Materials science and X-ray techniques
J.D. Brocka, and Mark Suttonb
aSchool of Applied and Engineering Physics, Cornell University, Ithaca, NY 14853, USA
bDepartment of Physics, McGill University, Montréal, Québec H3A 2T8, Canada
Available online 27 November 2008.
Many novel synchrotron-based X-ray techniques directly address the core questions of modern materials science but are not yet at the stage of being easy to use because of the lack of dedicated beamlines optimized for specific measurements. In this article, we highlight a few of these X-ray techniques and discuss why, with ongoing upgrades of existing synchrotrons and with new linear-accelerator-based sources under development, now is the time to ensure that these techniques are readily available to the larger materials research community.
For over 100 years, our fundamental understanding of the structure of materials on atomic length scales has been advanced by direct structural measurements using X-rays. However, the flux available from sealed-tube and rotating-anode laboratory X-ray sources limits the materials whose atomic-scale structure can be solved to those that can be crystallized.
In the mid-1970s, researchers began tapping the incredible intensity of synchrotron X-ray sources – synchrotrons are 6–8 orders of magnitude more intense than laboratory sources. Typically, this intensity is used to produce higher resolution: higher angular resolution for diffraction, higher energy resolution for spectroscopies, higher time resolution for dynamics and higher spatial resolution for imaging. Researchers often take advantage of the increased intensity to increase the resolution along two or more of these axes simultaneously.
Around the world, several new synchrotrons are coming online – Diamond in the UK, Soleil in France and the CLS in Canada. In addition, major upgrades to existing facilities (APS and ESRF) and the development of ‘ultimate’ synchrotrons (NSLS-II) are in the planning stage. Furthermore, two complimentary new types of linear accelerator (LINAC)-based coherent X-ray sources are being developed: the X-ray-Free Electron Laser (XFEL) and the Energy Recovery LINAC (ERL). Both have time-average fluxes equivalent to those of current sources. The primary distinction between the two is that in the XFEL, the X-rays will come in 102 bursts per second. In the ERL, they will come in 109 pulses per second. Consequently, the XFEL is the clear choice for single-shot or nonlinear (multiphoton) studies on timescales of just a few femtoseconds. The ERL is the choice for nondestructive, continuous, or high-repetition rate (100 kHz–1.3 GHz) studies on the 100 fs–1 ps timescale. In either case, the coherent hard X-ray beams produced by these sources will enable the atomic-scale structure of nonperiodic materials to be determined.
To oversimplify, but to give a perspective on what these spectacular increases mean, whatever one can measure with a conventional laboratory X-ray source, can now be measured with existing synchrotrons from micron-sized samples. Thus, some of the most powerful new techniques are variations on X-ray microprobes and microdiffraction. The next generation of sources will push the sample size to the nanometer scale. XFEL and ERL sources will enable studies of single molecules, nanoparticles, or grains on the timescale of atomic motion. To illustrate the potential, we offer a few examples of current research.
Local interconnects for VLSI
The local interconnects between active elements are one of the important components of Very-Large-Scale Integration (VLSI) technology underlying the current revolution in computers and microelectronics. With the ever-shrinking feature sizes, interconnect sizes also shrink. As feature sizes reached the micron scale, the material used for interconnects changed from doped silicon channels to metal channels. First TiSi2, then CoSi2, and finally NiSi were used. Sophisticated heat treatments are required to convert metal films to the correct silicide phase and the research required to develop these treatments was long and involved, , ,  and .
Here, we highlight the work of the IBM group of Lavoie et al. and  on NiSi interconnects. Working at the National Synchrotron Light Source (NSLS) on the IBM beamline X20C, which is optimized for time-resolved powder diffraction, data sets covering 14° of scattering angle (2) were acquired in less than 100 ms from 10 nm thick patterned thin films. As samples underwent Rapid Thermal Annealing (RTA) at 3°C/s in an ultrapure He gas environment to minimize oxidation, X-ray, optical reflectivity and four-probe measurement of electrical resistance were performed simultaneously. Representative X-ray measurements obtained during a single RTA scan are shown in Fig. 1. Each panel plots the intensity contours of the powder diffraction profiles versus temperature (1.5°C per profile).
Fig. 1. In situ X-ray diffraction comparing Ni–Si phase formation at 3°C/s for 10 nm Ni films with increasing amounts of Co: (a) pure Ni, (b) 5 at.% Co, (c) 10 at.% Co, and (d) 20 at.% Co. Films are deposited on poly-Si substrates. The lines show the gradual decrease in NiSi phase formation with Co content (Figure from Lavoie et al.2).
To understand what occurs, consider the transitions observed in pure Ni (Fig. 1a). At low temperatures, Ni(111) and Si(220) peaks are seen. At around 250°C, a sequence of Ni-rich phases appears. Although not easy to infer from this figure, the sequence of phases is (i) Ni2Si (until the Ni is consumed); (ii) a nonuniform phase (believed to be Ni3Si2) at about 380°C; (iii) NiSi, which remains up to 800°C; (iv) NiSi2. Fig. 1 shows how the desired phase (NiSi) occurs over a reduced temperature window which shrinks with increasing Co content of the initial film. By simply comparing the four panels (Fig. 1a–d), it is easy to see the effect of alloying Co on the temperature stability range of the NiSi phase. It is hard to imagine sorting out this sequence of phase transformations and how they are affected by changes to parameters under external control, without in situ time-resolved X-ray diffraction. In addition, the RTA procedure used in these measurements is identical to the processing of commercial VLSI wafers and does not have to be modified to make the measurements.
Intermediate-range order in a nominally amorphous molecular semiconductor film
The continuing demand for improvements in display and lighting technology has driven research in electroluminescent materials. Organic Light-Emitting Devices (OLEDs) are one of the alternatives to traditional semiconductor technology currently being studied. As in many other electronic technologies, OLED performance is correlated with the microstructure of the organic thin film. For this application, amorphous films are generally preferred over single crystals as they circumvent challenges associated with surface roughness and grain boundaries. Any changes in structure as the device operates or ages are also clearly of interest as performance is likely to be affected. However, since they scatter X-rays weakly, little is known about the atomic-scale structure of amorphous films of organic semiconductors. Furthermore, in the configurations utilized in devices, where the film thickness is typically of the order of 100 nm, strong scattering from the substrate can overwhelm the weak scattering from the film. Measurements utilizing standard laboratory-based X-ray generators are thus not feasible. These challenges can be overcome using synchrotron-based grazing-incidence scattering methods.
Blasini et al.6 used grazing-incidence scattering to study the microstructure of the ruthenium complex [Ru(bpy)3]2+(PF6−)2, where bpy is 2,2′-bipyridine. A grazing-incidence geometry was used to enhance the scattering from the thin-film scattering relative to the diffuse scattering from the substrate. Specifically, by working below the critical angle for the Indium Tin Oxide (ITO) substrate, the penetration depth and thus the strong ITO powder scattering were dramatically suppressed. These measurements demonstrate intermediate-range order, exhibiting crystalline domains with an average size of a few nanometers. Moreover, these crystalline domains grow larger upon exposure of the films to moisture, suggesting that a considerable degree of order can exist in molecular films previously thought to be amorphous.
X-ray imaging of shock waves generated by high-pressure fuel sprays7
The performance of high-pressure, high-speed fuel injectors used in modern automobile engines critically impacts both fuel efficiency and the pollutants produced. Thus, understanding the structure and dynamics of these sprays is of the utmost technological importance. The structure of these liquid sprays is difficult to measure optically because of the intense multiple scattering by the fuel droplets. The small scattering cross-section of X-rays makes them nearly ideal for radiographic (absorption) imaging but the combination of a high-flux synchrotron source and a high-speed area detector is essential.
The data shown in Fig. 2 were acquired at the Cornell High Energy Synchrotron Source (CHESS) using a novel high-speed Pixel Array Detector (PAD) developed by the Gruner group and . The specific device studied was a high-pressure common-rail diesel injection system typical of those found in passenger cars but with a specially fabricated single-orifice nozzle, 178 μm in diameter. The injection pressure could be varied between 50 and 135 MPa and the fuel was a blend of No. 2 diesel with an X-ray-contrast-enhancing cerium compound. For monochromatic X-ray beams, attenuation is an exact measure of the mass, enabling quantitative measurements.
Fig. 2. Time-resolved radiographic images of fuel sprays and the shock waves generated by the sprays at 38, 77, 115, 154, and 192 μs after the start of the injection. The exposure time per frame was set to 5.13 μs (twice the CHESS period) with subsequent images taken after an additional 2.56 μs delay. Each image shown is the average of images from 20 fuel injection cycles. (Figure from MacPhee et al.7).
The time-dependent structure of the fuel spray is clearly evident in Fig. 2. However, the observation of shock waves was completely unanticipated. Indeed, the manner in which shock waves affect the atomization and the combustion is currently unknown. Furthermore, as preinjection technology (a short pilot injection before the main injection) becomes more common, the generation of shock waves during the main fuel injection will become even more prevalent.
Other imaging modalities exist. For example, Fig. 3 shows a cut-away view of polystyrene microspheres (red) in a quartz capillary (white); the color green shows a residual amount of polyacrylate occurring sporadically throughout the system. It was expected that the polyacrylate should have been contained in the spheres but it has obviously leaked out. This image is a tomographic reconstruction taken by absorption measurements at two different energies giving chemical sensitivity10. Movies showing the full three-dimensional reconstruction are available at the website http://unicorn.mcmaster.ca/highlights/tomo2/tomo2.html.
Fig. 3. Cut-away view of latex spheres (1 μm diameter) in a quartz capillary. Resolution is better than 50 nm (Figure from Johansson et al.10).
The examples given so far can be considered as conventional X-ray studies carried to extremes in terms of resolution and sample size. The tremendous intensity of the new sources also allows one to use higher order X-ray interactions to access new kinds of information. Examples include scattering X-rays off magnetic moments, combining diffraction effects with Mössbauer interactions in 57Fe nuclei and many others. We highlight these kinds of experiments with the example of coherent X-ray diffraction. Transverse coherence leads to an effect called speckle which arises in the scattering of coherent light from disordered materials. Speckle is often seen with laser light and arises from constructive and destructive interference between rays with random path length differences. As the material fluctuates in time, so does its speckle pattern. Measuring time correlation functions from these fluctuations with light is called Dynamic Light Scattering (DLS) or photon correlation spectroscopy. With X-rays, it is generally called X-ray Photon Correlation Spectroscopy (XPCS) or X-ray Intensity Fluctuation Spectroscopy (XIFS). This is an important new technique to observe the dynamics of microstructure as described in a recent review11. Here, we would like to emphasize that extra structural information becomes available when coherent illumination is used. As first demonstrated using X-rays by Miao et al.12, the missing phase information that is needed to obtain the real space structure can be self-consistently reconstructed from the measured speckle diffraction profile. This leads to yet another mechanism on which to base an X-ray microscope. Shapiro et al.13 used this technique to image a yeast cell. Another example, of direct relevance to materials scientists, is the work by Pfeifer et al.14, imaging strain fields in a lead nanoparticle. In this work they measure the shape of the Q = (111) Bragg peak from a single 750 nm diameter lead particle. The particle's Bragg peak has been broadened by the finite size of the particle and contains information on the shape of the nanoparticle. One can easily see facets in the reconstructions as expected from the equilibrium crystal shapes. Under careful study, the reconstructed density has a residual complex phase in the interior that can be attributed to a term , where is the local strain. In addition to being a powerful technique for obtaining nanometer-scale structural resolution, there is much excitement in the field about the potential of using this technique with the XFELs currently under construction. It is estimated that the coherent flux from a single pulse of an XFEL, which are expected to be around 100 fs long, could be used to collect a speckle pattern. Potentially, such speckle patterns could be obtained from a single large biological molecule and this may provide a mechanism to solve the atomic structures of molecules that cannot be crystallized.
To fully describe even a small subset of the variety of new techniques and experimental results would require a much longer article than the present one. Although we realize that all researchers are fully occupied with their current research, we strongly recommend everyone to take some time to explore the activity reports and the annual reports of the various synchrotrons. Who knows, there may be some technique ideally suited to solving one of your most pressing problems. The website http://www.lightsources.org provides a suite of links into this vast literature.
Using the X-ray beams produced by current synchrotron sources, researchers have developed a suite of new X-ray techniques with significantly improved time, spatial, and energy resolution. The potential of these improvements to transform materials research has been illustrated by a few examples. However, the complexity of the techniques tends to discourage their adoption. Even when, as in the IBM example, there is a dedicated, optimized beamline, significant expertise is required to interpret the data. Thus, these techniques are frequently viewed as too difficult and too time consuming for general practitioners. Unfortunately, this barrier has often limited the participation of the broader materials research community in the current generation of synchrotron sources. Synchrotron facilities are working hard to overcome this obstacle. Other experiments consist primarily of imaging tools such as fluorescence, confocal, and diffraction microscopes. All of these technologies will be further advanced by the new sources currently being developed. Just as scanning electron microscopy enables diverse users to image the surface of almost any vacuum-compatible sample, X-ray techniques will enable three-dimensional imaging of thick, optically opaque samples.
There is currently a large amount of activity and planning for the new sources. Workshops and requests for people to serve on planning committees appear regularly. Materials researchers can benefit tremendously from these new machines; however, they need to actively participate in the specification and prioritization for these new sources to ensure that the facilities they incorporate are optimized for their needs.