 Hi, my name is Gregor, I'm a PhD student in chemical engineering at the department of chemical engineering, and we're working with membrane processes. So separation processes that can be used for every different separation process and also food processing. And then we had this one problem and we thought we have to search out for collaborators and found the money. So now we will present you what we've done so far. I would say, so should we start with the presentation. Go ahead. Go ahead. Good idea. Okay, so yes, today, we're going to present you our research on correlative x-ray tomographic imaging of membranes for improving food processing, so far away from the stomach that we just hear before. Before we start with our research and our results, the outline of this talk, what are actually membrane processes. Then we give a little introduction to correlative x-ray tomographic imaging and then we will finish with a take home message and the presentation has split up between me and Emmanuel. So why do you actually use membrane processes for food processing? Well, there's a lot of advantages that comes with membrane processes. They have a gentle product treatment because they can run at moderate temperatures and changes during the processing. They are highly selective and very unique separation mechanisms. They have a compact modular design, which means that it's easy to install and also to extend and they consume low energy and compared to other separation processes such as condensers and evaporators and it's easily to integrate it with other technologies. But of course, membrane processes also have downsides. There's a short membrane life cycle and it's a high cost for membranes usually. The products that have to be treated should be liquid or gaseous or vapors. And the membrane performance is often limited by the temperature, pH or the chemical resistance of the membrane material. And then they can be easily extended but they also scale only linearly so there's a limited economy of scale. And the topic that we want to talk about today is the tendency of membranes of membrane fouling and then the need for membrane cleaning to overcome this fouling. And here on the right side you can see a fouled membrane and a cleaned membrane. So what is actually membrane fouling? Well, this is the process resulting in loss of performance of a membrane due to the deposition of suspended or dissolved substances on its external surface at its pore opening or within its pores. So you can see that there's quite some difference between a clean and a fouled membrane. And why are we going to talk about this? Because fouling and cleaning is costly. So when the membrane is fouled and then has to be cleaned, there's production time losses and then also irreversible fouling and cleaning reduces the membrane lifetime. So that also takes time. So overall the total daily plant capacity has to be increased. So that means it's a fine balance between the cleaning frequency, the production efficiency and the membrane lifetime. And that's of course leads to costs for membrane cleaning. It's five to 20% of the capital expenditures of the membrane plant. So very significant costs and the membrane replacement also is two to 5% of the capital expenses. So how do we actually measure membrane fouling? Typically, membrane fouling is only measured by looking at the flux if you run at constant pressure. That means at the beginning of a filtration, the membrane has a pure water flux that doesn't change. But then once we start filtration, the flux declines and we have this capacity reduction. And this can be a bit compensated by rinsing the membrane and washing away loosely attached compounds. There's an irreversible fouling that is removed. But then you can still see that there's still a difference between the flux at the end of rinsing and at the beginning of the filtration. And this is due to irreversible fouling. So fouling that can only be removed by cleaning if at all. The problem with this method is that there's no information about the fouling layer structure and no information about the membrane altering due to fouling or cleaning. So with this, we collaborated with Emmanuel and wanted to and other people, of course, and wanted to look at the inner structure of the membrane. So we did this with correlative x-ray tomography and I present you the scheme how we did it here in the following. So first, we found and cleaned membranes. Then we put them, we cut them in small pieces and we had a sample mounting for lab scale micro tomography. Then we did a phase retrieval and at the end an image analysis where you can see the local thickness, for example, of the membrane sample. In parallel, we also did fibbing to prepare samples, membrane samples for synchrotron scans. And that's nice because that would allow us a higher contrast to noise ratio, a higher resolution and a higher flux. So less dose on the sample, which means the sample survives longer and faster imaging. However, you have to write the beam time proposal and it's important to need to motivate the need for this high resolution, which is quite tedious. So in the following manner, I will explain to you what actually is tomography and how you can use it. Yes, thank you. So I'm Emmanuel Larson and as Milena pointed out before I'm hired at the Division of Solid Mechanics and Stephen Hall, who is also links director is my boss. And within this, and I'm also affiliated to the Lunar to Computing Center so we use clusters to do all the analysis at the university. We have collaborated with Gregor and many others as well within this Vinova PhD project that Vinova announced the first time in 2009 when it comes to increasing the capacity and skills of PhD students regarding neutron and synchron based methods. And I can also mention that Shunyu, former colleague at Rice, he has given lectures to Gregor and on small x-ray scattering, which is actually underlying technique if you think about x-ray typography. And also Pablo Villanova-Perez has also given lectures in high resolution queried imaging. And then me and Gregor have also worked a little bit further both with lectures and educational tool kits that I developed to, let's say, make sure that Gregor learned all the aspects of tomographic acquisition, image processing, image analysis and so on. And I can start with highlighting that I have built this tomograph, which I based on a normal flashlight which I called kitchen based light tomography that I have at home. And this one compares very nicely to a real x-ray scanner, for example at the Division of Solid Mechanics or even at a synchron neutron, sorry, synchrotron x-ray microtomography or neutron tomography experiment. And we have used this setup for training, educating Gregor and also other PhD students postdocs. And you can see that if you scan an object like this Lego man with x-rays, you can really penetrate and see the inside of the object. You cannot do that with a normal flashlight, of course. But the nice thing is that all the data that we acquire, they're quite in the same manner as when you have a real experiment and also the data format, the images you have, you can treat them exactly as if you would have a real x-ray or neutron tomography experiment. And we can perform reconstruction, pre-reconstruction using Tomopai. We can also carry out a lot of image processing and quantitative image analysis. And we can, based on the KBLT dataset, render these nice images where you can see that the local thickness varies in the sample. And basically by training on this KBLT setup, you can actually directly transfer these skills to real experiment that you later carry on. And I will just show here a short 15 second video of how the setup works, which also train Gregor train on. And you see here that the sample starts to rotate for each rotation angle, we acquire an image, and then we will acquire 200 projection image over 360 degrees. And at the end, we will obtain this, free the rendering of the reconstruction. And we can also perform image analysis on this sample to estimate the pore size and so on. And further on, to give some more introduction to, let's say, x-ray and x-ray imaging techniques compared to normal light. I will present something called face contrast. And you can think of this very easy example where we have a glass very close to a paper screen and we have a flashlight on the other side. If you put the paper screen further away, we will actually be more prone to detecting small, let's say low absorbing differences. In this case, this impurities on top of the glass. And that is because we go from the absorption mode and we go more into the face contrast regime. And in another example, which I acquired both inside our home and also with the help of a large flashlight outside the house. You can see that on our wall, we get this pattern of this scotch. And what we basically see here is actually the face contrast effect. You can basically see all the air bubbles inside the scotch. And another version of this, taking a bit further, is that here we actually see the shadow of the branches in this tree. And these branches are more diffuse. So my qualified guess is that we have actually entered more of the frown of the regime. And in order to reconstruct this one nicely, we would have to do something called face retrieval, which is mandatory for frown of the regime while it's optional for the face contrast regime. And if we switch over to imaging with real x-rays, as in the samples that we show, we have image with let's say x-ray microtomography with a typical resolution of 0.7 microns. So we can image both in absorption and face contrast and you get more prone to detecting small features with this technique. Then there are also techniques such as nanomax, the full field x-ray holographic nanotomography, which is a lensless technique and the projection image is that you acquire of the sample. And you can see that they are basically blurred, let's say. I mean, because, and that depends on the distance from the sample to the detector. And in order to reconstruct this sample, you need to do this face retrieval, and you need to acquire several, let's say, projection images at several samples to detect their distances. A typical resolution around 100 nanometers. And then there is also a technique called full field x-ray nanotomography, which uses lenses, and it can also be combined with face rings to have more contrast in the Seneca face contrast regime. And this technique doesn't require face retrieval, you get a high contrast directly. Ultimately, there is this technique called typotomography, where you actually probe. So your source is very small and you have to scan it over the sample and you acquire the SUCKS pattern that you then overlap. You change back to a freer space and eventually you can actually obtain these projection images that you then back project and you can reconstruct the 3D object of the sample. And for this project, we aim to use all of these imaging techniques to have this correlative approach on different field of use and resolutions of these membrane samples. And we carried out a lot of scans on the lab-based scanner, and I can also show you here really the benefit of applying this face retrieval, where you can see that you really increase the contrast to noise ratio. And we can clearly see the difference between different components inside the sample, because they appear as taking something invisible and making it visible. And by applying this face retrieval, it of course becomes much easier to segment the data, carry out with image processing, image analysis. So this is, let's say, work in progress. And we're also reusing some of the codes from the KIM initiative. And then Torbeninsson Pingel, who is also here today, he has also prepared, I think, four or five, let's say, fibbed samples that have placed on these so-called omni-pins. That is quite a standard way to put the samples on top of these pins, and then you can scan them, for example, at the CSAC beam line at the PSI in Switzerland or at the Nanomax. And I can also mention that we have applied for several beam times, but in corona times it was a little bit difficult to get beam time granted. But for the CSAC beam line, we are currently on the waiting list, so we're hoping that there will be brighter times ahead. And with that, I will hand over again to Gregor. Yes, thank you. So back to the membrane samples itself. Here you can see that we've already scanned the different types of membranes, here tubular membrane and here hollow fibre membrane, so such a small spaghetti type of membrane. And this sample was found with pleat point efferents from a pipe mill, and you can see, you can nicely see the surface at the nonwoven support layer of the membrane and here the active layer. And here you can really nicely see the support layer in the inside of this hollow fibre membrane and outside the active layer. And now that we wanted to look at membrane fouling and cleaning, here you can see microfiltration membranes on the bottom that have been fouled and cleaned and ultra filtration membranes on the top. And they have been fouled with thermal mechanical piping process water. And what Emmanuel just explained, with phase retrieved and non-phase retrieved, we applied this approach here and we nicely see that in the fouled ultra filtration membrane, we have a bright fouling layer, so they are on top. And this is probably due to lignin absorption. So this is what I was talking about. So, but we actually wanted to tell you something about food processing. So just have a little outlook on food processing. There's this project that we are a part in rapeseed processing rapeseed are nice because you can push, press them out and get the nice rapeseed oil, and then two thirds of this stays left over as a press cake and there's high protein content in this. And this protein content has a high has a nice amino acid composition, but a bitter flavor. So here membrane processes come in in the game to separate first in the microfiltration step particles and microbes and then in the ultra filtration step to up concentrated proteins. So we now looked at this microfiltration step from the filter press cake filtration. And here I present you lab scans again from micro tomography. And we've scanned two different membranes one more open with 200 nanometers porcise and one more close with 100 nanometers porcise. Again, you see the non-woven structure nicely and on the top very thick, the membrane active layer. And if you zoom in, you can see that here on top, there's a white line. And this is probably due to fouling. So we're really happy that we saw this when we zoomed in and the manual just this week run another experiment analysis with much longer exposure time and also scan time. And what we can see here. So the first time that we actually see the fouling layer on top of this membrane. So this you can compare with the same image, but super nicely. I was very happy to see that here, this fouling layer line which you can see here and here on the top. You can also see already in the 3D construction rendering here at the bottom. So this really gives us a nice insight into the structure of the fouling layer and separates the fouling layer also from the active member layer. With this, I'm finishing this talk with my take home or take home messages. Extratermographic imaging provides really novel insight into the impact of membrane fouling and cleaning on the membrane structure. And here you can see the scan that we've done and the 3D rendering of ultra filtration membrane. However, we need high resolution that is only available at synchrotron facilities. So this is still a challenge. And last but not least, the technique requires a wide network of experts in collaboration. This might be challenging at the beginning, but I found it really fruitful because you learn so much from everybody. With this, I would like to thank everybody that was involved in this project. And here you maybe see that I'm the only one that is still a mystery. So tomorrow I can invite you for my PhD defense if you want to learn more about membrane fouling and cleaning. With this, I would like to thank you and finish this talk and open for discussion. Thank you Gregor. I think that that was very interesting and good luck tomorrow. Thanks. I don't think we have a lot of time for question but I have a little bit of a pitch for you to make. It's a, you know, you're talking to the converters, we know that imaging x-ray and neutron are very interesting and we should do more of it. Okay. But I'm just going to ask you if, you know, I, I do this all the time I work with membranes, and I do measure fouling and I can look at chemically what the composition is on the fouling membrane. What is the added value of what you're doing with imaging? Well, it's not so easy to get out the all the information that you were just describing. So it's really challenging to get an idea about the chemical composition of the fouling layer. And even when you know this fouling can happen at very different levels on the membrane. So it can happen on top of the membrane. That's more be the most obvious. But also membrane compounds can enter the pores and absorb on the pore walls. And that's the most critical, the biggest problem I would say because then they accumulate and eventually the pores are blocked. And you want to know at what time this happens and what compounds are actually entering those pores in order to optimize the cleaning procedure or running your process. So maybe you want to increase your shear forces on the membrane surface to take away those compounds that are getting there, or you want to adjust and clean earlier with a different, a different cleaning agent. So, yeah, it's a really complex topic membrane fouling and to optimize cleaning, which is the overall goal because we want to have a process that runs nicely. We really have to understand membrane fouling from all different perspectives. And maybe that's just the last point, just 10 seconds of your time is that with extra tomography we can get a nice scan, let's say within 50 minutes and it's also in 3D. And I mean if you compare it to some it will require much more time and you need to, you know, cut the sample in a way so I mean, there are so many positive aspects I could speak about it for half an hour. That's great. I just wanted a pitch. Yeah, okay. Thank you. If you have more questions, just contact me or Emmanuel.