Jessica M. Walkera,
Miguel A. Gomez-Gonzalez
a,
Johannes Ihli
b and
Julia E. Parker
*a
aDiamond Light Source, Diamond House, Harwell Science and Innovation Campus, Didcot, Oxfordshire OX11 0DE, UK. E-mail: julia.parker@diamond.ac.uk
bDepartment of Materials, University of Oxford, Parks Road, Oxford, OX1 3PH, UK
First published on 1st May 2025
Advances in X-ray nanoprobe beamlines at synchrotrons across the world present exciting opportunities for rich multimodal imaging of biomineral structures and their formation processes. The combination of techniques provides a sensitive probe of both chemistry and structure, making X-ray nanoprobes an important tool for investigating crystallite growth and orientations, interfaces, and assembly of building blocks into hierarchical structures. A discussion of these capabilities is presented with reference to recent examples using a range of nanoprobe imaging techniques for investigating enamel structure, as well as coccolith properties. Key opportunities for the use of X-ray nanoprobes lie in exploiting the penetrating power and coherence properties of synchrotron X-rays in order to image in situ processes or apply coherent diffractive imaging techniques to obtain higher resolutions. To this end initial results demonstrating the observation of calcium phosphate mineralisation, in a liquid environment, using nano-X-ray fluorescence mapping are presented, and the role of X-ray dose and beam induced effects is considered. Finally novel results from tomographic ptychography imaging of Mytilus edulis mussel shell calcite prisms are discussed, where the segmentation of the phase density into organic and mineral content gives insights into the mechanisms underlying mineral prism formation and the role of the organic matrix in biomineralisation.
Organisms are able to exert precise control over the mineralisation process, controlling the crystal phase and polymorph formed (even switching between polymorphs in a single structure), and the resulting structure and morphologies. It is proposed that this control occurs through; regulating of the chemical environment around cells; mediating ions and precursor availability; controlling pH levels; confinement of processes; and the presence of organic matrices which can act as nucleation and ion binding sites.6,7 Observations of biomineralisation processes, and the identification of stable amorphous precursor phases in biominerals, has contributed to a plethora of discussions about non-classical crystallisation pathways, encompassing prenucleation clusters, orientated attachment and mesocrystallinity.7,8 Hence, there continues to be much interest in imaging mineralisation processes, across multiple lengths and timescales, in order to gain further insights into the underlying mechanisms.
This fine control over mineralisation is often exhibited as exquisite structures and forms, from coccolithophore calcite formation to the spiral of the narwhal tusk, with highly organised hierarchical structures.9 This hierarchical structure contributes to the superior mechanical properties and functionality of biominerals. Understanding the organisation of biominerals, across the different structural levels, is therefore important for a thorough understanding of formation processes, properties, and development of biomimetic approaches for materials development.10
The structural complexity of biominerals presents significant challenges for characterisation. Understanding both structure and formation of these complex materials requires a range of characterisation and imaging techniques across multiple length scales and modalities. Here X-ray imaging can be used to complement other microscopies (such as light, Raman and electron microscopy) and provide a non-destructive means of probing the structure and composition of biominerals.
With focussed X-ray beam sizes of 25–100 nm, combined with the increased spatial resolution achievable through coherent diffractive imaging techniques such as ptychography, the experimental envelope of X-ray nanoprobes bridges the resolutions and fields of view between transmission electron microscopies and microscale techniques, whilst the penetrating power of hard X-rays increases the availability of options for in situ imaging studies. X-ray nanoprobes allow a combination of techniques including nano-X-ray fluorescence (n-XRF), spectroscopy, nano-X-ray diffraction (n-XRD), and advanced phase imaging, which can be applied to biomineralisation studies, revealing details of nanoscale elemental compositions, structures and morphologies.11
A key advantage of using X-rays is the ability to penetrate ‘thicker’ sample environments in order to image in situ processes.12–14 Of particular interest for biomineralisation is the ability to observe crystallisation processes and kinetics. Here, the design challenges of in situ liquid cells are discussed, and initial results demonstrating the observation of calcium phosphate mineralisation in a liquid environment, using n-XRF mapping are presented. These results illustrate the potential advantages and role of in situ nanoprobe techniques for mineralisation studies and highlight key aspects for future development and study. These include the role played by X-ray dose and beam induced effects, an important consideration when designing experimental approaches and interpreting results which has not yet been widely studied for X-ray nanoprobe in situ experiments.
The final factor in exploiting X-ray nanoprobes for biomineral imaging lies in the direction of applying coherent imaging techniques. Tomographic ptychography, a scanning coherent diffractive imaging technique, enables quantitative 3D visualisation of electron density, making it highly suited for exploring the impact of organic–mineral interfaces on structure and properties. Novel results from tomographic ptychography imaging of a Mytilus edulis mussel shell calcite prism are discussed, where the nanoscale segmentation of the phase density into organic and mineral content gives insights into the mechanisms underlying mineral prism formation and the role of the organic matrix in biomineralisation.
Nanoprobes are multi-modal probes, allowing simultaneous imaging using a combination of different techniques, at the same resolution, on the same sample. X-ray fluorescence (XRF) enables the mapping of elemental distributions in a sample, providing information on material compositions and heterogeneity. Phase contrast imaging techniques (such as differential phase contrast imaging, DPC)26 gives information on the sample morphology and density. This is often crucial in biological samples to provide a context for the elemental XRF maps, where lighter elements (including carbon, nitrogen and oxygen) are not observed. Thus a combination of DPC and XRF imaging can provide simultaneous imaging of the cellular structures and the distribution of metal nanoparticles or metallo–organic drugs, for example when investigating treated cancer cells.27,28 X-ray diffraction (XRD) probes the crystalline, atomic structure of the materials, and can be used to fingerprint the presence of different phases, such as mineral distributions in samples or, with a more detailed analysis, map lattice strain variations.
By repeating n-XRF maps at different energies the technique can be extended to give X-ray Absorption Near Edge Structure (XANES) spectroscopy maps (nano-XANES). 2D n-XRF maps are collected on a region at different energies through an elemental absorption edge. The elemental maps extracted from the XRF spectra at each energy are aligned and stacked to give a 3D image stack where each pixel (x, y) has a third coordinate, energy. The nano-XANES can be analysed, typically through principal component and cluster analysis, to group the spectra based on spectral similarities followed by fitting to reference data, to extract amounts and locations of distinct species. Repeating scans for different energies increases substantially the time and X-ray dose required to image a sample, limiting the use of this technique for in situ or biological samples. The X-ray dose and time can be reduced by the application of a variety of approaches, including sparse energy sampling29 or undersampling the raster scans.30 Implementation of these options begins to open the possibility of applying nano-XANES approaches to beam and time sensitive experiments.
The coherent nature of nanoprobe X-rays allows the application and development of coherent diffraction imaging techniques such as ptychography.31 In ptychography, far field diffraction patterns are collected at a series of overlapping scan positions. An iterative phase retrieval algorithm is used to recover the phase and amplitude of the scattered wave, reconstructing both the phase and adsorption image of the sample. Importantly, the image resolution is not limited by the size of the X-ray beam, offering the possibility of achieving higher resolution images than probe-limited techniques such as DPC.
These techniques can be extended to three dimensions through tomography – collecting maps or projections at a series of different angles as the sample is rotated and reconstructing to form a 3D volume. Current challenges for the implementation of tomographic techniques at nanoprobes centre around sample preparation and speed of acquisition. The demands of sample preparation, to produce a 10–20 μm pillar, extracted from a specific region in a larger sample and mounted on a pin, require specialised approaches, such as FIB-SEM, and dedicated time and expertise. Smaller samples are currently required in order to keep the sample in the focus of the beam and reduce the inherent scanning time limitations. In most cases the limitation on scanning speed is not a flux but rather a hardware, controls and detector limitation. However, the desire to extend to larger volume imaging means an increase in scanning speeds is a current development push.32 Overall, combining information from different X-ray modalities, through for example correlative ptychographic tomography and n-XRF, or n-XRF and n-XRD, is a powerful approach for imaging a wide range of materials, from biological cells,33 to catalyst34 and battery35 particles, and historic paint samples.36
The experimental hutch is located in a purpose built external building. At the endstation, Fig. 1(b), the X-ray beam is focused to 50 nm at the sample position (1) using a pair of in vacuum nano focusing KB mirrors. The sample, typically on a silicon nitride membrane window, or TEM grid, is mounted onto a magnetic holder and placed in the sample position (1). The sample spigot is attached via a kinematic mount to the scanning stages (2). The kinematic mounting system, combined with the long working distance focus of the KB optic, and in air environment, provides a flexible space for incorporating bespoke sample mounting solutions and a variety of in situ environments and facilitating easy changeover of experimental setups.
The scanning stage is a novel stage design concept, a Delta Robot.37 The delta robot uses three orthogonal voice coils, which actuate parallelogram flexures for a 3 mm travel range in x, y and z directions. The design enables high speed scanning at nm precision over mm travel ranges, with space and load capacity for flexible setups and sample environments. On top of the delta robot, underneath the kinematic mount, is a rotation stage to enable tomography measurements.
Once mounted a sample is aligned to the beam focus and region of interest found using the retractable sample alignment microscope (3). XRF data are collected in backscatter geometry using a four-element silicon drift detector (4). Downstream of the sample scattered or transmitted X-ray can be collected for DPC, ptychography or XRD analysis using silicon pixelated photon counting detectors; a quad merlin (Quantum Detectors), Excalibur 3M or an Eiger 500K (Dectris). These detectors are mounted on independent translation rails on a movable granite table to facilitate alignment and changeover between detector distances and setup – e.g. for ease of switching between DPC and XRD data collections in one experiment. Setup up of scanning regions (maps), and viewing of live data collections, is facilitated through an intuitive interface to the beamline controls system using a mapping GUI in the data acquisition system.
The capability of X-ray nanoprobe beamlines to collect correlative data across different modalities, across three dimensional and in situ experiments, and developments in faster scanning, present challenges in data processing and analysis. I14 has therefore placed a great deal of emphasis on ease of experiment and automated data processing pipelines, to enable live processing on Diamond's high performance computing cluster during a scan of; elemental maps from XRF spectra; processing of DPC data to give a quantitative phase image; stacking and alignment of XANES data; integration of 2D XRD images to 1D patterns and automated submission of ptychography reconstructions. This is complemented by offline processing tools using a web GUI for workflow submissions.
X-ray nanoprobe measurements at I14 were combined with soft X-ray ptychography, electron microscopy and larger scale X-ray tomography to investigate the chemical and structural properties of carious enamel.38,39 The hierarchical structure of human dental enamel has orientated hydroxyapatite (Hap) crystals arranged into rod and interrod regions where the nanoscale arrangement contributes to the mechanical strength of the enamel.40 Demineralization during caries formation leads to chemical and structural changes which reduce the stiffness and strength of the enamel. Besnard et al.39 investigated these changes in FIB-lamellae extracted from locations in tooth enamel corresponding to normal and demineralised areas. Using a I14, n-XRF mapping revealed calcium chemical gradients across the rod and interrod areas corresponding to different demineralization rates. The chemical information was correlated with crystal orientations extracted from n-XRD measurements and structural changes observed in DPC. Higher resolution images of the same sample using soft X-ray tomography (at Diamond's I08 beamline) allowed alignment of the n-XRF maps to start to reveal details of crystallite orientation gradients within the enamel.
Coccolithophores play an important role in global biogeochemical cycles.41 They produce a calcium carbonate exoskeleton, composed of coccoliths. Coccoliths are arrays of nanoscale crystals that form intracellularly, a key example of the fine control biomineralising organisms exhibit over the chemistry and morphology of their mineral phases. Understanding their formation necessitates information on elemental distribution and speciation at a nanoscale level. n-XRF tomography of a coccolith (the lopadoliths of Scyphosphaera apsteinii) at I14 revealed an uneven distribution of strontium,42 with stripes of different concentrations, as shown in Fig. 2. This is in contrast with current Sr fractionation models which predict an even distribution. The n-XRF tomography was complemented by nano-XANES analysis of the same sample to show that Sr resides in a Ca site in the calcite lattice in both high and low Sr stripes, confirming a central assumption of current Sr fractionation models.
![]() | ||
Fig. 2 3D volume renderings of the reconstructed 3D XRF data for (a) calcium and (b) strontium signals individually and (c) combined (calcium signal in red, strontium signal in green), showing both the full image of the lopadolith and a cut through along the growth direction. The corresponding reconstruction slices are also shown (d). Reproduced from ref. 42 with permission from the Royal Society of Chemistry. |
Nano-XANES at I14 has also been used, in combination with DPC and nano-XRF, to investigate magnetosome formation at the single-cell level in magnetotactic bacteria (MTB).43 MTB synthesize single-domain magnetic nanoparticles composed of magnetite (Fe3O4) or greigite (Fe3S4) within organelles known as magnetosomes, yet the mechanisms of biomineralization are still unclear. Recent work suggests a large fraction of intracellular iron, at least partly composed of ferrous species, is not incorporated into magnetosomes.44 Wild-type M. gryphiswaldense MSR-1 bacteria and a genetic variant lacking ferritin proteins were studied under varied iron concentrations and at different stages of magnetosome formation. DPC and nano-XRF mapping was first used to reconstruct both organic and inorganic components of bacterial cells to identify intracellular regions. Fe K-edge nano-XANES was then used to differentiate magnetosome particles from other intracellular iron species and determine their relative amounts, Fig. 3, confirming a significant presence of intracellular iron species distinct from magnetite during biomineralization.
![]() | ||
Fig. 3 Nano-XANES data for MTB with complete magnetosome chains. (A and F) WT-high-iron and WT-low-iron nano-XANES regions (white box), (B and G) PCA maps (red signal), (C and H) maps of cluster centers, (D and I) normalized offset Fe K-edge XANES spectra (inset with normalized spectra retaining edge jump values), and (E and J) a summarized composite of LC fitting results. Reproduced from ref. 43 under creative commons CC BY 4.0 licence. |
Repeat n-XRF maps were collected on beamline I14, Diamond Light Source16 using a 50 nm focused X-ray beam at an energy of 12 keV. XRF spectra were collected using a 4 element silicon drift detector in backscatter geometry. Ca elemental maps were produced by integrating the intensity in the XRF spectra over a 200 eV window around the Ca-Kα fluorescence peak (3.6 keV). The P signal can be mapped from the same spectra, however due to the lower X-ray energy and the absorption by the liquid and air path the signal is much weaker and so is not presented here, although the distributions match those of Ca.
![]() | ||
Fig. 5 Ca-Kα fluorescence maps extracted from n-XRF spectra at different time points during in situ CaP crystallisation in a liquid cell. Scale bar = 500 nm. |
![]() | ||
Fig. 6 (a) Ca-Kα XRF map indicating the location of Region 1 (white box) and Region 2 (red box) (b) the total Ca signal in Region 1, Region 2 and the whole mapped area as a function of time. |
The total Ca signal intensity in the entire mapped area decreased over time and we note that after 24 hours, the formed crystals in the field of view dissolved, and other larger crystals were found nearby when scanning nearby areas – raising the question of the role of beam effects in the dissolution process. Existing models of X-ray dose effects on liquids suggest that X-ray exposure can contribute to radiolysis of the solution, and generate radicals which change the local solution pH. Over time – and in the absence of flow – dispersion of these species is diffusion limited, so effects will accumulate.
These initial results can be compared to a TEM in situ liquid cell study of CaP crystallisation.45 As the same solution concentrations were used, similar kinetics might therefore be expected in the absence of beam induced effects and any confinement effects due to the different cell geometries. Wang et al. observed branched morphologies after 1 hour, these appear to be very similar in nature to the structures observed here.45 It is clear however that in situ TEM offers improved resolution, both spatial and temporal, above these initial nanoprobe measurements for imaging biomineral formation mechanisms.46–49 However, we suggest that they can be complementary in a multiscale approach; access to larger fields of view, as well as the chemical information that can be extracted, is important for examining at the mesoscale the assembly into hierarchically ordered structures. The greater flexibility offered in terms of sample environments opens interesting avenues in terms of customised microfluidic cells for controlling crystallisation, and bridging the gap between TEM and microscale microscopy techniques.
In order to progress from these initial results an improved understanding, and mitigation, of beam effects is required, building on ongoing modelling for electron dose effects including consideration of scanning paths, damage models and propagation. Improvements in speed can also be achieved by various routes, for example by continued application and development of sparse sampling approaches.29,30
Ptychographic X-ray tomography experiments were carried out at the cSAXS beamline of the Swiss Light Source.53 The photon energy was 6.2 keV. A Fresnel zone plate 220 μm in diameter with an outermost zone width of 60 nm was coherently illuminated, and the sample placed 1.2 mm downstream of the zone plate focal point. The sample was mounted in the in vacuum cryo-ONMY scanning instrument at −180 °C and scanned through the X-ray beam with a Fermat spiral covering 20 × 10 μm with a step size of 1.3 μm. Far-field coherent diffraction patterns were acquired with exposure times of 0.1 s per point using a Pilatus detector with 172 μm pixel size at a sample-detector distance of 7.342 m. A flight tube between sample and detector was used to reduce air scatter and absorption effects. 1000 angular tomography projections were acquired.
Ptychographic reconstructions were obtained using the difference map algorithm,54 with maximum-likelihood refinement.55 Tomographic reconstruction of the phase images was carried out following alignment using a modified filtered back-projection algorithm as described by Guizar-Sicairos.55
The X-ray dose imparted to a prism sample during tomogram acquisition was estimated to be on the order of 108–109 Gy. The estimated dose is based on the average area flux density of each scan and the assumed mass density of the specimen, assuming a composition of calcite and chitin.
The half-period spatial resolution of ptychographic tomograms was estimated by Fourier shell correlation (FSC). The threshold criteria for the FSC was the ½ bit criteria. From this analysis the resolution is estimated to be 75 nm.
![]() | ||
Fig. 7 SEM image of calcite prisms from M. edulis after dissolution of interprismatic organic matrix. Scale bar 10 um. |
Images from the ptychographic tomography reconstruction of the Mytilus edulis calcite prism are shown in Fig. 8(a)–(d). What was assumed to be a single prism here when mounting is in fact multiple smaller prisms, with their longitudinal axis co-aligned with the long axis of the larger needle-shaped prism being imaged. In contrast to micro- X-ray CT results examining the grain growth of Pinna nobilis calcite prisms,56 where each prism was observed to be a regular polygonal shape, these individual smaller prisms are much more irregular, Fig. 8(c) and (d). This may be due to interspecies variation (Pinna nobilis prisms are much larger than those of Mytilus edulis), the region of the shell from which the prisms were extracted (e.g. fresh growth vs. fully formed crystals) or the differing resolutions and length scales accessible via the different imaging techniques.
The 3D volume was segmented to separate the intensity based on the outermost 100 nm of each prism (Fig. 8(a) and (b)) compared to the internal prism structure. This reveals the intricate web of organic matrix around each individual prism, in agreement with electron microscopy studies that have imaged these prismatic envelopes in Pinna nobilis and Atrina rigida prismatic layers.57,58 Here we are able to observe them in 3D without prior dissolution of the mineral component, this enables visualisation of the close link between organic and mineral components at the outer regions of the prisms.
An advantage of ptychography is that the measured intensity corresponds to a quantitative electron density, this is shown for the internal prism and outermost 100 nm in the histogram in Fig. 8(e). Comparing to the known electron densities for chitin and calcite suggest the outermost layer is a good match to the electron density of chitin. The mineral phase of the prisms shows however a lower electron density than calcite, which is presumably due to differences in crystallinity and packing density, hydration and the presence of water as well as organics in the crystal structure.59
The electron density can be plotted with distance from the centre of a prism, as an average across the prisms imaged, Fig. 8(d). In addition to the higher organic content at the outer 100 nm, it can be seen that the inner 300 nm core of the prisms has a much lower electron density. The organic content of each voxel can be estimated using the known electron densities; segmentation based on 55–75 vol% organic is shown in Fig. 8(e) and (f). The presence of an organic rich core running through the centre of each prism can be clearly seen.
The observations of an organic web-like prismatic envelope, and a central organic core is suggestive of a formation mechanism in which calcite nucleation occurs within a dense organic fluid precursor phase, with the crystal growing to fill the space defined by the prismatic envelope, resulting in a calcite prism with high incorporated organic content. This is in broad agreement with existing results observing the organic web and a dense intra-crystalline network of chitin within Atrina rigida calcite prisms.58,60
Tomographic ptychography has provided novel insights into the formation mechanisms of Mytilus edulis calcite prisms, revealing the distinct contributions of organic and mineral components in biomineralization. These findings underscore the potential role of X-ray nanoprobes in biomaterials research, paving the way for future studies that exploit higher resolution imaging and in situ methodologies to deepen our understanding of biomineral formation processes.
This journal is © The Royal Society of Chemistry 2025 |