Committee Findings and Recommendations
In this report, chemical imaging is defined as the spatial and temporal characterization of the molecular composition, structure, and dynamics of any given sample—with the ultimate goal being able to both understand and control complex chemical processes. As illustrated by the case studies in Chapter 2, this ability to image or visualize chemical events in space and time is essential to the future development of many fields of science.
At present, imaging lies at the heart of the many advances taking place in our high-technology world. For example, microscopic imaging experiments have played a key role in the development of organic material devices used in electronics. Chemical imaging is also critical to understanding diseases such as Alzheimer’s, where chemical imaging provides the ability to determine molecular structure, cell structure, and communication and integrate these into obtaining information nondestructively from the human brain. Continued advances in these chemical imaging capabilities will result in more fundamental understanding of chemical processes. Concurrently, advances in other areas of research—such as nano-science and materials—underlie the developments needed to push chemical imaging ahead even further.
Chemical imaging techniques span a broad array of capabilities and applications. The findings and recommendations described below are offered as guidance for setting priorities and mapping plans toward fundamental breakthroughs in areas of imaging research as well as other areas that impact development of chemical imaging. Recommendations are presented in the order in which areas of chemical imaging research were examined in Chapters 2 and 3.
A GRAND CHALLENGE FOR CHEMICAL IMAGING
A very important goal for chemical imaging is to understand and control complex chemical processes. This ultimately requires the ability to perform multimodal or multitechnique imaging across all length and time scales. Complete characterization of a complex material requires information not only on the surface or in bulk chemical components, but also on stereometric features such as size, distance, and homogeneity in three-dimensional space. In chemical imaging, it is frequently difficult to uniquely distinguish between alternative surface morphologies using a single analytical method and routine data acquisition and analysis. Multitechnique image correlation allows for extending lateral and vertical spatial characterization of chemical phases. This approach improves spatial resolution by utilizing techniques with nanometer resolution to enhance data from techniques with micrometer resolution—such as atomic force microscopy (AFM) or scanning electron microscopy (SEM) combined with X-ray photoemission spectroscopy (XPS) or Fourier transform infrared (FTIR) spectroscopy. Multi-modal imaging also facilitates correlation of different physical properties such as phase information in AFM with chemical information in XPS. By combining techniques that use different physical principles and record different properties of the object space, complementary and better-quality information becomes available.
As in most cases of systems integration, multimodal imaging requires more than simply networking different imaging techniques. Advances in computational capabilities, for example, are fundamental to effective integration of imaging techniques. Data fusion is the name for the techniques used to combine data from multiple techniques to perform inferences that may not be possible from a single technique by itself. The goal is to combine image data to form a new image that contains more interpretable information than could be gained using the original information. Combining images to form a multimodal image requires—beyond the usual image processing for a single image—a compensation for changes in image alignment from one instrument to another due to slight movements of the specimens, slight differences in magnification, or imperfect centering of the sample.
There is a need to develop multitechnique correlations for various combinations of imaging techniques.
Two examples for techniques that may be combined are listed below, but they are by no means meant to be comprehensive.
Combining Surface Enhanced Raman Spectroscopy and Nanoscale Scanning Probe Techniques
Surface enhanced Raman spectroscopy (SERS) experiments on silver and gold nanoclusters have demonstrated large enhancement levels and field confinement of 5 nm or less for various samples such as single-walled carbon nanotubes.1 However, the locations of these conditions cannot be controlled but are instead determined by the specific nanostructures used. That is, the target molecules have
to be in the close vicinity of SERS-active nanometer-sized silver or gold substrates. On the other hand, location can be controlled in so-called tip-enhanced SERS experiments.2 Unfortunately, these experiments provide only small SERS enhancement factors compared with samples interacting with metal nanoparticles. Further development of tip-enhanced technology would benefit from experimental systems that combine high SERS enhancement factors and highly confined probed volumes with nanoscale-controlled scanning. This may be accomplished by combining nanoscale scanning probe techniques (such as modified atomic force microscopy [AFM] systems) with the techniques of single-molecule Raman spectroscopy.
Combining X-rays, Electrons, and Scanning Probe Microscopies
Scanning probe microscopy (SPM) techniques typically provide topographical, not chemical identification, so that combining other local spectroscopies with STM is typically necessary to identify the atoms and molecules present. Specialized approaches are being developed to address this. For example, progress has been made in joining transmission electron microscopy (TEM) and scanning electron microscopy (SEM) to scanning tunneling microscopy (STM). STM photoemission spectroscopy (PESTM) combined with inelastic electron tunneling spectroscopy yields vibrational and other information. Integrating the three techniques will enable investigation of the chemical (X-ray, infrared, or Raman), structural (EM), and topographic (SPM) nature of samples.
AREAS OF IMAGING RESEARCH
Understanding and controlling complex chemical processes also requires advances in more focused areas of imaging research. These chemical imaging techniques span a broad array of capabilities and applications, and are discussed in great detail within this report. Here, we briefly highlight the research and development that will best advance current capabilities—with a focus on applications in which investment would most likely lead to proportionally large returns. The main findings of the committee are:
Nuclear Magnetic Resonance
Nuclear magnetic resonance (NMR) and magnetic resonance imaging (MRI) represent mature technologies that have widespread impact on the materials, chemical, biochemical, and medical fields. NMR and MRI are very useful tools for obtaining structural and spatial information. It is clear that in the coming years, NMR and MRI will continue to expand rapidly and continue to be key tools for chemical imaging. However, the major limiting factor for application of these techniques to a broader range of problems is their relatively low sensitivity, which is a result of the low radio-frequency energy used. There are a number of
ways that need to be explored to obtain more signals from NMR and MRI, and these are provided below:
A major limiting factor of NMR and MRI is the relatively low sensitivity of their detectors. Progress has been made in increasing detector sensitivity by using supercooled detectors (increasing sensitivity by a factor of 2 to 4), and further progress could be made with new materials both for higher-temperature super-conductors and for better insulation. Work with other detector strategies such as superconducting quantum interference devices (SQUID) and other novel magnetometers should be encouraged, especially in light of progress in hyperpolarization (see below). Finally, force detection of magnetic resonance is a very promising area that is limited by detector design.
For MRI, gains in sensitivity by a factor of two- to fivefold have been realized primarily by building parallel arrays of MRI detectors. Cross-talk between the array detectors limits performance, and the configuration of these detectors is bulky. Novel approaches to building MRI array detectors will make dense arrays possible. Once the construction of dense arrays using small coils is achieved, the noise in the coils themselves will become limiting to signal to noise. It should be possible to cool the coils (as has been done in high-resolution NMR). Making cold, dense, parallel arrays feasible should enable at least a tenfold increase in sensitivity for both NMR and MRI. This will increase the resolution that can be obtained with MRI as well as with MR spectroscopic imaging of a large number of metabolites.
Increasing signal-to-noise ratios should be a chief focus of the efforts to improve the sensitivity of NMR and MRI detectors.
Another very promising avenue for increasing sensitivity in NMR and MRI is to increase signal from the molecules being detected. In recent years there has been growing interest in hyperpolarization techniques, which couple the nuclear spins being detected by NMR to other spins with a higher polarization, that increase NMR sensitivity by factors of 100-100,000. Exciting developments for hyperpolarization using a variety of techniques—such as dynamic nuclear polarization, laser-induced hyperpolarization of noble gases, and formation of parahydrogen—have realized extraordinary gains in sensitivity for applications to materials research and biomedical imaging. At present only a restricted set of molecules has been hyperpolarized using only a small set of possible techniques.
There is a need to expand the range of techniques useful for hyperpolarizing NMR and MRI signals, as well as the range of molecules that can be hyper-polarized.
Strategies for new MRI contrast agents parallel the development of fluorescence probes; however, MRI contrast agents are more typically used in animal models and humans. Therefore, more emphasis is placed on the safety of these contrast agents than on their usefulness as probes for wider uses. It is critical that the MRI relaxivity of contrast agents be improved so that they can be used in smaller quantities. In addition, much progress has been achieved over the past five years in the development of MRI probes that are site-specific as well as capable of tracing particular biological and chemical processes. Most of this has been proof-of-principle work; the more difficult task of optimizing these approaches to ensure their robustness must now be undertaken. Finally, preliminary work in identifying MRI-active proteins or protein assemblies that are equivalent to fluorescent proteins has begun. This is an area in which great gains can be made.
MRI probes need to have higher relaxivity, be more specific, and be deliverable to the site of action.
The sensitivity of magnetic resonance also increases with higher magnetic fields. In the range where detector noise dominates, sensitivity increases as approximately the square of the increase in field. In practice, this is difficult to realize, particularly because many samples of interest contribute to noise, leading to an increase in sensitivity that is linearly proportional to magnetic field strength. Nonetheless, much interest has been focused on producing higher magnetic fields, which means larger magnets and larger (expensive) dedicated facilities to house them. There is work now being done to decrease the siting requirement of high-field magnets, for example by employing innovative designs for superconducting wire that can carry higher current densities. These efforts could decrease the size of magnets, enabling very-high-field NMR and MRI to transition from dedicated laboratories to widespread use for applications such as advanced oil exploration, homeland security, and environmental study.
The miniaturization of high-field NMR and MRI magnets is needed to broaden the applicability of these techniques by reducing the need for dedicated facilities.
In contrast to NMR and MRI, optical spectroscopy imaging techniques utilize radiation at an energy level high enough to allow individual photons to be measured relatively easily with modern equipment at a detection sensitivity almost matched by the mammalian eye. As a result, imaging data are acquired at the sensitivity of individual molecules. The inherent temporal and spatial resolution is also increased proportionately, but the resonance itself is broad because envi-
ronmental influences are not averaged out within the inherent time scale of interaction between the molecules and this frequency of radiation. As a result, the chemical structural information content of optical spectra is considerably lower than that of magnetic resonance, particularly in the electronic region of the spectrum. Thus, research needs in the area of optical imaging are focused more on increasing chemical structural information.
New Probes Based on Metallic Particles
In terms of the high content of chemical structural information at desired spatial and temporal resolutions, Raman spectroscopy has the potential to be a very useful technique for chemical imaging. However, a disadvantage in many applications of Raman imaging results from relatively poor signal-to-noise ratios due to the extremely small cross section of the Raman process, 12 to 14 orders of magnitude lower than fluorescence cross sections. New methodologies such as localized SERS utilizing metallic nanoparticles can be used to overcome this shortcoming. Metallic nanoparticles have long been used in Mie scattering dark-field microscopy. However, nonspherical particles and their aggregates, which cannot be described by classic Mie scattering theory, offer rich optical properties associated with surface plasma-related phenomena. In recent years, this field has experienced much research activity due to the ability to fabricate new nanostructures, the emergence of sensitive microscopes and detectors, and the availability of tools for electromagnetism computation. However, these advances represent only the beginning of this research area; much work is still needed. For example, almost 30 years after its discovery, there exists little quantitative or even qualitative understanding of SERS, which arises in part from a strongly enhanced electric field in the close vicinity of gold and silver nanostructures. Better understanding of radiation signals—including Raman scattering, Mie scattering, and fluorescence—from the nanostructures or atomic clusters is an important prerequisite for the creation of new optical probes. In particular, probes that exploit SERS signals show promise in providing specific spectroscopic signatures and multiplex capabilities along with chemical specificity.
There is a need to develop a better theoretical understanding of the radiation signals of gold and silver nanostructures including Raman scattering, Mie scattering, and fluorescence. New probes composed of metal-based nano-particles or atomic clusters should be developed to provide improved sensitivity, specificity, and spatial localization capabilities.
Fluorescent Labels for Bioimaging
Unlike NMR spectroscopy and vibrational spectroscopy, electronic spectroscopy involves interactions with electromagnetic waves in the near-infrared, visible, and ultraviolet (UV) spectral regions. While electronic spectroscopy
is less enlightening about structural information than NMR and vibrational spectroscopy, the shorter wavelengths involved allow higher spatial resolution for imaging, and its stronger signal yields superb sensitivity. Fluorescence detection, with its background-free measurement, is especially sensitive and makes single fluorescent molecules detectable. On the other hand, particularly under ambient conditions, the amount of molecular structural information that can be obtained from fluorescence imaging is limited.
Organic fluorophores or labels that bind specifically to macromolecules, metabolites, and ions provide powerful tools for chemical imaging in cells and tissues. For example, green fluorescent protein and its derivatives allow live cell imaging and tracking of individual proteins. In addition, techniques such as fluorescence correlation spectroscopy (FCS), fluorescence resonance energy transfer (FRET), and multiphoton microscopy may be used for localization studies as well as for some cases of chemical reaction dynamics research. However, the efficiencies of chemical and biological labels are hampered by photobleaching. There is a great need to develop more robust labels. To accomplish this, one must understand the photophysics and photochemistry of fluorescent labels as well as the mechanisms of photobleaching. Suggested substitutes, such as semiconducting nanoparticle “quantum dots,” have been limited by their intermittent blinking and by the large size required for water-resistant coatings. Promising routes for increased label robustness include dye-molecule clusters fixed in silica shells and photochemically switchable labels.
Besides needing to be sufficiently robust, bright, nontoxic, and small in size, labels must also demonstrate chemical or biological specificity, which is of key importance. Further development of fluorescent labels for widespread application of dynamics research would also benefit the broader chemical imaging community. For example, in biological contexts the incorporation of unnatural fluorescent amino acids into nascent polypeptide chains by genetic encoding is a promising approach.
In order to probe chemical constituents and follow their biochemical reaction in cells and tissues, there is a need to make fluorescent labels more specific, brighter, and more robust. This will require greater understanding of the photophysics and photochemistry of fluorescent probes and the mechanisms of their photobleaching.
Nonlinear Optical Techniques
In addition to imaging based on single-photon excited or linear Raman scattering, vibrational images can also be generated using nonlinear coherent Raman spectroscopies. The most prominent nonlinear Raman process for imaging is coherent anti-Stokes Raman scattering (CARS). Like spontaneous Raman microscopy, CARS microscopy does not rely on natural or artificial fluorescent labels, thereby avoiding issues of toxicity and artifacts associated with staining
and photobleaching of fluorophores. Instead, it depends on a chemical contrast intrinsic to the samples. CARS microscopy offers two distinct advantages over conventional Raman microscopy: (1) The radiation damage is significantly less for CARS than for spontaneous Raman, especially when one is interested in following a dynamic process with short data collection time; and (2) it has threedimensional sectioning capability because the nonlinear CARS signal is generated only at the laser focus where laser intensities are highest. This is particularly useful for imaging thick tissues or cell structures. Techniques such as CARS microscopy and other nonlinear Raman methods offer the possibility of new contrast mechanisms with chemical sensitivity, but their potential depends critically on advances in laser sources, detection schemes, and new Raman labels. Efforts have been made to circumvent the diffraction limit by engineering the point spread function using nonlinear optical techniques. Continued developments in these nonlinear approaches will enable superhigh resolution using far-field optics without the need to employ proximal probes. Multiphoton fluorescence microscopy can also benefit from the development of more compact ultrafast lasers, fiber delivery, and improved fluorophores with larger nonlinear polarizability.
Nonlinear optical techniques need to be developed—with particular emphasis on improved ultrafast laser sources and special fluorophores, novel contrast mechanisms based on nonlinear methods for breaking the diffraction barrier without using proximal probes.
Ultrafast Optical Detectors
Current streak camera technologies allow one to measure the lifetime and spectral features of fluorescence with subpicosecond and subnanometer resolution, but they lack the sensitivity required for single-molecule applications. Charge-coupled device (CCD) cameras, on the other hand, can provide high spectral and/or spatial resolution at high quantum efficiency, but they lack temporal resolution and near-infrared (IR) sensitivity. Improvements in time resolution would also be a boon to lifetime imaging and single-molecule experiments using photon counting avalanche photodiodes, especially with the extension of the spectral range of the detector to near-IR (NIR) and UV regions. IR sensitivity, as has been pointed out, is an especially critical area for improvement, because IR and UV detectors have lagged behind visible detectors in most respects; yet IR provides some of the richest information about chemical structure. Although such advances are likely to be incremental, the effects of sustained successive improvements would be transformative. As yet, the tremendous power of combining modern ultrafast laser technology with high spectral resolution in spatially imaged measurements at nanoscale spatial resolution has not been realized, but this kind of multidimensional measurement is precisely what is required to follow the dynamics of complex interacting mixtures of chemical species.
There is a need for detectors to be developed that possess all of the following
attributes: (1) the ability to measure multiple dimensions in parallel fashion, (2) high time resolution, (3) high sensitivity, and (4) broad spectral range. IR and UV detector improvements, even if incremental, could catalyze new chemical insights.
Electron and X-ray Imaging
Techniques that probe samples with wavelengths much smaller than that of visible light provide high-resolution chemical and structural information below surfaces of materials. With wavelengths that are about 1,000 times smaller than that of visible light, electrons provide a high-resolution probe of chemical and structural information below surfaces of materials. Images of atomic arrangements over a large range of length scales can be obtained using EM techniques. X-rays can penetrate materials much more deeply than either visible light or electrons, producing chemical images that cannot be obtained by any other means.
Sources for Electron Microscopy
A limiting factor in EM is the quality of the electron beam used to probe the sample. Aberrations introduced by the optics limit both spatial resolution and analytical capabilities. Correcting for spherical and chromatic aberrations introduced by the electron optics will directly improve resolution and other analytical techniques. Imaging and diffraction will be directly improved by the greater coherence in the beam. In particular, advances in imaging techniques will permit the analysis of amorphous samples. Smaller beam sizes can be achieved, allowing for sub-Angstrom resolution chemical analysis of samples. In addition, aberration correction will relax the constraint on the sample volume, allowing for “a lab-in-the-microscope” approach to in situ microscopy. This will open a vast range of imaging environments such as reactivity measurements, mechanical deformation, and materials synthesis processes.
There is a need to develop higher-quality electron beams in order to broaden and deepen the application of electron microscopy.
Electron Microscopy Detectors
Improved detectors are also needed for EM to enable higher time resolution for imaging chemical kinetics. Higher sensitivity detectors will reduce the amount of electrons needed to image, therefore minimizing the damage that may occur from the electron beam. This will greatly expand the in situ environments for EM that are vital for imaging chemistry, such as reaction dynamics and electron-sensitive materials, including organics and biological samples. Such capabilities are likely to be achieved by developing high-density detector arrays coupled to fast discrimination electronics.
Spatial and temporal sensitivity of electron microscopy detectors need to be improved.
Optics for X-ray Microscopy
Zone plates are diffractive optics for X-ray microscopy that use constructive interference of light rays from adjacent zones to focus. Present zone plates are extremely inefficient (10-20 percent). As a result, a choice must often be made between (1) high efficiency with minimum radiation dose using lower-resolution optics and (2) the highest possible resolution. For trace element mapping in microprobes or mapping nanoscale chemical heterogeneities in spectromicroscopy, higher spatial resolution translates into the ability to carry out chemical imaging at a finer scale with less biasing of quantification. In addition, for immunolabeling, the label size must be comparable to the zone plate resolution in order to be detected; the development of higher-resolution optics will allow the use of smaller labels, which are dramatically easier to coax across the membrane of a cell. Efforts to improve resolution and efficiency require substantial work in nanofabrication in order to make zone plates that push the resolution frontier. Development of optics to correct spherical and chromatic aberrations will greatly improve the resolution of photoemission electron microscopes (PEEM) at third-generation synchrotrons. The development of ultrafast X-ray sources will expand the capabilities of X-ray imaging in terms of space and time.
There is a need to improve zone plate optics, which are presently the limiting factor for scanning transmission X-ray microscopes (STXM) and full-field X-ray microscopes (TXM).
The most common type of X-ray detector is the CCD. In its current incarnation, X-rays from the sample are imaged on a phosphor screen. This converts the X-rays to light, which is then transferred to the CCD chip by means of fiber optics or lens systems. Eliminating this conversion step, and imaging the X-rays directly onto a CCD chip with column parallel readout, will result in a detector with significantly greater sensitivity, higher resolution, and about a hundredfold faster readout speed than today’s generation of detectors. These devices would complement the current developments in novel, sophisticated, soft X-ray techniques, such as full-field, deconvolution tomography, which depends on the ability to collect large numbers of high-resolution images rapidly.
Great promise also lies in the further development of solid-state “pixel detectors” for X-ray imaging. Small versions of this detector type are now being used in several industrial and medical imaging applications. However, the technology is relatively immature and requires significant improvement to make it suitable for use in advanced research applications. Another important need is the
development of detectors for hard X-ray tomography. The primary focus should be in the area of scintillators for the conversion of X-rays to visible light. These scintillators have to be analyzed at high spatial resolution by light microscope techniques in order to reduce the pixel size of the X-ray to light conversion materials. Finally, an important new class of detectors for chemical imaging is the ultralow-temperature, high energy resolution, solid-state, energy-dispersive detector. A major effort is also required in the area of large-area, low-temperature detector arrays. The astronomy community has already made significant progress in this area, and it is timely to apply these impressive advances to chemical imaging. This has particular application to X-ray microscope-based fluorescence imaging and X-ray spectral imaging of cells and organometallic protein systems.
X-ray detectors—including solid-state pixel detectors and detectors for hard X-ray tomography—need to be improved through the development of scintillators that convert X-rays to visible light and detectors that image directly onto a CCD chip with column parallel readout, among other detector possibilities. The goal is to improve X-ray detector’s resolution, dynamic range, sensitivity, and readout speed.
Probes for X-ray Imaging
The use of X-ray microscopy to image chemical signals in biological materials requires probes that can be applied to both endogenous (naturally occurring) and exogenous (artificially introduced) molecules, particularly proteins. In most cases, endogenous proteins are identified using well-established immunochemistry techniques. However, application of these techniques to X-ray microscopy requires the conjugation of X-ray-dense moieties (that either absorb or become excited by X-rays) to the antigenic molecule. To date, gold-conjugated antibodies have been used to image a single species of protein inside a cell. The challenge now is to develop the capability to simultaneously detect multiple proteins inside the cell. This necessitates the development of multiple probes, each of which would contain a specific metal atom that could be excited at a different X-ray energy (e.g., nanocrystals containing atoms such as Ti, V, Fe, and Ni). Chemical signals from specific cellular structures can also be visualized, giving rise to high-resolution information about multiple proteins and interaction partnerships in context. Furthermore, based on the X-ray absorption coefficient, all measurements can be quantified in terms of concentration as well as location. On the other hand, the detection of specific chemical signals from exogenous molecules in the cell requires the development of probes analogous to green fluorescent protein (GFP) that also possess X-ray-absorbing powers. In both cases, these specific probes will facilitate three-dimensional localization of chemical signals at an isotropic resolution approaching 15 nm from whole, hydrated cells.
There is a need to advance X-ray-absorbing probes to specifically detect and localize chemical signals that are introduced into cells.
Since the advent of the scanning tunneling microscope in the early 1980s, a wide variety of related microscopies using similar experimental principles and instrumentation have been developed for imaging samples based on their electronic, optical, chemical, mechanical, and magnetic properties. All find broad applications in high-resolution chemical imaging experiments. Proximal probe microscopes employ a variety of materials—or probes—such as tungsten wire (STM), silicon nitride pyramid and cantilever (AFM), or optical fiber (near-field optical microscopy) in close proximity to the sample of interest for the purposes of recording an image of the sample, performing spectroscopic experiments, or manipulating the sample. These methods are especially useful for understanding the chemistry of surfaces—for example, the electrophilicity of individual surface atoms, the organization of atoms or molecules at or near the surface, and the electronic properties of atomic or molecular assemblies. Two areas of research that would help expand these capabilities are highlighted below.
Most high-resolution imaging techniques in materials science are limited to imaging surfaces or near-surface regions. Imaging below surfaces would allow studies of chemistry at the atomic or molecular level occurring at buried interfaces and/or defects sites in the bulk of samples. For example, the development of proximal probe methods (e.g., magnetic resonance force microscopy) by which images of samples can be recorded with high spatial resolution in all three dimensions would represent a major breakthrough in chemical imaging technology. At present, this method is limited primarily by the need for more sensitive cantilevers and stronger magnetic field gradients. Advances in these areas (e.g., force measurement tools for general applications) would allow more sensitive detection and greater spatial resolution to be achieved.
There is a need to develop methods for optical, X-ray, Raman, and other probe regimes that can image at depths of a few nanometers to macroscopic distances beneath a surface, especially for materials science applications.
Many interesting materials systems are chemically heterogeneous on a wide range of length scales down to atomic dimensions. The development of chemically selective proximal probe imaging methods has played a central role in uncovering sample heterogeneity and understanding its origins. Numerous chemically selective, spectroscopic proximal probe methods continue to emerge from a number of labs around the world. Both the evolution of existing methods and the further development of new ones promise significant advances in our ability to
obtain chemical information on heterogeneous samples on a variety of relevant length scales. One of the best examples of the use of proximal probe methods is the chemical bonding information that has been obtained on semiconductor surfaces by STM. However, in the biomedical realm and other application areas, the detection of discrete chemically specific binding interactions between proteins and peptides and/or drugs and receptors is of particular importance. Such measurements can be made by methods such as chemical force microscopy, but this approach lacks generality. That is, it provides significant specific information when much is already known about the sample surface composition. The development of new probes and chemical probe arrays will significantly advance these methods in the near future.
Contrast mechanisms that reveal chemical identity and function in surface characterization need to be improved for a wider variety of samples.
Molecular spectroscopies are usually restricted to length scales governed by the wave-like nature of light; specifically, spatial confinement of the source radiation is limited by the diffraction barrier to approximately one-half the wavelength of light. Near-field optical microscopy (optical proximal probe methods) overcomes this limitation and provides a means to extend optical spectroscopic techniques to the nanometer scale. However, the use of near-field microscopy to obtain chemical images of real-world samples remains hampered by issues of resolution and sensitivity. To overcome this limitation, we have to fabricate nanostructures that strongly localize and enhance the electric fields. Although recent results demonstrate the high potential of the field enhancement method, the technique is far from being well understood, reliable, or optimized.
There is a need to improve probe geometries for high-resolution chemical imaging beyond the diffraction limit. This includes design (theory) and realization (reproducibility, robustness, mass production) of controlled geometry near-field optics.
Image Processing and Analysis
Chemical imaging is used to selectively detect, analyze, and identify chemical and biological samples, followed by visualization of the data in the dimension of interest. The information of interest can range from composition, structure, and concentration to phase or conformational changes as a function of time or temperature. The expression “chemical image” describes a multidimensional dataset whose dimensions represent variables such as x, y, z spatial position, experimental wavelength, time, chemical species, and so forth. Image processing requires that the chemical images exist as digital images.
Initial Image Visualization
Frequently the first priority for the analyst is to generate an image or images that allow for visualization of heterogeneous chemical distributions in space or time. Image visualization methods vary from simply choosing a color scale for display of a single image to methods for displaying three-dimensional datasets. Simple gray-scale maps can be constructed from a single image. Different color scales can be utilized, and the contrast and brightness can be adjusted so that the information the analyst deems most important is emphasized. Multiple images from the same or different datasets can be viewed simultaneously for comparison. Scatter plots are frequently utilized for comparing two images. For more detailed comparisons among a small number of images, mapping individual images into red, green, and blue (RGB) channels creates composite color chemical images. For three-dimensional data, additional analysis tools are required, including the ability to extract spectra from a selected region of interest for multispectral imaging datasets or rendering a three-dimensional volume or projection for depth arrays.
Analysis tools for three-dimensional visualization need to be developed for various microscopies and materials analysis instrumentation.
Typically, data analysis is not considered until after an instrument is developed. This can often limit the imaging analysis or make it unnecessarily difficult. Particularly with quantitative techniques, questions such as, “How do you maintain calibration or correct for instrument drift over time?” have to be addressed from the initial stages of instrument design.
Researchers should be encouraged to integrate their data analysis with the development of their apparatus.
Multidimensional Image Processing
Present commercial multivariate analysis software is based on techniques that are more than 20 years old. Research on multivariate techniques, the development of chemometric analysis tools applied to imaging, and the deconvolution of hyperspectral images all need significant support. Developing these techniques to the point of routine use is an important challenge. Chemical imaging can benefit from other fields that depend on image analysis. The remote sensing field is rich in techniques for image analysis that are largely unexploited in chemical imaging; some kind of cross-fertilization between these two communities should be promoted. In addition, interactions with computer scientists and statisticians would enhance the development of chemical image analysis.
There is a need to develop better analysis and data extraction techniques for
elucidating more and different kinds of information from an image. In particular, this should include user-friendly multivariate analysis tools and hyperspectral imaging deconvolution and analysis.
Integrated Real-Time Analysis
As the sophistication and multimodality of imaging instrumentation increase and as the resolution of data collections improves, it will be necessary to perform some aspects of data reduction interactively during the measurement. Intelligent systems that can recognize information content and adjust resolution accordingly will have to be created. For example, high-resolution confocal imaging of large three-dimensional tissue sections in which specific, dispersed cell types are monitored for dynamic function requires programs that can automatically recognize appropriate cells and focus the dynamic data collection on those areas.
Integrated real-time analysis needs to be expanded for automated customization of data collection, particularly in multiscale imaging applications.
Electronic Structure and Molecular Dynamics Simulations
A quantitative understanding of the electronic structure of molecules and the theory that predicts the outcome of interactions of molecules with electromagnetic fields will aid in the development of chemical imaging probes for all imaging modalities. For example, it has been known since the first NMR experiments that chemical shifts are exquisitely sensitive to the electronic environment of a molecule. The ability to understand electronic structure well enough to predict NMR chemical shifts should address a variety of problems in chemistry such as predicting reaction and folding pathways.
A quantitative understanding of molecular electronic structure is needed to make advances in chemical imaging. Two chief ways in which this understanding can be furthered are through improving probes and better theory.
Molecular dynamics (MD) simulations can be performed to help better understand electronic and molecular structure. There exist at present several cybertools for performing MD simulations of various systems by experts or near experts. These tools are fairly mature in capability, but the user interfaces have only recently started to make them available to scientists other than the experts in the computational chemistry community.
Current MD simulations are thus not quite ready for use by nonexperts, particularly in cases involving multiple dimensions. One challenge will be to improve on these cybertools to make their use completely transparent. In particular, simulation packages taking MD approaches for modeling nonbiological systems lag behind comparable packages for biological systems. The packages should be able to interface with the common platforms (middleware) developed
in the context of chemical imaging cyberinfrastructure. New theory, new algorithms, and new programs are needed to enable such use in the future.
There is a need to develop a next generation of readily accessible, easy-to-use MD simulation packages.
In order to improve MD simulations, a number of specific areas should be addressed in the area of basic molecular dynamics theory. These include: (1) development of full quantum mechanical calculations on complex molecules and more robust ways to incorporate quantum mechanical calculations within larger-scale classical mechanics or statistical mechanics approaches; (2) development and refinement of transferable force fields between arbitrary atoms and molecules, which are necessary building blocks for MD simulations of general systems; and (3) development of multiscale theories and techniques for understanding systems. Moreover, the community must develop toolkits that allow general users to perform such simulations.
Chemical imaging would be invigorated by innovations in basic theory of molecular dynamics. At the same time, the specific needs of chemical imaging should play a role in guiding the development of MD theory.
All Imaging Techniques
In addition to targeted improvements for individual and multitechnique approaches, there are certain developments needed for overall advances in chemical imaging capabilities. These range from improved light sources to data management, and they are discussed briefly below.
Advances in light sources are providing new capabilities in chemical imaging. For example, recent developments in free-electron lasers have led to a rapidly growing interest in using the terahertz range (3-300 cm–1) for imaging. However, further development of terahertz spectroscopy as a powerful tool for imaging will depend on the development of convenient new terahertz light sources.
Brighter, tunable ultrafast light sources need to be developed, particularly infrared-terahertz vibrational and dynamic imaging, near-field scanning optical microscopy (NSOM), and X-ray imaging.
One of the emerging promises of chemical imaging is reversing the direction of information transfer. Many of the same approaches that are used to gather information at high resolution in spatial and temporal dimensions can also be used to control or manipulate chemical systems with similar spatial or temporal
resolution. Examples include photolithography in the semiconductor industry and light-directed synthesis of oligonucleotides in the DNA chip industry. This could be greatly extended to include high-speed synthesis and activity screening, initiating or controlling chemical reactivity, spatially patterning cellular growth and function, and ultimately mediating chemical activity throughout an entire organism.
There is a need to develop imaging methods for patterning complex spatial and temporal organization into chemical systems.
Lack of advances in optics has hampered improvements in microscopic imaging. Development of adaptable, inexpensive fiber optics to transmit high-energy femtosecond pulses from mode-locked lasers, custom phase plates, and miniature laser beam scanners for endoscopic microscopy instruments offer the potential for enormous advances in laser scanning microscopy for various applications, including medical diagnostics and surgery.
There is a need to develop optics for miniaturization and speeding of microscopic imaging instrumentation in order to improve chemical imaging capabilities.
Acquisition Speed and Efficiency
Higher-speed scanning probes that now reach video-rate imaging have been developed. Brighter probes will reduce the integration time in single-molecule detection techniques. Increasing the speed of microscopic image application instrumentation offers the potential for rapidly advancing and expanding chemical imaging capability.
Acquisition speeds need to be increased in order to provide improved time resolution. In addition, there is a need to provide more online analysis capabilities to improve the efficiency of imaging by allowing more directed investigations of samples.
With increasing acquisition and imaging resolutions, data set size and volume, and processing capabilities, improved software to process and correlate images across experiments and time points will be needed. At the same time, cyber infrastructure and its underlying theoretical frameworks can provide computer-generated images that provide insight into the structure and function of both particular and generic samples. Simulations can also allow test cases for new paradigms for the cybertools used to manipulate data, process metrics, and render images using the experimental measurements.
Theory needs to be developed and better utilized to address the data storage
and search problems associated with the increasingly large datasets generated by chemical imaging techniques.
As illustrated by the review of techniques and myriad examples given throughout this report, the field of chemical imaging is poised to provide fundamental breakthroughs in the basic understanding of molecular structure and function. The knowledge gained with these insights offers the potential for a paradigm shift in the ability to control and manipulate matter at its deepest levels. A strategic, focused research and development program in chemical imaging will best enable this transformation in our understanding of and power over the natural world. The committee has identified three key components of such a program: investigator approach, funding, and standards.
Some research efforts in chemical imaging (especially those at the interfaces of chemistry with other fields such as medicine, materials, and environment) may require a multidisciplinary team rather than individual investigator approach. For example, the realization of more safe and effective MRI contrast agents for clinical use requires a concerted effort of chemists, molecular biologists, radiologists, and MRI physicists. Forming a multidisciplinary team at an early stage and continued interaction at all stages of imaging technique development can be crucial. A key aspect of being able to form multidisciplinary teams is the development of reward systems at research institutions (funding, citations, and promotions or tenure) that recognize the equal importance of all facets of chemical imaging.
There is a need to encourage both individual investigator and multidisciplinary team approaches for the development of chemical imaging techniques.
In many aspects, the resource appropriation for chemical imaging research works quite well. For example, support for synchrotron light sources is very effective and should be continued. As described in the previous paragraph, however, breakthroughs in imaging often require the sustained effort of multidisciplinary teams of scientists. This process, which is vital for success, does not naturally match present funding schemes and cycles. New programs and channels for funding have to be established to support this mode of science.
Many of the opportunities for imaging often require coordination across several funding institutions. For example, the Department of Energy holds the primary mission for the design, construction, and operation of major facilities such as light sources, while other agencies such as the National Science Founda-
tion and National Institutes of Health support a significant number of science programs that utilize these facilities. Even greater cross-agency planning, coordination, and support will be vital to success. Support for this multidisciplinary research area in the form of (1) funding of proposals with multiple principal investigators, (2) financial support of the collaborative process (e.g. travel, meetings, training to facilitate interactions), and (3) larger grants of longer duration to finance extended startup periods could be crucial in facilitating chemical imaging development and advances.3 In addition, creation of special funding channels directed exclusively toward chemical imaging development offers great promise for advances in this field.
Novel approaches to funding mechanisms for chemical imaging need to be promoted.
At present, each subfield of imaging science develops its own structure for data storage and archiving. This has resulted in countless formats that inhibit the sharing of data. A broad effort to develop standards for data format and organization across experimental contexts will give a great boost to imaging science. Standards are needed for chemical imaging that allows image data to be effectively catalogued and shared between large numbers of researchers. These data should be organized in visually oriented, searchable, and parameterized databases, with common platforms for a vast array of image types. The implementation of such a system will be transformational. The failure to implement such a system will mean that chemical image data obsolesce almost instantly upon their production. To avoid redundant effort and the squandering of resources, solutions must be found to ensure chemical image longevity.
There is a need to develop standards for chemical image data formatting.
Imaging has a wide variety of applications that have relevance to almost every facet of our daily lives. These applications range from medical diagnosis and treatment to the study and design of material properties in novel products. To continue receiving benefits from these technologies, sustained efforts are needed to facilitate understanding and manipulation of complex chemical structures and processes. Chemical imaging offers a means by which this can be accomplished by allowing the acquisition of direct, observable information about the nature of these chemistries. By linking technological advances in chemical imaging with a science-based approach to using these new capabilities, it is likely that fundamental breakthroughs in our understanding of basic chemical processes in biology, the environment, and human creations will be achieved.