Aqueous ball milling of nacre constituents facilitates directional self-assembly of aragonite nanoparticles of the gastropod Haliotis glabra

Abstract

A ball-milling approach was developed to investigate the constituents of isolated nacre tablets of the gastropod Haliotis glabra in aqueous suspension without additional chemical additives. The obtained particle mixtures were characterized using X-ray crystallography as well as scanning and transmission electron microscopy. Aragonite nanoparticles retained their crystal structure even after 14 h of ball milling. The long-term stability of the particle mixtures varied as a function of the ball-milling duration. An increased milling time led to rod-like stable assemblies of aragonite nanoparticles. Selected area electron diffraction investigations revealed that the longitudinal axes in about one-third of these nanoparticle rods were oriented along the crystallographic c-axis of aragonite, indicating oriented attachment of the aragonite nanoparticles. These in vitro observations support the idea that a two-stage process, separated into crystallization of nanoparticles and oriented assembly of nanocrystals, could also occur in vivo.

1. Introduction

Ball milling is an efficient method to produce nanomaterials by conventional top-down approaches [1,2]. The properties of the resulting nanomaterial depend on the milling conditions (e.g. dry/wet milling, temperature, properties of the balls, and chemical additives) [3]. Material–ball collisions or turbulence in the stirred media [4–8] change the structure or the surface properties of the sample [9,10]. Ball milling has been applied to geological, biogenic and synthetic CaCO₃ minerals under dry conditions or using water, NaOCl/NaOH or toluene as solvents [11–15]. However, beside the production of nanomaterials, ball milling might serve as a suitable top-down approach to disintegrate and characterize biogenic nanoparticles from biominerals such as nacre.

The CaCO₃ biomineral nacre, formed by molluscs, is an organic–inorganic composite material [16–22] and its formation pathways are heavily debated [23–30]. Individual nacre tablets with diameters of a few micrometres, and less than 1 µm thick, are composed of aragonite nanograins or nanospheres with diameters of 15–180 nm [24,31–38]. This nanogranular texture might be an indication or an outcome of the formation process. Currently, it is suggested that nacre platelets form via amorphous CaCO₃ particles due to secondary nucleation [23,27,30]. Furthermore, the nanoparticle-assembled architecture of nacre tablets influences the mechanical properties of nacre and its order can be transformed by, for example, heat treatment [32,39,40]. Polished nacre surfaces or cross-sections provided evidence that organic material is incorporated into the aragonite tablets [35,41–43], but the exact composition and function of this intracrystalline material still has to be determined. Both the nanograin maturation and the nanograins' properties are difficult to study in vivo.

In this work, we applied ball milling on geological and biogenic aragonite in order to gain new insights into the structural level below individual nacre tablets, i.e. the aragonite nanograins.

2. Results

After removal of the organic surface coating, isolated nacre tablets (Haliotis glabra) and, for comparison, geological aragonite microparticles were mechanically disrupted in aqueous dispersion (pure water). Depending on the milling time, the morphology of the obtained particle mixtures changed significantly. Four milling fractions were defined. Fraction 1 (Fr1) contained isolated aragonite tablets in pure water before the milling. Fraction 2 (Fr2) was collected after 1–2 h, fraction 3 (Fr3) after 4–10 h, and fraction 4 (Fr4) after 14 h of ball milling. The morphology of samples without storage directly after the ball milling, samples frozen in liquid nitrogen, samples stored frozen at −80°C, or lyophilized samples was assessed by scanning electron microscopy (SEM). As shown in the electronic supplementary material, figure S1, drying, freezing or lyophilization did not induce any morphological changes in the particle mixtures. SEM analysis of nacre fraction Fr1 revealed that intact individual tablets displayed the characteristic polygonal shape and a slightly granular surface texture. Particle mixtures of nacre Fr2 contained tablet fragments, the size of which varied from about 30 to 300 nm in diameter (figure 1 and electronic supplementary material, figure S2). Aragonite nanoparticles of 50 nm in diameter or smaller were observed in Fr3 (figure 1 and electronic supplementary material, figure S2). Some nanoparticles tended to attach to each other, forming small rod-shaped nano-objects. In Fr4 these nanoparticle rods with dimensions of about 30–50 nm in diameter and up to 500 nm in length became the dominant fraction (figure 1 and electronic supplementary material, figure S3). Primary nanoparticles were identified as such in rod-shaped nano-objects by ‘bottlenecks’ in between the particles that appeared at regular distances of about 20–40 nm, which corresponds to the average thickness of the rods. Environmental SEM (electronic supplementary material, figure S3a) revealed that all nano-objects retain their morphology in water and in the dry state.

Figure 1.

Four sample fractions obtained after ball milling of isolated nacre tablets; SEM and TEM images of samples obtained after a certain time, Fr1 (t = 0 h) aragonite tablets, Fr2 (t = 1–2 h) aragonite tablets fragments/aragonite particles (SEM and TEM), Fr3 (t = 4–10 h) aragonite nanoparticles and nanoparticle rods, Fr4 (t = 14 h) nanoparticle rods (SEM and TEM); samples were stored at −80°C and observed uncoated.

X-ray powder diffraction (XRD) analysis (table 1 and electronic supplementary material, table S4) revealed that the aragonite crystal structure was almost completely preserved during the ball milling. It was merely reduced from 100 wt% to 86.5 ± 4.9 wt% after 14 h of ball milling. This is in strong contrast to the case of ball-milled geological aragonite, which rapidly converted into calcite, although an identical ball-milling protocol was used. There, the proportion of aragonite was reduced to 29 wt% after 6 h of milling. After 14 h the geological aragonite was entirely transformed into calcite (100 wt%) (electronic supplementary material, figure S4).

Table 1.

Aragonite and calcite content determined by XRD for the fractions Fr1–4 (XRD data and SEM images of geological aragonite are provided in the electronic supplementary material, figure S4).

fraction	Fr1	Fr2	Fr3	Fr4
milling time	t = 0 h	t = 1–2 h	t = 4–10 h	t = 14 h
aragonite (wt%)	100 ± 0	94 ± 3.6	87.5 ± 3.5	86.5 ± 4.9
calcite (wt%)	0 ± 0	9.3 ± 2.5	12.5 ± 3.5	13.5 ± 4.9

The nanoparticle rods were analysed by transmission electron microscopy (TEM) and selected area electron diffraction (SAED). The inset in figure 2a (and electronic supplementary material, figure S3b) shows a TEM micrograph of aragonite nanoparticles and nanoparticle rods (Fr4). The rod lengths varied between 60 and approximately 600 nm, independent of milling time. All rods were composed of nanoparticles with diameters of about 20–40 nm. These values differ slightly from those measured by SEM, but are more precise. XRD investigations revealed crystallite sizes in the same range (electronic supplementary material, figure S4). Energy-filtered TEM (EFTEM) was applied to verify the presence of organic fractions associated with the nanoparticle rods (electronic supplementary material, figure S5). A weak nitrogen signal was detected, indicating that a rather small fraction of proteins or aminosugars is still present. In fact, less than 1% of N would be difficult to exactly quantify. This, however, does not exclude the possibility that there are N-free organic compounds.

Figure 2.

TEM images and the diffraction pattern of a nanoparticle rod. (a) Individual aragonite nanoparticle rod; inset: a bundle of rods. (b) A selected area diffraction pattern for the nanoparticle rod shown in (a), in [100] zone axis orientation. The arrows in (a) and (b) indicate that the rod is oriented along the [001] direction of the aragonite crystal lattice. A selected area diffraction aperture with a diameter of 100 nm was used.

In order to elucidate possible attachment mechanisms for individual nanoparticles during the ball milling, SAED patterns of more than 30 individual nanoparticle rods were investigated. Approximately one-third of the rods gave rise to single crystalline aragonite diffraction patterns, indicating that the individual nanoparticles in those particular rods were crystallographically aligned. The crystallographic direction was determined by projecting the longitudinal axis of the nanoparticle rod in the real space image onto the diffraction pattern in such a way that it crossed through the central spot. This projection passed, in all cases, through the (002) reflection, indicating a nanoparticle attachment along the [001] direction of the aragonite crystal lattice. Other nanoparticle rods produced polycrystalline diffraction patterns or could not be indexed due to a poor alignment with respect to the electron beam.

The structural stability of the nacre tablet milling fractions Fr1–4 was analysed with respect to time. All samples were stored at 20°C in aqueous solution as they were collected during the milling process. The morphology of Fr1 did not change after 1 year of storage (figure 3). Rhombohedral particles were present in Fr2 and Fr3 after 3 days of storage, and their number increased remarkably after 52 days of storage (figure 3). XRD analysis revealed that these particles consisted of calcite. The content of calcite in Fr2 increased from 9 wt% directly after milling to 75 wt% for samples stored at 20°C for 52 days. In contrast, the amount of aragonite in Fr4 decreased slightly from 83 wt% directly after milling to 80 wt% after 52 days of storage. Compact, three-dimensional entities with edge lengths of approximately 2–5 µm covered with nanoparticle rods were formed in Fr4 (figure 3). Focused ion beam (FIB) prepared TEM lamellae of these entities revealed an inhomogeneous calcitic inner core to which aragonite nanoparticle rods were attached in large numbers (electronic supplementary material, figure S6).

Figure 3.

SEM images of stored sample fractions; Fr1 stored at 20°C in aqueous solution for 1 year, Fr2–4 stored at 20°C in aqueous solution for 52 days; rhombohedral-like minerals found in Fr2,3 are indicated by an arrow for Fr3.

3. Discussion

By milling the specimen in aqueous dispersion for longer than 4 h (Fr3, Fr4), nacre tablets disintegrated into aragonite nanoparticles of 20–40 nm in diameter. This narrow size range, obtained here with H. glabra shell constituents, perfectly matches previous reports of an apparently stable structural hierarchy level present in nacre tables (e.g. nanograins of about 32 nm in Haliotis rufescens [32]). Apart from the initial grinding phases with 15 wt% micron-sized particles (nacre tablets), ball milling was carried out at 2 wt%. According to Knieke et al. [3] the observed particle sizes would then represent a true grinding limit rather than a viscous dampening-related one. The calculation of the crystallite size (electronic supplementary material, figure S4) is mostly in agreement with the observed equilibrium particle diameter of 20–40 nm, suggesting that the number of defects, which would facilitate further breakage, in such small aragonite crystals tends to be rather low.

A previously unreported observation made here on H. glabra nacre was that continued ball milling caused these nanoparticles to assemble into nanoparticle rods. This phenomenon did not occur with nanoparticles stored or stirred in solution (see electronic supplementary material, figure S7), indicating the influence of energy impact or surface modifications due to the ball-milling process. Under ball-milling conditions, the diverse patterned surfaces of their particular constituents may individually suffer from exposure to harsh conditions that may lead to shearing off the attached biopolymers, removing hydration layers or inducing amorphous regions or very reactive surfaces. Because of the complex conditions, it is impossible to precisely calculate energy and stress distributions in the ball-milling system [3,6,7]. However, according to Blecher et al. [5] the majority of the energy is dissipated in only about 10% of the total volume. The stress energy input for our system is estimated to reach ∼10⁻⁵ N m per stress event, with collisions and velocity in the range of 10⁸ s⁻¹ and <10 m s⁻¹ [4]. Collision rather than cavitation effects are thus likely to dominate the observed product size distributions. However, potential cavitational effects may account for the temporary removal of a stable hydration layer surrounding each particle. Simulation experiments demonstrated that such hydration layers exist for CaCO₃ nanoparticles of about an order of magnitude smaller than the aragonite particles investigated here [44]. Attachment of crystallographically matching surfaces may occur faster than the recondensation of the water vapour. In this study, in about one-third of the nanoparticle rods the nanoparticles attached along the [001] direction of the aragonite crystal lattice. Colloidal model systems are typically not characterized, taking ball-milling conditions into account [45]. Mechanistic hypotheses could thus only be derived by comparing the geological and biogenic systems reported here, but the geological system was initially already unstable. For biogenic fractions, there is evidence that certain organic entities are still associated with the nanoparticles, thus either modifying or stabilizing hydration layers, or directly mediating centre-to-centre particle attachment.

Strong [001] oriented dipole interactions in synthetic aragonite rods [46] and alignment of 50–70 nm scleractinian coral aragonite nanoparticles in the [001] direction were reported [47]. Neither screwed meanders nor centre-to-centre oriented nanorods aggregated in vitro from mechanically disintegrated nacre mineral tablets have previously been reported [48,49]. With respect to nacre, crystalline aragonite nanoparticles have recently also been observed in vivo in the nacroprismatic transition zone of Pinna nobilis and the authors propose nanoparticle aggregation mechanisms for nacre growth [29]. In accordance with the current model for nacre formation, it was observed that amorphous CaCO₃ nanoparticles also tend to aggregate in vitro or in vivo and subsequently crystallize to calcite [50,51] or aragonite [30,52]. It might be worthwhile comparing crystalline nanoparticle properties from different molluscs [53,54] and other nanogranular biominerals [55] with respect to the biomineral formation process and considering possible self-assembly processes.

The fact that only one-third of the H. glabra nacre particles aligned in crystallographic orientation under in vitro conditions in a ball mill, whereas animals produce almost perfect alignment in three dimensions, indicates that the crystalline units must have been completely disintegrated prior to the subsequent reassembly. Exposing functional entities on particle surfaces of a certain size and pattern could be an important feature which could explain the geometrically centred alignment. Any hypothetical contributions of organic compounds, as indicated by traces of nitrogen associated also with the re-assembled nanoparticle rods, has to be examined in future studies.

Crystallographically, the biogenic aragonite of H. glabra nacre constituents, in contrast to geological aragonite, is remarkably stable in the presence of water and even during and after ball milling (Fr1–4). Nanoparticle rods were quite stable even stored in aqueous solution for more than 50 days. However, this long-term storage resulted in the formation of calcite crystals in Fr2 and Fr3. A dissolution and recrystallization process can be assumed but it remains unclear why these two fractions are less stable during long-term storage in aqueous solution after ball milling, especially since they retain their aragonite crystal structure during the milling process. So far, a transformation of biogenic aragonite to calcite via dissolution and recrystallization during ball milling was only observed in alkaline milling solutions [14]. Stabilizing mechanisms may include lower crystal defect density, intracrystalline biomolecules, biomolecules attached to the surface, and a very stable hydration layer. Organic fractions associated with the biogenic nanoparticles were detected by EFTEM and could potentially aid in the stability of the biogenic particles by surface modification or by modifying or stabilizing hydration layers. Each of them, individually or cooperatively, could prevent dissolution and recrystallization. Parameters such as size and consequently also surface roughness of the nanoparticles interfere with any of these hypothetical mechanisms.

4. Conclusion

This study represents the first ball-milling approach developed to characterize individual biogenic aragonite nanoparticles with respect to properties and behaviour. Aquerous ball milling of nacre tablets without any external additives has enabled the first characterization of relatively stable self-assembled aragonite nanoparticle rods as well as individual nanoparticles similar in size (20–40 nm) to the nanogranular texture of unmilled nacre tablets. These nanoparticles retain their original aragonite crystal structure, in contrast to milled geological aragonite.

The observed particle size, the aragonite stability as well as the rod-like assemblies of aragonite nanoparticles coincide with properties observed for the natural biogenic product, thus leaving us with the question of whether or not oriented attachment of aragonite particles is a phenomenon of relevance for the in vivo formation of nacre tablets and more generally for self-assembling crystalline materials.

5. Experimental section

5.1. Isolation of nacre tablets

Abalone shells were purchased from U.S. Shell Inc. (Texas, USA). The species was identified as Haliotis glabra. The shells were rinsed with water and dried. To remove the outer shell layer, the shells were brushed with 0.6 M HCl, then washed with water and dried. Isolation of nacre tablets was carried out as described previously [56]. Briefly, nacreous shells were crushed to pieces of about 0.5 cm and treated with NaClO for 24 h (12% active chlorine). The turbid supernatant containing the tablet fraction was centrifuged for 2 min at 1892 g and subsequently washed 20 times with ultrapure water. The final solution of ultrapure water containing about 30 wt% isolated nacre tablets was stored at 4°C.

5.2. Geological aragonite sample preparation

Pieces of geological aragonite were cleaned with NaClO and ultrapure water. To reduce the sample size, the aragonite was crushed and pre-milled using 20 mm and 5 mm balls. In this way the particle size was reduced to about 7 µm and smaller. This fraction was again treated with NaClO, centrifuged and washed with ultrapure water comparable to the treatment of isolated nacre tablets.

5.3. Ball milling and sample storage

The ball mill (MiniCer, Netzsch, Germany) was cleaned with HNO₃ and NaClO, followed by intensive washing with ultrapure water. Wet grinding was performed at 28°C and 3000 r.p.m. using ZrO₂ balls (300 µm in diameter, 180 g). Prior to ball milling the amount of nacre tablets and geological aragonite was reduced to 15 wt% with ultrapure water. After 1 h of milling, the solution was diluted to 2 wt% CaCO₃ material content, the ball mill was rinsed with ultrapure water and the milling procedure was continued for an additional 13 h at 3000 r.p.m. Two independently isolated fractions of nacre tablets were milled separately. Samples were taken after 0 h, 0.5 h, 1 h, 2 h, 4 h, 6 h, 8 h, 10 h, 12 h, 14 h of milling. Part of each sample was frozen in liquid nitrogen and stored at −80°C. Also, some sample fractions were lyophilized. Additionally, samples were stored in aqueous solution, as obtained, for up to two months at 20°C.

5.4. Scanning electron microscopy imaging

For SEM analysis, the solutions containing the different milling fractions were placed on silicon wafers and air dried. Lyophilized samples were placed on adhesive carbon plates. The samples were analysed without any sample coating using a Versa 3D scanning electron microscope (FEI) at acceleration voltages of 5 kV or 10 kV. Hydrated samples were imaged using an environmental SEM (Quanta 400 FEG; FEI).

5.5. Preparation of thin sections

Focused ion beam preparation of TEM lamella was carried out in a dual-beam FIB system (Versa 3D; FEI). The sample surface was protected with a Pt layer before a section (approximately 5 × 10 × 0.8 µm) was cut out using a Ga ion beam operated at 30 kV and a beam current of 1 nA. The lamella was attached to an in situ lift-out TEM grid by Pt deposition. Subsequent cleaning and thinning was carried out with sequentially smaller beam currents and a final current of 0.14 nA.

5.6. Transmission electron microscopy imaging

TEM and SAED studies were performed using a JEOL JEM 2100 TEM equipped with an LaB₆ electron source at 200 kV acceleration voltage. The selected area aperture had a diameter of 100 nm. Crystallographic data from selected area diffraction patterns were evaluated using the Pmcn notation of aragonite. The image rotation of 6° between the diffraction patterns and real space images was considered. Scanning transmission electron microscopy micrographs were recorded using a Cs corrected JEOL ARM200F at 200 kV acceleration voltage.

5.7. X-ray powder diffraction

Samples taken after 0, 1, 2, 4, 6, 8, 10 and 14 h of milling were analysed by XRD. They were stored at −80°C and lyophilized prior to the measurement. Samples stored for 52 days at 20°C were frozen in liquid nitrogen and lyophilized. XRD was carried out on a Bruker AXS D8 advance X-ray diffractometer (Lynxeye detector; Bruker) using Cu Kα radiation (30 mA and 40 kV) in a 2Θ range of 2.3° and 150° (step size 0.02°). Diffractograms were analysed using the software Highscore (Panalytical) and Topas (Bruker). The crystallite size was calculated using the Scherrer equation.

Data accessibility

The datasets used and/or analysed during the current study are available from the corresponding author on request.

Competing interests

The authors declare that they have no competing interests.

Funding

No funding has been received for this article.