Effect of surface treatment on the bioactivity of nickel–titanium

Available online at www.sciencedirect.com

Acta Biomaterialia 4 (2008) 1969–1984 www.elsevier.com/locate/actabiomat

Effect of surface treatment on the bioactivity of nickel–titanium Wojciech Chrzanowski a,c, Ensanya Ali Abou Neel a, David Andrew Armitage b, Jonathan Campbell Knowles a,* a

Division of Biomaterials and Tissue Engineering, UCL Eastman Dental Institute, 256 Gray’s Inn Road, London WC1X 8LD, UK b De Montfort University, Leicester School of Pharmacy, The Gateway, Leicester LE1 9BH, UK c The Silesian University of Technology, Faculty of Mechanical Engineering, ul. Konarskiego 18a, 44-100 Gliwice, Poland Received 16 January 2008; received in revised form 23 April 2008; accepted 9 May 2008 Available online 9 June 2008

Abstract In this paper, the bioactive properties of Ni–Ti alloy after different surface treatments were evaluated in different media (Hanks’ balanced salt solution, Dulbecco’s modified Eagle’s medium and osteogenic). Evaluation was performed on the basis of X-ray photoelectron spectroscopy and atomic force microscopy studies after immersing samples for up to 24 h in the relevant media. This allowed assessment of the kinetics of Ca2+ and P5+ precipitation and early interaction of the media with surfaces. In addition, the surface free energy was measured and the influence of heat treatment on phase transformation temperatures and rate of nickel and titanium ion release was investigated. The most favourable bioactive properties were observed for simply ground Ni–Ti samples when evaluated in HBSS, which showed similar properties to reference positive samples (BioactiveTi). On the other hand, samples heat-treated at 600 °C showed very low levels of precipitation of Ca and P. Most interestingly, evaluation in the media containing organic components (protein, vitamins, antibiotics and drugs) revealed that bioactivity for all the samples was at the same level (except for the reference negative) irrespective of the surface preparation method. It demonstrated that organic components interact with the surface rapidly, forming a thin protein layer, and this altered the surface properties of the samples, making them bioactive. No significant difference in kinetics of the Ca2+ and P5+ precipitation were observed. Nevertheless, further ion release and chemical composition evaluation revealed that alkali treatment and spark oxidation cannot be considered as a useful for biomedical application due to very high levels of Ni in the top layer (alkali-treated) and high rate of Ni release (spark-oxidized and alkali-treated). Ó 2008 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. Keywords: Bioactivity; Nitinol; Nickel–titanium alloy; Heat treatment; Spark oxidation

1. Introduction Titanium and its alloys have been important materials in the production of implants for hard tissues for several years. This is as a result of their mechanical properties (which are more favourable than those of stainless steel), very good tissue response and enhanced osseointegration. It is also possible to make the implant inert, which was demonstrated to be important for retrieving the implant and reducing the adherence of soft tissues (e.g. tendon) that *

Corresponding author. Tel.: +44 (0)207 915 1189; fax: +44 (0)207 915 1227. E-mail address: [email protected] (J.C. Knowles).

could alter their functions [1]. However, there is a considerable difference between the elastic properties of the metal compared to bone, and this difference plays a very important role during fracture healing for many types of fixations. It must be highlighted that it is not the mechanical properties of the material that are the most important but rather the properties of the entire construct which bears the load. Ideally, the fixator should replicate the natural rigidity of intact bone at the interface. This is because metal implants have a completely different shape and structure than tissue so its ‘‘shape” rigidity must be fitted to the tissue. Nevertheless, among the titanium alloys, one has exceptional mechanical properties including superelastic properties and shape memory effects; this is the nominally

1742-7061/$ - see front matter Ó 2008 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.actbio.2008.05.010

1970

W. Chrzanowski et al. / Acta Biomaterialia 4 (2008) 1969–1984

equiatomic Ni–Ti alloy (Nitinol), and these properties may improve the performance of some devices. The current fairly low popularity of Ni–Ti is caused by high material cost and very high content of Ni, which is known to be carcinogenic [2]. However, it has been demonstrated that its mechanical properties make this material especially attractive for bone fixators (bone clamps, plates, wires and pins), dental and laparoscopic devices. The biological properties of Nitinol can be improved by different surface treatments; it has also been shown that the negative effects of Ni can be eliminated by the same types of treatments [2–9]. For Nitinol, treatments such as oxidation, nitriding, different types of coating (e.g. carbon, hydroxyapatite and polymers), chemical etching, polishing and electrochemical methods [4,8,10–19] have been proposed. These treatments have been successfully used for commercially pure titanium (cpTi) and its alloys (Ti6Al4V, Ti15Zr4Ta4Nb, Ti6Al7Nb, Ti13Zr13Nb). Each of these methods can lead to different degrees of bioactivation or conversely passivation. The bioactive properties of materials can be interpreted in different ways. In general, the term bioactivity is related to the ability of the material to trigger a biological action, and the authors linked this term to cell response or more commonly to the formation of an apatite layer from simulated body fluid (SBF) [10,14,20–22]. To date, in the literature, there is little agreement as to how to evaluate bioactivity, and many authors follow Kokubo’s approach, immersing samples in SBF for 14 days and measuring the chemistry of the formed film using energy-dispersive Xray analysis and X-ray diffraction [22,23]. However, this evaluation does not involve consideration of immediate interactions of the surface with the fluids that were found to be the most important for cell response, i.e. for cells to ‘‘recognise” the surface, attach to it and proliferate. Most of the cell studies have been carried out over time periods ranging between a few hours and several days [24–26]. For the same reason evaluation of bioactivity should be performed over a similar period. Bioactivity evaluation after immersion for 14 days does not yield detailed information about the kinetics of apatite film formation. Nevertheless, most of the studies used zeta potential measurements, ion release or ion depletion from the solution as a means to characterize the bioactivity; these parameters enable better analysis of the bioactive behaviour of the material. Bioactivity can be altered by several different methods that change the topography, chemistry, wetting ability and surface energy. All these features were found separately and in combination to have an influence on Ca and P precipitation [4,15,27–30]. In this paper, the influence of surface treatments, heat, alkali treatment, and spark oxidation on the bioactive properties of Nitinol were evaluated. The results were compared with reference samples that were reported to be bioactive and passive in SBF. A new approach to evaluate bioactivity in medium containing not only inorganic but

also organic components such as peptides, antibiotics, proteins and vitamins on the basis of X-ray photoelectron spectroscopy (XPS) examination was proposed. 2. Material In the study 50 at.% Ni–50 at.% Ti alloy (also called Ni– Ti or Nitinol) in superelastic form was used (Johnson Matthey Inc./SMA). Three types of the surface treatments were chosen to alter the surface reactivity of the alloy with SBF: heat treatment, alkali treatment and spark oxidation (plasma electrolysis). These types of treatments were demonstrated to have a positive effect on bioactivity of Ti and Ti-based alloys such as Ti6Al4V and Ti6Al7Nb [4,16,17,19,23]. Prior to the treatments, Ni–Ti samples of dimensions 8  8  0.8 mm, were ground to a mirror finish and cleaned in isopropanol and ultrapure water. They were then soaked in 60% nitric acid, and finally ultrasonicated in ultrapure water (18 MX cm) and dried in compressed air. Heat treatment for Ni–Ti samples was carried out at three different temperatures (200, 400 and 600 °C in air), followed by cooling to room temperature. Alkali treatment was conducted in 10 M NaOH. Samples were immersed in the solution for 24 h at 80 °C then rinsed with water, dried and heat-treated at 600 °C in air. Spark oxidation was carried out in a solution with a ratio of two parts of 85% phosphoric and three parts of 25% sulphuric acid. During treatment, the temperature was controlled and kept at 25–35 °C. Oxidation was done using direct current (current density 2 mA cm2 for 1 min), and a Ti mesh electrode coated with Pt was used as a cathode. In total, six types of surfaces were produced:    

heat-treated—samples denotated 200, 400 and 600; alkali-treated—BNT; spark-oxidized—SP; ground to a mirror finish and cleaned—NT.

As reference samples, cpTi was used. The surface of the cpTi was prepared by:  grinding to a mirror finish—Ti, called reference negative (ve);  alkali treatment: 10 M NaOH for 24 h at 80 °C then rinsed with water, dried and heat-treated at 600 °C in air—BioactiveTi (BTi) (23), also called reference positive (+ve).

3. Methods For all of the Ni–Ti and reference samples, topography, roughness, surface chemistry, structure and surface energy were investigated and related to the bioactivity evaluation results. These factors have been reported to alter the behaviour of the material in contact with body fluids. The experiments were conducted on triplicate samples.

W. Chrzanowski et al. / Acta Biomaterialia 4 (2008) 1969–1984

3.1. Surface topography/roughness The topography of the surface was examined using atomic force microscope (AFM) (PSIA, XE-100). Images of the surface were recorded using the tapping mode with a silicon tip: force constant 40 N m1, resonance frequency 325 kHz, scan rate 1 Hz. For all samples, three sizes of images were recorded: 10  10 lm, 25  25 lm and 45  45 lm. The Ra parameter was evaluated on the basis of roughness measured from 45  45 lm images. However, it must be highlighted that the cutoff length for AFM measurements is significantly shorter than the length given by the standard (DIN 4768, ISO 4288). These measurements results represent ‘‘nanoscale roughness” which is not evaluated according to the standard. For this reason, laser profilometry measurements (Proscan 1000, Scantron) were conducted to provide information on the standard ‘‘macroscale” Ra values. The cutoff length was kc = 0.8 mm and the evaluation length was 4 mm. 3.2. Surface free energy (SFE) SFE for the tested samples was calculated on the basis of contact angle measurements which were conducted with two liquids with polar and non-polar characteristics: ultrapure water and di iodomethane. In this study, Foweks’ equation was used to calculate the energy: qffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffi cdL cdS þ cpL cpS cL ð1 þ cos hÞ ¼ 2 Static contact angles for the test liquids were measured using a KSV Cam200 contact angle system (LOT-Oriel Ltd., UK). Droplets of approximately 5 ll of liquids were placed on the sample surface using a manual syringe. The drop profile was recorded at 1 s intervals for 1 min, and the measurements were carried out on triplicate samples. The calculation of the surface free energy was carried out using Foweks’ method via SFE KSV software. 3.3. Surface chemistry The chemical composition of the surfaces was measured using XPS (Thermo Escalab 220iXL). Measurements were performed using an Al Ka monochromated X-ray source and quantified using CasaXPS (Casa Software Ltd.). For all the samples, both survey and detailed spectra were recorded. Detailed spectra enable precise evaluation of the chemical composition. In this evaluation, a Shirley background type was used. In the first instance, samples which were not immersed in body media were measured, as a reference for further bioactive study. 3.4. Transformation temperature Differential scanning calorimetry (DSC) measurements were conducted to find out the transformation behavior and the characteristic transformation temperatures, i.e.

1971

austenite and martensite start and finish (As, Af, Ms, Mf), for Nitinol samples treated at 200, 400 and 600 °C using a Pyris Diamond DSC (Perkin–Elmer Instruments) and compared with the non-treated ground samples as controls. For DSC, discs (n = 3) of each sample were heated from 50 to 150 °C at 10 °C min1, and the transformation temperature was calculated by the onset of change in the endothermic direction (upwards) of the heat flow of the heating ramp. All tests were carried out under nitrogen purge. 3.5. Bioactivity evaluation The evaluation of bioactivity, interpreted as Ca and P precipitation (in the form of apatite) on the samples, was performed on the basis of XPS examination of samples following immersion in Hanks’ balanced salt solution (HBSS, Lonza UK; Table 1), Dulbecco’s modified Eagle’s medium (DMEM, Lonza UK; Table 2), and osteogenic media (DMEM + ascorbic acid 5 mg l1, dexamethazone 5 mg l1) for 3 h, 24 h and 7 days at 37 ± 1 °C. The measurements enabled evaluation of the kinetics of apatite film formation in the early hours of media contact with the surface. This is especially important due to the fact that cells try to attach to the surface and proliferate immediately after implant insertion and the surface properties over this period are critical to their behaviour. DMEM and osteogenic media contain proteins, peptides, vitamins and antibiotics (naturally occurring in the body), which can alter the reactivity of the surface due to their very fast interactions with metallic surfaces. The kinetics of the film formation was evaluated also using AFM observations. To detect rapid interactions of the proteins with the surfaces, XPS examination was carried out for samples soaked for 5, 10 and 15 min. 3.6. Cumulative Ti4+ and Ni2+ ions release Nickel and titanium ion release was evaluated using inductively coupled plasma mass spectrometry (ICP-MS) (Spectromass 2000, Spectro). Samples for ion release evaluation were immersed in HBSS for 3 h, 24 h, 3 days and 7 days. At each time point, the media was collected for analysis, and the samples were placed in fresh media. Two isotopes of each element were measured (48Ti, 47Ti, 58Ni and

Table 1 Composition of the HBSS media Chemical compound

mg l1

NaCl KCl Na2HPO4 KH2PO4 CaCl2 MgSO4  7H2O NaHCO3 Glucose Phenol

8000 400 90 60 186 200 350 1000 RedNa

1972

W. Chrzanowski et al. / Acta Biomaterialia 4 (2008) 1969–1984

Table 2 Composition of the DMEM media

4. Results

Chemical compound

mg l1

Chemical compound

mg l1

CaCl2 Fe(NO3)3  9H2O KCl MgSO4  7H2O NaCl NaHCO3 NaH2PO4  H2O Glucose Phenol Red Sodium pyruvate L-Arginine HCl L-Cystine L-Glutamine Glycine L-Histidine HCl  H2O L-Isoleucin L-Lysine HCl L-Methionine

200 0.1 400 200 6400 3700 125 4500 15 110 84 48 584 30 42 104.8 146.2 30

L-Phentyalanine

66 42 95 16 72 93.6 4 4 4 7 4 4 0.4 4 50 10 1

60

L-Sarine L-Threonine L-Tryptophan L-Tyrosine L-Valine D-Ca

pantothenate Choline chloride Folic acid i-Inositol Nicotinamine Pyridoxine HCl Riboflavin Thiamine HCl b-Glycerophospate FCS Streptomycin

Ni), and the results were considered by taking the average of three repeats. The instrument was calibrated in the range 1 ppb–1 ppm using a multielement standard (Spex CertiPrep, Claritas PPT, Standard2). The total amount of release of either element was considered as a cumulative release. The data was plotted as cumulative release, lg l1 mm2, as a function of time (h), and the release rate was calculated from the slope of the linear fit of the cumulative degradation against time.

4.1. Results of the surface topography/roughness AFM and laser profilometry measurements showed differences in surface topography and roughness (Fig. 1 and Table 3). The roughness of the ground Ni–Ti samples was around Ra = 0.14 lm and was similar to the reference ve. Heat treatment at 400 and 600 °C caused an increase in roughness up to Ra = 0.3 lm. It is worth noting that the roughness measured using AFM did not show significant differences up to 400 °C and Ra was around 9 nm. However, treatment at 600 °C caused a significant increase in ‘‘nanoscale” roughness: Ra = 22 nm. Topography of the ground and heat-treated samples at 200 and 400 °C was uniform and the surfaces were smooth (Fig. 1a–c). Heat treatment at 600 °C resulted in formation of large number of uniform ‘‘nanoscale” nodules. In general, the greatest roughness was observed for alkali-treated, spark-oxidized and reference +ve samples (Table 3). Alkali-treated samples had irregular topography with a large number of scratches, which were formed during the preparation procedure. Topography of the sparkoxidized samples was rather regular with numerous pinholes and pits (Fig. 1d). 4.2. Results of SFE SFE, which reflects the wetting ability of the surface, was the greatest for alkali-treated, spark-oxidized and ref-

Fig. 1. Topography of the surface. Three-dimensional AFM images of: (a) ground Ni–Ti, (b) heat-treated Ni–Ti at 400 °C, (c) heat-treated Ni–Ti at 600 °C and (d) spark-oxidized Ni–Ti.

W. Chrzanowski et al. / Acta Biomaterialia 4 (2008) 1969–1984

1973

Table 3 Roughness and surface free energy results for tested samples Contact angle

cpTi BTi NT 200 400 600 BNT SP

Surface free energy 1

d

Roughness

SFE (mN.m-1)

tot

H2O (°)

DII (°)

SFE (mN m )

SFE (mN m )

SFE

47.18 ± 0.42 7.18 ± 0.58 32.71 ± 0.44 60.55 ± 4.42 55.77 ± 8.09 45.13 ± 1.84 14.52 ± 3.09 12.25 ± 6.20

46.12 ± 1.18 49.13 ± 3.21 37.00 ± 4.52 39.42 ± 1.98 39.65 ± 2.34 24.41 ± 3.46 39.84 ± 5.13 46.13 ± 2.14

36.40 ± 0.63 34.74 ± 1.78 41.03 ± 2.21 39.89 ± 0.99 39.77 ± 1.18 46.30 ± 1.23 39.62 ± 2.52 36.39 ± 1.15

21.31 ± 0.06 39.69 ± 1.30 27.08 ± 1.00 12.09 ± 2.02 14.88 ± 4.61 18.16 ± 0.75 34.97 ± 2.18 37.41 ± 2.34

57.71 ± 0.57 74.43 ± 0.48 68.11 ± 1.20 51.98 ± 3.01 54.65 ± 4.24 64.47 ± 1.85 74.59 ± 0.66 73.80 ± 1.21

1

(mN m )

Ra (lm)

RAFM (nm) a

0.25 ± 0.03 0.92 ± 0.04 0.14 ± 0.01 0.18 ± 0.06 0.32 ± 0.01 0.30 ± 0.09 0.79 ± 0.09 0.80 ± 0.08

6.10 ± 1.3 93.50 ± 7.5 10.28 ± 1.01 8.08 ± 1.32 7.06 ± 1.28 22.88 ± 3.42 42.55 ± 2.54 142.16 ± 24.57

liquids with the surface that could be related to the surface chemistry. When only heat-treated samples (200, 400 and 600 °C) were considered it was also observed that increasing heat treatment temperature caused an increase in SFE. The ground and heat-treated samples had a much lower polar fraction but a comparable dispersive component (SFEd) to the rest of the samples (Fig. 2). This suggests that chemical reactions did not play a major role for these samples when interacting with fluids.

80

60

1

p

SFEd p SFE SFEtot

40

20

4.3. Surface chemistry

0 cpTi

BTi

NT

BNT

SP

200

400

600

Fig. 2. Surface free energy for reference samples: cpTi, BTi and nickel titanium samples: NT, BNT, SP, 200, 400 and 600.

erence +ve samples (Fig. 2). However, the polar part of the energy (SFEp) was greater for the reference +ve sample, which may suggest stronger chemical interactions of the

XPS analysis of the ground and cleaned samples revealed the following elements on the surface: Ti, Ni, O, C, N and Na (Fig. 3). Detailed spectra were recorded for these elements and an accurate chemical composition was established (Table 4):  Carbon had three main peaks at energies of 284.6, 286.2 and 288.5 eV. The main peak at 284.6 eV corresponds to single-bonded carbon with carbon or hydrogen: C–C,

Fig. 3. XPS survey spectra for ground Ni–Ti sample (NT).

1974

W. Chrzanowski et al. / Acta Biomaterialia 4 (2008) 1969–1984

Table 4 Chemical composition of the samples after treatment as measured using XPS Sample

C (at.%)

Ca (at.%)

N (at.%)

Na (at.%)

Ni (at.%)

O (at.%)

P (at.%)

Ti (at.%)

Ti BTi NT 200 400 600 BNT SP

15.82 ± 1.20 5.15 ± 0.55 10.99 ± 1.30 7.93 ± 0.34 7.33 ± 0.62 13.65 ± 6.90 7.43 ± 0.44 12.71 ± 1.58

0.16 ± 0.04 3.93 ± 0.82 n/a n/a 0.16 ± 0.05 n/a 1.23 ± 0.27 0.36 ± 0.22

n/a n/a 0.25 ± 0.11 n/a n/a n/a n/a 0.71 ± 0.63

3.53 ± 0.48 21.27 ± 2.33 5.09 ± 0.44 6.22 ± 2.08 5.67 ± 3.09 4.14 ± 1.91 1.87 ± 0.20 0.75 ± 0.68

n/a n/a 8.00 ± 1.16 6.09 ± 0.98 6.13 ± 3.00 1.88 ± 0.57 34.83 ± 1.70 3.29 ± 0.37

58.84 ± 0.57 50.93 ± 3.69 57.52 ± 0.82 60.18 ± 1.09 60.63 ± 0.19 59.17 ± 4.42 52.55 ± 0.10 65.70 ± 1.20

0.62 ± 0.06 4.29 ± 1.20 n/a n/a n/a n/a n/a 10.93 ± 0.22

21.04 ± 0.43 16.40 ± 1.65 18.15 ± 0.40 19.58 ± 0.06 20.08 ± 0.29 21.16 ± 1.33 2.09 ± 0.69 5.55 ± 0.85

C–H. The next peak is associated with carbon singlebonded with oxygen: C–O. These two peaks are the major contribution to the total amount of carbon and were recognised as contaminations. The highest-energy carbon peak corresponds to double-bonded carbon C@O/O–C–O (Fig. 4a).  Titanium had two main peaks at 454.1 eV, which correspond to metallic titanium, and a second double peak at 458.5 and 646 eV, which corresponds to titanium dioxide (Fig. 4b).  Oxygen was recognized as O2 (530.0 eV), hydroxide (OH) and Ni2O3 (531.5 eV), and some water and/or double-bonded carbon oxygen (–C@O) was observed (532.8 eV) (Fig. 4c).

 Nickel: for the ground sample, nickel was recognized mainly in metallic form, with a binding energy of 852.4 eV (Fig. 4d).  Sodium had a main peak with an energy 1071.7 eV, which was recognized mainly as metallic sodium Na2+. Heat treatment at 200 °C caused some changes in the shape of the C, Ti and Ni lines. The C1s line had two main peaks at energies of 285 and 288.8 eV. The height of the second peak was significantly higher compared to non-heattreated samples. This suggests that the amount of doublebonded carbon with oxygen was significantly greater for these samples. The main peak of Ti (Ti2p) was at an energy of 458.6 eV (TiO2). In addition, a very small peak was

Fig. 4. XPS spectra obtained for ground sample: (a) C1s, (b) Ti2p, (c) O1s and (d) Ni2p.

W. Chrzanowski et al. / Acta Biomaterialia 4 (2008) 1969–1984

observed at 454.2 eV; this indicated traces of metallic Ti in this sample (Fig 5a). The total amount of Ti was 19.5 at.%. Ni (6 at.%) had a main peak at an energy of 856 eV and a small peak at 852 eV. This shape and shift of the energy suggests that Ni was oxidized (NiO, Ni2O3) though some remains in metallic form (Fig. 5b). A further increase in the heat treatment temperature caused subsequent changes in the line shapes. Carbon (C1s), for the samples treated at 400 °C, had one major peak (284.5 eV) and two small peaks at 286 and 289 eV. This suggests that the carbon was derived from contamination. Ti (20 at.%) and Ni (6.1 at.%) lines with peaks at 458.5 and 855.6 eV indicated that the elements were present only in the oxidized form (Fig. 5a and b). In addition, a small amount of Ca (0.15 at.%) was observed on the surface. The Ca2p line with peak at an energy of 348.2 eV corresponds with CaCl2 and more likely CaCO3. Heat treatment at 600 °C caused further oxide film growth, evaluated on the basis of the energy of the Ti and Ni peaks (Ti: 458.6 eV, Ni: 855.8 eV). A decrease in Ni content to 1.8 at.% in the layer was also observed. The most significant changes in chemical composition were caused by alkali treatment and spark oxidation. Alkali-treated samples had significantly higher amounts of Ni on the surface: 35 at.%. The main peak at an energy of 854.6 eV corresponded to NiO (Fig. 5c). However, another peak at 856.1 eV contributed to the main peak, and this corresponded to Ni2O3 or Ni(OH)2. For this type

1975

of sample, the amount of Ti (as TiO2: 458 eV (Fig. 5d)) and O (O2, OH and H2O: 530.3, 531.9 and 532.9 eV, respectively) dropped to 2 and 52 at.%, respectively. Additionally, 1 at.% Ca recognized as CaO (346.1 eV) was observed on the surface. For spark-oxidized samples further changes were observed. On the carbon line, an additional peak at a binding energy of 293.1 eV was observed, suggesting small amounts of potassium, which was present in the solution used for the oxidation. Another explanation could be that the layer is partially insulating and the charge compensation used was not effective in the valleys, leading to localized charging. However, this is unlikely as a similar effect would be expected in O1s, which was not observed. Ni was present as the oxide Ni2O3 and its content was about 3 at.% (Fig. 4c). Ti was recognized as TiO2 with a total amount of 5 at.% (Fig 4d). Most significantly, the sparkoxidized samples contained about 11 at.% phosphorous. The P2p line had one major peak at 134.2 eV which was consistent with P2O5. However, for all the spark-oxidized samples charge compensation needed to be used due to a relatively thick, non-conductive oxide layer on the surface, which could cause some uncertainty in interpretation of the spectra. Small amounts (0.3 at.%) of Ca were also detected on spark-oxidized samples. Ca2+ impurities came from the water that was used during sample preparation. Reference samples were mainly composed of C, Ti, O, Na, Ca and P. Carbon on cpTi (15 at.%) was recognized

Fig. 5. XPS detailed spectra: (a) N2p for NT, 200, 400 and 600, (b) Ti2p for NT, 200, 400, 600, (c) Ni2p for NT, BNT, SP and (d) Ti2p for Ti, BTi, NT, BNT and SP.

1976

W. Chrzanowski et al. / Acta Biomaterialia 4 (2008) 1969–1984

Fig. 6. XPS survey spectra for reference cpTi and BTi.

mainly as contamination (Fig. 6). Ti was present in both oxidized and metallic form. Ca (0.1 at.%), Na (3 at.%) and P (0.6 at.%) corresponded to CaO (347.1 eV), metallic Na (1071.7 eV), and NaPO3 and P2O5 (133.4 eV). Alkali treatment of Ti caused a drop in the Ti content to 5 at.% and significant increase of Ca (4 at.%), Na (21 at.%) and P (4 at.%). The binding energy of the Ca corresponded with CaO, Na with metallic Na and NaTiO3, and P with NaPO3 and P2O5. In addition water, hydroxide and oxygen contents were analyzed on the basis of component fitting over the O1s peak (Table 5 and Fig. 7), and the Marquardt technique was used. In general, minor differences were observed between samples, and both alkali-treated and spark-oxidized samples contained no water (on the basis of O1s XPS spectra fitting). On other hand, the ground Ni–Ti samples (NT) contained the highest amount of water and hydroxides, and thus the lowest amount of O2. The ratio O2/OH was significantly higher for samples treated at 400 and 600 °C. A lower concentration of OH- content may result in reduced wetting ability and lower bioactivity. It is noteworthy that the level of carbon contamination for

Fig. 7. Fitting of the oxygen components on the O1s line for Ni–Ti sample treated at 200 °C.

all the samples was low (5–15 at.%), indicating good preparation of the samples. 4.4. Results of transformation temperature

Table 5 Content of the water, hydroxide and oxygen on the basis of fitting done for O1s peak Sample

H2O (at.%)

O2 (at.%)

OH (at.%)

Ti BTI NT 200C 400C 600C BNT SP

5.99 ± 0.15 1.24 ± 0.18 6.60 ± 0.29 3.77 ± 0.36 3.43 ± 0.73 2.96 ± 0.88 1.58 ± 0.36 0.00 ± 0.00

86.86 ± 0.30 93.15 ± 0.21 84.44 ± 0.15 88.60 ± 0.60 91.95 ± 0.62 92.24 ± 0.88 91.53 ± 0.78 92.03 ± 1.50

7.14 ± 0.46 5.59 ± 0.03 8.94 ± 0.16 7.62 ± 0.31 4.60 ± 0.11 4.78 ± 1.12 6.88 ± 0.42 7.97 ± 1.51

DSC analysis showed that heat treatment altered the austenite and martensite transformation temperatures (start, finish and peak). The untreated polished samples had a transformation peak temperature of 18.25 ± 0.88 °C. Heat treatment caused an increase in the transformation temperatures (Table 6). A steady increase of the start, peak and finish temperatures for transformation austenite ? martensite during the heating run was observed up to a heat treatment temperature of 400 °C. For these samples, the recorded temperatures were the highest. A similar trend was observed for the cooling run, and an

W. Chrzanowski et al. / Acta Biomaterialia 4 (2008) 1969–1984

1977

Table 6 Onset, peak and finish temperatures for austenite ? martensite and martensite ? austenite transformation obtained from DSC measurements during heating and cooling of the ground and heat-treated samples

Ground 200 300 400 500 600

Heating

Cooling

As (°C)

Peak (°C)

Af (°C)

Ms (°C)

Peak (°C)

Mf (°C)

5.55 ± 2.74 8.47 ± 2.43 13.17 ± 2.27 39.62 ± 0.70 27.22 ± 1.85 12.09 ± 1.18

18.28 ± 0.94 18.78 ± 2.93 22.80 ± 1.50 48.74 ± 1.15 35.95 ± 0.02 30.79 ± 1.37

27.74 ± 1.39 28.04 ± 3.22 34.10 ± 0.23 57.54 ± 1.74 44.03 ± 0.69 42.30 ± 1.32

14.06 ± 0.62 17.96 ± 6.07 22.96 ± 5.16 41.66 ± 0.73 27.10 ± 0.20 9.19 ± 1.87

4.64 ± 0.61 7.24 ± 2.84 8.91 ± 2.38 32.05 ± 0.46 17.76 ± 0.63 14.64 ± 0.74

5.09 ± 0.80 1.95 ± 1.89 0.34 ± 2.05 22.09 ± 1.50 13.13 ± 1.15 19.66 ± 0.31

increase of the temperatures compared with ground samples was observed. Further increase of the heat treatment temperature caused a decrease in the transformation temperatures. For samples treated at 600 °C, the start of the austenite transformation appeared about 7 °C higher than for ground samples. However, the martensite formation started and finished at significantly lower temperatures ¼ 9:1  C; M600 ¼ 19:6  CÞ than observed for ðM600 s f Ground ¼ 4:6  C; MGround ¼ 5:1  CÞ ground samples ðMs f (Table 6). Thermally treated samples at 400 and 600 °C had significantly higher (P < 0.05) transformation temperatures compared to the untreated, ground samples (Fig. 8 and Table 6). 4.5. Results of bioactivity evaluation The amounts of Ca and P that precipitated on the samples immersed for 3 h in HBSS did not exceed 0.5 and 0.8 at.%, respectively, and were very similar for all of the samples except the spark-oxidized ones. For spark-oxidized samples, the precipitation of the Ca was significantly higher (2.8 at.% vs. 0.6 at.%), and the amount of P was also greater (11.8 at.%) (Table 7). However, since the initial amount of P was about 11 at.%, interpretation was difficult. It is not possible to determine whether P is being released into solution from the surface and being replaced by a calcium phosphate surface film or whether there is very little P addition to the surface. In this case, it was not possible to distinguish which component came from the HBSS as both appeared in an oxidized form which made analysis of Ca–P precipitation difficult for this type of sample. Analysis of the chemical composition of samples immersed for 24 h in HBSS revealed significant differences between them in terms of Ca and P content. The largest amounts of Ca and P precipitated were found on three types of samples: both references and ground Ni–Ti (Table 7). The amount of Ca was approximately 10 at.% and P 7.5 at.% for these samples. For spark-oxidized, heat-treated samples at 200 and 400 °C, the amount of Ca  5 at.% and P  4 at.% (spark-oxidized 11 at.%). Slightly lower amounts of both elements was observed for alkali-treated samples: Ca  4 at.%, P  3 at.%. Hardly any precipitation was observed for samples heat-treated at 600 °C. In addition, for all of the samples, Ca:P ratios were also evaluated.

80 60 Heat Flow (mW) - Endo UP

Sample

40 As

20

Af

0 Mf

-20

Ms

-40 -60 -100

-50

0

50

100

150

200

250

o

Temperature ( C)

Fig. 8. Transformation temperatures of ground nickel titanium (NT) and heat-treated at 200, 400 and 600 °C.

After 24 h immersion, all samples except for spark-oxidized and 600 had Ca:P ratios of approximately 1.2–1.4. These values are not very close to expected value of 1.67 found for hydroxyapatite. These results showed that in the early stages all the samples responded in a similar way; however, further film growth was accelerated only for ground Ni–Ti samples. Heat-treated at 200 and 400 °C and spark-oxidized and alkali-treated Ni–Ti showed lower rates of the film growth compared with ground and reference samples. Nevertheless, in general, a steady growth of Ca:P film on the surface of all tested samples was observed. The differences in kinetics of films grown on different samples were confirmed by the AFM observations (Fig. 9). Results of Ca and P content for samples immersed in DMEM showed significant differences in the kinetics of precipitation. At the early immersion time points, the fastest precipitation was observed for the reference samples, alkali-treated and also for samples heat-treated at 600 °C. The amount of Ca  4.5–5.8 at.% and P  3.2–3.8 at.%. Values of around half of these were observed for the rest of the samples (Fig. 9). Analysis of the chemistry after 24 h immersion showed that except for the reference ve there were no statistically significant differences in the amount of Ca and P between samples, i.e. all the samples

1978

W. Chrzanowski et al. / Acta Biomaterialia 4 (2008) 1969–1984

Table 7 Surface chemical composition of the samples before and after immersing in HBSS, DMEM and osteogenic media for 24 h Sample

Ca (at.%)

P (at.%)

Ni (at.%)

Ti (at.%)

Ca:P

Non-immersed Ti BTi NT 200 400 600 BNT SP

0.15 ± 0.04 3.93 ± 0.817 n/a n/a 0.15 ± 0.05 n/a 1.23 ± 0.27 0.36 ± 0.22

0.61 ± 0.01 4.28 ± 1.20 n/a n/a n/a n/a n/a 10.92 ± 0.21

n/a n/a 7.99 ± 1.15 6.08 ± 0.98 6.12 ± 3.01 1.88 ± 0.57 34.83 ± 1.70 3.29 ± 0.36

21.03 ± 0.43 16.40 ± 1.64 18.14 ± 0.40 19.57 ± 0.06 20.07 ± 0.28 21.16 ± 1.32 2.09 ± 0.68 5.54 ± 0.84

0.254 ± 0.01 0.917 ± 0.21 n/a n/a n/a n/a n/a 0.033 ± 0.01

Immersed for 24 h in HBSS Ti 24 h 10.31 ± 0.34 BTi 24 h 10.84 ± 0.44 NT 24 h 9.42 ± 0.71 400 24 h 4.75 ± 0.76 600 24 h 0.22 ± 0.13 BNT 24 h 3.56 ± 0.40 SP 24 h 4.77 ± 0.34

7.77 ± 0.52 7.60 ± 0.57 6.72 ± 0.42 3.62 ± 0.27 0.61 ± 0.08 2.94 ± 0.92 11.14 ± 0.03

n/a n/a 0.98 ± 1.15 5.82 ± 1.53 2.13 ± 0.48 19.36 ± 4.11 1.54 ± 0.03

2.76 ± 1.28 3.61 ± 2.03 4.70 ± 2.80 9.86 ± 2.53 24.17 ± 0.19 1.39 ± 0.65 4.72 ± 0.44

1.32 ± 0.04 1.42 ± 0.04 1.40 ± 0.01 1.30 ± 0.11 0.36 ± 0.17 1.25 ± 0.25 0.42 ± 0.03

Immersed for 24 h in DMEM Ti 24 h 4.68 ± 0.55 BTi 24 h 7.84 ± 0.880 NT 24 h 7.76 ± 0.75 400 24 h 6.61 ± 0.34 600 24 h 7.19 ± 0.27 BNT 24 h 7.32 ± 0.44 SP 24 h 8.16 ± 0.43

2.75 ± 0.46 4.61 ± 0.68 4.81 ± 0.98 4.29 ± 0.39 4.52 ± 0.27 4.69 ± 0.51 6.52 ± 0.31

n/a n/a 0.20 ± 0.05 1.23 ± 0.48 0.00 ± 0.00 0.67 ± 0.03 0.12 ± 0.16

3.55 ± 0.78 2.91 ± 0.70 0.60 ± 0.03 1.71 ± 0.19 1.55 ± 0.74 0.44 ± 0.20 0.88 ± 0.53

1.71 ± 0.06 1.70 ± 0.04 1.61 ± 0.09 1.54 ± 0.08 1.59 ± 0.03 1.56 ± 0.03 1.25 ± 0.05

Immersed for 24 h in osteogenic media Ti 24 h 1.60 ± 0.65 BTi 24 h 4.13 ± 1.57 NT 24 h 7.23 ± 0.40 400 24 h 7.59 ± 1.42 600 24 h 8.63 ± 0.07 BNT 24 h 8.09 ± 0.53 SP 24 h 8.912 ± 0.39

1.13 ± 0.54 2.53 ± 1.33 4.58 ± 0.19 5.12 ± 0.86 5.63 ± 0.25 5.46 ± 0.26 6.39 ± 0.31

n/a n/a 0.22 ± 0.31 n/a n/a 0.18 ± 0.07 n/a

9.85 ± 1.04 8.06 ± 3.37 1.20 ± 0.39 0.73 ± 0.46 0.57 ± 0.07 0.27 ± 0.04 0.50 ± 0.14

1.44 ± 0.07 1.68 ± 0.25 1.57 ± 0.06 1.48 ± 0.05 1.53 ± 0.11 1.48 ± 0.09 1.39 ± 0.09

Fig. 9. AFM images of the samples immersed for 24 h in HBSS: (a) heat-treated at 400 °C; (b) heat-treated at 600 °C.

reacted in a similar way. The detected Ca:P ratio was in the range of 1.6–1.7, which was close to the expected value for apatite. Further study carried out in osteogenic media showed that additional components (ascorbic acid and Na-b-glycerophosphate) slightly altered the reactivity of the surface (Table 7 and Fig. 10). After 3 h immersion in the osteogenic media, the highest amount of Ca and P was observed for both reference sample and the alkali-treated samples.

However, for the positive reference, we should bear in mind that the initial amount of Ca was 4% which may contribute to the total amount of Ca measured after immersion. This could indicate that amount of Ca deposited from the media was lower, as part of it could be a signal from the substrate. Furthermore, this suggests that the growth rate was not the fastest. The same argument holds true for alkali-treated Ni–Ti samples. Immersion for 24 h, however, resulted in further film growth. The fastest increase in

W. Chrzanowski et al. / Acta Biomaterialia 4 (2008) 1969–1984

1979

14

10

12 8

10

P (%)

Ca (%)

6

4

8

6

4 2

2

0

0 Ti

BTi

NT

200 C

400 C

600C

BNT

SP

Ti

BTi

NT

200 C

400 C

600C

BNT

SP

Fig. 10. Results of XPS analysis of Ca and P precipitated from the DMEM after 0, 3 and 24 h.

Fig. 11. Three-dimensional AFM images of the surface of ground Ni–Ti after immersing in osteogenic media for (a) 3 h and (b) 24 h.

Ca and P was observed for alkali-treated, spark-oxidized and 600 °C heat-treated samples. For all these samples, the amounts of Ca and P were the greatest: Ca, 8–9 at.%; P, 5.8–6.4 at.%. Approximately 2 at.% lower Ca concentration and 1.5 at.% lower P concentration were observed for ground and 400 °C heat-treated samples. For all Ni– Ti samples, steady growth of the apatite film was observed over the immersion time, which was additionally confirmed by the AFM images (Fig. 11). It was also noted that increasing heat treatment temperature produced an increase in the amount of Ca and P at both time points. Nevertheless, for both reference samples, no significant changes were observed at the measured immersion time points. Initial rapid film growth (first 3 h) was stopped and no further increase in film thickness was observed. Some differences were also observed between samples soaked in DMEM and osteogenic media. In general, osteogenic media facilitated higher precipitation of both Ca and P—particularly so for alkali-treated, spark-oxidized, 600 °C heat-treated and reference ve samples. For the rest of the samples, the differences were not statistically significant. AFM observations were used to confirm the analysis of the kinetics of film growth. Some differences were observed between samples immersed in media with different protein contents. After immersion in HBSS, precipitates were irreg-

ularly located on the surface and formed mountain-shaped groups with heights of up to 1.8 lm and widths of up to 6 lm (Fig. 12a). Immersion in DMEM and osteogenic media, however, resulted in a more uniform fine structure with large number of spike-shaped forms. The typical size after 24 h immersing was: height up to 600 nm and width up to 4.5 lm (Fig. 12b). XPS analysis of the samples immersed for 7 days did not yield any further information. On the surfaces, only Ca, P, O and C were observed, suggesting that the formed layer was too thick to be measured by XPS without sputtering. This layer covered the whole surface, and for this reason, these results were not taken into consideration. In addition to assessing the speed of the protein interactions from DMEM and osteogenic media with the surfaces, nitrogen content was assessed after immersing samples for 5, 10 and 15 min. For all of the samples, 0.5–1 at.% nitrogen was found after 5 min and an increase in this content was observed over time to reach 5 at.% after 15 min with no significant differences detected between samples. 4.6. Results of titanium and nickel ion release—ICP-MS The results of Ti4+ and Ni2+ ion release are presented in Fig. 13. It should be noted that the concentrations are cumulative. A gradual increase of Ti4+ cumulative ion release was

1980

W. Chrzanowski et al. / Acta Biomaterialia 4 (2008) 1969–1984

Fig. 12. AFM results. Line profile of the sample 400 immersed for 24 h in (a) HBSS and (b) DMEM.

observed with time. The highest Ti4+ release rate was observed for spark-oxidized samples: 0.14 lg l1 mm2 h1 (Fig. 13a). The rate of Ti4+ release from SP samples was significantly greater than for other samples. The other samples showed initial release rates of 0.01–0.04 lg l1 mm2 h1. More significantly, release rates between 1 and 7 days were 0.003–0.005 for the all samples except SP, for which the rate was 0.09 lg l1 mm2 h1, indicating sustained elevated Ti4+ release from these samples. Significant differences were observed in Ni2+ ion release. It is apparent from Fig. 13b that the release from the NT, 200 and 400 samples is low, and calculation of the release rates show that this is below 0.02 lg l1 mm2 h1 initially and less than 0.0002 lg l1 mm2 h1 at 7 days. Although the SP sample shows very high initial release (0.59 lg l1 mm2 h1), this falls to 0.046 lg l1 mm2 h1 at 7 days. The BNT sample shows an alarming Ni2+ release. The initial high levels of release (0.14 lg l1 mm2 h1) are sustained and indeed slightly elevated at 7 days to 0.18 lg l1

mm2 h1. The 600 sample also shows a similar trend although with lower overall release rates beginning at 0.023 lg l1 mm2 h1 and rising to 0.37 lg l1 mm2 h1. This trend is in contrast to the lower heat treatment temperatures which show decreasing nickel ion release with time. It must be highlighted that taking into account the size of the implant, the total amount of Ni2+ released could be significantly higher and may be expected to exceed a dose of 200–300 lg day1 for alkali-treated samples. The release of Ni at levels approximating the recommended daily amount for ingestion is of concern as release from a biomaterial is likely to result in far higher local concentrations than ingested Ni. 5. Discussion The aim of the study was to investigate the bioactive properties of Ni–Ti alloy after different surface treatments and examine their suitability for biomedical device preparation.

W. Chrzanowski et al. / Acta Biomaterialia 4 (2008) 1969–1984

NT 200 400 600 BNT SP cpTi BTi

15

Cumulative release Ti

4+

-1

-2

(μgL mm )

20

10

5

0

0

20

40

60

80

100

120

140

160

180

Time (hours)

-2 -1

25 20

Cumulative release Ni

30

2+

(μgL mm )

35 NT 200 400 600 BNT SP

15 10 5 0

0

20

40

60

80

100

120

140

160

180

Time (hours)

Fig. 13. ICP-MS results. Titanium (a) and nickel (b) cumulative ion release from reference (Ti, BTi) and Ni–Ti samples (NT, 200, 400, 600, BNT and SP) immersed in HBSS for 3 h, 24 h, 3 days and 7 days.

Bioactivity for the purpose of the study was interpreted as the ability of a material to encourage precipitation of Ca and P on the surface, predominantly in apatite form, from the surrounding environment. Many different factors, including surface chemistry, surface free energy, roughness and topography, have an impact on the bioactivity of the material. In this paper, all these properties and their correlation to the kinetics of Ca and P precipitation from HBSS, DMEM and osteogenic media were evaluated. Bioactivity was evaluated on the basis of chemical analysis of the sample surfaces following soaking for up to 7 days. However, after immersing samples for 7 days the layer thickness was significantly greater than the detection depth of XPS. The composition of all tested samples was similar. For these reasons, the kinetics of apatite film growth was analyzed only for the initial 24 h. Nevertheless, early reaction of the body media with implants seems to be critical for the subsequent cell interactions, and these results can yield important information for assessing the suitability of surface treatments for biomedical applications. Three groups of treatments were applied for the superelastic Ni–Ti alloy: heat treatment (200, 400 and 600 °C),

1981

alkali treatment (NaOH) and spark oxidation (H2SO4 + H3PO4). All the treatments altered the chemical composition and roughness. The topography of the samples heat-treated at 600 °C differed from the samples treated at lower temperatures; the roughness measured using AFM was significantly ¼ 22 nmÞ and the surface was covered with higher ðRAFM a a uniform, fine structure of large numbers of nodules (comprised of oxides). For this type of sample, a greater SFE was observed when compared to samples treated at lower temperatures. This high SFE can be related to this fine structure and thicker oxide layer (evaluated on the basis of XPS examination). Increased roughness and a thicker oxide layer may improve bioactivity. It was demonstrated that specific roughness can stimulate cells to attach and proliferate; additionally, rough structures seem to be more favourable for biomineralization due to their greater surface area compared to a smooth surface [8,15,30,31]. Higher SFE indicated better wetting ability which may translate to better bioactivity. However, further chemical analysis of the layers that were formed on the samples from different SBF did not confirm this assumption. Samples treated at 600 °C had far lower bioactivity compared to the samples treated at lower temperatures. One of the explanations for this could be the fact that roughness measured by laser profilometry (macroscale roughness) was similar to that of the other heat-treated samples, while nanoscale roughness (measured using AFM) was significantly greater. It could be assumed that this nanoscale roughness does not encourage Ca/P precipitation. In addition, the dispersive part of the SFE and Ni2+ ion release were the greatest among the heat-treated samples, which additionally could impair the bioactivity. It has been demonstrated that materials that are hydrophilic have better bioactivity in contact with body fluids [21]. However, many investigators have measured contact angle using pure water (a polar liquid) and related the results to the general wetting ability. Bearing in mind that body fluids contain organic and inorganic components that interact with the surface and create chemical bonds, measurement of SFE seems to be more appropriate as it includes both dispersive and polar forces. The dispersion forces (SFEd) are general forces acting between any atoms and molecules. In addition, chemical interactions can come into play when we deal with specific substances. The presence of these interactions translates into stronger adhesion between the liquid and the substrate due to additional contributions from these polar (SFEp) interactions [32]. Following alkali treatment, samples (both Ti and Ni–Ti) had very high wetting ability (SFEtot = 74 mN m2) and rough surfaces with irregular topography and a large number of scratches. Similar roughness (Ra = 0.8 lm) was also observed for spark-oxidized samples. The surface topography for spark-oxidized samples was very regular. Its structure was pitted with numerous pinholes, and the thickness of the oxide layer was 3.1 lm [33]. This treatment enriched the surface with 11 at.% P and resulted in very good wet-

1982

W. Chrzanowski et al. / Acta Biomaterialia 4 (2008) 1969–1984

ting ability, comparable with that of alkali-treated samples. This could be the combined result of surface topography and elimination of surface-bound water. High SFE (particularly the polar part), the incorporation of P into the layer, and roughness in the range which was reported to stimulate cells to attach could be recognized as positive findings. More detailed analysis of the surface free energy showed that three types of samples with the highest-energy (reference +ve, alkali-treated and spark-oxidized) had a ratio between the polar and dispersive part of the energy close to 1. While for heat-treated samples, the dispersive part was 2–3 times higher; the same was also observed for the reference ve sample. These results suggested that for both alkali-treated and spark-oxidized samples, the chemical interactions between the wetting liquid and the surface play an important role and are much stronger than on the rest of the samples. This may indicate better bioactivity of these types of surface. Information related to the transformation temperature and thus the structure of the material at body temperature may be important in terms of bioactivity and design of the implant. Bioactivity, which may be affected by the crystallographic structure of the top layer [4], could be altered by the presence of the martensitic phase. Thermal analysis of the samples revealed that heat treatment altered the transformation temperatures (As, Af, Ms, Mf and peak). The highest specific temperatures during heating and cooling runs were observed for samples treated at 400 °C, and further increase of treatment temperature caused a reduction in these temperatures. These results showed that samples analyzed at body temperature had two different structures: austenite (NT, 200, 600, BNT, SP) and a combination of austenite + martensite (400) (Fig. 8). Differences in structure can have an influence on the reactivity of the surface in body fluids due to a difference in crystallographic matching between inorganic components (hydroxyapatite) of the fluids and the surface [4]. It was found that the bioactivity was lowest for samples heat-treated at 600 °C for which the end of the martensitic transformation was below 37 °C, and austenitic transformation was found to commence at 37 °C. This could suggest that a metastable structure of Nitinol is not favourable for bioactivity. Analysis of the chemical composition of the samples showed a reduction of surface Ni content after applied treatments for all of the samples except the alkali-treated ones. Alkali treatment, however, resulted in a significant increase in Ni. This result is very negative in terms of biomedical applications. The surface was composed mainly of nickel oxides with small amounts of titanium dioxide. It is well known that nickel oxide is less stable in body fluids [2], which in turn can potentially result in a local increase of Ni content in surrounding tissues. In addition, on the surface of alkali-treated samples small amounts of Na and Ca were observed. Ni2+ ion release studies confirmed that the rate of Ni2+ release was the greatest for this type of sample. Despite the favourable roughness (Ra = 0.79 lm) and SFE (SFEtot = 74 mN m2), which could indicate good

bioactivity, the very high Ni2+ content on the surface which in turn resulted in high Ni2+ release, disqualifies the treatment as being useful for any biomedical application. Reference samples were covered with a titanium oxide layer, which naturally occurred on cpTi, and was chemically created for BTi samples. However, on both samples some Ca and P was found. These elements could precipitate on the surface from the water which was used for sample washing. In addition, BTi samples were heat-treated in the ceramic sample-holders, which could give rise to these elements on the surface. Bioactivity evaluation, which was conducted in three different media, showed that the type of media has an influence on the kinetics of Ca and P precipitation. Analysis of the XPS results for samples soaked in HBSS showed Ca–P film growth over the time for all the samples. The fastest growth together with the highest amount of both elements was observed for both reference samples and ground Ni– Ti. The Ca:P ratio varied between 1.2 and 1.4, which is relatively far from the expected value 1.67. The fast growth of the apatite film can be attributed to the titanium and its oxide present on the surface due to their high reactivity. Only one type of sample (heat-treated at 600 °C) resulted in hardly any precipitation. The main reason for the very low precipitation was assumed to be the structure, and additionally the nanoscale roughness seemed to discourage biomineralization. It is well known that higher roughness is beneficial from a bioactivity point of view; however, this is true for microscale roughness but not for nanoscale roughness. In this case, the 600 samples had very similar microscale roughness but significantly greater nanoscale roughness. This, in conjunction with structure and higher Ni2+ release rate compared with other heat-treated samples, resulted in zero bioactivity. The amount of Ca2+ that precipitated on SP and BNT samples was lower than that for ground and reference samples. This behaviour can be related to the significantly higher rate of both Ti4+ and Ni2+ release, which indicates the soluble nature of the layer. For spark-oxidized samples it must be highlighted that the Ca:P ratio was very low due to the fact that these types of samples contained initially about 11 at.% P, which made the analysis difficult. It is also noteworthy that the high Ni:Ti ratio for alkalitreated samples surged up to 72 after 3 h immersion in HBSS and then dropped back to the initial value of 16 after 24 h. In general, higher bioactivity was observed for pure Ti than for the Ni–Ti after surface treatment. One can say that in this case, Ni had a negative influence on bioactive properties. However, it was found that the greater Ni2+ content did not reduce the bioactivity. Thus, it seems to be more acceptable that the higher reactivity of pure Ti surface is favourable for apatite film formation, while surfaces with some Ni content may decrease the rate of film growth. The greatest activity observed for Ni–Ti was detected for ground samples. For ground samples, it was observed that Ni was predominantly in the metallic form, but in an oxidized form for the treated samples. It can be assumed that

W. Chrzanowski et al. / Acta Biomaterialia 4 (2008) 1969–1984

nickel oxides, which are not stable in body fluids, could affect the film formation. However, the metal form of Ti or Ni found in the ground samples has a positive influence on bioactivity due to their potential reactivity with oxygen and other elements. The manner of the Ca:P precipitation on the ground Ni–Ti was similar to that observed for bioactive Ti (BNT). This behaviour was rather unexpected due to its Ni content and very thin titanium oxide layer, which theoretically should not promote apatite film formation to such an extent. For these reasons further in-depth analysis is required for a more complete explanation. Nevertheless, the kinetics of precipitation significantly changed when the media contained proteins, vitamins and antibiotics. The main purpose of using these media is that they contain proteins and other organic components, which may influence the kinetics of apatite film formation and reactivity with the surface. It is commonly known that cells communicate with the surface via proteins and proteinaceous film formation, and its impact on further apatite layer formation seems to be very important from a biological point of view. Observations of film growth indicated that for all of the samples, except for the reference negative, the formation of the apatite-like film was similar. All samples reacted in the same way when soaked in DMEM and osteogenic media with no statistically significant differences seen between them. However, a slightly higher concentration of both elements was observed when samples were soaked in osteogenic media. In this situation when proteinaceous fluids were used, a more pronounced role of titanium oxide on apatite film formation was observed. In general, the samples with the oxidized surfaces behaved better than reference ve with very thin oxide layers. Moreover, it was observed that Ni–Ti reacted rapidly with the media components forming a thick, well-defined layer. In this case, it can be assumed that Ni had a positive effect on the formation of the apatite layer and reaction with complex fluid matrix. This also demonstrated that the organic component of the media altered the reactivity of the surface. It is assumed that very fast interaction of proteins (nitrogen deposition was observed after 5 min of immersing and increases with immersion time) and other organic components resulted in thin organic layers that covered the surface. Further, the slower reaction of inorganic elements of the media for all the samples was very similar. In fact, the organic layer encouraged precipitation of Ca and P and the Ca:P ratio was higher and closer to the ratio of hydroxyapatite compared to samples immersed in HBSS. It can also be concluded that proteins in body fluids can attach easily to the implant surface, and functionalize and assist formation of hydroxyapatite irrespective of the initial surface preparation. Despite the fact that Ni is an essential element in our body, the maximum daily intake should not exceed 200– 300 lg [2,34]. For this reason, it is very important to evaluate the amount of this element from the samples after different treatments. In addition, titanium ion release was evaluated. For two types of samples, alkali-treated and

1983

spark-oxidized, the total amount of released elements was significantly greater than for the other samples. The explanation of this fact is that the content of Ni was significantly greater on alkali-treated samples. The treatment introduced Ni to the surface, and etched out Ti. In addition, in both types of samples, this element was highly oxidized. This form of Ni is unstable and prone to decomposition in body fluids. From the ion release study, it can be seen that the three types of surface treatment increased Ni2+ ion release: alkali treatment, spark oxidation and heat treatment at 600 °C. It could be suggested that the highly oxidized layers of Ni–Ti alloy possess high potency for ion release, which could translate into ability to dissolve. This, however, is true only when Ni–Ti samples were considered. No increase of Ti4+ release was observed for reference samples, especially notable for the Ti samples after alkali treatment. Finally, it can be stated that Ni2+ ion release depends on the oxidation state of the element and amount of the oxide exposed to the media. Moreover, when the size of possible implant (wire, pins and clamps) was taken into consideration, it was assumed that the maximum recommended daily dose of Ni could be exceeded when the surface was prepared via alkali treatment. It showed that these types of treatments cannot be considered as useful for biomedical application due to the risk of high local concentrations of Ni in tissues. 6. Conclusions  Thin protein layers created on the surface very quickly after contact with DMEM and osteogenic media enhanced bioactive properties of all tested samples irrespective of the initial surface treatment.  Alkali treatment and spark oxidation (BNT, SP) produced an increase in roughness Ra and enhanced the surface with Ca and Na.  Thermal treatment at 400 °C caused a significant increase of transformation temperatures further temperature increase caused a reduction in these temperatures.  Alkali treatment and spark oxidation ensure very high SFE comparable to reference +ve (BTi).  The highest rate of Ca and P precipitation from HBSS was observed for ground Ni–Ti in a similar manner to reference +ve.  Bioactivity study in DMEM and osteogenic media showed that organic components changed the rate of Ca and P precipitation, which was at the same level for all tested samples.  Nickel ion release was significantly increased by two types of treatments: alkali treatment followed by heat treatment (BNT) and spark oxidation SP.  Ni–Ti was found to be highly bioactive with a simple ground, macroscopically smooth surface.  All applied surface treatments failed to enhance bioactivity of the Ni–Ti alloy in the initial phase of contact with simulated body media.

1984

W. Chrzanowski et al. / Acta Biomaterialia 4 (2008) 1969–1984

Acknowledgement This work was supported by EU FP6, IEF Marie Curie Action. References [1] Welton JL, Gwynn I, Richards RG. In vivo evaluation of polished titanium internal fixator surfaces to prevent soft tissue adhesion. In: Proceedings of the European society for biomaterials 2007; 2007. p. 185. [2] Ryha¨nen J. Biocompatibility evaluation of nickel–titanium shape memory alloy. Oulu: OulunYliopisto; 1999. [3] Wei DQ, Zhou Y, Jia DC, Wang YM. Characteristic and in vitro bioactivity of a microarc-oxidized TiO2-based coating after chemical treatment. Acta Biomater 2007;3(5):817–27. [4] Gu YW, Tay BY, Lim CS, Yong MS. Characterization of bioactive surface oxidation layer on NiTi alloy. Appl Surf Sci 2005;252(5):2038–49. [5] Tan L, Dodd RA, Crone WC. Corrosion and wear-corrosion behavior of NiTi modified by plasma source ion implantation. Biomaterials 2003;24:3931–9. [6] Vandenkerckhove R, Chandrasekaran M, Vermaut P, Portier R, Delaey L. Corrosion behaviour of a superelastic Ni–Ti alloy. Mater Sci Eng A: Struct Mater Proper Microstruct Process 2004;378(1–2):532–6. [7] Kawakita J, Stratmann M, Hassel AW. High voltage pulse anodization of a NiTi shape memory alloy. J Electrochem Soc 2007;154(6):C294–8. [8] Fujibayashi S, Neo M, Kim H-M, Kokubo T, Nakamura T. Osteoinduction of porous bioactive titanium metal. Biomaterials 2004;25:443–50. [9] Sul YT, Jeong Y, Johansson C, Albrektsson T. Oxidized, bioactive implants are rapidly and strongly integrated in bone. Part 1: Experimental implants. Clin Oral Implant Res 2006;17(5):521–6. [10] Armitage DA, Chrzanowski W. Surface preparation of Ni–Ti alloy using alkali, thermal treatments and spark oxidation. In: Proceedings of the European society for biomaterials 2007; 2007. [11] Chen MF, Yang XJ, Liu Y, Zhu SL, Cui ZD, Man HC. Study on the formation of an apatite layer on NiTi shape memory alloy using a chemical treatment method. Surf Coat Technol 2003;173:229–34. [12] Cheng Y, Zheng YF. Surface characterization and electrochemical studies of biomedical NiTi alloy coated with TiN by PIIID. Mater Sci Eng A 2006;438–440:1146–9. [13] Choi J, Bogdanski D, Koller M, Esenwein SA, Muller D, Muhr G, et al. Calcium phosphate coating of nickel–titanium shape-memory alloys. Coating procedure and adherence of leukocytes and platelets. Biomaterials 2003;24:3689–96. [14] Garcia C, Cere S, Duran A. Bioactive coatings deposited on titanium alloys. J Non-Cryst Solids 2006;352(32–35):3488–95. [15] Gu YW, Tay BY, Lim CS, Yong MS. Biomimetic deposition of apatite coating on surface-modified NiTi alloy. Biomaterials 2005:6916–23.

[16] Marciniak J, Chrzanowski W, Zak J, Rajchel B. Structure modification of surface layers of Ti6Al4V ELI implants. Key Eng Mater 2007;254–256:387–90. [17] Narayanan R, Seshadri SK. Phosphoric acid anodization of Ti–6Al– 4V—structural and corrosion aspects. Corros Sci 2007;49(2):542–58. [18] Shevchenko N, Pham M-T, Maitz MF. Studies of surface modified NiTi alloy. Appl Surf Sci 2004;235:126–31. [19] Wen HB, de Wijn JR, Cui FZ, de Groo K. Preparation of bioactive Ti6Al4V surfaces by a simple methods. Biomaterials 1998;19:215–21. [20] Chen MF, Yang XJ, Hu RX, Cui ZD, Man HC. Bioactive NiTi shape memory alloy used as bone bonding implants. Mater Sci Eng C 2004;24(4):497–502. [21] Kokubo T, Miyaji F, Kim HM. Spontaneous formation of bonelike apatite layer on chemically treated titanium metals. J Am Ceram Soc 1996;79:1127–9. [22] Kokubo T. Apatite formation on surfaces of ceramics, metals and polymers in body environment. Acta Mater 1998;46(7):2519–27. [23] Kokubo T. Design of bioactive bone substitutes based on biomineralization process. Mater Sci Eng C 2005;25:97–104. [24] Neel EAA, Mizoguchi T, Ito M, Bitar M, Salih V, Knowles JC. In vitro bioactivity and gene expression by cells cultured on titanium dioxide doped phosphate-based glasses. Biomaterials 2007;28(19): 2967–77. [25] Harle J, Kim HW, Mordan N, Knowles JC, Salih V. Initial responses of human osteoblasts to sol–gel modified titanium with hydroxyapatite and titania composition. Acta Biomater 2006;2(5):547–56. [26] Nebe B, Finke B, Luthen F, Bergemann C, Schroder K, Rychly J, et al. Improved initial osteoblast functions on amino-functionalized titanium surfaces. Biomol Eng 2007;24(5):447–54. [27] Bertheville B. Porous single-phase NiTi processed under Ca reducing vapor for use as a bone graft substitute. Biomaterials 2006;27:1246–50. [28] Chrzanowski W, Szewczenko J, Tyrlik-Held J, Marciniak J, Zak J. Influence of the anodic oxidation on the physicochemical properties of the Ti6Al4V ELI alloy. J Mater Process Technol 2005;162– 163:163–8. [29] Filip P, Lausma J, Musielak J, Mazanec K. Structure and surface of TiNi human implants. Biomaterials 2001;22:2131–8. [30] Hulbert SF, Morrison SJ, Klawitter JJ. Tissue reaction to three ceramics of porous and non-porous structures. J Biomed Mater Res 1976;6:347–74. [31] Takeuchi K, Saruwatari L, Nakamura HK, Yang J-M, Ogawa T. Enhanced intrinsic biomechanical properties of osteoblastic mineralized tissue on roughened titanium surface. Wiley–Interscience; 2005. doi:10.1002/jbm.a.30227. [32] KSV Instruments Ltd. Surface free energy – background, calculation and examples by using contact angle measurements. KSV Instruments Ltd., Application Note #108.2007. [33] Chrzanowski W, Neel EAA, Armitage DA, Lee K, Walke W, Knowles JC. Nanomechanical evaluation of nickel–titanium surface properties after alkali and electrochemical treatments. J R Soc Interf., in press. doi:10.1098/rsif.2007.1313. [34] Barrett RD, Bishara SE, Quinn JK. Biodegradation of orthodontic appliances. 1: Biodegradation of nickel and chromium invitro. Am J Orthod Dentofac Orthop 1993;103(1):8–14.