Available online at www.sciencedirect.com
Diamond & Related Materials 17 (2008) 1089 – 1099 www.elsevier.com/locate/diamond
Commercial developments of nano-crystalline diamond — Two prototypes as case studies F.R. Kloss a , L.A. Francis b , H. Sternschulte c , F. Klauser d , R. Gassner a , M. Rasse a , E. Bertel d , T. Lechleitner e , D. Steinmüller-Nethl c,⁎ a
e
Department for Cranio-Maxillofacial and Oral Surgery, Medical University of Innsbruck, Maximilianstr. 10, 6020 Innsbruck, Austria b IMEC, Kapeldreef 75, 3001 Leuven, Belgium c rho-BeSt coating Hartstoffbeschichtungs GmbH, Exlgasse 20a, 6020 Innsbruck, Austria d Institute of Physical Chemistry, Leopold-Franzens University of Innsbruck, Innrain 52a, 6020-Innsbruck, Austria Division of Physiology, Department for Physiology and Medical Physics, Medical University of Innsbruck, Fritz-Pregl-Str. 3, 6020 Innsbruck, Austria Available online 15 January 2008
Abstract Novel technologies for synthesis of nano-crystalline diamond (NCD) enable industrial production allowing large area deposition on a variety of substrate materials – at reasonable price. New perspectives for future innovative products emerge demonstrated by two case studies in the field of micro electro-mechanical systems (MEMS) sensors (case a) and medical implant devices (case b). a) This study comes as a preliminary step towards the integration of NCD thin film membranes in gravimetric sensors with low detection limits. We investigate theoretically and experimentally the mass sensing characteristics of composite thin Film Bulk Acoustic Resonator (FBAR) as a function of the side exposed to a mass perturbation. b) The aim of this study was to demonstrate the influence of different surface terminations of NCD on surface potentials and subsequently its influence on in vivo connective tissue healing. NCD-coated implants were evaluated by X-ray photoelectron spectroscopy (XPS) and atomic force microscopy (AFM)-surface potential measurements. After in vivo integration of the NCD-membranes into the subdermal layer of Wistar rats and subsequent histological evaluation it was demonstrated that the number of cells increased significantly at the O-terminated NCD and the scar tissue formed was less tight. Thus, a promising technique for controlling connective tissue adhesion in vivo is presented. © 2008 Elsevier B.V. All rights reserved. Keywords: Diamond film; Nano-crystalline; Biomedical applications; Micro electromechanical systems (MEMS)
1. Introduction Only during the last decade, a variety of cutting-edge innovations based on carbon and its respective allotropes have been further developed, in particular regarding applications in hightech areas such as physics and modern biomedicine [1]. Amongst the plethora of carbon configurations, diamond is outstanding, because it exhibits extraordinary physico-chemical properties. Hence, besides its unprecedented utilization in tool industry and tribology, many attempts are now being undertaken to put diamond in life science, sensor technology or
⁎ Corresponding author. Tel.: +43 512 283559 11; fax: +43 512 283559 99. E-mail address:
[email protected] (D. Steinmüller-Nethl). 0925-9635/$ - see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.diamond.2007.12.061
electronics into practice [2]. Progress in the broad application of diamond in life science devices was delayed, because of high costs for production or refinement. Development of novel technologies not only expedited the synthesis of nano-crystalline diamond (NCD) [3], which exhibits similarly remarkable properties as single crystal diamond, at reasonable price, yet more than that, opened new avenues for future diamond research. The latter clearly calls for innovative product development to be commercialized in a broad area of applications in the near future. The nano-crystalline and ultra nano-crystalline diamond (NCD/UNCD) films used in this study were obtained from rhoBeSt coating GmbH. Using a modified hot-filament technique and varying the deposition parameters, i.e. gas mixture, temperature, pressure and
1090
F.R. Kloss et al. / Diamond & Related Materials 17 (2008) 1089–1099
substrate seeding, CVD growth of diamond results in different types of films [4]. These are generally classified according to the crystal grain size as micro crystalline (grain size 1–5 μm), nano-crystalline (NCD) (grain size about 50 nm) and ultra nanocrystalline (UNCD) (grain size below 5 nm). Crystallites with sizes down to 3 nm display special advantages. Due to the small dimensions of nano-crystallites, the relation of volume-to-surface increases dramatically, leading to an exceptionally high amount of surface atoms, which results in distinct physical and chemical properties. Notably, NCD-films used in this study exhibit a particularly low content of sp2-hybridised carbon (≤3%) [5], a rootmean-square roughness of about 10 nm even for thicker films, a high transmissivity, when films are deposited on glass (thickness below 1 μm), a compressive stress below 3.3 MPa (for cantilever lengths up to 2 mm) as well as a controllable calibration of compressive as well as tensile stress. We here present our most recent results based on NCD prototypes from two case studies in the field of micro electromechanical system (MEMS) sensors (case a) as well as medical implant devices (case b). 1.1. Case a Mechanical and chemical properties of NCD may be further refined and exploited to develop sophisticated MEMS sensors. We therefore established means to integrate NCD in a thin film bulk acoustic resonator (FBAR). This structure has been described in recent theoretical and experimental work, corresponding to solidly mounted or membrane supported acoustic resonators [6–10]. FBAR is promising for the fabrication of highly mass sensitive miniature acoustic devices for gas detection and sensitive (bio)chemical assay systems. Our study was aimed at fabricating a four-layered air-backed membrane consisting of a piezoelectric layer sandwiched between two electrodes, which are stacked on a supporting layer that provides the necessary mechanical stability to the complex structure. The membrane is rigidly attached to a frame of micro-machined silicon. As a result of the electrical transduction evoked by the piezoelectric layer, a longitudinal bulk acoustic wave causes the membrane to vibrate at a certain resonance wavelength. Any disturbance of the acoustic field, e.g. when mass is adsorbed at the surfaces of the membrane, can thus be sensitively detected by recording/monitoring the shift in resonance frequency. The extremely high acoustic velocities transduced in NCD as supportive layer provide a variety of advantages over other commonly used materials [11]: increased resonance frequency, high values of the electromechanical coupling coefficient and a high mass sensitivity for mass adsorption on the piezoelectric actuation side of the composite MEMS structure. 1.2. Case b The influence of surface polarity on connective tissue healing in vivo was studied by comparing O-(O-NCD, hydrophilic) and H-terminated (H-NCD, hydrophobic) NCD coatings to pure titanium substrates. Endosseous implants for the rehabilitation of oral function or for anchorage of epitheses have a transmucosal/-
dermal connection to the external environment. This implant/soft tissue interface is characterized by two zones: the epithelial layer and the layer of connective tissue. The importance for the clinical outcome of a tight attachment of connective tissue to implants has been emphasized in numerous publications [12–14]. The epithelial layer is thereby supported and its apical migration is blocked [15]. Both the surface roughness/micro-architecture and the respective material surface properties have been the main target of investigations aiming at the improvement of the tight soft tissue attachment to implants. It has been demonstrated that micro-roughened titanium surfaces with grooves between 3 and 30 μm inhibit the epithelial down growth along the implant to a certain extent [12,16]. In contrast little is known about the influence of surface wettability, differences in chemical surface characteristics or electrical properties on the adherence of connective tissue and cells in vivo. It has been shown in vitro that electric field and surface charges improve cell adhesion and cell growth [17,18]. This phenomenon is believed to be caused by the binding of inorganic cations as well as cationic amino-acids and proteins to the negatively charged implant, which in turn makes the implant surface attractive for the negatively charged cell membrane [19]. Higher number of polar functional groups, in our case oxygen-containing groups, increased the number of H-bridge formation sites and thus the wettability by the adjacent aqueous medium or body fluid. It is assumed that the adsorption of extracellular matrix (ECM) which is needed by cells for binding, movement and communication will be facilitated by an appropriate surface potential and polarity. There is also further evidence for the positive influence of hydrophilic surfaces on cell adhesion and cell behaviour in vitro, which in due course may result in an improved osseointegration in vivo [20–22]. Data on the effect on the connective tissue adherence in vivo is however missing to date. 2. Experimental 2.1. Case a For experimental characterization, composite FBARs were fabricated as a four-layered air-backed membrane depicted in Fig. 1. First a 2.3 μm thick nano-crystalline diamond film was deposited on a clean 4 inch silicon substrate. Afterwards the piezoelectric layer and its electrodes were obtained by the reactive sputtering of an Al/AlN/Al stack above the diamond surface (Nimbus310, Nexx Systems, MA). The aluminum layers were 200 nm thick and the aluminum nitride layer was 800 nm thick. The top aluminum layer was patterned and wet etched to form electrodes shaped as squares of 200 and 300 μm edge sizes for the electrical RF probing of the device. The composite structure was micro-machined by subsequent back-side deposition of a masking layer (200 nm Al) with photo-defined patterns of mmlong side square membranes followed by a silicon deep reactive ion etching (DRIE, Bosch Process). Diamond is an automatic etch stop material for this process. After fabrication, the membranes were virtually flat thanks to a very low residual compressive stress in the multilayered stack of the membranes. Fig. 2 shows the optical mapping (Veeco Wyko NT3000) of the top structure of one of the membranes.
F.R. Kloss et al. / Diamond & Related Materials 17 (2008) 1089–1099
Fig. 1. Structure of the composite FBAR: a piezoelectric layer of thickness is sandwiched between two metal electrodes to electrically excite a longitudinal wave resonance mode in the stack formed by the piezoelectric and a “loading layer”.
2.2. Case b The etched and annealed titanium membranes with a diameter of 1 cm were manufactured and supplied by MedEl GmbH. These substrates were coated with a closed, 1 μm thick layer of nano-crystalline diamond (NCD). The NCD films were grown as described by means of a modified hot-filament chemical vapour deposition technique [4]. After the coating process, the dangling bonds at the surface were hydrogen terminated (H-NCD), hence these films exhibited hydrophobic properties. Thermal treatment of the NCD-surface at 400 °C for 4 h with 21% oxygen, replaced hydrogen for oxygen containing groups, i.e. carbonyl, ether or hydroxyl groups, rendering the film hydrophilic (O-NCD) [4,22]. Therefore the investigation compares three different groups within the study: pure titanium as provided (control), H-NCD (hydrophobic) and O-NCD (hydrophilic). Contact angles were measured on NCD films to get information about the hydrophobicity or hydrophilicity of the surfaces. The contact angle of 1 μl bidistilled water on the various surfaces was calculated using a digital image of the drop and measuring the ratio of its height and width. Morphology, grain size and roughness were characterized by AFM measurements performed on a Dimension 3100 AFM controlled by a Nanoscope IVa controller (Veeco Instruments Inc, Mannheim, Germany), using micro-fabricated scanning micro-
1091
scope probes of phosphourous (n) doped silicon (Veeco Instruments Inc, Mannheim, Germany). The cantilever constants are: thickness: 3.5–4.5 μm, length: 115–135 μm, width: 30– 40 μm with resonance frequency of 256–317 kHz and force constant of 20–80 N/m. The AFM was operated in tapping-mode with a vibration isolation system at room temperature in air. For the surface potential measurements Pt/Ir coated Antimony (n) doped Si cantilevers with the following cantilever constants were used: thickness: 2.5–3.5 μm, length: 200–250 μm, width: 23–33 μm with a resonance frequency of 60–100 kHz and a force constant of 1–5 N/m. Topographical data and surface potential were acquired with a two-pass technique. The topography was measured using normal tapping-mode. In the second pass on the same scan line, the AFM tip was lifted 50 nm off the surface and the previously scanned topography is retraced in order to maintain a constant vertical tip-sample distance. An ac-voltage was applied to the tip, with the frequency of the mechanical resonance of the cantilever and constant amplitude of 4 V. Differences in the tipsample potential caused a mechanical cantilever oscillation. A dcvoltage applied to the tip was used to compensate differences in the tip-sample potential thus nulling the mechanical cantilever oscillation. This compensating dc-voltage equalled the surface potential up to a constant. Conductive AFM measurements were performed in contact mode using n+ doped silicon tips with a conductive highly boron-doped diamond coating (resistivity 0.01–0.02 Ohm cm) and the following cantilever constants: thickness: 2.0 ± 1 μm, length: 450 ± 10 μm, width: 50 ± 7.5 μm with a resonance frequency of 6–21 kHz and a force constant of 0.02–0.77 N/m. A measurement bias of + 0.4 V was applied to the tip and the current between the tip and the grounded sample was monitored. XPS was used to quantify the surface oxygen content of the NCD-coatings. The measurements were performed in a Thermo MultiLab 2000 spectrometer equipped with a Mg/Al standard twin anode X-ray source. All spectra were collected using Mg Kα radiation (1253.6 eV). The energy resolution of the
Fig. 2. Optical mapping of the top electrode surface of one NCD FBAR. A small center deflection of 4 μm is caused by a residual compressive stress.
1092
F.R. Kloss et al. / Diamond & Related Materials 17 (2008) 1089–1099
spectrometer, determined by analyzing the full-width at half maximum of an Ag 3d5/2 peak on a sputter-cleaned silver foil, is about 1.00 eV. All XPS measurements were carried out at room temperature in ultra-high vacuum with a base pressure of about 3·10− 10 mbar. The oxygen to carbon content of the film was quantified by integrating over the C1s and O1s peak. The oxygen coverage was calculated from the atomic percentage assuming an average mean free path for the photoelectrons of 5 atomic layers. In the animal experiment 18 female Wistar rats (Charles River GmbH, Denmark) underwent surgical treatment (No. of approval: BMBWK-66.011/0107-BrGT/2004). For anaesthesia Ketasol (Dr. E. Graeub AG, Bern, Switzerland) (100 mg/kg/ body weight) and Xylasol (Dr. E. Graeub AG, Bern, Switzerland) (10 mg/kg/body weight) were used intra muscularly. The abdominal wall was shaved followed by local antiseptic treatment, to perform surgery under sterile conditions. A sagittal incision of the abdominal skin was performed and two pockets prepared into the subdermal soft tissue of each animal, into which the different membranes (groups as described above) were inserted facing the subdermis. A space of at least 1 cm was left between the two pockets to prevent from interaction between the different membranes. The membranes resembling the experimental groups were inserted randomly into the animals. Six animals each were sacrificed and evaluated after 1, 2 and 4 weeks respectively, resulting in 4 membranes per experimental group at each time period. For histological evaluation the membranes and the surrounding soft tissue were biopsied and embedded in Technovit 9100™ (Heraeus/Kulzer, Kulzer Dept., Wehrheim, Germany) for further preparation. From the embedded samples 20 μm sections were prepared using a precision saw and grinding machine (Exakt Gerätebau, Norderstedt, Germany) as described by Donath and Breuner [23]. The sections were stained with haematoxylin-eosin (HE) and van Gieson's staining. The cell number, i.e. the number of nuclear transsection profiles, was counted manually in three sections of each membrane. To do so, the cell nuclei in a distance of up to 10 μm to the experimental surface were taken into account. The percentage of connective tissue attachment to the surface was evaluated as connective tissue/implant contact ratio, analogue to the bone/implant contact ratio [24]. The fibrous tissue and scar formation was analysed histologically and morphologically in the van Giessons's staining. The cell number was given as mean value with standard deviation. The connective tissue/implant contact ratio was given as mean value (in percent) with standard deviation. The significance of the cell number and attachment was evaluated by non Table 1 Material data used for simulations Material
Density ρ [kg/m3]
Acoustic velocity V [m/s]
Acoustic impedance Z = ρV [MRayl.]
AlN (piezoelectric) NCD (support layer) Al (electrodes)
3260
10,400
34
3500
17,700
62
2700
5092
14
Fig. 3. Theoretical dispersion curves of the 3 first modes of resonance as a function of the NCD layer thickness for the structure NCD/AlN including the presence of the Al electrodes. The experimental points at the NCD thickness of 2.3 μm are reported in the graph to indicate the correlation between theoretical calculations and measurement results. The 1st and 3rd modes are closely placed to the theoretical estimation while the 2nd mode is below the expected value. The difference is probably caused by the difference between the assumed and the actual mechanical values of the materials, and the error on their actual thickness.
parametric tests, i.e. exact Wilcoxon rank-sum test, and Kruskal– Wallis H-Test (SPSS software, Chicago, IL, USA). A p-value of b0.05 was considered significant. 3. Results and discussion 3.1. Case a The main challenge in the development of acoustic sensors for the detection of surface adsorbed mass in gaseous and liquid environments is to reach the smallest detection limit possible. Most of the time, this challenge is tackled by a selection of materials and structures for which the acoustic energy is closely trapped at the sensing surface. In this case, a small perturbation due to the adsorption of a given mass causes a dramatic change in the acoustic field distribution in the entire vibrating structure and is detected through a given transduction mechanism. The thin film bulk acoustic resonator (FBAR) is an adequate structure for highly mass sensitive miniature acoustic devices with application to gas and (bio)chemical detection as described in recent theoretical and experimental work corresponding to solidly mounted or membrane supported acoustic resonators [5,7,8,25]. In that latter case, little effort has been devoted to theoretically investigating the influence of the supporting membrane material on the mass sensitivity of a FBAR sensor. We investigated theoretically the impact of introducing nanocrystalline diamond [26] (NCD) as a functional part of the device. The properties of a composite FBAR with NCD are presented. For its definition, a composite FBAR results from the electrical transduction of a longitudinal bulk acoustic wave in a piezoelectric layer mechanically supported by a second layer, referred to as the “loading layer”, and air-backed ends as depicted in Fig. 1. While the loading layer assures the mechanical stability of the structure by its finite thickness, it also influences acoustic and sensing properties of the resonator.
F.R. Kloss et al. / Diamond & Related Materials 17 (2008) 1089–1099
Fig. 4. Theoretical mass sensitivities of the two first modes of the composite AlN/NCD FBAR including the presence of the Al electrodes (perturbation on the top Al electrode side).
Since the composite FBAR presents two possible sensing surfaces, the mass adsorption can occur either on the piezoelectric side or on the loading layer side. In both cases, the properties of a given composite structure are theoretically assessed with the following generic approach: 1. Calculation of the relation of dispersion for the fundamental resonance mode; 2. Expression of the acoustic energy along the entire resonating structure; 3. Evaluation of the mass sensitivity by an analytical expression obtained from the perturbation theory. For practical considerations, a composite FBAR with aluminum nitride – as a widely used piezoelectric material in FBAR – supported by a NCD loading layer (AlN/NCD) was investigated and compared to AlN/AlN and AlN/SiO2. The mass sensitivity S of the acoustic resonator was defined by the relative change of frequency for a perturbing mass of density ρ and thickness Δh. A generalized expression of the mass sensitivity S of acoustic resonators is given by the relative change of frequency Δf caused by a perturbing surface mass change Δm: S ¼ lim
DhY0
Df : fDm
Fig. 5. Theoretical mass sensitivities of the two first modes of the composite AlN/NCD FBAR including the presence of the Al electrodes (perturbation on the NCD side).
should be obtained analytically or with help of computational methods, for instance by implementation of the electro-mechanical Mason's equivalent model of the stacked structure. The material data used for simulations are reported in Table 1. Simulated dispersion curves of the three first acoustic modes versus the NCD thickness are represented in Fig. 3, and experimentally measured frequencies are overlaid in the same figure. Experimental values of the 1st and 3rd modes are in good agreement with theoretical assumptions, while the 2nd mode is below the expected value. The error might be due to the difference between assumed and actual mechanical values of the materials and the error on the actual thickness of the various layers. Due to the absence of symmetry in the composite FBAR, the sensitivity differs according to the side of the membrane that would get perturbed. Figs. 4 and 5 present the evolution of the theoretical sensitivity of the two first modes of resonance for varying thickness of the NCD layer when the perturbation occurs, respectively, on the top electrode or on the diamond side of the resonator. As seen in Fig. 4, a maximum theoretical sensitivity
ð1Þ
By assuming an infinite lateral extension of the acoustic field and no acoustic losses, the perturbation theory allows us to write the sensitivity directly as a function of the particle velocity profile along the thickness (with an initial velocity v0 at the sensing surface x = 0) [27]: S¼
Rt
jv0 j2
ð2Þ 2
2 qð xÞjvð xÞj dx 0
The relation of dispersion, determining the resonance frequency and the velocity profile for various material thicknesses,
1093
Fig. 6. Variation of frequency.
1094
F.R. Kloss et al. / Diamond & Related Materials 17 (2008) 1089–1099
Fig. 7. AFM measurement of the surface roughness. A: the pure titanium membrane, B: the NCD coated membrane. The surface roughness is comparable. Sharp edges were smoothened and valley depths were slightly reduced. The micro-architecture is maintained. C: AFM image of a NCD coated borosilicate sample. The rms roughness of the coating is ~15 nm. D: 1 μm AFM image of the NCD coated borosilicate sample of B. The film consists of round shaped agglomerated particles with diameters ranging from 20 to 50 nm.
around 7000 cm2/g can be achieved by the 2nd mode of resonance, when using the top electrode side as sensing area (with additional advantage of high chemical stability of NCD in harsh environments), and that for a broad thickness band of the NCD comprised between 200 and 1500 nm. The operating frequency of such a device would be around 5 GHz as indicated in Fig. 3, and thus a detectable shift of 1 kHz above the noise level of the sensor would correspond roughly to a minimum detected surface mass of
30 pg/cm2. Above 2 μm, the 1st mode is getting more sensitive than the 2nd mode. The sensitivity of the 1st mode is stable around 1500 cm2/g over a wide range of NCD thicknesses. Such a characteristic makes the device robust to thickness variations usually encountered during thin film deposition processes. The sensitivity of the device can be further increased by reducing the thickness of the piezoelectric layer to a minimum and by optimizing the thickness of the NCD layer accordingly. A different
Fig. 8. A: AFM image of NCD coated quartz sample with an electrochemically generated alternating H- and O-termination. The 3-D-topography is coloured with the surface potential signal of this region. B: Conductive AFM image of the same region on sample depicted in A. The 3-D-topography is coloured with the measured C-AFM current. The oxidized areas are insulating (red) whereas H-terminated regions show a slightly higher surface conductivity. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
F.R. Kloss et al. / Diamond & Related Materials 17 (2008) 1089–1099
1095
Fig. 9. AFM images – 3-D-topography coloured with the corresponding surface potential signals of NCD coated titanium membranes with different oxygen content. A: H-terminated NCD-titanium membrane (10 min exposure in atomic hydrogen at a sample temperature of ~ 800 °C and a pressure of 5·10− 6 mbar). XPS measurements confirm an oxygen content of 2.57 at.%. B: O-terminated NCD titanium membrane (oxidation in an oxygen radiofrequency plasma) with an oxygen content of 18 at.%.
situation prevails when the mass is added to the diamond side of the device, as given in Fig. 5. In that case, the sensitivity is quickly decreased as the NCD layer is thickening. In this situation, materials with lower acoustic impedance should be preferred unless the properties of its surface chemistry are specifically desired. To assess the sensitivity of the device, 580 nm of the diamond has been etched using an O2 plasma. The resonance of the 1st mode before and after etching is given in Fig. 6: an upward frequency shift of 146 MHz is measured, corresponding to a sensitivity of 450 cm2/g. The theoretical median value of this particular example is 436 cm2/g. Although the experimental values are missing to assert that the theoretical predictions are correct, this initial result is promising for future works dedicated to FBAR sensors integrating the nano-crystalline diamond films used in this study. 3.2. Case b The dermal or mucosal seal of transdermal/-mucosal implants is of high importance to avoid adverse soft tissue reactions and thus ensure longevity of the implant. Thereby the roughness of the implant abutment plays a critical role for the tight connective tissue attachment [16,17,28]. While a microroughness of 3 to 30 μm inhibited the epithelial down-growth at an abutment significantly [16,17], the accumulation of subgingival plaque increased with a roughness of more than 0.2 μm [29]. Regarding the findings in the literature mentioned above a roughness between 100 nm and 300 nm seems most favourable for a sealing between connective tissue and implant surface in vivo. Thus, the coating of titanium surfaces with NCD can lead to a controlled fabrication of a nanostructure with a roughness within the rage range of 200 to 300 nm, which thereby might decrease the epithelial down-growth, without an increase of the plaque-accumulation. Therefore the surface roughness of the titanium and NCD coated implants was analysed by AFM. The measurements demonstrate the small change of roughness due to the 1 μm thick NCD
coating. In Fig. 7A the pure titanium surface with an average micro-roughness of 1.1 μm is shown. This micro-topography was essentially retained and is of the order of 0.9 μm after NCD coating (Fig. 7B). Typical grain sizes of diamond crystallites are in the order of 5–15 nm measured with AFM and X-ray diffraction (XRD) [4]. In addition to the micro-structure of the annealed titanium surface, which is basically reproduced by the diamond coating, the NCD film itself exhibits a nanoscale roughness of approximately 15 nm rms, which yields an additional contribution to the surface area. This is demonstrated by measuring the roughness of the NCD coatings on “smooth” substrates like Borosilicate glass or Silicon substrates (2 nm rms roughness). Such films yielded a nano-roughness in the range of 15 nm, as shown in Fig. 7C. In Fig. 7D the 1 μm-AFM image of the NCD coated borosilicate sample of 7C reveals that the film consists of round shaped agglomerated particles with diameters ranging from 20 to 50 nm. Therefore the nano-roughness of the NCD-coated titanium membrane is below the roughness of 0.088 μm which was demonstrated to inhibit accumulation and maturation of plaque formation [30]. Besides the inhibition of plaque accumulation, nano-structuring of surfaces has been proven
Fig. 10. Correlation between the oxygen surface-coverage of NCD coated titanium membranes and the root mean square deviation of the AFM-surface potential signal.
1096
F.R. Kloss et al. / Diamond & Related Materials 17 (2008) 1089–1099
Table 2 Correlation between oxygen coverage for different oxidation processes O Coverage [%]
Oxidation method
99.05 67.67 43.34 16.67 14.14
Plasma Thermal (atmosphere, 400 °C, 4 h) Thermal (vacuum, 650 °C, 1 h) As grown (H-term.) Atomic H-source, 10 min.
to have further benefits. Various cell types reproducibly showed distinct responses to nano-topography, e.g. improved proliferation and differentiation characteristics on nano-phase ceramics and nanometer diameter carbon fibers [29–32]. Fig. 8 shows 3-D-topography plots of a NCD film on quartz that was oxidized electrochemically using conductive-AFM (C-AFM). By applying a bias voltage of +9 V between the AFMtip and the sample surface a localized oxidation in the scanned areas was achieved and after several scans an oxidation pattern with alternating hydrogen and oxygen termination was generated. The colour code in Fig. 8A corresponds to the AFM-surface potential signal whereas in Fig. 8B the same 3-D-topography is coloured with the C-AFM current. The H-terminated regions show a significantly higher conductivity than the oxidized areas (Fig. 8B). The oxygen terminated regions exhibit a slightly more negative surface potential (Fig. 8A). However, one has to bear in mind that the potential contrast arising from the oxidationpattern is modified by a topographical contrast. The latter contrast is generated by differences in the tip-sample capacity between topographical “hills” and “valleys”. The topographical contrast is eliminated by averaging over H-terminated and oxidized areas. In this way an average surface potential difference of around 10 mV was calculated.
Fig. 9 shows 3-D-topography plots coloured with the corresponding AFM-surface potential signals of NCD coated titanium substrates with different surface-oxygen content. The sample in Fig. 9A was freshly H-terminated in a UHV chamber after 10 min exposure to 5·10− 6 mbar of atomic hydrogen at a sample temperature of ~ 800 °C. XPS measurements confirm an oxygen coverage of 14%. Sample 9B was oxidized in an oxygen radiofrequency plasma and reaches an oxygen coverage of 99%. The μm scale roughness of the titanium substrate and the nm scale roughness of the NCD coating itself affect the surface potential contrast in the images rendering determination of absolute surface potential values difficult. With increasing oxygen content, however, the surface potential contrast increases. In order to quantify the surface potential contrast, the root mean square (rms) deviation of the surface potential images was calculated. Fig. 10 shows the correlation of the oxygen surface coverage and the rms surface-potentials variation of NCD coated titanium membranes. The rms surface-potential variation increases almost linearly with the oxygen content. It is known that thermal- and plasma-oxidations, respectively, produce different oxygen species on NCD surfaces [33] (Table 2). Carbonyl- or ether-terminations exhibit a negative partial charge on the surface, whereas hydroxyl groups expose the positively charged hydrogen on the surface. Further investigations have to be performed in order to understand the correlation between oxygen content of NCD surfaces and the potential contrast measured in SP-AFM. In any case, all possible oxygen terminations exhibit a strong dipole moment and the ability of H-bridge formation which is evidenced by the wettability and the resulting contact angle α of the various surfaces. Untreated titanium (α = 70°) and H-terminated NCD (α = 85°–95°) are hydrophobic, whereas Oterminated NCD (α b 10°) is highly hydrophilic (data not shown).
Fig. 11. Schematical illustration of the analysed part of the histological section demonstrates the localisation and orientation of the implanted membranes in the tissue (van Gieson's staining). The region of interest describing the area, where fibrous tissue and scar formation was depicted.
F.R. Kloss et al. / Diamond & Related Materials 17 (2008) 1089–1099
It was demonstrated in vitro that the cell adhesion and migration is influenced by electric fields [18,19,34,35]. One reason is the decreased electrostatic repulsive force between the surface and the adhering cells. Several forces influence the adhesion of cells in a defined surrounding. Experimentally, any type of charged surface has a balance between pure electrostatic repulsive and attractive forces. This balance has been described in the DLVO theory [36]. The theory describes the force between charged surfaces interacting through a liquid medium. It combines the effects of the van der Waals attraction and the electrostatic repulsion due to the so called double-layer of counter-ions. Most of the theories of cell surface interaction are in vitro models in well defined systems. The cell-surface in-
1097
teraction in vivo is more complex due to the existence of a plethora of biological substrates which compete at the surface. In vivo, additional effects like the adhesion-relevant biofilm, consisting of ions and proteins, influencing the promotion of cell adhesion, occur. However, data on the effect of different surface energies on soft tissue healing in vivo are missing. In the light of this fact we evaluated the influence of different surface terminations on connective tissue healing in vivo. Fig. 11 demonstrates a section of the membrane embedded in the subdermal tissue after 4 weeks in the van Giesson staining. The region of interest is highlighted as the area where the fibrous tissue and the scar formation was evaluated histologically and morphologically. Regions adjacent to the subdermal connective
Fig. 12. A: Representative sections of the HE staining of titanium, hydrogen terminated NCD (H-NCD) and oxygen terminated NCD (O-NCD). Cell numbers were counted in a distance of 10 μm adjacent to the membranes. The increased cell number after two and four weeks at the O-NCD is obvious. B: van Gieson's staining demonstrates the collagenous connective tissue formation adjacent to the different substrates. Titanium and hydrogen terminated NCD (H-NCD) show a tight scar formation with a poor number of cells. In contrast the connective tissue formation at the oxygen terminated NCD (O-NCD) is loose, with a large number of cells.
1098
F.R. Kloss et al. / Diamond & Related Materials 17 (2008) 1089–1099
tissue site of the implant as highlighted in the scheme depicted in Fig. 11 are shown for titanium, hydrophobic H-NCD and hydrophilic O-NCD in Fig. 12. In Fig. 12A representative sections of histological preparations in HE staining derived from 1 week, 2 weeks and 4 weeks post operationem are displayed. Numbers of the fibroblasts and reticular cells for all groups at the surface per 500 μm cross section were assessed (Fig. 13A). No significant differences in the number of transsection profiles of cell nuclei for 1 week animals for H-NCD, O-NCD as well as the titanium control group were observed (Fig. 13A). Morphologically all groups showed signs of an inflammatory/foreign body reaction. Macrophages, lymphocytes and fibroblasts were dominating at the substrate surface (Fig. 12A). After two weeks a decrease of inflammatory cells was monitored, whereas the number of fibroblasts increased (Fig. 12A). The cells vicinal to the ONCD surface increased (27.68 cells/500 μm), representing an increase in the number of fibroblasts with no statistical significance, (Fig. 13A). Reticular cells were noticed morphologically after 4 weeks, both at the H-terminated and the O-terminated surface. At the hydrophobic surface a low number of these cells existed, arranged in a monolayer adjacent to the surface, whereas at the hydrophilic surface a multilayer of reticular cells was detected (Fig. 12B). The collagen fibres were oriented parallel to the surface in all groups after 4 weeks. Histologically a dense scar formation with a poor cell number occurred at the H-NCD
surface (Fig. 12B). In contrast the connective tissue contiguous to the hydrophilic surface was less dense and showed a remarkable number of fibroblasts and reticular cells (Fig. 12B). This finding is reflected in the number of cells adjacent to the surfaces. After 4 weeks there was a statistically significant higher number of cells at the hydrophilic O-NCD surface (18.87 cells/ 500 μm; p b 0.002), compared to the uncoated titanium substrate (10.35 cells/500 μm) and the H-terminated surface (11.24 cells/ 500 μm) (Fig. 13A). Investigating the direct contact of the connective tissue of those specimens that were derived from experimental animals 4 weeks after implant insertion, there was a statistically significant higher direct contact with the hydrophilic surface (89.3%; p b 0.001). The titanium (58.15%) and the hydrophobic surface (54.7%) showed a weak attachment (Fig. 13B). The in vivo experimental investigations clearly demonstrate that nano-structured, hydrophilic O-NCD improves the connective tissue response. After an initial inflammatory foreign body response within all groups, a significantly higher cell number and connective tissue/implant contact ratio was observed at O-NCD after 4 weeks. The reticular cells which could be observed at the hydrophobic surface in a monolayer and at the hydrophilic surface in a multilayer are morphologically meant to be fibroblastic reticular cells. Those cells convey the cell adhesion via collagen III. Thus the higher number of reticular cells might be an explanation for the improved connective tissue attachment at the O-NCD. Pure titanium and H-terminated NCD surfaces resulted in densely packed connective tissue with a low number of cells and minor attachment to the surface. Whereas at the O-NCD the collagen fibres were loose and attached tightly to the surface. In all groups the connective tissue fibres were still oriented in parallel direction to the surface. One reason might be that the natural orientation of the fibrous tissue in this animal is parallel to the implanted specimen. According to the present results, the surface roughness in the nanometer range, which is similar for H- as well as O-terminated NCD, seems to be of minor relevance for cell adhesion. Connective tissue response and cell growth are comparable for pure titanium and H-terminated NCD, resulting in a minor attachment to the surface. Furthermore the findings of this work demonstrate that all O-NCD surfaces reveal a strong dipole moment and the ability of H-bridge formation. Depending on the oxygencontaining groups the polarity changes between positive (e.g. hydroxyl-groups) and negative (e.g. carbonyl-groups) values [33]. In any case, the surface is polar, thus facilitating interaction of the O- or OH-terminated NCD implant with polar groups of the body fluid, connective tissue, the cell membranes and the cell attachment molecules. Hence, polarity, wettability and surface energy of O-NCD coated implants seems to be of higher importance for cell adhesion in vivo than surface morphology. 4. Conclusions
Fig. 13. A: Cell number adjacent to the different substrates after 1, 2 and 4 weeks. B: Direct connective tissue attachment to the different substrates after 4 weeks.
We reported the successful integration of nano-crystalline diamond in two different applications. Case a: A composite FBAR opens the way to a MEMS gravimetric sensor. A low detection limit of about 30 pg/cm2 is reached for the 2nd resonance mode
F.R. Kloss et al. / Diamond & Related Materials 17 (2008) 1089–1099
when the sensing event takes place on the piezoelectric side of the NCD-based structure. NCD makes also the characteristics of the structure less sensitive to thickness variations resulting usually from the fabrication process and more attractive for the achievement of real biosensor applications where the mechanical robustness of the membrane structure is of crucial importance. The increased miniaturization of this acoustic sensor with respect to other types of acoustic device is going along with an eventual integration to array technologies. Case b: The influence of different surface terminations of NCD on connective tissue was examined in vivo. It was demonstrated that the polar O-NCD surfaces produce a stronger lateral variation of the surface potential as compared to the H-NCD. The ability of H-bridge formation favours the binding of polar surface adsorbates. Thus O-NCD makes the implant surface attractive for the negatively charged cell membrane increasing the cell number at the subdermal connective tissue site of the implant surface with improved soft tissue/implant contact ratio in vivo. We therefore anticipate that NCD coated implant surfaces exhibit a higher biocompatibility and form an attractive interface promoting sustainable integration. Moreover, NCD provides carries the potential of sealing transmucosal or transdermal implants. This might drive the advancement of various other clinical applications beyond tight attachment of connective tissue to implants. In conclusion, both case studies demonstrate the high potential of commercial use of nano-crystalline diamond films in different fields of applications. Acknowledgements We thank MedEl GmbH, Innsbruck, Austria for the financial support of the present investigation and for providing the titanium substrates and Shasta Pelzer for her technical support in preparing the histological samples. DS and HS greatly acknowledge the RTN-project DRIVE (MRTN-CT-2004-512224) supported by the European Commission. References [1] D.M. Gruen, Ann. Rev. Mater. Sci. 29 (1999) 211. [2] J. Asmussen, D.K. Reinhard, Diamond Films Handbook, Marcel Dekker Inc., Michigan State University, (2002). [3] S. Bhattacharyya, O. Auciello, J. Birrell, J.A. Carlisle, L.A. Curtiss, A.N. Goyette, D.M. Gruen, A.R. Krauss, J. Schlueter, A. Sumant, P. Zapol, Appl. Phys. Lett. 79 (2001) 1441. [4] F.R. Kloss, M. Najam-Ul-Haq, M. Rainer, R. Gassner, G. Lepperdinger, C.W. Huck, G. Bonn, F. Klauser, X. Liu, N. Memmel, E. Bertel, J.A. Garrido, D. Steinmüller-Nethl, J. Nanosci. Nanotechnol. 7 (2007) 1. [5] X. Liu, F. Klauser, N. Memmel, E. Bertel, T. Pichler, M. Knupfer, A. Kromka, D. Steinmuller-Nethl, Diamond Relat. Mater. 16 (2007) 1463.
1099
[6] M. Benetti, D. Cannatà, A. DVAmico, F. Di Pietrantonio, V. Foglietti, E. Verona, IEEE Ultrason. Symp., 2004, p. 1581. [7] R. Gabl, E. Green, M. Shreiter, H.D. Feucht, H. Zeiniger, R. Primig, D. Pitzer, G. Eckstein, W. Wersing, W. Reichl, J. Runck, Proc. of IEEE Sensors 2003, vol. 2, 2003, p. 1184. [8] A. Reinhardt, V. Laude, Th. Pastureaud, S. Ballandras, IEEE Ultrason. Symp., 2002, p. 497. [9] S. Salgar, G. Kim, D.-H. Han, B. Kim, IEEE International Frequency Control Symposium and PDA Exhibition, 2002, p. 40. [10] H. Zhang, E.S. Kim, J. Microelectromechanical Syst. 14 (2005) 699. [11] L.A. Francis. Thin film acoustic waveguides and resonators for gravimetric sensing applications in liquid, PhD thesis, Université catholique de Louvain, (2006). [12] I. Abrahamson, N.U. Zitzmann, T. Berglundh, E. Linder, A. Wennerberg, J. Lindhe, J. Clin. Periodontol. 29 (2002) 448. [13] R. Glauser, P. Schupbach, J. Gottlow, Ch. Hammerle, Clin. Implant Dent. Relat. Res. 7 (Suppl 1) (2005) 44. [14] E. Rompen, O. Domken, M. Degidi, A.E.F. Pontes, A. Piattelli, Clin. Oral Implants Res. 17 (Suppl 2) (2006) 55. [15] B. Cheroudi, T. Gould, D.M. Brunette, J. Biomed. Mater. Res. 26 (1992) 493. [16] B. Cheroudi, T.R.L. Gould, D.M. Brunette, J. Biomed. Mater. Res. 24 (1990) 1203. [17] L. Erskine, C.D. McCaig, J. Cell Sci. 110 (1997) 1957. [18] M. Ohgaki, T. Kiziki, M. Katsura, K. Yamashita, J. Biomed. Mater. Res. 57 (2001) 366. [19] T.K. Monsees, R.H.W. Funk, Biomaterialien 6 (2005) 67. [20] D. Buser, N. Broggini, M. Wieland, R.K. Schenk, A.J. Denzer, D.L. Cochran, B. Hoffmann, A. Lussi, S.G. Steinemann, J. Dent. Res. 83 (2004) 529. [21] M. Wieland, N. Olshanska, L. Scheideler, J. Geis-Gerstorfer, G. Zhao, Z. Schwartz, M.A. Duran, D. Chochran, B.D. Boyan, Biomaterialien 6 (2005) 16. [22] D. Steinmueller-Nethl, F.R. Kloss, M. Najam-Ul-Haq, M. Rainer, K. Larsson, C. Linsmeier, G. Koehler, R. Gassner, C. Fehrer, G. Lepperdinger, C.W. Huck, G. Bonn, Biomaterials 27 (2006) 4547. [23] K. Donath, G.A. Breuner, Int. J. Oral Maxillofac. Implants 10 (1995) 312. [24] K.A. Schlegel, F.R. Kloss, P. Kessler, S. Schultz-Mosgau, E. Nkenke, J. Wiltfang, Int. J. Oral Maxillofac. Implants 18 (2003) 505. [25] J.Y. Pan, P. Lin, F. Maseeh, S.D. Senturia, IEEE Solid-State Sensor and Actuator Workshop, 1990. 4th Technical Digest., 1990, p. 70. [26] J. Philip, P. Hess, T. Feygelson, J.E. Butler, S. Chattopadhyay, K.H. Chen, L.C. Chen, J. Appl. Phys. 93 (2003) 2164. [27] J. Zhang, S. O'Shea, Sens. Actuators, B, Chem. 94 (2003) 65. [28] A. Wennerberg, L. Sennerby, C. Kultje, U. Lekholm, J. Clin. Periodontol. 30 (2003) 88. [29] M. Quirynen, M. Marechal, H.J. Busscher, A.H. Weerkamp, P.L. Dariius, D. van Steenberghe, J. Clin. Periodontol. 17 (1990) 138. [30] L. Rimondini, S. Fare, E. Brambilla, A. Felloni, C. Consonni, F. Brossa, A. Carrassi, J. Periodontol. 68 (1997) 556. [31] T.J. Webster, C. Ergun, R.H. Doremus, R.W. Siegel, R. Bizios, Biomaterials 22 (2001) 1327. [32] A.S.G. Curtis, N. Gadegaard, M.J. Dalby, M.O. Riehle, C.D.W. Wilkinson, G. Aitchison, IEEE Trans. Nanobioscience 3 (2004) 61. [33] A. Krüger, Neue Kohlenstoffmaterialien, Teubner Verlag, 2007. [34] A.M. Rajnicek, K.R. Robinson, C.D. McCaig, Dev. Biol. 203 (1998) 412. [35] K. Larsson, P.-O. Glantz, Acta Odontol. Scand. 39 (1981) 79. [36] P.-O. Glantz, T. Arnebrant, T. Nylander, R.E. Baier, Acta Odontol. Scand. 57 (1999) 238.
