Acta Biomaterialia 10 (2014) 940–950

Contents lists available at ScienceDirect

Acta Biomaterialia journal homepage: www.elsevier.com/locate/actabiomat

High-frequency impedance measurement as a relevant tool for monitoring the apatitic cement setting reaction Christelle Despas a,⇑, Verena Schnitzler b, Pascal Janvier c, Franck Fayon d,e, Dominique Massiot d,e, Jean-Michel Bouler f, Bruno Bujoli c, Alain Walcarius a a

LCPME, Université de Lorraine CNRS UMR 7564, 405 rue de Vandoeuvre 54600 Villers Lès Nancy, France Graftys SARL Eiffel Park bâtiment D, 415 rue Claude Nicolas Ledoux, pôle d’activités d’Aix en Provence, 13854 Aix en Provence Cedex 3, France CEISAM, Université de Nantes CNRS UMR 6230, 2 rue de la Houssinière, BP 92208, 44322 Nantes Cedex 03, France d CEMHTI, CNRS, UPR 3079, 1D Avenue de la Recherche Scientifique, 45071 Orléans Cedex 02, France e Université d’Orléans, Faculté des Sciences, Avenue du Parc Floral, 45067 Orléans Cedex 02, France f LIOAD, Université de Nantes, INSERM UMR 791, Faculté de chirurgie dentaire, BP 84215, 44042 Nantes Cedex 1, France b c

a r t i c l e

i n f o

Article history: Received 30 January 2013 Received in revised form 18 October 2013 Accepted 18 October 2013 Available online 25 October 2013 Keywords: Calcium phosphate cement High-frequency impedance In situ monitoring Initial and final setting times Bisphosphonates

a b s t r a c t This work reports the development of a relevant and general method, based on high-frequency impedance measurements, for the in situ monitoring of the alpha-tricalcium phosphate to calcium-deficient hydroxyapatite transformation that is the driving force of the hardening processes of some calcium phosphate cements (CPCs) used as bone substitutes. The three main steps of the setting reaction are identified in a non-invasive way through variation of the dielectric permittivity and dielectric losses. The method is also likely to characterize the effect of the incorporation of additives (i.e. antiosteoporotic bisphosphonate drugs such as Alendronate) in the CPC formulation on the hydration process. It allows us not only to confirm the retarding effect of bisphosphonate by an accurate determination of the setting times, but also to assess the phenomena that take place when Alendronate is added in the liquid phase or combined to the solid phase of the cement composition. Compared to the conventional Gillmore needle test, the present method offers the advantage of accurate, user-independent, in situ and real-time determination of the initial and final times of the chemical hardening process, which are important parameters when considering surgical applications. Ó 2013 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

1. Introduction The discovery of calcium phosphate cements (CPCs) constitutes a significant advance in the field of biomaterials for bone reconstruction, especially for injectable compositions which allow implantation using minimally invasive surgical techniques [1–5]. These formulations most often consist of combinations of calcium orthophosphates. They have been extensively developed to obtain biocompatible materials showing improved mechanical properties in comparison to ceramics, while preserving their ability to be resorbed in vivo and replaced over time by newly formed bone. To be of practical use in a surgery room, the cements must reach a suitable mechanical strength after an acceptable time. Indeed, from the moment it is prepared, the cement paste changes from a fluid to a rigid state until its plasticity is completely lost and a specific mechanical resistance is reached (generally between 5 and 50 MPa in compression). Throughout this setting period, there is

⇑ Corresponding author. Tel.: +33 3 83 68 52 22; fax: +33 3 83 27 54 44. E-mail address: [email protected] (C. Despas).

a time window during which the cement can be handled (e.g. for surgical implantation). The hardening process then continues while the mechanical strength of the cement paste gradually increases. Therefore, determining the initial and final setting times of injectable bone cements is of great importance. For this purpose, two standardized methods are classically used: the Gillmore needles [6,7] or Vicat needle [8] tests. These methods are based on the concept of ‘‘visible indentation’’, which is operator-dependent and can result in poor reproducibility from one research group to another. Furthermore, the methods are unable to provide information about the progress of the chemical reaction controlling the setting process. Monitoring chemical characteristics (i.e. nature of the products and their relative amount) and physical properties (i.e. porosity and mechanical strength) of the cement as a function of time most often entails the collection of samples at regular intervals and requires rather sophisticated characterization techniques, including X-ray diffraction, solid-state NMR, FTIR, mercury intrusion porosimetry, calorimetric analyses, scanning electron microscopy, compressive strength measurements or gas adsorption/desorption for surface area determination. In all cases the conditions are destructive, since the samples are, for example,

1742-7061/$ - see front matter Ó 2013 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.actbio.2013.10.019

C. Despas et al. / Acta Biomaterialia 10 (2014) 940–950

soaked in acetone to quench the setting reaction, especially in its early stage. The design of analytical methods allowing non-destructive and in situ dynamic monitoring of the setting reaction of cements dedicated to orthopedic applications is thus of great interest. In this context, ultrasonic techniques have been proposed to monitor the change in the viscoelastic properties of hydroxyapatite–polymethylmethacrylate systems [9] and calcium sulphate-based bone cements [10], but the high sound attenuation induced by the high content of liquid in the initial cement paste largely limits the potential of such techniques to investigate the initial stages of the setting reaction. On the other hand, impedance spectroscopy based on the electrical properties of the studied medium was exploited as a non-destructive technique for studying hydration processes occurring in other kinds of cement-based materials, even for low conductivity media like mortars [11–18]. Several approaches have been developed to correlate the electrical parameters with the mechanisms related to dissolution–precipitation reactions and microstructure changes occurring in Portland cement. They include DC electrical conductivity [11,12], measurements of the reflection coefficient of an electromagnetic wave sent perpendicularly to the flat surface of the sample [13–15] and impedance spectroscopy in the low frequency range (with major contributions of ionic species and charges present on the surface of particles) [16] or in the high frequency range (where the response is essentially related to the behaviour of the solid/liquid interface and structural characteristics) [17,18]. Computer models have been proposed to simulate the electrical response of cement pastes with equivalent circuits and to discriminate the contribution of the solid phase from that related to the electrolyte filling the material pores [19–21], or to distinguish between free water and water molecules either hydrogen-bonded to the surface of the cement particles or combined to hydration products [14,15,22–25]. Although the literature contains a number of contributions of the dielectric technique for the characterization of concrete, mortar and similar materials, only a few articles report on monitoring the hydration process of calcium phosphate cements by AC impedance, and these have been restricted to the low frequency range (kHz). For instance, Liu and co-workers [26,27] demonstrated that the introduction of crystal seeds accelerates the setting reaction but concomitantly reduces the compressive strength of CPCs. Other studies have focused on the relationship between the dielectric response of hydroxyapatite (HA) and its structure by considering the influence of temperature, the porosity and degree of hydration of HA, or the frequency of the applied electrical field [28–30]. In this paper, we propose to demonstrate that high-frequency impedance measurements provide an efficient way to monitor in real time the alpha-tricalcium phosphate (a-TCP) to calcium-deficient hydroxyapatite (CDA) transformation, which is the driving force of the setting reaction of some apatitic cements. Indeed, this technique in such a frequency range was previously reported to be able to show chemical changes at solid/liquid interfaces [31–35]. Recently, preliminary observations have demonstrated that significant variations of the dielectric parameters took place during the hardening process of a bisphosphonate (BP)-modified CPC when the frequency range of the applied electrical field was between 0.4 and 100 MHz [36]. This original approach enables us to investigate under non-destructive conditions, very close to the situation of in vivo implantation (body temperature, humid environment), the influence of additives (i.e. Alendronate) depending on the way they are introduced into the formulation. The evolution of dielectric parameters for data collection in this high frequency range allows us to assess the change in the chemical composition of the cement during the reaction, as well as the initial and final setting times.

941

2. Experimental 2.1. Calcium phosphate cement The apatitic CPC is a mixture of 78 wt.% a-TCP (Ca3(PO4)2), 5 wt.% dicalcium phosphate dihydrate (DCPD, CaHPO42H2O; Fluka), 5 wt.% monocalcium monohydrate (MCPM, Ca(H2PO4)2H2O; Fluka), 10 wt.% CDA (Ca10–x[ ]x(HPO4)y(PO4)6–y(OH)2–z[ ]z), prepared as described previously [37], and 2 wt.% hydroxypropyl methyl cellulose (E4MÒ, Colorcon-Dow Chemical, Bougival, France). Alendronate-doped CDA was obtained by suspending 1 g of CDA in 10 ml of an aqueous Alendronate solution ([Alendronate] = 1.23  102 M– Alendronate sodium trihydrate (Sigma–Aldrich)), as previously reported [36]. The resulting doped CDA was then collected by centrifugation, rinsed with water and dried at room temperature to a constant weight before use. Each CPC powder was milled for 30 min to obtain a similar particle size distribution, controlled using a Beckman-Coulter LS 230 laser granulometer. This grinding step leads to an almost complete transformation of DCPD into CaHPO4 (DCPA) and of MCPM into Ca(H2PO4)2 (MCPA), as observed by 31P MAS(magic angle spinning) and CP (cross-polarization) MAS NMR measurements. All cement paste samples were prepared by mixing 8 g of the powdered preparation with 4 ml of a 5 wt.% Na2HPO4 (Fluka) aqueous solution in a specific syringe (Fig. 1C) for 2 min to ensure the homogeneity of the obtained paste before analysis. This liquid/ solid ratio (R = 0.5 ml g1) was found to be optimal for providing suitable injectability and resorbability for its practical use in orthopedic surgery.

2.2. Instrumentation The high-frequency impedance measurements were recorded between 0.4 and 100 MHz, using an HP 4194 A impedance/gainphase analyser (Hewlett–Packard). Our experimental set-up allowed us to concomitantly perform complex impedance and Gillmore needles measurements at 37 °C. The thermostated axial capacitive cell and the thermostated cell for Gillmore measurements were home-made (Fig. 1). The bottom of the dielectric cell was first filled by injecting, via a syringe, 4–5 ml of the cement preparation, which was then covered with 5 ml of a 0.9 wt.% NaCl aqueous solution to mimic a real-life implantation. It was verified that the NaCl solution added did not affect the impedance response and that the quantity of the cement paste introduced into the dielectric cell was sufficient to avoid disturbing the dielectric measurements by contributions other than those of the cement paste. The top of the cell was closed to avoid evaporation. The remaining paste was deposited in the second cell for the Gillmore test, so that comparison of the dielectric and Gillmore results could be strictly made on the same sample and under the same experimental conditions. Each experiment was monitored until stabilization of the values of each dielectric parameter was observed (i.e. experiments were conducted for up to 12 and 36 h for undoped and Alendronate-modified cement, respectively). The experiments were run twice and led to similar results, giving evidence of the good reproducibility and accuracy of this technique. The experimental device was completed by a computer allowing automatic data acquisition and real-time calculation of the complex impedance, Z⁄, from which the dielectric permittivity, e’ (related to dipole variation), and dielectric losses, e’’ (related to the motion of free charges), were computed [38]. The evolution of -Z’’ vs. Z’ (the imaginary part and the real counterpart of Z⁄, respectively) is called the Nyquist plot. For the Gillmore standard method, the initial setting time (ti) is defined as the time elapsed until the small needle (diameter

942

C. Despas et al. / Acta Biomaterialia 10 (2014) 940–950

(B) (A)

Flow of water (37°C) NaCl solution Cement paste Insulator

(2)

1st electrode

(3)

2d electrode

(C) (4)

(1)

(5)

Fig. 1. (A) Experimental set-up combining dielectric measurements and the Gillmore needles test. Representations of (B) the dielectric cell and (C) the ancillary device. (A) (1) Home-made thermostated dielectric cell. (2) Control of the temperature. (3) Gillmore needles test apparatus. (4) In-house thermostated cell. (5) Thermostated bath (37 °C). (C) Two independent chambers containing (1) the powder and (2) the hardening liquid. (3) A mixer is placed in the powder compartment. (4) A multi-position selector which enables the transfer of the hardening liquid into the powder chamber, then the injection of the cement. (5) A Luer-Lock tip to which a catheter or trocart can be attached in order to guide the injection. (6) A pushrod or plunger, which enables pressure to be applied to the piston of the chamber containing the liquid in order to transfer it into the compartment containing the powder.

2.12 mm, weight 113.4 g) fails to indent the surface of the sample, while the final setting time (tf) is the corresponding value when using the large needle (diameter 1.06 mm, weight 453.6 g). In order to follow the progressive transformation of a-TCP to CDA ex situ, 31P MAS (magic angle spinning) and 31P-1H CP (crosspolarization) MAS 1H-decoupled NMR spectra were recorded with a Bruker Advance I spectrometer operating at 7.0 T (1H and 31P Larmor frequencies of 300 and 121.5 MHz, respectively) using a 4 mm double-resonance MAS probe head. All experiments were performed after quenching the hydration process by grounding the samples in acetone. The quantitative 31P MAS spectra of these samples were fitted using five independent contributions, four of them corresponding to the 31P experimental spectra of a-TCP, DCPA, DCPD and MCPA, respectively, and the remaining one being a Lorentzian line centred at 2.9 ppm, characteristic of the CDA phase. The relative amount (in wt.%) of each phase as a function of time was determined assuming that the composition of the formed CDA phase was close to Ca10–x[]x(PO4)6–y(HPO4)y(OH)2–z[]z. Zeta potential measurements were performed using an Anton Paar SurPASS electrokinetic analyser. Cement samples were observed using a scanning electron microscope (Leo 1450VP, Zeiss, Germany; magnification = 20,000). Images were acquired on the secondary electron mode, with a 5 pA beam current and a 7 kV accelerated voltage. 3. Results and discussion 3.1. Dielectric monitoring of the setting of the undoped apatitic cement 3.1.1. Impedance response Typical variations of the dielectric permittivity, e’, and dielectric losses, e’’, during the setting reaction of the undoped cement are

shown in Fig. 2A. Three successive stages could be observed in agreement with the different steps of the hardening reaction of the apatitic CPC. The overall reaction consists in the transformation of the main calcium phosphate component, a-TCP, into a CDA via a dissolution–precipitation process Eq. (1) [1]:

3a  Ca3 ðPO4 Þ2 þ H2 O ! Ca10x ½x ðPO4 Þ6y ðHPO4 Þy ðOHÞ2z ½z

ð1Þ

The hydrolysis of a-TCP, catalysed by Na2HPO4, starts immediately after mixing together the liquid and solid components, thus leading to the release of calcium and phosphate ions into the solution [39,40], resulting in the high initial dielectric losses value. This corresponds to an induction period (stage 1) which is rather short for the present sample, but can be much longer in other cases (discussed below). The linear increase in e’ demonstrates the accumulation of mobile unbound charged species on the surface of the solid reactants and the formation of a supersaturated medium. Concomitantly, calcium and phosphate ions precipitate into CDA crystals on the surface of a-TCP (stage 2), as shown by the rapid decrease in the dielectric losses, e’’. The comparison of the 31P solidstate NMR spectra in the early stage of the setting reaction (see Supplementary Supporting Information Fig. Si-1(a and b)) shows an increase in the intensity of the CDA resonance after reaction for 17 min. The quantified relative amount of CDA (i.e. 11 wt.% after the 17 min reaction compared to 6.9 wt.% in the initial powder, Fig. 3a) confirmed unambiguously the rapid a-TCP to CDA transformation and the short induction period in the case of the undoped cement. Such transformation is also supported by XRD analyses (Fig. Si-2). Because of the growth of CDA crystals, which contributes to the reduction in the intergrain distance via the formation of an interpenetrated network of CDA grains (Fig. 4) responsible for the cement hardening process (stage 3), the

943

C. Despas et al. / Acta Biomaterialia 10 (2014) 940–950

160

3500

120 100

800

3000

80 Gillmore

60 40

tf

Gillmore

ti

20 0 0

80

dielectric losses, ε"

dielectric permittivity, ε'

140

100

dielectric losses, ε"

dielectric permittivity, ε'

(A)

10 20 30 40 50

reaction time (min)

60 40

2500 2000

50

100

150

200

250

1000

0

50

100

3000

90 60 30 0 0

2000

4000

6000

reaction time (min)

60 40

200

250

300

Gillmore

tf

50

100

ti

2500 Gillmore

ti

2000

1000 500 0 0

tf

6000

0 150

200

250

300

0

50

100

3000 Gillmore

tf Gillmore

dielectric permittivity, ε'

60 40 20

180 150 120 90 60 30 0 0

500 1000 1500 2000

2000

0 200

300

Gillmore

1500 1000 500

3500 3000 2500 2000 1500 0

250

300

reaction time (min)

500 1000 1500 2000

reaction time (min)

0 150

250

tf

reaction time (min)

100

200

Gillmore

ti

2500

dielectric losses, ε"

dielectric losses, ε"

120

ti

150

reaction time (min)

140

50

4000

1000

3500

0

2000

reaction time (min)

1500

160

80

1500

Gillmore

reaction time (min)

100

2000

500

Gillmore

20

dielectric losses, ε"

120

dielectric losses, ε"

dielectric permittivity, ε'

dielectric permittivity, ε'

150

80

0

dielectric permittivity, ε'

150

reaction time (min) 3500

0

(C)

10 20 30 40 50

1500

300

160

100

0 0

reaction time (min)

reaction time (min)

120

200

0 0

140

400

500

20 0

(B)

tGillmore i Gillmore tf

600

0

50

100

150

200

250

300

reaction time (min)

Fig. 2. Variation of dielectric permittivity, e’ (left), and dielectric losses, e’’ (right), vs. reaction time for (A) the undoped calcium phosphate cement, (B, C) cement doped with 0.3 wt.% Alendronate (with respect to the solid phase) either dissolved in the liquid phase (B) or chemisorbed on CDA particles (C). Frequency: 10 MHz; temperature: 37 °C. For comparison, the arrows marked the initial ðtGillmore Þ and final ðt Gillmore Þ setting times, respectively, as determined by the Gillmore needles test. The y axes have been i f adjusted to the same scale for comparison purposes. Insets: evolution on the short term for (A) and on the long term for (B) and (C).

a-TCP surface is rapidly covered by a CDA shell. The created semipermeable layer then controls the motion of water and ions towards the inner part of the a-TCP particles for the propagation of the hardening reaction [1,41–43]. The slower decrease in dielectric losses towards very low values is consistent with a less charged solution. In parallel, the less conductive final product (the zeta potentials for CDA, a-TCP and cement were 0.03, 4.4 and 2.5 mV, respectively) hinders the charges polarizability in an alternating electrical field, thus decreasing the dielectric permittivity. Monitoring the dielectric permittivity values at longer times showed another evolution (a slow decrease in e’ value), suggesting an aging process due to the continuous and slow formation of CDA, limited by the restricted transport of water through the semi-permeable CDA shell. Finally, when the diffusion of reactive species through the CDA layer gets too limited, dielectric parameters reach a constant value (e’ = 54 and e’’ = 31 for t > 167 min). The dielectric variations are in agreement with the ex situ NMR analyses. During the first 20 min, the CDA formation is fast (with an estimated

conversion rate equal to 0.4% min1) and the dielectric parameters change rapidly before slowing down (the reaction rate is approximately 13 times slower). Hence, the a-TCP to CDA transformation after 100 min at 37 °C was only 8%, and was only 64% even after 2 days. 3.1.2. Induction period evidenced by impedance measurement Many factors can slow down the setting reaction and lead to an induction period, including the Na2HPO4 concentration in the liquid phase, the reaction temperature or the granulometry of the solid phase. The induction period of the hydration process of CPC cements can be unambiguously evidenced by high-frequency impedance measurements in the 0.4–100 MHz range. When less concentrated Na2HPO4 solutions were used as the liquid phase (i.e. 2 wt.% Na2HPO4, Fig. 5(A and B)), the dielectric parameters remained almost constant for the first 19–20 h and then increased (e’) or decreased (e’’) as a result of the starting of the CDA precipitation (stage 2). When increasing the Na2HPO4

944

C. Despas et al. / Acta Biomaterialia 10 (2014) 940–950

(A)

2,0 1,8

CDA

1,6

α-TCP

1,4

log(w.%)

1,2 1,0 0,8

DCPA

0,6

MCPA

0,4 0,2

DCPD

0,0 0

500

1000

1500

2000

2500

3000

reaction time (min)

(B)

2,0 1,8

CDA

1,6

α-TCP

1,4

log(w.%)

1,2 1,0

DCPA 0,8 0,6 0,4

MCPA

0,2 0,0

DCPD 0

500

1000

1500

2000

2500

3000

reaction time (min)

(C)

2,0

CDA

1,8 1,6

α-TCP

1,4

log(w.%)

1,2 1,0

DCPA

0,8 0,6 0,4 0,2 0,0

MCPA 0

500

DCPD 1000

1500

2000

2500

3000

reaction time (min) Fig. 3. Evolution of the logarithm of the relative amount (log(wt.%)) of a-TCP (squares), DCPA (up triangles), DCPD (circles), MCPA (star – wt.% of MCPA < detection limit except at the initial time) and CDA (down triangles) vs. reaction time determined by 31P MAS NMR for (A) the undoped cement compared with (B, C) cements loaded with 0.3 wt.% Alendronate (with respect to the solid phase) either (B) dissolved in the liquid phase or (C) chemisorbed onto CDA.

concentration in the liquid phase (i.e. the 5 wt.% Na2HPO4 used in the optimized formulation), the induction period was significantly reduced to less than 2 min and was thus unobservable by dielectric analysis (Fig. 2A) (2 min is the minimal necessary mixing time for CPC preparation before monitoring of the setting reaction). Liu et al. [44]have reported that, in the case of a CPC based on tetracalcium phosphate and anhydrous dicalcium phosphate, CDA formation did not occur until a supersaturation threshold was reached. This was consistent with our observation since, when increasing the phosphate ion concentration in the liquid phase, saturation of the liquid phase at the solid/solution interface was reached more rapidly. The Na2HPO4 component thus acted as an accelerator for CDA precipitation, thereby suppressing the induction period [45,46]. A similar effect was observed when increasing the particle size of the solid phase (i.e. no milling of the powder mixture), leading to a decrease in the surface area, which slowed the dissolution step down enough to delay the precipitation of a-TCP to CDA after homogenization and the introduction of the cement paste into the dielectric cell. Even in the presence of 5 wt.% Na2HPO4 in the liquid phase, an induction period of 25 min was observed, consistent with the absence of evolution of the dielectric parameters (Fig. 5(C and D)). When both calcium and phosphate concentrations reached a critical level, typical variations of dielectric values, which characterized the hydration process ("e’ and ;e’’) and then the aging period, could be observed. 3.1.3. Nyquist analysis Fig. 6A depicts the Nyquist plots (-Z’’ plotted vs. Z’) for the undoped cement. Their arc form is in good agreement with an electrical circuit combining resistance and capacitor in parallel to simulate the solid–liquid interface, and resistance in series for the contact solution (Fig. 7), as evidenced by the good concordance between the experimental data and the fitted curves (Fig. 6). At the earliest stage of the hydration process, a unique depressed semicircle is observed, the diameter of which increases with reaction time. This high-frequency arc diameter corresponds to the interfacial solid–liquid resistance, Ri. At the same time, the center of semicircles below the Z’ axis indicates the presence of a distribution of relaxation times that should be associated with the mixed composition of the sample. This result is also consistent with the early precipitation of scattered CDA crystals which interconnect and form a shell on the surface of the a-TCP particles as the reaction proceeds. As the CDA layer becomes thicker, a more interconnected network of CDA crystals (more porous and tortuous [47–50]) is present, with the replacement of the pore fluid by a more resistive final product, leading to an increase in the associated resistance. The interfacial resistance of the undoped cement increases during the hydration process until reaching a value above 6800 X. During the reaction, new electrical contributions appear, as shown by the deformation of the semicircle and the shift of the relaxation frequency to lower values in association with an increase in the heterogeneity of the sample (Fig. 6A). The geometry of the pores and their surface chemistry affected the reorientation of the charges and/or their relaxation time. Models from the literature for the dielectric response of Portland cement during its hardening process indicate the existence of two contributions which differ greatly in conductivity, one from the pore–fluid phase and the other from the reaction products [20,51]. Similarly, the increase in the h angle (called ‘‘depressed angle’’) along with the shift of the relaxation frequency towards lower values confirms such changes. Values of between 25 and 30° are achieved, corroborating those reported previously [18], and lead to a factor of heterogeneity, a, of 0.25 (Table 1). In the presence of additives, the variation of the interfacial resistance is less important, especially when the BP was previously

945

C. Despas et al. / Acta Biomaterialia 10 (2014) 940–950

A

B

1μm

1μm

Fig. 4. Scanning electronic micrographs of the hardened undoped cement after setting periods of 15 min (A) and 11 days (B) (same magnification). Note the progressive growth of precipitated nanocrystals of CDA at the surface of the a-TCP particles.

(A)

(B)

250

2000 induction period

1500

dielectric losses, ε"

dielectric permittivity, ε'

200 150 100 50

1000

500

induction period

0

0 0

2000

4000

6000

0

8000 10000

2000

(C)

(D)

350 300

6000

8000 10000

3000

induction period

2500

250

dielectric losses, ε"

dielectric permittivity, ε'

4000

reaction time (min)

reaction time (min)

200 150 100 induction 50 period

2000 1500 1000 500

0

0 0

50

100

150

200

reaction time (min)

0

50

100

150

200

reaction time (min)

Fig. 5. Evolution of the (A, C) dielectric permittivity and (B, D) dielectric losses vs. reaction time for (A, B) a milled undoped cement in the presence of 2 wt.% Na2HPO4 and (C, D) a non-milled undoped cement in the presence of 5 wt.% Na2HPO4.

chemisorbed onto CDA (Table 1). This small variation observed could be explained by the persistence of the surface charge brought by the additive. The deformation of the semi-circle from the beginning of the recording shows the complexity of the system in the presence of BP, confirmed by the quasi-stability of a regardless of the introduction mode of the Alendronate.

3.2. Alendronate-loaded cements

concentration and the number of PO3 groups on the phosphonic acid backbone [47]. The retarding effect was attributed to the ability of phosphonate groups to complex calcium and silicon ions, both in solution when released during the dissolution step and on the surface of the solid particles [47,54–56]. The following section demonstrates that the impedance method is undoubtedly appropriate for the real-time in situ monitoring of these effects (delay in the setting time, hardening process as a function of the introduction mode of Alendronate).

The use of CPCs having the dual roles of bone substitute and carrier for local and controlled delivery of drugs is very attractive, in particular for the treatment of bone diseases such as infections, tumors or osteoporosis. In this context, the combination of a bisphosphonate antiosteoporotic drug [52,53] with an apatitic CPC was recently reported. However, bisphosphonates (i.e. Alendronate) are known to affect the hardening process of Portland cements by increasing the setting time in direct proportion to their

3.2.1. Case 1: Alendronate dissolved in the liquid phase The dielectric responses give clear evidence that the presence of Alendronate strongly delays the setting reaction by the formation of a precipitated calcium–Alendronate complex onto the surface of a-TCP particles. The induction time, which was not visible in the absence of Alendronate (Fig. 2A), is now clearly observed (Fig. 2B). 31P MAS NMR analysis confirmed that no measurable

946

C. Despas et al. / Acta Biomaterialia 10 (2014) 940–950

350

3500

300

3000

250

2500

200

2000

-z" (ohm)

-z" (ohm)

(A)

ω

150

ω 150 min

1500 1000

100

17 min

50

500

5 min

0 0

44 min

0 0

50 100 150 200 250 300 350

500 1000 1500 2000 2500 3000 3500

z' (ohm)

z' (ohm)

(B)

1000

30 25

800

15 10

ω

139 min 112min 9 min

5

-z" (ohm)

-z" (ohm)

20

ω

600

450 min 400 200

0 -5

240 min

0 0

5

10

15

20

25

30

35

0

200

z' (ohm)

(C)

400

600

800

1000

z' (ohm)

12

16

10 12 6 250 min

4

-z" (ohm)

-z" (ohm)

8

ω

80 min 15 min

2

ω

8

1800min 1018min

4 0

0 0

2

4

6

8

10

12

14

0

4

8

12

16

20

z' (ohm)

z' (ohm)

Fig. 6. Nyquist plots observed during (left) the first part (initial and/or induction step) and (right) the conversion stage for (A) the undoped cement, (B, C) cements loaded with 0.3 wt.% Alendronate (with respect to the solid phase) (B) dissolved in the liquid phase or (C) chemisorbed onto CDA. Experimental data (scatter) and modelling results (line).

(A)

(B)

Fig. 7. Schematic representation of the equivalent electrical circuit (A) and the Nyquist plot (B).

transformation of the a-TCP phase into CDA is observed in the first 5 h (Fig. 3B). A last point to be mentioned is a singularity of the dielectric parameters evolution at the early beginning of the process. Their variation during the first few minutes agrees well with a dissolution step leading to the release of calcium and phosphate ions ("e’) and the concomitant formation and precipitation of a

calcium–Alendronate complex (;e’’), thus forming a passivation layer [47] that slows down the hydration of the cement particles. According to the 31P isotropic chemical shift (18.6 ppm) and the chemical shift anisotropy of the bisphosphonate resonance observed in the 31P CP-MAS NMR spectra recorded after a reaction time of 75 min, this complex can be identified as Ca4(Alendronate)3

C. Despas et al. / Acta Biomaterialia 10 (2014) 940–950 Table 1 Electrical parameters deduced from Nyquist plots. Reaction time (min) Undoped cement 5 17 44 150

Rinterfacial (X)

aa

51 249 760 6850

0.01 0.20 n.d. 0.25

Cement loaded with 0.3 wt.% Alendronate dissolved in the liquid phase 9 17 0.30 112 23 0.20 139 32 0.20 240 217 0.30 450 1498 0.30 Cement loaded with 0.3 wt.% Alendronate chemisorbed onto CDA 15 6 0.10 80 6 0.10 250 7 0.20 1018 5 0.20 1800 14 0.20 a a, the factor of heterogeneity, was calculated from the ‘‘depressed angle’’ value, h: a = 2hp.

[36] (Supporting Information, Fig. Si-3). The Nyquist diagrams (Fig. 6B) are consistent with this mechanism: (i) the semicircles shift towards higher frequencies, which suggests the formation of a product different from CDA; and (ii) a small inductive loop (negative part) is present at low frequencies, as previously observed in the case of adsorption/desorption processes [48,49,57], which agrees well with the precipitation of a calcium–bisphosphonate complex onto the surface of the mineral particles [55]. A second step is observed, with dielectric permittivity values remaining almost unchanged and dielectric losses decreasing slowly (during the first 100–120 min; see Fig. 2B). It can therefore be assumed that the Ca4(Alendronate)3 layer acts as a semi-permeable barrier which hinders water and cation diffusion, thus limiting the formation of CDA throughout this induction period (about 3% conversion ratio). However, this inhibition effect is only temporary due to the subsequent reorientation and/or rearrangement of the Ca-bisphosphonate product on the mineral surface [58], once again allowing the diffusion of aqueous species. After this induction time, the variation of both dielectric parameters is similar to that of undoped cement ("e’ and ;e’’ values), indicating that conversion of aTCP to CDA has started. Then comes the aging period, with a slow decrease in the dielectric permittivity values (inset of Fig. 2B, left), as confirmed by 31P MAS NMR results (Fig. 3B). The conversion ratio after 48 h is quite similar to that of the undoped cement (ca. 59%). The Nyquist plots appear as more complete and less deformed semicircles with relaxation times moving towards lower frequencies as the a-TCP to CDA conversion progresses, with a concomitant increase in the interfacial resistance. 3.2.2. Case 2: Alendronate chemisorbed onto CDA particles Schnitzler et al. [36] recently demonstrated that the chemisorption of Alendronate on the CDA surface is the best route to prepare cements containing up to 0.1–0.3 wt.% Alendronate with respect to the solid phase, whilst limit the retarding effect of the bisphosphonate. In this case, the induction time was about three times shorter than that previously observed when Alendronate was dissolved in the liquid phase (see parts B and C of Fig. 2). This fact is also supported by ex situ 31P solid-state NMR analyses (Fig. Si-4). The evolution of the dielectric parameters also provides evidence that the hydration mechanism differs significantly. Dielectric permittivity does not evolve during the first 75 min, while dielectric losses first increase before stabilizing. Given that Alendronate is chemisorbed onto CDA via

947

a substitution of surface phosphate groups by phosphonate moieties of the BP [36,59], it is very likely that the dissolution of a-TCP begins ("e’’) but that the drug blocks nucleation sites and alters the CDA crystal growth, as suggested for BaSO4 crystallization in the presence of diphosphonates [60,61], leading to constant values of dielectric parameters. The complexity of the setting reaction is evidenced by the fact that the increase in the intensity of the 31P MAS NMR signal of CDA is detected rapidly after the induction period and before the dielectric permittivity reaches its maximum. The conversion ratio after 48 h is equivalent to that observed for the undoped cement (ca. 67%; see Fig. 3C). However, for the last part of the process (the aging step), the deduced Nyquist plots are not consistent with those previously obtained (for the undoped cement as well as the BP-doped cement when Alendronate is introduced in the liquid phase), thus making it difficult to propose a mechanism in this case. 3.3. Determination of initial and final setting times Because the dielectric technique allows the continuous monitoring of the CPC hydration process by providing in situ data in agreement with the reaction mechanism (aging, influence of additives), the specific times of the evolution can be defined with more relevance than those obtained by the Gillmore needles test. For biomaterial applications, the most interesting point of the method lies in the fact that the initial (ti) and final (tf) evolution times can be easily assessed from the e’ and e’’ curves. The ti and tf values are determined from the intersection points of the tangent and asymptote (see the detailed examples in the Supporting Information, Fig. Si-8), and account for the beginning of the CDA precipitation and the appearance of the aging period, respectively. The corresponding results are presented in Table 2, where they are compared with setting time values obtained from the Gillmore needles method. The data confirm that the cement setting time is delayed upon introduction of Alendronate, with the most pronounced effect observed when the BP is dissolved in the liquid phase. The evolution times deduced from the e’ and e’’ curves are consistent, but differ markedly from those obtained by the Gillmore normalized method, whatever the nature of the sample. In the case of the undoped cement, the initial and final setting times measured by the Gillmore needles method correspond to the sharp increase in the e’ values (Fig. 2A), namely at the early stage of the a-TCP to CDA transformation (ca. 3%), while impedance data clearly demonstrate that the cement paste continues to evolve. When microstructural changes of the undoped cement as a function of the reaction time are observed by SEM, it appears that transformation of the surface of the a-TCP particles is already obtained within 7 min (i.e. ti < t < tf) in comparison with the initial powder (Fig. 8A and B, 7 min). This is also supported by the dielectric evolution (ti < 4 min) (Table 1). For longer reaction times (t > tf), a-TCP particles are completely covered by CDA crystals, and their growth and entanglement (Fig. 8B, 15 min) result in an increase in the mechanical strength. More strikingly, when the BP is dissolved in the liquid phase, the final setting time (ca. 100 min, Gillmore needle test) is observed during the induction period while at this stage the a-TCP to CDA conversion has not started (Fig. 3B), since 31P MAS NMR spectra indicate that no significant transformation is observed before 280 min. Fig. 8C shows that full growth and entanglement of the CDA crystals formed on the surface of the cement particles are not reached after 180 min. The same conclusion can be made after 60 min of reaction time for the cement loaded with BP chemisorbed onto CDA particles (Fig. 8D). It can thus be concluded that the Gillmore method is not adapted to characterizing the setting reaction in such cases (i.e. in the presence of additives). Indeed, the initial and final setting times only denote the resistance to needle penetration, but

948

C. Despas et al. / Acta Biomaterialia 10 (2014) 940–950

Table 2 Initial and final setting times measured from high-frequency impedance curves, compared with Gillmore needles test data. Undoped cement

Alendronate doped cement (0.3 wt.%) BP dissolved in the liquid phase

a

BP combined to CDA

ti (Gillmore test) (min) tf (Gillmore test) (min)

10 ± 2 17 ± 2

35 ± 2 100 ± 5

ti(e’) (min) tf(e’) (min)

High Frequency Impedance Measurement as a Relevant Tool to Monitor Apatitic Cement Setting Reaction.

This work reports the development of a relevant and general method based on high frequency impedance measurements, for the in situ monitoring of the a...
2MB Sizes 0 Downloads 0 Views