Analytical and Bioanalytical Chemistry

, Volume 400, Issue 6, pp 1637–1644

Trends in hard X-ray fluorescence mapping: environmental applications in the age of fast detectors

Authors

    • Centre for Environmental Risk Assessment and RemediationUniversity of South Australia
    • CRC CARE
  • M. D. de Jonge
    • Australian Synchrotron, X-ray Fluorescence Microscopy
  • E. Donner
    • Centre for Environmental Risk Assessment and RemediationUniversity of South Australia
    • CRC CARE
  • C. G. Ryan
    • CSIRO Earth Science and Resource Engineering
  • D. Paterson
    • Australian Synchrotron, X-ray Fluorescence Microscopy
Trends

DOI: 10.1007/s00216-011-4829-2

Cite this article as:
Lombi, E., de Jonge, M.D., Donner, E. et al. Anal Bioanal Chem (2011) 400: 1637. doi:10.1007/s00216-011-4829-2

Abstract

Environmental samples are extremely diverse but share a tendency for heterogeneity and complexity. This heterogeneity poses methodological challenges when investigating biogeochemical processes. In recent years, the development of analytical tools capable of probing element distribution and speciation at the microscale have allowed this challenge to be addressed. Of these available tools, laterally resolved synchrotron techniques such as X-ray fluorescence mapping are key methods for the in situ investigation of micronutrients and inorganic contaminants in environmental samples. This article demonstrates how recent advances in X-ray fluorescence detector technology are bringing new possibilities to environmental research. Fast detectors are helping to circumvent major issues such as X-ray beam damage of hydrated samples, as dwell times during scanning are reduced. They are also helping to reduce temporal beamtime requirements, making particularly time-consuming techniques such as micro X-ray fluorescence (μXRF) tomography increasingly feasible. This article focuses on μXRF mapping of nutrients and metalloids in environmental samples, and suggests that the current divide between mapping and speciation techniques will be increasingly blurred by the development of combined approaches.

https://static-content.springer.com/image/art%3A10.1007%2Fs00216-011-4829-2/MediaObjects/216_2011_4829_Figa_HTML.jpg
Figure

Tricolour maps of elemental distributions in a barley grain: Zn (red), Compton (green) and Mn (blue)

Keywords

Element distributionX-ray fluorescenceTomographyImagingXANESSpeciation

Introduction

Environmental samples are extremely diverse, ranging from living organisms to minerals, soils and sediments. Nevertheless, most environmental specimens share at least one common characteristic in their tendency to be extremely complex and heterogeneous over a wide range of length scales. In organisms, for example, heterogeneity varies from the tens of micrometres (μm) scale when considering different tissues to the sub-μm scale in the case of subcellular features. Similarly, heterogeneity in soils extends over a continuum from the field scale down to that of individual particles in the sub-μm range. This inherent heterogeneity plays a key role in determining the mechanisms controlling biogeochemical processes. Indeed, Hesterberg et al. [1] recently suggested that each soil microsite could represent a “micro-chemical reactor”, with emergent macroscopic processes simply representing the combined result of heterogeneity at the microscale. Yet biogeochemical processes are still frequently investigated at the macroscale even though this approach is increasingly criticised. As early as 2001, Sparks [2] reviewed the limitations of “bulk” approaches for investigating soil processes, whilst more recently, Lobinski et al. [3] stated that a macroscale approach to investigating processes in biological environments is “as frequent as useless” because it ignores both the speciation and cellular distribution of an element.

Although the importance of heterogeneity in controlling biogeochemical processes in environmental systems is now widely accepted, the methodological challenges of addressing this issue remain. For example, in the case of both micronutrients and inorganic contaminants (e.g. metals), target analyte concentrations in environmental samples may be low (e.g. Fe or As concentrations in the low- to sub-mg/kg range); the materials are typically heterogeneous in 3D (e.g. in the case of a soil aggregate or a plant leaf); and element speciation and distribution can be altered during sample preparation (ranging from drying to embedding) as well as by the analytical process itself (e.g., X-ray beam damage). Furthermore, in the case of most biological samples, the highly hydrated nature of the specimens is not compatible with major analytical techniques operating under vacuum. However, in recent years, a number of techniques have been developed for investigating the distribution, and in some cases speciation, of nutrients and contaminants at the microscale (for reviews, see [4, 5]). Among these, laterally resolved synchrotron techniques are rapidly becoming the methods of choice for the investigation of micronutrient and inorganic contaminant distributions and speciation in environmental samples. In particular, laterally resolved micro-X-ray fluorescence spectroscopy (μXRF) and micro-X-ray absorption spectroscopy (μXAS) are arguably the most popular methods for element mapping and speciation, respectively. This article focuses on the μXRF mapping of nutrients and metal(loid)s, but suggests that in the near future, the current divide between mapping and speciation techniques will be blurred by the development of combined approaches.

Capabilities and common limitations of hard X-ray fluorescence mapping

The XRF mapping technique is based on core shell ionization, which is also the foundation for other microspectroscopy approaches such as particle-induced X-ray emission (PIXE) and scanning electron microscopy/energy-dispersive X-ray analysis (SEM-EDX). In this specific case, an X-ray beam is used to excite the core electrons of the atoms in a sample, resulting in the expulsion of photoelectrons. The vacancy left by each expelled photoelectron is filled by an electron from a higher energy level. Fluorescence is emitted to compensate for the difference between the energies of the two electron levels. This fluorescence is thus specific to different electron transitions and elements. When a focused X-ray beam is used, detection and recording of the fluorescence emitted as a sample is scanned through the X-ray beam allows the generation of element maps.

Synchrotron-based hard X-ray μXRF has the following advantages:
  • High lateral resolution (down to a few tens of nanometres)

  • High sensitivity (sub-fg absolute detection limits)

  • Potential for element quantification

  • Suitability of thick specimens, up to 100–200 μm for biological/low-Z matrices

  • Minimal requirements in terms of experimental conditions (e.g. analysis can be conducted at ambient pressure)

  • Multielement information (albeit limited by the energy of the incoming photons, the fluorescence energy, the detector characteristics, and the sample environment)

  • The possibility to obtain 3D information using fluorescence μ-tomography

On the other hand, this technique, as with any method, has its own limitations and drawbacks. In this case, we argue that a substantial proportion of the present limitations result from the long analysis/dwell times required for each single point measurement. Even in its simplest form (i.e., scanning-mode 2D μXRF), a size vs. resolution conundrum is imposed by beamtime availability. In practice an experimenter has to decide whether to investigate an area in detail (using a small step size), or whether to cover a larger area at a lower lateral resolution. This issue becomes increasingly problematic when analysing samples that are highly heterogeneous at the μm or sub-μm level. Yet, these types of samples represent the norm rather than the exception in environmental systems.

Currently, scanning time also tends to limit the information that can be generated by μXRF. By using the tunability of synchrotron radiation (SR), μXRF maps collected at different energies could be used to generate chemical maps. Early examples of this approach have focussed on elements with dominant near-edge features, such as elements with different oxidation states (e.g. [6, 7]). This is because chemical maps can be inferred by collecting information at 2–3 energies. However, when near-edge features are more subtle, the need to collect XRF images at several energies has generally precluded this approach due to the extended acquisition time required. The same issue exists in the case of fluorescence μ-tomography, whereby a sample needs to be scanned at different angles in order to obtain information about the internal elemental distribution (for a review, see [8, 9]). In this case though, it should be noted that interesting alternative approaches based on the use of glass polycapillary half-lenses have been developed. These systems have been used to realize a confocal detection scheme, in fluorescence mode, that is able to provide 3D and internal 2D elemental mapping (e.g. [10, 11]).

Analysis time is also problematic in terms of beam damage. Even with dwell times of 1 second per pixel, various authors have reported beam damage to biological samples [5, 12]. This issue is even more challenging in the case of 3D studies, such as X-ray absorption near edge structure (XANES) imaging and tomography, where the need for repeated scanning means that extensive beam damage is likely.

Beating damage: imaging before the loss of structural/chemical integrity

Room-temperature, hydrated conditions are the goal for most environmental studies attempting to preserve in situ conditions, but as outlined above, these studies must also contend with the effects of beam damage. Inner-shell ionisation and subsequent relaxation can leave chemical bonds frayed, with atoms free to re-establish their binding configuration. Chemical damage can be quite rapid in unstable systems, where the extra energy input may make alternative binding states accessible (for an example of changes in speciation, see [5]). Rearrangement of chemical bonds in the presence of liquid water may also result in significant elemental redistribution. Rescanning of a specimen may thus determine a completely different elemental distribution, with chemical damage followed by elemental migration being the primary underlying mechanism. Measurement under cryogenic conditions has been shown to preserve elemental distribution. Alternatively, where the signal is sufficient, rapid scanning approaches can be used to beat the onset of damage, or to characterise damage in unstable systems. Rapid scanning approaches may also potentially reduce the problems associated with dehydration of non-frozen hydrated specimens during relatively slow scanning. Until recently, sample shrinkage due to dehydration has presented a major barrier to 3D studies of fresh hydrated tissues. In fact, fluorescence tomography of highly hydrated biological materials has only been conducted on freeze-dried samples to date (e.g. [13]).

Fast detection

The quest for low sample damage demands a short dwell time per pixel or a lower beam flux during scanning. However, both of these approaches degrade sensitivity. Optimum detector system design therefore aims for maximum detector efficiency or solid angle, and minimum data acquisition overheads. Such a detector system must be able to handle the high count rates that result from an extended collection solid angle. To address these demands, innovative X-ray detector systems such as the CSIRO-BNL Maia-384 massively parallel detector [14] have been developed.

The Maia detector strategy uses a large detector array (96 or 384 detectors) to achieve a large detector solid angle and high count-rate capacity, and utilizes “on-the-fly” (continuous) scanning and a nuclear physics approach to data acquisition in order to minimize overheads. In this approach, each event is tagged by detector identity and position in the scan, and the event stream is processed in real time in the Maia FPGA-based processor. The result is essentially zero overheads for transit times per pixel down to ∼50 μs and a count rate capacity greater than 10 M/s [1416]. The position that is recorded is now generalised, and can be spatial coordinates for 2D mapping, beam energy for spectroscopy, or specimen angle for tomography. Furthermore, any (and potentially more than one) of these generalised coordinates can be the fast axis, as determined by experimental requirement, measurement stability, equipment performance, or specimen radiation tolerance.

High data rates, zero overheads and short transit times per pixel enable 2D images to be acquired in minutes rather than hours. This makes the collection of 3D data sets practical, such as fluorescence tomography and XANES image stacks collected as a series of 2D images for a sequence of specimen rotation angles and incident beam energies, respectively. Demonstrated 2D imaging to beyond 108 pixels suggests that a similar number of voxels may be acquired for 3D imaging [16].

The Maia FPGA processing approach also enables the dynamic analysis method for synchrotron XRF spectral deconvolution to be applied in real time, thereby producing peak overlap-separated elemental images while scanning the specimen. This provides real-time feedback to the user to help guide an analytical session with an elemental image quality equivalent to linear least-squares fitting of pixel spectra [16, 17].

The Maia detector utilises an annular geometry with the X-ray beam passing through the detector onto the specimen at the normal incidence angle (Fig. 1). While this approach does increase scattered beam intensity relative to a large solid-angle detector located at 90°, and has necessitated care in avoiding charge-sharing between detector elements to maintain high peak-to-background ratios, it enables a large detection solid angle of ∼1.3 sr to be achieved without placing any constraint on specimen size or movement [18]. In practice, large specimens and large XY scanning ranges can be employed to image areas of 100 × 100 mm or even larger.
https://static-content.springer.com/image/art%3A10.1007%2Fs00216-011-4829-2/MediaObjects/216_2011_4829_Fig1_HTML.gif
Fig. 1

Schematic of the Maia-384 detector system novel annular geometry. The X-ray probe beam passes through the detector array and then strikes the sample, which is scanned perpendicular to the probe beam. This geometry enables a large detector solid angle while still affording almost unlimited lateral scanning of the sample

In addition to the development of the Maia detector, the present limitations resulting from the use of slow pixel rate data acquisition systems are being tackled by a number of research groups. For example, Rivers [19] has written an EPICS driver for the digital X-ray spectroscopy modules from XIA LLC (see http://www.xia.com). For the xMAP and Mercury modules, this driver provides fast access to double-buffered memory, allowing the collection of complete spectra with millisecond dwell times in “mapping mode”. It can also collect regions of interest (ROI) mapping data with a 100-microsecond dwell time, and implement list (event) mode with sub-microsecond time resolution. These developments are so recent that we are not aware of any related publications. The mapping mode appears compatible with existing X-ray detectors, and so it seems likely that the back-end electronics will be upgraded by most of the community, and that present-generation detectors will thus be retained. However, these existing detectors use the traditional 90°, limited solid-angle geometry, and often do not have sufficient count-rate capabilities to capitalise fully on these pixel rates (1 ms dwell, 250 kHz count rate on a SDD detector is 250 photons). A transmission geometry, four-element silicon drift-diode detector has been developed by Bruker with a high solid angle (1.1. sr) and a throughput of around 1 MHz. We anticipate that such customised multi-element detectors will be required before ultimate pixel rate and sensitivity is realised with biological and environmental specimens. We note, however, that such detectors have not yet been demonstrated with X-ray probes.

New possibilities

The following examples demonstrate how the opportunities offered by fast detection systems can be used to overcome some of the challenges associated with hard X-ray fluorescence analysis of environmental samples.

Megapixel μXRF mapping of heterogeneous environmental samples

Heterogeneity in environmental samples can result from numerous different processes. For instance, in the case of biological samples, heterogeneity is partly due to genetically controlled developmental processes that lead to the differentiation of tissues or subcellular compartments. In the case of soil, sediments or minerals, heterogeneity is the result of biogeochemical reactions operating on mixed constituents over long time scales. Irrespective of the mechanism driving heterogeneity, the overall result is a system complexity that requires the acquisition of large datasets in order to be properly understood. With regards to element distribution and association, it is obvious that the development of fast acquisition systems will go a long way toward enabling such datasets to be collected within reasonable experimental timeframes. Some examples of research utilising the Maia detector are provided below.
  1. 1)
    Micronutrient distribution in cereal grains. Cereals are the world’s staple food, yet they are often poor in micronutrients such as Fe, Zn and Se. Over the last decade a large research effort has been dedicated to solving this problem through agronomic, breeding and biotechnology approaches (e.g. [20]). Yet, our general understanding of the distribution of micronutrients in cereal grains is still extremely limited. This may be due to the analytical difficulties of mapping elements present in the low mg/kg range in samples of several mm2. Until very recently, the most detailed micronutrient maps available for whole cereal grains had a lateral resolution of 10–15 μm [21, 22]. In addition to the relatively coarse lateral resolution, most studies greatly undersampled the grains, as changing the size of the X-ray beam is often a time-consuming process. For instance, a beam of approximately 1.1 μm2 was used to map rice grains with a step size of 10–50 μm [22]. This results in an effective mapping of only 1–0.4% of the sample. With the advent of fast detection, these issues of low resolution and undersampling have been overcome. A high-resolution elemental map of a thin section of a barley grain is presented in Fig. 2. This 18.4 megapixel image (8.3 × 3.3 mm collected at 1.25 μm lateral resolution) was collected in approximately 3 h using the Maia-96 detector [23]. Importantly, 100% of the sample surface was probed, as the size of the beam was the same as the pixel size.
    https://static-content.springer.com/image/art%3A10.1007%2Fs00216-011-4829-2/MediaObjects/216_2011_4829_Fig2_HTML.jpg
    Fig. 2

    Tricolour maps of elemental distributions in a barley grain (replotted from [23]). Top: Fe (red), Cu (green) and Zn (blue). Bottom: K (red), Ca (green) and Mn (blue)

     
  2. 2)

    Mapping elemental microparticles. Precious elements such as gold (Au) often occur as microparticles finely dispersed among other minerals. An understanding of ore body formation requires knowledge about the distribution of these particles in relation to subtle textural features. However, this is a challenging task, as finding these particles is like looking for the proverbial “needle in a haystack”. Recently, Ryan et al. [16] have demonstrated how the use of the Maia-96 detector enabled the distribution of Au microparticles (estimated to be between 0.3 and 0.5 μm in diameter) to be viewed in relation to textural features at the scale of mm to cm.

     
  3. 3)

    Contaminant distribution in reclaimed waste materials. Contaminants present in wastes are of major concern in relation to their management/disposal. This is particularly the case for biosolids, the product of wastewater treatment. These materials are rich in organic matter and plant nutrients, and are therefore potentially valuable in an agricultural context, but the presence of metals such as Cu, Zn and Cd in them creates concern. One fundamental question relates to the long-term bioavailability of the metals in biosolids. This is largely controlled by whether they are associated with organic matter, which decomposes over time, or with more stable constituents such as Fe oxides. So far, the only published synchrotron-based paper [24] on the subject showed two elemental maps suggesting the association of metals with Fe oxides. However, these maps only covered areas of up to 400 × 400 μm (mapped in 10 μm steps). Recently, Donner et al. [Donner E, Howard D, de Jonge M, Paterson D, Naidu R, Lombi E (2011) X-ray absorption and micro X-ray fluorescence investigation of copper and zinc speciation in biosolids, submitted] collected elemental distribution maps for six biosolids over an area of 4×4 mm with a lateral resolution of 2 μm. These large maps revealed that metal distribution is highly heterogeneous and that potentially toxic metals are not consistently associated with Fe oxides. This finding has significant repercussions in terms of risk assessment and can provide information regarding possible modifications of wastewater treatment processes.

     

Chemical speciation mapping or XANES “stack” imaging

XANES is a powerful technique for chemical microanalysis that can be used to determine the chemical speciation or valence state of an element of interest. XANES can be sensitive to charge transfer, orbital occupancy, and site symmetry. In the soft X-ray region, scanning transmission X-ray microscopy (STXM) is used in an approach called XANES “stack” imaging, as a stack of repeated 2D scans is acquired. Changing the incident photon energy and taking images with other photon energies gives an image sequence that includes both chemical information and topographical information. The third dimension becomes the XANES spectra, and each spatial pixel in the image contains a XANES spectrum and, potentially, spatially resolved chemical speciation information.

In the hard X-ray region, the fluorescence signal is often used when acquiring XANES spectra. This can have several advantages. A principal advantage is that the element of interest does not need to be a major constituent but can be embedded in a complex matrix of other major and minor elements. Trace elements may thus be analysed. However, rastering the image at multiple energies to create the XANES stacks takes considerable time, and the limitation on acquisition times in traditional detector schemes has meant that, until recently, XANES spectra have generally been restricted to point spectra taken at a few locations [25]. Fast fluorescence detection schemes such as the Maia overcome this dwell time restriction and have enabled XANES imaging and other 3D studies to be conducted in realistic time frames.

XANES imaging using the fluorescence signal has recently been demonstrated by mapping the arsenic oxidation state in two geological samples [26]. In this work, XANES imaging of thin sections of an oxidized pisolitic regolith and a metamorphosed, sedimentary exhalative Mn-Fe ore were conducted. The area imaged on the regolith sample was 2.0 × 4.5 mm2 with a pixel size of 2.5 × 2.5 μm2. Each image in the 35-incident-energy stack series took ∼18 mins to collect. A similar sized area was scanned for the Mn-Fe ore. The oxidation state of As in both samples examined proved to be much more complex than expected, with reduced As identified in both samples despite the fact that they represent oxidized environments. The XANES imaging approach has also been successfully demonstrated on biological samples with biologically relevant concentrations of the element of interest. For example, the chemical speciation of platinum in tumour sections has been mapped very recently (personal communication from Prof. Trevor Hambley, University of Sydney's School of Chemistry).

XANES imaging allows the exploration of subtle gradations, boundaries and transition zones in the chemical speciation of heterogeneous samples. An imaging approach can also ensure that entire regions of interest are sampled. This is preferable than relying on a restricted number of point spectra (usually collected at “hot spots”) to adequately represent the diversity of chemical speciation. Binning of spatial pixels can also be performed post-measurement to improve the statistical quality of low-concentration regions. A further advantage of XANES imaging is that other elements (typically lower Z) can still be mapped in parallel, providing valuable elemental correlation information. For example, while mapping As speciation around the As K edge (11.867 keV) it is also usual to map many of the lower Z elements (e.g. Ca to Ge) and potentially some higher Z elements (e.g. Pt) as well. A final potential advantage of the XANES imaging approach centres on avoiding photoreduction and reducing the possibility of radiation damage. As noted previously, dwell times per pixel can be sub-ms, meaning that although the total integrated dose to a scanned area may be large, the per pixel dose can be constrained to acceptable levels.

Micro X-ray fluorescence tomography

A suite of techniques widely referred to as “tomography” promise nondestructive imaging of internal elemental distributions [8, 9, 27]. Tomographic data is usually obtained by recording a “rotation series” of two-dimensional projections with the specimen presented to the beam at a number of angles. A variety of reconstruction algorithms can interpret the measured projections [28]. Despite the great benefit that would derive from internal element maps, μXRF tomography has not been widely realised due to the slow operation and poor efficiency of present-generation fluorescence detectors [9]. To illustrate, a CT scan of a 100-μm cube at 1 μm resolution requires the measurement of around one million pixels; if limited to a 1-s dwell time, this requires about 12 days, delivering an immense radiation dose to the specimen in the process. However, with fast fluorescence detection, such a dataset could be acquired in only 15 min using a 1-ms dwell time. Here we explore some of the consequences of this new imaging domain for X-ray fluorescence tomography of environmental specimens.

Every projection in a rotation series must be consistent, representing the same distribution of material but seen from a different angle. However, radiation damage can result in specimen evolution and so may invalidate this assumption. In particular, highly hydrated specimens such as young roots may be very radiation intolerant, showing the effects of exposure with very low dose. Using fast fluorescence detectors, a large detection solid angle can be used to optimize the collection of fluorescence photons and measurement rates can be tuned to beat the onset of damage for individual specimens, enabling the measurement of an entire rotation series before radiation damage is manifest.

Is there a limit to how quickly one might record a rotation series? The Maia detector has demonstrated 20 kHz pixel rates, and in principle extension into the MHz domain is feasible [14]. However, the signal-limiting step seems to be the count rate capabilities of fluorescence detectors, which typically lies somewhere between 50 k/s and 10 M/s in the case of Maia. In the age of the fast detector with zero readout overheads, emphasis should be given to large voxel counts to best locate all events in 3D space. Re-binning can be used post-analysis to trade off spatial resolution for elemental sensitivity, with the balance being determined by the limitations of the dose fractionation theorem [29].

Recent applications of μXRF tomography for biological/environmental specimens [27, 30, 31] using existing detector technology all indicate that time was a limiting factor in the measurement. Figure 3 shows recent results of de Jonge et al. [27] for the measurement of a 10 μm diatom Cyclotella meneghiniana. 3-D spatial resolution has been used to probe the heterogeneity of elemental correlations in detail, with quantitative analysis leading to chemical insight. Twenty-four two-dimensional projections were measured with a step-size of 150 nm using an X-ray beam of 270 nm; around 250,000 fluorescence spectra were recorded in total. The X-ray dose delivered to the specimen was around 5 × 107 Gy, and the total measurement time was around 35 hours. While this time is prohibitive, the main component of this (12 hours) was due to the use of a prototype setup with poorly-optimised mechanics. Of the remaining 23 h, only 7 h were used for X-ray fluorescence collection (dwell); 8 h were required for detector readout, and a further 8 h for stage motion. Thus, the measurement time could be reduced to 7 h if an optimised system were to be used with a fast data acquisition system. Use of a high solid-angle detector could further reduce measurement times by a factor of 10 to around 36 min without compromising the measurement accuracy.
https://static-content.springer.com/image/art%3A10.1007%2Fs00216-011-4829-2/MediaObjects/216_2011_4829_Fig3_HTML.gif
Fig. 3

3-D tomographic reconstruction of the freshwater diatom Cyclotella meneghiniana, clearly showing distributions of elements between Si and Zn with a resolution of below 400 nm [from 27]

Outlook

The imaging approach using the Maia detector system has demonstrated that with the benefit of zero overhead delays, images can be acquired at the limiting spatial resolution (as dictated by the beam size). Then, post data reduction, individual elemental images may be re-binned to trade off sensitivity against spatial resolution where required. These trace element images can then be located in a very detailed spatial context with reference to the major element images that retain the full spatial detail. The same approach is being taken for the 3D imaging mode of XANES imaging and fluorescence tomography. Future development is aimed at software and processing approaches to dealing with the necessary data-set sizes, which exceed 108–9 pixel or voxel elements.

Environmental applications of hard X-ray fluorescence mapping are set to grow as the advent of fast detectors facilitates more regular deployment of powerful mapping techniques such as XANES imaging and μXRF tomography. Indeed, it is likely that the current divide between mapping and speciation techniques will be increasingly blurred by the development of combined approaches. Greater integration of molecular biology studies with XRF techniques, as in the case of Kim et al. [30], can also be envisaged. With faster scanning and less beam damage, it is not only possible to investigate more hydrated samples, but also to re-image the same samples and thus investigate in vivo the temporal variation in dynamic processes such as functional gene expression and its relation to element speciation.

Another area that is likely to benefit from these developments is environmental risk assessments of manufactured nanoparticles. As occurs with many products of human activity, the intentional or accidental release of nanoparticles into the environment is largely unavoidable (e.g. [32]). In response to this increasing cause for concern, the scientific community has begun investigating the environmental consequences of nanotechnologies. Due to the small size of the material to be investigated (<100 nm) and their dispersion in natural media, it is envisaged that the use of a nanoprobe will be required. As these investigations represent another version of the “needle in the haystack” problem discussed earlier, we expect that fast detectors will provide a powerful tool in this growing field of environmental science.

Acknowledgements

The authors would like to thank Mark Rivers (University of Chicago) and Stefan Vogt (Argonne National Laboratory) for the discussion of developments at the Advanced Photon Source, Argonne, IL, USA. Some of the research presented here was undertaken on the X-ray fluorescence microscopy beamline at the Australian Synchrotron, Victoria, Australia. Further synchrotron installations of Maia are anticipated; for more information, please contact C.G.R. (chris.ryan@csiro.au). E.L. gratefully acknowledges financial support by FT100100337 (Australian Research Council Future Fellowship).

Copyright information

© Springer-Verlag 2011