1 Introduction

The amorphous phase of sp3 bonded C atoms is known in literature as “diamond-like carbon” (DLC) [1–3]. DLC is an extremely tough and resistant material and is used in the form of thin films, usually for coating metal parts exposed to high friction [4, 5]. Common methods of DLC films synthesis are plasma enhanced chemical vapour deposition [6–8], filtered pulsed arc discharge [9, 10], magnetron sputtering [2, 11], ion beam deposition [12] and pulsed laser deposition [13, 14].

Besides high wear resistance coatings for metallic parts, DLC also proved useful in coating implants due to some specific surface properties (low surface energy values and chemical inertness) that prevent blood coagulation and favour osteoblasts adhesion [1, 15]. The higher the sp3 content in DLC, the larger the internal stress in thin films [16] causing peeling from substrate. Two main solutions were reported for improving the bonding strength: synthesis of (i) multilayer structures (e.g. Ti–TiN–TiC–DLC) [17, 18] or of (ii) functionally graded materials [19–21]. The head of the femoral prosthesis is made of alloys with high wear resistance (Co–Cr). Inside the human body, the friction with the acetabular cup is accentuated by the presence of liquids that attack the metal and enhance the wear rate. DLC proved biocompatible and can be applied as a protective layer against wear and as a barrier against metal ion release [1, 22].

However, in the biomedical field, the main necessity for DLC coatings comes however from vascular prostheses. In the case of interaction with blood it seems that DLC quality has a major influence upon clotting time. During the blood flow through these tubes, the erythrocytes and thrombocytes (platelets) aggregate in certain spots and may eventually block the blood passage. To compensate for this general weakness of vascular prostheses, DLC films can bind albumin molecules from the sanguine plasma forming a passive layer that makes the surfaces less adhesive for blood platelets [23].

The blood compatibility with carbon-based films is extremely complex and for the moment there is no relation found between haemocompatibility and surface properties such as surface energy, atomic bond structure of carbon or composition of material. Contradictory data have been reported regarding the behaviour of the material in terms of blood clotting, the adherence of platelets or protein adsorption to the surfaces.

The relationship between the sp3 bonds content of DLC and its anti-thrombogenicity properties is still not well understood. Many in vitro [2] and in vivo [13, 24] studies indicate that better results can be obtained for a higher sp3 content. Nevertheless, an opposite trend has also been reported for: (i) tetrahedral amorphous carbon films deposited by pulsed vacuum arc [25], and (ii) hydrogenated amorphous carbon (a-C:H) films deposited by plasma immersion ion implantation, when the blood compatibility tended to improve with the increase of sp2/sp3 ratio [26].

Kwok et al. [27] pointed out that a higher surface energy of phosphorous doped a-C:H films is associated with a low adsorption of proteins, among them the albumin being the preferential one. Similar findings in terms of protein adsorption were presented by Ma et al. [28] who reported a higher albumin to fibrinogen adsorption ratios on surfaces with higher surface energy. On the other hand Hasebe et al. [29, 30] observed that haemocompatibility improves with the decrease of surface energy due to fluorine incorporation in DLC films. Saito et al. [31] also observed that the increase in hydrophobicity increases the anti-thrombogenicity of their fluorinated DLC films. Jones et al. [32] explored platelet attachment on Ti, TiN, TiC and DLC surfaces and reported that the more hydrophilic surfaces present a greater platelet spreading and fibrinogen adsorption. They suggested that the better haemocompatibility of DLC surface is linked to its low surface energy and thus high hydrophobicity. Okpalugo et al. [33] also noted that improved blood compatibility can be obtained when surface energy is lowered in silicon doped a-C:H films.

We report on the fabrication of highly adherent multilayer coatings Ti/TixTiC1−x/DLC using combined radio-frequency magnetron co-sputtering and plasma enhanced chemical vapour deposition, with an intermediate buffer layer having a gradual composition (TixTiC1−x; x = 0–1) synthesized by a new approach [34, 35]. We studied the correlation between activated partial thromboplastin time (aPTT) and surface energy of DLC structures with different sp3/sp2 bonds ratio. Attention was paid to the investigation of protein adsorption and platelets adherence to the surface, both acting as crucial factors for material haemocompatibility.

2 Materials and methods

2.1 Substrate preparation

Mirror-polished titanium (Mateck GmbH) discs (Φ = 15 mm) were used as deposition substrates. The substrates were ultrasonically cleaned in acetone and iso-propanol for 10 min and then dried in argon flow.

2.2 Deposition procedure

The DLC structures were deposited with a UVN-75R1 (1.78 MHz) type system equipped with two RF magnetron cathodes and with one plasma enhanced chemical vapour deposition (PECVD) position. This configuration allows preparing different kinds of structures, like composites, multilayer or graded structures by using both magnetron sputtering (MS) and PECVD deposition methods (Fig. 1). Because of the well-known poor adherence of the DLC films to metallic substrates, we attempted a new architectural coating design: a functionally graded structure (FGS). MS was used to obtain the FGS adherence buffer layer, and PECVD to prepare the top biofunctional DLC film.

Fig. 1
figure 1

Schematic diagram of magnetron co-sputtering deposition set-up

Full size image

The FGS buffer layer was prepared by co-sputtering from two targets, titanium (Ti) and titanium carbide (TiC), in a pure argon atmosphere (0.35 Pa pressure). The Ti and TiC targets were fixed on the vertical and on the horizontal cathode, respectively (Fig. 1). Initially, the substrate holder was placed in a parallel position to the Ti target, at a separation distance of 50 mm. During the deposition process, it was gradually displaced towards the TiC target by a planetary movement. The rotating holder was placed in five different equidistant positions between the Ti and the TiC targets, each for a 3-min deposition time. This way we fabricated a structure with composition gradient (TixTiC1−x) finished with titanium carbide surface. After the 15 min of co-sputtering a 30 nm thick TixTiC1−x functional layer with graded composition was obtained (as determined by X-ray Reflectometry measurements performed on a structure deposited on a glass substrate).

In a second session, the deposited structures were fixed with silver paint onto the PECVD horizontal cathode. The DLC structures were prepared using three different dilutions of methane in argon (20, 60 and 100 %) at a total pressure of 15 Pa. The samples are further denoted as D20, D60, and D100, respectively. During deposition, the DCbias value was kept at ~400 V. When the RF power of 75 W was supplied, the substrate temperature increased due to plasma bombardment, but never exceeding 50 °C, as monitored with a Chromel–Alumel thermocouple system.

For comparison, samples without the TixTiC1−x graded buffer layer were deposited in identical conditions.

Prior to each deposition session the chamber was evacuated down to a base pressure of ≈10−4 Pa.

2.3 Characterization techniques

2.3.1 Morphology and composition

The DLC films thickness was measured by profilometry.

In order to study the vibrations of carbonic matrix, Raman spectroscopy measurements were performed with a Renishaw 1,000 micro Raman system excited with the 532 nm line of an Elforlight laser diode operating at 30 mW. For each sample, a set of 20 measurements of 5 s integration time were performed.

XPS analysis, indicating the C 1s core level variation was used in order to assess the amount of sp2 and sp3 bonded C in the three types of samples (D20, D60, and D100). XPS spectra were recorded in a dedicated chamber (Specs, Germany), under ultra high vacuum (base pressure ~10−10 mbar), using Al Kα = 1486.71 eV monochromatized radiation. The electrons were collected using a hemispherical electron energy analyzer (Phoibos 150) operated at a 30 eV pass energy. Resolution (in terms of full width at half maximum) of 0.45 eV was achieved. During measurements, a flood gun operating at 1 eV acceleration energy and 100 μA electron current was used in order to ensure sample neutralization. The spectra were recorded both prior and after sputter cleaning for 10 min using an ion current of 10 μA in order to analyse the ion-beam induced changes in the surface composition and bonding. During sputtering, the pressure inside the XPS analysis chamber was kept at 1 × 10−7 mbar. The experimental data were fitted with Voigt profiles, allowing different values for the background. C 1s line (E = 284.5 eV) was used as reference for the charge shift correction.

The crystallographic structure of each film was investigated by X-ray diffraction (XRD), using a Bruker D8 Advance diffractometer, in parallel beam setting, equipped with copper anode X-ray tube. The measurements were carried out in grazing incidence geometry at 1° in the 2θ range from 20° to 65°, with a step size of 0.04° and 90 s acquisition time per step.

2.3.2 Pull-out bonding strength

The pull-out measurements were carried out using a DFD Instruments PAT MICRO AT101 (maximum pull force=1 kN) adhesion tester equipped with 2.8 mm diameter stainless steel test elements. The test elements were glued to the film surface with E1100S cyanoacrylate one-component epoxy adhesive. The stub surface was first polished, ultrasonically degreased in acetone and ethanol and dried in a nitrogen flow. After gluing, the samples were placed in an oven for thermal curing (130 °C/1 h). Each test element was pulled-out vertically with a calibrated hydraulic pump until detachment. The experimental procedure was conducted in accordance with the ASTM D4541-09e1 standard. The mechanical adhesion strength was determined from the recorded failure value divided by the quantified detached surface area. Five samples were tested independently for each type of deposition condition. Mean value and standard deviation were computed.

2.3.3 Surface energy

We performed solid surface energy calculations based on contact angle measurements, using deionized water and formamide as standard solvents. The measurements of the prepared DLC structures were carried out using a goniometric method, the two solvents being dropped onto the surface and the contact angle estimated. The drop size and the drip distance were kept constant in all cases. The contact angle values were determined by the evaluation of the tangent angle of a sessile liquid drop on the DLC solid surface. The surface energy was calculated using the Owens–Wendt approximation [36, 37]. Five experiments were performed for each DLC surface. Mean value and standard deviation were computed.

2.3.4 Biological assays

2.3.4.1 Important note

In order to eliminate the potential biological variations, blood was prelevated from 5 healthy non-smoking patients. They signed an informed consent allowing us to use their blood and derivatives of their blood in our studies in accordance with the ethical rulings of the Helsinki declaration (6th revision, World Medical Association 59th Meeting, Seoul 2008).

2.3.4.2 Surface preparation

Surfaces were cleaned in ultrasonic bath with 80 % ethanol. Then, surfaces involved in cell testing were sterilized by autoclavation (for 30 min, at 130 °C and 0.1 MPa) in individual sealed sterilization bags. All procedures were carried out in a sterile environment and all biohazard safety measures were used.

2.3.4.3 Whole blood, plasma and platelets solutions

We have sampled blood in standard test-tubes with or without anticoagulant substances (sodium citrate or ethylenediaminetetraacetic (EDTA)).

2.3.4.4 Partial thromboplastin time test

In order to asses interaction with blood coagulation systems, we conducted a slightly modified protocol for manual detection of aPTT as introduced by Margolis [38]: fresh blood was harvested in sodium citrate (3.2 % final concentration) using Vacutainer medical devices from five healthy non-smoking patients.

Prior to the test, CaCl2 was added to plasma to counter the effects of sodium citrate and enable clotting. The samples were placed on a plane surface and drops of plasma (50 µl) were deposited from a distance of 1 cm. The aPTT was considered as the time for plasma clotting on surface [38, 39].

For comparison of DLC coatings characteristics with commercial biomaterials, the same analysis was performed on medical grade poly(methyl-methacrylate) (PMMA) and Ti plates. Five experiments were performed for each type of surface. Mean value and standard deviation were computed.

2.3.4.5 Platelet adhesion quantification

Healthy patients’ blood was collected in Vacutainer tubes containing EDTA as anticoagulant. Platelets were obtained by centrifugation of whole blood. First, tubes were centrifuged at 250×g for 5 min, at room temperature in order to separate red and white blood cells. Then, the platelet-rich plasma was removed and centrifuged in a second tube at 800×g in order to obtain platelets. The plasma was removed and the platelets were washed three times with cell culture medium M199 and centrifuged at 800×g as before. Platelets were resuspended with medium M199 at a density of 1 × 108 platelets/ml.

300 μl medium M199 containing platelets was deposited in a uniform layer onto samples (Ti, D20, D60, D100). They were put in sterile Petri dishes and transferred in a humidified atmosphere incubator at 37 °C with 5 % CO2 saturation. After 2 h the discs were washed twice with phosphate buffer saline and overlaid with 100 μl lysis buffer (50 mM Tris HCl, 0.15 mM NaCl, 1 % Triton X-100, 0.5 % Na Deoxicolate, 0.1 % SDS, 2 mM EDTA, 10 μg/ml aprotinin, 10 μg/ml pepstatin, 1 mM PMSF). The platelets lysate was centrifuged at 14,000×g for 15 min, and supernatants were removed and stored at 80 °C.

2.3.4.6 Western blot analysis

Samples were denaturated by boiling for 10 min in 2 % DDT. Samples of 10 μl were subjected to SDS-PAGE with 12 % acrylamide-bisacrylamide gels. The transfer was carried out in a wet cell, on polyvinylidene fluoride (PVDF) membranes in a Mini Protean system (Bio-Rad, USA). The membranes were blocked overnight in a 5 % dry milk solution in TBST (50 mM Tris/HCl, 150 mM NaCl, 0.1 % Tween-20). Incubation with anti-beta actin antibody produced in rabbit (Santa Cruz, USA) diluted 1:200 on TBS with 2 % BSA, was performed at room temperature (RT), for 1 h. After three times washing, the membranes were incubated at RT for 1 h with corresponding horseradish peroxidase (HRP) coupled secondary antibody (goat anti rabbit and goat anti mouse from Santa Cruz) at a dilution of 1:20,000. Chemiluminiscence reagents (enhanced chemiluminescence kit, Pierce, USA) were added to membranes for 2 min and results were acquired on BioMax films (Kodak, USA) developed automatically according to the manufacture’s indication. Membranes were stripped in 5 % beta-mercaptoethanol for 1 h at 45 °C and probed with anti-aprotinin mouse antibody (Thermo Scientific, USA) for loading control. Film analysis was conducted with Gel Scan 7,000 and Quantity One software (both from Bio-Rad, USA).

2.3.4.7 Adsorbed protein mass spectroscopy

Blood from healthy patients was collected in Vacutainer test-tubes containing EDTA, and centrifuged at 1,000×g for 10 min in order to obtain fresh plasma. Plasma from five patients was pooled and 100 μl of plasma was put in contact with each type of material: Ti, D20, D60, and D100. After 1 h the discs were washed three times with ice cold double distilled water and overlaid with 100 μl analysis buffer (7 M ureea, 2 M thioureea, 2 % CHAPS, 2 % beta mercaptoethanol, 10 μg/ml pepstatin, 1 mM PMSF). Samples were sonicated and 16 μl of each were mixed in a test tube with 4 μl energy absorbing molecule buffer (5 mg sinapinic acid dissolved in 400 μl solvent composed of: 50 % acetonitrile, 0.5 % trifluoroacetic acid, 49.5 % H2O HPLC grade). 10 μl of each sample were loaded onto spots of ProteinChip Gold Array and allowed to dry in the laminar flow hood used for cell cultures. After drying a quantity of 1 μl energy absorbing molecule buffer was added to each spot and allowed to dry under the same conditions.

Samples were analysed with a ProteinChip Surface Enhanced Laser Desorption/Ionization-Time of Flight mass spectroscopy (SELDI-ToF) platform (Bio-Rad, USA) and ProteinChip DATA Manager software. All materials used in this experiment were purchased from Bio-Rad, USA. Parameters were: mass range 6,000–2,00,000; focus mass: 25,000; matrix attenuation: 6,000; sampling rate 400; calibration method: PCS4000 instrument default; acquisition method: SELDI Quantization; warming: energy 4,400 (1 shot); data: energy 4,000 (10 shots).

2.3.4.8 Statistical analysis

For the pull-out, surface energy and aPTT measurements the statistical significance was determined using an unpaired Student’s t test. The differences were considered significant when p < 0.05.

3 Results

3.1 Film thickness

Thickness measurements by profilometry revealed a linear dependence of the DLC deposition rates by PECVD on the methane dilution in the working atmosphere: 5 nm/min for D20, 9.1 nm/min for D60, and 15 nm/min for D100. For conformity of experiments, all the top biofunctional DLC layers were prepared at a 300 nm thickness.

3.2 Raman measurements

Raman spectroscopy was used to characterize the quality of the DLC films, obtained at different methane dilution, in terms of diamond carbon-phase purity (Fig. 2). This technique proved very useful in correlating the electronic properties of carbonaceous materials with their vibrational signature, given the high sensitivity of vibrational modes (located in the 1,200–1,600 cm−1 spectral range) to modifications in the electronic configurations of C atoms.

Fig. 2
figure 2

Raman spectra of DLC films deposited in: a 20 % CH4 + 80 % Ar atmosphere b 60 % CH4 + 40 % Ar atmosphere, and c 100 % CH4 atmosphere

Full size image

The Raman spectra were deconvoluted with three components, corresponding to: nanocrystalline carbon contribution, D band and G band respectively (Fig. 2). The G-peak corresponds to the fundamental E2g mode of graphite, and D-peak to disordered graphite [40, 41].

The peak between 1,100 and 1,300 cm−1, as in our case, is often attributed to nanocrystalline or amorphous diamond [42–45]. Accordingly, one can assume the presence of nanocrystalline diamond clusters with sp3 -bonding embedded in the non-crystalline matrix of our carbonaceous films. In the Raman spectrum of D100 sample, an enhancement of the band associated to nanocrystalline form of C may be observed, which suggests a crystalline C form appearance.

The position of the G peak and the ratio of the integrated areas under the G and D peaks (IG/ID) (Table 1) is correlated with the sp2/sp3 bonding ratio [40, 46, 47]. The decrease of the IG/ID ratio, the shifting of the G peak towards higher wave numbers and the broadening of the D-peak are widely accepted as indicators for the increase of the graphite-like phase concentration in the carbon films [46, 48, 49].

Table 1 Raman deconvolution data for the DLC films (relative bands’ intensities, FWHM, IG/ID ratios)
Full size table

We can observe that as the methane dilution is increased, the IG/D ratio increases too (Fig. 2 and Table 1). According to Zhang et al. [46] and Ferrari et al. [40, 47] a higher IG/ID ratio corresponds to a higher percent of sp3–C bonds. This variation is accompanied by a slight continuous shift of the G band from ~1,529 cm−1 (D20) to ~1,522 cm−1 (D100). In the Raman spectra of carbon-based materials the G-peak positions move to higher wave numbers due to two processes: the increase of the sp2 content and/or a higher compressive stress [50]. Additionally, the full-width at half-maximum (FWHM) of D-peak also showed a progressive decrease with the increase of methane concentration (Table 1), in good correlation with the aforementioned IG/ID ratio and G band position trends. Although the main indicative furnished by Raman spectroscopy concerning the sp3-C amount is the IG/ID ratio, the widening of the D-peak, which is correlated with a distribution of carbonaceous clusters having different dimensions [47], may as well serve as a structural “gauge”.

3.3 XPS characterization

From the XPS analysis we could extract the amount of sp3 -bonded C and sp2 -C as the ratio between the integral intensities of each component. The XPS spectra exhibited a very complex shape indicating the existence of different chemical states for C 1s (Fig. 3). The experimental data were fitted with Voigt profiles, allowing different values for the integral background instead of using just one integrated spectrum [51]. Three components were needed in order to assure a good fit, associated with the sp3 -C (~286 eV) and sp2 -C (~284.3 eV) contributions, as well as to C–O, C=O and/or O–C=O bonds (287.5–289.9 eV) owing most probably to the contamination of the sample surface [52–54]. The results of the fitting are presented in Table 2.

Fig. 3
figure 3

High resolution XPS spectra for C 1s core level photoelectron after sputter cleaning: samples a D20; b D60; and c D100

Full size image
Table 2 XPS peak separation data for the C 1s line of DLC films
Full size table

The deconvolution studies of the C 1s spectra generally reveal two main distinct peaks assignable to sp2 - and sp3 -C hybridizations [54]. The peak situated at a higher binding energy (BE) is assigned to sp3 -bonded carbon (C–C and C–H), and that at lower BE corresponds to the sp2 hybridization state of carbon.

From the analysis of the main components of C 1s core level spectra we can conclude that the amount of sp2 bonded C decreases from ~36 % in D20 sample to about 10 % in D100 sample.

We note that the result for D20 sample is in agreement with the Raman measurements which suggested that a higher content of sp2 is present in these films. When the methane dilution is increased (D60 and D100) the sp3 -C concentration strongly increases (to around 78 and 87 %, respectively).

3.4 XRD characterization

The XRD patterns of the DLC samples are presented in Fig. 4. The diffraction peak centred on 2θ = 44.2° (d = 2.04 Å) which appears in the patterns of samples D60 and D100 can be assigned to the most intense and unique peak of diamond (face-centred cubic, ICDD # 01-75-410) accessible in grazing incidence with the current experimental set-up (2θ < 65°). This is the only diffraction peak that does not belong to titanium in D60. In the case of D100 there are several other weak maxima, which fit satisfactorily to two quite unusual (obtained at high temperature and high pressure) carbon-based phases (ICDD # 46-943 and ICDD # 46-944), reported by Shterenberg and Bogdanova [55].

Fig. 4
figure 4

GIXRD patterns of the Ti/TixTiC1−x/DLC structures [filled circle–Diamond (ICDD # 01-75-410); filled triangle–Carbon (ICDD # 00-46-943); empty triangle–Carbon (ICDD # 00-46-944)]

Full size image

3.5 Pull-out test adherence

The adherence of the coating to the metallic substrate is a very important parameter for haemocompatible applications. A value of ~40 MPa is the minimum required for this kind of applications (ISO 13485:2003).

A strong enough epoxy adhesive (E1100S) was used to evaluate the suitability of the DLC coating adherence. Prior to the DLC coating adherence evaluation, the glue bonding strength was tested independently on bare titanium plates. Each time, the failure occurred in the glue volume at 80–85 MPa, confirming its technical specification.

The simple DLC/Ti structures presented unsatisfactory adherence values, well below the threshold of 40 MPa, being easily detached by a common scotch-test (D20 and D60) or even by simple scratching (D100). On the other hand, the prepared FGS coatings demonstrated promising values of adherence (Fig. 5), even higher than those required by mandatory international standards for the load-bearing orthopaedic implants (ISO 13779–2:2008). The lowest value of bonding strength (58 ± 4.2 MPa) was recorded for D100 structures. The mean adherence value increased slightly for D60 (60 ± 3.5 MPa), and more for D20 (66.7 ± 2.2 MPa) (see Fig. 5). The two tailed t-testing with unequal variances showed no statistically significant difference (p > 0.05) between the recorded adherence values of D60 and D100, whilst the difference was statistically significant (p < 0.05) between D20 versus D60 and D100.

Fig. 5
figure 5

Histogram depicting the variation of the mean pull-out adherence values of the DLC films

Full size image

The insertion of the co-sputtered TixTiC1−x transition stratum aims to reduce the interface discontinuity between the two different materials (Ti and DLC), which improves significantly the adhesion strength. The functionally graded interlayer acts as an anchor, contributing to the adaptation at the interface area. Moreover, a carbide surface can react with the depositing gaseous radicals, promoting a “chemical bonding” of the nucleating DLC layer, that can further enhance the film adhesion.

3.6 Surface energy

The surface energy values recorded for DLC/Ti structures were lower than those of the bare medical grade Ti and PMMA control substrates (Table 3). One can also notice a decrease of the surface energy with the increase of methane dilution in the reactor chamber (Table 3). The two tailed t-testing showed statistically significant differences (p < 0.05) between the surface energy values recorded for all the samples. An important decrease (with ~25 %) of the surface energy was obtained when applying the DLC coating: from 37.85 ± 0.94 mJ/m2 for the bare Ti substrate down to 28.7 ± 0.34 mJ/m2 for the D100 structure.

Table 3 Surface energy values recorded for the DLC films, and for the Ti and PMMA controls
Full size table

3.7 Biological assay

In order to assess the haematological biocompatibility, the coagulation time was measured, followed by more complex tests: platelet adherence and protein adsorption to the surface.

3.7.1 apTT tests

Generally, higher partial thromboplastin times were recorded on DLC coated surfaces with respect to two commercial control surfaces (Ti and PMMA) (see Table 4). The two tailed t-testing, assuming unequal variances, showed statistically significant differences (p < 0.05) between the aPTT values. A remarkably higher aPTT value was obtained in the case of D100 coating, which will be discussed later in conjunction with the other physical and biological results.

Table 4 aPTT values measured for the DLC films, and for the Ti and PMMA controls
Full size table

3.7.2 Platelet adherence

The amount of beta-actin, a structural protein present in all cells, was investigated by Western blot method. The obtained signal is proportional to the quantity of beta actin and, therefore, to the number of platelets adhered on the sample surface at the moment of lysis. The PVDF membranes were also probed with aprotinin (a protease inhibitor with proteic structure and mass of ~6 kDa) which was present in the same concentration in all samples, since it was added to the lysis buffer formulation. This is an internal quality control which ensures that all steps of the technique are properly done. As visible from Fig. 6, there was almost the same number of platelets present on the surface of bare titanium and D20 samples. The number of platelets adhered on D60 and D100 was significantly lower.

Fig. 6
figure 6

a Western blot analysis of beta-actin and aprotinin present in platelets adhered on DLC and bare Ti samples; b Optical density histograms of normalized quantity of beta-actin present in platelets adhered on DLC and bare Ti samples

Full size image

3.7.3 Protein adsorption

To study the adsorption pattern of proteins to the samples’ surface a state-of-the-art technique was used: Surface Enhanced Laser Desorption/Ionization-Time of Flight mass spectroscopy (SELDI-ToF).

As one can observe (Fig. 7a), serum albumin was adsorbed in greater quantities on all DLC surfaces than on the bare titanium surface [Albumin molecular weight (MW) ≈ 66483 Da]. Other proteins (G immunoglobulins) presented a roughly similar pattern (G immunoglobulins MW ≈ 134350 Da). Another important peak is that of 28,900 Da which can be assigned to the factor XIIa light chain and is more prominent on the titanium sample (Fig. 7b). The factor XII, the activator of surface contact coagulation cascade, could not be identified because it had a mass similar to that of albumin (Factor XII MW ≈ 67792 Da).

Fig. 7
figure 7

a SELDI-ToF complete spectra of proteins adsorbed on DLC and bare Ti samples from fresh blood plasma; b SELDI-ToF detailed spectra in MW range 8,000–30,000 Da

Full size image

4 Discussion

The sustainability of a biocompatible film, to be applied as biofunctional coating layer of implantological elements used to remedy various cardiologic, orthopaedic or dental medical conditions, is essentially conditioned by the adherence to substrate.

Unfortunately the high thermal expansion mismatch between DLC and Ti, responsible for the high tensile stress in the film, causes a relatively poor film adhesion (as shown in Sect. 3.5 and Refs [17, 56]). Moreover, a higher sp3 content in the DLC films, which is believed to play an important role on the blood biocompatibility, also induced greater internal stress and reduction of film adhesion [16]. The deposition of the DLC film on a rather thick inter-layer buffer (e.g., titanium carbide or titanium nitride) and not directly on the metallic surface resulted in better adherence results [17, 18, 57–59]. However, when using TiN as an intermediary buffer layer a significant haemolytic effect was observed after long assessment times [17, 18].

In this study we propose an alternative solution which resorts to the introduction of a graded transition from the titanium support to the DLC top biofunctional film by co-sputtering from two targets: Ti and TiC. The 30 nm thick TixTiC1−x (x = 0–1) buffer layer significantly improved the top DLC film adherence. The slightly lower adherence of the D100 samples, compared to D60 and D20, could be related to its higher sp3 bond content, which can induce a higher internal stress. It is worth noting, that all the recorded pull-out adhesion values are superior to the value imposed by the international standards for biocompatible load-bearing coatings to withstand dynamical mechanical loads (ISO 13779–2:2008; ASTM F 1185-03: 2009).

The increase in the sp3 content with the increase of the methane concentration in the working atmosphere has been confirmed both by Raman and XPS. We measured a significant sp3 content augmentation from sample D20 to D100, which could be the effect of the initial sp3-hybridization of carbon in the methane molecule. Bugaev et al. [60] also reported that high quality DLC films can be obtained from pure methane, their results pointing that most probably methyl mechanism is favourable for diamond-like bonds formation. It is known that CH3 are the most abundant species in pure methane discharges, while carbon dimer C2 is the most abundant in methane highly diluted in argon discharges [61, 62]. It is not surprising, neither unprecedented [62] that at low temperature the formation of diamond-like bonds from a reactive environment having as majority species C2 dimers is more difficult.

In order to better predict the behaviour of materials in vivo we designed a range of in vitro tests to investigate the events that happen on the surface when put in contact with blood. We first analysed the aPTT, which is generally used to detect the degree of activation of the endogenic clotting system, to measure the surface dependent blood clotting. Next, we dissociated the elements involved in coagulation—plasma and platelets—and separately investigated their interaction with surfaces.

As visible from Table 3, the surface energy is decreasing with the methane concentration in the deposition atmosphere. In DLC, the graphene regions of sp2 carbons are heavily conjugated and partially charged boundaries appear, which could promote adsorption of charged blood proteins such as coagulation factors, kininogens, and various globulins [63]. We hypothesize that the more boundaries of graphene regions or the more distorted graphene regions exist, the more partial charges will appear, the faster the charged proteins will be adsorbed and the more quickly the blood/plasma will coagulate.

One can notice that for DLC functionalized structures the tendency of the partial thromboplastin time is to increase with the decreasing of surface energy values (Tables 3 and 4). The maximum thromboplastin time was recorded for the DLC structures prepared in pure methane atmosphere (D100). The clotting process can usually be initiated by surface adsorption and contact activation of coagulation factors of the endogenic clotting system [64]. Unlike bare Ti, the contact with the DLC coated surface suppresses the activation of the endogenic clotting system.

The adherence of platelets to the materials plays a key role in platelet activation and subsequently in blood clotting activation. Platelet adherence was studied by Western blot analysis of beta actin, a structural protein present in all cells. In our opinion this technique allows for a more correct determination of the variations of platelet adherence than optical approaches often used in literature. However, there are some groups that have attempted to better simulate the natural conditions [65, 66], in platelet adherence studies and have assembled an enclosed system which ensures a laminar flow through parallel plates of material. This system is not infallible, since this way the blood is subjected to an important shear stress, due to circulation via a pump through tubings which can activate platelets, along with other surfaces interactions thus conducting to additional errors. As seen in the platelet adhesion experiment (Fig. 6), the DLC coatings create conditions for a weaker platelet-surface interaction, which in vivo can conduct to a lower platelet activation and subsequently a prolonged time of coagulation (see Fig. 8). One can assert that this effect derives from the fact that all cells have a negatively charged cellular membrane which tends to interact/adhere to positively charged surfaces (hydrophilic surfaces) rather than to hydrophobic ones.

Fig. 8
figure 8

The DLC coatings and Ti bare control aPTT values plotted against the normalized beta-actin values

Full size image

In this work, the interaction of blood with the surface was investigated by SELDI-ToF mass spectroscopy which generates a profile of protein adsorption at interface with materials, as seen in Fig. 7.

The other current approaches in literature put the resolubilised powder of bovine serum albumin or fibrinogen in Phosphate Buffer Saline, in contact with a material and determine the adsorption. This is a rather simplistic way to study protein adsorption behaviour because these determinations are focused on one protein at once, making it difficult to predict these phenomena in their complexity. Our approach is in pace with nowadays proteomics, since it uses fresh blood plasma from healthy patients which is a complex mixture of proteins, peptides, aminoacids and many organic and inorganic substances.

Albumin, with a molecular weight of ~66.5 kDa represents roughly 50–60 % of plasma proteins and is not glycosylated like other proteins in blood. The G immunoglobulins (~134.5 kDa) account for roughly 15 % of plasma proteins. Fibrinogen is found at concentrations in range of 150–400 mg/dl and other coagulation factors are enzymes present in very small quantities. Albumin is a protein which has hydrophobic moieties, being a blood carrier for many hydrophobic molecules. Since our DLC surfaces tend to be more hydrophobic it is expected to find more adsorbed albumin than on titanium as confirmed by mass spectroscopy spectra. The vast majority of proteins in blood are glycosylated which makes them more hydrophilic and more susceptible to polar interactions. The quantity of albumin adsorbed on the surface shields the surface of the sample, making it difficult for the different proteins and coagulation factors to reach the sample and activate the coagulation cascade (Fig. 7b). These findings are in line with Liu et al. [67], who showed that the albumin adsorption on DLC inactivates the surface for blood clotting.

We can state that the coagulation time for each material is in line with surface energy data, with the platelet–surface adherence properties and protein adsorption profiles, and so advocates for a cause-effect relationship between these factors.

5 Conclusions

Adherent and haemocompatible DLC films have been synthesized onto titanium substrates using combined radio-frequency MS and PECVD methods. The effects of methane concentration in the reactor chamber on the structural properties of the DLC films and on the biological response of the samples can be concluded as follows:

  1. 1.

    The introduction of a graded transition from the titanium support to the DLC top biofunctional film by co-sputtering from two targets (Ti and TiC) through a TixTiC1−x (x = 0–1) buffer layer resulted in a significant improvement of the DLC films’ adherence. The adherence values are similar to or even higher than those imposed by international standards and are therefore encouraging for implantology applications.

  2. 2.

    The sp3 content increased with the methane concentration in the deposition atmosphere.

  3. 3.

    The relationship between the sp3 content and anti-thrombogenicity of DLC films was confirmed by the partial thromboplastin time. The haemocompatibility is strongly dependent on the methane concentration: a higher concentration of methane reduces the platelet activation and thrombus formation;

  4. 4.

    We studied protein adsorption using fresh blood plasma from healthy patients. In the case of DLC films with the highest sp3 content albumin was preferentially adsorbed (due to the affinity between surface and the hydrophobic moieties of this protein), thus shielding the surface and preventing the immobilization of coagulation factors.

Next, wear/abrasion measurements together with antimicrobial tests are projected in order to fully asses this coating design feasibility for implantology development.