Periodic Reporting for period 2 - 3DX-FLASH (Probing MHz processes in 3D with X-ray microscopy)
Période du rapport: 2022-09-01 au 2024-02-29
I aim to develop an X-ray imaging technique capable of filming processes in 3D, with a temporal resolution several orders of magnitude faster than up-to-date 3D X-ray imaging techniques.
The unique penetration power of X-rays allows us to study systems in their native environment. This property has led to the development of X-ray microtomography (µCT). µCT acquires 3D information, which determines the functionality and mechanical properties of nature, by rotating a sample with respect to the X-ray source. µCT is a crucial tool for several scientific disciplines, such as physics, biology, and chemistry.
Over the last decade, µCT has become a technique capable of not only recording 3D information but also filming processes. Several breakthroughs have made this possible: i) intense X-ray sources (synchrotron light sources), ii) efficient and fast X-ray detectors, and iii) fast 3D reconstruction algorithms. Despite these developments, the acquisition protocols remain unchanged, i.e. the sample is only rotated faster. This fast rotation introduces forces that may alter the studied dynamics and ultimately limit the achievable temporal resolution.
My project is to establish an X-ray microscope that avoids the sample rotation, obtaining 3D information from a single X-ray flash by splitting it into several angularly resolved beams that illuminate the sample simultaneously. When implemented at intense X-ray sources such as synchrotron light sources and X-ray free-electron lasers, this approach will allow the filming of natural processes with a micrometer to nanometer resolution and resolve dynamics from microseconds to femtoseconds. To demonstrate its capabilities, I will study fundamental processes in cellulose fibers, a renewable biomaterial that can replace fossil-based materials, such as plastics. This technique will open up the possibility to film dynamics in 3D to answer questions coming from industry and natural sciences at rates not accessible today.
The unique penetration power of X-rays allows us to study systems in their native environment. This property has led to the development of X-ray microtomography (µCT). µCT acquires 3D information, which determines the functionality and mechanical properties of nature, by rotating a sample with respect to the X-ray source. µCT is a crucial tool for several scientific disciplines, such as physics, biology, and chemistry.
Over the last decade, µCT has become a technique capable of not only recording 3D information but also filming processes. Several breakthroughs have made this possible: i) intense X-ray sources (synchrotron light sources), ii) efficient and fast X-ray detectors, and iii) fast 3D reconstruction algorithms. Despite these developments, the acquisition protocols remain unchanged, i.e. the sample is only rotated faster. This fast rotation introduces forces that may alter the studied dynamics and ultimately limit the achievable temporal resolution.
My project is to establish an X-ray microscope that avoids the sample rotation, obtaining 3D information from a single X-ray flash by splitting it into several angularly resolved beams that illuminate the sample simultaneously. When implemented at intense X-ray sources such as synchrotron light sources and X-ray free-electron lasers, this approach will allow the filming of natural processes with a micrometer to nanometer resolution and resolve dynamics from microseconds to femtoseconds. To demonstrate its capabilities, I will study fundamental processes in cellulose fibers, a renewable biomaterial that can replace fossil-based materials, such as plastics. This technique will open up the possibility to film dynamics in 3D to answer questions coming from industry and natural sciences at rates not accessible today.
Most of the work has been devoted to designing and fabricating an optimized X-ray multi-projection imaging setup to retrieve the highest spatiotemporal resolution beyond what is possible with state-of-the-art time-resolved approaches. We have succeeded in proposing and fabricating a novel setup. This setup has been commissioned at large-scale facilities: X-ray free-electron lasers and storage rings. We have recently demonstrated that X-ray multi-projection imaging, when combined with the unique capabilities of the European X-ray free-electron laser, can record 3D movies 1000 times faster than current approaches. Furthermore, it can also record rotation-free 3D movies faster than state-of-the-art approaches when implemented at storage rings. The latter will open new research opportunities that were hindered by the shear forces introduced with conventional time-resolved imaging techniques.
The reconstruction subproject has also successfully delivered a self-supervised deep-learning algorithm that can generate 3D reconstructions with eight or fewer projections. This will open new opportunities in X-ray imaging and has the potential to be applied to fields like medicine to reduce the dose and acquisition time of current 3D approaches. Moreover, we have extended this algorithm to reconstruct not only 3D but also 4D (3D+time), opening the possibility to better constrain the physical processes of interest.
The reconstruction subproject has also successfully delivered a self-supervised deep-learning algorithm that can generate 3D reconstructions with eight or fewer projections. This will open new opportunities in X-ray imaging and has the potential to be applied to fields like medicine to reduce the dose and acquisition time of current 3D approaches. Moreover, we have extended this algorithm to reconstruct not only 3D but also 4D (3D+time), opening the possibility to better constrain the physical processes of interest.
- We have developed a new splitting scheme that enhances the spatiotemporal resolution of X-ray multi-projection imaging. This concept has led to a patent application, and it is the one to be implemented in this project. We expect to be able to have acquisition rates ten times faster than the initially proposed setup up to the MHz rate at the European XFEL.
- We expect to have algorithms to reconstruct physical parameters from 4D reconstructions. Compared to current approaches that rely on tomography and lengthy reconstruction pipelines, our approach can potentially deliver the physical parameters of interest directly.
- We expect to study not only cellulose fibers but also other scientific problems concerning sustainable development. For example, we plan to study failure and kinking mechanisms in composite materials for sustainable transport and additive manufacturing for cost-effective manufacturing.
- We expect to have algorithms to reconstruct physical parameters from 4D reconstructions. Compared to current approaches that rely on tomography and lengthy reconstruction pipelines, our approach can potentially deliver the physical parameters of interest directly.
- We expect to study not only cellulose fibers but also other scientific problems concerning sustainable development. For example, we plan to study failure and kinking mechanisms in composite materials for sustainable transport and additive manufacturing for cost-effective manufacturing.