UNSW SPREE 201305-02 Martin Green – Evolution of High Efficiency Silicon Solar Cell Design 2

Just another WordPress site

UNSW SPREE 201305-02 Martin Green – Evolution of High Efficiency Silicon Solar Cell Design 2

okay good morning everyone today I’ve got to continue on where I left off last on Tuesday talking about the evolution of high-efficiency silicon solar cell design as I mentioned I’m with the new Australian Centre for advanced the photovoltaics and this is the first public airing of its new logo so this is what I’ve got to go over but I’ll start off recapping some of what I mentioned last week and then start talking about some of the principles underlying the high-efficiency design itself so this is what I mentioned last week by the early sixties cell technology had evolved to this stage and there was a standard design to centimeter by two centimeters actually six fingers that you can see here and for about a decade this was the way solar cells were made so no one deviated from that because the types of missions this is the first space mission but the types of missions were very critical that nothing go wrong so stuck with the standard cell design so as I mentioned he’d had a problem in that had a very heavy diffusion so this is phosphorus versus depth so you notice one micron is a long way for these impurities to diffuse into the silicon material so this doped region near the surface is very shallow considering that the wafer is several hundred microns thick so it’s only the top percent or so of the device that is actually doped even though a lot of these drawings the regions are looking pretty similar in size also these lateral dimensions are very much larger than than the vertical ones so these fingers as I think we elucidated on Tuesday a couple hundred microns wide and spaced several millimeters apart so six of them over two centimeters goes three point three I guess millimeters apart in the standard design and then came the varlet cell so in the early seventies people started reassessing the cell design with all the bounces that happened in photo in micro electronics so photographic techniques were used to find these contacts went from six fingers to eight and then to twelve so a big leap forward this cell was known as the varlet cell which started getting the efficiency moving up again this point here about 15% efficient under terrestrial sunlight so the things that were done and we’ll talk more about this in the next phase of lecture a lighter diffusion here to get rid of that dead layer I mentioned that formed at the surface and then photographic techniques were defined used to define these fingers so they could be made much narrower put closer together and this reduced the resistance penalty from going to a light a diffusion in this region here this doped region near the rear by heating the an aluminium layer on the silicon rear you can actually cause it get it to get it to cause the silicon to melt at quite low temperature and then when it resolidify dit was doped with aluminium which is a p-type dopant in silicon high-index anti-reflection coating the ones we used before were were a little bit low and as I mentioned last week it was also less absorbing materials so silicon monoxide used to be used and titanium dioxide tantalum pentoxide and a range of other less absorbing anti-reflection coating started to be used and because the cell could now respond better to blue light because there’s no it was no dead layer along the surface the anti-reflection coating was sensible to make it thinner because it could a cell previously couldn’t respond to blue light so you designed the end reflection coating for longer wavelengths but now that you could respond to blue was more sensible to have a thinner and reflection coating and the cells sort of look violet so that’s why they are known as the vial itself and then the substrate was more heavily doped so increasing the majority carrier concentration reduces the minority carrier concentration so the N P product of thermal equilibrium is a constant so push up the doping you push down the minority carriers and as we mentioned last week current flows are due to gradients in minority carriers so if you push down the one already carrier concentration gradients are going to be and current flows are going to be smaller so we’ll talk a little bit more about that as well the next big step forward and it was quite a jump as you can see in 1974 which is what I joined the University here this cell was reported the black cell because it looked like black velvet when you looked at it but the major advance over this one was the use of this texturing of the surface to reduce reflection so I think the reflection technique reduction technique is quite obvious like light

coming down will first strike one of the sides of these pyramids and most of it will be refracted into the silicon material but what isn’t reflected what isn’t reflect refracted in is reflected but the angles are such that it has to get reflected downwards so that means it has at least two chances of getting coupled into the silicon and if you think about it if you’ve got 10 percent reflection on your first loss if you have two bounces you get 10 percent of 10 percent loss on the after the second bounce so can reduce a cell with 10 percent reflection down to a cell with 1 percent reflection by that technique so it’s a very good technique for getting very broadband anti reflection properties and it’s used in most commercial cells speaking of which this is my standard diagram for what a commercial cell looks like today it’s quite an old version these finger widths have steadily come down as the technology has improved but basically the cells that are sold commercially now are black cells so they you know this using the technology that was pretty well known in 1974 and what’s going to happen over the coming years is a lot of the improvements that I’ve talked about that I’ll talk about but we demonstrate in the laboratory subsequently to this will be what happens in commercial production so what I’m telling you is going to be very relevant to what you see in commercial production over the coming years so this is a standard commercial cell the main differences between this and a black cell is the way the contacts have been put on some of the improvements I’ve talked about I’ll talk about sooner but already be incorporating to these so a couple of years ago they were about 17 percent efficient as well now some of these can getting up to 18 hi eighteenth nineteen percent efficiency you know higher than the laboratory cells that were sort of their prototypes so this is the way the metal contact is applied commercially you just have this screen and the metal is sort of ground up into a paste mixed up with other stuff so that’s a thick ish type paste and then it’s just squeezed through this screen screen onto the cell to give you the defined patterns the other technique I was talking about last lecture used a similar type of mask to this but the metal was just sort of evaporated down here so that it just got through where the holes were whereas this is more of a physical process where you push the pace through the mask so this screen printing process works very well and that’s why cell technology has been stagnant commercially for quite a while because this screen printing is a really nice process that works quite well and it’s been improved quite a bit over the recent years okay just looking a bit more detail that they at the cells I’ve talked about now this was the conventional cell 60s to 70s that was ruled so the open circuit voltage was about 550 Billy volts going to the violet and black cell went to bore heavily doped substrates and then increase the voltage to close to 600 millivolts now that’s pretty pathetic by present standard so most commercial cells will have 620 630 millivolts voltage these days so you know that’s was an obvious deficiency in the performance of the device the low voltage but people were struggling to find ways of increasing a big boost in the current so this boast of this boost is the removal of the dead layer and their improved anti-reflection coatings so that gives most of this boost and then the better reflection in the black cell gives the additional current boost and these are the spectral responses that I’ll skip over for the moment so I want to talk about three different principles tonight of increasing system sophistication the first one is a superposition principle that I think most of you will be ready familiar with but as I mentioned last week here’s Walter Schottky again who I love this theory you can treat the light and dark currents within a cell separately so that’s really good so I’ll be talking about what happens in a PN Junction the dark and everything in you might ask how relevant is that to a solar cell that works in the light because of this principle of superposition position you can separate the two out so do a design a good solar so you want to have a good dark characteristic here you’d particularly you’ll want this bit of the curve to be pushed out as far as you can in this direction so this in black or electronics this is called the D voltage of the diode where the current starts increasing so you want a high knee voltage and to get that you can do things like dope the substrate more heavily so that’s what’s done with diodes so you want to suppress all the dark currents in a solar cell so you want to push this curve out as far as you can all this light generated current is determined by generally other design parameters of the cell so you can sort of optimize the voltage there’s open circuit voltage and the current separately at least that’s what we

thought but I’ll show it’s a little bit more interrelated than that so this principle of superposition you know it applies pretty well you’ve already got to look for examples where it doesn’t apply you know so you know more or less it applies from those cells series resistance is the main cause of departure because series resistance effects are different in the dark from when illuminator is here o’clock anyway oh thanks what I’m going to talk about now is the emitter design and you probably all know but this is the emitter of the cell the top thin diffusion region near the surface and this is our termination originated with Volta here because a lot of this technology development was already was originally done for his NPN transistor shown here so this was the emitter it emitted the carriers within the PN Junction transistor this was the collector over here that collected them and this middle region was called the base so that terminology is sort of carried over into solar cell so this is always called the emitter on you know what sort of mid but it that’s the emitter and this is the base so what I’m going to talk about now is the effect of these surfaces because the first thing we did when trying to improve that black cell design we’ll start looking at the surfaces and how they influence cell performance and how we could reduce that influence so I’ve got to start looking at the surface of the emitter so next bit improving emitters so just to recap this is um from the red book but this just shows the carrier concentrations under a dark PN Junction diode under forward bias and as I said that’s relevant to a solar cell because everything gets superimposed so the this is a linear scale here the carriers that spill over from either side when you fought bias the junction decay exponentially in their respective regions as shown here so this is what happens if the regions go on for infinite extent you know you this will decay back to the equilibrium value if you go to heavier doping like this side is heavier doped than this side in this case you can suppress this drawings are all that good this should actually be bigger than this this drawings more representative here but the the proportionate increase here is smaller than the proportionate increase in this case so going to heavy doping sort of suppresses how high this concentration gets the current flows depend on the gradient here so d n DX if you call excess Direction minus DN DX determines how many electrons flow in that direction there there’s a constant the diffusivity D that determines you know how much flow you get for a given gradient okay we’ll just blow up that left-hand side so this is just your blowing up the p-type side in that previous diagram and this is the exponential decay down to the thermal equilibrium value so you’re causing a disturbance by forward biasing the diode particularly in the regions near the junctions in that disturbance just decays as you move through the device so pretty straightforward so what happens if there’s surface in the right way so you stick a surface like that and it depends very much on the properties of the surface so if you have a really bad surface if you you know scratch the surface with sandpaper or something you can get a really bad surface and that gives you what’s called a infinite surface recombination velocity any minority carrier getting that surfaces is instantly gobbled up so it means that you can’t maintain an excess of them if there’s any any carriers there above this equilibrium value if there’s any excess above that they just get coupled up and infinitely quickly if you’ve got an infinite recombination right so what’s that do you can see that well wow you’ve increased the gradient here so you you’re getting more current associated with that so that that that dark knee voltage is moving in so that’s going the wrong direction so you don’t want infinite surface recombination velocity if you want a high open circuit voltage solar cell and remembering that as as the design was evolving this thickness here was reducing so going from the conventional space cell to the violet cell you know this thickness room was substantially reduced if you go the other direction if you have a perfect surface if you have a surface that’s inert called minority carriers you have what’s called a zero surface recombination velocity it doesn’t matter how many carriers get to that surface they’re just not going to recombine there’s no process for them available there to recombine and what happens there is you get a an enhancement of the carrier concentration throughout this region here and you’ll note the gradient

here is reduced very substantially so very much less you know like it might be five or ten times less in this drawing here but that means five or ten times less current flow so your D voltage can increase accordingly it turns out a ten times decrease in current flows current flows I’m talking about corresponds to a 60 millivolts increase in open circuit voltage so if you got a solar cell with six hundred millivolts open second voltage a sixty million volt improvement is quite substantial so this is the type of thing you can get by controlling the surface recombination velocity are like a 10% supercharging of the cell performance if you have the value of the surface recombination velocity just right and in this case it’s equal to this diffusivity I mentioned divided by the diffusion lengths that decay constant here so if it has that particular value you don’t change anything the carriers get to the surface and it just looks like to them as if the material continued on indefinitely so that’s that’s the critical bit in deciding whether that’s the critical value in deciding you get behavior like this or behavior like that okay now comes something really interesting when we did this work we were thinking about the effects upon voltage primarily our initial aim was to improve the voltage of the cells above that 600 millivolts and when we started this work I think 620 millivolts was the record the world record for the highest voltage that ever been obtained from silicon so initial aim was to try and push up that voltage to two much higher values and and this was the type of thinking that that come in that work if we go back to the UM the conventional cell it had this dead layer at the surface so this was more like what happened in the conventional space so you had this dead layer where there was just no chance of minority carriers surviving they just got suppressed to their thermal equilibrium value but that increases the gradient further as you can see there so you’ve you’ve wiped out this region of the cell by doping it too heavily and that increases the current flows to that region which reduces voltage okay so what I’m going to talk about now is a principle not too many people in the photovoltaic area know about so if you understand what I’m talking about here you’ll be ahead of I reckon ninety-five percent of the people that think about these things in the sole area but it’s a reciprocal relation between the dark properties of the cell and the light properties this is a diagram from my red book I didn’t know about this relationship when I did this work and if I hadn’t known about it it would have made everything a bit simpler I think but if you look at my red book I’ve got a exercise or like I guess it’s a theoretical development that I do there and I don’t think I hadn’t seen it done before and before that time but I assume there’s a you know this is an unphysical situation although you might be able to think of ways of getting it experimentally but in assume you’ve got a plane within the cell and I’ve done it on the thick base region to sell because that some wider less influence from surfaces but you’ve just got a so this is the generation rate of carriers you’ve got none um than a impulse function of generation right at that point so that turns out to be quite a simple situation to analyze which is why I did it in the textbook but let’s figure eight point one figure eight point three has a solution so this is the probability a carrier is generated you know distance X from the junction is collected by the junction or contributes to the short-circuit current of the cell so if you generate right in the junction that’s great that carriers got a good chance to getting collected there’s some situation of a really bad material that won’t occur but practical silicon solar cells 100 percent of carriers generated in the depletion region get collected and then you get these exponential decays on either side if I had have been a bit more astute then we said well that looks pretty much like the decays of the whole concentrate already carrier concentrations we saw in some of those earlier overheads and in fact it turns out that that’s the case the there’s a relationship between this decay here and the distribution this is this is you know for a device that’s illuminated there’s a relationship between this collection probability and the distributions you get in the device in the dark so you know there’s no way you could have guessed that or anything there’s no physical arguing that you could that argument that you could make that would show that relationship but it just happens it’s a fundamental relationship due to the nature of the differential equations that govern everything so it’s a property of the mathematics involved in formulating these solutions but that’s um you know so I’m you know you can go one step

further you can the effect of finite sum widths of these region and these Exponential’s get modified as we saw for the case of the carrier distributions and I even go to the trouble in the textbook of giving the solutions there again if I had a given the solutions for the carrier concentrations I might have notice the connection but if we go back to this diagram here it turns out you know like now we’re talking about the carrier distributions throughout the device but they’re exactly equal to the probability that if you generate a carrier at this plane here for example you know like just the concrete example if you had a plane of generation here and you had an infinite surface recombination velocity the ratio of this to the value here would give the probability of collection so you know we didn’t know that at the time so we’re sort of thinking a current and voltage in the seller sort of separate but they’re fundamentally linked by this relationship so you know like if you look halfway across the year this will have you know about half the value that it does at the junction so that gives you a 50% probability of collection of carriers which is which is you know about right because this is acting as a sync for carriers if your short circuit of a junction that means there’s no excess carriers there so that’ll act as a sync so the the sync here and the sync here just battle for the carriers and you have a 50% chance of collection however if you do something that’s good for the voltage and try and reduce the surface recombination velocity you can see that you bump up the probability of collection of the carriers as well so you know doodling with these little diagrams of carrier distributions in the dark you know gives you a lot of information about what’s happening in the cell I might add this is for the case of uniform doping it gets a little bit more complex if the doping is non uniform but the same principle applied so these expressions here that were derived for the carrier collection probability given in the red book actually described these carrier concentrations in these two limiting cases as well so you know if if the diffusion length is reasonably small this might be quite a straight line it will bow down a little bit from a straight line and be described by that relationship there so that’s something that if you understood that you know that a lot of people working in photovoltaics don’t know so you had a many of them okay now getting back to the technology side of things I’ll talk about how we applied all this in improving the cell design this is the chart of efficiency versus time and we started making a contribution in 1983 this was our first world record for silicon cell efficiency but what we did to try and control the surface recombination velocity we you know we were trying to fix up the emitter of the cell first so we said we got to get that s close to zero as we can it turns out a native oxide on silicon is a really good way of neutralizing the electronic activity of a silicon surface like that had been known for decades and that’s why silicon you know was so successful in micro electronics it you know the this native oxide or oxide the oxide that forms if you heat silicon oxygen it it really forms a very good interface with silicon and that was the underpinned the micro electronics industry for decades it’s moved on from that now but that was the reason it got to where it is now so we initially put oxide along the surface even under the metal contacts so metal middle to silicon forms an infinite recombination velocity contact so that’s really bad news wherever you have a metal contact to silicon s equals infinity and you get all those linear gradients and extra current flow etc by putting this oxide ID under there we could we could make that look less damaging to the minority carriers that region so this was the structure that gave us our first world record 1983 so mainly it was a reduction both along this surface and here it turns out you know have you know like we’ve got a metal here and then an insulator and then semiconductors over oh you might not get any current flow there because you’ve got an insulator but we had to make these oxide regions very thin here so the carriers could flow through that insulator by quantum mechanical tunneling so that’s one of the successes of the of the Quattro mechanics developed in the 1920s was you know it predicted currents electrons could flow through regions that were classically forbidden and going through an insulator is a region that’s classically forbidden but if it’s thin enough the electrons can get truly so had to be about 290 meters 20 angstroms for that to occur in this region here we could have the oxide thicker but you’re in two different problems if the silicon index is very high it’s about three point five relative refractive index the better of

oxides about 1.4 1.5 so the it turns out optically you what a gradation in index going from silicon to air to reduce kept the lowest possible reflection so having a low index layer you know in the stack isn’t a good idea so these and reflection coating materials that we chose were progressively we had a double layer coating so we had one layer that was high index zinc sulfide which is a quite a high index and then another layer of magnesium chloride to try and grade the index of silicon down to here having this oxide in the way wrecked your X the optical design but if you keep it very much smaller than a wavelength the light doesn’t really see it so so here we had different constraints on how we couldn’t think we could make the oxide so it was about three times thicker than in this region you know just keeping it so it didn’t influence the optics so that structure and you’ll notice we have no texturing it’s just the planar device structure but we’re still over to beat the the black cell efficiency record so essentially what we’re doing by this approach is we’re just reducing the flows of carriers both in this region here across to the contacts across to the surface so you can regard this as sort of a you know a lot of one-dimensional things all together in parallel but this flows to the surface here of holes that’s going to be reduced if you have a good interface here and then the flow to the contact is reduced by having this oxide layer here so you get improvement both in contacted areas and non contacted areas of this approach oops I think I’ll skip one day yep I did the following year we we simplified the structure a bit and got better performance so often making things simpler allows you to to get better performance a less exacting less demanding to make the devices and you can pump them through more quickly and your simplification of what you’re doing is always a good idea so what what we did here we retained this oxide here so it’s thickness as I said was determined by optical constraints but instead of having the oxide underneath the metal we used a different strategy to reduce the influence of those contexts we made the contact region very small so what we have now is we have good surfaces all along here but bad surfaces right in the contacted areas but if you reduce the contact area so as I said this isn’t drawn to scale this width of this region here in our devices was probably about 2 microns the depth of this region here was about 4.4 of a micron so these contacts are actually you know spread much over much over you know if I drew it to scale you’d have them spread over a much larger thickness than this so they’d go you know about the extent of this thing here so it merely a geometrical thing you’ve got regions where you’ve got enhanced current flows to the contacts here but you’ve kept that the total flow down by keeping those regions small so that was a different way to improve your meter pro properties and again most of our improvement came through improved voltages in these structures better move on the next big jump was by introducing texturing so all those little pyramids that a standard approach don’t work well well we had trouble making them work with the photographic approaches that we’re using to define a lot of the features here so we work with more controlled structures in this case we work with what we called slaps or micro grooves so these are performed by the same crystal planes and the silicon have the same angles and everything but we just confine them to a you know I guess it’s a two-dimensional structure rather than three-dimensional and that got us to the first twenty percent efficient silicon solar cell so that was that was a real milestone because back when the first cells were developed which you know particularly in the 1950s where there’s a lot of excitement and that page headlines on the New York Times and everything people started trying to work out what was the efficiency limit on a silicon cell and a lot of people came up with 20% that was about the best you could do so that was like the four-minute mile in athletics you know a bound that could be sort but probably never attained so this was a real milestone achieving the first twenty percent efficient cell and there’s a very happy group here that that did achieve it so you might recognize some of the faces visit Stuart Wenham who was my number three PhD student and II Blake is who was number two PhD student he’s now a head of a group similar to ours at Anu Chi Minh who’s here at the back he was number eight PhD student she went Mohan was number five gel was number six but

all these people have gone on to do great things within the industry so this is just where some of them have ended up so gene Marcia of course working with technology transfer Mohan’s been CTO of a number of the major solar companies including Trina and now CTO of Hanwha Saleh one Ted here he has been very he’s probably done the most work in terms of getting the industry to where it is now he established the production lines at two of the major solar cell manufacturers Sun tech and J a solar as well as a couple of other ones that are in the top twenty and so on Joe is some vice president of technology at China’s energy so everybody’s calling on to do something interesting in there into solar area okay next topic I’d like to talk about it’s light trapping so we we had some questions aren’t last week this is Eli your bono that you did you know the most interesting work with life trapping although Adolph goats burger it’s sort of thinking the same ideas at a similar time but like trapping relies on the fact that this the index of silicon is very much higher than that of air so 3.5 times that of air approximately so you might remember back to your physics says you know total internal reflection if you have light troubling in high index materials so if you get light into a silica and if this was just a flat reflector which is what people had been well what what they use in space cells is flat reflectors on the back if any light reflected from that rear reflector just bounces straight back out again just follows the same optical path but if you do something to you know roughing the rear surface or something or other you can get the light sort of reflected off in other directions and if it’s reflected within the escape cone of the silicon it’ll go straight back out so the escape cone of silicon lives within about 16 degrees of the of the surface in a range of about 6 I think it’s 8 degrees half angle so if it’s reflected within that half that half angle it will you know at this angle here is small enough it’ll just go out of the device getting refracted this way when it hits the interface however if it’s some reflected a little bit more oblique angle it gets totally internally to reflect that reflected at the surface internal total internal reflection is like a hundred percent efficient process you’re unless you have dust or something on the surface of the device that can help couple the light out but you know if you have nice pristine surfaces total internal reflections effectively a 100 percent reflector so you can’t do too much better than that so eli developed the theory of what are known as Ghatak light trapping schemes which means schemes that randomize the direction of light you know sort evenly over all angles so that’s you know just completely randomizing the light as random as it can get and he showed that the light trapping if this was a thickness W by using a randomizing reflector here you could increase the op pathlength in that material by a factor for n squared so as I said n is very high their effective index for silicon is 3.5 so if you figure out for N squared you get something like 50 so by doing something like this you can increase the path length 50 times over that of a single pass across that layer so that’s quite a important boost if you’ve got a solar cell that’s um you know 300 microns thick and you multiply that by 50 you know it’s something like what is it 1.5 centimeters that’s sort of the absorption depth that you that you’re working with so it’s really quite a substantial improvement in terms of trapping light within the device we were interested in these schemes that were based more on geometrical structures and we we thought we could probably do better than that because you know like randomizing things is like not using your engineering skills where we were engineers so is that all we can engineer something better than the randomizing stuff so we tried to to see if using these geometrical features we could get better than that and Patrick Campbell who’s still within the group and myself worked out that you could you could get even better than 50 times you get a lot better than 50 times you if you but you had to pay a penalty which is what I mentioned last week if you restrict the range of angles that you’re willing to accept like from and you know that’s that might be all that silly to do you can correspondingly increase your path length so you know as shown here so 50s is just like a lower limit since then there’s been an integral form of this

equation developed so if you have a variable angular sis sensitivity of a cell you can now derive what the path length enhancement would be so that you know that so that sort of established the baseline I might add these are all for very thick cells much thicker than a wave length and there’s been a lot of recent work on what happens when the cell thickness is small compared to a wave length can you get enhancements beyond this and the answer is not in the average case but in sub extreme cases yes very very much so so that’s what’s meant by light trapping and it’s a function of the high index of of silicon you know one way of no white going to that point of time so we started to think about that incorporating that into our designs around this stage so that led to the second era of cell design most of the work we’d been doing up till then was was based on fixing up the emitter like the front surface of the cell the back of the cell was pretty standard but then our our run of efficiency records we got our 20% and then pushing beyond just refining what we’re doing and then this group at Stanford University um interrupted our run of records with this structure here which was a double sided cell structure it’s quite unusual this is one of their drawings but the sunlight comes from this direction here I always used to point out your being in the southern hemisphere the sunlight sort of comes from on top of you rather than from underneath but now this this structure here you know it’s quite interesting just got a silicon wafer and all the actions on the back of the of the cell so this is the rear Junction cell is probably the correct terminology for it it’s got both the n-type Junction and the p-type where is that it seems to have dropped off this drawing oh it’s underneath these regions here so you can’t see it in the story but it’s a little p-type doping here so your European Junction activities on the rear it turns out you need very good material for this structure to work because most of your carriers are getting generated near this surface they’ve got to find their way over to this rear surface to to work so you need very long diffusion links for the structure to work and very low surface recombination velocity along these surfaces so it’s quite a demanding structure but it sort of pushed the technology in that in the direction of going to more sophisticated approaches so this is what the commercial cell looks like drawing up the right way up this time you know all the action on the rear so if you look at one of these cells on the front it’s just black bell but there’s nothing there apart from this texturing and a native reflection coating so virtually non reflective surface and then on the rear that’s all this action metal contacts you know running the length of the cell and so on so all the actions confined to the rear so that raised the stakes we then started to look at what we call double sided structure so we working on the emitter now we have to look holistically at the whole cell structure and this is the device that ultimately resulted from that which we call our pearl cell passivated emitter you know we’ve done all the work on the emitter really locally diffused so we we went to this idea of just having you know Rhea contacts in selected areas similar to what the Stanford group had used so that’s the pearl cell structure so we used all the tricks that we found for the that we used in our 20% efficient cell we’ve gone to a more sophisticated texturing approach sort of a 3d approach and this this one was chosen and designed to maximize the light trapping in the device so we incorporated features into the exact patterning and so on that that enhanced the amount of life that get trapped in and we get cloaked quite close to that 50 times trapping of the light within the material wherever we made contact to the device we took one extra step we did a heavy diffusion which was what was used in the Stanford structure no gene we’re running out of time so you remember this is p-type this is P plus the electron concentration has to be lower in the P plus compared to the P you know the the product in thermal equilibrium you know is the NP product is constant so this P being higher means n has to be low so so what doping in the contact region does it suppresses the carrier concentration in that region it suppresses the minority code and so that’s going to suppress all those gradients to the to the contact so you know that’s a way of reducing the contact recombination so we’ve incorporated that in the front side so wherever we have a middle contact we have it sort of protected by this heavily doped region that stops the minority carriers from getting there and then apply it to the rear as well and then we use oxide in shroud meant to pass about the whole surface initial initial an additional refinement here is that we

use the electrostatic effect of this rear contact to control the recombination in this intermediate region here but I won’t have time to go into it okay moving on another reciprocal relationship I’ll just briefly touch upon what I want to talk about is the shockley-queisser limit on efficiency so this shows the limiting efficiency for a single Junction cell versus the bandgap and in 1961 I think Shockley and Queisser derived a limiting theory that it apply for any single Junction solar cell and this is a shown here these are a lot of experimental results for different cell technologies including our twenty five percent crystalline silicon device but shockley-queisser looked at the case where the only recombination property processes within the device for radiative so in our silicon devices you get a lot of radiation like this defects in the material so you get recombination occurring through defects and this all j process becomes very critical if you’ve managed to reduce this this radiative process is quite interesting in that you the extra energy of the electron that’s released when the recombination event occurs causes a photon and that photon has energy above the band gap of silicon of the material that it’s occurring in so it can be reabsorbed within the material somewhere else in these processes you know you lose the carrier you’ve lost it in this process that sort of carrier sort of lives on in the form of this photon so the photo can bounce around in the device and it’s not necessarily dead but that makes it very complicated to analyze what’s happening in the device and people had sort of struggled with that blackbody radiation I hope them for you familiar with this this is just the energy distribution from the Sun versus wavelength very similar to a blackbody radiation you get this wavelength distribution that Peaks that particular wavelength depending upon the temperature of the blackbody I’ll just go over this and probably skip over the rest because this is quite important fundamental theory that describes what the limits on self performance what really determines him but shockley-queisser did quite an elegant analysis they said you know we’ve got a solar cell here we want to work out what the limiting open circuit voltage in particular of that structure is you know the currents quite easy to calculate you know how many photons are coming in if each gives you a carrier you know you’ve worked out the current but it but what determines the voltage so they said you know if this device you know assuming it was a black solar cell and had good and reflection coating and textured surface or whatever if it was just sitting in the in the room like this room at room temperature how much like would be emitting and if we remember back to Kirchhoff’s law you can work that out you know a black body emits radiation characteristics of its temperature so you can work out exactly how much radiation this solar cell is emitting at at room temperature you know sitting in the dark if you then play a voltage to the device you know what happens oh we’re going back one step if you look at the strongest optical processes within the structure at the energies corresponding to the band gap of silicon these are these absorption processes across the band gap so there’s nothing else that comes close to matching that in terms of the strength of the process so if you look at the blackbody that’s being emitted by this solar cell sitting here in the room without any voltage or light shining on you most of that radiation that’s being emitted at energies around the bandgap is occurring from these radiative recombination and regeneration processes across backwards and forwards across the band gap so automatically taking into account all these photon recycling and total internal reflection effects that really complicated analysis looking from the inside out looking from the outside in simplifies things down enormously so if you then apply a voltage you know we know from what I’ve said before all the carrier concentrations increase exponentially throughout the device structure and if you’ve got good surfaces you know they they increase just about uniformly throughout the device structure in fact if you want to collect all the photo carriers they have to increase uniformly throughout the device so what will be the what will be the total recombination the device in the dark with a given voltage across it it’s just going to be that blackbody radiation that lies at energies above this band gap here exponentially enhanced by the voltage so this is the expression here yeah this is the radiation from the Sun multiplied by a black as multiplied by the young boy at 6,000 Kelvin which is the sun’s temperature approximately and this is the if you like the radiation being emitted by the the cell it’s the blackbody radiation that’s above the bandgap this integrals just

done above the bandgap for both exponentially enhanced by this factor here which is the factor that increases their carrier concentration to the device so you can work out the limiting efficiency of a solar cell just by looking at blackbody radiation characteristics that’s what shopping quasar did looking at experimental devices like this is a blackbody emission this is a blackbody spectrum at 300 Kelvin I showed it for the Sun before so this is the photons per square meter or something or other up here so this is a number of photons being emitted by a black body you know our point of our solar cell sitting here of room temperature so you and this is wavelength so 10 microns here this is the visible region here which we can have a closer look at the visible region here you know you won’t see much radiation from device yeah I can see you now not because you’re emitting radiation but because radiation is getting reflected from you so the eyes aren’t responsive to this region that these bodies are emitting but if you look at where silicon’s bandgap is on this scale you start getting this little bit of radiation emitted so this about a radiation here determines the about a recombination within the device that’s above the bandgap at you know without any voltage there and if you increase the voltage this just gets exponentially enhanced if you look at one of our solar cells this is a radiation emit it turns out that it’s you know this is wavelength here it turns out it’s similar to one of those little blips it gets refined a bit because um you know because of some of the details involved but it’s pretty much the same so solar cells actually do behave in their shockley-queisser mode but one very important relationship that I’ll just mention but it’s important it’s probably new like again I’d say 99% of the people working in the field not aware of this relationship so if you’re cotton under what’s being said you know you’ll be ahead of most of people but it turns out the solar cells don’t emit quite as much as that exponentially enhanced blackbody radiation only a perfect solar cell will do that and you were al found just you know a few years ago that the actual expression for the amount of light emitted by a solar cell was the blackbody term multiplied by something that were perhaps familiar with the eqe of the cell acting as a photovoltaic device so again sort of a coupling of different types of applications this is type sort of an LED type of application you’re looking at light emitted but it depends on a parameter that’s you know that’s relevant to a device operating as a light detector so I’m you know quite interesting relationship so this is our um this is the e QE which is just a number of electrons you get in the circuit for incident photon this is the e QE at one of our 25% efficient cells very good across most of the visible spectrum drops off at the bandgap and this fall off here determines the shape of this curve very important relationship some of the members of our team have developed a commercial equipment based on that that now can inspect silicon cells looking at the light emission so this is showing a multi crystalline cell and the pattern you get is showing you the eqe variation over the cell surface or the quality of the cell over as it vary spatially so very handy tool diagnostic tool final bit just talk about the future I’ll just have to skip some of this some of the cell designs that people looking at like within the group here we’re working on laser dope selective emitter structures where we automatically we get away from that screen printing process automate align the context to regions that were heavily doped using lasers to provide that patterning that’s another structure sub tech been our most prominent licensee had a lot of these cells in production although that says scaled back a bit over recent years the other cell you might have heard about as a hit cell structure uses higher bandgap material in this case hydrogen doped amorphous silicon which has a higher bandgap than the normal silicon to form the emitter and the base contact emitter wrap through cells a lot of interest in other approaches getting real contacts apart from the Sun power approach so having little holes through the wafers that brings the contact through or you can bring the whole metal pattern through the contact as well so that’s the metal wrap through cell so looking further ahead into the future where do I think see things heading it turns out that the thermodynamic limit on solar conversion is not 25 percent yeah better stop about 74 percent for global sunlight is sunlight coming from the whole sky lots of approaches have been suggested for getting from a single cell limit the Shockley limit shop the quasar limit up to that 74 percent but the best way of doing is stacking cells on top of each

other and it turns out that silicon is not a bad choice for the bottom cell in the stack like if you stack two cells on top of silicon this is your limiting efficiency with the silicon as the bottom cell compared to an unconstrained choice so pretty good choices of bottom cell so this is where I think things might be going actually finding techniques for growing thin film cells thin film cells are very good quality on top of silicon to sort of supercharge the performance of the device taking the limiting efficiency you know from 30 percent which has holes for present devices to something like 50 percent for a 2 or 3 cell stack so I better finish there but thanks for your attention