the elements that are indicated with an asterisk we actually look pretty good. This is a log scale, so the dotted bar indicates the California threshold for the element. At, you know, 100% of that threshold you can see as you step through the elements with the asterisks, so arsenic, barium, cadmium, etc., most of them are down at 10% or less of the relevant threshold, and so are not really – you know, wouldn't be expected to be in danger of failing for that element or exceeding the threshold for that element and being classified as hazardous waste.

There are a few exceptions; probably the most common one was actually for lead in some of the CFLs. Otherwise where you see the threshold being approached or exceeded, there's a handful of these, and it actually happens to pretty much all the different lamp types, mostly with CFLs and LEDs for antimony, but then all lamp types for copper. There's also some approaching or slightly exceeding for nickel and a handful or most of them exceeding for zinc.

One thing to note is that even though we gave special attention to mercury using specialized test procedures for this element, apparently its volatility caused it to escape detection. It's not exactly clear necessarily what exactly the cause of this was, but it seems pretty clear that the levels that we detected in terms of mercury were not what we should've found, and we've actually found, as indicated in the report, that we were not the first to run into this problem, although we pretty clearly come in lower than expected for mercury. And again, that's for the CFLs, where we know that some small amount should be present, but we did not even see that small amount.

The tables on the next several slides help to give a little bit better of a breakdown in terms of which samples were problematic, were closest to the thresholds, and which – they also summarize the other tests that were performed, the results from the other tests that were performed, so we can see how to compare it versus the STLC and the FRL.

Some trends or apparent trends do emerge. Again, copper was, you know, looking first tier, thinking of lamps, we look at copper, the incandescent lamps actually fairly consistently exceeded the California threshold for copper. A couple of the halogens did exceed for nickel and also had one of them exceed for zinc. Looking at the CFLs, again, we've got – just looking at the California criteria, we've got antimony, copper, nickel, zinc being exceeded at the California level; not even evaluated at the federal level. But then at the federal level we see what might be a trend in terms of lead, where they didn't all fail outright, but when we continued testing found that some of these lamps were exceeding these solubility thresholds. And in some of the cases where there was a question mark, we had run out of material and just weren't able to continue testing to confirm whether or not there was actually an issue with that criterion.

Looking at the LEDs we see a similar pattern, except in the case of lead, where there wasn't really much found. But copper, again, something of a pattern, as well as with zinc. Looking at the components we see that basically what we're finding, the source of the problem, it's not necessarily the LED light sources themselves; typically it's not, it's going to be some of their components. And this goes for the other product types as well. We've got screw bases or drivers or ballasts; these are the things that are causing these products to exceed thresholds.

We've compared our results with some other testing, which is actually very limited. When we started testing nobody else had really done this kind of testing on these products, but then right before we finalized this report another one was published by University of California, and our results compare pretty well with them. As well as when you compare with other types of electronic devices, there's definitely more data here. We're looking at things like cell phones and basically found that these LED products actually are pretty comparable to other types of electronic devices; they are electronic devices themselves. For instance, the drivers.

One thing to note is that testing by others often excludes things. We included all components in our testing. And it's worth noting that when they tested these other electronic devices they found that they also exceeded these California thresholds.

So with this diagram, this basically illustrates some of the same patterns we saw with our testing. We did not conduct any testing of cell phones as part of our testing.

The report goes into quite a bit more detail than you're going to see in this presentation, discussions of things like homogeneity and where that can introduce some error. We also make it very clear that this is an exploratory test and very limited in its scale. Even though we tested 22 samples, this can't really, you know, it's not enough to give you an adequate characterization of a given product or a given technology; you can't say one of these lamps would fail or that LED or CFL would fail as an overall market characteristic. And if it doesn't work that well you've got to do much more testing and much more structured testing to get a definitive answer on these things.

It's also worth noting that we only looked at elements; there are other things that are included in the regulations that were not evaluated here. And I want to reinforce that this testing is intended for – this type of testing is used for classifying waste, not for testing products since they're being ground up and being dropped in acid, that does not – you know, that's not the kind of testing you would do to get a sense of whether it's safe in terms of handling, you know, just when you're installing it in the ceiling or something. This is not informative in that regard.

So we also took a quick look at what happens when you consider the effect of lamp life. Basically if you're, you know, it's possible that you might have a product that lasts 1,000 hours and has a low concentration of elements in it, but then needs to be replaced every 1,000 hours, versus a product that has somewhat higher concentrations, but lasts a long time and doesn't need to be replaced for that same period. So we made some quick adjustments to give you a sense of what would happen if you were to just look at the total amount of the given element that's being deposited in a landfill over time and show that in this chart here. The story doesn't change all that much; there are a few exceptions, such as in the case of nickel, where incandescent did not do as well once you look at it in this light.

So to wrap things up, we found that, again, across all these models, lamps did pretty well relative to federal regulations, but then we found in looking at the California criteria that all of the lamps exceeded at least one of these criteria applicable in California, where they've got more stringent criteria. Against these same criteria you have things like cell phones also failing, so it's good to keep that in perspective.

Again, the culprits causing the lamps to fail were not the LED light sources themselves. This is not something specific to LED technology. It was typically some other component, including in the other lamps, you know, it might've just been the screw base that was causing the problem.

And then again, want to keep in mind that, you know, electronic waste, e-waste is already supposed to be recycled to help divert these materials from going into the waste stream, and based on this testing, it appears that, you know, LED lamps; CFLs, which incorporate electronic devices, electronic components, should probably be treated in a similar manner.

And in closing, whereas parts one and two of the study showed that the most important thing with these products is to look at – the critical thing is to look at the overall picture, not just one aspect; not just the material content or the end-of-life aspect, but to look at the overall lifecycle impacts, which were reported on extensively in parts one and two. And what they showed was that the most important feature of a product is its efficacy, you know, how much energy is it going to use while it's in use. You know, but that said, you do want to also consider, or you need to consider at some point what's going to happen at end-of-life. And so we can really start taking a look at things like recycling to see what we can do, this will improve the overall life cycle performance of these products as well as removing some of these elements from the waste stream. And it's going to become more important as these products gain traction in the market.

That's probably all I can address at this point. I want to thank everyone again for joining us.