Marc-Georg
Willinger
*a,
Julien
Polleux
c,
Markus
Antonietti
*c,
Helmut
Cölfen
ce,
Nicola
Pinna
b and
Nadine
Nassif
*cd
aFritz Haber Institute of the Max Planck Society, Department of Inorganic Chemistry, Faradayweg 4–6, 14195 Berlin, Germany. E-mail: willinger@fhi-berlin.mpg.de
bHumboldt-Universität zu Berlin, Institut für Chemie, Brook-Taylor-Str. 2, 12489 Berlin, Germany
cMax-Planck Institute of Colloids and Interfaces, Colloid Chemistry, Research Campus Golm, D-14424 Potsdam, Germany. E-mail: nadine.nassif@upmc.fr
dLaboratoire Chimie de la Matière Condensée de Paris, UMR 7574 CNRS, UPMC, Collège de France, 11 place Marcellin Berthelot, 75231 Paris Cedex 05, France
eUniversität Konstanz, Physikalische Chemie, Universitätsstr. 10, D-78457 Konstanz, Germany
First published on 9th March 2015
The effect of amphoteric siderophore deferoxamine on the crystallization behavior of calcium carbonate was investigated under bioinspired conditions. Amphoteric siderophore deferoxamine possesses self-organization ability, surface activity and ion-chelating properties. It induces in the present case the formation of unusual, highly organized aragonite mesocrystals during the precipitation of calcium carbonate from liquid phase. A detailed investigation of the structure and growth of the particles provides an insight into the role of the deferoxamine and its function in the crystallization process.
Especially, aragonite attracts much interest mainly due to its presence in nacre.10 Nevertheless, the described routes for aragonite synthesis depend on an intricate set-up or well-defined reaction conditions. So far, aragonite was synthesized under high concentration of Mg2+ ions,11,12 with additives such as extracted glycoproteins,13 which are associated with a β-chitin-silk matrix,14 and specific imidazolium-based ionic liquids,15 or under harsh conditions such as elevated temperatures,16,17 high-power ultrasonic18 or microwave irradiation.19 Moreover, aragonite could be synthesized at the air–water interface under compressed monolayers7,20via an amorphous precursor route,21 at the liquid–liquid interface in a radial Hele-Shaw cell22 or by using block copolymer microgels23,24 where aragonite bundles are obtained. A synthesis in alcoholic medium in the presence of sodium stearate was also proposed to obtain lamellar aragonite.25
In this work, a novel model system for complex morphosynthesis is presented. More specifically, the influence of deferoxamine on the crystallization behavior of calcium carbonate is reported. Deferoxamine belongs to the family of siderophores, a class of low-molecular-weight, iron-coordinating agents produced by plants and microbial organisms. It is clinically used for the treatment of iron overload disease.26 With its three hydroxamic acid groups (RCONR′OH) interspaced by two amide groups and terminated with an amine, this biomolecule is a chelating agent27 with a rich coordination chemistry.28 Applied to the nonaqueous synthesis of metal oxides, deferoxamine can also provide intra- and intermolecular amide–amide interactions with the potential for supramolecular hybrid nanostructure formation.29 Our intention to use deferoxamine was to set a model system for a siderophore that is able to bind weakly and reversibly to Ca2+ providing high local availability of non-clustered Ca2+, which is an interesting in the frame of natural mineralization processes.
Scanning electron microscopy (SEM) observations of the formed calcium carbonate collected after 24 hours from the bottom of the reaction flasks revealed intricate superstructures of spherical shape as the predominant morphology. In agreement with the XRD results, a few crystals with a rhombohedral morphology were observed (Fig. 1a) as a minority phase. Their morphology is typical for calcium carbonate that precipitates in the absence of additives (Fig. S1†). This suggests that some crystals grow from the continuous phase without being controlled by deferoxamine.
At higher magnification, the SEM images reveal further details of the peculiar crystal superstructure of the obtained particles (Fig. 1b, c and d). Their morphology is characterized by two parts with distinct aspects. The main part is formed by a globular body of 20–50 μm in diameter. It is characterized by striation lines that run from one pole to the other. The other characteristic feature is formed by a bundle of outgrowing structures that generally appear at one of the poles. The length of these outgrowing structures varies from particle to particle but generally reaches a length that is comparable to the diameter of the globular body. Besides a majority of isolated particles, there are some that are connected to each other, either at a common plane or through a complex intergrowth (Fig. 1b, c and S2†).
It is worth noting that predominant aragonite formation was already observed in previous experiments using polytertiary amines.23,24 Interestingly, precipitation performed with suberohydroxamic acid, made of two hydroxamic acid groups and without amines, does not lead to the formation of unusual crystal morphologies, and instead of spherical aggregates, only the default event of rhombohedral calcite is observed (Fig. S3†). In order to exclude the influence of the counter-anion present in the commercially available deferoxamine in the crystal formation, a control experiment was performed in the presence of methanesulfonic acid. No characteristic changes were observed in the morphology compared to free crystallization, and mainly classical calcite rhombohedra were obtained (few crystals possess a central hole on one face) (Fig. S4†).
After 24 hours, the rough surface structure of the lamella sidewalls has completely disappeared, and crystallographically flat facets have developed. They show numerous gaps and discontinuities (Fig. 2e and f).
The morphology of the globular body differs from the columnar structures that grow at the poles. Fig. 2g and h show the aspect of these structures in more detail. Each individual outgrowth appears as a stack of thin platelets or pillars that are aligned and oriented towards the tip of the outgrowth. The diameter of each individual outgrowth shrinks through a progressive termination of the outmost crystallites at their growth front. Stepwise, deeper lying platelets are exposed until the axis is reached at the tip of the outgrowth. Looking at the tips of the outgrowing structures from above reveals a well-defined six-fold symmetry (Fig. 2h). Similar pseudo-hexagonal crystals were previously obtained using lithium niobate as a carbonate free single-crystal substrate. In that case, epitaxial aragonite growth occurred through the arrangement of cations on the substrate.30
The primary crystal seed is known to play a very important role in the determination of whether calcite or aragonite forms, and the polymorph can even be encoded in an amorphous CaCO3 precursor phase.32 Indeed, our data shows that once the growth of a calcite particle is initiated, it would just keep on growing while ignoring the presence of the siderophore. It is probable that siderophore molecules interact differently with the growing structures in a time-dependent manner. Indeed, the total charge of deferoxamine at different pH values may define its binding properties and thus control the mode of aragonite growth. During the mineralization protocol, ammonia and CO2 diffuse into the CaCl2 solution, leading to an increase in pH from 5.8 to 9.5 throughout the crystallization process. With dissociation constants equal to pH 8.30, 9.00, 9.46 and 10.4 that correspond to that of the three hydroxamic acids and the amine group, respectively,31 deferoxamine is positively charged at pH 5.8 and may electrostatically bind to hydrogen carbonate (HCO3−) via the ammonium group but not to calcium(II) at the beginning of the crystallization process. During the increase in pH, although the total charge of deferoxamine becomes neutral and then negative, the monochelation of calcium(II) to this siderophore remains rather weak, as the stability constant of the complex is not significantly changed.31
In spite of the rather high deferoxamine concentrations, it may come as a surprise that the organic content of the final structures is very low, as determined by TGA. This just means that the siderophore is only active as a growth modifier, but is effectively released at the final pH and within the final structure. This low consumption makes such a modifier also useful for technical mineralization recipes.
Putting the crushed pieces in the context of complete particles was, however, quite challenging and could only be achieved by analysing distorted growth events. Some of the particles that nucleated at the wall of the reaction flask grew such that the glass cut a plane through their center. Particles “cut” in this manner reveal their inner structure and allow the study of changes in the growth modes from the center to the surface. In Fig. 4, two particles are shown revealing their equatorial and longitudinal inner structures, respectively.
Fig. 4 SEM images recorded form two particles in which the nucleation at the wall of the flask led to a longitudinal and equatorial cut, respectively. Crystallographic directions are indicated in d). |
Interestingly, in both cases, the shape of the final particle seems to be unaffected by the presence of the glass substrate on which they grew. Indeed, the usual morphology is obtained, such as if the particles were cut after growth. The presence of both longitudinally and equatorially cut particles at the glass surface could be a hint on the role that the inner, central region of the particles plays in defining the growth directions and crystallographic arrangement of crystalline domains of the complex shape and thus the final shape. In the very center, the nanoparticle aggregate does not show any signs of pre-determined directionality or anisotropy. From the center to the outside, the structural density gradually increases and orientation relationships emerge. This is clearly shown in Fig. 4b. The aspect of the central region, marked as square I in Fig. 4b, is similar to that of the agglomeration shown in Fig. 3a. Further away from the center, in the region marked as square II, a pillar-like, dendritic structure, similar to that in Fig. 3c, can be identified. Hence, the crushed samples shown in Fig. 3 are indeed from the inner volume and resemble the first growth stages of the particles. Formation of the particles therefore starts with a loose agglomeration of nanoparticles at the center. It is followed by the formation of nanoparticulate, dendritic structures that radiate outwards. This is the region where first signs of local crystallographic alignment can be observed (Fig. 3b), presumably following the so-called oriented attachment. It can finally result in the formation of a so-called mesocrystal, which however cannot be proven in the present case.33
As ordinary aragonite nanoparticles hardly show oriented attachment, we assume that the nanoparticles observed at this growth stage are still stabilized by the siderophore, locally liberated throughout the consumption of Ca2+ by the precipitating nanoparticles. Nucleation and attachment of further nanoparticles lead to further densification and formation of extended crystallographic coherent domains and the formation of sheet-like crystals growing radially in the [001] direction. A possible underlying concept for the orientation alignment is crystal growth competition,34,35 according to which particles with a fast growing direction (here, the [001]) pointing towards the growth front successfully compete and dominate over differently oriented particles. The principle is schematically illustrated in Fig. 5. Once the fast growing direction is determined, the local alignment of the remaining crystallographic axes follows under the constraint of confined space. The radial symmetry is then a product of all the crystal slices growing radially outwards in the [001] direction. With increasing density and probably also with a change in the pH and a drop in Ca concentration, the growth mode changes.
Surface energy minimization then results in the formation of joint, extended crystallographic facets terminating the superstructure. The transformation of the rough surface observed at an earlier growth stage into flat crystallographic facets can carefully be followed by comparing Fig. 2d with f.
Observation of the striation lines at the surface of the globular body by high resolution SEM (Fig. 6a) reveals numerous growth discontinuities and a nanoscale surface texture, being the memory of the primary nanoparticulate building blocks. However, the densely packed domains terminate with common planes, similar to what is expected for the case of a single crystal. Indeed, the single crystal-like behaviour is confirmed by high resolution TEM imaging. The image shown in Fig. 6b reveals continuous lattice fringes and an agreement in crystallographic orientation even across gaps. Thus, the behaviour is very similar to the case of nacre, in which the individual aragonite platelets diffract like single crystals although they are made up of densely packed nanocrystalline units.36
Another interesting point to note here is the presence of an amorphous surface layer, similar to what was described by Nassif et al. for the case of nacre37 and synthetic aragonite.23 The current, repeated observation indicates that the amorphous surface layer is a natural feature of aragonite crystals grown from water in the equilibrium state.
For a deeper understanding of the growth of the particles and their final shape, the relationship of the crystal orientation of aragonite to the particulate morphology has to be revealed.
In order to do so, detailed HRTEM imaging was performed along the surface of the globular body of individual particles. The combination of SEM and TEM and observations over a large range of magnifications was crucial for relating information obtained on the atomic scale to the overall structural motif. The SEM overview image and the corresponding TEM images are shown in Fig. 7. The HRTEM investigations and the analysis of lattice fringes conclusively revealed that the particles indeed grow radially along the c axis of aragonite. The striation lines running in the longitudinal direction from one of the poles to the other coincide with the crystalline b axis. Consequently, the remaining a direction runs around the globe in a plane parallel to the equator. Crystallographic orientations are indicated by arrows in Fig. 5c, 6a, b, and 7b.
Once the crystallographic relationships have been revealed, the initial reason for the formation of the outgrowth has to be elaborated. On a sphere, radial growth with an alignment of the crystallographic axis faces some constraints (similar to the “hairy ball theorem”38).
It has already been shown by SEM images recorded at an early growth stage (Fig. 2) that the same lamellas that give rise to the striations at the surface of the globular body are also found at the outgrowth. However, while they are running parallel to the equatorial belt, with their b axis pointing to the longitudinal direction, they are crossing each other at the pole, as indicated in Fig. 9. Aragonite is known to form polycyclic twins at the (110) planes. These conditions are met when [010] or [100] directions cross at 120°. Indeed, outgrowing structures are formed specifically at positions where lamellae cross at 120°. Under such conditions, a twin is formed. This is a straightforward explanation for the six-fold symmetry that is revealed when the outgrowing structures are viewed from the top as shown in Fig. 2h and 9c.
Fig. 9 The image on the left side shows a montage made from the half spheres shown in Fig. 4a and c. Crystallographic directions as well as crystalline domains and striation lines are indicated. Aragonite grows radially from the center, with the c axis pointing outwards and the b axis pointing longitudinally towards the poles. Hence, crystalline domains are oriented parallel to the equatorial belt, but crossing each other in the region close to the poles, as schematically shown in the diagram on the right side. Small pictures show that polycyclic twinning occurs at the poles. |
Therefore, the outgrowing structures are formed as a consequence of the radial growth and crystallographic alignment combined with the ability of aragonite to form twins. As shown in Fig. 4a, one can see that the outgrowth on the left shows its roots deep inside the particle. The first signs of the formation of outgrowing structures are therefore manifested in a relatively early growth stage (i.e., after less than 9 h, Fig. 2a). The corresponding region, which is characterized by intergrowth and twinning, expands faster due to an increased growth speed and hence requires more space. Presumably, the attachment of the growth carrying nanoparticles adds up at the point where lamellae intercept in a preferred angle and where a cyclic twin is formed. This is possibly due to a combination of rather good stabilization of the nanoparticles by the siderophore and field effects which allow the particles in later stages of reaction to discriminate between different positions of aggregation, thus preferring twinning positions where the fields are not compensated.
A strong indication of the presence of field effects is found in view of the fact that the outgrowth is exclusively formed on only one of the poles. As seen in Fig. 5c, aragonite grows preferably along [001], which is the fast growth direction. The {001} faces are, however, highly polar faces and not stabilized by the siderophore. Consequently, the surfaces on the globular body are mainly defined by (101) and some (100) planes (see Fig. 4d). However, the faces of outgrowing structures at the pole mainly constitute (100) planes, defining the thin direction of the aragonite slices (thereby greatly stabilizing and also loosing against [001] in the growth competition). The pole-to-pole direction of the striation lines is in the [010] direction, associated with polar surfaces. As the siderophore binds to carbonate especially at the more acidic pH at the beginning of the crystallization reaction but not or only weakly to calcium ions,31 one of the (010) faces of the primary nanoparticles, the negative side, is hindered in growth causing a dipole field along [010]. If we define the equatorial cut through the superstructure to be along the [100] direction, this gives an exact cycle, i.e. all [100] directions sum up into the same circular direction. The (010) cut is more of an ellipse, and the structure must be twinned (as all the [100] directions are aligned already). Indeed, the “cut through the particle” reveals this pole-to-pole distortion by its lower structural density (see Fig. 4a) along the pole-to-pole axis. This means that all meridians are oriented in the same direction, supported by twinning structures along the axis. By superstructure formation, this sums up to the discussed fields at the poles where the “negative pole” does not outgrow, while the “positive pole” attracts further nanoparticles and leads to the found pronounced structured outgrowth. Such inner field effects were found for CaCO3 before and can lead to very specific superstructures of remarkable beauty.39,40 The fact that the siderophore, contrary to the previously reported cases, binds here, mostly to the (100) and (010) faces of aragonite (instead of the (001) of calcite), is attributed to appropriate charge distances in the surface and ligand and pattern recognition.
Electron microscopy investigations have been performed at the Department of Ceramics and Glass Engineering at the University of Aveiro, Portugal, using a JEOL 2200FS transmission electron microscope and a Hitachi SU70 scanning electron microscope.
Dynamic light scattering (DLS) experiments were performed using a Malvern Nano ZS90 equipped with a HeNe laser (λ = 632.8 nm) and a power of 10 mW. The scattered light was collected at an angle of 90°. A polystyrene curve was used for each measurement and filled with 3 mL of the freshly prepared deposition solution through a 0.02 μm size filter. The time t = 0 is defined as the introduction of the sample in the apparatus. The temperature of this equipment is maintained by a Peltier element. A reaction temperature of 20 °C was reached after about 3 min. Correlation functions were recovered every 150 s for a total duration of about 90 min. Data evaluation was performed with the software supplied by the manufacturer employing the analysis mode for the multimodal sample, which yields an average value for the hydrodynamic radius and the polydispersity index, which are characteristics of the particle size distribution. DLS measurements were triplicated for each sample to ensure the reproducibility under the preparation conditions.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ce00186b |
This journal is © The Royal Society of Chemistry 2015 |