Periodic Reporting for period 2 - MHz-TOMOSCOPY (MHz rate mulTiple prOjection X-ray MicrOSCOPY)
Periodo di rendicontazione: 2023-06-01 al 2024-05-31
Modern enabling technologies, such as additive manufacturing or cavitation peening used in the aerospace and automotive industries, suffer from a lack of diagnostic tools. To date, one cannot provide relevant volumetric information about the fast processes involved. The realization of this project will break the current limits in fast, 4D X-ray microscopy by three orders of magnitude. It will be possible to visualize and characterize dynamics reaching velocities up to ~km/s for the first time with micron-scale resolutions. Instead of sample rotation, we will generate multiple X-ray probes and virtually rotate them around the sample to obtain with a single exposure multiple angular views simultaneously. Using modern X-ray sources with very high brilliance, each such 3D frame may be sampled at kHz rates at synchrotrons and even MHz rates at X-ray free-electron laser sources. This will unlock access to 4D observation of processes with velocities never before possible. 4D imaging of opaque samples at MHz rates enables insights across a range of sectors and industries.
In cavitation peening, an industrially relevant phenomenon for aerospace and new materials, we have no volumetric information, due to its high speed. This breakthrough will be achieved by the construction of a prototype that will demonstrate MHz rate tomoscopy at the European XFEL, taking advantage of world-unique European laboratories for the benefit of industry. Observing MHz-fast phenomena in opaque samples enables an entirely new branch of research, with possibilities for all sectors where such fast phenomena has, to date, been left to simulations and speculations. For industry and society, it would open new possibilities in the development and management of several techniques, including laser driven additive manufacturing, shock waves, fractures, evaporation, light alloy metallurgy, fast fluid dynamics, and cavitation phenomena.
In cavitation peening, an industrially relevant phenomenon for aerospace and new materials, we have no volumetric information, due to its high speed. This breakthrough will be achieved by the construction of a prototype that will demonstrate MHz rate tomoscopy at the European XFEL, taking advantage of world-unique European laboratories for the benefit of industry. Observing MHz-fast phenomena in opaque samples enables an entirely new branch of research, with possibilities for all sectors where such fast phenomena has, to date, been left to simulations and speculations. For industry and society, it would open new possibilities in the development and management of several techniques, including laser driven additive manufacturing, shock waves, fractures, evaporation, light alloy metallurgy, fast fluid dynamics, and cavitation phenomena.
The prototype of MHz Multi-projection X-ray imaging has been successfully installed. Initial commissioning was performed in April at the SPB/SFX instrument at EuXFEL. Follow up commissioning with fully functional setup will be performed in Autumn 2024. Data analysis is in the progress and the results are suggesting the feasibility of the method with minimum 4 projections feasible. Consortial beamtime proposal for the measurement has been submitted to European XFEL. Setup was placed on granite table, which was installed in April at in April. The prototype consist of motorized xyz,tip/tilt,theta stages for crystal splitters mounted on the rail system at the granite table. Total 6 stages has been designed and fabricated by SUNA. Detector system consists of 2 angular segments on left and right side of the granite table. Each segment can hold up to 3 MHz detectors in total 6 would be available. However due to large cost of one detection arm ~300k we have been able to realize only 4 detectors. Each detector consists of diffraction limited microscope coupled to scintillator via 90deg visible mirror, high NA visible objectives projecting image on fast MHz FT-CMOS cameras. This work was submitted as a deliverable 1.3 We have purchased 2 such cameras via this project and other two has been provided by EuXFEL as in-kind contribution for the project duration. All mechatronics part is controlled by control system designed and assembled by SUNA precision. Preliminary test of this setup shows very good mechanical stability especially for the crystal spliter positioners. Furthermore the setup contains motorized positioner for the sample environments. During Aprill 2024 commissing we have mounted the cavitation setup which was developed through WP3 with successful data acquisition. As for the crystal optics we have available as a backup plan 4 Diamond HPHT crystals as in-kind contribution from DESY and EuXFEL. The diamond crystals for the project have been specified and purchased with delivery awaiting around September 2024. Another development is the collaboration with INFN for the Silicon crystal optics. We have tested various thin membranes: with frame then without frame. Experimental data shows that Si splitters with frames solved the issue of mechanical stability however thickness is still the issue. So far we tested Si spitters with down to 30um however they suffer from heating which is deforming image already after first frame. Analyzing experimental data with aim to predict conditions: thickness photon energy under which Si splitters can work during MHz train illumination. We are confident that method is feasible as 400 diamond reflection provided enough intensity which assures that lower reflections 111, 113, 220 will provide even more signal as they have higher bandwidth.
.
We successfully improved the simulation framework and the fast data interface. The fast data interface was implemented in several increments and delivered (not yet published) to a Git-Lab repository. The developed framework is a computer program called FATRA (FAst Train Review Application) which was installed at SPB SFX beamline and is ready to be used by not just MHz-TOMOSCOPY apparaturs but for purpose of any experiment where fast 2D cameras. We have successfully developed a deep-learning framework, named 4D-ONIX, for 4D (3D+time) reconstruction from sparse projections. This method was validated using simulated data of binary droplet collisions due to the need for ground truth and the noisy, scarce data from the commissioning experiment. The results demonstrated the first MHz 4D reconstruction at XFELs and are under review in Nature Communications (with received positive review), with the algorithm described in an article available on ArXiV. Efforts are ongoing to make the algorithm more memory-efficient and faster, explore novel constraints to reduce the number of required experiments, and prepare simulations for cavitation peening studies in collaboration with WP3. The Dynamical tools for 4D analysis (wp2.4) is in progress.
The portable experimental setup and the sample environment were developed based on the technical parameters provided in Task 1.1. The experimental setup was equipped with accurate sensors, including pressure transmitters, pressure gauges, a mass flowmeter, a control valve, and acoustic sensors. Furthermore, the sample environment featured several venturi tubes manufactured using a state-of-the-art 3D printer and clear resin. After integrating the sample environment into the setup, we characterized the cavitation properties of different Venturi tubes using direct imaging techniques. This characterization aimed to select the optimal Venturi tubes and the corresponding operating conditions for further investigations using the MHz_TOMOSCOPY technique. This sample environment is now ready to be used for cavitation peening experiments in future beamtime.
The overall status and perspective of is with positive outlook to fullfill all project objectives.
.
We successfully improved the simulation framework and the fast data interface. The fast data interface was implemented in several increments and delivered (not yet published) to a Git-Lab repository. The developed framework is a computer program called FATRA (FAst Train Review Application) which was installed at SPB SFX beamline and is ready to be used by not just MHz-TOMOSCOPY apparaturs but for purpose of any experiment where fast 2D cameras. We have successfully developed a deep-learning framework, named 4D-ONIX, for 4D (3D+time) reconstruction from sparse projections. This method was validated using simulated data of binary droplet collisions due to the need for ground truth and the noisy, scarce data from the commissioning experiment. The results demonstrated the first MHz 4D reconstruction at XFELs and are under review in Nature Communications (with received positive review), with the algorithm described in an article available on ArXiV. Efforts are ongoing to make the algorithm more memory-efficient and faster, explore novel constraints to reduce the number of required experiments, and prepare simulations for cavitation peening studies in collaboration with WP3. The Dynamical tools for 4D analysis (wp2.4) is in progress.
The portable experimental setup and the sample environment were developed based on the technical parameters provided in Task 1.1. The experimental setup was equipped with accurate sensors, including pressure transmitters, pressure gauges, a mass flowmeter, a control valve, and acoustic sensors. Furthermore, the sample environment featured several venturi tubes manufactured using a state-of-the-art 3D printer and clear resin. After integrating the sample environment into the setup, we characterized the cavitation properties of different Venturi tubes using direct imaging techniques. This characterization aimed to select the optimal Venturi tubes and the corresponding operating conditions for further investigations using the MHz_TOMOSCOPY technique. This sample environment is now ready to be used for cavitation peening experiments in future beamtime.
The overall status and perspective of is with positive outlook to fullfill all project objectives.
We have demonstrated for the first time 1.125 megahertz X-ray multiprojection imaging of binary water dropet collisions with 4D recostruction of the structural dynamics. Only two X-ray projections has been achieved but so far this work demonstrate three orders faster 4D X-ray imaging then curent state of the art X-ray computed tomography. This work has been resubmitted to Nature Communication with positive review already received.