Thermal Treatment of FlameSynthesized ... - Wiley Online Library

2 downloads 0 Views 2MB Size Report
Thermal Treatment of Flame-Synthesized Amorphous. Tricalcium Phosphate Nanoparticles. Nicola Do¨belin, w,z. Tobias J. Brunner, y,ww. Wendelin J. Stark,.
J. Am. Ceram. Soc., 93 [10] 3455–3463 (2010) DOI: 10.1111/j.1551-2916.2010.03856.x r 2010 The American Ceramic Society

Journal

Thermal Treatment of Flame-Synthesized Amorphous Tricalcium Phosphate Nanoparticles

Nicola Do¨belin,w,z Tobias J. Brunner,y,ww Wendelin J. Stark,y Martin Fisch,z Egle Conforto,J and Marc Bohnerz z

RMS Foundation, CH-2544 Bettlach, Switzerland

y

Department of Chemistry and Applied Biosciences, Institute for Chemical and Bioengineering, ETH Zu¨rich, CH-8093 Zu¨rich, Switzerland z

Institut fu¨r Geologie, Universita¨t Bern, CH-3012 Bern, Switzerland

J

Fe´de´ration de Recherche en Environnement pour le De´veloppement Durable (FR-EDD), FR CNRS 3097, Centre Commun d’Analyses, F-17071 La Rochelle Cedex 9, France

possess a higher SSA than phases obtained by high-temperature processes (c). In case of high-temperature formation, on the other hand, sinter effects lead to a reduction of the SSA, an increased particle growth, a high degree of crystallinity, and densification and shrinkage by particle fusion. These sinter processes often result in hard and brittle solid materials. b-tricalcium phosphate (b-TCP; Ca3(PO4)2) is one of the most prominent CaP phases in modern implants. It is classically obtained from various precursor materials with a Ca:P ratio of 1.50 by thermal treatment above 7501C.9,10 Above 11201–11501C, it transforms into the high-temperature polymorph a-TCP11–14 by undergoing a reconstructive reversible phase transformation.15 The reverse transformation to b-TCP can be suppressed by rapid cooling to room temperature, which maintains a-TCP as a metastable phase. In fact, it was shown that even at moderate cooling rates ( 21C/min) no reverse transformation from a- to b-TCP occurs.11,14,15 Another less known high-temperature polymorph of a- and b-TCP was observed by optical microscopy16 and neutron diffraction17 at temperatures above 14301C. The latter authors refined the crystal structure; however, to our best knowledge no atomic coordinates have been published to date. The phase has been called  -TCP,12 or more recently a0 -TCP.17 This ‘‘super-a-TCP,’’16 a phase could not be maintained in a metastable form at room temperature by quenching. Metastable a-TCP is soluble and reacts with water to give calcium-deficient hydroxyapatite (CDHA) at low temperatures.18–20 Several bone cements have been developed based on this setting reaction.21–23 Besides solubility-controlling chemical parameters such as pH and chemical composition of the liquid component, the kinetics of the hydration reaction depend on physical parameters such as the SSA and particle size of the solid component.24–26 Being able to control these parameters allows to control the setting and handling properties of the cement without having to add retarding or accelerating agents. However, due to the pronounced sintering effects (crystallite growth, particle fusion, reduction of the SSA) in the stability field of a-TCP above 11501C, it is not possible to obtain nanoparticulate a-TCP with a high SSA by conventional high-temperature preparation route. Milling of larger particles is possible, but it further increases the enormous cost of a-TCP production and may require further thermal treatment.27 A theory known as the Ostwald step rule28 provides a solution to this problem. It states that in the crystallization from a solution or of an amorphous solid phase, the least stable phase often occurs first, followed by transformations to thermodynamically more stable phases. Indeed, it was shown by Somrani et al.29 that the crystalline phases of a precipitated amorphous TCP (ATCP) powder obtained after calcination in the range

Flame-spray synthesis was used to produce nanoparticulate X-ray amorphous tricalcium phosphate. Upon heating, the material crystallized at temperatures between 5251 and 6001C, yielding a-tricalcium phosphate (TCP) with minor amounts of b-TCP and hydroxyapatite (HA). Further heating induced a gradual transformation of a- to b-TCP accompanied by crystallite growth and particle fusion. a-TCP was completely transformed into b-TCP at 9501C. The high-temperature polymorph a0 -TCP was not observed. In transmission electron microscopy and selected area electron diffraction analyses, the raw material appeared completely amorphous and samples heated previously to 6001 and 7001C comprised single-crystalline particles of a- , b-TCP, and HA. The results of this study demonstrate that nanoparticulate a-TCP can be obtained from flame-spray-synthesized amorphous nanoparticles at temperatures where sintering effects such as particle growth and fusion are moderate, if not negligible. I. Introduction

S

several decades calcium phosphate (CaP) ceramics have been a popular synthetic alternative to bone autografts, allografts, and xenografts from natural sources in the reconstruction of large bone defects.1–4 Some of these synthetic ceramics show excellent biocompatibility due to their similarity to the mineral phase of bone. The degradation rate after implantation depends on the solubility of the CaP phases in the human body,5 and on the physical properties of the implant such as the degree of porosity, pore diameter, pore interconnectivity, specific surface area (SSA), and particle size.6 The biological performance and resorption characteristics of bone substituting CaP products can thus be tailored by optimizing these parameters,7 and by combining phases with favorable properties. CaP materials are obtained from various processes including: (a) precipitation from aqueous solutions, (b) cement reactions of solid phases with an aqueous component, and (c) solid-state dehydration reactions or phase transformations at high temperature.8 Several physical properties of the synthesis products depend on the formation conditions: phases synthesized by low-temperature processes (a) and (b) are less crystalline and INCE

J. Ferreira—contributing editor

Manuscript No. 26512. Received July 21, 2009; approved April 17, 2010. This work was supported by the Gebert Ru¨f Foundation under grant no. GRS-048/04. w Author to whom correspondence should be addressed. e-mail: nicola.doebelin@ rms-foundation.ch ww Present address: Biotronik AG, CH-8180 Bu¨lach, Switzerland

3455

3456

Journal of the American Ceramic Society—Do¨belin et al.

from 6001 to 8001C comprise more than 90 mol% a-TCP, even though the maximum temperature of the formation process was far below the stability field of a-TCP. However, Li et al.30 showed that the crystalline phase obtained after heating of precipitated amorphous TCP at 8001C depended on synthesis parameters, namely on the pH of the mother solution, and on the aging time of the precipitate in the mother solution. Pure a-TCP phase was obtained by heating precipitated amorphous TCP prepared at highly alkaline pH, and/or aged for a short time. Mildly alkaline pH of the mother solution, as well as longer aging times, favored the direct formation of b-TCP instead. Their results demonstrate that short-range order and synthesis parameters of X-ray amorphous TCP affect the phase evolution occurring at early stages of temperature-induced crystallization. The assumption that any amorphous TCP is bound to form metastable a-TCP as a first crystalline phase at low temperature appears to be not valid. According to Li et al.,30 amorphous TCP materials may possess different degrees and types of short-range structures, which predetermine the crystalline phase occurring first after heat-induced crystallization. This theory could also explain why a0 -TCP was so far never observed below 14501C, despite the prediction of the Ostwald step rule. In the present study, in situ high-temperature X-ray powder diffraction (XRD) was used to monitor the phase composition and crystallite growth induced by thermal treatment up to 10001C of a nanoparticulate amorphous TCP powder prepared by flame-spray synthesis.31 The preparation method of the material was vastly different from the precipitation methods frequently described in the literature. It was thus of interest to investigate how an amorphous TCP material formed at a very high temperature in a water-free environment would crystallize, and whether a0 -TCP could be obtained. With increasing particle growth and fusion at higher temperatures, particularly in the stability field of a-TCP above 11501C, the material starts to lose its unique and outstanding features of nanosized weakly agglomerated particles. The temperature range analyzed in the present study was thus restricted to temperatures up to 10001C and we concentrated on the temperature of the initial crystallization. Preliminary results of the study focusing on the occurrence of a0 -TCP,32 on variations of the Ca:P ratio,33 and on the reactivity of as-prepared phases34 have been published before. In this follow-up study, we provide a more in-depth discussion of the results. We also used scanning and transmission electron microscope techniques to provide a thorough characterization of the material, in particular the phase composition, particle morphology, and crystal domain size, in its amorphous and crystalline states.

Vol. 93, No. 10

in y/y configuration using Ni-filtered CuKa radiation and a fixed divergence slit of 1/81 opening. A heating chamber (HTK-1200, Anton Paar, Ostfildern, Germany) was attached to the goniometer. The samples were prepared in a ceramic sample holder spinning at approximately 1 rps. Data sets were collected from 91–651 2y with a step size of 0.0161 2y and a counting time of 1 s per step at the following temperatures: 251, 5001, 5251, 5501, 5751, 6001, 6501, 7001, 7501, 8001, 8501, 9001, 9501, 10001C, and again at 251C. The heating rate was set to 11C/min for all heating sequences, which was slow enough to prevent overshooting within the accuracy of the furnace control unit (711C). During data collection, the temperature was held constant, and for the last data acquisition the sample was cooled from 10001 to 251C at 601C/min. XRD patterns were analyzed by Rietveld refinement using the software FullProf.2k.36 Starting models for the crystal structures were taken from Mathew et al.37 for a-TCP, Dickens et al.38 for b-TCP, and Sudarsanan and Young39 for hydroxyapatite (HA). No additional crystalline phases were observed in the diffraction patterns. The refinements were performed in three steps, starting with vertical sample displacement, scale factors for all phases, and sampled background points with linear interpolation until convergence was reached. In the second step, the cell parameters of all phases were released for refinement, and in the last step isotropic peak broadening of all phases and one common overall isotropic displacement parameter (Boverall) were optimized. The instrument resolution function and peak asymmetry were determined with an ‘‘NIST SRM 660a—Line Profile LaB6’’ standard before the refinements of the sample data sets. Peak shapes were modeled with a pseudo-Voigt function accounting for peak asymmetry due to axial divergence.40 Average crystallite sizes were calculated from isotropic Gaussian peak broadening using the Scherrer equation. The most common method to determine X-ray amorphous fractions in a specimen is to add a known quantity of a crystalline internal standard. However, at temperatures up to 10001C, solid-state reactions of the specimen with the standard powder are to be expected. Relative weight fractions WP of the crystalline phase P, on the other hand, can be calculated from refined parameters and chemical information of all phases as follows41: SP ðZMVÞP WP ¼ P i Si ðZMVÞi

(1)

where S is the Rietveld scale factor, Z the number of formula units per unit cell, M the molecular mass of the formula unit, V the unit cell volume, and i iterates over all crystalline phases. The absolute mass mP of each crystalline component in the specimen is proportional to

II. Experimental Procedure ATCP were prepared by flame-spray synthesis as described by Loher et al.31 and Brunner et al.35 A liquid precursor solution was prepared from calcium hydroxide (Ph. Eur., Riedel-deHae¨n, Seelze, Germany) dissolved in 2-ethylhexanoic acid (technical grade, Soctech, Bucharest, Romania) and tributyl phosphate (Acros Organics, Geel, Belgium). The liquid mixture was fed through a capillary (0.4 mm in diameter) into a burning methane (1.13 L/min)/oxygen (2.4 L/min, both technical grade, Pan Gas, Dagmersellen, Switzerland) supporting flame at a rate of 10 mL/ min using a gear-ring pump (HNP Mikrosysteme, Parchim, Germany). Oxygen (10 L/min) was used to disperse the liquid leaving the capillary. During flame-spray synthesis, the nanoparticles form in a high-temperature environment (core part of the flame: 15001–20001C; residence time: 1–3 ms) and are rapidly quenched by air entrainment from the surrounding. This synthesis procedure favors the formation of kinetically controlled phase distributions. The particles were separated from the gas phase using PTFE filters and subsequently sieved with 250 mm mesh sieves. In situ XRD data sets were collected using a powder diffractometer (X’Pert Pro MPD, Panalytical, Almelo, the Netherlands)

mP / SP ðZMVÞP

(2)

By making the reasonable assumptions that (i) the intensity of the primary beam and the device configuration were identical for all data sets, (ii) the total amount of powder in the specimen remained constant during the experiment, and (iii) all phases involved, including the amorphous powder, have similar particle sizes and absorption coefficients, the dependency of SP could be reduced to the phase abundance and the overall atomic displacement parameter. The latter, as well as the unit cell volume, are temperature dependent. The value mP calculated from Eq. (2) for a constant amount of a phase P will thus show a slight linear decrease with the increasing temperature. In order to determine the amount of X-ray amorphous material in the specimen, a 100% crystalline reference was needed. As the sum of mP values of all crystalline phases remained constant between 8001 and 10001C, except for a steady decrease caused by the temperature-dependent parameters, it was assumed that all amorphous material was crystallized below 8001C. A linear regression through the sums of mP values

October 2010

Thermal Treatment of Flame-Synthesized ATCP Nanoparticles

between 8001 and 10001C thus served as references for 100% crystallinity for the calculation of absolute phase quantities. Scanning electron micrographs (SEM) were recorded using a LEO 1530 Gemini (Zeiss/LEO, Oberkochen, Germany) after sputtering the samples with B4 nm of platinum. Particle sizes were measured manually by determining the diameter of 50 particles per sample using the software Image Access Premium 7 (Imagic, Glattbrugg, Switzerland). For transmission electron microscopy (TEM) and selected area electron diffraction (SAED), one sample was used without thermal treatment, a second one was heated for 1 h at 6001C (11C/min), and a third one was heated for 1 h at 7001C (11C/ min). Figures 5–7 were recorded using a TEM (JEM-2011, JEOL Ltd., Tokyo, Japan) operated at 200 kV in the brightfield mode with a resolution of 0.25 nm. The instrument was equipped with a Link ISIS EDS system (Oxford Instruments, Abingdon, U.K.) and a Dualview CDD camera (Gatan France, Grandchamp, France). A double-tilt specimen holder was used to spatially orient crystals in the SAED mode. This procedure allowed the identification of crystalline phases though the indexation of mono- and polycrystal diffractograms. CaP particles suspended in ultra pure ethanol were submitted to ultra sound agitation for 15 min to separate particles forming agglomerates and to obtain homogeneous dispersion. The suspensions were then dropped on TEM grids covered with carbon films. Figure 8 was recorded using a CM30 ST (LaB6 cathode, operated at 300 kV, point resolution 0.20 nm), transmission electron microscope (Philips Electron Optics, Eindhoven, the Netherlands). Particles were deposited onto a carbon foil supported on a copper grid.

III. Results In XRD patterns, the material appeared X-ray amorphous up to 5001C. First diffraction peaks were observed at 5251C (Fig. 1), and at 5501C the first refinable diffraction pattern was obtained for which the phase quantities and crystallite sizes could be determined despite considerable peak broadening. The dominant phase at early stages of the crystallization was identified as a-TCP, and the minor phases as b-TCP and HA (Fig. 2). The highest absolute amount of a-TCP was 69 wt% found at 6001C. Further heating induced a gradual transformation of a- into b-TCP and narrowing of the peak shape due to crystal domain growth. Above 9001C, no a-TCP could be detected. HA concentrations converged to 7–8 wt% above 7501C, indicating a slight mismatch of the Ca:P ratio in the bulk material. Calculation of absolute phase quantities showed that the degree of

Fig. 1. In situ X-ray diffraction patterns and theoretical peak positions for a-tricalcium phosphate (TCP), b-TCP, and hydroxyapatite (HA). Just after crystallization, the main phase was a-TCP, followed by a gradual transformation to b-TCP. The peak shift of the top pattern was caused by the thermal contraction of the unit cell when cooled from 10001C to room temperature.

3457

Fig. 2. Absolute concentrations of crystalline phases as refined from the X-ray powder diffraction patterns. The pattern observed at 5251C could not be quantified. The degree of crystallinity reached 52 wt% at 5501C and 100 wt% at 6001C. The standard deviation of the sum of crystalline phases is 1.5 wt% for three-phase compositions, and 1.0 wt% for two-phase compositions.

crystallinity reached 52 wt% at 5501C, and 88 wt% at 5751C. At 6001C and above, the material was 100% crystalline and no amorphous content was detectable. Average crystallite sizes shown in Fig. 3 were calculated from isotropic peak broadening. The broad peak shapes observed in the 5501C X-ray diffraction pattern corresponded to an average spherical crystallite size of 30–40 nm. This was slightly larger than the values measured from SEM images before thermal treatment (Fig. 4). With the increasing temperature, the crystallites grew, reaching 150 nm at 10001C and thus exceeding the nanorange (o100 nm). Particle growth was faster than crystallite growth, which became particularly pronounced above 8001C. At high temperatures, some reflections of the b-TCP pattern lost intensity due to excessive thermal vibration of some atomic sites. It was not attempted to refine individual atomic displacement parameters, instead higher RBragg values were accepted, as peak shape and phase quantity refinements did not appear to be negatively influenced. TEM observations of the amorphous raw material confirmed that the particles were originally amorphous (Fig. 5). However, they were very sensitive to the electron beam irradiation and started to crystallize after exposure of some seconds up to 1 min.

Fig. 3. Average crystallite sizes were refined from X-ray powder diffraction (XRD) data by measuring isotropic peak broadening. Particle sizes were measured from scanning electron microscopy (SEM) images by averaging the diameters of 50 particles.

3458

Journal of the American Ceramic Society—Do¨belin et al.

Vol. 93, No. 10

Bright-field images showed morphological transformations from the amorphous spherical particles to the structured ones. In Fig. 6, we observed the corresponding electron diffraction pattern evolution from amorphous particles before irradiation to a high degree of crystallinity after irradiation for several minutes. The sample calcined at 6001C was crystalline, but tended to become amorphous under the electron beam. Electron diffraction patterns of a- , b-TCP, and HA were observed (Fig. 7) for particle ensembles, but no individual particles containing more than one phase simultaneously were found.

IV. Discussion

Fig. 4. Scanning electron microscopy images of the amorphous powder before thermal treatment showed spherical particles with an average diameter of 28 nm (standard deviation 5 4 nm, determined from 50 particles).

(1) Phase Evolution Despite the difference in the preparation route, the temperaturedependent phase evolution observed in our experiment generally agrees with the results published in previous studies.29,31,34 The first phases observed were a- , b-TCP, and HA. The a-TCP content was relatively high with an a:b-TCP ratio between 76:24

Fig. 5. Bright-field transmission electron microscopy images of the amorphous sample: not irradiated (A), after irradiation during 1 s (B), 20 s (C), 40 s (D), and 1 min (E).

October 2010

Thermal Treatment of Flame-Synthesized ATCP Nanoparticles

3459

Fig. 6. Selected area electron diffraction patterns of the amorphous sample: not irradiated (A), after some seconds of irradiation (B), and particles exposed to the electron beam for different (increasing) times until almost complete crystallization (C–F).

and 85:15 from the beginning of crystallization up to 6501C (Table I). a0 -TCP, which is less stable than a-TCP and should thus, according to the Ostwald step rule,28 occur first in the crystallization sequence could not be observed in the present study. The gradual transformation of a-TCP into b-TCP after the initial crystallization was expected according to thermodynamics and the Ostwald step rule. It represents the transition from a metastable to a thermodynamically stable phase. The high-temperature XRD setup used in this study, even though allowing in situ data collection at high temperatures, was not a real-time analysis technique. It showed the static phase composition at a given temperature, but no real-time information about ongoing reactions. In fact, the design of the experiment was chosen in a way that phase transformations primarily occurred between data acquisitions when the temperature was increased rather than during data collections when the temperature was kept constant. From the XRD data, we learned that a0 -TCP does not occur as a metastable or stable phase in the crystallization sequence of the amorphous flame-synthesized TCP. But we cannot exclude that it occurred temporarily at

early stages of the crystallization, and if it does, a certain size of coherent crystal domains would still be required in order to generate a diffraction pattern.

(2) Particle Growth Before thermal treatment, the raw material was composed of weakly connected spherical particles with a mean particle diameter of approximately 28 nm (Fig. 4). XRD data show that the powder was transformed from a purely X-ray amorphous state at 5001C into purely crystalline material with crystallite sizes of approximately 50 nm at 6001C (Fig. 3). At temperatures up to 7001C, the crystallite sizes determined from isotropic peak broadening were slightly larger than the particle sizes measured from SEM images. Because this constellation has no physical meaning, it is rather an indication of a calibration error of the diffractometer resolution function, or an interpretation error of the SEM images. Considering the large number of parameters that can influence the interpretation of isotropic peak broadening (strain, anisotropic crystallite shapes), it is known and no

3460

Journal of the American Ceramic Society—Do¨belin et al.

Vol. 93, No. 10

Fig. 7. Six examples of crystalline phases identified through selected area electron diffraction patterns in the sample calcined at 6001C: particle ensembles containing a-TCP and hydroxyapatite HA (A, B), particles containing only a-TCP (C, D), and particles containing only b-TCP (E, F).

surprise that the calculated values obtained by the Scherrer equation do not reach the accuracy of optical measurements. The data, however, show that crystallite and particle sizes are in the same range. In other words, just after crystallization the particles were single crystals with a coherent crystal structure. TEM images of the 7001C data set (Fig. 8) confirmed this interpretation by showing coherent crystal domains throughout the particles, and SAED results supported the data by proving the single-phase composition of the particles. With the increasing temperature, it became increasingly difficult to define the size of the particles in SEM images due to the formation of necks and particle intergrowth. As the particles agglomerated, the discrepancy between particle size and crystallite size increased, and so did the particle size distribution. The XRD data showed no sign of anisotropic crystal growth such as needles or platelets.

In a gradual phase transformation as it was observed in this study, one would expect to find partially transformed particles. The fact that a- and b-TCP coexisted and transformations from a- to b-TCP occurred but no multiphase particles were found is remarkable, but it can be explained by new insights on the b-2a-TCP phase transformation recently published by Carrodeguas et al.15 These authors suggested that the transition is reconstructive and requires a considerable amount of activation energy to proceed in both directions. Our observations confirm that once the activation barrier is surpassed, the transformation proceeds throughout the particle too quickly to appear as crystal domains of sub-particle size in XRD patterns or to be ‘‘frozen’’ in the samples used for ex situ TEM analysis. Hence, we only observed untransformed a-TCP or fully transformed b-TCP particles.

October 2010

Thermal Treatment of Flame-Synthesized ATCP Nanoparticles

3461

Table I. Absolute Phase Quantities and Calculated Ca:P Molar Ratios of the Crystalline Content Temperature (1C) a-TCP (wt%) b-TCP (wt%)

25, 500, 525 550 575 600 650 700 750 800 850 900 950 1000

— 31.0 (0.5) 63.5 (0.5) 69.7 (0.5) 68.4 (0.5) 62.6 (0.5) 38.4 (0.5) 21.2 (0.5) 10.2 (0.5) 5.0 (0.5) 0 0

— 9.9 (0.5) 11.1 (0.5) 13.3 (0.5) 17.7 (0.5) 29.0 (0.5) 57.2 (0.5) 72.2 (0.5) 81.5 (0.5) 88.1 (0.5) 91.6 (0.5) 92.3 (0.5)

HA (wt%)

Ca:P ratio

— 11.2 (0.5) 13.1 (0.5) 15.5 (0.5) 12.3 (0.5) 9.6 (0.5) 8.1 (0.5) 7.9 (0.5) 7.2 (0.5) 7.3 (0.5) 7.3 (0.5) 7.0 (0.5)

— 1.54 (0.01) 1.52 (0.01) 1.53 (0.01) 1.52 (0.01) 1.52 (0.01) 1.51 (0.01) 1.51 (0.01) 1.51 (0.01) 1.51 (0.01) 1.51 (0.01) 1.51 (0.01)

Compositions were obtained from one data set (n 5 1). Standard deviations based on results published by Stutzman42 are given in parentheses. TCP, tricalcium phosphate, HA, hydroxyapatite.

(3) The Amorphous State The term ‘‘amorphous’’ in this study primarily refers to the absence of diffraction peaks in XRD and SAED patterns. It does not preclude the existence of local order such as cluster arrangements. In fact, it was shown that [Ca3(PO4)2]n clusters, in particular Posner’s clusters (Ca9(PO4)6), closely packed to spherical particles with water heals in the interstices, are a reasonable structural model for amorphous TCP obtained by precipitation from aqueous solutions.43–45 The precipitation process itself was described as ‘‘mass crystallization’’, and the presence of hydroxyl groups was found to stabilize the Posner’s clusters in C3 symmetry, an arrangement similar to the crystal structure of HA.46 These models rely on the presence of OH groups and

Fig. 8. Transmission electron microscopy images of the powder calcined at 7001C show coherent crystal structures throughout the particles.

water molecules, which are readily available in aqueous precipitation processes, but not in flame-spray synthesis. It was shown that apatites formed under water-free conditions can be completely free of OH,47 and even fully hydrated HA easily dehydrates to a certain degree at high temperatures.48 Considering the formation conditions of our amorphous powder and the experimental conditions during the formation of the crystalline

Fig. 9. During selected area electron diffraction analysis, weak a-tricalcium phosphate reflections disappeared and diffuse rings occurred after irradiation for 30 s up to 3 min.

3462

Vol. 93, No. 10

Journal of the American Ceramic Society—Do¨belin et al.

phases, it is to be expected that the material, including the minor HA phase, is in fact completely dehydrated. This is a major difference between our starting material and the one used in other studies, e.g. by Somrani et al.29 and Li et al.,30 who used amorphous TCP precipitated from an aqueous solution thus being fully hydrated before thermal treatment. These considerations, as well as the fact that no signs of order were observed in XRD and SAED patterns, suggest the absence of cluster arrangements in the amorphous flame-spray-synthesized material. In order to understand the phase changes occurring under the TEM electron beam, it is important to observe that: first, both types of transformation are induced by the interaction of a lowintensity (pA–nA), high-energy (200 keV) electron probe with the matter under vacuum (105 Pa). These transformations are not general but local, and induced by the kinetic energy transferred from electrons to the sample atoms. Second, the degree of crystalline order of samples, as well as the crystalline phases involved in the amorphization and crystallization processes is different. In the crystalline sample calcined at 6001C, a complex transformation occurs: The crystalline order of the a-TCP phase decreases under the beam irradiation, and simultaneously, a new amorphous phase appears. This is highlighted by the fact that, on one hand, a-TCP diffraction spots become less visible, and on the other, a diffuse diffraction ring is observed (Fig. 9). HA and b-TCP phases remain unchanged. In the uncalcined amorphous sample, however, the crystalline phase obtained after irradiation was identified (Fig. 6) as being mainly HA.

opposite trend was observed in a previous study analyzing the reactivity of flame-spray-synthesized nanoparticles produced in the same reactor as the material used in the present study,34 where compositions with small amounts of HA showed slightly less a-TCP and more b-TCP than HA-free samples after heat treatment below 10001C. Ca excess appears to have a different influence on the formation of a- and b-TCP depending on whether it crystallized below 10001C or at the phase boundary at 11501C.

V. Conclusion Previous studies have shown that flame-spray synthesis is a suitable method for the production of X-ray amorphous CaP nanoparticles, and that the process can be scaled up to produce larger quantities. Our study demonstrates that the as-prepared powders can be transformed into a nanoparticulate crystalline powder with high a-TCP content by calcination at relatively low temperatures. The process forms single-crystalline a-, b-TCP, and HA particles with minimum particle fusion and intergrowth. Nanoparticles sometimes allow to modify the reaction kinetics and flow properties of hydraulic cement pastes. The material characterized in the present study could thus be an interesting reactant for hydraulic bone cements to optimize the setting reaction and injectability.

Acknowledgments (4) Ca:P Ratio The Ca:P ratios of the bulk material calculated from the refined phase compositions remained constant at 7501C and above, but a higher ratio was determined for lower temperatures (Table I). For the calculations, it was assumed that all crystalline phases were stoichiometric, i.e. the Ca:P ratios were assumed to be 1.50 for the TCP phases, and 1.667 for HA. However, in the observed chemical system, no known reaction could explain the apparent change in the bulk Ca:P ratio. Obviously, the assumption that all phases were stoichiometric was not correct. There are two tendencies coinciding with the calculated change in the bulk Ca:P ratio: (i) a slight reduction of the HA content, and (ii) a massive reduction of the a-TCP content (Fig. 2). If one of these phases was deficient in Ca, the calculation assuming stoichiometric composition would have over estimated the bulk Ca content, but the error would diminish as the Ca-deficient phase diminishes. Ca deficiency in HA is a common phenomenon. It was shown by Raynaud et al.49 that nonstoichiometric HA with a Ca:P ratio below 1.667 decomposes into b-TCP and stoichiometric HA between 7001 and 7501C. This stays in agreement with the observed decrease of HA between 6001 and 7501C and the over estimated bulk Ca:P ratios calculated below 7501C. However, the charge balance in CDHA (Ca10x(HPO4)x (PO4)6x(OH)2x) is maintained by the incorporation of HPO4 substituting for PO4, and the phase is typically formed by hydration or precipitation reactions such as the hydration of a-TCP at low temperatures.20 It is thus questionable whether CDHA forms in a heat-induced crystallization of a fired raw material. On the other hand, HA is not the only phase in the observed system known for a variable Ca content; a certain degree of Ca deficiency has also been reported for both a- and b-TCP,12,50 but the data available in this study are not sufficient to determine the precise Ca:P ratios of the individual phases and to reveal the origin of the Ca:P over-estimation was not the scope of the present study. The slight excess of Ca as compared with the stoichiometric TCP composition may have had an effect on the crystalline phase composition of the thermally treated material, in particular on the formation of a-TCP. It was shown by Lemaıˆ tre et al.51 that Ca:P ratios slightly above 1.50 promoted the formation of a-TCP when heat treated at 13001C, whereas Ca-deficient compositions tended to form b-TCP instead. The

The authors would like to thank Dr. Urs Eggenberger from the University of Bern for his support with the XRD data collection, and Norman Lu¨chinger from the ETH Zu¨rich for SEM images.

References 1

H. Albee and S. J. Morrison, ‘‘Studies in Bone Growth Triple Calcium Phosphate as a Stimulus to Osteogenesis,’’ Ann. Surg., 71 [1] 32–9 (1920). 2 T. Driskell, ‘‘Calcium Phosphate Resorbable Ceramics: A Potential Alternative to Bone Grafting,’’ J. Dent. Res., 52, 123 (1973). 3 J. J. Klawitter and S. F. Hulbert, ‘‘Application of Porous Ceramics for the Attachment of Load Bearing Internal Orthopedic Applications,’’ J. Biomed. Mater. Res. Symp., 2 [1] 161–229 (1971). 4 S. F. Hulbert, F. A. Young, R. S. Mathews, J. J. Klawitter, C. D. Talbert, and F. H. Stelling, ‘‘Potential of Ceramic Materials as Permanently Implantable Skeletal Prostheses,’’ J. Biomed. Mater. Res., 4 [3] 433–56 (1970). 5 F. C. M. Driessens and R. M. Verbeeck, ‘‘Relation Between Physico-Chemical Solubility and Biodegradability of Calcium Phosphates’’; pp. 105–11 in Advances in Biomaterials, Vol. 8, Edited by C. de Putter, G. L. de Lange, K. De Groot, and A. J. C. Lee. Elsevier Science Publishers, Amsterdam, 1988. 6 R. Z. LeGeros, ‘‘Biodegradation and Bioresorption of Calcium Phosphate Ceramics,’’ Clin. Mater., 14 [1] 65–88 (1993). 7 M. C. von Doernberg, B. von Rechenberg, M. Bohner, S. Grunenfelder, G. H. van Lenthe, R. Muller, B. Gasser, R. Mathys, G. Baroud, and J. Auer, ‘‘In Vivo Behavior of Calcium Phosphate Scaffolds with Four Different Pore Sizes,’’ Biomaterials, 27 [30] 5186–98 (2006). 8 M. Bohner, ‘‘Calcium Orthophosphates in Medicine: From Ceramics to Calcium Phosphate Cements,’’ Injury, 31 [Suppl. 4] 37–47 (2000). 9 A. Tampieri, G. Celotti, F. Szontagh, and E. Landi, ‘‘Sintering and Characterization of HA and TCP Bioceramics with Control of their Strength and Phase Purity,’’ J. Mater. Sci. Mater. Med., 8 [1] 29–37 (1997). 10 I. R. Gibson, I. Rehman, S. M. Best, and W. Bonfield, ‘‘Characterization of the Transformation from Calcium-Deficient Apatite to Beta-Tricalcium Phosphate,’’ J. Mater. Sci. Mater. Med., 11 [9] 533–9 (2000). 11 H. Monma and M. Goto, ‘‘Behavior of the a–b Phase Transformation in Tricalcium Phosphate,’’ Yogyo-Kyokai-Shi, 91 [10] 473–5 (1983). 12 J. H. Welch and W. Gutt, ‘‘High-Temperature Studies of the System Calcium Oxide–Phosphorus Pentoxide,’’ J. Chem. Soc., 874, 4442–4 (1961). 13 W. Fix, H. Heymann, and R. Heinke, ‘‘Subsolidus Relations in the System 2CaO  SiO2–3CaO  P2O5,’’ J. Am. Ceram. Soc., 52 [6] 346–7 (1969). 14 R. Enderle, F. Gotz-Neunhoeffer, M. Gobbels, F. A. Muller, and P. Greil, ‘‘Influence of Magnesium Doping on the Phase Transformation Temperature of Beta-TCP Ceramics Examined by Rietveld Refinement,’’ Biomaterials, 26 [17] 3379–84 (2005). 15 R. G. Carrodeguas, A. H. De Aza, X. Turrillas, P. Pena, and S. De Aza, ‘‘New Approach to the Beta-Alpha Polymorphic Transformation in MagnesiumSubstituted Tricalcium Phosphate and its Practical Implications,’’ J. Am. Ceram. Soc., 91 [4] 1281–6 (2008). 16 R. W. Nurse, J. H. Welch, and W. Gutt, ‘‘A New Form of Tricalcium Phosphate,’’ Nature, 182 [1] 1230 (1958). 17 M. Yashima and A. Sakai, ‘‘High-Temperature Neutron Powder Diffraction Study of the Structural Phase Transition between a and a0 Phases in Tricalcium Phosphate Ca3(PO4)2,’’ Chem. Phys. Lett., 372 [5–6] 779–83 (2003).

October 2010 18

Thermal Treatment of Flame-Synthesized ATCP Nanoparticles

H. Monma and T. Kanazawa, ‘‘The Hydration of a-Tricalcium Phosphate,’’ Yogyo-Kyokai-Shi, 84 [4] 209–13 (1976). 19 C. Durucan and P. W. Brown, ‘‘Alpha-Tricalcium Phosphate Hydrolysis to Hydroxyapatite at and Near Physiological Temperature,’’ J. Mater. Sci. Mater. Med., 11 [6] 365–71 (2000). 20 C. Durucan and P. W. Brown, ‘‘Kinetic Model for Alpha-Tricalcium Phosphate Hydrolysis,’’ J. Am. Ceram. Soc., 85 [8] 2013–8 (2002). 21 F. C. M. Driessens, M. G. Boltong, O. Bermudez, J. A. Planell, M. P. Ginebra, and E. Fernandez, ‘‘Effective Formulations for the Preparation of Calcium Phosphate Bone Cements,’’ J. Mater. Sci. Mater. Med., 5, 164–70 (1994). 22 F. C. Driessens, J. A. Planell, M. G. Boltong, I. Khairoun, and M. P. Ginebra, ‘‘Osteotransductive Bone Cements,’’ Proc. Inst. Mech. Eng. [H], 212 [6] 427–35 (1998). 23 M. Bohner, ‘‘Physical and Chemical Aspects of Calcium Phosphates Used in Spinal Surgery,’’ Eur. Spine J., 10 [Suppl. 2] S114–21 (2001). 24 M. P. Ginebra, E. Fernandez, F. C. M. Driessens, and J. A. Planell, ‘‘Modeling of the Hydrolysis of Alpha-Tricalcium Phosphate,’’ J. Am. Ceram. Soc., 82 [10] 2808–12 (1999). 25 C. Durucan and P. W. Brown, ‘‘Reactivity of Alpha-Tricalcium Phosphate,’’ J. Mater. Sci., 37 [5] 963–9 (1999). 26 M. Bohner, A. K. Malsy, C. L. Camire, and U. Gbureck, ‘‘Combining Particle Size Distribution and Isothermal Calorimetry Data to Determine the Reaction Kinetics of a-Tricalcium Phosphate–Water Mixtures,’’ Acta Biomater., 2 [3] 343–8 (2006). 27 M. Bohner, R. Luginbuhl, C. Reber, N. Doebelin, G. Baroud, and E. Conforto, ‘‘A Physical Approach to Modify the Hydraulic Reactivity of AlphaTricalcium Phosphate Powder,’’ Acta Biomater., 5 [9] 3524–35 (2009). 28 R. A. Van Santen, ‘‘The Ostwald Step Rule,’’ J. Phys. Chem., 88 [24] 5768–9 (1984). 29 S. Somrani, C. Rey, and M. Jemal, ‘‘Thermal Evolution of Amorphous Tricalcium Phosphate,’’ J. Mater. Chem., 13 [4] 888–92 (2003). 30 Y. Li, W. Weng, and K. C. Tam, ‘‘Novel Highly Biodegradable Biphasic Tricalcium Phosphates Composed of a-Tricalcium Phosphate and b-Tricalcium Phosphate,’’ Acta Biomater., 3 [2] 251–4 (2007). 31 S. Loher, W. J. Stark, M. Maciejewski, A. Baiker, S. E. Pratsinis, D. Reichardt, F. Maspero, F. Krumeich, and D. Gu¨nther, ‘‘Fluoro-Apatite and Calcium Phosphate Nanoparticles by Flame Synthesis,’’ Chem. Mater., 17, 36–42 (2005). 32 N. Do¨belin, T. J. Brunner, W. J. Stark, M. Eggimann, M. Fisch, and M. Bohner, ‘‘Phase Evolution of Thermally Treated Amorphous Tricalcium Phosphate Nanoparticles,’’ Key Eng. Mater., 396–398, 595–8 (2009). 33 M. Maciejewski, T. J. Brunner, S. F. Loher, W. J. Stark, and A. Baiker, ‘‘Phase Transitions in Amorphous Calcium Phosphates with Different Ca/P Ratios,’’ Thermochim. Acta, 468 [1–2] 75–80 (2008). 34 M. Bohner, T. J. Brunner, N. Doebelin, R. Tang, and W. J. Stark, ‘‘Effect of Thermal Treatments on the Reactivity of Nanosized Tricalcium Phosphate Powders,’’ J. Mater. Chem., 18 [37] 4460–7 (2008).

35

3463

T. J. Brunner, M. Bohner, C. Dora, C. Gerber, and W. J. Stark, ‘‘Comparison of Amorphous TCP Nanoparticles to Micron-Sized alpha-TCP as Starting Materials for Calcium Phosphate Cements,’’ J. Biomed. Mater. Res. B Appl. Biomater., 83 [2] 400–7 (2007). 36 J. Rodriguez-Carvajal, ‘‘Recent Developments of the Program FULLPROF,’’ Commission on Powder Diffraction (IUCr),’’ Newsletter, 26, 12–9 (2001). 37 M. Mathew, L. W. Schroeder, B. Dickens, and W. E. Brown, ‘‘The Crystal Structure of a-Ca3(PO4)2,’’ Acta Crystallogr., B33, 1325–33 (1977). 38 B. Dickens, L. W. Schroeder, and W. E. Brown, ‘‘Crystallographic Studies on the Role of Mg as a Stabilizing Impurity in b-Ca3(PO4)2 I. The Crystal Structure of Pure b-Ca3(PO4)2,’’ J. Solid State Chem., 10, 232–48 (1974). 39 K. Sudarsanan and R. A. Young, ‘‘Significant Precision in Crystal Structure Details: Holly Springs Hydroxyapatite,’’ Acta Crystallogr., B25, 1534–43 (1969). 40 L. W. Finger, D. E. Cox, and A. P. Jephcoat, ‘‘A Correction for Powder Diffraction Peak Asymmetry Due to Axial Divergence,’’ J. Appl. Cryst., 27, 892–900 (1994). 41 R. J. Hill and C. J. Howard, ‘‘Quantitative Phase Analysis from Neutron Powder Diffraction Data using the Rietveld Method,’’ J. Appl. Cryst., 20, 467–74 (1987). 42 P. Stutzman, ‘‘Powder Diffraction Analysis of Hydraulic Cements: ASTM Rietveld Round-Robin Results on Precision,’’ Powder Diffr., 20 [2] 97– 100 (2005). 43 G. Treboux, P. Layrolle, N. Kanzaki, K. Onuma, and A. Ito, ‘‘Existence of Posner’s Cluster in Vacuum,’’ J. Phys. Chem. A, 104 [21] 5111–4 (2000). 44 N. Kanzaki, G. Treboux, K. Onuma, S. Tsutsumi, and A. Ito, ‘‘Calcium Phosphate Clusters,’’ Biomaterials, 22 [21] 2921–9 (2001). 45 A. S. Posner and F. Betts, ‘‘Synthetic Amorphous Calcium Phosphate and its Relation to Bone Mineral Structure,’’ Acc. Chem. Res., 8 [8] 273–81 (1975). 46 E. I. Suvorova and P. A. Buffat, ‘‘Size Effect in X-Ray and Electron Diffraction Patterns from Hydroxyapatite Particles,’’ Crystallog. Rep., 46 [5] 722–9 (2001). 47 J. C. Trombe and G. Montel, ‘‘Some Features of the Incorporation of Oxygen in Different Oxidation States in the Apatitic Lattice—I,’’ J. Inorg. Nucl. Chem., 40, 15–21 (1978). 48 P. V. Riboud, ‘‘Composition et Stabilite´ des Phases a Structure d’ Apatite Dans le Syste`me CaO–P2O5–Oxyde de Fer–H2O a Haute Tempe´rature,’’ Ann. Chim., 8, 381–90 (1973). 49 S. Raynaud, E. Champion, D. Bernache-Assollant, and P. Thomas, ‘‘Calcium Phosphate Apatites with Variable Ca/P Atomic Ratio I. Synthesis, Characterisation and Thermal Stability of Powders,’’ Biomaterials, 23 [4] 1065–72 (2002). 50 J. R. Van Wazer, ‘‘Orthophosphoric Acid, its Salts and Esters’’; pp. 479–599 in Phosphorus and its Compounds, Vol. 1: Chemistry, Edited by J. R. Van Wazer. Interscience Publishers, New York, 1958. 51 J. Lemaıˆ tre, H. Andrianjatovo, C. Biourge, K. Ohura, and P. Harodouin, ‘‘A New High Strength, Resorbable Biocement,’’ Oberfla¨chen Werkstoffe, 9, 13–8 (1994). &