7262
Langmuir 2008, 24, 7262-7268
Interactions of Hydroxyapatite Surfaces: Conditioning Films of Human Whole Saliva Marite´ Ca´rdenas,*,† Juan Jose´ Valle-Delgado,‡,§ Jildiz Hamit,† Mark W. Rutland,‡ and Thomas Arnebrant† Biomedical Laboratory Science and Technology, Health and Society, Malmoe UniVersity, SE-20506 Malmoe, Sweden, and Department of Chemistry, Surface Chemistry, Royal Institute of Technology, SE-10044 Stockholm, Sweden ReceiVed February 5, 2008. ReVised Manuscript ReceiVed April 7, 2008 Hydroxyapatite is a very interesting material given that it is the main component in tooth enamel and because of its uses in bone implant applications. Therefore, not only the characterization of its surface is of high relevance but also designing reliable methods to study the interfacial properties of films adsorbed onto it. In this paper we apply the colloidal probe atomic force microscopy method to investigate the surface properties of commercially available hydroxyapatite surfaces (both microscopic particles and macroscopic discs) in terms of interfacial and frictional forces. In this way, we find that hydroxyapatite surfaces at physiological relevant conditions are slightly negatively charged. The surfaces were then exposed to human whole saliva, and the surface properties were re-evaluated. A thick film was formed that was very resistant to mechanical stress. The frictional measurements demonstrated that the film was indeed highly lubricating, supporting the argument that this system may prove to be a relevant model for evaluating dental and implant systems.
Introduction The mineral hydroxyapatite (HA) is the stable phase of calcium phosphate (Ca5(PO4)3OH) at body temperature and pH > 4.2.1 In some cases, some calcium ions are substituted by other metal ions2 altering the behavior of the mineral.3,4 Currently, there is large interest in hydroxyapatite substrates due to their possible use in bone implant applications.5 Moreover, crystalline HA is the main component of tooth enamel accounting for up to 90% of the volume in its mature state.6,7 Thus, HA can also be used as a model surface for dental studies. When HA is placed in contact with a biological fluid, adsorption of ions and biomolecules readily occurs. This is a common process occurring onto most solid surfaces in aqueous medium and is often referred to the formation of conditioning films.8 A prime example may be the proteinaceous layer called “acquired pellicle”, which develops on tooth enamel surfaces in the oral cavity.9 The acquired pellicle is believed to impart a wide variety of functions including lubrication10 and the control of demineralization and remineralization processes of enamel11 and is also believed to be * Corresponding author. E-mail:
[email protected]. † Malmoe University. ‡ Royal Institute of Technology. § Present address: Institute for Bioengineering of Catalonia, Barcelona Science Park, 08028 Barcelona, Spain.
(1) Suchanek, W.; Yoahimura, M. J. Mater. Res. 1998, 13, 94–117. (2) Robinson, C.,Chemistry of mature enamel. In Dental enamel from formation to destruction; Robinson, C., Kirkham, J., Shore, R. C., Eds.; CRC Press: Boca Raton, FL, 1995; pp 167-191. (3) Kiyoshi, Y.; et al. J Jpn. Stomatol. Soc. 2001, 50, 90–97. (4) Tao, S. X. Durhan Anthropol. J. 2005, 12, http://www.dur.ac.uk/ anthropology.journal/vol12/iss2-3/tao/tao.html. (5) de Bruijn, J. D.; et al. J. Biomed. Mater. Res. 1992, 26, 1365–1382. (6) Jenkins, G. N. The physiology and biochemistry of the mouth, 4th ed.; Blackwell Scientific Publ.: Oxford, 1978. (7) Robinson, C.; et al. Arch. Oral Biol. 2005, 50, 267–270. (8) Loeb, G. I.; Neihog, R. A. AdV. Chem. 1975, 145, 319–335. (9) Hay, D. I. Arch. Oral Biol. 1967, 12, 937–946. (10) Hahn Berg, I. C.; Rutland, M. W.; Arnebrant, T Biofouling 2003, 19, 365–369. (11) Moreno, E. C.; Zahradnik, R. T. J. Dent. Res. 1979, 58, 896–903.
determining for various other processes such as tooth staining12 and plaque formation.13 Recently, HA crystals have been the subject of a wide variety of physical chemical studies to determine their surface properties given that they will profoundly influence the subsequent Biofilm formation.14 Most of the publications deal with measurements on hydroxyapatite particle suspensions through potentiometric measurements,15 surface titration,16 and electrophoretic mobilities,17 although a few of them deal with high resolution measurements by force spectroscopy.14 In general, HA particles have a point of zero charge very close to neutrality ranging from pH 6.5 to 7.3.18 Moreover, it has been demonstrated that phosphate groups predominate at the hydroxyapatite surface both in pure water and at physiological conditions.17 Lately, spatially resolved specific force spectroscopic measurements on the surface of HA pellets have been performed using nanosized probe tips functionalized with COO-- and NH3+-terminated alkanethiol groups.14 They found that for the asymmetric system under study, the data fitted reasonably well to the DLVO (DerjaguinLandau-Verwey-Overbeek) theory yielding a mean negative surface charge density of -0.005 C/m2 at I ) 10 mM although this value varied quite a lot depending on the position along the grain structure. The authors suggested that this effect was due to the existence of different crystal planes at the various facets of the HA grains. The AFM colloid probe is suitable for studying interactions between surfaces at nanoscopic scale.10,19,20 An advantage of (12) Joiner, A.; Jones, N. M.; Raven, S. J. AdV. Dent. Res. 1995, 9, 471–476. (13) Carle´n, A. In SaliVa and bacterial adherence. Studies on components in saliVa and pellicles which may bind bacteria to teeth; Department of Cariology, Go¨teborg University: Go¨teborg, 1996; p 44. (14) Vandiver, J.; et al. Biomaterials 2005, 26, 271–283. (15) Skartsila, K.; Spanos, N. J. Colloid Interface Sci. 2007, 308, 405–412. (16) Harding, I. S.; Rashid, N.; Hing, K. A. Biomaterials 2005, 26, 6818– 6826. (17) Garcı´a Rodenas, L; et al. J. Colloid Interface Sci. 2005, 290, 145–154. (18) Saleeb, F. Z.; de Bruyn, P. L. Electroanal. Chem. Interface Electrochem. 1972, 37, 99. (19) Ducker, W. A.; Senden, T. J.; Pashely, R. M. Langmuir 1992, 8, 1831– 1836.
10.1021/la800402s CCC: $40.75 2008 American Chemical Society Published on Web 06/12/2008
Interactions of Hydroxyapatite Surfaces
the colloid probe technique is not only that well-known model surfaces such as mica or silica can be used21,22 but also the fact that mechanistic information between single or polyasperity contacts can be obtained. Since the use of hydroxyapatite surfaces in this context is novel, it is necessary to establish that the methodology is appropriate and that the obtained results are reasonable. To this end it is possible to extract surface charge and/or surface potential values from the force-distance data by fitting the results to, for example, the Poisson-Botzmann theory. In some cases, particles may be too rough or irregular to allow friction measurements to be performed at all. Therefore, it is necessary to demonstrate that reproducible measurements of friction coefficients and, ideally, reproducible response to the changes in bulk conditions are obtained. In this work, we demonstrate the feasibility of using HA particles for nanoidentation studies, and we monitor the changes in friction behavior in the presence of HWS as both a system of study and benchmarking.
Materials and Methods Materials. Hydroxyapatite (HA) particles were obtained from CAM Implants B V (The Netherlands) and were used and cleaned as described below. The particle diameter ranged typically between 30-40 µm as measured by an optical microscope. Sintered hydroxyapatite discs (HA, diameter of 16 mm, Calcitek, Carlsbad, CA) were used as a comparison. The oxidized silicon surfaces (∼300 Å silicon oxide layer) were obtained from Stefan Klintstro¨m at Linko¨ping University (Linko¨ping, Sweden) and were cleaned as described elsewhere.23 The buffer solution (denoted PBS) was a 10 mM phosphate buffer supplemented with 50 mM NaCl and adjusted to pH 7.0. This was chosen to match the ionic strength of morning saliva as reported by Dawes et al.24 All chemicals used were of at least analytical grade (VWR International, Stockholm, Sweden, Sigma-Aldrich Sweden AB, Stockholm, Sweden). SDS (L6026, >99% pure) used for the desorption studies was purchased from Sigma-Aldrich Sweden AB (Stockholm Sweden) and used without further purification. Water used for salt (nonbuffered) and buffer solutions and surface preparation was of ultra high quality (UHQ), processed in Elgastat UHQ II (Elga Ltd., High Wycombe, Bucks, England). Human whole unstimulated saliva (HWS) was collected from one healthy male adult donor in the morning on the day of experiments as described by Dawes et al.24 The collection was performed 2 h after breakfast. The donor was considered to be in good oral health upon clinical examination. The sampling method used has established reproducible results,25 and it has previously been demonstrated that no statistically significant differences was found in adsorbed amounts of salivary components from a number of individual donors.26 The saliva diluted 10 times with PBS buffer was used without further treatment. The protein content of the saliva was determined to be 1.1 mg ml-1 using a BioRad Micro assay (Bio-Rad Laboratories AB, Sundbyberg, Sweden). The committee on research ethics at Lund University has approved this study (No: LU 518-02). Substrate Preparation. The HA particles were glued to either a freshly cleaved mica surface or a tipless cantilever with a spring constant of ∼0.1 N/m and left to dry overnight. The procedure for (20) Attard, P.; Carambassis, A.; Rutland, M. W. Langmuir 1999, 15, 553– 563. (21) Bogdanovic, G.; Tiberg, F.; Rutland, M. W. Langmuir 2001, 17, 5911– 5916. (22) Sternstedt, J.; Brumer, H.; Zhou, Q. Biomacromolecules 2006, 7, 2147– 2153. (23) Ca´rdenas, M.; Elofsson, U.; Lindh, L. Biomacromolecules 2007, 8, 1149– 1156. (24) Dawes, C. Int. J. Chronobiol. 1974, 2, 253–279. (25) Vassilakos, N. Some biophysical aspects of salivary film formation. Studies of salivary adsorption at solid/liquid and air/liquid interfaces. Ph.D. Dissertation, Lund University, 1992. (26) Lindh, L.; et al. Biofouling 1999, 14, 189–196.
Langmuir, Vol. 24, No. 14, 2008 7263 colloidal probe preparation is described elsewhere.27 The particles and cantilevers were rinsed carefully with Millipore water and ethanol, dried in N2, and plasma cleaned for 30 s prior to use. The macroscopic HA surfaces were polished on wet 1200-grit silicon carbide paper (3 M United Kingdom, Bracknell, Berks, UK) followed by successive polishings with 3, 1, and 0.25 µm diamond polishes (Kemet International, Maidstone, Kent, UK) using a Buehler Metaserv Grinder-Polisher (Buehler UK, Coventry, Warwickshire, UK) for 10 min. The HA surfaces were first rinsed in ethanol and then sonicated in ethanol for 30 min. This was followed by extensive rinsing in water, treated in detergent solution, and finally rinsed extensively in water, ethanol, and water. The cleaned surfaces were dried under nitrogen and kept in a sealed Petri dish. Prior to the adsorption measurements the surfaces were plasma cleaned for 5 min. Atomic Force Microscopy. AFM measurements were performed using a scanning probe microscope (Picoforce multimode SPM with a Nanoscope IV control unit, Veeco). A Picoforce scanner with closed loop in z direction was used both in imaging and spectroscopic mode. All measurements were performed in a multimode liquid cell. In the imaging mode, a silicon nitride tip (Veeco, type DNP) with cantilever spring constant of 0.06 N/m was used. The height images were subjected to first order flattening to remove offset and tilt of each line and the mean roughness was calculated using the software provided by nanoscope. In the spectroscopic mode, tipless cantilevers carrying a HA particle were used to acquire force curves and perform frictional measurements. Both the normal and torsional spring constants were measured for each cantilever using the methods proposed by Sader et al.28 and Bogdanovic et al.,29 respectively. The software AFM Tune IT v2.5 (ForceIT, Sweden) was used in the determination of the normal spring constants of the cantilevers. The colloid probe studies followed the guidelines of an IUPAC report on the subject.30 The relative merits of the different torsional spring constant techniques are discussed in a recent paper.31 In the experiments of HA particle-particle interaction, the particle on the cantilever was carefully aligned on top of the particle glued on the mica substrate before bringing the probe into contact with the substrate. Then, the liquid cell was flushed with various aqueous solutions of distinct ionic strength (I), and force and friction measurements were carried out in each solution. In the experiments with saliva, saliva samples were diluted to 10 (v/v) % in 10 mM phosphate buffer containing 50 mM NaCl and injected in the liquid cell to allow the adsorption on HA surfaces for 1 h. Once the salivary film was formed, force and friction measurements were performed after the liquid cell was subjected to extensive rinsing with buffer solution to avoid problems due to light scattering and due to the high viscosity of the salivary solution. The normal forces were measured at a frequency of 1 Hz and a typical scan size of 5 µm. The friction force was calculated from the so-called friction force loops, in which the lateral voltage signal from the AFM (related to the twist of the cantilever) is recorded as the substrate slides under the colloid probe at a fixed applied load. The friction force is proportional to the difference between the lateral voltage signal values obtained in the forward and reverse scan directions.30,32 The friction was measured as a function of increasing and then decreasing load. In all measurements a scan size of 2 µm was used for the relative movement of the surfaces. Characterization of HA Particles. Hydroxyapatite particles were glued on mica and imaged using AFM in contact mode. A typical deflection image of the flattest part of the particle is given in Figure 1. The figure represents a surface area of 5 × 5 µm2. The HA particle surface is composed of smaller grains having an apparent diameter of 2.2 µm. The mean square roughness of the flattest part of the (27) Feiler, A.; Plunkett, M. A.; Rutland, M. W. Langmuir 2003, 19, 4173– 4179. (28) Sader, J. E.; Chon, J. W. M.; Mulvaney, P. ReV. Sci. Instrum. 1999, 70, p. 3967–3969. (29) Bogdanovic, G.; Meurk, A.; Rutland, M. W. Colloids Surf. B 2000, 19, p. 397–405. (30) Raltson, J.; et al. Pure Appl. Chem. 2005, 77, 2149–2170. (31) Pettersson, T.; et al. ReV. Sci. Instrum. 2007, 78, 093702. (32) Mate, C. M. J. Res. DeV. 1995, 39, 617–627.
7264 Langmuir, Vol. 24, No. 14, 2008
Cárdenas et al.
Figure 1. The surface of a HA particle as imaged by deflection contact mode AFM before (a) and after (b) adsorption of HWS. An area of 5 µm × 5 µm is shown. The size distribution of grains in presented in c. The mean square roughness for flat part of the HA grains before and after adsorption of HWS was 4 ( 2 nm and 12 ( 3 nm, respectively.
grains was measured to be 4 ( 2 nm (average between 26 different grains). Similar results were reported earlier for hydroxyapatite pellets.14,33 For our purposes, i.e. the analysis of the interaction forces, the radius of curvature of the contact region of the grains of the HA particles is the value of interest. In order to get an estimate of the real radius of curvature of the grains, more than 90 height profiles of different grains were fitted to the equation of a circumference. In this way, a radius of curvature of 2.2 ( 2.5 µm was obtained. The extremely large uncertainty in this value is a result of the high polydispersity of the grain size and geometry, as evidenced in Figure 1c. Although most of the grains are below 6 µm, a few of them were very large and reached 13 µm. We decided to include such large grains in the calculation of the radius of curvature to get a force analysis that reflects the real system, although the results still have to be considered as an estimate of the real values. As an alternative substrate, a flat surface of HA was probed and a typical image is given in Figure 2. The surface presents areas where the roughness is quite low (2 nm) although a few defects are present that are as deep as 1.5 µm. For the force measurements, the flat surface was first imaged and a suitable spot with low roughness was located. Then, the colloidal probe cantilever was carefully aligned in the selected area to ensure a smooth contact area. Measurements within these regions provide more precise force-distance data than for HA particles. However, the use of the irregular HA particles implies that the studied system is closer to the real biological system. (33) Vandiver, J.; et al. J. Biomed. Mater. Res. 2006, 78A, 352–363.
Figure 2. Surface plot of a deflection image of the macroscopically flat HA surfaces.
Results Typical approach force curves between a HA particle and a HA flat substrate at various ionic strength conditions (1 mM NaCl, 10 mM NaCl, and 10 mM phosphate buffer containing 50 mM NaCl) are given in Figure 3, including their corresponding fits to the DLVO theory.34,35 Force curves before and after friction experiments did not show any significant difference. This is an indication that no polishing of the HA surfaces occurs during the friction measurements.
Interactions of Hydroxyapatite Surfaces
Langmuir, Vol. 24, No. 14, 2008 7265
Figure 3. Approach force curves for hydroxyapatite surfaces immersed in 1 mM NaCl (a), 10 mM NaCl (b), and PBS buffer (c). Solid lines are best fits to the DLVO theory (see Table 1). Approach force curves for the asymmetric system HA particle and silica substrate immersed in 1 mM NaCl is given in (d).
As is usual, the force curves have been normalized by the effective radius R, given by:
R)
R1R2 R1 + R2
(1)
where R1 and R2 are the radii of curvature of the surfaces in the contact area. The topography of the HA particles suggests that the radii used to describe the interaction should be that of two smooth grains rather than assuming that the interaction occurs between two smooth particles of about 35 µm in diameter. Therefore, the average radius of curvature of a grain (2.2 ( 2.5 µm) was used for the calculation of the effective radius R according to eq 1. Thus, in the case of HA particle-particle interaction a value R ) 1.1 ( 1.3 µm was used (R1 ) R2 ) 2.2 ( 2.5 µm), whereas for the system HA particle-flat substrate eq (1) yields R ) 2.2 ( 2.5 µm (the radius of curvature of the HA flat substrate is infinite). According to the DLVO theory, in the limit of low surface potentials the interaction force F between two surfaces is given by:36
2 Ψ1Ψ2 - exp(-κ x)(Ψ21 + Ψ22) F ) 2πε0εrκ exp(-κ x) R 1 - exp(-2κ x) A (2) 6x2 where R is the effective radius given by eq 1, x is the separation distance between the surfaces, Ψ1 and Ψ2 are the electrical surface potentials of each surface, A is the Hamaker constant of the system, εo is the dielectric constant in vacuum, εr is the relative (34) Verwey, E. J. W.; Overbeek, J. T. G. Theory of the stability of lyophobic colloids; Elsevier: New York, 1948; p 205. (35) Derjaguin, B. V.; Landau, L Acta Phys. Chim. USSR 1941, 14, 633–662. (36) Israelachvili, J. N. Intermolecular and Surface Forces, 2nd ed.; Academic Press: London, 1991. (37) Hough, D. B.; White, L. R. AdV. Colloid Interface Sci. 1980, 14, 3–41.
dielectric constant of the medium (aqueous medium in our case), and κ, the inverse Debye length, is given by
κ)
∑ zi2e2ni0 i
ε0εrkT
(3)
where e is the charge of the electron (absolute value), k the Boltzmann constant, T the absolute temperature, ni0 the concentration of ions (number of ions per unit volume) of type i in the bulk, and zi the valence of such ions. The negative sign in the numerator of the first term in eq 2 corresponds to interaction at constant surface potential, whereas the positive sign corresponds to interaction at constant surface charge. Both Hamaker constant and electrical surface potentials were used as fitting parameters. The same value for Ψ1 and Ψ2 (Ψ1 ) Ψ2 ) Ψ0) was obtained in each individual fit, which was expected since the interacting surfaces are of the same nature (HA). The best fits to the experimental data were obtained with the DLVO theory by assuming interaction at constant surface potential. The values of the fitting parameters in that case were also much more reliable than those obtained by assuming interaction at constant surface charge. In the latter case, the results of surface charge density does not follow the expected trend with ionic strength and the values of Hamaker constant were largely inconsistent within the various cases studied and were, in general, much higher than expected. Therefore, only the results of the DLVO fits under the condition of interaction at constant surface potential are presented in this paper. Note that given the uncertainty in the radius, the values from the fits can be considered at best an estimate. Table 1 shows a summary of the values of the electrical surface potential Ψ0 obtained for the systems HA particle-particle and HA particle-flat substrate in different media. In all the cases, a value of A ) 1 × 10-21 J gives rise to good fits. This value for the Hamaker constant of HA in aqueous medium should be considered as a rough estimate, which is of the same order of
7266 Langmuir, Vol. 24, No. 14, 2008
Cárdenas et al.
Table 1. Electrical Surface Potentials of HA Surfaces Obtained by Fitting the Experimental Force Curves to the DLVO Theory under the Condition of Interaction at Constant Surface Potentiala aqueous medium
HA particleparticle, mV
HA particle-flat substrate, mV
1 mM NaCl 10 mM NaCl PBS buffer
Ψ0 ) -24 ( 2 Ψ0 ) -12 ( 3
Ψ0 ) -33 ( 2 Ψ0 ) -26 ( 1 Ψ0 ) -22 ( 2
a In all the cases, a value of A ) 1 × 10-21 J was assumed for the Hamaker constant of the system.
magnitude as the Hamaker constants of many materials in aqueous medium.37 For the asymmetric system HA particle-silica substrate in 1 mM NaCl (Figure 3d), the DLVO theory was used to fit surface potentials of both HA and silica surfaces. In this case, the Hamaker constant of the system HA-water-silica was fixed to the value obtained by the following equation:36
AHA-water-silica ) √Asilica-water-silica × √AHA-water-HA (4) By taking AHA-water-HA ) 1 × 10-21 J (from the symmetric case) and Asilica-water-silica ) 8.3 × 10-21 J,36,38 a value of AHA-water-silica ) 2.9 × 10-21 J was obtained. The best fit to the DLVO theory yields surface potentials of -33 mV for the HA surfaces (in agreement with values in Table 1) and of -45 mV for the silica surface (in good agreement with those reported earlier19,39). Figure 1b shows a deflection image of the HA surfaces coated by the HWS film. As can be observed, the film surface seems more porous than the clean particle, yielding a much larger mean square roughness (12 ( 3 nm) than for the clean particle. Typical approach force curves for HA surfaces coated with a salivary film before and after treatment with surfactant are given in Figure 4. Note that the given distances are relative and not absolute. One of the main disadvantages of the AFM in a surface force context is the impossibility of directly measuring the separation distance between the interacting surfaces. The zero of contact is inferred as the point at which the surfaces achieve “constant compliance”, i.e. the movement of one surface is transmitted directly to the other. This does not take account of any adsorbed films but otherwise is a reasonable approximation for hard surfaces. In the case of compliant films, the thickness of the compressible layer is experimentally inaccessible with the AFM, which gives rise to an uncertainty in the origin of separation distances. However, in order to compare the force curves obtained under different conditions, null separation between the HA surfaces coated with a salivary film at the highest applied load (8 mN/m) has been assumed. Force curves for the salivary film were also acquired after the friction measurement, and no significant changes were observed (data not shown). The force curves for saliva-coated HA surfaces cannot be fitted to DLVO theory since the steric repulsion due to the compression of the salivary film is not taken into account in that theory. The curves present a long-range repulsive force with onset at ∼70 nm of separation. On separation, a significant attractive force regime exists (inset Figure 4). Such attractive force has a minimum at about -0.4 mN/m at a separation that ranges from 40-60 nm. At distances larger than this point, the attractive force starts to decrease in a step-like manner, presenting at least one additional smaller minimum. The total range of the attractive force, extending (38) Lyklema, J. Fundamentals of Interface and Colloid Science; Academic Press: London, 1995; Vol. 2. (39) Horn, R. G.; Smith, D. T.; Haller, W. Chem. Phys. Lett. 1989, 162, 404– 408.
Figure 4. Approach force curves on HA surfaces coated with a salivary film before (crosses) and after (filled circles) treatment with surfactant. The corresponding data for the clean system (open circles) is also included in the figure. The inset gives the approach and retract force curve for HA surfaces coated with a salivary film (closed and open symbols, respectively). The aqueous medium is PBS buffer.
Figure 5. Force curves upon approach for HA surfaces that have been exposed to HWS, SDS, and buffer before (crosses) and after (filled circles) the friction measurement. The corresponding data for the clean system is also given in the figure (open circles). The aqueous medium is PBS buffer.
out to ∼250 nm, is much longer ranged than the repulsive force upon approach. Major changes in the force curve are observed when the saliva-coated HA surface is washed with a 0.5 (v/v) % SDS solution followed by extensive rinsing with buffer solution. In particular, the range of the repulsive force is considerably reduced presenting an onset at ∼30 nm. This repulsive force is not of electrostatic origin since it cannot be satisfactorily fitted to the DLVO theory. In the case of the particle-particle system, significant differences in the force curves before and after the friction were found with the salivary film after surfactant treatment as shown in Figure 5. The shape of force curve (postfriction) is rather similar to that of the clean system, and indeed a reasonable fit to the DLVO theory can be obtained, although yielding slightly higher surface potential values. Note that no significant adhesion upon retraction was measured for the clean system or the coated surface after treatment with SDS (data not shown).
Interactions of Hydroxyapatite Surfaces
Langmuir, Vol. 24, No. 14, 2008 7267
Figure 6. Friction force Ffriction versus applied load Fload for the HA particle-particle system before and after the formation of HWS film, as well as after the HWS film was treated with surfactant. The aqueous medium is PBS buffer.
Frictional measurements were performed for the particleparticle system before and after the formation of a salivary film, as well as after the HWS film was treated with SDS. Typical results are shown in Figure 6. The friction loop was measured along a 2 µm line given that the diameter of an individual grain is 2 µm. From the friction loop it was easy to determine whether the contact area included one or more grains, but the measured friction force was independent of this. The friction coefficient (calculated from the slope of friction versus load forces) decreased from 0.44 ( 0.08 for the clean HA-HA system to 0.20 ( 0.07 after the formation of a salivary film onto the HA surfaces. This implies a reduction of ∼55% in the friction coefficient. If the salivary film was treated with an anionic surfactant (SDS) and subjected to extensive rinsing with buffer, the friction coefficient measured was 0.42 ( 0.04. This indicates that SDS treatment efficiently eliminates the lubricating properties of salivary films. Friction forces were also measured for the HA particle/HA macroscopically flat surface, and the same effect of salivary film formation and surfactant treatment was observed (data not shown), although the reduction in friction coefficient due to salivary film formation was considerably larger (∼89%). After SDS treatment, the friction coefficient was only 10% lower than for clean HA surfaces.
Discussion The best fit for the approach force data to the DLVO theory (Figure 3) indicates that the repulsive force is better described by modeling the surface to a constant surface potential. The results of the best fit (Table 1) show that both the microscopic particles and flat macroscopic surfaces are negatively charged. The surface potential values obtained from the DLVO fits are absolute values. The negative sign is deduced by the fact that the interaction with silica surfaces of known negative charge is repulsive. This is indeed in agreement with the fact that phosphate groups are predominant on HA surfaces at physiological conditions.17 Note that increasing the ionic strength, I, decreases the surface potential as predicted by the DLVO theory. In order to calculate the corresponding surface charge density σo, we can use the following equation valid for low surface potentials Ψ0:
σ0 ) ε0εrκ Ψ0
(5)
A diffuse layer model (DLM) for the charge distribution in the electrical double layer around the hydroxyapatite surfaces is
assumed in eq 5. This model was chosen because it is the simplest model that is able to explain the experimental results of our system. Other models like the basic Stern model (BSM) or the triple layer model (TLM) require the inclusion of additional fitting parameters which are difficult to corroborate experimentally. Furthermore, the constant capacitance model (CCM) is clearly not appropriate for our system because the presence of an electrical double layer repulsion (long-range repulsion that depends on the ionic strength, see Figure 3) in the force curves between hydroxyapatite surfaces implies the overlap of existing diffuse layers.40 For the case of HA particle-HA flat macroscopic substrate immersed in 10 mM NaCl, Ψ0 ) -26 mV (see Table 1), which yields σo ) -0.006 C/m2. This value is very close to that reported by Vandimer et al.14 for HA pellets at similar I (σo ) -0.005 C/m2). The value for the Hamaker constant of HA used in the fits is low (1 × 10-21 J) but of the same order of magnitude than that for other materials in aqueous solutions. Vandimer et al.14 also found a similar value of Hamaker constant (1.3 × 10-21 J) for HA pellets. Slightly larger surface potentials are obtained when the flat HA surface was used. This could be a result of the better defined geometry of the system, which should give a better defined plane of charge. It is also possible that, given the high polydispersity of grain’s geometry, the size of the grains in contact is different in each case. In any case, the assumption of the grain size as the interaction radius for HA particles is validated by three separate facts: (i) that the surface charges obtained by this approach agree well with independent measurements, (ii) that changing the geometry of the lower surface from particle to flat surface does not appreciably change the value of the potential extracted, and (iii) that the surface potential extracted from the HA-silica system returns a sensible value for the silica potential, which has been extensively measured over the years.27,41,42 Given that the particle size and grain size differ by a factor of 10, none of these three validations, or the reasonable Hamaker constant would have been possible if the particle radius had been used. Vandiver et al.14 showed that the grains on the HA particle presented facets with a net negative surface charge density although the absolute value varied. Heterogeneous charge distributions that would give irreproducible force curves and also the presence of attractive double layer forces, which are not present in our force curves. Thus, our results do not give evidence of varied charged density on the grains. This could be due to two different reasons: (1) only particular facets are important for the interaction (i.e., same facets always outermost) or (2) that the surface charge does not in fact differ greatly between facets. Recent publications have shown the presence of nanosubstructures in crystalline hydroxyapatite that carry opposite charge (see as an example ref 7). Oppositely charged nanosubstructures would lead to unmistakeable attractive forces in the force curves, which are not observed in our force curves. This implies that such nanostructures are either extremely rare in our particles, or sufficiently small that they can be considered homogeneous on the scale of the contact area of our measurements. Given that we have determined the relevant surface properties of HA particles for the interpretation of surface and frictional forces, we can use them for the study of “conditioning films”. In the following, we will show that the spontaneous film formation from human whole saliva on hydroxyapatite allows us a better understanding of the adsorbed film structure and frictional properties. The approach force curves after a salivary film was formed (Figure (40) Lu¨tzenkirchen, J. J. Colloid Interface Sci. 1999, 217, 8–18. (41) Hahn Berg, I.; Rutland, M.; Senden, T. Langmuir 1993, 9, 412–418. (42) Ducker, W. A.; Rutland, M. W.; Pashely, R. M. J. Colloid Interface Sci. 1989, 130, 448–456.
7268 Langmuir, Vol. 24, No. 14, 2008
4) show that long-range repulsive forces are found between the two coated surfaces, due to steric rather than electrostatic forces. The outer region of the repulsive force in the approach curve presents an exponential behavior with a decay of tens of nanometers. This value is much larger than the Debye length in the system (1 nm) and therefore can be interpreted as being related to the thickness of the adsorbed layer.43,44 In this way, a 25 nm apparent layer thickness is estimated. This value is in accordance with that measured by neutron reflectivity for HWS on sapphire surfaces at similar conditions, which are believed to have similar surface properties (being hydrophilic and possessing a similar charge) as those of hydroxyapatite.45 A similar type of behavior, including the observation of the decay length being much larger than expected at the ionic strength conditions, was observed for salivary films adsorbed to mica surfaces studied by the interferometric surface force apparatus.46,47 On retraction (inset in Figure 4), a long-range, shallow attraction was measured, more likely due to multiple bridging of biomacromolecules. Note that this bridging extends 250 nm from contact. This means that very large biomacromolecules or aggregates of them are present at the interface. At this stage, it is relevant to point out that the film structure is not affected by the friction experiment since no major changes in the force curves could be detected afterward. The latter agrees with previous reports on salivary films in which AFM imaging under hard contact conditions (at force loads equivalent to the compliance region) did not succeed in scratching the film.23,45 Thus, salivary films on various substrates (hydroxyapatite, sapphire, and hydrophobically modified silica) show high compressibility and mechanical resistance. After reviewing the load force curves, we can now understand the frictional properties of the films on HA. Figure 6 clearly shows the good lubricating properties of saliva. The friction coefficient is reduced to half of that of the clean particle-particle system although the effect is much larger for the particle-flat surface system. The particle surface is much more irregular than the flat substrate, which might induce larger friction even after the adsorbed film was formed. Sternstedt et al.22 demonstrated that the friction coefficient was determined by the roughness (log rms) and the nature of the film. In the particle case, the roughness is so high that its contribution is still significant after the HWS film was formed, even though the extremely efficient lubrication capacity of this film. The lower roughness of the flat HA surface leads to the lubricating effect of the HWS film to be proportionally larger and thus give a larger reduction of the friction coefficient. If we compare our data with the friction data previously reported for salivary films on a silica particle and a macroscopic silica surface,10 a much higher reduction in friction was obtained in their case. This is likely due at least in part to the smoother surfaces but in addition to the different nature of the underlying surfaces. It has been shown earlier that the morphology of the salivary films differ considerably depending on the type of substrate.23,45 Few but very large aggregates were present in the salivary film on substrates with hydrophobic character while much smoother films formed on hydrophilic ones. AFM imaging (43) Braem, A.; et al. Langmuir 2003, 19, 2736–2744. (44) Braem, A. D.; Campos-Teran, J.; Lindman, B. Langmuir 2004, 20, 6407– 6413. (45) Ca´rdenas, M.; et al. Biomacromolecules 2007, 8, 65–69. (46) Nylander, T.; et al. Prog. Colloid Polym. Sci. 1998, 108, 34–39. (47) Nylander, T.; Arnebrant, T.; Glantz, P.-O. Colloids Surf. A: 1997, 129-130, 339–344.
Cárdenas et al.
(Figure 1b) shows a film that covers the entire surface, giving the surface a more porous appearance. The film surface has a higher roughness than the clean particle. The fact that the friction coefficient decreases upon salivary film formation even though the film has a higher roughness reflects that this interface is softer than that of the clean particles. Anionic surfactants are commonly present in toothpaste and mouthrinse formulations. Therefore, studying the effect of anionic surfactants on salivary films is of high relevance. SDS treatment largely removes the adsorbed salivary film although not to completeness (Figure 4 and Figure 5). The remaining material is, however, efficiently removed during the friction experiment (Figure 5). This indicates that the mechanical stability of the salivary film is lost by the solubilization of certain components by SDS. Moreover, SDS treatment eliminates the lubricating properties of saliva (Figure 6), in agreement with the idea of layer removal and the fact that the normal force curves change after the friction measurements, becoming more similar to those of the clean system (Figure 5). By using the normal forces and frictional information in combination with previous imaging AFM and neutron reflectivity data,23,45 we can now outline a model of the salivary film on HA. It can be considered as a densely packed adsorbed layer onto which a more diffuse outer layer, more likely composed by larger clusters, sits. The inner and denser layer is quite thick and should be responsible for the good lubrication of salivary films. The outer layer, on the other hand, might protect the inner layer from mechanical stress. SDS treatment removes the outer layer and thus the mechanical stability of the film. The presence of clusters in the film would explain the friction and force data as well as the neutron reflectivity data, although these results do not provide direct proof of their existence. The AFM imaging on HA did not show the presence of such aggregates but instead showed a rougher interface. However, AFM imaging is restricted to a very small area and therefore we could have failed to visualize the aggregates. On sapphire, on the other hand, AFM imaging showed the presence of a few but very large aggregates that decorated the inner HWS film.
Conclusions Hydroxyapatite has been used for first time in colloid probe AFM mode. The HA particles have been shown to consist of grains, and the surface force measurements clearly indicate that the grain size is the relevant unit to use to scale the forces. Films of human whole saliva were formed on HA, and they were shown to resist rinsing with buffer solution. Such film formation significantly reduced the friction, as expected. Elution of the salivary film with the anionic surfactant SDS was shown to disrupt the film. Finally, the particle-particle HA system proved a useful model that can be used to probe e.g. bioadhesion and biolubrication. Indeed, the approach could be used to study how individual proteins interact with HA, or how HA interacts with, for example, implant materials, etc. Acknowledgment. The authors thank Liselott Lindh for advice during saliva sampling. M.C. and T.A. thank the Knowledge Foundation (KK-stiftelsen, Biofilms - Research Centre for Biointerfaces), Malmo¨ University, and the Swedish Larynx Foundation for financial support. J.J.V.D. acknowledges the financial support from the Spanish “Fundacio´n Ramo´n Areces”. M.W.R. is a research fellow of the Swedish Research Council. LA800402S