The continuous development of the technology and medicine leads to newer and a better solutions for the existing health problems. Both doctors and engineers are trying to develop new treatment methods, enabling shortening of recovery periods, less invasive surgeries and overall improvement of patients’ life comfort. Human tissues have limited ability to regenerate, which declines with increasing age [1
]. Therefore, we can observe huge interest in tissue engineering (TE): the sub-field of biomaterials that combines the life and materials sciences and engineering to create a functional replacements of damaged tissues or organs [2
]. In one of the TE approaches, cells are seeded on a three-dimensional (3D) substrate that acts as a scaffold for cells growth and proliferation. Some cell types, e.g., mesenchymal stem cells (MSC), can be further stimulated to differentiate into specified tissues by appropriate mechanical or chemical stimuli [3
]. Materials used for fabrication of the 3D scaffolds should meet a number of requirements, which include: biocompatibility, adequate strength, surface physicochemical properties, thermal and electrical conductivity, and processability. Nowadays, the most investigated 3D structures are scaffolds for TE of bone, cartilage, and tendon [5
]. Scaffolds envisioned for substitution of bone defects should exhibit relatively high mechanical properties to match those of bone and, at the same time, high open porosity to ensure bone and vasculature ingrowth [8
]. Those two features cancel each other out and require utilization of materials with high bulk stiffness and strength, such as, for example, metals. By varying pore size, thus porosity, the mechanical properties can be further optimized to a specific application. For example, porous implants made of small-pore core and big-pore shell combine enhanced mechanical properties of the small-pore structures with stabilization of the implant by bone ingrown into the big-pore shell. The numerical optimizations of open-porous scaffolds showed that meeting mechanical properties of the custom bone defect needs fabrication of implants with porosity gradient [10
]. Moreover, the porosity gradient can be used to promote development of different tissues, for example bone and cartilage [12
The appropriate porosity of the metallic scaffold for bone TE is one of the most important properties and must be addressed during their development. It makes it possible to adjust Young’s modulus of the implant to that of a human bone, thus prevents the stress shielding effect in vivo
]. The interconnected system of pores is also required for cells and tissue ingrowth, transport of nutrients and metabolites, as well as vascularization. It was shown that open porosity is an important factor in osteogenesis [9
]. The optimal pore size was a subject of many studies. In general, it is believed that the minimal pores size required for bone tissue regeneration should be around 100 µm [13
]. An in vivo
study in dogs showed that implants with the pore sizes below 100 µm favored formation of the fibrous tissue or overgrowth with ossein. The same study reviled that the pore sizes above 150 µm favored bone tissue formation [14
]. Another in vivo
study showed that the increase of the pore size from 180 to 300 µm resulted in improved response of the bone tissue in rabbits [15
]. On the other hand, the highest cell proliferation was observed for the 200 µm pore sizes during in vitro
tests on a laser melted Ti-6Al-4V alloy scaffolds [16
]. Similarly, the study conducted by Stangl et al.
] showed that 200 µm pore sizes are the most preferred while 100 µm are the least preferred for in vitro
osteoblast proliferation on commercially pure (CP) titanium implants. In the studies of Warnke et al.
] the Ti-6Al-4V alloy scaffolds produced by SLM were filled by cells after 6 weeks in 100%, 44%, 10%, 0% for 500, 600, 700, 900 µm pores, respectively. In addition, the study conducted by Bael et al.
] showed that pore shape and size in the range from 500 to 1000 µm influenced the cell differentiation. The authors suggested fabrication of the functionally graded scaffolds to enhance cell seeding efficiency.
Till now, titanium functionally graded materials (FGM) were fabricated by conventional or pulsed sintering processes. However, those techniques did not ensured neither appropriate porosity nor control of implant shape [20
]. On the other hand, manufacturing of the 3D patient-specific metallic scaffold with controlled architecture is possible by additive manufacturing (AM) methods such as selective laser melting (SLM) or selective electron beam melting (SEBM) [22
]. In the AM, objects can be fabricated from elemental powders, based on the computer aided design (CAD) models. The models can be created using numerical algorithms [24
] or µ-CT imagining [25
]. Laser/electron printing in a powder bed provides a possibility to fabricate objects of any shape in one production step. However, it also carries some disadvantages. One drawback is the requirement to generate support for the fabricated parts. Such support should dissipate process heat generated during 3D printing from metallic powders and minimize geometrical distortions induced by internal stresses [26
]. The second disadvantage is the necessity of post-processing to improve the surface parameters, such as roughness, and to remove the unmelted powder particles [25
]. However, in the case of a cellular pore structure of a few hundred micrometers, post-processing cannot be performed by means of machining or vibro-abrasive machining. Removal of the unmelted particles can be done by chemical or electrochemical methods instead of the conventional post-processing. The chemical polishing of AM titanium scaffolds using hydrofluoric acid (HF) or hydrofluoric/nitric acid (HF/HNO3
) solutions was reported [29
]. The polishing effect depended on composition and concentration of the acid bath. In general, an increase in the HF concentration promoted formation of soluble compounds, while HNO3
enhanced titanium passivation [33
]. Another factors influencing the polishing process is structures’ cell geometry or employing an ultrasonic waves during cleaning. For example, polishing of scaffolds consisting of hexagon pores with size around 600 µm resulted in powder particles trapped within the structures after polishing [34
]. Contrarily, scaffolds composed of elemental cubes and pore size around 400–420 µm were successfully polished using ultrasonic cleaner [30
]. This clearly shows that development of novel architectures of porous scaffolds has to go hand in hand with optimization of polishing parameters. Another disadvantage of metal 3D printing are difficulties in predicting open porosity and obtaining struts size matching that of the CAD [16
]. The usage of protocols basing on µ-CT was proposed for better reproduction of the scaffold CAD model during SLM fabrication. Such protocol involves adjustment of the processing parameters based on extrapolation of data obtained by the µ-CT [36
]. However, this protocol was not checked for all elemental cell geometries, and its usage is pointless when further chemical polishing is needed.
In this study, we tried to meet the dimensions of the scaffolds CAD model by employing various chemical polishing baths. We have investigated the effect of the chemical polishing of titanium scaffolds with different pore sizes—200 µm, 500 µm and bimodal 200 + 500 µm on their morphology, porosity, mechanical properties and in vitro cell response. We described and discussed the influence of the chemical polishing of scaffolds with diamond cell structures on changes of mass, pore and strut size, as well as mechanical properties and the cells behavior. A detailed guideline was proposed for fabrication of the porous scaffolds by SLM and removal of the unmelted powders from their surface to achieve dimensional match to that of the CAD model. To the best of our knowledge, this is the first work studying the influence of polishing on the properties of the scaffolds with pore sizes bellow 400 µm or bimodal pore size distribution in the core and shell of a scaffold.
2. Materials and Methods
2.1. Scaffolds Modeling
Titanium scaffolds were designed in Magics (Materialiese NV, Leuven, Belgium) software using the structures toolbox. To generate 3D scaffold model a cylindrical model with the height of 4 mm and diameter 6 mm were filled with a diamond elemental structure (Figure 1
a). The size of the diamond cell was equal 0.47 mm and 0.87 mm to obtain construct with pore sizes 200 µm and 500 µm, respectively. Both designed scaffold models had 70% relative density, after filling them with the diamond structure. The bimodal scaffold (Figure 1
b) was generated by Boolean operations using pipes and cylinders filled with structures of different sizes.
2.2. Scaffolds Fabrication
The titanium scaffolds were fabricated from CP Ti powder (Starbond Ti4, Scheftner Dental Alloys GmbH, Mainz, Germany) on a SLM 50 3D printer (Realizer GmbH, Borchen, Germany) for metals and their alloys. According to the manufacturer’s data, the Ti4 powder had a diameter of 10–45 µm, met titanium Grade 4 requirements and its purity was minimal 98.95 wt %. (max impurities: 0.5% Fe, 0.4% O, 0.08% C, 0.05% N and 0.125% H). The powder size distribution before the process was investigated in ethanol using a laser scattered light analyzer Horiba LA-950 (Horiba Ltd., Kyoto, Japan). During the fabrication, an argon inert atmosphere was used and the oxygen value was kept in the volume in the range of 0.2%–0.4%. Three series of scaffolds with various designed pore sizes: (1) 200 µm, (2) 500 µm and (3) bimodal 200 µm core + 500 µm shell, were fabricated. During fabrication, the scaffolds were oriented on the edge to minimalize the support area (Figure 2
). The laser scanning speed and energy density delivered to the scaffolds for their consolidation were in the range of 325–375 mm/s and 91–151 J/mm3
, respectively. A layer thickness of 25 µm was used. The detailed process parameters are summarized in the Table 1
2.3. Chemical Polishing
Chemical polishing was performed in the hydrofluoric acid solutions [30
] (1%–5% HF) for time 1–6 min. and in mixtures of hydrofluoric and nitric acids [33
] (2.0/20%, 1.3/9.0%, 4.0/16%, 2.2/20% HF/HNO3
respectively) for 3–9 min. Each sample was polished in a separate beaker in an ultrasonic cleaner. Acid concentration and composition of the baths are summarized in the Table 2
and Table 3
. The powders trapped between scaffolds struts were removed from the scaffolds before chemical polishing by compressed air and multiple washings in a distilled water using an ultrasonic cleaner. The mass of the samples was measured before and after chemical polishing. The microscopic observations before and after polishing were performed using HITACHI SU8000 (Hitachi Ltd., Tokyo, Japan) and Phenom PROX microscopes (Phenom-World BV, Eindhoven, Holland). Based on the mass change and microscopic observations, the bath composition was optimized for each scaffold type. The characterization was performed on non-polished and scaffolds.
2.4. Contact Angle Measurements
Contact angle measurements were made using the Contact Angle System OCA (DataPhysics, Filderstadt, Germany). The measurements were performed on the thin (~1 mm) solid plates fabricated with the same process parameters and energy densities as delivered to the material for a porous scaffolds. The 9 measurements were performed on each chemically polished and non-polished material type 1 µL of distilled water was placed on the samples, photographed and analyzed using software delivered with the goniometer. Each measurement was performed 6 times.
2.5. Density Measurements and µ-CT Reconstruction
The SKYSCAN 1172 X-Ray µ-CT (Bruker, Billerica, MA, USA) was used for open/closed porosity, struts/pore size and surface area calculations using NRecon, DataViewer, CTAn, CTVol and CTVox software (Bruker, Billerica, MA, USA). The X-ray tube voltage was 100 kV and the current was 100 µA. The Al + Cu filter was used during measurements. The X-ray projections were obtained at 0.3° intervals with a scanning angular rotation of 180°. 8 frames were averaged for each rotation. The exposure time was 1160 ms and pixel size was set on 4.90 µm. Additionally, ring artifacts were reduced through selection of a random movement amplitude of 50. The open/closed porosity and struts/pore size µ-CT results were compared to ImageJ measurements on SEM pictures which were made using the Hitachi SU8000 (Hitachi Ltd., Tokyo, Japan).
2.6. Mechanical Tests
The compression tests were performed using the electromechanical testing machine Zwick Roell Z005 (Zwick GmbH & Co. KG, Ulm, Germany). 5 chemically polished and 5 non-polished scaffolds with different porosities were tested. The used strain rate was 10−3 1/s, which corresponds to the crossbar travel speed of 0.004 mm/s. Young’s modulus was calculated as the slope of the initial linear range of the stress-strain curve. Strain measurements were done by using non-contact digital image correlation method.
2.7. In Vitro Cell Response
Samples were sterilized in 70% ethanol for 1 h, followed by four washes with phosphate buffered saline (PBS; Sigma, Steinheim, Germany) and overnight incubation in an expansion medium (minimum essential medium alpha, α-MEM; Gibco, Thermo Fisher Scientific, Paisley, UK) supplemented with 10% FBS (fetal bovine serum; South American Origin, Biowest, Nuaillé, France), 1% PSN (5 units/ml of penicillin, 5 mg/ml of streptomycin, 10 mg/ml neomycin; Sigma, St. Louis, MO, USA) and 1 ng/mL human basic fibroblast growth factor (Sigma, Jerusalem, Israel). Normal human bone marrow derived mesenchymal stem cells (hMSCs) were purchased from Lonza (Walkersville, MD, USA). The cells were isolated from bone marrow of a 40-year-old female. The cells used in the experiments were from passage 4. MSCs were seeded at a density of 2.5 × 105 per scaffold and incubated in expansion for 7 days and in osteogenic medium (α-MEM supplemented with 10% FBS, 1% PSN, 50 µM ascorbic acid phosphate (Sigma, Osaka, Japan), 2 mM β-glycerophosphate (Sigma, St. Louis, MO, USA), 10 nM 1,25-dihydroxy-vitamin D3 (Sigma, Jerusalem, Israel) and 10 nM dexamethasone (Sigma, Shanghai, China) for an additional 7 days.
2.7.1. Cell Viability
The MTS assay (CellTiter 96® AQueous one solution cell proliferation assay; Promega, Madison, WI, USA) was carried out as an index of a viable cell number. After 24 h, scaffolds with cells were washed with α-MEM without FBS and placed in a new 24-well plate containing 700 µL of α-MEM w/o FBS. Then, 140 µL of MTS was added to each well and incubated for 2 h. The MTS assay was also conducted in wells used for a 24 h incubation of cell-laden scaffolds to estimate cell retention within scaffolds with different pore sizes. After 7 days of culture, hMSC viability was assessed qualitatively using live-dead fluorescent cell imagining kit (R37601, Molecular Probes, Thermo Fisher Scientific, Eugene, OR, USA). The scaffolds were washed twice with PBS and incubated with the mixture of live-dead dyes at 37 °C for 30 min. The cells were visualized using a fluorescent microscope Leica TCS SP8 (Leica Microsystems GmbH, Wetzlar, Germany) and filter sets corresponding to the fluorescence of interest.
2.7.2. Cell Differentiation
Alkaline phosphatase (ALP) activity was used as an indicator of hMSCs differentiation towards osteogenic lineage. After 7 and 14 days in culture, the samples were washed three times with PBS and frozen in 1 mL of deionized water. After thawing, the samples were vortexed and diluted with deionized water to yield concentrations suitable for further analysis. Total protein concentration (TPC) was determined using micro-BCA assay (QuantiPro BCA Assay Kit; Sigma, St. Louis, MO, USA). The ALP activity was measured spectrophotometrically using para-nitrophenyl phosphate (pNPP) as a substrate (Phosphatase Substrate Kit, Thermo Fisher Scientific, Rockford, IL, USA). Concentration of para-nitrophenol was calculated from a standard curve and normalized to the TPC.
2.7.3. Confocal and Scanning Electron Microscopy
Cells cultured for 1 day were fixed with a 3% glutaraldehyde solution. For confocal microscopy, the cells were labeled with DiD dye (D307; Molecular Probes, Thermo Fisher Scientific) and visualized at a 633 nm wavelength using a Leica TCS SP8 confocal microscope (Leica Microsystems GmbH, Wetzlar, Germany). For scanning electron microscopy, the fixed cells were washed 3 times with deionized water, followed by dehydration with an ethanol gradient (50%, 70%, 90% and 99%) and hexamethyldisilazane treatment (Sigma, St. Louis, MO, USA). The cells were visualized at an acceleration voltage of 5–10 keV in secondary electrons (SE) mode at various magnifications using a Hitachi SU8000 microscope (Hitachi Ltd., Tokyo, Japan).