On Grounds of the Memory Effect in Amorphous and Crystalline

Research Article Cite This: ACS Appl. Mater. Interfaces 2018, 10, 14491−14508

www.acsami.org

On Grounds of the Memory Effect in Amorphous and Crystalline Apatite: Kinetics of Crystallization and Biological Response Vuk Uskoković,*,†,‡ Sean Tang,‡ and Victoria M. Wu‡ †

Advanced Materials and Nanobiotechnology Laboratory, Department of Bioengineering, University of Illinois, 851 South Morgan Street, Chicago, Illinois 60607-7052, United States ‡ Advanced Materials and Nanobiotechnology Laboratory, Department of Biomedical and Pharmaceutical Sciences, Center for Targeted Drug Delivery, Chapman University, 9401 Jeronimo Road, Irvine, California 92618-1908, United States ABSTRACT: Memory effects, despite being intrinsic to biological systems, are rarely potentiated in biomaterials. By exploring the transition between amorphous calcium phosphate (ACP) and hydroxyapatite (HAp) from different empirical angles, here, we attempt to set the basis for elicitation of structural memory effects in CPs. Two CPs precipitated under different degrees of saturation (DS), yielding HAp at a low DS and ACP at a high DS, were shown to evolve into structures with a high level of crystallographic similarity after their prolonged aging in the solution and served as the basis for this study. Amorphous-tocrystalline transition was abrupt in both precipitates, indicating an autocatalytic process preceded by considerable nucleation lag times, but it was more dynamic and proceeded in multiple stages in the precipitate formed at a higher DS, involving a greater degree of lattice rearrangements. ACP was found to exist in one of the two stoichiometrically and crystallographically different forms, one of which, amounting to ≥60 wt %, resembled tricalcium phosphate and transformed to HAp through the surface dissolution/reprecipitation mechanism and the other one, amounting to ≤20 wt %, was apatitic, enabling the transformation of ACP to HAp via martensitic, bulk lattice reordering phenomena. Large density of stacking faults was responsible for the comparatively high lattice strain, the property to which biogenic apatite owes its ability to accommodate foreign ions and act as a mineral reservoir for the body. Being the precursor for biogenic apatite during biomineralization and a thermodynamically logical intermediate in the ripening of HAp per the Ostwald law of stages, ACP proved to be more prone to structural transformation than the final and the most stable of the CP phases in this sequence of events: HAp. Amorphized upon gelation, two CPs transformed into HAp, albeit at different rates, which were higher for the material that had been crystalline prior to amorphization than for the one that had initially been amorphous, indicating the presence of a definite memory effect. The two HAp powders with different histories of formation also elicited different biological responses, including a Runx2 transcription factor expression in MC3T3-E1 osteoblasts, cell uptake efficiency, and antibacterial activity, extending the memory effect in HAp to the biological domain. The biological response was typically indistinct between the final products and their respective precursors but markedly different between the two products obtained by following different formation paths, confirming the presence of the given memory effect. It is suggested that the key to explaining the difference in the response between the materials differing in their route of formation lies in the direct dependence between the DS at which precipitation occurs and the rate of exchange of hydrated ions and ionic clusters across the particle surface in contact with a solution. KEYWORDS: amorphous calcium phosphate, antibacterial, crystal growth, gene expression, FTIR, hydroxyapatite, kinetics, memory, nucleation, uptake, XRD ity to protons, polarizability and convertibility to an electret,2 and so on. Hydroxyapatite (HAp) is also both the least soluble pure CP and its only hydroxylated stoichiometry, going against the common sense expectation that hydroxylation would cause an increase in solubility. Studies reporting the absence of hydroxyl groups in biogenic apatite,3,4 a compound native to a

1. INTRODUCTION Calcium orthophosphates (CPs) are known for exhibiting an array of peculiar and anomalous phenomena, which is the reason they may be justifiably considered the analogues to water in the domain of solid materials.1 The list of these phenomena is long and includes effects such as: (i) exceptional foreign ion accommodation capacity, (ii) complex nucleation and growth pathways, (iii) natural endosomal escape propensity and “smart” drug delivery properties associated therewith, (iv) finite piezoelectricity, pyroelectricity, conductiv© 2018 American Chemical Society

Received: February 9, 2018 Accepted: April 6, 2018 Published: April 6, 2018 14491

DOI: 10.1021/acsami.8b02520 ACS Appl. Mater. Interfaces 2018, 10, 14491−14508

Research Article

ACS Applied Materials & Interfaces

kinetics of formation of two otherwise identical HAp components,26 showing for the first time that the history of formation of a CP phase can have a key effect on its properties. Our goal in this study was to explore phenomena that may be the basis for the elicitation of structural memory effects in CPs. Memory, in the sense in which we discovered it in CPs, refers to the ability of a material to follow a trajectory toward a particular structural state differently depending on the history of reaching the starting point of this trajectory. In this effort, we delved into the transitions between ACP and HAp from different empirical angles and using different experimental techniques. ACP is widely accepted to be the first phase that forms during the morphogenesis of apatite in the bone27,28 as well as an intermediate in its resorption during bone remodeling,29,30 implying the potential biomedical meaning of this basic study. The fact that ACP is most abundant in the actively metabolized regions of the bone31,32 signifies the importance of understanding its transitions to be able to therapeutically interfere with them and produce desirable health outcomes. Moreover, the prominence of the transitory amorphous phase in HAp may be the key to explaining not only its inherent pleiotropy but also its being more structurally pliable and “alive” than other minerals composed of Earthabundant ions, which is the reason it interfaces so well with the living tissues. Therefore, findings deduced in the course of this investigation aim at expanding the repertoire of unusual properties of CPs and understanding their idiosyncrasies mechanistically, for the purpose of their potential utilization in a number of possible biotechnological and medical applications. We expect that these and similar findings will contribute to the revival of interest in this extraordinary material whose use in the biomedical domain has currently been reduced mainly to that of an osteoconductive component of tissue engineering constructs. Such a revival may endow CP with similarly versatile and pleiotropic functionalities as those which it exhibits in our bodies.

highly hydrated chemical environment, further add to this veil of perplexity surrounding HAp. All in all, with strange phenomena entailing the formation and evolution of CPs in various physicochemical contexts and conditions, it comes as no surprise that biological systems, especially the most complex ones, which are themselves known for their functional complexity and ontological inscrutability, chose it as the material to most commonly coevolve with.5 Complex mechanisms of formation and phase transformations entailing phase transitions in CPs, involving up to a dozen different orthophosphate stoichiometries,6 are some of the hallmarks of complexity in these materials. La Mer’s classical, diffusional growth7,8 was dismissed as a real possibility long ago in favor of the aggregational growth conforming to the Ostwald law of stages.9 Confirmed by high-resolution microscopies,10 the brick-and-mortar analogy is frequently used to describe the way in which the amorphous surface of brick-like HAp particles facilitates the coalescence of individual particulate units. On one hand, this phenomenon is at the core of practically all the major weaknesses of CPs, which are routinely ignored in the literature11 and include the following: (i) high aggregation propensity of CP particles and their correspondingly low colloidal stability, (ii) relatively weak morphological control, and (iii) high hydration of the particle surface, predisposing it for instability and low capacity for stable conjugation with organic ligands. On the other hand, it can be turned into strength if precisely harnessed. For example, the fact that this brick-and-mortar mechanism is the basis of the self-assembly of HAp nanoparticles12 hints at the great potential of controlling the interaction between the amorphous and the crystalline components in these systems. The medical applicability of this fundamental finding may be best illustrated by the use of amorphous CP (ACP) as an efficient agent for remineralization of enamel13 and prevention of its demineralization.14 However, the exact chemical pathways taken on by CP particles during nucleation and growth are still subject to debate, partly because of their pleiotropy, that is, the propensity to follow a number of different routes depending on the chemical conditions at work. At the same time, the state-of-theart materials science is mostly focused on elucidating the materials structure at the finest scales, albeit statically, in 3D, and not so much on the routes followed by the atomic and nanoparticulate entities to specific structural states,15 that is, in a spatially and temporally resolved, 4D manner.16 The latter is particularly important to understand for materials that dynamically coevolve with their environments, one example of which is the bone mineral, whose viable state is associated with constant dissolution and recrystallization during the perpetually ongoing bone remodeling process. In an attempt to make biomaterials ever more biocompatible and capable of engaging in an intimate and functional interface with living tissues, materials are often designed to mimic some of the properties of animate matter. However, memory is normally not potentiated in biomaterials, even though it is an essential biological property, exhibited not only at the central nervous system level but also at the level of organs or single cells, which are both intrinsically prone to priming,17,18 a form of an implicit memory effect. Memory effects, in turn, have been observed in a number of different materials, including pure metals such as bismuth,19 metallic alloys such as nitinol,20,21 polymers such as urethanes,22 styrenes,23 and even lactides,24 and, of course, water.25 Recently, we were able to tune the setting and drug release kinetics by controlling the

2. MATERIALS AND METHODS 2.1. Synthesis. Two different CP nanopowders were synthesized and compared in this study: HAp (Ca5(PO4)3OH) and ACP. Their synthesis involved precipitation from aqueous solutions. Specifically, to make HAp, 400 mL of 0.1 M aqueous solution of calcium nitrate [Ca(NO3)2, Fisher Scientific] containing 50 mL of 28% NH4OH was added dropwise to the same volume of 0.06 M aqueous solution of monoammonium phosphate (NH4H2PO4, Fisher Scientific) heated to 80 °C and containing 25 mL of 28 vol % ammonia (NH4OH, SigmaAldrich), vigorously stirred with a magnetic bar (400 rpm). After the addition of Ca(NO3)2 was complete, stirring was suspended, and the precipitate was left to age in atmospheric conditions together with its parent solution for 24 h unless noted otherwise. After the given time, the precipitate was washed once with deionized H2O, centrifuged (5 min at 3500 rpm), and let to dry overnight in air. ACP was made by abruptly adding a solution containing 100 mL of 0.5 M Ca(NO3)2 and 7 mL of 28 vol % NH4OH into a solution comprising 100 mL of 0.2 M NH4H2PO4 and 4 mL of 28% NH4OH. The fine precipitate formed upon mixing was aged for 15 s before it was collected, centrifuged, washed with water, centrifuged again, washed with ethanol, then dried overnight in air, and stored at 4 °C to prevent spontaneous transformation to HAp. The gelation process involved the mixing of 2 g of either HAp or ACP powders with 75 μL of 2 wt % Na2HPO4 aqueous solution set to pH 5.0 with the addition of HCl. Mixing was followed by a 5 min agitation on a digital vortex mixer (Fisher Scientific, 2000 rpm) in closed Eppendorf tubes. Precursor HAp and ACP powders were freshly prepared and then, without any aging in the solution, washed 14492


Research Article


Table 1. Ionic Concentrations of the Reactant Solutions, Initial and the Final pH Values and DSs during the Synthesis of Crystalline and Amorphous Apatite Powders by Precipitation, along with the Average Crystallite Sizes of the Precipitated Powders precipitate

phase (if washed and dried)

[Ca2+] (M)

[HxPO43−x] (M)

initial pH

final pH

DS

d (nm)

1 2

HAp ACP

0.1 0.5

0.06 0.2

11 10

10.3 9.3

21.5 28.8

12

1 4 ⎛ h2 + hk + k 2 ⎞ l2 = ⎜ ⎟+ 2 2 2 3⎝ ⎠ c d a

and dried in air for 2 h before being mixed with the Na2HPO4 solution. The degrees of saturation (DSs) were calculated using an algorithm based on the Debye−Hückel equation DS = pK sp − pQ

(1)

Q = {Ca 2 +}x {PO4 3 −}y{H+}z {OH−}w

(2)

Williamson−Hall (WH) plots were constructed to differentiate the effects of lattice strain from the effects of the crystallite size on the broadening of diffraction lines. (β cos θ)/λ was plotted as a function of d*, where β was the integrated breadth calculated as a ratio of the integrated area of a diffraction peak to its height, θ was the diffraction angle, and d* was the reciprocal Bragg distance, 1/d, corresponding to the given diffraction line. The following equation,34 where D is the average crystallite size and ε is the measure of the microstrain distribution, was used to derive the strain from the slope of the curve

Q is the ionic activity product of the solution, and pKsp is the negative logarithm of the solubility product of a given CAP phase, equaling 117.3 for HAp (Ca10(PO4)6(OH)2). Activity coefficients were calculated through log γ = −Azi2/(I1/2 + Bai), where zi is the charge number of ion species i, I is the ionic strength of the solution, and A is a temperature-dependent constant equal to 0.5115 at 25 °C. B and ai are, like A, constants depending on temperature, dielectric constant of the solution, and Debye screening length; ai = 6 × 10−8 for Ca2+; 9 × 10−8 for H+; and 4 × 10−8 for HxPO4x−3/CaH2xPO42x−1 ions. Appropriate dissociation constants for H2O, HxPO4x−3, HxCO3x−2, and HF and association constants for CaHxPO4x−1, CaHxCO3x, and Ca··· OH were taken into account as functions of pH.33 Table 1 summarizes the different ionic concentrations in reactant solutions used to obtain DS of solutions precipitating CP nanoparticles. The DS for the supersaturated solution in the case of synthesis of ACP was calculated by taking into account complete concentrations of the reactant solutions. The DS for the supersaturated solution in the case of synthesis of HAp was calculated by assuming a dropwise addition of the phosphate solution and approximating it at 1 mL of it for each precipitation event to estimate the upper limit of DS. 2.2. Physicochemical Characterization and Modeling. Scanning electron microscopy (SEM) studies were carried out on JEOL JSM 6320F-FESEM operated at 4.2 kV voltage and 8 μA beam current. Sample preparation involved depositing powders or pastes on clean aluminum stubs using the carbon tape and subsequently sputtercoating them with gold to reduce the charging effects. X-ray diffraction (XRD) was carried out on a Bruker D2 PHASER diffractometer using polychromatic Cu as the irradiation source. The Kβ line was stripped off with an inbuilt filter, whereas the Kα2 line, the frequent source of peak asymmetry artifact at high 2θ angles, was stripped off automatically, together with the instrumental line broadening. The step size was 0.01° and the irradiation time per step was 1 s. Particle size distribution histograms for the different types of cements and HAp precursors were obtained from their corresponding scanning electron micrographs using ImageJ (NIH, Bethesda, MD) and the sample size of 350−1500 particles. High-resolution transmission electron microscopy (HR-TEM) analysis was performed by smearing a droplet of the CP dispersion over a carbon-coated copper grid (Ted Pella), blotting off the excess liquid with filter paper after 1 min and imaging on a FEI monochromated F20 UT Tecnai HR-TEM under the electron acceleration voltage of 200 kV. Interplanar spacing (dhkl) was measured using the Bragg relation where λ is the wavelength of Cu Kα radiation, 1.5416 Å, and Θ is the diffraction angle for (hkl) reflection

dhkl =

λ 2 sin Θhkl

(4)

(β cos θ)/λ = D−1 + 2εd* The root-mean-square strain using the following relation

(5) (⟨ε02⟩1/2)

was subsequently calculated

⟨ε0 2⟩1/2 = (2/π )1/2 ε

(6) 35

The Johnson−Mehl−Avrami−Kolmogorov model was used to estimate the kinetics of nucleation/growth from the aqueous solutions. The following relationship was used to derive the Avrami rate constant, k, for crystallization of HAp and the Avrami exponent, n, related to the nucleation mechanism, where the extent of the reaction, α, was determined as the ratio of the integrated intensity of the diffraction peak (304) at time t and at its maximal value during the reaction

ln(− ln[1 − α]) = n ln(k) + n ln(t )

(7)

Absolute intensities of the diffraction peaks were measured as their maximal height, while integrated intensities were measured by calculating a background function and the deconvoluted peak profile using an automated Gaussian-fitting routine (OriginPro 2016). Fourier transform infrared (FT-IR) spectroscopy measurements on nonannealed samples were performed by sampling out a portion of the precipitate after various aging times and recording spectra in the 400− 4000 cm−1 wavenumber range with the maximal spectral resolution of 2 cm−1 on a Bruker ALPHA Platinum attenuated total reflection (ATR) spectrometer with a single-reflection diamond/WC composite ATR module. Adsorption isotherms were constructed by incubating 5 mg/mL of different CP nanoparticles in aqueous solutions containing different concentrations (0.25, 1, 2, and 4 mg/mL) of either bovine serum albumin (BSA) as the model negatively charged protein or lysozyme as the model positively charged protein for 1 h, after which the particles were spun down by centrifuging for 5 min at 7000 rpm. The amount of adsorbed protein after centrifugation was measured by comparing the absorbance at 278 and 282 nm (BMG LABTECH FLUOstar Omega) for BSA and lysozyme, respectively, in the initial solution and in the supernatant after the incubation. Absorbance was converted to protein concentration based on the standard curve constructed using the following concentrations: 0.125, 0.25, 1, 2, 4, and 10 mg/mL. 2.3. Biological Assays. The mouse calvarial preosteoblastic cell line, MC3T3-E1 subclone 4, was purchased from American Tissue Culture Collection (ATCC, Rockville, MD) and cultured in alpha minimum essential medium (α-MEM; Gibco) supplemented with 10% fetal bovine serum (FBS, Invitrogen) and no ascorbic acid (AA). The medium was replaced every 48 h, and the cultures were incubated at 37 °C in a humidified atmosphere containing 5% CO2. Every 7 days, the cells were detached from the surface of the 75 cm2 cell culture flask (Greiner Bio-One) using 0.25 wt % trypsin, washed, centrifuged (1000

(3)

The hexagonal lattice parameters of HAp, a and c, were calculated from dhkl using the following equation, assuming additionally that c/a = 0.72998 for HAp 14493


Research Article


Figure 1. HR-TEM images of ACP (a) and HAp (b) nanoparticles at the atomic scale.

Figure 2. SEM images of HAp (a) and ACP (b) powders. White arrows indicate the crystal growth by coalescence of submicron particles with larger, microscale particulates. Both materials additionally exhibited aggregational growth in the transition of the ACP intermediate to HAp at the nanoscale (c,d). rpm × 3 min), resuspended in 10 mL of α-MEM and subcultured in a 1:7 volume ratio. Cultures were regularly examined under an optical microscope to monitor growth and possible contamination. For the purpose of staining with fluorescent markers, MC3T3-E1 cells were seeded on glass cover slips placed in 24-well plates and 500 μL of α-MEM supplemented with 10% FBS and no AA at the density of 6 × 104 cells per well. After 5 days of incubation, nearly confluent cells were treated with α-MEM containing 50 μg/mL of AA as the mineralization inductor. At the same time, 2−4 mg/cm2 of particles were added to the cells. After 5 days of incubation with the particles, cells were stained for collagen and nucleus. The staining procedure began with washing the cells with phosphate buffered saline (PBS, pH 7.4) and fixing them for 15 min in 3.7% paraformaldehyde. The cells were then washed with PBS 3 × 5 min and then with the blocking solution (PBT = 1% BSA, 0.1% Triton X-100 in PBS) 2 × 5 min. The cells were then blocked and permeabilized in PBT for 1 h. The cells were then incubated with 100 μL/well of the primary antibody, 10 μg/ mL rabbit anticollagen-type-1 (Abcam) in PBT for 1 h. The cells were then washed with PBS 3 × 10 min and incubated with 100 μL/well of the secondary antibody, 10 μg/mL Alexa Fluor 555 goat antirabbit IgG (Invitrogen), and 20 μg/mL 4′,6-diamidino-2-phenylindole (DAPI) dihydrochloride nuclear counterstain (Invitrogen), all in PBT for 1 h and then washed with PBS 3 × 5 min. The DAPI/secondary-antibody solution also contained either 2 μM calcein AM as a CP-particle-

staining compound or 20 μg/mL FITC-BSA as a nonspecific polymerstaining compound. In the case of staining for F-actin, the cells were incubated with 100 μL/well of 10 μg/mL phalloidin−tetramethylrhodamine (Alexa Fluor 555, Invitrogen) and 20 μg/mL DAPI, all in PBT for 1 h and then washed with PBS 3 × 5 min. The cover slips containing the fixed and stained cells were mounted onto glass slides using HardSet VECTASHIELD and nail polish and were subsequently imaged on a confocal laser scanning microscopeC1si (UCSF Nikon Imaging Center) at 20−100× magnification in oil. The particle uptake was analyzed using flow cytometry (Becton Dickinson FACSVerse). MC3T3-E1 cells were grown to confluency in aforementioned growth conditions in 24-well plates before 5 mg/mL of CP particles were added to them. After 24 h incubation at 37 °C, the cells were rinsed with PBS and trypsinized using 0.25% trypsin− ethylenediaminetetraacetic acid. The cells that had uptaken CPs were sorted from the control MC3T3-E1 cells based on changes in the increased side scatter, which is indicative of increased granularity in the cells due to intracellular localization of CP particles uptaken by the cells. Total RNA was extracted from MC3T3-E1 cells following a 10 day incubation with the particles using the RNeasy kit (Qiagen). cDNA was synthesized using the High-Capacity cDNA reverse transcription kit (Applied Biosystems) from 100 ng of total RNA. cDNA was quantified using custom TaqMan probes for osteocalcin (BGLAP), 14494


Research Article


Figure 3. XRD patterns of the precipitates 1 (a,c) and 2 (b,d) sampled out of their parent solutions after different periods of time following precipitation, showing the course of the transformation of the initial precipitates into crystalline HAp through amorphous intermediates, along with the timeline of appearance of the (002) diffraction peak centered at 2θ = 26.53° in place of the amorphous hump for aging precipitates 1 (c) and 2 (d) and the extent of the reaction of disappearance of the amorphous component (e) and the appearance of the crystalline apatite component (f) for the two precipitates as a function of time. Reaction extent was measured on the amorphous hump centered at 2θ = 28.56° for ACP and the (304) peak centered at 2θ = 64.52° for HAp. Dashed lines in (c,d) indicate the diffraction angle, 2θ, of the (002) reflection maxima, and the arrow in (d) indicates a shift toward the lower lattice parameter, d. Numbers 1−3 in (f) denote the three different line profile families described in the text and in the subsequent figure. Runx2, β-actin (ACTB), and POLR2 on a StepOne real-time PCR system (Applied Biosystems). The real-time PCR results were analyzed using the ΔΔCt method, and all the data were normalized to the expression levels of ACTB as the housekeeping gene. All the samples were analyzed in three experimental triplicates and six analytical replicates (n = 3 × 6 = 18). Antibacterial assays were performed using a liquid inoculation test where a single colony of Escherichia coli cultured on a blood agar plate over a period of 24 h was stabbed with a pipette tip, placed in 5 mL of lysogeny broth (Sigma Life Sciences), and incubated overnight at 37 °C and 250 rpm. This overnight stock was diluted the next day down

to identical concentrations in the broth and had different concentrations of different CP powders, ranging from 10 to 100 mg/mL, added to them. This was followed by 24 h incubation at 37 °C and 250 rpm. All the samples were analyzed in triplicates and compared against the infected, particle-free control broth. To prevent the CP particles from interfering with the optical density (OD) measurements, after the 24 h incubation 1 M HCl was added to the particle-containing broths until CP particles were dissolved. The OD of broths was measured at λ = 600 nm (BMG LABTECH FLUOstar Omega) and converted to the number of colony-forming units. 14495


Research Article


3. RESULTS AND DISCUSSION 3.1. Precipitate Growth and Aging Analysis. To confirm the crystallinity of HAp and the amorphousness of ACP, the dried precipitates were analyzed under HR-TEM (Figure 1). While ACP particles displayed smeared lattice fringes (Figure 1a), which is indicative of the merely low-range order, HAp displayed more distinct regions occupied by an ordered lattice (Figure 1b). In spite of this, the fully amorphous regions were present in HAp, indicating that the precipitate was poorly crystalline HAp. SEM images shown in Figure 2 capture the growth of larger, microsized particles from the finer, nanosized units (Figure 2a,b), which, in turn, are built from even finer nanoparticles (Figure 2c,d). Aggregational growth presents the major mode of growth of CPs and only the finest, Posner’s clusters as nonequilibrium, prenucleation species with circa 9 Å in size are expected to form via diffusional growth. This type of growth is fostered by the intensely hydrated and diffusive particle interface that is characteristic of the hydroxylated crystal structure of HAp. Such aggregational growth is relatively slow, as shown by the first evidence of the crystalline apatite phase coming 90 min after the precipitation has occurred. Although the white precipitate forms continuously during the mixing of the two reactants, the process of formation continues well after the mixing is done. These results refute two common preconceptions: (i) that HAp forms immediately upon precipitation and (ii) that ACP does not transform into crystalline HAp if left in contact with its parent solution. For crystalline HAp to form, sufficient aging of the precipitate is required, while for ACP to form, prompt washing and drying are required or else the precipitate will crystallize. Although poorly crystalline HAp will form and ACP will remain amorphous if the precipitates are washed and dried right after their formation, their prolonged aging together with the supernatant changes their identity over time thanks to the complex interrelation of thermally induced bulk lattice reordering, aggregational growth, Ostwald ripening, and dissolution/reprecipitation phenomena. The nucleation rate is directly proportional to supersaturation at the onset of precipitation and is, per classical theory, expected to obey the following relation, where γ is the interfacial tension for the formation of the critical nucleus (∼1.2 nm), S is the supersaturation, and A and B are preexponential and exponential constants, respectively J = A exp( −Bγ 3/S2)

Because both nucleation density and the nucleation rate are proportional to S2, whereas the crystal growth rate is proportional to Sx and x = 1 for diffusion-limited growth, typically taking place at high S, increasing S in a relatively high S range will lead to a higher increase in the nucleation rate than in the crystal growth rate. This indicates that the greater concentration of CP nuclei forms at a higher DS, and they take more time to grow into stable crystalline units than when precipitation is run at a lower DS. As a result, this aggregational growth is, naturally, lengthier for the amorphous phase precipitated at a higher DS. This is evidenced by the XRD studies, which show that while the amorphous structure is present for 60−90 min in the precipitate formed at higher DS (28.8, Table 1, Figure 3a), it is twice as long, that is, for 150− 180 min, in the material formed at a lower DS (21.5, Table 1, Figure 3b). Both precipitates eventually adopt almost indistinct patterns indicative of a very high level of crystallographic similarity (Figure 3a,b). The kinetics of the transformation from the ACP intermediate to HAp as the final phase is, however, different depending on the DS. For example, following the timeline of the appearance of the (002) diffraction peak centered at 2θ = 26.53° in place of the amorphous hump centered at 2θ = 28.56° demonstrates a quicker but also more dynamic transition in case of the precipitate formed at a higher DS (Figure 3c) compared to the precipitate formed at a lower DS (Figure 3d). Thus, while the (002) peak is first seen after 120 min of aging in the former precipitate, it is first detected 60 min later in the latter precipitate. Also, the broad precursor doublet preceded the detection of the (002) reflection in both precipitates but was shifted to lower diffraction angles (2θ = 26.09° compared to 26.53°) and higher lattice parameters in the precipitate originating from a higher DS, and such a shift was not noticed in the precipitate originating from a lower DS. This is suggestive of the more dynamic transition, involving a greater degree of lattice rearrangements in the precipitate formed at a higher DS. The kinetics of crystallization was analyzed in more detail by following the evolution of the integrated intensities and line profiles of a selected broad reflection of ACP (Figure 3e), centered at 2θ = 28.56° and with the Lorentzian base range of 10 < 2θ < 40°, and a nonoverlapping, high-angle (304) reflection of HAp (Figure 3f), peaking at 2θ = 64.52°. As far as the former peak is concerned, it is maintained at a consistent high intensity before rapidly decreasing down to negligible levels (Figure 3e). This analysis, therefore, shows that the amorphous-to-crystalline transition in both precipitates is abrupt, indicating an autocatalytic process preceded by considerable nucleation lag times, otherwise typical for the crystal growth of this material.37 Apatites, namely, are characterized by the relative slowness of phase transformations, coinciding with their being a material of choice by the living systems whose biological viability depends on the comparative slowness of chemical reactions comprising them.38 However, following the evolution of the (304) peak of HAp demonstrates its steady increase during the crystallization of the solid phase precipitated at a lower DS and an increase in stages during the crystallization of the solid phase precipitated at a higher DS (Figure 3f). This smoother transition observed in the evolution of the high-index (304) reflection is in agreement with the abovementioned smoother lattice rearrangements observed in the evolution of the low-index (002) reflection (Figure 3c,d). The sigmoid shape of the kinetic curve in the case of the

(8)

For the polynucleation model, which HAp dissolution and growth are assumed to follow,36 the following equation can be used as well, where T is the temperature, k is the Boltzmann constant, ν is the molecular volume of the crystallizing phase (5.287 × 10−28 m3 for HAp), and β is a geometric factor related to the perimeter of the surface (for cubic shaped nuclei equal to 4) J = A exp(−βγ 2ν 4/3/3k 2T 2(ln S)2 )

(9)

Nucleation density, N, in turn, is described by the classical nucleation theory with the following term, in which ΔG* is the activation energy for nucleation, Ω is a pre-exponential factor, and C is a constant N = Ω exp(−ΔG* /kT ) = C exp(−βγ 3ν 2 /k3T 3(ln S)2 ) (10) 14496


Research Article

ACS Applied Materials & Interfaces precipitate formed at a lower DS fits the assumed division of the process to three stages: (i) the induction time period; (ii) the cooperative growth process, which is indicative of the interface-controlled, heterogenous growth; and (iii) deceleration due to depletion of the concentration of growth units. The disparity between the plateauing of the ACP diffraction peak intensity (Figure 3e) and the steady, sigmoid, concentration-dependent increase in the presence of HAp in the solid mixture (Figure 3f) can be explained by the presence of dual forms of ACP. Although each of the dozen or so calcium phosphate phases is theoretically able to exist in an amorphous state, it can be hypothesized that in the 1.3 < Ca/P < 1.67 range of molar concentrations, ACP exists in either of the two major stoichiometrically and crystallographically different forms, one of which resembles tricalcium phosphate (TCP) [Ca3(PO4)2, Ca/P = 1.5] and the other one resembles HAp [Ca5(PO4)3OH, Ca/P = 1.67]. This is in agreement with the previously reported existence of two distinct ACP phases, one bearing resemblance to β-TCP and the other one to HAp.39 Multiple structural forms of the amorphous phase were found not only in another major biomineral, amorphous calcium carbonate40 but also in other materials, including the oxides of titanium41 and iridium,42 justifying the use of the term “polyamorphism”.43 Being the dominant form of water in the outer space, amorphous ice, like ACP, presents another material showing two distinct X-ray glassy forms.44 Every amorphous structure, correspondingly, exhibits a finite level of order,45 which, albeit difficult to assess and quantify, is assumed to be rooted in the prenucleation cluster ordering phenomenon, which is distinct for different amorphous phases despite being breakable down to identical elementary units.46 In the case of the growth of titania, TiO6 tetrahedra were discerned as such units,47 whereas the same building block role was ascribed to Cu(OH) 2 (H 2 O) 4 “tectons” for copper-based solids. 48 Ca9(PO4)6 hexagons a.k.a. Posner’s clusters are expected to play the same role in the growth of various forms of apatitic structures, amorphous or crystalline. Ion association complexes, such as [Ca(HPO4)3]4−, were proposed as even more elementary building blocks that combine with Ca2+ ions and undergo concomitant deprotonation to yield ACP and HAp.49 Whereas the TCP-resembling ACP phase is expected to yield the diffuse XRD pattern with two distinct amorphous humps, such as those present in Figure 3a,b, the diffuse diffraction pattern of HAp-resembling ACP, such as that forming through amorphization effects caused by surface additives,50 is typically devoid of any peaks or humps. Coinciding with the most intense reflections of dicalcium phosphate dihydrate (brushite), not HAp, the two peaks in the diffractogram of ACP indicate the predominant presence of the amorphous nonapatitic phase in the mixture. On the basis of previous studies on the composition of the amorphous phase, this stoichiometry could be approximated as CaxHy(PO4)z·nH2O.51 At the same time, it is a common assumption that only if the XRD profile of the amorphous structure is related to a crystalline polymorph can it act as a seed for its nucleation and growth.52 Also, the fact that the solubility product, Ksp, for this standard, TCP-resembling ACP is in the similar range, circa 20−25 for −log Ksp, as that for both TCP polymorphs53,54 is another indicator of their stoichiometric and crystallographic similarity. Given the similarity in the hexagonal lattice symmetry between the HAp-resembling ACP and HAp, the initial phase transition yielding HAp in the precipitate formed at the higher DS is expected to proceed via lattice reorganization and reordering,

that is, without the involvement of alternate dissolution and reprecipitation of the growth units. The depletion of this apatitic form of ACP, detectable as sheer noise and a flat pattern in diffractograms, will produce no apparent change in the XRD pattern. This type of phase transition that proceeds independently of its complementary mechanism comes earliest in the timeline of the process but presents a minor percentage of the total ACP-to-HAp transition. The majority of it proceeds during the abrupt drop in the ACP concentration and involves the transition of the TCP-resembling ACP into both HAp and HAp-resembling ACP through the surface dissolution/ reprecipitation mechanism. This latter HAp-resembling form of ACP presents the source of the subsequent, gradual increase in the crystallinity of HAp, succeeding in time the point of the abrupt drop in the ACP amount (Figure 3e,f). On the basis of the reaction extent values separating the individual steps depicted in Figure 3f, it can be estimated that the amount of the HAp-resembling ACP component in ACP is no less than 20 wt %, whereas that of the TCP-resembling ACP component is no less than 60 wt %. These kinetic effects suggest the presence of two different amorphous phases involved in the process, one of which at least is HAp-resembling in stoichiometry and crystal structure, enabling the transformation to HAp via martensitic, bulk lattice reordering phenomena. The transitions between them, when viewed as crystallographic projections on the basal, (001) planes, are schematized in Figure 4. These findings agree

Figure 4. Phase transformation sequence occurring during the aging of the primary, amorphous CP precipitate, starting with the formation of Posner’s clusters with the Ca9(PO4)6 stoichiometry (a), followed by their growing into centers of the amorphous phase resembling hexagonal β-tricalcium phosphate, R3cH, in the ionic arrangement (b) and its transformation into an amorphous phase, structurally resembling hexagonal HAp, P63/m (c), which eventually rearranges to form poorly crystalline HAp as the final phase (d). All the structures show the projection parallel with the c-axis and perpendicular to the (001) plane.

with the previous detection of two microstructurally distinct types of ACP forming during the ACP-to-HAp conversion process using X-ray photon correlation spectroscopy,55 the initial of which was more structurally amorphous than the intermediate, apatite-resembling ACP, which eventually yielded poorly crystalline HAp. A more detailed comparison of the evolution of the (304) diffraction peak of HAp (2θ = 64.52°, d = 1.444 Å) between the two forms of CP during crystallization demonstrates a more erratic and faster progress toward HAp in the case of the phase precipitated at a lower DS as compared to the one formed at a 14497


Research Article


Figure 5. Evolution of the (304) diffraction peak of HAp centered at 2θ = 64.52° as a function of the aging of the fresh precipitates 1 (a) and 2 (b) in contact with their parent solutions. Peaks are manually shifted along the y-axis to avoid the overlap and accentuate the fluctuant (a) or stable (b) patterns over time. Three types of peaks, differing from one another in terms of the shape, are labeled with 1, 2, and 3. The degree of the (304) peak asymmetry is measured as the difference between the halfwidths in the low (Hl) and the high (Hr) diffraction angle ranges normalized to the total halfwidth (Hl + Hr) during the aging of precipitate 1 (right y-axis, black symbols) and precipitate 2 (left y-axis, red symbols) (c). WH plots constructed in the 0.1−0.5 Å−1 range of the reciprocal lattice spacing, d, for the final product of the solidification reaction, HAp, depending on whether it is yielded in precipitate 1 or 2 (d).

higher DS, in which case the transition proceeded smoother but slower (Figure 5a,b). Three characteristic profiles of the (304) peak were noticed, progressing from one to another form as the precipitate matures and transitions from ACP to HAp. With two different line profiles (1 and 2) corresponding to ACP and with the transition from them being consistent between samples, they might signify the aforementioned two different forms of ACP, one resembling TCP in structure and stoichiometry and the other one resembling HAp. Expectedly, the transition from 1 to 2 proceeds faster during crystallization of HAp in the precipitate formed at the lower DS than in its higher DS counterpart, while the peak shape 3 is also attained faster in the former than in the latter material. The integrated intensities and shapes of peak 3, corresponding to the final product, HAp, were highly similar for both materials, demonstrating a minor effect of the route of formation and proving the reliability of the analysis. A detailed analysis of the diffraction line profile and asymmetry showed that there was no shift in the peak maximum occurring during the transition between any of the three profiles, indicating no shift in the lattice parameters associated with it. Lattice parameters calculated from this position were identical for both materialsa = b = 9.361 Å and c = 6.833 Åand lower than both the values for mineral HAp (a = b = 9.418 Å and c =

6.875 Å) by 0.6% and the typical values for synthetic HAp reported in the literature56 (a = b = 9.370 nm and c = 6.881 nm) by up to 0.1%. Although most materials abundant with an amorphous component exhibit the expanded lattice compared to their high crystallinity counterparts, in part because of the surface voids; this distinction is not clear-cut, and amorphization caused by the surface relaxation due to interfacial tension, for example, can contract the lattice and produce locally densified solid pockets. Still, this effect is more pronounced in metallic materials, whereas ceramics, including HAp, undergo surface reconstruction whose effect on the lattice constants and translational symmetry cannot be easily intuited. In contrast to the (304) peak position, the peak asymmetry measured as (Hl − Hr)/(Hl + Hr), where Hl and Hr are the left and the right components of the total peak halfwidths, respectively, indicated considerable skewing toward lower angles. Moreover, a sigmoid function was observed in the inverse of the degree of this skewing during the progression of the transformation of ACP into HAp in the precipitate formed at a higher DS. As seen in Figure 5c, starting from 26%, this degree decreased down to 8.6% after 90 min, 6.3% after 120 min, and eventually reached 5.4% at the point at which HAp was detected in the system. Interestingly, the same degree of skewing was noticed in HAp upon its first appearance in the 14498


Research Article


depending on the exact nature of nucleation and its limiting factors (interface vs diffusion), is composed of two additive factors, nd and nn, the former of which represents the dimensionality of the growing units, while the latter represents the time dependence of the nucleation. With the value of nn equaling 0 for heterogeneous nucleation, as opposed to 1 for the homogeneous, nd has the value of 2, denoting planar growth, as in agreement with the platelet morphology of crystals that is typical for apatite in the bone. According to a different categorization,61 the Avrami exponent value of 2 should be indicative of the edge-controlled nucleation, as opposed to 1 for the surface and 3 for point sites. In both scenarios, uni- or biaxial, anisotropic growth, complying with the hexagonal lattice of HAp and the favored growth along the c-axis, is indicated. Aside from showing the expected dominance of OH− bands in the hydrated gels and the presence of all the major vibration modes of HAp (Figure 8a,b), the FT-IR analysis of the precipitates aged for different periods of time showed that there was a distinct blue shift of the asymmetric ν3(PO43−) mode centered at 1017 cm−1 for the ideal phosphate tetrahedron (Figure 8c), the dominant vibration in the IR spectrum of HAp, accompanying the aging process and evolution toward the HAp structure in the precipitate formed at a higher DS. In contrast, this shift was absent in the precipitate formed at the lower DS (Figure 8d), thus eliminating the reduced hydrogen bonding of the phosphate tetrahedra due to solvent evaporation as a possible cause. A more probable scenario indicates the transition of the TCP-resembling ACP to a HAp-resembling ACP in the course of the crystallization of HAp. In this scenario, a reduced coordination of the phosphate tetrahedra should accompany this transition, proceeding despite the continual increase in the Ca/P stoichiometric ratio from 1.3 to 1.5 to 1.67, as one moves from the Posner’s cluster structure to TCP-resembling ACP to HAp-resembling HAp, respectively (Figure 4). Another possible effect pertains to the fact that the increased concentration of off-center phonon scattering on defects, more pronounced in the amorphous structure, leads to characteristic shifts to lower frequencies; as the structure partially recovers during its transformation to crystalline HAp, the vibrational mode also shifts to higher frequencies. Even if these explanations do not match the real effects, a greater degree of structural change can be invariably concluded for the transformations involving the precipitate formed at a higher DS. The same effect is confirmed by the analysis of the symmetric, ν4(PO43−) bending mode centered at 567 cm−1 for the ideal phosphate tetrahedron; namely, whereas a doublet with insignificant intensity fluctuations was detected during the crystallization of HAp from the precipitate formed at a lower DS (Figure 9a), the disappearance of the doublet between 5 and 30 min time points and its reappearance at the 60 min time point as well as the increasing intensity of absorbance throughout the process were detected during the crystallization of HAp from the precipitate formed at a higher DS (Figure 9b). Another major difference between the FT-IR spectra of the two precipitates was the sharper and more distinct difference between the adjacent symmetric, nondegenerated stretching mode, ν1, and the asymmetric, triply degenerated stretching mode, ν3, bands centered at 938 and 1017 cm−1, respectively, in the precipitate formed at a lower DS compared to a continuous, indistinct band structure observed in the precipitate formed at a higher DS. Such IR band splitting is a typical accompaniment of the amorphous-to-crystalline transitions in CPs.62 The major

material precipitated at the lower DS. However, in this case, the change in the degree of skewing was not as consistent and was also highly fluctuant, agreeing with the idea of the presence of at least two different amorphous phases and of multiple competing nucleation and growth mechanisms at work, as opposed to the smoother transition in the precipitate formed at a higher DS, following the ideal nucleation lag time pattern, such as that observed by Boskey and Posner.57 In general, the asymmetric broadening of diffraction peaks can be caused by the random distribution of a polarized screw or other types of ordered dislocations.58 The broad distribution of lattice parameters due to the presence of the amorphous phase, which, as already mentioned, would be virtually undetectable in diffractograms because of its crystallographic resemblance to HAp, can present another reason for this asymmetry. Hence, to differentiate the lattice strain effects from the crystallite domain size effects on the line broadening, WH plots were constructed for diffraction peaks in the 0.1−0.5 Å−1 range of the reciprocal lattice spacing, d, of the final, HAp product obtained through prolonged aging of precipitates 1 and 2, that is, following two different reaction pathways (Figure 5d). Despite the negative intercept, disabling the derivation of the crystalline domain size, the slope of the linear fit could still be used to deduce the lattice strain,59 which was estimated at ⟨ε02⟩1/2 = 2.24 and 2.57% using eqs 3 and 4 for HAp formed at lower and higher DSs, respectively. Additionally, the inability to derive a correlation between data points in broader ranges of the lattice spacing, even after classifying the data points into axial families of planes, for example, (11l), (31l), (00l), and so forth, is strongly suggestive of anisotropic line broadening, an effect that is expected in view of the directional growth that the hexagonal lattice of HAp is intrinsically prone to. A frequent cause of asymmetric peak broadening that varies for different reflections and disobeys whole pattern fitting is stacking faults, a high density of which was observed during the TEM analysis of HAp nanoparticles (Figure 6). Taking into calculation both low- and

Figure 6. TEM imaging of stacking faults and kinks in the structurally defective region of the lattice of a HAp nanoparticle. An extrinsic stacking fault and a kink are denoted within red and white dashes, respectively.

high-index planes outside the range given in Figure 5d lowers the estimated strain by two orders of magnitude, down to ⟨ε02⟩1/2 = 0.22%, and provides a more realistic value. It should be noted that the ability to undergo immense lattice strains is a hallmark of HAp and a property thanks to which it can accommodate a plethora of foreign ions and act as a mineral reservoir for the body in the bone. With the calculated value of 1.8, the Avrami constant, n, corresponding to the recrystallization of HAp via an amorphous intermediate, is closest to the integer of 2 (Figure 7) and is in agreement with our previous estimation of this factor for brushite formation during the hardening of a β-TCP cement.60 This exponent, although having an ambiguous meaning 14499


Research Article


Figure 7. Linear fit of the extent of crystallization of HAp following its amorphization/gelation as a function of time in the 30−120 min time window of the reaction, along with the selected kinetic parameters derived from the fit, including the Avrami exponent, n, and the Avrami rate constant, k. The reaction extent, α, was measured on the (304) diffraction peak of HAp.

Figure 8. FT-IR spectra of the fresh precipitates 1 (a,c) and 2 (b,d), showing total spectra along with the assignment of the most prominent bands (a,b) and the evolution of the ν3(PO43−) band (c,d) as a function of the aging time in contact with the parent solution.

cm−1, indicating B type carbonated HAp, the dominant form under room-temperature synthesis regimens and involving PO43− → CO32− substitution. In contrast, A type HAp usually forms at elevated temperatures and involves OH− → CO32− substitution, whereas biological apatite is considered a

hydroxyl vibration modes detected were the hydrogen-bonded OH− stretch with the broad absorbance in the 3200−3550 cm−1 range and the H−O−H scissors bending mode at ∼1600 cm−1 (Figure 8a,b). The major carbonate band, the asymmetric ν3(CO32−) stretch, was present with the maximum at 1460 14500


Research Article


Figure 9. FT-IR spectra of the fresh precipitates 1 (a,c) and 2 (b,d), showing the evolution of the asymmetric bending mode ν4(PO43−) (a,b) and ν3(CO32−) (c,d) bands as a function of the aging time in contact with the parent solution. Spectra in (c,d) are manually shifted along the y-axis to avoid the overlap.

combination of A and B types.63 While the A type is recognized by the given carbonate band shifted to ∼1530 cm−1, the mixed A and B type adopts the middle range, between 1460 cm−1 (B type) and 1530 cm−1 (A type).64 Interestingly, this carbonate band was markedly more pronounced in the precipitate forming at a higher DS (Figure 9c,d). It appears as if a more rapid crystal growth leads to the capture of a greater amount of CO32− ions, whose intracrystalline presence is otherwise minimized when more time is allowed for the atomic growth units to rearrange in space. In total, the transitions indicative of the amorphous-to-crystalline transition detected in the FT-IR analysis preceded in time those observed under XRD, suggesting that, naturally, the short-scale atomic and molecular-scale ordering phenomena precede the long-range crystallographic ordering. 3.2. Evolution of the Reamorphized Precipitates. To further explore the crystallization kinetics and mechanisms, the two precipitates formed at different DS values were washed and dried immediately after the precipitation to preserve the ACP structure in the precipitate formed at a higher DS. The precipitate formed at a lower DS formed the HAp structure regardless of whether aging was extended or none. The two powders, HAp and ACP, were then subjected to gelation and turned into amorphous pastes by admixing an aqueous NaH2PO4 solution with an appropriate acidity (pH 5.0), after which their solidification was followed in situ. The amount and the acidity of the liquid phase in this paste formation process

were carefully chosen and set between an overly low free proton concentration, which would be incapable of amorphizing HAp, and an overly high free proton concentration, which would exceed the hydroxyls released upon the dissolution of HAp and initiate the formation of brushite upon solidification. The partial dissolution and an increase in the diffusivity of the particulate units upon gelation foster grain coarsening, a type of growth that relies heavily on the aforementioned hydration and ionic mobility at the particle/solution interface. The average particle sizes calculated from SEM histograms displayed significantly lower numbers for the amorphous precursor precipitate as compared to crystalline HAp (Figure 10). Moreover, while gelation and subsequent setting decreased the nanoparticle size of HAp by a minor degree, that is, from 93.5 to 74.4 nm, they led to an almost fivefold increase in the nanoparticle size of ACP, that is, from 24.5 to 107.1 nm. This is a direct evidence of the higher reactivity of the amorphous phase of CPs as compared to the crystalline phase. Being the precursor for biological apatite during biomineralization and a thermodynamically logical intermediate in the ripening of HAp per the Ostwald law of stages, ACP is naturally more prone to structural and morphological transformation than the final and the most stable of the CP phases in this sequence of events: HAp. XRD studies, first of all, demonstrated that HAp and ACP powders both became amorphized upon gelation, displaying indistinct diffraction patterns of the amorphous intermediate 14501


Research Article


(Figure 12c). Different concentrations of ultrafine crystalline and semicrystalline domains remaining in the material following amorphization and acting as nucleation centers present the probable basis for this memory effect. Remarkable is also the ability of both crystalline phases to be transformed back to the amorphous state under ambient and moderately acidic conditions. With pH ≥ 5, these conditions were less acidic than those created by bone-lysing osteoclasts, which drop pH down to 3−465 and thus must be physiologically relevant. 3.3. Biological Effect Analysis. To test whether the memory effect observed at the microstructural level applies to the material/cell interaction, four powders were analyzed in parallel: two CP powders obtained by drying precipitates 1 and 2, labeled A and B, respectively, and two CP powders obtained by drying the precipitates 1 and 2 aged for 3 h in the parent solution, that is, until they formed an indistinct HAp phase, labeled A → C and B → C, respectively (Figure 3a,b). As shown in Figure 13a, the gene expression of the osteogenic differentiation transcription factor, Runx2, in MC3T3-E1 fibroblasts treated with no chemical osteoblastic differentiation agents and only 5 mg/mL of the given particles was lower than in cells treated with no particles and chemical osteoblastic differentiation agents, that is, 50 μg/mL AA and 10 mM βglycerophosphate. To confirm the differentiation of the MC3T3-E1 cells toward the osteoblastic phenotype, the expression of the housekeeping gene, RNA polymerase POLR2, was measured (Figure 13b). Similar levels of expression in the conventionally differentiated cells and in cells subjected to the treatment with the CP powder expressing the least amount of Runx2, along with no expression in undifferentiated cells, indicated a successful transition toward the osteoblastic phenotype. More importantly, Runx2 expression levels in cells treated with the two structurally indistinct forms of HAp, A → C and B → C were significantly different from each other. Moreover, the expression in cell population treated with either of these two powders was highly similar to that in the cells treated with their respective precursors. That is, no statistically significant difference was detected between the expression in A and A → C populations or between the expression in B and B → C populations. At the same time, expressions in both A and B populations were closest in value to A → C and B → C populations, respectively, than to any other cell population tested. This correlation serves as a direct evidence in favor of

Figure 10. Average particle size determined from SEM histograms for HAp and ACP powders obtained by precipitation and for the corresponding materials after being gelled and completely hardened. n refers to the sample size. Data points represent mean, while error bars represent standard deviation. Asterisks connecting comparable sample groups denote statistical significance: ** represents very statistically significant difference (p < 0.01), while *** represents extremely statistically significant difference (p < 0.0001).

(Figure 11a,b). Although the precursor ACP powder displayed two distinct short-range order humps indicative of the aforementioned TCP-resembling amorphous structure, the further amorphized powder displayed a completely flat pattern indicative of a lesser short-range order than the initial ACP and of an amorphous structure possessing HAp-like stoichiometry. Both materials transformed into HAp as the final product of the solidification reaction, albeit at different rates. Thus, the material that had been HAp prior to amorphization transformed back to HAp faster than the one that has been initially amorphous, indicating the presence of a definite memory effect. Specifically, as shown in Figure 12a,b, while the material that had been HAp prior to amorphization yielded HAp between 1 and 2 h of the setting time, the initially amorphous material yielded HAp 1 h later on average. The rate of increase of the integrated intensities of the two most prominent diffraction peaks in the 2θ range studied, 20−45°, (002) and (211), was correspondingly higher in the reaction of hardening of the material that had been HAp prior to amorphization than in the reaction of hardening of the initially amorphous material

Figure 11. XRD patterns of the two types of materials investigated for ACP → HAp phase transition, one of which (a) was HAp initially, before transforming to ACP in the gelation stage and then back to HAp, and the other one of which (b) was ACP initially, before remaining in the ACP form during gelation and then transforming to HAp. Patterns of the gelled materials were taken immediately after the onset of gelation and the setting reaction, whereas the patterns of the final, fully set gels were taken 80 h later. 14502


Research Article


Figure 12. Time lapse, in situ XRD patterns taken over 2−3 h of the transformation of the crystalline HAp amorphized by mixing with 2 wt % aqueous NaH2PO4 at pH 5.0 (a) and of the amorphous CP powder that underwent the same gelation treatment (b), along with the increase in the integrated intensities of (002) and (211) reflections of HAp as a function of the setting time (c).

the biological relevance of the aforementioned memory effect in CPs. The percentage of MC3T3-E1 cells uptaking the CP particles exceeded 98% for all the sample groups based on flow cytometry data (Figure 13c). Interestingly, although no statistically significant difference was observed between the final HAp powders, A → C and B → C, and their precursors, A and B, respectively, and no statistically significant difference was observed between the two different precursors either, a statistically significant difference was observed between the uptake efficiency of the two final powders, A → C and B → C, in spite of the extremely high uptake values for both (Figure 13c). These results indicate that the pathway of formation of a CP compound significantly affects its interaction with cells as well as that there could be a greater similarity in response between a precursor and its product than between two indistinct products formed through different reaction kinetics. That this finding applies not only to eukaryotic but also to prokaryotic cells was evident following the liquid broth E. coli inoculation assay run in the presence of the two indistinct forms of HAp, A → C and B → C, and their precursor precipitates, A and B; namely, no statistically significant difference was detected between the number of colony-forming units in infected broths incubated overnight with the following comparative powders: A versus B, A versus A → C, and B versus B → C (Figure 13d). However, this bacterial number was significantly different between broths treated with A → C and broths treated with B → C (Figure 13d), reiterating the biologically different properties of HAp depending on its formation pathway.

The question naturally posable at this point is what structural features of the two seemingly indistinct HAp powders are actually distinct and causing the observed difference in the biological response. As seen in Figure 13a, the more active powder from the osteogenic standpoint is the one spending more time in its amorphous form before transitioning to the final, crystalline form, that is, B → C. Apparently, this intrinsic activity of the amorphous form becomes retained to a greater degree in this powder compared to the one spending lesser time in the amorphous state during formation, that is, A → C. It could be hypothesized that the more active and soluble former powder is typified by a more intense dissolution and reprecipitation of the surface growth units in the form of solitary ions, ionic complexes, Posner’s clusters, or bigger nanoparticles. One such difference in the surface dynamics was captured during an immunofluorescent cell/particle interface analysis (Figure 14a), showing a greater concentration of finer nanoparticulate units separating from and anchoring onto the bigger particle conglomerates in HAp prepared in a process involving a lengthier transition from the amorphous to the crystalline phase, that is, B → C. Although units comparable in size to Posner’s clusters are smaller than those captured in Figure 14a, aggregation phenomena in fine particles, especially favored at low surface charges such as those exhibited by CPs, are usually hierarchical, bearing resemblance in mechanism across multiple scales.66−68 Notwithstanding the possible involvement of other mechanisms, if this idea of hierarchical aggregation of subunits holds, then the surface activity between two internally indistinct powders may be responsible for this drastic difference in their biological response. Thanks to its 14503


Research Article


Figure 13. (a) Comparative effect of two indistinct forms of HAp, A → C and B → C, formed by aging different amorphous precipitates, 1 (A) and 2 (B) on the ACTB-normalized mRNA expression of the osteogenic transcription factor Runx2 in preosteoblastic MC3T3-E1 cells. mRNA expression in the differentiation population was used as a control. mRNA expression was detected by quantitative reverse transcription-polymerase chain reaction relative to the housekeeping gene, β-actin (ACTB). The expression in all the sample groups was very significantly (p < 0.005) downregulated with respect to the control. (b) Comparative expression of the housekeeping gene, RNA polymerase, POLR2 in the lowest Runx2 expressing sample group (A) against the undifferentiated and differentiated control MC3T3-E1 cell populations. (c) Percentage of MC3T3-E1 cells uptaking particles comprising the two indistinct forms of HAp, A → C and B → C, and their precursor precipitates, 1 (A) and 2 (B) after 24 h of coincubation time measured in a flow cytometry analysis. (d) Number of colony forming units of E. coli in untreated, negative control broths and in broths treated overnight with 100 mg/mL of the two indistinct forms of HAp, A → C and B → C, and their precursor precipitates, 1 (A) and 2 (B). Data points are shown as averages [n = 3 × 6 in (a,b), n = 4 in (c) and n = 3 in (d)] with error bars representing standard deviation. Statistically insignificant difference between comparative sample groups (p > 0.05) is marked with n.s. (not significant). Significant (p < 0.05) and very significant (p < 0.005) differences between comparative sample groups are marked with * and **, respectively.

intrinsically hydroxylated and extrinsically hydrated nature, HAp is typified by a more dynamic interface in aqueous media than oxides or metals.69 Also, compared to many other biominerals, including silica and goethite, which all carry a pronounced covalent character, HAp is almost entirely ionic in nature,70 which additionally enforces these dynamic fluctuations and inconstancy of the surface in contact with a solution. This dynamics, determined by the surface structure, extending over dozens of atomic layers, may vary depending on the synthesis route and predispose the material for a unique response. This higher surface volatility of the B → C powder compared to A → C is demonstrated by its lower capacity to bind negatively charged BSA by physisorption (Figure 14b). A surface undergoing a greater degree of ionic exchange with the dispersion medium would allow for a more intense desorption of the adsorbate than the more stable surface, hinting at the possible key to explaining the difference between the two seemingly identical materials, A → C and B → C, differing in their route of formation. The lower entropic contribution to surface energy and the more intense surface hydration are additional contributors to the higher desorption potential of the amorphous phase. Interestingly, no difference in the adsorption capacity toward positively charged lysozyme was detected between the different C powders (i.e., A → C and B → C) and

their precursors (Figure 14c), suggesting that lighter and more mobile Ca2+ ionsthe predominant binding sites for negatively charged BSA as opposed to PO43− groups predominantly binding lysozymeare the major ions transferring across the solid/liquid interface in this exchange process. This is in agreement with the network of phosphate ions being the central structural support in HAp,71,72 a sort of skeleton filled with smaller calcium ions and hydroxyls. One such structure is analogous to that of specific, typically low pressure polymorphs of other ionic crystals, including entropically stabilized α-AgI, in which case iodide anions provide a rigid framework to the lattice and silver cations diffuse through it at comparatively high rates.73 The same effect explains the accelerated transformation of ACP to HAp in PBS, a phosphate-rich solvent, when compared to pure water;74 namely, by electrostatically attracting Ca2+ ions across the particle/solution interface, the phosphate solutes speed up the surface-mediated processes that lead to phase transformation. Dynamics of the CP particle surface is incontrovertible, and although it gradually drops as the system progresses down the sequence Ca 2+ (aq) / HxPO4x−3(aq)/OH− ⇆ prenucleation clusters ⇆ liquid ACP ⇆ solid ACP ⇆ anhydrous crystalline polymorphs, the question hinted at by these results is how much of this dynamics may be preserved in a system depending on the 14504


Research Article


Figure 14. (a) Immunofluorescent optical micrographs showing the interaction of the HAp powders C obtained through different pathways of formation: A → C and B → C. HAp particles and their conglomerates are stained in green, collagen in red, and cell nucleus in blue. Adsorption isotherms showing concentration-dependent binding of BSA (b) and lysozyme (c) on two indistinct forms of HAp, A → C and B → C, and their precursor precipitates, 1 (A) and 2 (B) after 1 h of incubation time. Significantly (p < 0.05) and extremely significantly (p < 0.0001) different expressions between A → C and B → C sample groups are marked with * and ***, respectively.

must be different depending on the characteristics of its initial “birth”. This primal process leaves traces in a material that determine its fate in interactions with chemical agents and biological entities, and the goal of this study is to provoke interest in reaching a deeper understanding of this enigmatic effect. Even though the mercurial material that HAp is has been dropping in favor of more structurally flexible or chemically robust materials, such as polymers or metal oxides, it continues to perplex in accordance with its etymology, originating from the Greek απαταο: “I am misleading”. Hope underlying this study is that the teleological background of these unusual memory phenomena may contain the sources for novel properties of HAp and other calcium phosphates applicable across a wide array of disciplines.

history of its formation, rendering seemingly indistinct materials to behave differently, from both physicochemical and biological standpoints.

4. SUMMARY Materials science literature contains a decent number of reports evidencing memory effects in solid structures. They include the effect of supersaturation and other growth conditions in the past on the mode and the rate of future crystal growth,75 precursor phenomena in the martensitic transformation between different polymorphs in shape memory alloys,76 magnetic and electrical hystereses in polarizable materials, for example, memristors,77 and others. In spite of this, memory effects have neither been explored nor exploited in amounts equaling their indisputably large applicative potential in the context of biomaterials science. In this study, we explored some of the properties affecting, accompanying, and entailing the memory effect observed in HAp, the most abundant solid in the mammalian bodies and traditionally the most beloved of all the biological materials by the biomaterials science community. Thus, we showed that nascent calcium phosphate precipitates as a dual amorphous phase and subsequently transforms to crystalline HAp through either internal lattice rearrangements or reprecipitation of constantly dissipating units in contact with an aqueous solution. Recrystallization rates following gelation depended on the rate of crystallization of the original powders and were higher for the initially crystalline material than for its amorphous counterpart, indicating the presence of a memory effect in the material. The observed difference in the biological response between two apatites with different histories of formation indirectly demonstrates that this material is not only a static structure in its native, aqueous environment but a dynamic one too, constantly reforming itself in a process that

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Vuk Uskoković: 0000-0003-3256-1606 Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS NIH grant R00-DE021416 is acknowledged for support. We thank Shreya Ghosh and Eric Huynh from the Uskokovic Lab at the University of Illinois in Chicago and Chapman University, respectively, for the synthesis and a portion of characterization of CP powders and gels. Confocal microscopy data were acquired at the Nikon Imaging Center at UCSF. HRTEM analyses were performed at the National Center for Electron Microscopy (Berkeley, CA) supported by the Office of 14505


Research Article


activating transcription: T cell memory is maintained by DNA elements which stably prime inducible genes without activating steady state transcription. BioEssays 2017, 39, 1600184. (19) Shu, Y.; Yu, D.; Hu, W.; Wang, Y.; Shen, G.; Kono, Y.; Xu, B.; He, J.; Liu, Z.; Tian, Y. Deep melting reveals liquid structural memory and anomalous ferromagnetism in bismuth. Proc. Natl. Acad. Sci. U.S.A. 2017, 114, 3375−3380. (20) Sevcikova, J.; Goldbergova, M. P. Biocompatibility of NiTi alloys in the cell behavior. BioMetals 2017, 30, 163−169. (21) Lester, B. T.; Baxevanis, T.; Chemisky, Y.; Lagoudas, D. C. Review and perspectives: shape memory alloy composite systems. Acta Mech. 2015, 226, 3907−3960. (22) Yu, J.; Xia, H.; Teramoto, A.; Ni, Q.-Q. The effect of hydroxyapatite nanoparticles on mechanical behavior and biological performance of porous shape memory polyurethane scaffolds. J. Biomed. Mater. Res., Part A 2018, 106, 244−254. (23) Bai, J.; Shi, Z. Dynamically Cross-linked Elastomer Hybrids with Light-Induced Rapid and Efficient Self-Healing Ability and Reprogrammable Shape Memory Behavior. ACS Appl. Mater. Interfaces 2017, 9, 27213−27222. (24) Balk, M.; Behl, M.; Wischke, C.; Zotzmann, J.; Lendlein, A. Recent advances in degradable lactide-based shape-memory polymers. Adv. Drug Delivery Rev. 2016, 107, 136−152. (25) Chaplin, M. F. The Memory of Water: an overview. Homeopathy 2007, 96, 143−150. (26) Ghosh, S.; Wu, V.; Pernal, S.; Uskoković, V. Self-Setting Calcium Phosphate Cements with Tunable Antibiotic Release Rates for Advanced Bone Graft Applications. ACS Appl. Mater. Interfaces 2016, 8, 7691−7708. (27) Grynpas, M. D.; Omelon, S. Transient precursor strategy or very small biological apatite crystal? Bone 2007, 41, 162−164. (28) Rodriguez, D. E.; Thula-Mata, T.; Toro, E. J.; Yeh, Y.-W.; Holt, C.; Holliday, L. S.; Gower, L. B. Multifunctional role of osteopontin in directing intrafibrillar mineralization of collagen and activation of osteoclasts. Acta Biomater. 2014, 10, 494−507. (29) Semba, I.; Ishigami, T.; Sugihara, K.; Kitano, M. Higher osteoclastic demineralization and highly mineralized cement lines with osteocalcin deposition in a mandibular cortical bone of autosomal dominant osteopetrosis type II: ultrastructural and undecalcified histological investigations. Bone 2000, 27, 389−395. (30) Sasaki, T.; Motegi, N.; Suzuki, H.; Watanabe, C.; Tadokoro, K.; Yanagisawa, T.; Higashi, S. Dentin resorption mediated by odontoclasts in physiological root resorption of human deciduous teeth. Am. J. Anat. 1988, 183, 303−315. (31) Molnar, Z. Development of the parietal bone of young mice. I. Crystals of bone mineral in frozen-dried preparations. J. Ultrastruct. Res. 1959, 3, 39−45. (32) Thyberg, J. Electron microscopic studies on the initial phase of calcification in guinea pig epiphyseal cartilage. J. Ultrastruct. Res. 1974, 46, 206−218. (33) Larsen, M. J. An investigation of the theoretical background for the stability of the calcium-phosphate salts and their mutual conversion in aqueous solutions. Arch. Oral Biol. 1986, 31, 757−761. (34) Mittemeijer, E. J.; Welzel, U. The “state of the art” of the diffraction analysis of crystallite size and lattice strain. Z. Kristallogr. 2008, 223, 552−560. (35) Borkiewicz, O.; Rakovan, J.; Cahill, C. L. Time-resolved in situ studies of apatite formation in aqueous solutions. Am. Mineral. 2010, 95, 1224−1236. (36) Christoffersen, J.; Christoffersen, M. R. Kinetics of dissolution of calcium hydroxyapatite. V. The acidity constant for the hydrogen phosphate surface complex. J. Cryst. Growth 1982, 57, 21−26. (37) Xie, B.; Halter, T. J.; Borah, B. M.; Nancollas, G. H. Tracking Amorphous Precursor Formation and Transformation during Induction Stages of Nucleation. Cryst. Growth Des. 2014, 14, 1659− 1665. (38) Viney, C.; Bell, F. I. Inspiration versus Duplication with Biomolecular Fibrous Materials: Learning Nature’s Lessons without

Science, Office of Basic Energy Sciences of the US Department of Energy under contract no. DE-AC02-05CH11231.

■

REFERENCES

(1) Uskoković, V. The Role of Hydroxyl Channel in Defining Selected Physicochemical Peculiarities Exhibited by Hydroxyapatite. RSC Adv. 2015, 5, 36614−36633. (2) Horiuchi, N.; Madokoro, K.; Nozaki, K.; Nakamura, M.; Katayama, K.; Nagai, A.; Yamashita, K. Electrical conductivity of polycrystalline hydroxyapatite and its application to electret formation. Solid State Ionics 2018, 315, 19−25. (3) Rey, C.; Miquel, J. L.; Facchini, L.; Legrand, A. P.; Glimcher, M. J. Hydroxyl groups in bone mineral. Bone 1995, 16, 583−586. (4) Pasteris, J. D.; Wopenka, B.; Freeman, J. J.; Rogers, K.; ValsamiJones, E.; van der Houwen, J. A. M.; Silva, M. J. Lack of OH in nanocrystalline apatite as a function of degree of atomic order: implications for bone and biomaterials. Biomaterials 2004, 25, 229− 238. (5) Desai, T. A.; Uskoković, V. Calcium Phosphate Nanoparticles: A Future Therapeutic Platform for the Treatment of Osteomyelitis? Ther. Delivery 2013, 4, 643−645. (6) Uskoković, V.; Desai, T. A. Phase Composition Control of Calcium Phosphate Nanoparticles for Tunable Drug Delivery Kinetics and Treatment of Osteomyelitis. I. Preparation and Drug Release. J. Biomed. Mater. Res., Part A 2013, 101, 1416−1426. (7) LaMer, V. K.; Dinegar, R. H. Theory, Production and Mechanism of Formation of Monodispersed Hydrosols. J. Am. Chem. Soc. 1950, 72, 4847−4854. (8) La Mer, V. K. Nucleation in Phase Transitions. Ind. Eng. Chem. 1952, 44, 1270−1277. (9) Nudelman, F.; Pieterse, K.; George, A.; Bomans, P. H. H.; Friedrich, H.; Brylka, L. J.; Hilbers, P. A.; de With, G.; Sommerdijk, N. A. J. M. The role of collagen in bone apatite formation in the presence of hydroxyapatite nucleation inhibitors. Nat. Mater. 2010, 9, 1004− 1009. (10) Nielsen, M. H.; Lee, J. R. I.; Hu, Q.; Han, T. Y.-J.; De Yoreo, J. J. Structural evolution, formation pathways and energetic controls during template-directed nucleation of CaCO3. Faraday Discuss. 2012, 159, 105. (11) Qi, C.; Lin, J.; Fu, L.-H.; Huang, P. Calcium-based biomaterials for diagnosis, treatment, and theranostics. Chem. Soc. Rev. 2018, 47, 357−403. (12) Tao, J.; Pan, H.; Zeng, Y.; Xu, R.; Tang, R. Roles of amorphous calcium phosphate and biological additives in the assembly of hydroxyapatite nanoparticles. J. Phys. Chem. B 2007, 111, 13410− 13418. (13) Shaik, Z. A.; Rambabu, T.; Sajjan, G.; Varma, M.; Satish, K.; Raju, V. B.; Ganguru, S. Quantitative analysis of remineralization of artificial carious lesions with commercially available newer remineralizing agents using SEM-EDX in vitro study. J. Clin. Diagn. Res. 2017, 11, ZC20−ZC23. (14) Zawaideh, F. I.; Owais, A. I.; Kawaja, W. Ability of Pit and Fissure Sealant-Containing amorphous calcium phosphate to inhibit enamel demineralization. Int. J. Clin. Pediatr. Dent. 2016, 9, 10−14. (15) Eres, G. Exploring nanomaterials synthesis on the length scale of fundamental building blocks. Presented at the 19th YUCOMAT Conference, Herceg-Novi, Montenegro, 2017. (16) Challenges at the Frontiers of Matter and Energy: Transformative Opportunities for Discovery Science: A Report from the Basic Energy Sciences Advisory Committee; U.S. Department of Energy; p vi. Retrieved from https://science.energy.gov/?/media/bes/ besac/pdf/Reports/Challenges_at_the_Frontiers_of_Matter_and_ Energy_rpt.pdf (November 2015). (17) Larsen, H. L.; Martín-Coll, L.; Nielsen, A. V.; Wright, C. V. E.; Trusina, A.; Kim, Y. H.; Grapin-Botton, A. Stochastic priming and spatial cues orchestrate heterogeneous clonal contribution to mouse pancreas organogenesis. Nat. Commun. 2017, 8, 605. (18) Bevington, S. L.; Cauchy, P.; Cockerill, P. N. Chromatin priming elements establish immunological memory in T cells without 14506


Research Article

ACS Applied Materials & Interfaces Copying Nature’s Limitations. Curr. Opin. Solid State Mater. Sci. 2004, 8, 165−171. (39) Liu, S.; Weng, W.; Li, Z.; Pan, L.; Cheng, K.; Song, C.; Du, P.; Shen, G.; Han, G. Effect of PEG amount in amorphous calcium phosphate on its crystallized products. J. Mater. Sci.: Mater. Med. 2009, 20, 359−363. (40) Cartwright, J. H. E.; Checa, A. G.; Gale, J. D.; Gebauer, D.; Sainz-Díaz, C. I. Calcium Carbonate Polyamorphism and Its Role in Biomineralization: How Many Amorphous Calcium Carbonates Are There? Angew. Chem., Int. Ed. 2012, 51, 11960−11970. (41) Machon, D.; Daniel, M.; Pischedda, V.; Daniele, S.; Bouvier, P.; LeFloch, S. Pressure-induced polyamorphism in TiO2 nanoparticles. Phys. Rev. B 2010, 82, No. 140102(R). (42) Willinger, E.; Massué, C.; Schlögl, R.; Willinger, M. G. Identifying Key Structural Features of IrOx Water Splitting Catalysts. J. Am. Chem. Soc. 2017, 139, 12093−12101. (43) Ganguli, D. Polyamorphism in Liquids and Amorphous Substances : An Analogue of Polymorphism in Crystalline Solids. Trans. Indian Ceram. Soc. 2009, 68, 65−80. (44) Loerting, T.; Winkel, K.; Seidl, M.; Bauer, M.; Mitterdorfer, C.; Handle, P. H.; Salzmann, C. G.; Mayer, E.; Finney, J. L.; Bowron, D. T. How many amorphous ices are there? Phys. Chem. Chem. Phys. 2011, 13, 8783−8794. (45) Köhler, T.; Turowski, M.; Ehlers, H.; Landmann, M.; Ristau, D.; Frauenheim, T. Computational approach for structure design and prediction of optical properties in amorphous TiO2 thin-film coatings. J. Appl. Phys. 2013, 46, 325302. (46) Tian, M.; Mahjouri-Samani, M.; Eres, G.; Sachan, R.; Yoon, M.; Chisholm, M. F.; Wang, K.; Puretzky, A. A.; Rouleau, C. M.; Geohegan, D. B.; Duscher, G. Structure and Formation Mechanism of Black TiO2 Nanoparticles. ACS Nano 2015, 9, 10482−10488. (47) Yoshida, K.; Miao, L.; Tanaka, N.; Tanemura, S. Direct observation of TiO6 octahedron forming titanate nanotube by advanced transmission electron microscopy. Nanotechnology 2009, 20, 405709. (48) Singh, M.; Thomas, J.; Ramanan, A. Understanding supramolecular interactions provides clues for building molecules into minerals and materials: A retrosynthetic analysis of copper-based solids. Aust. J. Chem. 2010, 63, 565−572. (49) Habraken, W. J. E. M.; Tao, J.; Brylka, L. J.; Friedrich, H.; Bertinetti, L.; Schenk, A. S.; Verch, A.; Dmitrovic, V.; Bomans, P. H. H.; Frederik, P. M.; Laven, J.; van der Schoot, P.; Aichmayer, B.; de With, G.; DeYoreo, J. J.; Sommerdijk, N. A. J. M. Ion-association complexes unite classical and non-classical theories for the biomimetic nucleation of calcium phosphate. Nat. Commun. 2013, 4, 1507. (50) Wu, V. M.; Mickens, J.; Uskoković, V. BisphosphonateFunctionalized Calcium Phosphate Nanoparticles for the Delivery of the Bromodomain Inhibitor JQ1 in the Treatment of Osteosarcoma. ACS Appl. Mater. Interfaces 2017, 9, 25887−25904. (51) Dorozhkin, S. V. Amorphous calcium (ortho)phosphates. Acta Biomater. 2010, 6, 4457−4475. (52) Bates, S. Amorphous materials: A structural perspective. Presentation at the 11th PPXRDPharmaceutical Powder X-ray Diffraction Symposium, 2013. (53) Combes, C.; Rey, C. Amorphous calcium phosphate: synthesis, properties and uses in biomaterials. Acta Biomater. 2010, 6, 3362− 3378. (54) Dorozhkin, S. V. Calcium orthophosphates (CaPO 4 ): occurrence and properties. Prog. Biomater. 2015, 5, 9−70. (55) Zhang, F.; Allen, A. J.; Levine, L. E.; Vaudin, M. D.; Skrtic, D.; Antonucci, J. M.; Hoffman, K. M.; Giuseppetti, A. A.; Ilavsky, J. Structural and dynamical studies of acid-mediated conversion in amorphous-to-calcium-phosphate based dental composites. Dent. Biomater. 2014, 30, 1113−1125. (56) Uskoković, V.; Uskoković, D. P. Nanosized Hydroxyapatite and Other Calcium Phosphates: Chemistry of Formation and Application as Drug and Gene Delivery Agents. J. Biomed. Mater. Res., Part B 2011, 96, 152−191.

(57) Boskey, A. L.; Posner, A. S. Formation of hydroxyapatite at low supersaturation. J. Phys. Chem. 1976, 80, 40−45. (58) Groma, I.; Monnet, G. Analysis of asymmetric broadening of Xray diffraction peak profiles caused by randomly distributed polarized dislocation dipoles and dislocation walls. J. Appl. Crystallogr. 2002, 35, 589−593. (59) Zhao, Y.; Zhang, J. Microstrain and grain-size analysis from diffraction peak width and graphical derivation of high-pressure thermomechanics. J. Appl. Crystallogr. 2008, 41, 1095−1108. (60) Uskoković, V.; Rau, J. V. Nonlinear Oscillatory Dynamics of the Hardening of Calcium Phosphate Cements. RSC Adv. 2017, 7, 40517− 40532. (61) Cahn, J. W. Transformation kinetics during continuous cooling. Acta Metall. 1956, 4, 572−575. (62) Poralan, G. M., Jr.; Gambe, J. E.; Alcantara, E. M.; Vequizo, R. M. X-ray diffraction and infrared spectroscopy analyses on the crystallinity of engineered biological hydroxyapatite for medical application. IOP Conf. Ser.: Mater. Sci. Eng. 2015, 79, 012028. (63) Rey, C.; Collins, B.; Goehl, T.; Dickson, I. R.; Glimcher, M. J. The Carbonate Environment in Bone Mineral: A ResolutionEnhanced Fourier Transform Infrared Spectroscopic Study. Calcif. Tissue Int. 1989, 45, 157−164. (64) Berzina-Cimdina, L.; Borodajenko, N. Research of Calcium Phosphates using Fourier Transform Infrared Spectroscopy. In Infrared SpectroscopyMaterials Science, Engineering and Technology; Theophanides, T., Ed.; InTech: Rijeka, 2012. (65) Silver, I. A.; Murrills, R. J.; Etherington, D. J. Microelectrode studies on the acid microenvironment beneath adherent macrophages and osteoclasts. Exp. Cell Res. 1988, 175, 266−276. (66) Matijevic, E. Preparation and Properties of Uniform Size Colloids. Chem. Mater. 1993, 5, 412−426. (67) Niederberger, M.; Cö lfen, H. Oriented attachment and mesocrystals: non-classical crystallization mechanisms based on nanoparticle assembly. Phys. Chem. Chem. Phys. 2006, 8, 3271−3287. (68) De Yoreo, J. J.; Gilbert, P. U. P. A.; Sommerdijk, N. A. J. M.; Penn, R. L.; Whitelam, S.; Joester, D.; Zhang, H.; Rimer, J. D.; Navrotsky, A.; Banfield, J. F.; Wallace, A. F.; Michel, F. M.; Meldrum, F. C.; Cölfen, H.; Dove, P. M. Crystal Growth. Crystallization by particle attachment in synthetic, biogenic, and geologic environments. Science 2015, 349, aaa6760. (69) Wopenka, B.; Pasteris, J. D. A mineralogical perspective on the apatite in bone. Mater. Sci. Eng., C 2005, 25, 131−143. (70) Gebauer, D.; Kellermeier, M.; Gale, J. D.; Bergström, L.; Cölfen, H. Pre-nucleation clusters as solute precursors in crystallization. Chem. Soc. Rev. 2014, 43, 2348−2371. (71) Orme, C. A.; Giocondi, J. L. The Use of Scanning Probe Microscopy to Investigate Crystal-Fluid Interfaces. In Perspectives on Inorganic, Organic, and Biological Crystal Growth: From Fundamentals to Applications; Skowronski, M., DeYoreo, J. J., Wang, C. A., Eds.; American Institute of Physics: Melville, 2007. (72) Uskoković, V.; Li, W.; Habelitz, S. Amelogenin as a Promoter of Nucleation and Crystal Growth of Apatite. J. Cryst. Growth 2011, 316, 106−117. (73) Mendels, D.; McCarty, J.; Piaggi, P. M.; Parrinello, M. Searching for Entropically Stabilized Phases: The Case of Silver Iodide. J. Phys. Chem. C 2018, 122, 1786−1790. (74) Chatzipanagis, K.; Iafisco, M.; Roncal-Herrero, T.; Bilton, M.; Tampieri, A.; Kröger, R.; Delgado-López, J. M. Crystallization of citrate-stabilized amorphous calcium phosphate to nanocrystalline apatite: a surface-mediated transformation. CrystEngComm 2016, 18, 3170−3173. (75) Pantaraks, P.; Matsuoka, M.; Flood, A. E. Effect of Growth Rate History on Current Crystal Growth. 2. Crystal Growth of Sucrose, Al(SO4)2 12H2O, KH2PO4, and K2SO4. Cryst. Growth Des. 2007, 7, 2635−2642. (76) Niitsu, K.; Kimura, Y.; Kainuma, R. Transformation entropy change and precursor phenomena in Ni-rich Ti-Ni shape memory alloys. J. Mater. Res. 2017, 32, 3822−3830. 14507


Research Article

ACS Applied Materials & Interfaces (77) Dongale, T. D.; Mohite, S. V.; Bagade, A. A.; Kamat, R. K.; Rajpure, K. Y. Bio-mimicking the synaptic weights, analog memory, and forgetting effect using spray deposited WO3 memristor device. Microelectron. Eng. 2017, 183, 12−18.

14508


On Grounds of the Memory Effect in Amorphous and Crystalline

Recommend Documents