3D Structure of Carbon Soot
August 11, 2010 – Researchers at Ulm University have resolved individual graphitic layers inside carbon soot in three dimensions. Aberration-corrected, energy-filtered HRTEM at 80 kV with a strict dose budget and a new marker-free alignment reached a resolution of at least 340 pm in the tomogram, and the spacing, length and curvature of 3400 layers around catalytic platinum particles were measured.
Carbon soot is the black backbone of many fuel cells: it carries the platinum nanoparticles that do the catalysis and keeps them from clumping together during operation[1]. How its graphitic layers are arranged inside a soot particle, and how they wrap around the platinum, had never been seen in three dimensions. Electron tomography, which reconstructs an object from a series of tilted images, routinely reaches a resolution of about 1 nm3[2], too coarse for layers 340 pm apart, and the electron beam reshapes soot long before a high-resolution tilt series is complete. Jens Leschner, Johannes Biskupek, Andrey Chuvilin and Ute Kaiser at Ulm University have now combined an aberration-corrected, energy-filtered transmission electron microscope at 80 kV with a strict dose budget and a new alignment that needs no markers. In commercial Vulcan XC72 soot they resolved individual graphitic layers in 3D and measured the spacing, length and curvature of 3400 of them. An independent study from Jülich, made with the same type of microscope at the same voltage, reached a similar sub-nanometer resolution on MoS2 nanoparticles.
A fragile target
Atomic resolution in all three dimensions had been shown only for a simulated gold nanocrystal, using the prior knowledge that its atoms sit on a regular lattice[3], an approach that does not carry over to layered, disordered materials. Soot is such a material: unlike graphite, whose layers lie 340 pm apart in perfect order, it consists of curved layers of varying length and spacing[4],[5]. Platinum particles in carbon black had been mapped in 3D by tomography in the scanning transmission electron microscope (STEM) with respect to their shape, volume and distribution[6], but the inner structure of the support itself remained unknown.
The obstacle is radiation damage. Fast electrons knock carbon atoms out of place, heat the sample and etch it chemically[7],[8]; depending on how the carbon atoms are bonded, knock-on displacement sets in between 27 and 95 kV[9]. Revealing soot in its pristine state therefore requires a microscope run near or below that threshold, an electron dose kept as low as possible[10],[11], and an imaging mode that delivers high resolution while meeting the demands of tomography. The Ulm team chose 80 kV, the lowest voltage then commonly available on aberration-corrected microscopes.
How little soot can take was measured first. While graphene permits high-resolution exposures with a cumulative dose of 109 e−/nm2, soot needs conditions two orders of magnitude stricter. After 3.1 × 107 e−/nm2 a carbon layer moved by half the graphitic layer spacing, which sets the dose limit for a tilt series that is to resolve layers 340 pm apart; after 6.4 × 107 e−/nm2 a platinum particle had shifted by a third of its diameter (Fig. 1). The weakest spot of a soot particle, not the sturdiest layer, decides whether a tilt series succeeds.
Choosing the imaging mode
Tomography needs a signal that rises steadily with the thickness of the sample and the scattering power of its atoms, the so-called projection requirement, and enough signal per dose. In materials science the usual choice is high-angle annular dark-field (HAADF) STEM, whose Z-contrast meets the projection requirement for crystalline samples[12]. Aberration correction and brighter sources had brought STEM to atomic resolution[13], but at the cost of a much higher dose; a low-dose STEM roadmap had been reported, so far only for low-angle dark-field imaging[14], which shows less Z-contrast than sharp platinum on carbon requires[6].
Phase-contrast imaging in the transmission electron microscope (TEM), by contrast, relates the intensity linearly to the projected potential for weakly scattering, thin objects[15],[16]. Zero-loss energy filtering removes the electrons that lost energy in the sample, extends the range in which this weak-phase approximation holds, improves the contrast and keeps a good signal-to-noise ratio even at low dose[17]. Low-dose, aberration-corrected phase-contrast HRTEM had already given atomic resolution on beam-sensitive organic and inorganic test specimens[18]. For the carbon layers the projection requirement is fulfilled[19]. Low-dose protocols, in turn, came from biological tomography, where searching, tracking, focusing and exposure run at different settings[20] and a whole tilt series receives less than about 2 × 103 e−/nm2 at a resolution of about 2 nm[21]. The Ulm team combined both worlds.
A dose budget for 46 images
The researchers spread the tolerable dose carefully over the series. Tracking needed only a faint view of the field, stigmation a visible first Thon ring in the Fourier transform, focusing a faint view of the carbon layers; only the final exposure had to deliver full resolution. Each step ran at its own condenser setting, a beam blanker in front of the sample let electrons through only while the camera was exposing, and two cameras were used, one for tracking and one behind the energy filter for the tomographic images, so that the magnification never had to be changed and re-tuned. A semi-automatic script logged the dose at every step. In this way the team recorded 46 images from −57° to +57° in steps of 3° with a total dose of 3.0 × 107 e−/nm2 (Fig. 2): 1.7 × 107 e−/nm2 for the images themselves and 1.3 × 107 e−/nm2 for tracking, stigmation and focusing. Throughout the series the graphitic layers kept their pristine arrangement. The microscope was set to an information limit better than 120 pm with special illumination conditions[22]; microscope, cameras and sample are listed in a box further below.
Aligning without markers
Before a tilt series can be back-projected into a volume, every image has to be shifted into register with the others. The standard procedure, cross-correlation of successive images followed by tracking of features, here the platinum particles, in the IMOD package[23], left visible shifts: the particles are not spherical and overlap, which limited the accuracy. The team therefore developed an iterative, self-consistent alignment that uses the reconstruction itself as reference, explained step by step in the box below.
The gain shows in slices through the reconstruction along the direction most sensitive to alignment errors: with the standard alignment the layers at the tip of the soot particle are blurred, with the new one they are smooth and distinct (Fig. 3). The Fourier ring correlation, a measure of how well two halves of the data agree[27], rose by 138 %, and the spectral signal-to-noise ratio[28] by 35 %, both at a spatial frequency of 2.9 nm−1, corresponding to 340 pm.
Single layers in 3D
Individual carbon layers stand out clearly in the reconstruction, so a resolution of at least 340 pm, the typical spacing of graphite layers, was reached throughout. A linescan across the layers in the interior of the particle shows that even spacings of 320 pm are resolved in the tomogram and that deviations from the average spacing can be measured inside the volume (Fig. 4). The crystalline platinum particles diffract the beam in some directions; their intensity then overlaps the surrounding layers with less contrast, still high enough to characterize them.
How the layers wrap the platinum
The reconstruction shows the carbon layers wrapped tightly around the platinum particles, sometimes irregularly, sometimes in regular shells that bend around the particle; a few layers further out, disordered and curled layers already join the overall soot structure, occasionally with larger spacings (Fig. 5). Because each layer is located in 3D, its spacing, length and tortuosity (the ratio of the path length of a layer to the distance between its ends) can be linked to its position in the particle (Table 1).
| Value | What it shows | |
|---|---|---|
| Layers measured | 3400 | whole soot particle |
| Layer spacing | 350 ± 60 pm | close to graphite (340 pm); single spacings down to 320 pm resolved |
| Layer length | 4.6 ± 2.7 nm | longer layers near the surface |
| Tortuosity | 1.20 ± 0.22 | 1.0 = straight layer |
| Near the surface | mostly 1.0 to 1.3, within about 3 nm of the surface | long, straight layers hold the particle together |
| Tip of the particle | 1.2 to 1.4 | strongly curved layers |
| Interior | higher tortuosity in places | two ordered regions merge around a disordered one |
| Around platinum | layers tightly wrapped | irregular, or regular shells; disordered layers a few layers further out |
Table 1: Graphitic layers in Vulcan XC72 soot measured in the 3D reconstruction (this work). Mean ± standard deviation; tortuosity = path length of a layer divided by the distance between its ends. Surface values read from the paper’s histogram of tortuosity against distance from the soot surface.
Long, straight layers near the surface hold the soot particle together. Higher tortuosity belongs both to the tip of the particle, where layers reach a tortuosity of 1.2 to 1.4, and to parts of the interior, where two long-range ordered regions merge and enclose a region of highly disordered layers.
The same approach, low-voltage aberration-corrected bright-field tomography, had been tested two years earlier on a quite different material: nested MoS2 nanooctahedra, studied at the Ernst Ruska-Centre in Jülich with a Titan at 80 kV. How the two studies compare is shown in the box “Validation: Titan at 80 kV in Ulm vs. Titan at 80 kV in Jülich” below, which also holds slices through both tomograms side by side (Fig. 6).
Microscope, cameras, sample and software are listed in the box below.
Challenges named by the authors
The tilt range was limited to ±57°, so missing-wedge artifacts remain and show up as streaks at the top and bottom of the volume, whatever the alignment. Deviations of the true tilt angles from the nominal ones were not yet corrected, and the rotation of the tilt axis was set by hand. Where the crystalline platinum particles diffract, their intensity overlaps nearby carbon layers. Above all, the tolerable dose still limits the resolution: future work, the authors write, will develop strategies to further reduce the influence of beam damage, which may then allow an even higher resolution of beam-sensitive carbon materials in 3D.
Outlook
The study turned 80 kV aberration-corrected, energy-filtered HRTEM into a tomography tool for carbon that the beam would otherwise destroy: a dose budget of 3.0 × 107 e−/nm2 for a full tilt series and an alignment that needs no markers gave individual graphitic layers in 3D, and with them numbers for spacing, length and tortuosity tied to their position in the particle. The measured dose consumption also allows estimating in advance whether a tilt series of another structure will succeed. “The presented method allows further insights into the formation process of carbon soot and into the influence of the carbon support to the catalytic activity of nanoparticles,” the authors conclude. The protocol should carry over to other beam-sensitive, graphite-like carbon nanomaterials.
Resource:
Leschner, J., Biskupek, J., Chuvilin, A., & Kaiser, U. (2010).
Accessing the local three-dimensional structure of carbon materials sensitive to an electron beam.
Carbon 48(14), 4042–4048.
https://doi.org/10.1016/j.carbon.2010.07.009
Update – March 30, 2013
The news article of August 11, 2010 ends above. Later work that builds on it is listed below.
Projection matching in materials science
In a review of computational methods for electron tomography in materials science, Jose-Jesus Fernandez from the National Centre for Biotechnology (CNB-CSIC) in Madrid named alignment as a key need for atomic-scale tomography[29]. He cited the Ulm alignment as one of the few uses in materials science of projection matching, in which images are aligned to projections of the tomogram itself, a strategy common in the life sciences. Next to it he placed the marker-free refinement of Houben and Bar Sadan from Jülich, which optimizes the image shifts for the sharpest features in trial tomograms[30].
Update – July 7, 2016
An early example of sub-nanometer tomography
Introducing an MRS Bulletin issue on advanced tomography, James E. Evans from the Pacific Northwest National Laboratory and Heiner Friedrich from Eindhoven University of Technology counted the Ulm work among the first of its kind: “A few early examples of sub-nm resolution imaging of crystalline lattice planes in molybdenum disulfide and graphitic carbon have been reported,” they write, citing a study of an industrial nickel–molybdenum catalyst[31] and the Ulm soot tomograms[32].
Update – January 17, 2023
Soft nanomaterials under the electron beam
Zhiheng Lyu, Qian Chen and colleagues at the University of Illinois Urbana-Champaign reviewed electron microscopy of soft nanomaterials in Chemical Reviews[33]. On tomography they write: “Compared to hard materials, electron tomography of soft nanomaterials faces a unique challenge due to their beam sensitivity given the needed large number of tilt images and imaging time. This challenge can be overcome by using low-dose imaging and cryogenic sample preparation.” For low-dose imaging they cite two studies side by side: the Ulm soot tomograms and their own tomography of polyamide desalination membranes, carried out with Nestor J. Zaluzec of Argonne National Laboratory[34].
Set next to each other, the two studies show how differently the materials tolerate the beam. The membranes were imaged at 200 kV in a JEOL 2100 Cryo with 2.2 × 103 e−/nm2 per image, 61 images from −60° to +60°, aligned without markers by patch tracking in IMOD and reconstructed with voxels of 0.68 nm, enough to map polyamide folds 15 to 20 nm thick. The whole series received about 1.4 × 105 e−/nm2. The Ulm soot series took more than 200 times that dose and resolved single graphitic layers 340 pm apart.
Download this news article as PDF – for reading on paper, taking notes or filing.
-
Steele, B. C. H., Heinzel, A. (2001). Materials for Fuel-Cell Technologies. Nature, 414, 345–352. https://doi.org/10.1038/35104620
-
Midgley, P. A., Dunin-Borkowski, R. E. (2009). Electron Tomography and Holography in Materials Science. Nature Materials, 8, 271–280. https://doi.org/10.1038/nmat2406
-
Jinschek, J., Batenburg, K., Calderon, H., Kilaas, R., Radmilovic, V., Kisielowski, C. (2008). 3-D Reconstruction of the Atomic Positions in a Simulated Gold Nanocrystal Based on Discrete Tomography: Prospects of Atomic Resolution Electron Tomography. Ultramicroscopy, 108, 589–604. https://doi.org/10.1016/j.ultramic.2007.10.002
-
Vander Wal, R. L., Tomasek, A. J., Pamphlet, M. I., Taylor, C. D., Thompson, W. K. (2004). Analysis of HRTEM Images for Carbon Nanostructure Quantification. Journal of Nanoparticle Research, 6, 555–568. https://doi.org/10.1007/s11051-004-3724-6
-
Shim, H.-S., Hurt, R. H., Yang, N. Y. C. (2000). A Methodology for Analysis of 002 Lattice Fringe Images and Its Application to Combustion-Derived Carbons. Carbon, 38, 29–45. https://doi.org/10.1016/S0008-6223(99)00096-2
-
Cervera Gontard, L., Dunin-Borkowski, R. E., Ozkaya, D. (2008). Three-Dimensional Shapes and Spatial Distributions of Pt and PtCr Catalyst Nanoparticles on Carbon Black. Journal of Microscopy, 232, 248–259. https://doi.org/10.1111/j.1365-2818.2008.02096.x
-
Banhart, F. (1999). Irradiation Effects in Carbon Nanostructures. Reports on Progress in Physics, 62, 1181–1221. https://doi.org/10.1088/0034-4885/62/8/201
-
Telling, R. H., Heggie, M. I. (2007). Radiation Defects in Graphite. Philosophical Magazine, 87, 4797–4846. https://doi.org/10.1080/14786430701210023
-
Kamimura, O., Dobashi, T., Kawahara, K., Abe, T., Gohara, K. (2010). 10-kV Diffractive Imaging Using Newly Developed Electron Diffraction Microscope. Ultramicroscopy, 110, 130–133. https://doi.org/10.1016/j.ultramic.2009.10.010
-
Egerton, R., Li, P., Malac, M. (2004). Radiation Damage in the TEM and SEM. Micron, 35, 399–409. https://doi.org/10.1016/j.micron.2004.02.003
-
SALVE news (2009). Microscopy at the Bottom. salve-project.de, August 30, 2009. https://www.salve-project.de/newspress/microscopy-at-the-bottom.html
-
Midgley, P. A., Weyland, M., Thomas, J. M., Johnson, B. F. G. (2001). Z-Contrast Tomography: A Technique in Three-Dimensional Nanostructural Analysis Based on Rutherford Scattering. Chemical Communications, , 907–908. https://doi.org/10.1039/b101819c
-
Krivanek, O., Nellist, P., Dellby, N., Murfitt, M., Szilagyi, Z. (2003). Towards Sub-0.5Å Electron Beams. Ultramicroscopy, 96, 229–237. https://doi.org/10.1016/S0304-3991(03)00090-1
-
Buban, J. P., Ramasse, Q., Gipson, B., Browning, N. D., Stahlberg, H. (2009). High-Resolution Low-Dose Scanning Transmission Electron Microscopy. Journal of Electron Microscopy, 59, 103–112. https://doi.org/10.1093/jmicro/dfp052
-
Ho, M.-H., Jap, B. K., Glaeser, R. M. (1988). Validity Domain of the Weak-Phase-Object Approximation for Electron Diffraction of Thin Protein Cystals. Acta Crystallographica Section A Foundations of Crystallography, 44, 878–884. https://doi.org/10.1107/S0108767388002831
-
Bar Sadan, M., Houben, L., Wolf, S. G., Enyashin, A., Seifert, G., Tenne, R., Urban, K. (2008). Toward Atomic-Scale Bright-Field Electron Tomography for the Study of Fullerene-Like Nanostructures. Nano Letters, 8, 891–896. https://doi.org/10.1021/nl073149i
-
Grimm, R., Koster, A. J., Ziese, U., Typke, D., Baumeister, W. (1996). Zero-Loss Energy Filtering Under Low-Dose Conditions Using a Post-Column Energy Filter. Journal of Microscopy, 183, 60–68. https://doi.org/10.1046/j.1365-2818.1996.77441.x
-
Evans, J. E., Hetherington, C., Kirkland, A., Chang, L.-Y., Stahlberg, H., Browning, N. (2008). Low-Dose Aberration Corrected Cryo-Electron Microscopy of Organic Specimens. Ultramicroscopy, 108, 1636–1644. https://doi.org/10.1016/j.ultramic.2008.06.004
-
Götz, A. (2010). Bestimmung optimaler Untersuchungsbedingungen für Elektronentomographie an Nanokohlenstoffobjekten. Diploma thesis, Ulm University.
-
Koster, A. J., Grimm, R., Typke, D., Hegerl, R., Stoschek, A., Walz, J., Baumeister, W. (1997). Perspectives of Molecular and Cellular Electron Tomography. Journal of Structural Biology, 120, 276–308. https://doi.org/10.1006/jsbi.1997.3933
-
Conway, J., Trus, B., Booy, F., Newcomb, W., Brown, J., Steven, A. (1993). The Effects of Radiation Damage on the Structure of Frozen Hydrated HSV-1 Capsids. Journal of Structural Biology, 111, 222–233. https://doi.org/10.1006/jsbi.1993.1052
-
Meyer, J. C., Chuvilin, A., Algara-Siller, G., Biskupek, J., Kaiser, U. (2009). Selective Sputtering and Atomic Resolution Imaging of Atomically Thin Boron Nitride Membranes. Nano Letters, 9, 2683–2689. https://doi.org/10.1021/nl9011497
-
Kremer, J. R., Mastronarde, D. N., McIntosh, J. (1996). Computer Visualization of Three-Dimensional Image Data Using IMOD. Journal of Structural Biology, 116, 71–76. https://doi.org/10.1006/jsbi.1996.0013
-
Midgley, P., Weyland, M. (2003). 3D Electron Microscopy in the Physical Sciences: The Development of Z-Contrast and EFTEM Tomography. Ultramicroscopy, 96, 413–431. https://doi.org/10.1016/S0304-3991(03)00105-0
-
Herman, G. T., Lent, A., Rowland, S. W. (1973). ART: Mathematics and Applications. Journal of Theoretical Biology, 42, 1–32. https://doi.org/10.1016/0022-5193(73)90145-8
-
Marabini, R., Masegosa, I., San Martín, M. C., Marco, S., Fernández, J., de la Fraga, L. G., Vaquerizo, C., Carazo, J. (1996). Xmipp: An Image Processing Package for Electron Microscopy. Journal of Structural Biology, 116, 237–240. https://doi.org/10.1006/jsbi.1996.0036
-
van Heel, M. (1987). Similarity Measures Between Images. Ultramicroscopy, 21, 95–100. https://doi.org/10.1016/0304-3991(87)90010-6
-
Unser, M., Sorzano, C., Thévenaz, P., Jonić, S., El-Bez, C., De Carlo, S., Conway, J., Trus, B. (2005). Spectral Signal-to-Noise Ratio and Resolution Assessment of 3D Reconstructions. Journal of Structural Biology, 149, 243–255. https://doi.org/10.1016/j.jsb.2004.10.011
-
Fernandez, J.-J. (2013). Computational Methods for Materials Characterization by Electron Tomography. Curr. Opin. Solid State Mater. Sci., 17, 93–106. https://doi.org/10.1016/j.cossms.2013.03.002
-
Houben, L., Bar Sadan, M. (2011). Refinement Procedure for the Image Alignment in High-Resolution Electron Tomography. Ultramicroscopy, 111, 1512–1520. https://doi.org/10.1016/j.ultramic.2011.06.001
-
de Jong, K. P., van den Oetelaar, L. C. A., Vogt, E. T. C., Eijsbouts, S., Koster, A. J., Friedrich, H., de Jongh, P. E. (2006). High-Resolution Electron Tomography Study of an Industrial Ni-Mo/γ-Al2O3 Hydrotreating Catalyst. J. Phys. Chem. B, 110, 10209–10212. https://doi.org/10.1021/jp061584f
-
Evans, J. E., Friedrich, H. (2016). Advanced Tomography Techniques for Inorganic, Organic, and Biological Materials. MRS Bull., 41, 516–521. https://doi.org/10.1557/mrs.2016.134
-
Lyu, Z., Yao, L., Chen, W., Kalutantirige, F. C., Chen, Q. (2023). Electron Microscopy Studies of Soft Nanomaterials. Chem. Rev., 123, 4051–4145. https://doi.org/10.1021/acs.chemrev.2c00461
-
Song, X., Smith, J. W., Kim, J., Zaluzec, N. J., Chen, W., An, H., Dennison, J. M., Cahill, D. G., Kulzick, M. A., Chen, Q. (2019). Unraveling the Morphology–Function Relationships of Polyamide Membranes Using Quantitative Electron Tomography. ACS Appl. Mater. Interfaces, 11, 8517–8526. https://doi.org/10.1021/acsami.8b20826






