Time-resolved in situ tomography for the analysis of evolving metal-foam granulates
An experimental setup has been developed that allows for capturing up to 25 tomograms s−1 using the white X-ray beam at the experimental station EDDI of BESSY II, Berlin, Germany. The key points are the use of a newly developed, precise and fast rotation stage, a very efficient scintillator and a fast CMOS camera. As a first application, the foaming of aluminium alloy granules at 923 K was investigated in situ. Formation and growth of bubbles in the liquid material were observed and found to be influenced by the limited thermal conductivity in the bulk granules. Changes that took place between two tomographic frames separated in time by 39 ms could be detected and analysed quantitatively.
Dynamic processes in evolving metallic foams have been studied in situ using fast X-ray radioscopy (Garcia-Moreno et al., 2008, 2012; Rack et al., 2009). Although acquisition rates over 100000 frames s−1 have been reached (Garcia-Moreno et al., 2012), and valuable insights into the dynamics of foaming and the corresponding morphological changes in foams have been gained, radiographic information is often ambiguous and limited to flat sample configurations. In situ tomography lifts some of the restrictions but is too slow for real-time studies.
Recent developments have led to higher temporal resolutions through improved instrument hardware and data processing capabilities. Acquisition times are also becoming shorter because of the use of high-brilliance third-generation synchrotron sources, highly efficient scintillators and improved optics. As a consequence, fast tomography with up to 20 tomograms s−1 (20 Hz) has become available at different beamlines (Rack et al., 2010; Momose et al., 2011; Takano et al., 2013; Mokso et al., 2015; Maire et al., 2016; Kamm et al., 2017; dos Santos Rolo et al., 2014; Lovric et al., 2016). Evolving metallic foams (García-Moreno et al., 2013; Jiménez et al., 2018; Kamm et al., 2017) can now be studied in real time with in situ tomography. Furthermore, operando studies such as cycling of rechargeable batteries (Schröder et al., 2016) or operating fuel cells (Alrwashdeh et al., 2017) are currently the focus of research.
In this work, we present the latest development in imaging acquisition speed at the EDDI beamline at BESSY II, Berlin, Germany, where we are now able to perform time-resolved in situ tomography up to 25 Hz, which is the fastest tomography acquisition rate reported at synchrotrons so far. As a case study, we analyse the foaming behaviour of a metallic foam granulate. Similar to polymeric foams, the use of expandable granulates is interesting from an application point of view. As the surface-to-volume ratio of granules compared with standard bulk precursors is much larger, their foaming behaviour might suffer from insufficient gas nucleation or increased gas losses, incomplete particle bonding, and result in non-uniform pore sizes and foam densities (Duarte et al., 2013; Nosko et al., 2010). Therefore we study the foaming behaviour of single granulates and the evolution of the time-resolved bubble size.
A new high-speed rotation stage (see Fig. 1) jointly developed at the Helmholtz–Zentrum, Berlin, and the Technische Universität Berlin is used. Here, we explore fast phenomena in evolving metal foams with a rotation frequency of ∼12.5 Hz, which allows us to capture tomograms at ∼25 Hz, each of which requires a rotation by 180°. The stage features a precision rotation axis with ball bearings and a servo drive regulated by an angular PID speed controller with sensor feedback from the servo drive, programmable by self-developed Labview-based software. No electric sliding contacts for the rotation system are available, which is why heating and temperature measurements are conducted contactless. Special attention has been paid to centering samples and crucibles precisely as no xy table was used, and centrifugal forces reach 0.6g at the inner side of the crucible of radius ri ≃ 1 mm.
AlSi10 + 0.5TiH2 (all in wt%) foamable precursor granulate (equivalent diameter ≃ 1 mm) is produced following the powder metallurgical route described elsewhere (Banhart, 2006). The granules are placed in an X-ray-transparent boron nitride (BN) crucible of 2.2 mm inner and 3 mm outer diameter. Foaming is induced by heating the samples in the crucible with a ramp of ∼2 K s−1 to around 923 K using a 150 W IR lamp (Osram, Germany) from the top (see Fig. 1). Foaming starts during the heating phase at around 853 K, close to the eutectic temperature of the Al–Si alloy (850 K) when the system is not under isothermal conditions.
A white X-ray beam (about 6–120 keV) is provided by a superconducting 7 T multipole wiggler. A 200 µm-thick LuAG:Ce scintillator is used to convert the transmitted images into visible light, after which a mirror projects the light onto a PCO Dimax CMOS camera. A more detailed explanation of the imaging system can be found elsewhere (Jiménez et al., 2018; García-Moreno et al., 2013).
A compromise between pixel size, field of view (FOV), number of projections and rotation speed is necessary depending on the experiment type and requirements. In the present case, the acquisition parameters are: a spatial resolution of 5 µm measured on a micro-resolution test pattern from JIMA, Japan, an FOV of 3.44 mm × 1.85 mm (given by the pixel size of 2.5 µm), an exposure time of 0.48 ms for each single projection, 78 projections per tomogram (sufficient for our purpose), and a time resolution of ∼39 ms per tomogram. The Common Unified Device Architecture (CUDA)-accelerated filtered back-projection (FBP) algorithm-based reconstruction is performed using the open source ASTRA Toolbox (van Aarle et al., 2016). Tomograms at 25.6 Hz are acquired in 15.5 s.
A total of 396 tomograms were acquired in ∼15.5 s in order to follow the stage of gas nucleation and subsequent bubble evolution in AlSi10 + 0.5TiH2 granules as they were heated. The high rate of ∼25 Hz allows us to resolve, in situ and in detail, the bubble evolution inside each single granule. Fig. 2(c) shows six xy-slices at a particular height (z = 1.4 mm) extracted from the series of 396 tomograms. Nucleation probably starts around t ≃ 0 s in the granule (see lower right image), but in fact we only observe bubble growth in the early stages because nucleation itself is not observable due to insufficient spatial resolution. Pores of 1–2 µm diameter contribute most to the total bubble volume in the nucleation stage of Zn foams as measured by neutron scattering but pores as small as 50 nm could also be detected (Banhart et al., 2001). We assume that each appearing bubble corresponds to a single nucleus (Kamm et al., 2017). At t ≃ 2.5 s and 5 s, the second (lower left) and third (middle) granules follow as manifested by the appearance of the first bubbles, respectively. The different onset times are understandable because the heat applied is transmitted into the grains from the direction of the heating lamp and through the crucible walls, and there is a small radial and a larger axial temperature gradient inside the crucible. Moreover, intergranular heat transfer is low because of the small contact area between granules, which leads to a slight temperature delay in some particles depending on their location. This observation underlines a difference to bulk samples and can explain the varying pore sizes and densities of solidified foams made from granules reported in the literature (Nosko et al., 2010). On the other hand, intragranular nucleation takes place almost simultaneously as a result of good heat transfer within individual granules similar to bulk samples (Kamm et al., 2017).
The tomogram at t = 7.695 s shown in Fig. 2(a) as a three-dimensional rendering features several granules in transparent grey and a selected grain with individual bubbles internally highlighted by different colours, obtained by binarization and segmentation of the image. It also shows the same single granule again four more times within time intervals of 39 ms (i.e. at t = 7.734 s, 7.773 s, 7.812 s and 7.851 s), thus exploiting the maximal temporal resolution of our experiment. This visual comparison together with the corresponding analysis of the mean equivalent diameter of all bubbles (Fig. 2b) reveals the changes in bubble size and distribution within these very small time intervals. This result underlines the highly dynamic nature of foaming, although the difference in this short period is smaller than the standard error of the mean diameters. The number of bubbles remains nearly constant at 1298 ± 24, between t = 7.695 s and t = 7.851 s, suggesting that already at this early foaming stage there is no more nucleation or that there is a balance between newly appearing bubbles and those that already start to merge or disappear. The foam bubble evolution at a later stage (t = 12.558 s and t = 12.597 s) is depicted in Fig. 2(d), where the difference image shows in red the changes between the two slices (e.g. shrinkage of the central bubble and rupture of a smaller one), thus emphasizing the need of such high spatio-temporal resolution.
Time-resolved tomography up to 25.6 Hz is now available as a powerful characterization tool for in situ analyses in material science.
Granules of AlSi10 alloy + TiH2 of ∼1 mm equivalent diameter have been foamed and their foaming sequence is found to be dependent on their position. This can be explained by an inferior intergranular thermal contact, although intragranular bubble size evolution is found to be comparable with that of bulk precursors.
The distribution of equivalent bubble-size diameters and their mean values evolve even within the short time between two tomograms (39 ms), thus emphasizing the need for high temporal resolution.
Funding by the European Space Agency (project AO-99-075) and German Research Foundation (project BA 1170/35-1 and GA 1304/5-1) is gratefully acknowledged.
Aarle, W. van, Palenstijn, W. J., Cant, J., Janssens, E., Bleichrodt, F., Dabravolski, A., De Beenhouwer, J., Joost Batenburg, K. & Sijbers, J. (2016). Opt. Express, 24, 25129–25147. Web of Science PubMed Google Scholar
Alrwashdeh, S. S., Manke, I., Markötter, H., Haußmann, J., Arlt, T., Hilger, A., Al-Falahat, A. M., Klages, M., Scholta, J. & Banhart, J. (2017). Energ. Technol. 5, 1612–1618. Web of Science CrossRef Google Scholar
Banhart, J. (2006). Adv. Eng. Mater. 8, 781–794. Web of Science CrossRef Google Scholar
Banhart, J., Bellmann, D. & Clemens, H. (2001). Acta Mater. 49, 3409–3420. Web of Science CrossRef Google Scholar
Duarte, I., Oliveira, M., Garcia-Moreno, F., Mukherjee, M. & Banhart, J. (2013). Colloids Surf. A Physicochem. Eng. Asp. 438, 47–55. Web of Science CrossRef Google Scholar
García-Moreno, F., Jiménez, C., Kamm, P. H., Klaus, M., Wagener, G., Banhart, J. & Genzel, C. (2013). J. Synchrotron Rad. 20, 809–810. Web of Science CrossRef IUCr Journals Google Scholar
Garcia-Moreno, F., Mukherjee, M., Jiménez, C., Rack, A. & Banhart, J. (2012). Metals, 2, 10–21. Google Scholar
Garcia-Moreno, F., Rack, A., Helfen, L., Baumbach, T., Zabler, S., Babcsan, N., Banhart, J., Martin, T., Ponchut, C. & Di Michiel, M. (2008). Appl. Phys. Lett. 92, 3. Google Scholar
Jiménez, C., Kamm, P. H., Plaepow, M., Neu, T., Klaus, M., Wagner, G., Banhart, J., Genzel, C. & García-Moreno, F. (2018). J. Synchrotron Rad. Accepted. Google Scholar
Kamm, P. H., García-Moreno, F., Neu, T. R., Heim, K., Mokso, R. & Banhart, J. (2017). Adv. Eng. Mater. 19, 1600550. Web of Science CrossRef Google Scholar
Lovric, G., Mokso, R., Schlepütz, C. M. & Stampanoni, M. (2016). Phys. Med. 32, 1771–1778. Web of Science CrossRef Google Scholar
Maire, E., Le Bourlot, C., Adrien, J., Mortensen, A. & Mokso, R. (2016). Int. J. Fract. 200, 3–12. Web of Science CrossRef CAS Google Scholar
Mokso, R., Schwyn, D. A., Walker, S. M., Doube, M., Wicklein, M., Müller, T., Stampanoni, M., Taylor, G. K. & Krapp, H. G. (2015). Sci. Rep. 5, 8727. Web of Science CrossRef PubMed Google Scholar
Momose, A., Yashiro, W., Harasse, S. & Kuwabara, H. (2011). Opt. Express, 19, 8423–8432. Web of Science CrossRef CAS PubMed Google Scholar
Nosko, M., Simančík, F. & Florek, R. (2010). Mater. Sci. Eng. A, 527, 5900–5908. Web of Science CrossRef Google Scholar
Rack, A., García-Moreno, F., Baumbach, T. & Banhart, J. (2009). J. Synchrotron Rad. 16, 432–434. Web of Science CrossRef CAS IUCr Journals Google Scholar
Rack, A., Garcia-Moreno, F., Schmitt, C., Betz, O., Cecilia, A., Ershov, A., Rack, T., Banhart, J. & Zabler, S. (2010). J. Xray Sci. Technol. 18, 429–441. Google Scholar
Santos Rolo, T. dos, Ershov, A., van de Kamp, T. & Baumbach, T. (2014). Proc. Natl Acad. Sci. 111, 3921–3926. Web of Science PubMed Google Scholar
Schröder, D., Bender, C., Arlt, T., Osenberg, M., Hilger, A., Risse, S., Ballauff, M., Manke, I. & Janek, J. (2016). J. Phys. D Appl. Phys. 49, 404001. Google Scholar
Takano, H., Morikawa, M., Konishi, S., Azuma, H., Shimomura, S., Tsusaka, Y., Nakano, S., Kosaka, N., Yamamoto, K. & Kagoshima, Y. (2013). J. Phys. Conf. Ser. 463, 012025. CrossRef Google Scholar
© International Union of Crystallography. Prior permission is not required to reproduce short quotations, tables and figures from this article, provided the original authors and source are cited. For more information, click here.