Electrical properties of hydroxyapatite

Introduction

In the last decade, there has been a relative upsurge in the interest towards electrical properties of a well-known biomaterial, hydroxyapatite (HA), Ca₅(PO4)₆(OH)₂ widely used in cementless implant coatings and bone graft substitutes. This is in contrast to a huge body of literature that exists in the field of chemical synthesis, processing and functionalization of HA and related materials and devices. An emphasis on chemical-biological properties of HA probably originates from the general perception that HA is the synthetic chemical analog of the mineral component, apatite, of bone and teeth in vertebrae. Both bone and dentine show interesting electrical properties foremost of which is piezoelectricity, the ability of generating electrical charge at the surface when stressed [1]. This pressure induced surface charge has been considered to be connected with the functional shape generation of bone in accordance with the Wolff’s law. Bone also possesses a spontaneous dipolar electrical polarization due to its pyroelectric property [2], which again is believed to have functional distribution and purposes.

Bone piezoelectricity and its potential physiological role have been subject to vigorous investigations since the 1960s. The interest in further investigations subsided towards the end of 1980s with an implicit consensus that the observed unusual electrical properties of bone originates from its organic constitute, collagen. Collagen is a fibrilar protein and has been found to exhibit both piezoelectricity [3] and pyroelectricty [2]. The presence of apatite in bone was considered to make bone stiff due to its mechanical strength, and to act as a reservoir for calcium (Ca) to maintain calcium homeostasis. The electrical influence of apatite in bone’s piezoelectric behavior had been explained as a natural composite between apatite as a nonpiezoelectric dielectric within a piezoelectric collagen matrix [4]. The interest in studying electrical properties of hydroxyapatite, in general, waned. Since then, the success in poling hydroxyapatite using a direct current (dc) to induce electret nature in it, the selective calcification on the negatively poled side of such electrets [5], the discovery of two piezoelectric hydroxyapatite phases [6], and the ability of creating discrete domains of electrostatic charges [7] are key important developments that have driven further investigations of electrical properties of hydroxyapatite.

This article intends to provide a critical insight into the works conducted and the concepts developed within the authors’ group in the field of electrical properties of hydroxyapatite and its potential applications both as a bioceramic but also, hopefully, as a future lead free technical ceramics. The article then provides an outlook towards the future investigations that can be conducted in this new frontier of biomaterials research. For this purpose, after setting the scope of the article in its introduction, we describe the crystal structure, piezo-pyro-ferroelectricity, poling and electret behavior of hydroxyapatite and electrostatic domain creation there on. We then discuss the significance of these properties with respect to bone growth stimulation. Finally, we will provide with our view on the possible chemical modifications and processes that can be attempted to investigate electrical properties of HA further.

Hydroxyapatite crystal structure and symmetry

The question of the structure and symmetry of hydroxyapatite has been considered settled since the 1970s with a consensus that hydroxyapatite can exist in two stoichiometric polymorphs in monoclinic P21/b and hexagonal P63/m symmetries, both non-polar due to the presence of a centre of symmetry.

P63/m symmetry is a fundamentally incorrect description of stoichiometric hydroxyapatite as the suggested reflection of the OH dipole about the mirror plane is unrealistic and would make the structure unstable. The removal of the mirror plane would lead towards the P63 symmetry where the OH ions are in ferroelectric order, i.e., pointing in the same direction within the adjacent channels defined by the edge oxygen atoms of the PO₄ tetrahedra. These oxygen atoms as well as the OH dipoles within the channels lie along the crystallographic c-axis thus making the caxis or the [001] direction as the polar direction. The possibility of P63 symmetry was ruled out by the proponents of the P63/m symmetry due to a better empirical fitting obtained with the P63/m structure.

In monoclinic P21/b structure, the reflection of OH dipoles at the mirror plane has been obtained by removing the mirror plane and reversing the direction of the OH dipoles within the adjacent channel. This needed, along with the removal of the mirror plane, a doubling of the unit cell along the b axis. Even so, the off the mirror plane location of the OH dipole would influence the surrounding electrostatic balance such that the edge oxygen atoms of the respective phosphate tetrahedra surrounding the OH dipole could have been rotated. Such rotation would lower the symmetry to a P21 symmetry, which being a lower symmetry could have two configurations of OH dipole between the adjacent channels: a unidirectional orientation of OH dipoles giving a ferroelectric ordering and an antiparallel orientation giving an anti-ferroelectric ordering.

Quantum mechanical simulations found the antiferroelectric ordering energetically most stable at absolute zero temperature and in single crystals. The energy difference between this configuration and ferroelectric ordering within monoclinic and hexagonal structures was however small, of the order of a few kJ mol^–1, which was postulated to be available from surface energy at finite temepratures. Two polar phases of hydroxyapatite, P21 and P63 were considered feasible thus leading to the discovery of piezoelectric and pyroelectric phases of hydroxyapatite. X-ray diffraction and Rietveld analysis revealed over three-quarter of ferroelectrically ordered phase (denoted as P21) in polycrystalline, stoichiometric hydroxyapatite of Standard Reference Material (SRM) 2910 obtained from the National Institute of Standard and Testing, USA to the contrary of the certification that it was 78 % hexagonal and 22 % monoclinic.

Piezo-pyro-ferroelctricity

The experimental evidence of the stability of polar phases of hydroxyapatite led to the possibility of realizing piezoelectricity and pyroelectricity in hydroxyapatite due to the ferroelectric ordering of OH dipoles in these phases. Importantly, while piezoelectricity and pyroelectricity can be anticipated on the basis of crystallographic arguments only ferroelectricity cannot be predicted in this way and needs experimental determination. Semi-empirical atomistic simulation estimated piezoelectric constants that can be obtained in single crystal hydroxyapatite [8].

Finding piezoelectricity in polycrystalline aggregate was not trivial. This was primarily because conventional piezoelectricity measurements required reasonably rigid solid forms that could withstand mechanical pressure or electrical force that would have been applied during such measurements. Sintering of hydroxyapatite is well known but conventional free-standing sintering cause grain growth and leads to near-spherical grains and isotropy in the final aggregate. By careful uniaxial pressing and relatively shorter duration sintering a small anisotropy was retained in sintered ceramic of hydroxyapatite made from SRM 2910. The rotational symmetry of such aggregate was postulated to be ∞/2 symmetry. This symmetry allows only shear piezoelectricity. A pseudo-tensile piezoelectricity measurement (d₃₁) on a face that is cut at 45° to the pressing axis during the green body preparation was needed to measure the so called d₁₄ constant. The polarity of the voltage measured switched when the sample was flipped between the pressing faces thus qualitatively confirming piezoelectricity [9]. A quantitative value could not be established simply because the measured values were too low, in the order of 10^–6 pC N^–1.

Conventional ferroelectric ceramics are also known to exhibit very low piezoelectricity in their unpoled conditions. So it was necessary to pole hydroxyapatite ceramics. The thermally stimulated depolarization current (TSDC) experiment exhibited very low current coming from hydroxyapatite ceramics that were poled below 250 °C. Hydroxyapatite ceramics poled in this way needed a thermal preconditioning to show pyroelectricity. Texturing the ceramics and the retention of enough OH ions in the structure by resorting to very fast sintering (a few minutes instead of a few hours) helped to obtain appreciable pyroelectricity in hydroxyapatite [10]. As pyroelectricity is a subset of piezoelectricity, by definition these ceramics too were piezoelectric but no direct measurement of piezoelectricity was conducted at the temperatures where pyroelectricity was observed (300–500 °C). Interestingly two samples exhibited pyroelectricity at lower temperature 27–60 °C after keeping sandwiched between two electrodes for sufficiently long time (∼1 week). Overall, textured and poled ceramics showed higher pyroelectricity. The average pyroelectric constant (0.1–40 nC cm^–2 K^–1) between 300 and 500 °C comparable to conventional pyroelectric ceramics such as LiTaO3 or lead zirconate titanate (PZT) but thermal stability of pyroelectricity must be improved for technical applications as repeated thermal cycles showed decay in the polarization leading to an eventual loss of pyroelectricity. The pyroelectric constants measured between 27 and 60 °C were four orders higher than those observed in bone and tendon.

Much more significant piezoelectricity, as high as six orders of magnitude higher than that measured as shear piezoelectricity, has been obtained recently in textured hydroxyapatite ceramic that was poled under much higher electric field but at relatively lower temperature, 95 °C [11]. The piezoelectric strain coefficient, d33, is an order higher than that of bone and of the same order as that of quartz crystal and uncalcified collagen in tendon. The (e₃₃) piezoelectric charge coefficient is an order higher than bone and three orders higher than that of tendon. Most importantly, piezoelectricity in these ceramics can be tuned by controlling texture, and poling parameters. Similar poling method has recently been applied to a textured ceramic of 70/30 mixture of hydroxyapatite with β-tricalcium phosphate and the composite exhibited piezoelectricity of similar order to that observed in stoichiometric hydroxyapatite. Piezoelectric constants of poled HA ceramics are high enough to have physiological relevance. Examples include the use of bone grafts for stimulated calcification or as a potential candidate material for in situ energy harvesting from physiological motions. The low temperature poled hydroxyapatite has shown weak pyroelectricity (0.015 nC cm^–1 K^–1) and ferroelectricity.

Thin films of nanocrystalline hydroxyapatite on silicon have shown very strong piezo, pyro and ferroelectricity in their natural unpoled state. The magnitude of piezoelectricity and pyroelectricity are of the similar order of that shown by piezoelectric material zinc oxide (ZnO) and ferroelectric material polyvinylidene fluoride (PVDF), respectively [12]. When poled at 70 °C for an hour with a 1 kV cm^–1 dc electric field, a six orders of magnitude higher current than an unpoled film was observed (Fig. 1). These films have also shown both macroscopic and microscopic ferroelectricity [13].

Fig. 1

TSDC of nanocrystalline hydroxyapatite thin films on Si in (top) unpoled condition, three cycles showing pyroelectricity and (bottom) after poling at 70 °C, one cycle showing pyroelectricity and space charge.

Electrostatic micro- and submicro-domains

While electrical properties such as local electrostatic charge distribution at biomaterial surfaces are accepted as to play an important role in defining biological interactions, the exact nature of such interactions still remains to be understood. Physiological environment will polarize the surface of a medical implant by forming an electrical double layer. Similar double layers are formed around proteins and other biomolecules too. The thickness of these double layers lies in the nanometer range and depends on the magnitude of the surface charge. Most inorganic surfaces are negatively charged at normal pH, but the surface is far from being homogeneous as they can have patches of different polarities and magnitudes of surface charge as well as hydrophobic and hydrophilic regions even on an ideally flat surface. To discern the true role of surface charge interactions of implants with biology, it is important to create homogenous surface charge, which is practically difficult to create on real surfaces. A pattern array of discrete surface charge has been proposed so that the role of surface charge can be isolated from the effect of surface chemistry and roughness [14]. A number of methods have been outlined to aide such creation of these discrete areas of surface charge, which, in analogy to magnetic domains, can be termed as electrostatic domains.

Among the methods of creating such domains, electron irradiation has shown the ability to modify the surface electric potential and wettability of hydroxyapatite surface and consequent protein and bacterial adhesion thereon. Electron beam is known to inject negative charge into the surface and subsurface region of an insulator but the interactions of electron beam with the materials surface is complex and needs careful investigation. A move towards using the electron beam available in a laboratory scanning electron microscope was made so as to suit the practicality of biological assay analysis that requires a number of samples within a realistic timeframe. This was initially achieved by a direct microscopic patterning [7, 15] (i.e., without any pre-patterned mask/template), and then also by exploiting the raster scan of the electron beam used in electron microscopic imaging [16]. These works were able to show both positive and negative surface potentials within the micro and submicron sized domains with respect to the unexposed area. Figure 2 shows creation of submicron domains of negative surface potentials created on a hydroxyapatite thin film exposed to low magnification scanning electron beam through a mask made of nanochannel alumina (NCA) having an average pore size of 200 nm. Selective adsorptions of fluorescent tagged lysozyme proteins have been observed on negative microdomains with almost one to one correspondence of Coulombic interactions [17].

Fig. 2

Submicron domains of negative surface potential on thin hydroxyapatite films measured by Kelvin force probe microscopy (bottom). The film has very little topographical change as measured by atomic force microscopy (middle). The NCA filter used as the mask has an average pore diameter of 200 nm (top). The location of the pores and corresponding areas of surface potentials are highlighted in circles.

The ability of creating these microdomains comes with two interesting features: the tunability of the surface wettability [18] and the ability to visualize these domains in a typical confocal laser scanning microscope (c-LSM) widely used in characterizing biological species [19]. Arrays of micron-sized domains on hydroxyapatite showed a gradual increase of the water contact angle (from 57° to 93°) as the electron dosage increased. This was due to the elimination of the polar component of the surface free energy when the dosage increased. Surface contamination by carbonaceous species, assuming it is present, could not be held as fully responsible for such behavior at lower dosage of electron beam. This brought to the fore an interesting phenomenon that confronted current electro-wetting theory. A transfer of free surface charge to water and an electron beam induced disruption of polar orientation of OH ions might have been responsible for such behavior and deserves further investigations perhaps with the help from micro-droplet contact angle measurements. The size and shape of these microdomains are particularly relevant to that of mammalian cells and can be used further for elucidating the role of surface charge in mediating biological interactions in situ by monitoring the photoluminescence of the domains.

Potential for bone growth stimulation on piezoelectric hydroxyapatite

Observations of bone’s parallel architectural and electrical responses to applied forces have led to suggestions that one or both of the electromechanical signals mediate osteogenesis [20]. In the case of piezoelectricity, the growth of bone could occur in response to the surface polarization itself, or to the subsequent neutralization-ion kinetics [21]. Piezoelectricity was identified as the primary mechanism for stress generated potentials (SGPs) in dry bone but for wet bone its role was less clear. The relaxation time for piezoelectric charge was measured in the microsecond range, while that for SGP in wet bone was in the range of a few seconds [22]. While this prompted many to shed the idea of piezoelectricity of bone playing any major physiological role, Guzelsu and Walsh [23, 24] still considered it important in triggering the bone forming actions and proposed a detailed feedback mechanism in which both piezoelectricity and streaming potential was seen to play significant but different roles. Recently, Ahn and Grodzinsky [25] have put forward the hypothesis that by changing the surface charge as a result of mechanical stress, the piezoelectricity of collagen may influence the magnitude of streaming potential (measured as zeta potential) during compression. This mechanism may indirectly modify the stiffness and fluid dynamics of bone.

Bone is a living tissue which undergoes constant modeling and remodeling in vivo. While the mechanism of bone growth associated with modeling/remodeling cannot be directly employed to prosthetic devices, piezoelectric materials have been investigated as a potential biomaterial to influence bone growth behaviour. Examples of such materials include poly(lactic-co-glycolic acid) (PMLGA) [26], PVDF [27] and poled BaTiO₃. Deproteinated bone (mainly bone apatite) and decalcified bone (mainly collagen) have been used commercially as grafts for bone augmentations, perhaps to exploit the natural complex architecture, porosity and chemistry of such engineered tissues. Decalcified bone is piezoelectric while piezoelectricity of deproteinated bone has not been confirmed. Noris-Suarez et al. [28] have reported statistically significant in vitro HA deposition within 3 weeks on the compressed internal surface of the demineralised bone (negatively polarized due to bending piezoelectricity), while the outer surface (positively polarized due to bending piezoelectricity), which was under tensile strain, showed no difference in apatite precipitates.

If piezoelectricity is the driving force behind such stimulated calcification, one can expect similar, if not higher calcification on the surface of a piezoelectric HA. This expectation is based on the assumption that calcification occurs at least partially through redox reactions where a positively charged calcium ion (Ca⁺²) can gain two electrons from a negatively charged surface and deposit on the surface. As such deposition should scale with the number of available Coulombic charge in accordance with Faraday’s law of electro-deposition, the surface charge density generated in a piezoelectric hydroxyapatite becomes a more important figure of merit (FoM) than an absolute measure of surface charge only. From this perspective, the piezoelectric stress coefficients, denoted by ‘e’ and measured in C cm^–2(or C m^–2) is a better FoM than the commonly used strain coefficients, denoted by ‘d’. The stress coefficient can be obtained by a simple multiplication of the d constants by the respective elastic moduli.

For a macroscopic longitudinal piezoelectricity of 5 pC/N and the longitudinal modulus of elasticity of 91.28 GPa [9] a stress coefficient of ∼45 μC cm^–2 can be obtained for a poled and textured hydroxyapatite ceramic. This is three times higher than that in a fluid saturated bone, the longitudinal piezoelectricity and stiffness of which are 0.45 pC/N [29] and 27.40 GPa [9], respectively. The stress coefficient for demineralised bone is considerably lower, ∼10 nC/cm² due to its much lower longitudinal elastic modulus, 49 MPa [30]. This simple calculation signifies that piezoelectricity in poled HA is capable of much more calcification than demineralised bone on the basis of piezoelectric polarization only. This comparison is based on the piezoelectricity of dense hydroxyapatite ceramics where the piezoelectricity of individual microcrystals can become low due to averaging. Local piezoelectricity of individual microcrystals can be even higher, 24–240 μC cm^–2 when taking the estimated single crystal piezoelectricity. This leads us to believe that a suitably engineered hydroxyapatite bone graft is capable of generating significant local piezoelectric charge to stimulate bone growth.

Future directions

The research on piezo-pyro and ferroelectricity in hydroxyapatite has been focused mainly on stoichiometric compositions although substitution can be beneficial to enhance piezoelectricity coefficients and increase thermal and water stability. The later two issues are important for biomedical applications especially to sustain piezoelectricity after autoclaving beyond 135 °C or gamma radiation beyond 20–30 Gy. There are not many studies conducted in this field to decide on the chemistry and the extent of substitution a priori. However, below we will make the case for a few substitutions that can be attempted as a starting point.

The first candidate for substitution can be Cl for OH end member. The Cl ion in chlorapatite, unlike F ion in fluoroapatite, sits farther away from the mirror plane in P63/m structure despite of its spherical shape. This will cause the two corner oxygen atoms in the phosphate tetrahedra in the vicinity of Cl ion to move away further thus causing more rotation of the tetrahedra, which will make the polar phase more stable. In fact, single crystal chlorapatite has shown ferroelectric loop albeit with some leakiness which could have happened due to improper electroding. Figure 3 shows some preliminary data on the high temperature pyroelectricity in chlorapatite. We are confident that both synthesis and sintering methods can be further optimized as well as a suitable poling temperature/field will be found to impart piezoelectricity in chlorapatite both at room temperature and high temperature.

Fig. 3

Pyroelectricity in sintered chlorapatite polled at 375 °C. Pyroelectricity diminishes with thermal cycling.

Strontium-substitution for calcium is promising from this perspective that it is stable in the piezoelectric P63 state in a wide range of substitution and it narrows the channel wherein the OH dipoles order [31]. So, once the OH dipoles are oriented using an electric field at a slightly elevated temperature, upon cooling, the dipoles will find it difficult to flip back and would stay ordered.

As for the electrostatic microdomains, great promises are lying ahead in terms of using such domains to investigate the nature of electrical charge mediated biological interactions on medical devices. The work so far has been carried out using model proteins in a model environment. Proteins and cellular systems of practical significance such as fibronectin, osteopontin, osteocalcin, bone morphogenic proteins, osteoblasts, osteoclasts and human mesenchymal stem cells must be studied to understand the role of static electric charge at the surface in bone related activity. This will then lead towards the ability to understand the role of the dynamic nature of piezoelectric surface charge due to the fact that it is commensurate with a stress cycle. So far this dynamic nature of piezoelectricity has been largely ignored in experimental designs and must be corrected to obtain any meaningful interpretation of the physiological role of piezoelectricity.

Conclusions

In the last decade a few interesting studies took place that resurrected interest in the interest of electrical properties of hydroxyapatite. The key developments in the field of piezoelectric hydroxyapatite and electrostatic domain creations and their potential physiological relevance have been discussed here. There is now enough evidence that investigating electrical properties of hydroxyapatite is an important area of research which now should endeavor to the study of substituted apatite to make possible appropriate applications of piezoelectric hydroxyapatite and elestrostatic domains.