International Colloquia on Thermal Innovations #17: Thermophotovoltaic and Thermionic Technologies

okay um so good morning good evening good day wherever you are and welcome to mit innotherm i'm gang-chen i will be moderating today's panel and my colleagues asking henry john linghard everyone organizer for this qualcomm series and i think some of you probably in the in the series from the beginning and this is actually the colloquial number 17. and today we'll talk about the high temperature direct energy conversion uh including thermal photovoltaic and thermian ionic technology but before we jump in i just want to say we launched this colloquium during the pandemic and uh we're really uh very grateful for people to attend those colloquium we hope to bring people together during this difficult time and we're happy to say that the qualcomm so far has a very good attendance i think avg uh was around the 400 people because we have other 6000 people attending according to a record and we had so far 16 uh colloquial already so somewhere near 400 people on average and also online there are a lot of views uh i uh i'll counter reset recently because we're migrating but if you go to youtube you probably can see somehow some of those uh really have very good review say views so uh so today this is a topic uh i think many of us are interested and uh before we uh jump into introducing the people i want to mention two stories in fact one is a thermal photovoltaic the field was started by dick swanson and dick swanson is say uh we know in u.s they probably the two biggest solar companies one is sun power and he uh started the sun power and so this is the second bit the photo will take so actually uh originally if i claim uh from his working thermal photovoltaics at the time it was a work on sitcom you'll hear more on this and the other is on thermionic technology this uh was there was a com there currently there's a company many of us buy scientific equipment from this thermal fissure and thermal fissure is a combination of two components thermal electron and the fissure right and the thermal electrons started by george and he at the beginning was aiming to commercialize the serbian technology so we have this really great stories from starting commercializing tpv and thermal alex and grow big companies not exactly this technology but today say we're happy to have really leading people discuss the ideas and the current status of this technology and the commercial efforts so i'll introduce our panelists uh andre starting with in order for today's presentation first is andre nana he's currently assistant professor at the university of michigan in chemical engineering department he was a graduate from mit roosevelt lab we're very proud to have him on board and he received the 3m non-tenure faculty award and was named by forbes isis one of the 30 under 30. and his research area is radiative heat transfer energy conversion transport and storage and the nanostructure materials and uh after android we have jarrah and he is the founder and ceo at spark thermonix and he graduated from stanford with a phd and did some really interesting work in combining full time with the thermal elegant effect and he joined the lawrenceburg election lab as part of their first cohort of the cyclotron road and his research area is energy conversion and heat transfer semiconductors and solar devices micro nano fabrication and he will talk about of course thermionics and uh we have also panelists uh um say alejandro he already attended this uh series as a panelist on summer storage and he is an assistant professor at technical university madrid i couldn't really speak very well the spanish so sorry alejandro you taught me but i know he's an electrical engineer applied physics and he has a experience at the tokyo institute technology mit and worked on thermal photovoltaic photovoltaic thermionic and high temperature thermal energy storage so today he will talk about the idea of combining these two technologies so our game plan is we'll have this three speakers each talk about the 15 minutes and then we hope to use 30 minutes for q and a so this is weblog uh on your screen there are q and a button and you can send in your questions along the way and we hope to use more time to discuss after everybody's presentation and uh if you want to have your name mentioned um we'll assume your name i say well mention your name if a question from you but if you want to stay anonymous you can stay so or you can just send the alumnus comments so after this i will post the recording on youtube and if there are questions so we couldn't handle uh as in the past that's like most of times the case uh we will post all the questions on youtube so people can continue the discussion so with that uh let me hand over to andre andre all right yeah yes do you see my screen okay yes all right well thank you gong thank you for having me i'm thrilled to be a part of this series and to be able to share with you work that really grew out of the rosenhower kendall lab at mit and talk about some recent advances in tpv conversion i i want to start off by acknowledging uh that this work that has been done recently in in my group is really in collaboration with steve forrest and one of his students dejofan who were pioneers in developing new tpv devices and also acknowledge my student tobias berger who is almost ready to graduate as well as support for this work from a variety of sources at nsf in particular epmd and thermal transport processes as well as support for some of steve's efforts from the army research office so um since i'm since i'm going first i i will give this very short explanation of how thermal photovoltaics work and if you are familiar with the solar cell and how it works you're well on your way just replace the sun with a local heat source a a hot emitter and and uh just take into consideration that that emitter is closer and takes up more of the solid angle and so these devices have much higher power density uh to heat up that emitter we can use a variety of of sources including direct solar heating fuel chemical fuels nuclear fuels or even electrical heating and i i star solar and electrical because they were the topic of an earlier innotherm on may 13th which discussed the brayton battery and approaches that use high temperature thermal storage to take surplus renewable electricity convert it and store it in that form before using heat engines to put it back on the grid and one of the suitable heat engines for higher temperature storage are tpvs so the my talk today is going to mostly focus on the efficiency of this conversion scheme and efficiency is an important metric at this stage in this technology readiness level and there are other metrics that are important as well such as power density and cost but i'll focus mostly on efficiency and efficiency can be thought of as the typical heat engine form which is the electrical power that's generated divided by the heat that's consumed by the emitter and we can also break it up into two factors one is the absorption efficiency which takes a look at um how much of the energy that's absorbed is converted into electricity and the cavity efficiency which describes how well radiation is transferred from the emitter to the cell and i'm going to focus mostly on absorption efficiency and recently we published a review paper in joule that looked at efficiencies across the years and one of the things that we found is that we have been making improvements recently but also notably and gong mentioned this uh back in the late 70s and early 80s dick swanson uh really pushed the envelope in terms of conversion efficiency with his silicon tpv cells which were able to reach about 29 absorption efficiency um now looking at the progress my my focus of the focus of my talk today is going to be on these last two data points in particular work out of uc berkeley that was able to achieve 29.1 with an in-gas tpv cell and work out of our group that was able to achieve 32 percent and was published earlier uh this year in nature and so uh i think if if you were just to look at this progression of efficiencies you uh you may say that um there hasn't we haven't really come that far over the last 30 years only a few percentage points relative to what dick swanson was able to do but hopefully i'll be able to convince you today that if we dig a little deeper we'll see that uh the the the future for tpvs is much brighter and and i'll be able to answer this question of what is the the real significance of these recent tpv advances and provide some context for where context for where we're going next um so to do that uh it's a it's important to look at efficiency as a product of spectral and carrier management and to compare each efficiency point to its respective thermodynamic limit which in this case is the radiative limit and only considers radiative recombination and so spectral management describes how well the black body spectrum is utilized to excite carriers across a specific band gap compared to how much energy is absorbed by the cell and the carrier management describes how much power is produced relative to sort of the maximum ideal power that could be produced and so it's a ratio of this blue square on an iv curve to this gray square and if we start if we look at at tpv efficiencies along these two metrics and we compare them to the radiative limit we see now that the recent advance in 2019 gained a lot of ground in terms of carrier management with respect to the silicon the dick swanson silicon tpv cell and i'll show today that we were able to improve tpv conversion along another facet and i'll be able to totally convince you that going forward we should be able to combine these two to reach a very high efficiencies so the metric that we were going after in in our group is to improve spectral efficiency so the ability to convert this broad spectrum that's radiated by a heat source and i just want to note without going into too much detail that tpvs are best suited at converting the thermal tail of the blackbody distribution of planks distribution and that's because as we push the band gap higher we can produce higher voltages and then we're less susceptible to non-radiative recombination and ohmic losses and these are very important losses in tpvs so as we push the band gap higher these become less important but what we're left with is we have to manage a large amount of radiative radiated power that's out of band that's at energies below the band gap usually this accounts for 75 to 80 percent of the overall radiated power so the approach is to do this to suppress out-of-band uh radiation or radiative heat transfer can be categorized as either uh emission suppression so emitters that use structures or filters to suppress emission or another approach is to reflect that low energy radiation selectively back to the emitter and in terms of what's been in terms of the efforts to accomplish this mitigation of out-of-band loss which is so critical to efficiency there's been work out of mit to develop selective emitters there's also been some some more older work looking at front surface filters so these are filters that are put on top of the cell and that through interference and plasma achieve reflection across a broadband and uh a and really the devices that have achieved the highest deficiencies rely on back surface reflectors so these are really good mirrors that are placed behind the absorber and this architecture is the one that was used by the berkeley group to achieve 29.1 efficiency however looking at what's been done so far what we see is that current strategies only suppress about 95 or less of this out-of-band power and you may say that seems pretty good but actually if we if we look at the sensitivity to this metric what we see is that in the radiative limit and at these prototypical conditions uh by by changing by improving the reflectance by a few percentage points we can reach very high efficiencies so at 95 reflectance we're bound to 47 efficiency uh but if we go to 100 we max out at 60 efficiency and so these are large gains for marginal changes in out-of-band reflectance so key metric to improve upon our initial work in 2018 tried to do this by developing an architecture where an in-gas absorber was put on top of a low index spacer and then on a mirror and we were trying to improve reflection in this structure out of band and theory suggests that the autoband reflectance should be close to 98 or 99 but in practice we saw parasitic absorption in this magnesium fluoride spacer and this is something that we've noticed and other groups have seen as well as their issues once you start to actually deploy a dielectric spacer a low index dielectric spacer behind the absorber so we came across this bold idea of trying to replace the dielectric spacer with nothing and when i say nothing i mean air so try to put a small air cavity behind the absorber and the back gold reflector and now this the this geometry or this architecture this airbridge architecture as we call it can achieve lossless phenol reflectance at each interface air is in a way an ideal dielectric it doesn't absorb and it maximizes the difference in refractive index between the semiconductor air interface and the air air metal interface and so overall we can achieve very high reflectance and simulations show that that reflectance can approach 99 and uh in the next couple of months i just want to show you how we achieved uh this fabrication of this device how we realized it and i want to demonstrate that it's relatively straightforward and i expect that it's something that can be replicated in other labs or at least i encourage people that it's something that they can they can accomplish as well so i just want to highlight the simplicity of the process for making this airbridge architecture because there are similar architectures that have been reported but were used what was used was a much more complicated process to make them so we start by growing a a epi layer on a lattice match substrate indium phosphide and we define the grid lines that will end up being the supports of the absorber of the air bridge then that wafer is flipped over and bonded to a planar gold mirror and once that bond is established we can actuate the parent substrate the indium phosphide substrate that was used to grow the epi film and then we can we can pattern out the cell and put a top contact and there's a resulting architecture is shown here this is a cross section where we can see the active absorber layer suspended over the planar gold mirror by this over this air cavity and we can achieve spans of close to 100 microns while maintaining the 600 nanometer air gap using this simple fabrication process and now we can realize a reflectance very high reflectance in practice that matches our simulations so now the out-of-band reflectance is as high as 98.5 percent

and that high out-of-band reflectance should lead to high efficiencies so to do this we developed a basically a tpv simulator setup sort of analogous to a solar simulator setup but here we're using a glow bar source at about 1450 kelvin and we can vary the illumination levels by changing the distance relative to the tpv cell and what we see is that as we increase the illumination level and consequently the short circuit current we see is voltage gains up to some point in which all mclosses begin to dominate and then we see plateauing and even a downturn in efficiency but at the peak efficiency the peak absorption efficiency that we see in this device is around 32 percent and the iv curves that are correspond to that 32 percent are shown here with an open circuit voltage of around 0.45 all right so let's go back to this plot of spectral versus carrier management and see where we ended up so our work pushed as mentioned pushed into this high spectral management high efficiency spectral management regime and we made significant gains in that direction and we were able to surpass and approach about 55 percent of the radiative limit and i want to say that uh i want to acknowledge that we pushed along different facets of this problem relative to the berkeley group but there's no reason why these two can't be merged together and so if we were to come up with a hypothetical scenario that does this i would expect that in the very near term we would see single junction deficiencies at 1500 kelvin for in gas that are approaching 40 percent about 38 um and also i want to acknowledge that this is single junction efficiency efficiencies can be higher for multi-junctions and there's work out of nrel and mit that have shown about 36 percent absorption efficiency with a multi-junction cell and they using their models they estimate that if the reflectance goes to 98 they can exceed 40 efficiency so this paints an optimistic picture but i want to paint an even more optimistic picture by saying that the urbridge architecture allows us to now think about operating in an entirely new regime device regime and this is best shown by these bands gray and blue bands that show that are on a plot of spectral efficiency versus band gap divided by the thermal energy of the emitter and what i want to say is that these gray bands show the trajectory of efficiency spectral efficiency as a function of increasing band gap or decreasing temperature and if we are in the conventional reflectance regime as we increase band gap we take a hit on on spectral efficiency but if we are start to approach above 98 or 99 reflectance which can be enabled by the air bridge architecture we see that we enter a regime where we're much less sensitive to band gap and temperature and so this opens up a very exciting possibility of using cheaper wider bandgap cells that have already been produced for solar applications or enabling the use of lower temperature heat sources and i think both of these are very exciting as a path forward for tpv technology so i'll end there and turn it over to jared sorry i'm just trying to end my screen share okay i hope everyone can see my screen yep great um so it's a pleasure to be involved in this panel so thank you so much for the invitation to be here um and today i'm going to be presenting a general introduction to thermodynamic energy conversion and within the context of this panel to explain some of the differences similarities uh to your thorough photo impacts so you might not be familiar with thermionic energy conversion but you're actually quite familiar with devices they use thermion emission and it's just a source of electron evaporation so if you'll remember your old crt monitor there's a little filament that gave off electrons that were rastered back and forth and if you have fluorescent lighting above you there's thermion convection going on above your head at the end of this there's a little filament that's giving up electrons that then help accept the mercury uh within the fluorescent pole and so what we do is we use this electron evaporation as a direct source of dc current so instead of a filament we have in this case a plate that emits electrons across a small vacuum gap that is close to a another material which we call it so this thermodynamic cathode and its electrons directly to across this gap to this uh anode and that's a source of dc current um so thermionics actually has a very long history that uh gong alluded to and the 1960s was part of a space race multiple groups were achieving mid-teens efficiencies really enormous power densities above 10 watts per square centimeter and in those days the leading folks in the field that's uh the encephalitis the guy alluded to with normal electron and wilson was at ge believe that there is a ton of potential left on the table for these thorough energy conversion classes so what happened well what happened was they uh got captured by a single application that was deep space nuclear power and within that a single application well efficiency matters but it wasn't necessarily uh the most important to do is one portion of a system um and at the end of the day uh the only interest from that was from nasa and nasa had a good enough solution which is great for nasa which is thermoelectrics and so in the case of uh deep space nuclear power thermionics lost now but that means that today there's an enormous opportunity to apply decades of innovations and materials and microfabrication that i'll describe a little bit in general later in this talk but before i do that i just want to give books a a status of where conventional thermionics got to because i feel that much of this early literature has not been uh looked at in the last couple of decades so here you can see the characteristic curves the efficiency that was possible in the 1960s and in the 1990s there were vacuum chamber measurements that put the efficiency around 20 the power density and what temperature was required for that power density you can see here but for me the most important thing perhaps is the lifetime in the 1960s they figured out a lot of the metallurgy and showed that there is no fundamentally life-limiting mechanism for thermometer converters that was present by 1973 there is a five-year lifetime test that uh the device was operating the same at the end of that test as it was at the beginning as you can see here in the 1980s as part of the doe work they demonstrated well over 10 000 hours in a combustion environment with really enormous um year over year change so that's where we come in spark was founded in 2015 and it was really based off of a lot of the work during my phd at stanford and those of my stanford colleagues after i graduated we became part of the founding cohort at cyclotron roads spun in to berkeley lab at which point we gained support from rpe and also for darpa as well as some of our first investment and using this rpe support uh and this other investment support we were able to demonstrate first fully encapsulated devices which resulted in our first sales in 2019 and if anyone would like to test some of these there ionic devices themselves feel free to reach out since that time we've expanded from lawrence berkeley labs spun out into our own facility of memory california is still quite close to just a couple of miles from the lab and we still maintain active collaborations there and while i won't have too much time to talk about them today i did want to mention that there's been continued work by ian lim uh and collaborators at warrensburg lab by andre calderon and andrea schmidt and this uh work was focused on understanding electron reflection and these neuronal converters and then minimizing it with nanoscale order and since that time we've also gained the support of the army air force uh increased our collaborations outside of the lab to army research lab and most recently oak ridge and also received continued support from california energy commission as well as the doe so going a little bit more detail in terms of the basic operations the emission current follows what's known as the richards endushment equation and as you can see it's a very sensitive function of the work function of that cathode in terms of the voltage that we get out in the ideal case it's simply the difference between the cathode and the other word function however in reality electrons have charge and so once they begin to emit into this vacuum gap there's typically some amount of space charge that builds up that begins to suppress that emission current that means that you often have to apply some voltage accelerating voltage within the gap in order to mitigate that space charge uh and that brings us to the basic figure of varying thermal devices which is the sum of the voltage within the gap as well as the anode sometimes called the collector work function so this is known as the back voltage or barrier index and i'll be spending a couple of slides now kind of describing what the opportunities within the tar so you can essentially just write down what is the possible efficiency of a thermodynamic device as a function of temperature parameterized by this barrier index and the historical folks were typically at barrier indices around 2eb however that was as we'll describe part of the limitations of the technology of the deck it's quite straightforward to create a work function around 1.5 an operating device and in a vacuum chamber that's very easy to get down below one and i'll highlight some of those as well so what i want to emphasize here is well it's very uh challenging to integrate these barrier indices integrate integrate these low work function materials it's not at all crazy to be talking about a barrier index of habitable to a volt in fact we've demonstrated work functions as low as 0.6 eb this would be if you're

familiar with the electric competitive with the zt of 10 it really shows the potential for the opportunity for reducing work function and reducing this uh voltage that we have to apply within the gap so because this is a handle that compares thermometics and thermophotovoltaics i wanted to give a very brief discussion on what the sort of differences are between them because probably from a black box approach you're saying well one side is very hot the other side's cold electricity comes out and at least until recently the performance in terms of raw efficiency is actually quite similar so what i want to emphasize is that both thermodynamics and thermophobable tags depend on a flux of either electrons or photons above some threshold energy and the case of thermal photovoltaics that's often a band gap or the desired spectral emitter and within the thermionic field that's the work function of the cathode and so i just broke down the basic stat mech derivation of what that actual flux is and you can see there's some dimensionless integral from some threshold energy and then a bunch of uh constants uh in front where we've made this unit list so that all the constants are and what you can see is that they share a lot of similarities these constants which basically have to do with the density of states of electrons oh no when did this drop out now i'm okay now it's okay okay okay um so to summarize uh i don't know when i dropped out but uh there's some um unitless integral uh and some unit pole constants in front that have to do with the density of states of these photons or electrons are traversing that and if you just divide these three factors you end up with m c squared over kt so the thermionic version of this is about a million times larger than the thermophobic tab and what that means is that despite the black box similarity of these technologies we're actually operating in very very different regimes as you heard in the previous talk a band gap in a thermophobic cell is often just a handful of kt and of course the voltage out is just a pkt as well kt cockpit should specify and at the typical emitter temperatures uh that you'll see in literature of course these can be much higher for um energy storage applications you're looking at a pretty small bandgap as andre reported of 0.6 to 0.75 db and this really forces you to operate at a very cold collector temperature which has implications for system level design and efficiency um in contrast thermionics as long as you can get them across the band gap we have more electrons than we know what to do with so the space charge is the critical thing to overcome and then that barrier index tells you for these electrons of the heat that they're carrying from the cathode of that voltage what fraction is actually converted into useful power and so this barrier index is a primary challenge for reducing the thermalization loss of electrons coming from vacuum and then thermalizing at the fermi level it also means that for the time being as we're still developing ultra low work function materials we're extraordinarily insensitive to the rejection temperature t cold which has again important system level implications so now i just wanted to cover and i'll refer to a forthcoming review and a couple of slides how is it that you can actually achieve a better barrier index and this is a simplified energy diagram that's got a typical work function at 1800 kelvin for historical thermometic devices and typical anode and voltage in the gap for these historical plasma based thoracic devices so the first thing to know is that the ideal gap size for a thermometic device to mitigate space charge is on the order of a few microns half to five is all pretty good which was very difficult to achieve in the 1960s and so instead they relied on plasma based space charge screening which worked incredibly well but it did require a half volt overpotential in those games what that means is that if we can reclaim that we can go from here to this maximum power point which directly increases the voltage and voltage all the other things being equal directly translates to efficiency there are several ways to do this i don't want to stress that uh uh ideal you know the micron gap size is the only way there are schemes to improve plasma based screening that folks have looked at there are also gate-based designs such as this one you see on the left but we're focused on this microgap technique and on the right side you can see some of our data this shows that for the first time we can actually get a power densities that are of relevance for thermodynamic devices it's patrick in 1993 uh achieved a 10 micron gap which is fairly good enough in order to get a reasonable current out and what you can see is that a very low temperature from a historical thermal perspective of about 1300 kelvin or just above a thousand celsius we're able to achieve with micron gap spacers uh over a walker square centimeter which is a i'll refer to later as a very important threshold and i'll uh remind you again that this is just with conventional thermionic materials uh the both emitter and collector in this case were just polyphenol with a low work function product but of course you're not limited to chunks of refractory metal with coatings there's several decades of materials innovation that now we believe that we can apply to this and like i said there's in principle no uh known reason that we can't achieve this low work function of 0.5 like i said we have personally achieved 0.6 of course it's a very long road between that uh vacuum chamber demonstration and getting that stability into a device and i'll just highlight a few of them here the there are a few schemes in literature in order to use conventional codings but then shift the fermi level within the device such as here with back gating and a graphene collector and while it shares some uh differences uh and you'll hear more about it in the following i also look at uh techniques to combine thermophobic tags and thermolinks as part of this permeate level splitting and shifting because at the end of the day it's all towards increasing voltage but there are other techniques as well there is a lot of work that has been uh demonstrated in a diamond for instance that gets work functions within this low level but i will mention that these will be limited currently to fairly low current densities and that should be an active area of research to overcome so i did want to mention that there is a forthcoming review that that should arrive in just a few weeks at this point it's been accepted into advanced science and the major purposes besides reviewing some of the most recent work some of it was highlighted on the previous slides is to define a core efficiency our goal there is to include realistic thermal pathways and practice they're both standardized across different work but to also discourage overly optimistic reporting of experimental results and i'll briefly go through what that is on the following side and based on those thermal pathways we identify a rough threshold of on the order of one watt per square centimeter as a region of interest if you can get above one walkers per centimeter we think that this is practical but if your uh concept will struggle to get to that level um it's really difficult to justify integrating that into a practical final and encapsulated thermal converting and the other final area that we wanted to do is that we feel that a lot of the thermion literature is just not as well known and so while there are some very good plasma reviews in the early 1990s in particular some of the work since then has not really appeared in too many reviews so we wanted to highlight and unearth some of the recent but also some of the more lesser known work over the last 20 to 30 years and i'll just briefly explain why we think it's so important to consider a core efficiency which is that there are a number of additional thermal pathways within a practical device here we've picked a metering collector materials that are typical for historical devices that somewhat exaggerates what i'm about to say but there's radiation if you're using cesium plasma there may be vapor uh there is conduction between basic encapsulation as well as any structures or spacers to uh structure the emitter from the collector and all of these can be on the order it depends on your design of 1 to 10 watts per square centimeter which is where this one watt threshold basically comes from and while many report this weed and radiation some recently have uh started to report just an electronic efficiency so ignore all of these normal pathways and assume that electrons are carrying all the heat from the emitter collection uh and we recommend against this because we think that it uh overstates the potential and also points you in a direction towards a lower power density as opposed to a higher one and so i won't go too much into these individual curves but what you can see is that if you only focus on the electronic efficiency you could be very excited about say a 30 uh plus efficient device um and i'll just mention that these are typical id curves for microgap devices and we picked a 10 micron gap to exaggerate these effects but if you look at what actually happens when you include radiation or vapor you could be at 13 if not below 10 once you consider these thresholds of these devices so the summary is that devices with very high power density tend to be able to overcome these parasitics much better but in addition the system has to be considered and this also means what's required for cooling or if you're fueling it what pumps or blowers will be required for supplying that fuel to your device so there's too much to talk about so i'll just summarize here um so the first is that thermionics has a very long history of development to build from uh compared to tpv there's a very high power density that's possible and flexibility and heat rejection which is important for system level performance but to date i think there's an enormous opportunity to apply materials innovation and modern fabrication in order to really unlock it and get past the sort of teens the 20 that was possible in the 1960s and 1990s we've also hopefully convinced you that a core efficiency the standardized reporting the literature is important uh which also sets a threshold of one walkers per centimeter as a target for these new concepts and for additional information i'll point and that's all from us thanks for the opportunity again to speak to all of you thank you alejandro if you can see my screen please yes yes okay perfect um whoa and we have had uh very interesting talks about thermionics thermophototax so my role will be trying to put these things together um so i will talk about the concept of fermionic enhanced thermoplastics this is a concept that has been proposed very recently in the last few years and i will start with a brief introduction and and you probably noticed that the previous two talks about thermophobic center munich have also focused on different metric figures in the case of thermophotics the main metric was efficiency the case of harmonic was power density and this this is related to because uh its technology is performing better in each of these metric figures so i'm illustrating here efficiency as a function of power density uh for two technologies one is a thermophotol types and the and the other is thermoelectric generators i'm not including thermionics here because of the lack of data in the literature so the review that is preparing would be very nice to complete this graph but we could say that permians will lie more or less where thermoelectrics technology is is represented in this figure with the difference that the municipally operated at extreme temperatures much higher than 1007 degrees celsius but in general you can see that uh well a thermophototype performs very well in efficiency you have here these two recent results that in hundred years commented with efficiencies already over 30 percent and fermionics or thermoelectrics as well perform very well in power density but not as good in conversion efficiency and and we can understand this behavior very easily if we look at these two devices thermostatic center thermoelectric converters and how they they are arranged for instance in the case of thermoelectric generators you have a direct contact between the hot side and the cold side so it means that there is a huge flux of energy carriers and those energy carriers could be useful or useless but at the end there is a lot of them and then the power is huge but because there is a lot of carriers in this case phonons that are conducted from the hot side to the call side this brings a lot of heat losses and then the efficiency is limited well the strategy in photovoltaic thermophobics is a completely different here you have a deep possible arrangement where you have the hot emitter and the photovoltaic cell and you place a filter in between them also you could use a backside reflector as andre recommended in in his presentation but the the aim at the end is to have a very restrict a spectral band of emission that is reaching the cell so the next uh relative transfer between the meter and the cell is restricted to very narrow range of energy photons so it reduces quite a lot the power density but it enables to use high even gut materials and well the tendency to increase the efficiency basically is to to make make this spectral control as efficient as possible and this enable to use higher band gap materials but these again increase the efficiency that's okay but the power lcd is quite limited is lower than one by the square centimeter so the the the question now is whether we could go to this region here so so the question will be if we can develop devices that can operate in this region and very importantly at low temperatures where most of the thermal application exists because obviously you can reach this region if we operate the tpb cells or the terminic converters are extreme temperatures for instance if we go to 2000 degrees celsius we can operate in this region but we want to do it with low temperature heat sources where most of applications exist and this is what my talk is going to be about and well there is a concept that is uh has a potential for approaching this range or this region of high power density and high efficiency and this is near field thermo photovoltaics the idea that was established in the early 2000s so it has been already 20 years around is that you can place a thermometer very close to a foothold like so so the gap between them is very very small nanometric distance in this case if all the radiation sources inside the emitter well typically they are some of these modes are totally reflected internally reflected in this material but uh if you place these two materials very very close and there are some photons and can tune this this gap and then the absorb at this side and in this decide you have a photovoltaic cell then you can produce power out of this near field radiation so you have a contribution of evanescent waves which enhance the relative transfer and very importantly you don't have any phone propagation so there is no heat conduction because the hot side and the coal side are interrupted by a nano scale batten gap so only photons can cross this this gap and well very recently there have been very um encouraging results as for instance in this publication in by purino in 2018 where you can see that when you reduce the the gaps distance then you have a drastic increase in the power density in the power output of the device in this case up to 40 times when you reach distances as low as 100 nanometers um just to make a brief overview of the state of the art i'm writing here the four main experimental results proving this concept this is the one that i saw before but there have been other other three and remarkably i want to emphasize the date of the publication so these are really really recent results so you have different divisive materials with different bundling up energies from extremely low band gaps of 0.2 or 0.3 electron balls to more convenient perhaps van gaaps of 0.74 0.67

or germanium you have also the distances around 100 nanometers this is the separation distance between the meter and the cell the temperature variance between the hot and the cold side so the cell and the meter are around 800 kelvin in in this case in the location of you know in a way and the power gain is in the range of 10 40 or something in that range of one order of magnitude power enhancement and the efficiencies are of course very low for the moment except for this publication in this uh thesis in this year that represents the record conversion efficiency for this technology but the rest of the publications are reporting very low conversion efficiencies for the moment but i want you to pay attention especially to these two parameters which are the size and the power that is produced by this experimental proof of concept you see that the size is extremely small so all devices are very very small and consequently the power that is produced in the near field is also very very small so you can imagine that there are reasons for making very small devices you see all the implementations of these publications here uh well you can solve planarity issues and things like that but there is another key challenge of making these devices at larger scale which is the ohmic losses and this is what i'm gonna try to explain now well if you want to make a near-field ppv device at large scale let's say one by one square centimeter device uh well you will need to incorporate some electrodes in the cell uh i have figured out here one possibility which is to place the electrodes uh in the in the front side of course they have to be hidden let's say from the from the meter in order to be able to create this a small distance at the so at the end um well you have uh photogeneration of course and you have a collection of electrons and and holes in these two electrodes well if you do the simulation of this device um you can get this this curve here for the power density that increase substantially when you go below one micron distance and then up to six watts square centimeter but you have to pay attention to the current density because we are using in this case a very low and that cell of 0.35 electron volts which is needed to tackle the spectrum of the emitter which in this case is 1000 kelvin so you need very low bandwidth cells and then um well you will have huge uh current densities to produce these powers in the range of 10 to 40 amperes per square centimeter and this is of course very challenging regarding the electrodes so if you incorporate a typical ohmic losses uh for these electrodes of 10 million square centimeter then you get this real performance of the near fuel tpv device so you see that you lose quite significantly the the capability of increasing the power density so so this points out a problem of scalability for near fuel tpv devices i want to remark that i have made this for front side electric electrodes but this will apply also if you make the electrodes on the rear because at the end the problem with the ohmic losses is the lateral conduction so this um this trajectory of the minority carriers when they have to travel to the local electrodes is the main contribution to the to the ohmic losses together with the conduction of course through the metal grid so and here it comes the idea of making this hybrid of thermionics and thermophotics that has a particular interest for applications in the near field so here is basically the idea if you want more details you can read these publications but the idea is that you allow the meter to emit also electrons so not only photons so this is like a thermionic regular terminic device but so you have these two energy carriers you have electrons and you still have photons as well so the process at the end what you have is that the electrons are collected in the front surface and they are injected in the balance band of your photovoltaic cell so all the holes uh are [Music] combined with these electrons in the surface and then the electrons are collected in the back so you need a piece over and a photovoltaic structure if you look at the results if if you assume a work function for the meter of 1.3 electron ball so is very low in order to tackle this low temperature operation and anode or collector or function of one electron volt so it's opening three electron balls of different you can get this power density with well you can see that is quite significantly higher than what you can get in the real near field tpv device so the idea is that with this concept you avoid any electrodes in the front so there is a wireless electrical connection between the meter and the photo that is photovoltaic so there is 1d carrier a collection and this is really really important because 1d correction avoids the lateral conduction of the minerality carriers in in the in the in the photovoltaic cell and then it enables an illegal losses and of course you also have an extraterminic power which is basically due to the voltage that is established between the emitter and the front side of the cell so then you get this much higher power density well so we have done uh the first experimental proof of concept uh in a collaboration with the cnr in italy uh this is the setup that our italian friends have in rome uh where you can see this is a the cathode which is heated by a laser source and here there is the anode which is a macroscopic distance here so you can approach the cathode to the anode to a distances like a few hundreds of microns not much less than that you can see the results in this paper that has been published very recently in asus energy letters this year and the summary of the results are here on the on the right well you have the the current versus the voltage of course this is the logarithmic logarithmic of the current and you have the iv curve for three different devices this the base a is simply a conventional let's say thermionic converter where you have tungsten as cathode and a p-type gallium arsenide anode you have this other converter which is already this permeanic enhanced device or photos like anodes where you have a p over n junction of gallium arsenide and again tungsten emitter and an improved version of all of them which is the c option where you have the same photovoltaic anode but this time with a session coating to reduce the work function of the ammo and and you can see that well in these cyber curves you see different different things uh first well this is the direct operation regime where you can see that all devices have the same saturation current so they can be compared so in terms of terminal current they are identical this other side is a reverse operation regime where you see that these two devices have uh the same current because basically they don't they have the same surface properties so this tungsten and amp type glancing in both cases but in this other case you have a higher reverse current basically because you have this other anode surface treatment where you have session and then you have a higher reverse current so the current that goes from the unknown to the cathode but the most remarkable thing is the the voltage improvement in particular if we look at device device a and b we see that there is an increase in the voltage for the same current you you get higher voltage in the in the case where you have a photovoltaic anode and this voltage is is even higher if we use a session on top of the photovoltaic cell due to the reduction of the word function so this represents basically the proof of this concept and i will go very very briefly obviously i'm not going into details of all the symbols in this graph but the the basic idea this is the ban diagram of this device uh is to illustrate that you have a photovoltaic cell from here to right and you have a thermionic device from here to left and the two contributions to voltage are the thermionic states voltage here which is basically the difference between the fermi levels in the cathode and in the anode in this case this is a p plus a gallium arsenide layer and then in the photovoltaic side you have this other voltage which is the difference in the fermi levels of the of the photovoltaic cell so at the output you have this higher voltage which is the addition of these other two voltages and and that's basically uh represents uh well what we have seen here this is the increment uh that we see from from the thermionic states alone to the terminic plus photovoltaic cell a combination well and of course this is a proof of concept and and there there are a number of challenges to be addressed in the future one of them is uh to establish large area nanogaps with very low thermal conductivity uh that withstands large temperature gradients and we we have seen that um jared have shown that they are working on that and there is an interesting link here between thermionics and near-field thermal phototypes intermediates typically they need a bit larger gap distances of typically greater than one micron but there is uh uh well an interest in reducing this distance to to get more power and then you have in the terminic side you have a problem with the near field radiation so the idea is that having this uh hybrid concept you can take make a good use of these photons that otherwise otherwise were wasted of course the very important spectral control in the near field you you can take advantage of resonances in the surfaces uh plasmon phonon fluoritons you can also use rear reflectors so they work by andre could be also applied to this concept basically to to improve the spectral management of the cell and as jared said lower function materials for the cathode and the anode is very important and obviously at the end the size system configuration is to to make high temperature and high vacuum standalone devices assembly so just let me acknowledge our funding program from europe uh you're in some 2020 program that open this is a models project that ended in 2019 and make one announcement because we are having a sympathy on thermophotics in emrs this spring meeting in may 31 to june 4 and there are the the abstract mission is open already until 19 january so this is all thank you very much oh thank you all uh for all the uh three really very stimulant talk i would like to ask our parents or turn on your uh see video and we will answer questions uh if you have questions please uh input your questions through qna and i want to uh also make a uh two announcements uh before i say because some people may have to leave uh this will be our last uh colloquium this year so we will restart the new qualcomm in 2021 on february 3rd so that's our new semester start and we will also say the next semester or the colloquial will be bi-weekly and it will be say the same time but it's on wednesday right on tuesday we have to juggle through our teaching so that's why the change so look forward to the announcement and with that let's uh go back to the questions and let's see uh we have a few questions already so uh i'll uh first uh uh i think the uh first two questions i share with the see more geared towards uh andrei so i'll combine them and uh one is from an anonymous attendee how would you comment the effect of air property change when it's heated up and pass through by electric current so that's a and then the second question is from anina lapodin and she asked thanks for the very interesting presentation a question for andre on the time measurement efficiency is the efficiency measurement done at the full intensity at the emitter temperature or it's some percentage of the food intensity for your spectrum all right yes for the first question the air properties um the cell has to remain relatively cold so we're not planning on heating up the cell to very high temperatures during processing is probably the highest temperature it experiences that's around 150 celsius during the bonding process so there is air that's trapped in those cavities and and it can survive that temperature range hopefully that addresses your question sometimes a question about thermal resistance comes up and it turns out that a simple back of the envelope calculation will show you that thermal resistance through that air is is not that significant and most of the heat flows through the gold grids just as the charge flows through the gold grids and in terms of the other question yes our tpv simulator does not reach full solid angle illumination of the cell so that's one of the compromises that we had to make in in the design of that variable illumination experiment one of our goals was to really make tpv efficiency measurements accessible to a broader audience i think if the the buy-in is to have to develop a large-scale prototype in order to characterize the device i think that's too high of a nask and so i think there are steps along the way in milestones along the way that can be proven out and i think that's that's the intention of our tpv simulator experiment ah great uh so we have a question for jarrod how come the temperature of collector is ignored in ceramic similarity emission performance calculation doesn't cause a backing machine if it's too high yeah that's a great question so it does cause back emission if it's too high the rule of thumb is that emission turns on at 700 kelvin previous work function so if you're at a conventional material that's at say 1.5 ev that means you can largely neglect back emission until you're at something like a thousand kelvin so that's really what i mean about this being a very insensitive is that the performance is quite similar uh and i'll i'll be mixing going back to celsius but the performance is quite similar whether your collector is at 400 500 600 celsius and actually that five-year lifetime data was at 800 celsius now as we uh create better and better uh collector materials at one ev of course you'll need to lower that if you get all the way down to that half fold you'll need to have similarly cold collectors as in thermophotovoltaics but for the immediate future at practical work functions of one to 1.5

we can easily handle hundreds of degrees above ambient which creates a lot of advantages in terms of the overall system thank you i think the next question is from muhammad taha mansour and just a general query can the emitter be designed to emit only in the desired wavelengths i guess uh this is a unbridged you mentioned this probably yeah um in theory yes in practice it's very difficult because you have to also operate the emitter at high temperatures at which point the emissivity out of the wavelength range that you're interested in grows particularly if you rely on a dope semiconductor metal emitter uh then the emissivity increases with electrical resistivity and it's really hard to control in practice thank you um so we have a question from uh uh lin xiao zhu uh thanks for the exciting presentation this is us nicholas joe from penn state i have a question on the combination of thermionic and thermal photovoltaic would the electron bombarding the insermionic operation need to material degradation on the pv cells so this is alejandro if you could you and so uh you are muted so yeah sorry well um maybe this is a mixture between and me because well i i i can't tell much about the um the reliability of terminus probably here it can tell um in the anode which is well the main difference of this combination is is the material that you use to make the anode and you are using well we are using gallium arsenide now but that means that we are using that in the future uh because it could be germanium it could be indium gallium arsenide uh basically three five semiconductors or well if we go to extreme temperatures why not silicon as well so we haven't done any tests of this reaction of of well we have been assessing uh with gallium arsenide we haven't seen any degradation after the experiment but there could be some reactions there i think jared will comment on that but uh well at the end uh in the pv cell you you typically also have a window layer to prevent recombination on the front surface so you can use materials like amorphous silicon or even well some transparent conducting oxides so you have a white flexibility of design in your anode even if it's phototyx in order to reduce this interaction with whatever hermionic coatings you you can place on top of that so since uh our continuous uh thermionic question and one from alumnus attendee i wonder how important the lead loss would be in practical thermionica system great question and i didn't allude to that and i'll also point folks to this uh review with our long-standing collaborators at upend that's a forthcoming evil bargainton and matt campbell the short story is that you choose the lead based on the other heat flows within the device typically you use a metal which is usually limited by vitamin fronts of course you could do better if you had a nice p-type uh semiconductor to create this into a thermocouple leg but if you're using a typical metal that's limited by vitamin fronds then you have a trade-off between how much resistive loss that you want and how much heat conductance and as a rule of thumb you can imagine that you lose about 10 percent of each so the voltage that you measure versus the voltage that you get out it's on the order of 90 of that and you only want to add a 10 or 20 percent additional heat flow through this as well but the the way that you do this is that um you basically just decide on the dimensions of your leads if your emitter is uh some distance from uh whatever collector insulating seal then maybe you want a thicker or thinner sheet of metal essentially connecting the top to the bottom so ah that's great so let me group a few uh some more photo would take uh questions uh and then there's a very important question i want to bring up uh later on so uh uh see uh one of this question is from nema what's the key point to have a less expensive tpv system to be successful uh in the market and actually say uh let me combine this with another one and because we're talking about market would it be possible for each of these of you you comment on economics and most likely application of technologies uh sure should i start kong yeah okay so in terms of the economics so usually the biggest lever that people use in the economics of tpvs is power density and it's a way as you increase power density the dollar per watt metric will go down um and and that's that's been the main driver to bring down costs and why you know why semiconductors that are typically too expensive for solar applications would be considered for tpv applications is because of a 100 fold difference in power density i think the other question was you know just generally about why you would want to make it cheaper so by going to more established technologies with wider band gaps you can certainly make it more cheaper and you can access production at larger scales which is one of the bottlenecks to entry into an application space like energy storage where you need to deliver very large systems and i think that would be probably the most exciting first application for tpvs but realistically there are other lower hanging fruits that don't actually depend on cost as much as power density or the weight of the of the device in aerospace applications okay here they want to go sure and i think andre covered it quite well and this is also why we point very much towards the power density because as long as you are still using high temperature materials the even if the materials themselves are cheap the processing and putting together a high temperature emitter that say compatible if you're interested in combustion and combustion environment is quite challenging it's actually a big focus and i wanted to mention that this is uh the purpose of our california energy commission uh grant is to start scaling up the production of our converters so i just wanted to give them a shout out that it's a really interesting program that i hope is replicated in states outside of california as well the final thing that i'll just mention is just like andre said it's making power in difficult places that really depend on size weight and power and so this has been a lot of our interest from the department of defense whether it's the army or the air force is that if we're successful this power density means that we'll create the smallest lightest generator on the planet uh and that has uh which is by the way silent um and that has certain interesting defense applications for our first uh uh first applications uh alejandro yo yeah well it has been very well covered but the question is is really important perhaps it's the most important question if we can make this economical and well just to maybe summarize what happens inside i i see three strategies to to reduce the cost uh one is increase the temperature well it's related with increasing the the power density but if increasing the temperature typically means that you are targeting temperatures that are not typically in any any application i mean in an industry typically waste heater is typically lower temperature so you have to create

2020-12-20 11:27

Other news