In this presentation, Prof Nigel Browning will discuss the current state-of-the-art in dynamic in-situ TEM/STEM and describe the advances anticipated with the RUEDI instrument.
Definitely thank you to the organizers for inviting me give me a chance here to talk about Rudy um Rudy is relativistic ultra fast electron diffraction and imaging this is a project which is part of the Wave 3 infrastructure funding strategy through ukri and we’ll finally get into the final stages on Friday
Although while I was stuck on the m6 coming down here we had 20 odd emails about how to change the proposal with only two days to go so um what I will talk about is what we’re trying to do with Rudy how it applies to electron microscopy how it applies to
Catalysis and what the overall goals are moving forward with the system and in particular how everyone here can help us to get this to work out okay well I spent many years in my career developing uh electron microscopy and trying to get things to work out to get Atomic resolution and really if
You’re thinking about doing electron microscopy in terms of scanning transmission microscopy or TM really there’s no point doing it unless you’re going to see atoms and so the goal really is to be able to quantify what is happening to all of the atoms in the structure how do they move under the
Conditions that you’re you’re you’re running a reaction and how does that lead to the functionality that you’re expecting to get out of the system now all of this requires quite a lot of work in terms of electron microscopy to get that to work it’s a very very difficult experiment we need to
Understand what the temporal signature is um you know if a reaction or if the whole process is actually developing on the millisecond time scale there’s no point using femtoseconds to measure it and vice versa we have to understand uh particular and I’ll go through this in
Quite a lot of detail what is happening to the Electron Beam it’s one of these things where the Electron Beam is a blessing and a curse if we didn’t have an electron beam we wouldn’t see anything but the fact we have an electron beam means we’re changing
Everything and so how do we go through the process of understanding exactly what’s going on with and without the beam being there we also have to understand the starting structure problem I’ll show this little movie in a minute um of where the starting structure problem comes from we
Have to control the environment the environment needs to be something relative to realistic environmental conditions in terms of temperatures pressures whatever you want to do in terms of the reaction and then of course we’re going to generate lots and lots of images and Spectra that comes out of
That and the question is how do we correlate everything together how do we understand what each of the individual measurements is contributing to the overall problem so I have here is a little movie Taken in a stem quite a long time ago now um just showing osmium clusters sitting
On a MAG magnesium oxide support and you can see the Clusters are moving around all over the place you can see individual atoms and we run it again see individual atoms shooting around all over the place and you think oh this is absolutely fantastic we can do in situ
Or operando microscopy and get it to it the only problem with this is we’re actually not doing anything in that experiment all that’s happening here is the changes that are happening to the surface structure in taking one image after another using the Electron Beam now when you go through and you think
About this and what’s happening in terms of the changing the structure this is something that happens independent of electron voltage it happens independent of the damage mechanism it’s really just related to a charging and a heating effect that’s taking place inside of the microscope when you’re taking the images
Now the reality of this is of course that electron microscopy has to deal with the fact that it can see all of the changes and so we’re creating our own problem of having to deal with the changes that are occurring on the surfaces of structures when we’re trying
To image them that doesn’t mean that these things don’t occur with other experimental methods it just means with electron microscopy we have to be able to control it because we’re going to be under the process of looking for these things okay so what I’m going to talk about is
How do we deal with that how does that lead into the design structure for Rudy and what do we expect to get out of Rudy in the end we’ll be able to deal with these things okay so very simply when we go through the process of dealing with stem images
Of supported clusters the easiest thing is just to turn the beam current down the problem is that most people refuse to do that because when you take turn the beam current down your images don’t look very nice and so your images don’t look as nice but the interpretation becomes a little bit more
Straightforward to do so what you can see here is exactly the same structure that we had in that movie except now what we’ve done is turn the beam current down to the point where we can still see Atomic resolution but we’re not driving the motion of the atoms on the surface
And now you can actually work out very clearly what the orientation of the Clusters are what the general configuration is how do they sit on the surface of the magnesium oxide and what are the interaction points during those interactions that take place so it’s pretty straightforward if you go
Through that process now it turns out that you can do that for multiple clusters and this is just a summary of results here it doesn’t matter what the cluster size is and of course here we’re talking about very very small clusters where one to ten atoms but as you go
Through the small cluster sizes you can actually work out that as you get bigger and bigger you start to get different orientations obviously the more atoms that are involved excuse me you get different configurations those different configurations lead to multiple images that look slightly different but effectively are the same structures or
The same number of clusters or atoms in the cluster they’re just configured on the surface differently so this also indicates the problem that we’re going to have as we move up to clusters that are one to two nanometers in size is that you’re more and more atoms you put
Into the system the more the configurations are going to be the same but look different in our experimental images when we try and interpret them so that goes into our data and our data analysis side that we’re going to have to be able to deal with
Now this applies even if we go to more complex support So This is actually a set of images from a zeolite structure and this is the idea where we have a structure before catalysis is being performed and we have a structure over here that after the experiment has been
Performed so here you can see single gold atoms we can actually using it’s not Atomic resolution it’s very difficult to get Atomic resolution images out of zeolites simply because of the damage mechanism that goes into it there are some zeolys that damage less than others but effectively you always
Get to a point where you’d love to get higher resolution but you can’t do it because you damage the system in this case actually Again by turning the beam current down a little bit working towards the use of the Symmetry we know what the zeolite is in the
Structure you can actually identify the locations in the structure of where the atoms sit and so you can see a change in the general configuration of where the dopen atom sit after before and after the actual reaction is taking place but in this process this is not an
Operando experiment it’s a before and after what you’d really like to know is why do the atoms move from one side to another what is it about that configuration that’s giving rise to the most of the atoms so that in general you end up with more on one side before and
More on the other side afterwards so that’s where we can start to think about um oh actually one more on the high resolution again it turns out that it it really when you start to use the knowledge that we have about the symmetry of the structures that we’re
Looking at you actually don’t have to um get Atomic resolutions to be able to do interpretations at the atomic scale and this is something that a lot of electron microscopy has missed out over the years as soon as we get to Atomic resolution as soon as we get to
Aberration correctors as soon as we get this power to get these fantastic images we forget that we don’t actually have to use that power to do a scientific interpretation and actually just backing it off a little bit will give us the information that we want and again here
This is actually a platinum case you can actually just look at the particular sites you can find the individual Platinum atoms or clusters wherever they sit this example is just some single items but you can do it for clusters as well and as you go through that process
You can identify where they’re going to go and as part of the reaction but again it’s not in situ or operando what we’d like to be able to do is to get operando experiments to work all right so the example of an operando experiment I’m going to show is one of
These multi materials um where this is a case of an oxidative dehydration reaction um the idea is that these uh catalysts exist as these sort of nanoscale particles that have different surfaces you can go through and you can characterize the atomic structure it’s shown at low magnification here but this
Is atomic resolution image we can find the surfaces when you go through and you look at plotting the activity versus the General Distribution of the system you find that there is a general relationship between activity and the types of surfaces but if you just integrate over several experiments what
You find out from this is that essentially there is an increase in activity with an increase in surface area which is not really a surprise in terms of catalytic activity but what we’d like to do is to do some in-situ operandiodynamics to see if we can work out what it is about the
Reaction that’s giving rise to this performance so this is done with a set of in-situ gas stages um the experiments I’m going to show were done on an etem for some of the experiments and on some of the other experiments were done in a windowed gas stage these are designs for electron
Microscopy that have been around for about 10 years now allow you to get the info to put a gas pressure in the windowed cells you can get up to about one atmosphere of pressure in the e-tems it’s much much less than that it’s about 20 millibar in terms of what you can get
But effectively when you do this and you change the pressure what you find immediately is actually the Electron Beam scatters from the gas as well as scattering from the Catalyst you’re actually trying to look at now what this means is effectively that you get rid of your contrast mechanism so effectively
The background that you get gets rid of your contrast mechanism so that your image gets less contrast when you’re trying to look at the Catalyst the way to overcome this of course is to turn our beam current back up so we can get more contrast into our image but the
Reason we don’t do that is because you just ionize the gas under that circumstance and you create more ionizing species which means your reaction will just go a lot faster because you’re actually giving free electrons to the system and then all of these Reactions where I’m talking about the effect of the beam
It’s not the primary energy of the beam that’s the important thing what happens is when the beam that we put into the sample is hitting the sample it creates secondary electrons those secondary electrons have an energy of about one to two EV that means they’re exactly where
You want to be for a chemical reaction so if we think about what’s happening when the Electron Beam is there what it’s doing if we don’t control it is it’s giving free electrons to the Catalyst giving free electrons to the catalyst is doing what a catalyst would
Want to do anyway and it’s providing those elections those electrons to the reaction and that’s making all the reactions go faster so effectively what’s happening when we try and do operando is we’re shrinking everything down we can’t fix it by going to a higher beam current because if we
Do that we’re going to change all the kinetics of the reaction so we have to work in this very low area of very poor contrast but in this case it’s still enough for us to get Atomic resolution images the other part of this which becomes very important and this is something I
Think is pretty standard now is that you don’t want to use any form of phase contrast Imaging if you’re trying to look at a catalytic simple samples or heterogeneous Catalyst samples simply because the the phase effect is being washed out by the thickness of the catalyst so you need High current to get
Good phase contrast Imaging if we want to get good Atomic resolution imaging then we need to use Z contrast or some form of incoherent Imaging method to be able to get that to work out and that again reduces the effect of the background scattering in terms of what we can see
Now again this all ties in together whether it’s a windowed cell or whether it’s an etem effectively all that’s happening is that the effect of the windowed cell that we’re looking at gives us an increase in the thickness of our sample so that’s just like having an
Extra bit of gas pressure there it’s usually amorphous silicon nitride so it’s a pretty heavy gas that we’re talking about so it’s just going to reduce the contrast even more but it’s easy to understand what that’s going to be and then we can go through and look at the effect of the
Um of the actual reaction so in these systems we can actually set up the multis under the the ethane reaction conditions so this is about 450 degrees we can get the pressure up in an e term as I said to about 25 millivar up to an atmosphere pressure
Inside of the etem when you don’t have any gas pressure you can see here is the atomic resolution image here you can pick out all the individual elements where the malignant Vanadium delirium is going to be you have these motifs where you can actually you know identify the structure
Um when we try and do it inside the gas and it starts to get a little bit more blurry you get things that look like this again these images in a lot of cases are just put through a low pass filter so you have the resolution the microscope and you reduce the noise
Effect there are better ways of dealing with noise which I’ll talk about as we get a little bit further into talking about Rudy but effectively we can see the changes in the structure and in this case the interesting part of it is that as this reaction progresses the sample
Loses tellurium so tellurium diffuses out of the system now in the catalytic reactions that were being done there would always be tellurium on the inside of the reaction vessel and when we do the atomic resolution we actually see the same thing is happening here there’s very little changes in the structure
There is oxygen that’s going to change at high temperature and then the main effect in terms of the metal atoms is that we get seeing tellurium coming out of this Central part let me putting this in the right direction coming out of this central part just here so these Bond lengths are
Changing we can see that in terms of the oxygen and we can see it in some artillery if you go through the DFT calculations what’s happening is when the tellurium is coming out you’re effectively making the oxygen more reactive so the bridging oxygen axons that sit around the possession of the
Tellurium are actually telluriums going away the bonding to the oxygen is going to be less so it goes more negative which makes the oxygen more reactive when you have the ethene in that vicinity and so effectively the diffusion of the tellurium in the sample is what’s helping to change the
Reactivity of the oxygen atoms which goes through the process to give us the catalytic reaction and this is effectively the process that you can come up with if you go through the whole idea of what’s going on in terms of touring now you could say well why don’t
You just put tellurium not put tellurium in there to start with and then the whole thing would go in that way but the structure changes if you don’t have tellurium in it you don’t end up with the bridging atoms and so you don’t have the oxygen in the position where the
Diffusion of the tellurium during the reaction can actually lead to the the activation of the oxygen acid so this is one of the cases where we’re going through and looking at Dynamic ocarandotype experiments where the mechanism of what is actually taking place is something that you would never
Be able to get from a before and after type experiment so in all of these things that we’ve gone through with the electron microscope there are a few things that we we sort of can come up with we need to be able to look through thick samples
If we’re going to do operando under gases and liquids we need to be able to reduce the dose we need to be able to look at very high speed because if you think about surface diffusion or even bulk diffusion of single atoms inside of a support then the speed that they’re
Taking place at means that they’re going to go through individual hops on the scale of microseconds to nanoseconds so the time to see each of those individual hubs means we have to get to a very fast way of doing the Imaging so all of this combines together into the plans that we
Have for the design for Rudy so Rudy is this idea of having a national user facility based at darsbury the goal is to make a microscope that will have all of the things I just talked about good Imaging fast be able to look through thick samples get high spatial resolution or high temporal
Resolution so if we look at the the actual design of Rudy this is the sort of bits that go through it this little bit at the top here this is actually a picture of the slack uh Mev system this thing is about 20 meters long from one
End to the other doesn’t look anything like an electron microscope that any of you will have seen before um but it effectively allows us to create these very high uh energy pulse beams now the reason that we’re going to high energy and going to Mev beams is we
Need to look through thick samples you want to be able to look through something that is thick so that you can get high pressures so effectively inside of our cells we should be able to go to high pressure because we can look through thick things with the Mev system
The other bit which I’ll talk about as we go through is that Mev system when you’re at two to four megavolts then relativity helps you in terms of understanding what happens to the electron system we need to control the electron Bunch when you’re looking at things like time dilation length
Contraction all of that helps you to have higher B managers which allows you to manipulate the beam so what we’re going to do in the system effectively is take something which is one of these electron diffraction system add the type of control for in situ or
Operando stages that you get out of an electron microscope and to do that you have to add oops an electron microscope lens system to it so you can have the goniometer control so that we can have a cryo stage we can have a gas stage a
Liquid station open cell stage we can do heating we can do straining all of the things that you can do in a regular microscope but now because we can look at a thicker sample the biggest problem in conventional microscopy is to make the stage small enough so that you can
Get the electrons to go through it that ceases to be a problem here so we can actually have more flexibility in designing the experiments the other part that goes into this is there’s been a lot of development over the last few years in terms of CMOS detectors CMOS detectors are also
Ideally suited for looking at things with very high energies and so we’ll bring in some of the highest energy detectors into the system as well and that will allow us to get better signals better resolution now we have five themes that are basically involved in this
Um the first one is dynamics of chemical change this is related to fundamental chemical reactions so the idea is to look at chemical reactions in gases in liquids and in solids and in many cases you can actually look at very similar reactions and understand what the physical state of matter is doing in
Terms of controlling the speed of a reaction and so this is related a lot to diffusion mechanisms so understand the physical principles of what’s happening when you get diffusion under various systems when we talk about materials and extremes effectively inside of the Rudy system we can focus very high power
Lasers into it we have a thick specimen stage so you create shock fronts these are the kinds of propagations where things are exceeding the speed of sound so it’s the type of reaction that occurs in earthquakes it’s also the type of reaction that occurs during explosions
And so you can look at how to control materials that are if you like earthquake proof or resistant to damage under explosive forces now you can also create plasmas using those electron beams so we can create the same kind of plasmas that exist in astrophysical environments and look at
The Dynamics of plasma interactions using the Electron Beam to image those systems so that gives us a lot of flexibility in terms of what we’ll go into that um in the system that we will hire we’ll also have a neutron line coming in so we can look at ion interactions with
Materials now that will affect both things in terms of nuclear energy and how to understand uh making materials to make safe nuclear energy also it fits in with medical Advanced Medical Sciences as well when you’re looking at using your radiation for various cancer treatments and all those kinds of processes
That ties into the biosciences the biosciences I’ll get back to the other two in a minute the biosciences allows us to effectively because we have a thick cell we can look at what happens inside of an actual organic cell so in a lot of cases where people are looking at
Protein Dynamics when they’re looking at chemical dynamics that occur for the biosciences you have to purify the proteins you have to purify the structures that’s something that takes a lot of time you don’t have to crystallize it and you have to go through the interaction that is really
The limitation in terms of the advancing of the science using biological systems because we have high voltages we can put the whole cell inside of the microscope and we can image through a whole cell and look at Dynamics as it occurs within the cell now the resolution will be different
Than it will be for our protein Dynamics but there’s lots of people doing protein Dynamics by looking at the whole cell we can understand how those individual protein Dynamics transfers into the actual reaction that’s taking place in the organic system um Quantum materials um there’s a lot of things related to
Electronic and magnetic phase changes um most of the other uh systems that are here chemical change materials and extremes biosciences are talking about Atomic movements when atoms move it actually quite slow even though a chemical reaction is on the order of femtoseconds the movement of the atoms that actually after that reaction takes
Place is on the order of nanoseconds so you don’t actually have to get to femtoseconds to be able to see Atomic motion when it comes to things like electric fields and magnetic fields however things move a lot faster and when they move on those kinds of speeds you’re talking about femtosecond types
Interactions so here we have a research theme that’s looking at understanding how to control electric and magnetic fields in modern semiconductor systems or nanoscale systems which can be used for Quantum Computing and various other applications the theme that’s probably more relevant here is energy generation storage and
Conversion and this is looking at things like we can use lasers to propagate any reaction so we can look at photo cells we can look at the process of both the excitation when you actually absorb the light and we can look at the process of de-excitation of how the electron comes
Out of the system so under a lot of cases we can look at individual interfaces individual defects individual systems to understand that process when it comes to conversion obviously catalysis will fit in into that side of things and we also have storage effects of looking at batteries so if you want
To look at what happens to the electric double layer how that changes around interfaces we can use light to propagate a particular reaction create a very rapid change and then look and see how those interactions propagate across any kind of interface whether that’s a solid liquid interface in a conventional
Battery or whether you want to go into something like a solid solid battery and look at those interfaces and the diffusion of ions so effectively there’s a lot of different science that will fit into this system it also involves a lot of updates in terms of the technology of
Electron microscopy taking it into what is a modern type of electron microscope now there are quite a few of these facilities around the world um they exist to do diffraction but not to do imaging and so all of these facilities that are uh Japan us China in
Europe there’s none of them in the UK but they’re all trying to do diffraction and none of them are doing the Imaging and in all those five themes that I talked about the biosciences was all Imaging a lot of the energy theme is Imaging materials and extremes is all
Imaging these are things that you can’t do using the equipment at various places around the world and so what we’re going to try and do with Rudy is to build the lens system um the lens system is the most complex thing that you can try and do under Mega
Volts and so effectively what happens is when you try and build a lens at this kind of voltages you get saturation of the magnetic field and so the focusing effect goes down so we have to go into novel designs lenses interactions with quadrupose octopoles build those onto
The whole system and that will allow us to get sub nanometers Imaging with probably around one picosecond of temporal resolution which should be plenty of of temporal resolution to see diffusion on surfaces and in bulk should be plenty of resolution for looking at shop interactions all of that will allow
Us to to get that to go forward now on the diffraction side of things we should also be able to get less than 20 come to seconds to fraction results and this is actually the fastest you would ever get of any of those facilities anywhere in the world in
Terms of doing the diffraction okay so all of this is tied together the instrument itself is designed to go to darsbury um it has to be on a national lab facility because you can’t Shield um for Mev electrons using the metal casing of the microscope so it has to go
Into a bunker which means that we can’t build it at the University of Liverpool and stick it on top of the campus which is on a train line and all these kinds of things will go horribly wrong if we tried to do that so the instrument
Itself is going to go to darsbury but it’s supposed to be a hub and spoke facility and the reason for that is that we are not going to put all of the specimen preparation equipment at darsbury is going to go through the collaborative institutions and as I’ll
Go through as we get towards the end there’s a lot of things that we want to do in terms of image interpretation where high resolution imaging on the conventional microscopy will help us to do a lot of the time result Imaging okay in terms of ultra fast electron microscopy and imaging there’s actually
Quite a lot of Imaging facilities that are out in the world but these are basically all related to the former group of the late Ahmed soil so um zoell did a lot of work on femtochemistry it went through that process you set up a group at Caltech I
Had a lot of graduate students and postdocs to design an ultra-class microscope those students and postdocs have now dispersed all around the world and there’s a lot of these microscopes that are out there the difference between what we’ve got and Rudy is all of these microscopes run at about 100
Kilovolts which means none of the experiments I just talked about are actually feasible under those voltages because you can’t get fast enough you can’t look through thick enough sample you’re not going to get the resolution to get Atomic spatial resolution okay so the goal in the hub and spoke
Model is also to leverage existing uh infrastructures um uh we have the materials Innovation Factory which is part of the Royce Institute of Liverpool um on the Harwell campus we have RFI which is just across the street in one of these directions um and then we have the cockcroft
Institute in darsbury now this is a process in terms of putting this proposal together and we’re not limited in terms of the infrastructures that can be involved in this this is just the three of us that came together to try and build the microscope the collaborations and things going forward
And and and me being here is part of uh discussions with the catalysis Hub about how to interact in doing these experiments um there’s a lot of work that still needs to be done to get this thing to work out right so in darsbury the the expertise
That’s going to be leveraged is that this is uh something that darsby does it builds uh these systems um let me just sell the right one yep okay and that’s what leads to uh the initial design uh for Rudy which is going to have two beam lines one for diffraction one for Imaging
Um when it comes to it this thing is about 20 meters in size the other thing is that it is about 130 million pound project to build something of this size um so where we’re at moment as I said at the start the proposal is going going in
It’s ranked highly by both dpsrc and sdfc and bbsrc um so we we we’re in as good a shape as we can possibly be obviously none of that means that there is any guaranteed funding um but the process itself means that we should be starting in about 2025 to
Build this thing and then from then onwards it will take about three to five years to get everything up and running now one of the advantages of something like Rudy which is running at Mev system as opposed to a conventional microscope is that we can make for the diffraction
Line and for the Imaging side line much bigger stages so a lot of the problem of getting reactions into a TM is that you only have a three millimeter size sample it has to be loaded in a side entry rod and you have to be able to tilt it which
You can’t do very much um if it gets too big and so that creates a very difficult micro Machining problem when it comes to Rudy none of those problems exists it’s much easier to scale things to be bigger than to scale them to be smaller and so one of
The things that will come out of this is that each of the of the themes will be designing their own stages to go into particular interactions so one of the things for example we can do is we can get a dilution fridge into a diffraction line so that means we can get
Millikelvin control over temperatures and have 20 ft second time resolution that goes through it so you can really look at the kinds of interactions that will occur in Quantum materials now each of those themes that we have is going to be designing stages the goal through
Epsrc in terms of doing this is the Strategic infrastructure I think that’s what’s called now strategic infrastructure program is going to allow people to develop stages that will then be applied into the facility much is the same for existing National facilities um various modes of operation basically it’s all about temporal and spatial
Resolution um the goal in the facility The Hub and spoke is also to set up an ecosystem of related facilities the reason for this is that the Rudy piece of equipment is is a monster of an experiment there is no way that a graduate student postdoc
Or scientist would turn up and do the experiment themselves it’s just going to take too much effort to get that to work now we are going to have postdocs on site that will be part of the of the team there but the reality is is most people are going to have experiments
Done by somebody else but that doesn’t mean that all of the ultra fast experiments need to be done by someone else Rudy is an Mev system that means it’s not by definition sensitive to surfaces so if we want to correlate surface uh work with what comes out of Rudy we’re going to need
Things like ued and leam and new leam type experiments where those allow us to look at the Dynamics of the surfaces and correlate all those together with uh with Rudy so this is where places like RFI comes in for a microscope Liverpool from microscope leads for the ulim and
The U lead along with Swansea and we already have a collaboration so that with Imperial for their Ultra fast electron diffraction and Ultra past x-ray diffraction type experiments I’ve already talked a little bit about where the energy transformation um uh experiments come in these are the kinds of things that we’ll do basically
Through that process and um most of us working on this project are either Electron Beam physicists electron Accelerator designers things related to lasers not all of us are working on catalysis and we would love to have as much input and help in designing experiments for this system as we could possibly get um to say where this is then going and why even though I’ve said it starts in
2025 and it’s going to take five years to build we still need help in designing these things right now um and the reason for that is a lot of the the interpretation is going to use artificial intelligence to help deal with the fact that we don’t want to have
Very many electrons in our system and that means that we’re going to use things like compressive sensing we’re going to use dictionary learning the better way to do that is to get the biggest training set that we can possibly get and that involves doing experiments outside of Rudy to be able
To get that training set so the reason we have to do this is that effectively to get our temporal resolution um this is it’s not difficult to to understand electrons are charged particles they do not like to be squeezed together in space and time and they push apart what those electron electron
Interactions do is effectively diffuse the the momentum transfer that comes from the specimen so the transfer that comes from the specimen that goes through the lenses onto a detector is how we get contrast in a microscope when all the electrons are bouncing off each other that’s diffusing that so that’s
Blurring our image in space and time so we want to reduce that and to get fewer electrons so it doesn’t overcome the problem that we have in stem where we’re going to have beam damage and therefore we have to use fewer electrons what it means is it’s the same problem in stem
As it is in high high temporal resolution we need to develop methods where we’re looking at low beam currents so just an example of what a low beam current does in terms of the resolution here is just an image a very low resolution image acquired from a TM this
Is what happens when you have about 10 to the nine electrons in a pulse beam and this is what happens when you have 10 to the six electrons if effectively if you don’t do anything to interpret the image or make the image better you’re going to be dealing with not
Enough signal to get the interpretation that you want all right so that means that we have to go through various forms of artificial intelligence I’m just going to show one because that will help to to understand the types of interactions that we want to try and spur over the next few years
So a typical way of doing an image the one that every physical scientist is aware of is this one over here which is that you have an object you have an image and you sample every position on the object to make up the image and that’s a pretty standard way of
Doing any kind of analysis the bit that the mathematicians do which makes it difficult for physical scientists to follow is to say that if you have something on your sample which has redundant information you do not have to sample your sample with as much detail
And you can get an image which is a lower amount of information and then you can use that to reconstruct your full image what this effectively is doing is using a Bayesian method throwing away bits of information that cause damage to your sample with electron microscope to then
Have a better set but a smaller set of information that you can then reconstruct this is basically like jpeg compression in in Reverse right we all know that when you take a big image on your camera and you try and put it on the Internet
It’s too much data and so it won’t go so what you do is you use a JPEG compression what that compression does is it shrinks down the amount of information in that image and then allows it to be transferred which then can be reconstructed into a full image
Now most the time we have to over compress to the point where you lose information but you can compress the point where you can reconstruct perfectly the information and still make the whole thing smaller what this approach in electron microscopy will do is to say Well if you
Know you have to compress the image why don’t you just acquire the compressed image to start with makes all the data transfer easier makes it faster makes it less damage and then you decompress it when you want to actually see what’s going on and this actually mathematically can be
I’m not going to talk about the math um can do do that pretty easily we have a new microscope coming to the University of Liverpool to be able to do a lot of the things that you can do on Rudy but doing it with high spatial
Resolution so this is a Joel 300 KV cryo stem so it has all the optical excitation it has the compressor sensing which I’ll talk about a second has all of the eels has everything associated with it um microscopes at the University of Liverpool are not on a recharge basis so
They are free for any academic to come and use the reason for that is we want to generate the data for the compressive sensing so if you have any experiment where you want to do high resolution microscopy um by all means just get in touch with
Me and you can get time on our microscope if you already have a microscope facility and you want to try some of these experiments we’re more than happy to show how to do the experiments and then get the data so we can do some of the analysis anyway so compressive sentencing what
Does it look like so just two images here easy to recognize uh one is the famous Barbara one is just the color image of a castle um essentially you can acquire the images that you see here very noisy images basically throwing away a lot of the pixels and you can reconstruct a
Full image without any problem if you then actually go to a microscope you can do the same thing so this is an atomic resolution image being acquired actually on one of the microscopes at the RFI so across the street you can see it’s a much poorer quality image that’s
Coming out because it’s compressed but you can easily reconstruct the atomic resolution image you can also deal with Focus changes you can deal with shifting you can go through that entire process and so that allows us to basically correct for a lot of errors in electron microscopy this case is not particularly
Low sampled it’s only 20 of the data so it’s about a reduction of a factor of five in terms of the total dose allows you to get the Reconstruction to go through another advantage of this is the dictionary that you create you can take from simulations and so you can run
Atomic resolution simulations use that to interpret your experimental data at that lower sampling rate and that will actually fix artifacts in their image so this is is using an atomic resolution simulation of silicon to fix a slightly tilted image to get Atomic resolution so all of these processes allow us to
Get higher resolution out of both high resolution stem images and our Dynamic images but to get those to work the best that we can we need to have a library of as many data sources as we can to get that to work and that’s the reason why
The ecosystem and the Hub and spoke method that we want to do for Rudy is already in process it’s already working we already want to get people to try and do experiments using the microscopes or send us the results from their experiments so that we can use the dictionaries to understand how to
Improve our experimental methods so in conclusion basically Rudy will be completely unique in the world in terms of the spatial resolution the temporal resolution um we think we could probably get up to 20 or 30 atmospheres of pressure in our cells if we wanted to uh to do the experiments with that
Um the first experiment is probably 2028 um fingers crossed if we’re lucky and we get funded um but we’re starting to prepare for all of these experiments now because even if Rudy isn’t funded a lot of this work will still go into Dynamics and operando experiments and then we’re open to all
Forms of UK and international collaboration final slide other than the Rudy Project these are all the people that that basically are involved with the work I talked about essentially the atomic resolution stuff was done with Bruce Gates group at UC Davis and then all the Institute Dynamics was Johannes
Lucas group at Munich and with that I will thank you very much for your attention and answer any questions thank you well mind using the question it’s just a very general comment in idle and first thank you for a really exciting election in this facility I think will be very
Very important for the catalysis community and I think we should the Hulk should engage with you fantastic as soon as possible and maybe organize a workshop to discuss how we can interact and contribute to the project yeah absolutely that would be great um we are told we will have an answer from the
Infrastructure fund by the end of this year just after that would be great times to start yeah thanks just Echo there’s absolutely wonderful as hope it’s a positive outcome obviously we can try and pick up a piece that isn’t that positive I’ve got a specific question I suppose
Because you mentioned you can stick a cell in this and when you’re doing your extreme editions can you then put some other things in I mean obviously cell needs to be in a liquid environment um I’d be very interested in looking at its cell death things that you know yeah
It’s possible to put some gases in there yes so basically the the suite right so if you if you have a high resolution like the 300 KV microscope I showed you can you can get gas cell liquid cells the gas cells are you can connect up to a
Mass spectrometer you can look at reactions and all that kind of stuff we will have the same thing and it’s just easier because instead of having to fit into Gap this size you have a gap this size so that’s why we can get all of the
Different bits in so instead of being an individual cell for one thing you can actually do multiple things within the same system so all of that is within the design parameters to do and obviously within that 130 million is money to build a whole Suite of in-situ stages
And once it’s starting then then you know if it’s funded then the collaboration to figure out what we should build with those stages is going to be a great thing to have because I think there’s a lot of interest in sort of membrane damage and you know sort of
Trying to show how that happens in the situated yep we should be able to do that perfect thank you very much for the talk I I was very um intrigued to see that you also are planning to have a lead or lean facility there and now when you that probably
Means you you need to slow down the electrons and in front of the sample can you still retain the the high time resolution and spatial resolution there yes and no so so um uh Klaus ropes um has done a lot of work on a an ultra fast lean an ultra fast lead
Um what happens with that is that the goal that we want for Rudy is to be able to do single shot type experiments where you have enough electrons in the system to be able to get all the information from a single shot a lot of the stuff
For the ulim and the U lead will be more of a stroboscopic type experiment where you can still get the temporal resolution in a single electron but you have to do more pump probe type experiments where they’re highly reversible type interactions to be able to look at how surfaces change in their
Structure where you’re not really trying to get massive Atomic rearrangements it’s more of interactions in that surface so we can get the temporal resolution but it’s more on a reversible type interaction more than an irreversible interaction now in terms of a lot of the diffusion effects though
Those are going to be nanoseconds and those they should be enough electrons to get those um nobody’s really tried it though at this point and it’s not clear whether you can get the combined spatial and temporal resolution of nanoseconds and to be able to see the surface with subnam resolution but what’s the
Um what’s the advantage of combining that with a with a mega electron uh Source uh it wouldn’t it make more sense to basically have two instruments for that or although they won’t they won’t they will be two instruments they’ll be completely separate all right okay so so the idea is that
Um effectively the partners Swansea leads Liverpool RFI would have an instrument which is an entry instrument into the system where if you were more interested on Surface interactions you would enter through a lead or you lean type experiment and then figure out how to go into Rudy from that if you were a
Microscopist you would come through Liverpool or RFI RFI for the biology Liverpool for the Material Science and then into designing the experiment because um you know a lot of microscopy has moved away from Mev beams over the last 30 40 years and there is no experience of working with these things anymore and
So effectively it’s not clear how to make the best samples that would go into that system and so it’s all part of a learning process of iterating towards better samples for Rudy if you’re going to have just one more from Josie one question yeah so I have a an online
Question someone can say it starts I thank you for a very nice talk and then says with regards to this simulated library to collect correct tilt in experimental data what happens if you have a defect in the 2D material will the defect be corrected by the perfect
Library uh that that is a good question it’s the question that we’re always asked with the compressive sensing as well because you know if you’re sampling and you had a small defect you could miss it with the sub sampling uh now it’s reasonably straightforward to look
At the sampling that you need for the resolution and set that correctly for the microscope and then it’s also a way to set how much you want to correct so obviously in everything that we’ve done we’ve never corrected a defect um but we’ve never tried to take something that’s very low magnification
And make it into something which has very high resolution so there is like most techniques where you have some kind of manipulation to correct for something you can overdo the minimum bit and then you get an oscillation where you know you could pick something here and it’s
The right one but the oscillation is the problem so as long as we stick within the resolution limit of the microscope then that oscillation and that correction of a defects shouldn’t happen thank you um thank you very much I think in the interest of time we should be involved