Recently in Crystallography Category

X-ray crystallography is remarkably successful at yielding atomic-resolution structures of proteins and other biological molecules. But that success relies on growing macroscopic crystals. Unfortunately, some molecules crystallize with difficulty or not at all. A decade ago, researchers predicted that the femtosecond pulses from an x-ray laser would be short enough and intense enough to produce a useful diffraction pattern from an uncrystallized biomolecular cluster before vaporizing it. An international collaboration of more than 80 scientists has now used SLAC’s Linac Coherent Light Source free-electron laser to perform two proof-of-concept demonstrations of the feat. In one study, the researchers squirted a suspension of nanocrystals (the photosynthetic protein photosystem I) across the 1.8-keV x-ray beam, recorded the two-dimensional diffraction pattern each time a crystal intersected the beam, and then combined 15 000 single-crystal snapshots to form the 3D projection shown here. From the data, the team reconstructed the protein’s structure at the near-atomic resolution of 8.5 Å, just over the x-ray wavelength. In the second study, the team injected an aerosol stream of 0.45-µm-diameter noncrystalline mimivirus particles across the beam. Thanks to the beam intensity—nearly 10¹³ photons per pulse—the collaboration was able to transform the diffraction pattern of a single virus particle into a real-space projection of its interior, though at the more modest resolution of 32 nm. The studies represent a step toward making molecular movies at atom-by-atom detail using harder and shorter pulses. (H. N. Chapman et al., Nature 470, 73, 2011; M. M. Seibert et al., Nature 470, 78, 2011.)—R. Mark Wilson

Since Andre Geim and Konstantin Novoselov first touched off the graphene “gold rush” in 2004—their pioneering work earned them this year’s Nobel Prize in Physics—researchers have been pursuing ways to scale up its production. Among graphene’s remarkable properties is its roughly 100-GPa tensile strength, which is 40 times greater than the value for steel. That, however, is for defect-free graphene sheets; when formed by chemical vapor deposition, a proven industrial technique, graphene sheets contain crystallites separated by grain boundaries (see the news story in Physics Today, August 2010, page 15). Now, a computational study by Rassin Grantab and Vivek Shenoy at Brown University and Rodney Ruoff at the University of Texas at Austin reveals that graphene sheets with highly misaligned boundaries are actually stronger than slightly misaligned ones. As the image shows, misaligned grain boundaries consist of repeating pairs of 5- and 7-member rings separated by hexagonal rings. In simulations of the stress–strain curves as a function of the misalignment, the researchers found that, surprisingly, tensile strength increases with increasing misalignment angle. According to their model, stress failure begins at critical bonds within the 7-member rings; and critical bond length, which decreases with increasing misalignment angle, is proportional to initial material strain. In one simulation, a graphene sheet with a boundary angle of 28.7° and strained by 15% resisted stress up to 95 GPa; conceivably, it might be more efficient for researchers to engineer controlled defects into a graphene sheet rather than trying to make a perfect one. (R. Grantab, V. B. Shenoy, R. S. Ruoff, Science 330, 946, 2010.)—Jermey N. A. Matthews

A team from Japan has measured the crystal structure of iron under conditions that prevail in Earth's solid inner core—that is, at temperatures and pressures higher than 5000 K and 300 GPa. To reach those extreme values, Shigehiko Tateno and Kei Hirose of the Tokyo Institute of Technology and their collaborators placed Fe powder inside the 20-μm-wide cell of a diamond anvil. Tightening the anvil's screw squeezed the sample to pressures up to 377 GPa, while two 100-W ytterbium fiber lasers raised the sample's temperature as high as 5700 K. Placing the cell in a beamline at the SPring-8 synchrotron in Sayo, Japan, yielded the structural information and enabled the team to fill in the uncharted top corner of Fe's phase diagram. Under ambient conditions, Fe adopts a body-centered cubic (bcc) structure (the red region in the bottom left corner of the phase diagram). As the temperature increases, the pressure needed to forestall melting increases too. Previous measurements (solid diamonds) had shown that Fe switches from a bcc to a hexagonal close-packed (hcp) structure (blue region) at modest temperatures and pressures. That the hcp structure survives at inner core conditions has now been established by the SPring-8 measurements (open symbols). If Fe in Earth's inner core really is hcp, then the lengthening of the crystal's c-axis parallel to Earth's rotation would naturally account for a certain anomaly in seismic signals that pass through the core. (S. Tateno, K. Hirose, Y. Ohishi, Y. Tatsumi, Science 330, 350, 2010. )—Charles Day

X-ray diffraction has long been an important tool for finding crystal structures by mapping their electron densities. In recent decades, time-resolved x-ray diffraction has probed ever-faster structural changes in single crystals, including atomic motions on the femtosecond time scale. But many materials of interest, such as the transition metal complexes used in organic photovoltaic cells, can’t easily be made into crystals of sufficient size and quality. Now Michael Woerner, Thomas Elsaesser, and colleagues at the Max Born Institute in Berlin have demonstrated femtosecond x-ray powder diffraction, in which the sample is an ensemble of randomly oriented microcrystals of ammonium sulfate, (NH₄)₂SO₄, and the diffraction pattern is composed of concentric rings rather than discrete peaks. The innovation was in engineering the x-ray source—a laser-driven plasma source that produced an ultrafast x-ray pulse from an equally brief optical pulse—to operate stably and at high repetition rate for long enough to reveal small changes in the diffraction ring intensities. From those changes, the researchers calculated the change in the sample’s electron density. As shown in the figure, which depicts the equilibrium electron density and the resulting changes over one slice through the crystal, electrons briefly pool (red blobs) where no nucleus exists in the equilibrium structure—so a nucleus, specifically a proton, must have migrated there. Ultrafast IR spectroscopy confirmed that NH₄⁺ ions were reversibly breaking apart; surprisingly, the observed structural change bears no resemblance to either of ammonium sulfate’s known phase transitions. (M. Woerner et al., J. Chem. Phys., in press.)—Johanna Miller

In the presence of a suitable nucleating agent, a liquid in a metastable state below its thermodynamically defined melting point freezes. That’s what happens when atmospheric aerosol particles cause supercooled water droplets in clouds to form snowflakes. Researchers have suspected that the atomic surface structure of such seeding particles acts as a template, inducing local order in the disordered liquid and catalyzing its crystallization. Conversely, a solid with a different structure can inhibit crystallization, as has now been observed at the European Synchrotron Radiation Facility in Grenoble, France, by Tobias Schülli and his colleagues. The researchers coupled x-ray scattering data with molecular-dynamics simulations to study supercooled gold-silicon droplets on a silicon substrate, a system that is used to grow Si nanowires. Surprising results emerged when they heated the AuSi alloy above 676 K: As it cooled, the Si atoms leached onto the substrate and, as the figure shows, rearranged its surface atoms into pentagonal clusters. The alloy’s atoms near the interface mimicked the substrate’s surface structure (see inset), but the resulting local order did not promote crystallization in the droplets, which froze at 513 K, about 120 K below the freezing temperature for the AuSi alloy. Apparently, the pentagonal geometry inhibits freezing because it is not conducive to crystal packing. That finding suggests that substrates with such atomic structures offer a simpler method of maintaining and observing the supercooling process than such techniques as magnetically levitating or otherwise suspending the liquid droplet. (T. Schülli et al., Nature 464, 1174, 2010.)—Jermey N. A. Matthews

In a single cell, thousands of simultaneously occurring biochemical reactions carry out such functions as converting and storing energy and regulating nutrient levels; together, those processes make up the cell’s metabolic network. Computational biology involves, among other things, the linking of metabolic pathways to form a metabolic network model, a promising tool for preclinical drug studies and other medical research. However, such computational models do not traditionally include the function-determining structural details of a network’s macromolecules; for example, an enzyme’s ability to catalyze reactions and regulate the cell’s response to external stimuli is determined by its three-dimensional configuration. Now, an international team led by Adam Godzik at the Burnham Institute for Medical Research in California has taken a rare step and introduced atomic-level protein structural data to the metabolic network model of an ancient thermophilic bacterium, Thermotoga maritima, shown in this optical microscope image. The image also shows schematics of proteins in their 3D configurations, which, when they were expressed in the reconstructed metabolic network, helped the research team solve the puzzle of how proteins evolve when their cell networks grow larger.

It turns out that only 37% of T. maritima’s proteins are essential to the formation of its metabolic network; those “core-essential” proteins adopt the bulk—61%—of the bacterium’s relatively few unique 3D configurations. The finding suggests that the core-essential proteins evolved their structure to perform additional functions in distinct pathways. (Y. Zhang et al., Science 325, 1544, 2009.)—Jermey N.A. Matthews

Concrete is the most prevalent synthetic material on Earth, yet the detailed nature of its primary binding constituent, hydrated cement, is only poorly understood. When cement, a dry powder that consists mostly of calcium oxide and silicate, is mixed with water, the material hardens through the formation of a complex hydrated oxide called calcium-silicate-hydrate. But the microscopic structure of C-S-H is largely unknown—even its stoichiometry, as suggested by convention with hyphens. C-S-H's structure had been thought to be related to that of two naturally occurring calcium silicate minerals, but those minerals can't explain C-S-H's observed properties. Armed with recent measurements of C-S-H's density and its ratio of calcium to silicon atoms, a team of researchers at MIT has proposed a new molecular model for C-S-H based on atom-scale simulations: Layers of calcium ions (gray in the figure) are surrounded by silicon (yellow) and oxygen (red) arranged as short silica chains one, two, and five units long; between those layers are water (oxygen in blue, hydrogen in white) and interlayer calcium ions (green) that ensure overall neutrality. The model's chemical composition, (CaO)_1.65(SiO₂)(H₂O)_1.75, agrees well with results from neutron scattering experiments. In addition to reproducing the known structural properties of the material, the model also suggests that at short length scales C-S-H should be viewed as a glassy phase. With an atom-level model of the C-S-H structure now in hand, the researchers hope to be able to manipulate the macroscopic properties of concrete, such as its strength and temperature resistance. (R. J.-M. Pellenq et al., Proc. Natl. Acad. Sci. USA, doi:10.1073/pnas.0902180106, in press.)—Richard J. Fitzgerald

Dispersed in the brains of Alzheimer's patients are disk-shaped lesions, about 100 µm across. Whether those lesions, or plaques, are a cause or a consequence of Alzheimer's disease is controversial, but their composition is clear. The plaques are made from fibrous aggregates—amyloid—of protein or their shorter cousins, peptides. Once sequestered in amyloid, a protein or a peptide can no longer perform its function. Even if amyloid does not directly cause Alzheimer's and other diseases, it seems at best a useless, dead-end repository of proteinaceous material. But as a new paper exemplifies, a less malign view of amyloid is emerging. Roland Riek of ETH Zürich and his collaborators have demonstrated that our bodies exploit amyloid as a temporary storage medium for a wide range of peptide hormones. Riek suspected a hormone–amyloid connection when he found that a stress hormone formed amyloid fibrils. He and his collaborators then subjected 41 other peptide hormones to a battery of biochemical, biophysical, and crystallographic tests. The finding: 75% of the peptide hormones form amyloid; and, as befits a storage medium, the amyloid can also disaggregate to release the peptides. In a final test, the team stained slices of mouse brain with hormone-sensitive and amyloid-sensitive dyes. The stained regions coincided. Riek's discovery adds to the modest but growing list of examples of so-called functional amyloid that perform useful tasks in living organisms. Evidently, amyloid is not always pernicious. (S. K. Maji et al., Science 325, 328, 2009.)—Charles Day