Fatigue Property and Cytocompatibility of a Biomedical Co–Cr–Mo Alloy Subjected to a High Pressure Torsion and a Subsequent Short Time Annealing

Abstract

In the present study, we evaluated the effects of high pressure torsion (HPT) and subsequent short time annealing processing on fatigue properties and cytocompatibility of the biomedical Co–Cr–Mo alloy (CCM). Before processing, CCM was solution treated (CCM_ST) to achieve a microstructure composed of coarse single γ-phase equiaxed grains with no internal strain. Through HPT processing, an inhomogeneous microstructure containing both micro- and nano-scaled grains is obtained in CCM specimens, which were named as CCM_HPT, accompanied by high internal strain and extensive ε martensite. Following a subsequent short time annealing, a uniform single γ-phase ultrafine-grained microstructure with small local strain fields dispersed forms in CCM specimens, which were named as CCM_HPTA. This microstructure change improves fatigue strength in CCM_HPT, and further in CCM_HPTA, because of the enhanced crack initiation and/or propagation resistance. For cytocompatibility evaluation, the cells cultured on CCM_ST show an immobilization tendency, while those cultured on CCM_HPT exhibit a locomotion tendency. The cells cultured on CCM_HPTA have an intermediate pattern. Compared with CCM_ST, much larger numbers of cells are proliferated in both CCM_HPT and CCM_HPTA. All these results demonstrate that the CCM_HPTA offers an improved fatigue property and a good cytocompatibility. Therefore, it is promising for use in biomedical applications.

Fig. 3 EBSD results of CCM_ST,¹³⁾ CCM_HPT, and CCM_HPTA.

1. Introduction

Because Co–Cr–Mo alloys exhibit good mechanical properties, great biocompatibility, as well as excellent wear resistance, they have been commercially applied as implant materials in abundant biomedical applications such as artificial dental wires and hip and knee joints.^1–3) However, mechanical failure still occurs sometimes in the practical uses of the Co–Cr–Mo alloys, and the wear debris of these alloys, which induce metallosis, might lead to an allergy issue.⁴⁾ Therefore, it is required to further enhance the mechanical properties of the Co–Cr–Mo alloys to reduce their wear and failure risks. This improvement will benefit biomedical applications.

High pressure torsion (HPT) is one of the severe plastic deformation (SPD) methods, in which a high pressure and a large torsional straining are applied on a disk specimen so that the extremely high dislocation density and extremely large strain can be introduced into the material to facilitate significant grain refinement and/or phase transformations.^5–11) In a previous study,¹²⁾ HPT processing was adopted into a Co–Cr–Mo alloy, Co–28Cr–6Mo (mass%, hereafter abbreviated as CCM), at a rotation number of 0.25, a rotation speed of 1 rpm, and a high pressure of 6 GPa, to improve the mechanical properties by introducing significant grain refinement into nano-scale and extensive γ (face-centered cubic, FCC) to ε (hexagonal closed-packed, HCP) martensitic transformation. The mechanical strength of the CCM alloy was indeed significantly improved via nano-scaled grain refinement, large internal strain associated with high dislocation density, and hard ε martensitic transformation. However, all of these inevitably deteriorate the ductility of the CCM alloy because they play a strong effect on impeding the dislocation gliding and dislocation multiplication. In addition, the transformed ε martensite with an HCP structure having limited slip systems shows poor ductility which also contributes to the reduced ductility of the CCM alloy.¹²⁾ A short time annealing was carried out on this HPT processed CCM alloy (CCM_HPT) to improve the ductility of the CCM_HPT alloy while maintaining its high mechanical strength by properly removing the excessive hard ε martensite and releasing the large internal strain while maintaining the ultrafine-grained microstructure.¹³⁾ The CCM_HPT alloy subjected to the short time annealing at 1273 K for a duration of 0.3 ks (CCM_HPTA) indeed exhibits a recovered elongation of ∼12% with a maintained high tensile strength of ∼1600 MPa compared with the CCM_HPT alloy (elongation: ∼1%, tensile strength: ∼1700 MPa) and the solution treated CCM alloy (CCM_ST, elongation: ∼15%, tensile strength: ∼1000 MPa) as illustrated in Fig. 1.¹³⁾ However, the fatigue properties and cytocompatibility, both of which are extremely crucial properties for the CCM_HPTA alloy to be practically used in biological applications, have still not been investigated in detail. In this study, the fatigue properties and cytocompatibility of the CCM_ST, CCM_HPT, as well as CCM_HPTA alloys were comparatively investigated to evaluate the applicability of the CCM_HPTA alloy in biological uses. The relationship between processing conditions, including HPT and short time annealing, microstructure, and properties including fatigue properties and cytocompatibility of the CCM alloy was discussed.

Fig. 1

Tensile properties of CCM_ST, CCM_HPT, and CCM_HPTA.¹³⁾

2. Experimental Procedures

2.1 Materials preparation

A hot-forged CCM alloy rod having a diameter of 25 mm was applied in the present study. The chemical composition of the alloy is the same with that shown in the previous study.¹³⁾ The hot-forged CCM rod was firstly subjected to a solution treatment at 1473 K for a duration of ∼3.6 ks in vacuum (CCM_ST). The obtained CCM_ST rod was then cut into disk-shaped specimens with the diameter of 10 mm and the thickness of 1 mm. A HPT processing was conducted on the CCM_ST disk-shaped specimens under a quasi-constrained condition at a rotation number of 0.25, a rotation speed of 1 rpm, and a high pressure of 6 GPa at room temperature (CCM_HPT). The CCM_HPT specimens were subsequently subjected to a short time annealing at 1273 K for a duration of 0.3 ks in vacuum (CCM_HPTA).

2.2 Microstructural analysis

The microstructure of the CCM_ST, CCM_HPT, and CCM_HPTA specimens was investigated using an electron backscatter diffraction (EBSD) at the designated position as schematically illustrated in Fig. 2(a). The specimens were subjected to a mechanical polishing using the SiC waterproof emery papers up to 2400 grits, and then a mirror polishing using the colloidal SiO₂ suspension.

Fig. 2

Schematic illustrations of specimen preparations for (a) EBSD analysis and (b) fatigue tests.

2.3 Fatigue property evaluation

The fatigue specimens with dimensions based on the standard (ASTM E466) were prepared from the CCM_ST, CCM_HPT, and CCM_HPTA specimens, as schematically illustrated in Fig. 2(b). The specimen surfaces were mechanically polished using the SiC waterproof emery papers, then buff polished to a mirror finishing using the colloidal SiO₂ suspension. The mirror-polished specimens were subsequently subjected to the fatigue tests with a frequency of 10 Hz and a stress ratio of 0.1, under the tension-tension mode at room temperature using an electro-servo-machine. In this study, the fatigue limit was defined as the maximum cyclic stress, at which the specimen still not fatigue fractured under 10⁷ cycles. After fatigue failure, the fracture surface morphologies of the specimens were inspected by a scanning electron microscope (SEM). Additionally, the specimens before and after fatigue tests were subjected to an X-ray diffraction (XRD) analysis to quantitatively clarify the volume fraction of the ε martensite in CCM specimens.

2.4 Cytocompatibility evaluation

2.4.1 Cell culture

A mouse preosteoblast cell (MC3T3-E1), was purchased from RIKEN BioResource Center and used directly as described previously.¹⁴⁾ Briefly, cells were maintained in an alpha modified Eagle’s minimum essential medium (α-MEM; Gibco, Carlsbad, CA) supplemented with 10% fetal bovine serum (Gibco) and an antibiotic/antimycotic (Gibco). Before cell seeding, all CCM specimens were immersed in deionized water over 1 week to form a stable oxide surface layer. Then 70% ethanol was used to sterilize all specimens for 20 min and rinsed thoroughly with deionized water, following which cells were seeded onto the specimens with seeding density of 6000 cells·cm⁻². Cells were cultured at 37°C in a humidified atmosphere of 5% CO₂ in air.

2.4.2 Attached cell counts

According to manufacturer instructions, the attached cell number on each specimen was counted using Cell Counting Kit-8 (Dojindo Laboratories, Kumamoto, Japan), which also described elsewhere.¹⁵⁾ Because the cell number and absorbance at 450 nm showed a linear relationship, the attached cell number was calculated based on a prepared standard curve.

2.4.3 Fluorescent staining and imaging

The morphologies of cells on each specimen were visualized by immunofluorescence staining after 6 h incubation, as described in our previous report.¹⁶⁾ Briefly, cells were fixed and blocked before staining. Cellular F-actin was stained with rhodamine phalloidin (Cytoskeleton Inc., Denver, CO), and nuclei were counterstained with 4′,6-diamidino-2-phenylindole (DAPI, Invitrogen, Carlsbad, CA). Adhesion plaques were stained with labeling of vinculin using a monoclonal anti-vinculin antibody (Sigma-Aldrich) and then conjugated with Alexa Fluor 488-conjugated goat anti-mouse IgG (Invitrogen). IX71 microscope with DP70 charge-coupled device camera (Olympus, Tokyo, Japan) was used to acquire digital images.

2.4.4 Statistical analysis

The results of cell proliferation were repeated three independent times ($\text{n} \geqslant 3$) and analyzed by ANOVA followed by a S-N-K test using SPSS. A value of p value <0.05 was considered statistically significant.

3. Results and Discussion

3.1 Microstructure

Figure 3 shows the EBSD micrographs of the CCM_ST, CCM_HPT, and CCM_HPTA specimens. The CCM_ST specimen¹³⁾ shows a microstructure composed of single γ-phase (FCC) equiaxed grains having an average grain diameter of ∼70 µm. Annealing twins are extensively identified in the interior of the grains in the CCM_ST specimen, which is owing to the low stacking fault energy. The corresponding kernel average misorientation (KAM) map shows that almost no internal strain is detected in the CCM_ST specimen, which is because of the stress-released solution treatment. After HPT, deformation-induced γ to ε (HCP) martensitic transformation extensively occurred in the CCM_HPT specimen. A large number of black areas, where the crystallographic information is unable to be identified by EBSD because of the existence of nanoscale-grained microstructure and/or the high internal strain, are observed in the CCM_HPT specimen. These results suggest that an inhomogeneous microstructure containing both micro-scaled grains and nano-scaled grains formed in the CCM_HPT specimen. The KAM map indicates that the CCM_HPT specimen shows high internal strain induced by severe plastic deformation throughout the entire specimen. When subjecting the CCM_HPT specimen to a short time annealing, the excessive ε martensite was mostly removed, and a uniform single γ-phase ultrafine-grained microstructure having an average grain diameter of ∼6.5 µm formed in the CCM_HPTA specimen. The KAM map reveals that the internal strain in the CCM_HPTA specimen was released significantly compared with that in the CCM_HPT specimen, but still leaves small local strain fields dispersed throughout the CCM_HPTA specimen. These results match well with those reported in previous works.^11–13)

Fig. 3

EBSD results of CCM_ST,¹³⁾ CCM_HPT, and CCM_HPTA.

Figure 4 shows the S-N curves of the CCM_ST, CCM_HPT, and CCM_HPTA specimens, in which the maximum cyclic stress are plotted as a function of the number of cycles to failure. The CCM_ST specimen shows the lowest fatigue strength, and the lowest fatigue limit of ∼400 MPa among the examined specimens. After HPT, the CCM_HPT specimen exhibits higher fatigue strength compared to the CCM_ST specimen in both low-cycle fatigue-life region where failure occurs below 10⁵ cycles, and high-cycle fatigue-life region, in which the number of cycles to failure exceeds 10⁵ cycles; the fatigue limit of the CCM_HPT specimen is thus improved to ∼520 MPa. When the short time annealing was conducted following HPT processing, the fatigue strength is further improved in the CCM_HPTA specimen in both low-cycle and high-cycle fatigue-life region compared to that of the CCM_HPT specimen; the CCM_HPTA specimen shows the highest fatigue limit of ∼630 MPa among the examined specimens.

Fig. 4

S-N curves of CCM_ST, CCM_HPT, and CCM_HPTA.

3.2 Fatigue properties

The fatigue fractographs of the CCM_ST, CCM_HPT, and CCM_HPTA specimens fractured in the high-cycle fatigue-life region were inspected by SEM (the fractured specimens chosen for observations are marked by circles in Fig. 4); the corresponding results are demonstrated in Fig. 5. Because the fatigue fractographs of the specimens fractured in the low-cycle fatigue-life region show similar morphologies to those of the specimens fractured in the high-cycle fatigue-life region, only the results of the latter are exhibited in the present study. In the CCM_ST specimen, two crack initiation sites, indicated by the red arrows in Fig. 5(a), are identified near the specimen surface, which suggests that the fatigue cracks initiated on the specimen surface in the CCM_ST specimen during fatigue tests. Facets with a facet width of ∼25 µm are identified in the fatigue crack propagation area in the CCM_ST specimen, as marked by the rectangle in Fig. 5(b). Furthermore, the striations can be clearly observed on the observed facets, as shown in Fig. 5(c). In the CCM_HPT specimen, several crack initiation sites are identified near the specimen surface, as pointed out by the red arrows in Fig. 5(d). Facets are also detected in the fatigue crack propagation area in the CCM_HPT specimen, as marked by the rectangle in Fig. 5(e), whereas the width of the facets observed in the CCM_HPT specimen is smaller than that in the CCM_ST specimen. Moreover, the striations are also visible in the CCM_HPT specimen, as shown in Fig. 5(f). It is thus known that both of the CCM_ST and CCM_HPT specimens show facet and striation morphologies in the fatigue fracture surfaces, whereas the CCM_HPTA specimen exhibits a completely different fatigue fracture morphology. In the CCM_HPTA specimen, only one crack initiation site is observed on the specimen surface, as indicated by the red arrow in Fig. 5(g). In the fatigue crack propagation area, neither facets nor striations, but only small dimples are observed in the CCM_HPTA specimen, as shown in Figs. 5(h) and 5(i).

Fig. 5

SEM fractographs of CCM_ST, CCM_HPT, and CCM_HPTA tested in high-cycle fatigue-life region.

As shown by the above results, the fatigue cracks initiated on the specimen surfaces in all the three investigated specimens. The CCM_HPTA specimen shows the highest crack initiation resistance because it shows the least crack initiation sites even though it is under the highest cyclic stress. According to Fig. 3, the CCM_ST specimen shows a microstructure consisting of coarse single-γ-phase equiaxed grains with no internal strain, while the CCM_HPT specimen has an inhomogeneous microstructure composed of micro-scaled γ grains, ε martensite and local areas containing nano-scaled grains, and/or extremely high internal strain. The increased number of crack initiation sites in the CCM_HPT specimen compared to the CCM_ST specimen is thus considered to be attributed to the applied high cyclic stress and/or the inhomogeneous microstructure in which the stress concentration can easily occur in the border of the hard brittle regions and soft ductile regions, thus causing crack initiations. In addition, the CCM_HPTA specimen shows a uniform single γ-phase ultrafine-grained microstructure with small local strain fields uniformly distributed. This homogeneous ultrafine microstructure effectively disperses the stress uniformly throughout the entire specimen and reduces the risk of the occurrence of stress concentration, thereby improving the crack initiation resistance in the CCM_HPTA specimen.

In the fatigue crack propagation area, facets and striations can be observed in both the CCM_ST and CCM_HPT specimens. It is likely that the facets formed because the fracture occurred along the phase boundaries between γ phases and ε martensite lamellar. Table 1 lists the calculated volume fraction of the ε martensite in the CCM_ST, CCM_HPT, and CCM_HPTA specimens before and after high cycle fatigue tests based on the XRD analysis (The calculation procedure is the same as that reported in previous works¹¹⁾ and not shown here). After fatigue tests, it is observed that the volume fraction of the ε martensite significantly increased from 0% to 49% in the CCM_ST specimen. However, no obvious variation in ε martensite volume fraction can be identified in either the CCM_HPT specimen (from 87% to 89%), which is because of the large volume fraction of ε martensite originally formed before fatigue tests, or the CCM_HPTA specimen (from 2% to 5%), which is considered to be attributed to the ultrafine-grained microstructure that may stabilize the γ phase and increase the resistance for γ to ε martensitic transformation.^17,18) These results reveal that the stress-induced martensitic transformation occurred in the CCM_ST specimen during fatigue cyclic loading, which explains the facet formation in the fracture surface of the CCM_ST specimen. The volume fraction of the ε martensite is much larger in the CCM_HPT specimen compared to that in the CCM_ST specimen, even after fatigue tests. Therefore, the more martensite lamellar in the CCM_HPT specimen provides more barriers and more complicated paths for fatigue cracks to propagate, thereby enhancing the fatigue crack propagation resistance in the CCM_HPT specimen. In the CCM_HPTA specimen, because almost no ε martensite formed before or after fatigue tests, no facet can be identified in the fracture surface. The dimple-type fracture morphology suggests that the uniform ultrafine-grained microstructure with uniformly distributed small local strain fields in the CCM_HPTA specimen seems to provide a high resistance for the fatigue crack propagation because stress is difficult to concentrate within this microstructure.

Table 1 Volume fractions of ε phase in CCM_ST, CCM_HPT, and CCM_HPTA before fatigue tests and after high cycle fatigue tests based on XRD analyses.

3.3 Cytocompatibility

Cell adhesions on the CCM_ST, CCM_HPT, and CCM_HPTA specimens were visualized by immunofluorescence staining after 6 h incubation; the results are shown in Fig. 6. Spreading shapes showed with cytoskeleton staining are similar for all CCM specimens. Although it has been reported that HPT processed Ti alloy (TNTZ) promoted cell adhesion and migration by visualization of actin filaments,¹⁹⁾ in this study, the interactions between cells and CCM substrates were detected by visualization of adhesion plaques through vinculin staining.^20–22) Interestingly, for adhesion plaques vinculin, depending on the processing, differences in the distribution and shape were detected. Cells cultured on CCM_ST specimen and CCM_HPTA specimen present a fibrous structure at the margin of the pseudopodia, compared with those cultured on the CCM_HPT specimen. By contrast, cells growing on the CCM_HPT specimen and the CCM_HPTA specimen display a pointed distribution of adhesion plaques in cells. As known the role of vinculin during cell adhesion behavior,^20,21) the shapes of cells cultured on the CCM_HPT specimen indicate a stronger cell locomotion trend, on the other hand, those on the CCM_ST specimen present a stronger immobilization trend. Cells cultured on the CCM_HPTA specimen show an intermediate pattern. As reported previously that nano-scaled grains (∼100 nm) promoted cellular locomotion,²⁰⁾ cells cultured on the CCM_HPT specimen show a locomotion trend compared with those cultured on CCM specimens with micro-scaled grains, where CCM_ST exhibits an average grain diameter of ∼70 µm, and CCM_HPTA has an average grain diameter of ∼6.5 µm (Fig. 3). However, the different cellular adhesion behaviors of the CCM_ST and CCM_HPTA specimens were detected, which indicates that the uniformly distributed small local strain fields or newly formed grain boundaries in CCM_HPTA could have an effect on the promotion of cell locomotion.

Fig. 6

Initial cellular morphologies of MC3T3-E1 attached to CCM_ST, CCM_HPT, and CCM_HPTA after a 6-h incubation. With a fluorescent staining, cellular nuclei, F-actin, and vinculin were visualized with blue, red, and green, respectively.

Cell proliferation was evaluated with all CCM specimens for 5 days, as shown in Fig. 7. With the increasing incubation time, the cell numbers increase for all specimens. No significant difference was observed in cell attachment number among all the examined CCM specimens, after 1-d culture. However, more cells were harvested by the CCM_HPT specimen than those by the CCM_ST or CCM_HPTA specimens were on the second and third days. After 4 to 5 days’ culture, a significantly lower number of cells are observed on the CCM_ST specimen compared with those on both the CCM_HPT and CCM_HPTA specimens. These results indicate that compared with CCM with a larger grain diameter (CCM_ST), the CCM with a smaller grain diameter corresponding to more newly formed grain boundaries and higher internal strain, which relates to more generated defects such as dislocations (CCM_HPT and CCM_HPTA), are beneficial to cell proliferation. In addition, it is worth noting that, compared with the CCM_HPTA specimen, the CCM_HPT specimen with nano-scaled grains, significantly higher internal strain, and extensive ε martensite lamellar showed faster cell growth at the initial cultivation (2∼3 days).

Fig. 7

Cellular proliferation of MC3T3-E1 cultured on CCM_ST, CCM_HPT, and CCM_HPTA. The “*” represents statistically significant, and the “n.s.” represents non-significance.

4. Conclusions

Herein, effects of HPT and subsequent short time annealing processing on fatigue properties and cytocompatibility of the CCM alloy were investigated. Before processing, CCM specimens were solution treated (CCM_ST) to achieve a microstructure composed of coarse single γ-phase equiaxed grains. After HPT, CCM_HPT shows an inhomogeneous microstructure containing both micro- and nano-scaled grains, high internal strain, and extensive ε martensite; by contrast, a uniform single γ-phase ultrafine-grained microstructure with dispersed small local strain fields forms after a subsequent short time annealing in CCM_HPTA. This microstructure change improves fatigue strength in CCM_HPT and further in CCM_HPTA because of the enhanced crack initiation and/or propagation resistance. For cell adhesion behavior, a locomotion trend and an immobilization trend are presented by cells cultured on CCM_HPT and cells cultured on CCM_ST, respectively. The cells on CCM_HPTA show an intermediate pattern. For cell proliferation behavior, after 4 days culture, a similar much larger number of cells are harvested by both CCM_HPT and CCM_HPTA in comparison with that harvested by CCM_ST. Our results indicate that the CCM with an HPT processing followed by a short time annealing (CCM_HPTA) shows an improved fatigue property and a good cytocompatibility, which could be a promising material applied for biomedical using.

Acknowledgment

This work was supported by Creation of Life Innovation Materials for Interdisciplinary and International Researcher Development (iLIM) from the Ministry of Education, Culture, Sports, Science and Technology (MEXT), Japan.

REFERENCES

• 1)   S.H.  Sun,  Y.  Koizumi,  S.  Kurosu,  Y.P.  Li,  H.  Matsumoto and  A.  Chiba: Acta Mater. 64 (2014) 154–168.
• 2)   K.  Yamanaka,  M.  Mori and  A.  Chiba: Acta Biomater. 9 (2013) 6259–6267.
• 3)   A.  Chiba,  K.  Kumagai,  N.  Nomura and  S.  Miyakawa: Acta Mater. 55 (2007) 1309–1318.
• 4)   Y.  Koizumi,  S.  Suzuki,  K.  Yamanaka,  B.S.  Lee,  K.  Sato,  Y.  Li,  S.  Kurosu,  H.  Matsumoto and  A.  Chiba: Acta Mater. 61 (2013) 1648–1661.
• 5)   A.P.  Zhilyaev and  T.G.  Langdon: Prog. Mater. Sci. 53 (2008) 893–979.
• 6)   H.  Yilmazer,  M.  Niinomi,  K.  Cho,  M.  Nakai,  J.  Hieda,  S.  Sato and  Y.  Todaka: Acta Mater. 80 (2014) 172–182.
• 7)   R.B.  Figueiredo and  T.G.  Langdon: Mater. Sci. Eng. A 528 (2011) 4500–4506.
• 8)   S.  Scheriau,  Z.  Zhang,  S.  Kleber and  R.  Pippan: Mater. Sci. Eng. A 528 (2011) 2776–2786.
• 9)   M.  Kawasaki,  B.  Ahn and  T.G.  Langdon: Acta Mater. 58 (2010) 919–930.
• 10)   Y.  Ito and  Z.  Horita: Mater. Sci. Eng. A 503 (2009) 32–36.
• 11)   M.  Isik,  M.  Niinomi,  H.H.  Liu,  K.  Cho,  M.  Nakai,  Z.  Horita,  S.  Sato,  T.  Narushima,  H.  Yilmazer and  M.  Nagasako: Mater. Trans. 57 (2016) 1109–1118.
• 12)   M.  Isik,  M.  Niinomi,  K.  Cho,  M.  Nakai,  H.H.  Liu,  H.  Yilmazer,  Z.  Horita,  S.  Sato and  T.  Narushima: J. Mech. Behav. Biomed. Mater. 59 (2016) 226–235.
• 13)   M.  Isik,  M.  Niinomi,  H.H.  Liu,  K.  Cho,  M.  Nakai,  Z.  Horita,  T.  Narushima and  K.  Ueda: Mater. Trans. 57 (2016) 1887–1896.
• 14)   P.  Chen,  A.  Nagai,  Y.  Tsutsumi,  M.  Ashida,  H.  Doi and  T.  Hanawa: J. Biomed. Mater. Res. A 104 (2016) 639–651.
• 15)   P.  Chen,  M.  Miyake,  M.  Tsukamoto,  Y.  Tsutsumi and  T.  Hanawa: J. Biomed. Mater. Res. A 105 (2017) 3456–3464.
• 16)   P.  Chen,  T.  Aso,  R.  Sasaki,  M.  Ashida,  Y.  Tsutsumi,  H.  Doi and  T.  Hanawa: J. Biomed. Mater. Res. A 106 (2018) 2735–2743.
• 17)   E.A.  Owen and  D.M.  Jones: Proc. Phys. Soc. B 67 (1954) 456–466.
• 18)   J.Y.  Huang,  Y.K.  Wu,  H.Q.  Ye and  K.  Lu: Nanostruct. Mater. 6 (1995) 723–726.
• 19)   H.  Yilmazer,  M.  Sen,  M.  Niinomi,  M.  Nakai,  H.H.  Liu,  K.  Cho,  Y.  Todaka,  H.  Shiku and  T.  Matsue: RSC Advances 6 (2016) 7426–7430.
• 20)   P.  Chen,  M.  Ashida,  H.  Doi,  Y.  Tsutsumi,  Z.  Horita and  T.  Hanawa: Mater. Trans. 57 (2016) 2020–2025.
• 21)   P.  Chen,  T.  Aso,  R.  Sasaki,  Y.  Tsutsumi,  M.  Ashida,  H.  Doi and  T.  Hanawa: J. Biomed. Nanotechnol. 13 (2017) 324–336.
• 22)   J.L.  Coll,  A.  Ben-Ze’ev,  R.M.  Ezzell,  J.L.  Rodríguez Fernández,  H.  Baribault,  R.G.  Oshima and  E.D.  Adamson: Proc. Natl. Acad. Sci. USA 92 (1995) 9161–9165.